diff options
Diffstat (limited to 'drivers/gpu/drm/amd/amdgpu')
172 files changed, 10211 insertions, 2169 deletions
diff --git a/drivers/gpu/drm/amd/amdgpu/Kconfig b/drivers/gpu/drm/amd/amdgpu/Kconfig index 9375e7f12420..74a8105fd2c0 100644 --- a/drivers/gpu/drm/amd/amdgpu/Kconfig +++ b/drivers/gpu/drm/amd/amdgpu/Kconfig @@ -34,15 +34,6 @@ config DRM_AMDGPU_USERPTR This option selects CONFIG_HMM and CONFIG_HMM_MIRROR if it isn't already selected to enabled full userptr support. -config DRM_AMDGPU_GART_DEBUGFS - bool "Allow GART access through debugfs" - depends on DRM_AMDGPU - depends on DEBUG_FS - default n - help - Selecting this option creates a debugfs file to inspect the mapped - pages. Uses more memory for housekeeping, enable only for debugging. - source "drivers/gpu/drm/amd/acp/Kconfig" source "drivers/gpu/drm/amd/display/Kconfig" source "drivers/gpu/drm/amd/amdkfd/Kconfig" diff --git a/drivers/gpu/drm/amd/amdgpu/Makefile b/drivers/gpu/drm/amd/amdgpu/Makefile index 13ebb1f71e49..ee85e8aba636 100644 --- a/drivers/gpu/drm/amd/amdgpu/Makefile +++ b/drivers/gpu/drm/amd/amdgpu/Makefile @@ -71,7 +71,7 @@ amdgpu-y += \ vi.o mxgpu_vi.o nbio_v6_1.o soc15.o emu_soc.o mxgpu_ai.o nbio_v7_0.o vega10_reg_init.o \ vega20_reg_init.o nbio_v7_4.o nbio_v2_3.o nv.o navi10_reg_init.o navi14_reg_init.o \ arct_reg_init.o navi12_reg_init.o mxgpu_nv.o sienna_cichlid_reg_init.o vangogh_reg_init.o \ - nbio_v7_2.o dimgrey_cavefish_reg_init.o hdp_v4_0.o hdp_v5_0.o + nbio_v7_2.o dimgrey_cavefish_reg_init.o hdp_v4_0.o hdp_v5_0.o aldebaran_reg_init.o aldebaran.o # add DF block amdgpu-y += \ @@ -83,11 +83,12 @@ amdgpu-y += \ gmc_v7_0.o \ gmc_v8_0.o \ gfxhub_v1_0.o mmhub_v1_0.o gmc_v9_0.o gfxhub_v1_1.o mmhub_v9_4.o \ - gfxhub_v2_0.o mmhub_v2_0.o gmc_v10_0.o gfxhub_v2_1.o mmhub_v2_3.o + gfxhub_v2_0.o mmhub_v2_0.o gmc_v10_0.o gfxhub_v2_1.o mmhub_v2_3.o \ + mmhub_v1_7.o # add UMC block amdgpu-y += \ - umc_v6_1.o umc_v6_0.o umc_v8_7.o + umc_v6_0.o umc_v6_1.o umc_v6_7.o umc_v8_7.o # add IH block amdgpu-y += \ @@ -106,7 +107,8 @@ amdgpu-y += \ psp_v3_1.o \ psp_v10_0.o \ psp_v11_0.o \ - psp_v12_0.o + psp_v12_0.o \ + psp_v13_0.o # add DCE block amdgpu-y += \ @@ -121,6 +123,7 @@ amdgpu-y += \ gfx_v8_0.o \ gfx_v9_0.o \ gfx_v9_4.o \ + gfx_v9_4_2.o \ gfx_v10_0.o # add async DMA block @@ -129,6 +132,7 @@ amdgpu-y += \ sdma_v2_4.o \ sdma_v3_0.o \ sdma_v4_0.o \ + sdma_v4_4.o \ sdma_v5_0.o \ sdma_v5_2.o @@ -172,11 +176,17 @@ amdgpu-y += \ amdgpu-y += \ smuio_v9_0.o \ smuio_v11_0.o \ - smuio_v11_0_6.o + smuio_v11_0_6.o \ + smuio_v13_0.o + +# add reset block +amdgpu-y += \ + amdgpu_reset.o # add amdkfd interfaces amdgpu-y += amdgpu_amdkfd.o + ifneq ($(CONFIG_HSA_AMD),) AMDKFD_PATH := ../amdkfd include $(FULL_AMD_PATH)/amdkfd/Makefile @@ -187,6 +197,7 @@ amdgpu-y += \ amdgpu_amdkfd_gfx_v8.o \ amdgpu_amdkfd_gfx_v9.o \ amdgpu_amdkfd_arcturus.o \ + amdgpu_amdkfd_aldebaran.o \ amdgpu_amdkfd_gfx_v10.o \ amdgpu_amdkfd_gfx_v10_3.o diff --git a/drivers/gpu/drm/amd/amdgpu/aldebaran.c b/drivers/gpu/drm/amd/amdgpu/aldebaran.c new file mode 100644 index 000000000000..65b1dca4b02e --- /dev/null +++ b/drivers/gpu/drm/amd/amdgpu/aldebaran.c @@ -0,0 +1,407 @@ +/* + * Copyright 2021 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + */ + +#include "aldebaran.h" +#include "amdgpu_reset.h" +#include "amdgpu_amdkfd.h" +#include "amdgpu_dpm.h" +#include "amdgpu_job.h" +#include "amdgpu_ring.h" +#include "amdgpu_ras.h" +#include "amdgpu_psp.h" +#include "amdgpu_xgmi.h" + +static struct amdgpu_reset_handler * +aldebaran_get_reset_handler(struct amdgpu_reset_control *reset_ctl, + struct amdgpu_reset_context *reset_context) +{ + struct amdgpu_reset_handler *handler; + struct amdgpu_device *adev = (struct amdgpu_device *)reset_ctl->handle; + + if (reset_context->method != AMD_RESET_METHOD_NONE) { + dev_dbg(adev->dev, "Getting reset handler for method %d\n", + reset_context->method); + list_for_each_entry(handler, &reset_ctl->reset_handlers, + handler_list) { + if (handler->reset_method == reset_context->method) + return handler; + } + } + + if (adev->gmc.xgmi.connected_to_cpu) { + list_for_each_entry(handler, &reset_ctl->reset_handlers, + handler_list) { + if (handler->reset_method == AMD_RESET_METHOD_MODE2) { + reset_context->method = AMD_RESET_METHOD_MODE2; + return handler; + } + } + } + + dev_dbg(adev->dev, "Reset handler not found!\n"); + + return NULL; +} + +static int aldebaran_mode2_suspend_ip(struct amdgpu_device *adev) +{ + int r, i; + + amdgpu_device_set_pg_state(adev, AMD_PG_STATE_UNGATE); + amdgpu_device_set_cg_state(adev, AMD_CG_STATE_UNGATE); + + for (i = adev->num_ip_blocks - 1; i >= 0; i--) { + if (!(adev->ip_blocks[i].version->type == + AMD_IP_BLOCK_TYPE_GFX || + adev->ip_blocks[i].version->type == + AMD_IP_BLOCK_TYPE_SDMA)) + continue; + + r = adev->ip_blocks[i].version->funcs->suspend(adev); + + if (r) { + dev_err(adev->dev, + "suspend of IP block <%s> failed %d\n", + adev->ip_blocks[i].version->funcs->name, r); + return r; + } + + adev->ip_blocks[i].status.hw = false; + } + + return r; +} + +static int +aldebaran_mode2_prepare_hwcontext(struct amdgpu_reset_control *reset_ctl, + struct amdgpu_reset_context *reset_context) +{ + int r = 0; + struct amdgpu_device *adev = (struct amdgpu_device *)reset_ctl->handle; + + dev_dbg(adev->dev, "Aldebaran prepare hw context\n"); + /* Don't suspend on bare metal if we are not going to HW reset the ASIC */ + if (!amdgpu_sriov_vf(adev)) + r = aldebaran_mode2_suspend_ip(adev); + + return r; +} + +static void aldebaran_async_reset(struct work_struct *work) +{ + struct amdgpu_reset_handler *handler; + struct amdgpu_reset_control *reset_ctl = + container_of(work, struct amdgpu_reset_control, reset_work); + struct amdgpu_device *adev = (struct amdgpu_device *)reset_ctl->handle; + + list_for_each_entry(handler, &reset_ctl->reset_handlers, + handler_list) { + if (handler->reset_method == reset_ctl->active_reset) { + dev_dbg(adev->dev, "Resetting device\n"); + handler->do_reset(adev); + break; + } + } +} + +static int aldebaran_mode2_reset(struct amdgpu_device *adev) +{ + /* disable BM */ + pci_clear_master(adev->pdev); + adev->asic_reset_res = amdgpu_dpm_mode2_reset(adev); + return adev->asic_reset_res; +} + +static int +aldebaran_mode2_perform_reset(struct amdgpu_reset_control *reset_ctl, + struct amdgpu_reset_context *reset_context) +{ + struct amdgpu_device *tmp_adev = NULL; + struct amdgpu_device *adev = (struct amdgpu_device *)reset_ctl->handle; + int r = 0; + + dev_dbg(adev->dev, "aldebaran perform hw reset\n"); + if (reset_context->hive == NULL) { + /* Wrong context, return error */ + return -EINVAL; + } + + list_for_each_entry(tmp_adev, &reset_context->hive->device_list, + gmc.xgmi.head) { + mutex_lock(&tmp_adev->reset_cntl->reset_lock); + tmp_adev->reset_cntl->active_reset = AMD_RESET_METHOD_MODE2; + } + /* + * Mode2 reset doesn't need any sync between nodes in XGMI hive, instead launch + * them together so that they can be completed asynchronously on multiple nodes + */ + list_for_each_entry(tmp_adev, &reset_context->hive->device_list, + gmc.xgmi.head) { + /* For XGMI run all resets in parallel to speed up the process */ + if (tmp_adev->gmc.xgmi.num_physical_nodes > 1) { + if (!queue_work(system_unbound_wq, + &tmp_adev->reset_cntl->reset_work)) + r = -EALREADY; + } else + r = aldebaran_mode2_reset(tmp_adev); + if (r) { + dev_err(tmp_adev->dev, + "ASIC reset failed with error, %d for drm dev, %s", + r, adev_to_drm(tmp_adev)->unique); + break; + } + } + + /* For XGMI wait for all resets to complete before proceed */ + if (!r) { + list_for_each_entry(tmp_adev, + &reset_context->hive->device_list, + gmc.xgmi.head) { + if (tmp_adev->gmc.xgmi.num_physical_nodes > 1) { + flush_work(&tmp_adev->reset_cntl->reset_work); + r = tmp_adev->asic_reset_res; + if (r) + break; + } + } + } + + list_for_each_entry(tmp_adev, &reset_context->hive->device_list, + gmc.xgmi.head) { + mutex_unlock(&tmp_adev->reset_cntl->reset_lock); + tmp_adev->reset_cntl->active_reset = AMD_RESET_METHOD_NONE; + } + + return r; +} + +static int aldebaran_mode2_restore_ip(struct amdgpu_device *adev) +{ + struct amdgpu_firmware_info *ucode_list[AMDGPU_UCODE_ID_MAXIMUM]; + struct amdgpu_firmware_info *ucode; + struct amdgpu_ip_block *cmn_block; + int ucode_count = 0; + int i, r; + + dev_dbg(adev->dev, "Reloading ucodes after reset\n"); + for (i = 0; i < adev->firmware.max_ucodes; i++) { + ucode = &adev->firmware.ucode[i]; + if (!ucode->fw) + continue; + switch (ucode->ucode_id) { + case AMDGPU_UCODE_ID_SDMA0: + case AMDGPU_UCODE_ID_SDMA1: + case AMDGPU_UCODE_ID_SDMA2: + case AMDGPU_UCODE_ID_SDMA3: + case AMDGPU_UCODE_ID_SDMA4: + case AMDGPU_UCODE_ID_SDMA5: + case AMDGPU_UCODE_ID_SDMA6: + case AMDGPU_UCODE_ID_SDMA7: + case AMDGPU_UCODE_ID_CP_MEC1: + case AMDGPU_UCODE_ID_CP_MEC1_JT: + case AMDGPU_UCODE_ID_RLC_RESTORE_LIST_CNTL: + case AMDGPU_UCODE_ID_RLC_RESTORE_LIST_GPM_MEM: + case AMDGPU_UCODE_ID_RLC_RESTORE_LIST_SRM_MEM: + case AMDGPU_UCODE_ID_RLC_G: + ucode_list[ucode_count++] = ucode; + break; + default: + break; + }; + } + + /* Reinit NBIF block */ + cmn_block = + amdgpu_device_ip_get_ip_block(adev, AMD_IP_BLOCK_TYPE_COMMON); + if (unlikely(!cmn_block)) { + dev_err(adev->dev, "Failed to get BIF handle\n"); + return -EINVAL; + } + r = cmn_block->version->funcs->resume(adev); + if (r) + return r; + + /* Reinit GFXHUB */ + adev->gfxhub.funcs->init(adev); + r = adev->gfxhub.funcs->gart_enable(adev); + if (r) { + dev_err(adev->dev, "GFXHUB gart reenable failed after reset\n"); + return r; + } + + /* Reload GFX firmware */ + r = psp_load_fw_list(&adev->psp, ucode_list, ucode_count); + if (r) { + dev_err(adev->dev, "GFX ucode load failed after reset\n"); + return r; + } + + /* Resume RLC, FW needs RLC alive to complete reset process */ + adev->gfx.rlc.funcs->resume(adev); + + /* Wait for FW reset event complete */ + r = smu_wait_for_event(adev, SMU_EVENT_RESET_COMPLETE, 0); + if (r) { + dev_err(adev->dev, + "Failed to get response from firmware after reset\n"); + return r; + } + + for (i = 0; i < adev->num_ip_blocks; i++) { + if (!(adev->ip_blocks[i].version->type == + AMD_IP_BLOCK_TYPE_GFX || + adev->ip_blocks[i].version->type == + AMD_IP_BLOCK_TYPE_SDMA)) + continue; + r = adev->ip_blocks[i].version->funcs->resume(adev); + if (r) { + dev_err(adev->dev, + "resume of IP block <%s> failed %d\n", + adev->ip_blocks[i].version->funcs->name, r); + return r; + } + + adev->ip_blocks[i].status.hw = true; + } + + for (i = 0; i < adev->num_ip_blocks; i++) { + if (!(adev->ip_blocks[i].version->type == + AMD_IP_BLOCK_TYPE_GFX || + adev->ip_blocks[i].version->type == + AMD_IP_BLOCK_TYPE_SDMA || + adev->ip_blocks[i].version->type == + AMD_IP_BLOCK_TYPE_COMMON)) + continue; + + if (adev->ip_blocks[i].version->funcs->late_init) { + r = adev->ip_blocks[i].version->funcs->late_init( + (void *)adev); + if (r) { + dev_err(adev->dev, + "late_init of IP block <%s> failed %d after reset\n", + adev->ip_blocks[i].version->funcs->name, + r); + return r; + } + } + adev->ip_blocks[i].status.late_initialized = true; + } + + amdgpu_device_set_cg_state(adev, AMD_CG_STATE_GATE); + amdgpu_device_set_pg_state(adev, AMD_PG_STATE_GATE); + + return r; +} + +static int +aldebaran_mode2_restore_hwcontext(struct amdgpu_reset_control *reset_ctl, + struct amdgpu_reset_context *reset_context) +{ + int r; + struct amdgpu_device *tmp_adev = NULL; + + if (reset_context->hive == NULL) { + /* Wrong context, return error */ + return -EINVAL; + } + + list_for_each_entry(tmp_adev, &reset_context->hive->device_list, + gmc.xgmi.head) { + dev_info(tmp_adev->dev, + "GPU reset succeeded, trying to resume\n"); + r = aldebaran_mode2_restore_ip(tmp_adev); + if (r) + goto end; + + /* + * Add this ASIC as tracked as reset was already + * complete successfully. + */ + amdgpu_register_gpu_instance(tmp_adev); + + /* Resume RAS */ + amdgpu_ras_resume(tmp_adev); + + /* Update PSP FW topology after reset */ + if (reset_context->hive && + tmp_adev->gmc.xgmi.num_physical_nodes > 1) + r = amdgpu_xgmi_update_topology(reset_context->hive, + tmp_adev); + + if (!r) { + amdgpu_irq_gpu_reset_resume_helper(tmp_adev); + + r = amdgpu_ib_ring_tests(tmp_adev); + if (r) { + dev_err(tmp_adev->dev, + "ib ring test failed (%d).\n", r); + r = -EAGAIN; + tmp_adev->asic_reset_res = r; + goto end; + } + } + } + +end: + return r; +} + +static struct amdgpu_reset_handler aldebaran_mode2_handler = { + .reset_method = AMD_RESET_METHOD_MODE2, + .prepare_env = NULL, + .prepare_hwcontext = aldebaran_mode2_prepare_hwcontext, + .perform_reset = aldebaran_mode2_perform_reset, + .restore_hwcontext = aldebaran_mode2_restore_hwcontext, + .restore_env = NULL, + .do_reset = aldebaran_mode2_reset, +}; + +int aldebaran_reset_init(struct amdgpu_device *adev) +{ + struct amdgpu_reset_control *reset_ctl; + + reset_ctl = kzalloc(sizeof(*reset_ctl), GFP_KERNEL); + if (!reset_ctl) + return -ENOMEM; + + reset_ctl->handle = adev; + reset_ctl->async_reset = aldebaran_async_reset; + reset_ctl->active_reset = AMD_RESET_METHOD_NONE; + reset_ctl->get_reset_handler = aldebaran_get_reset_handler; + + INIT_LIST_HEAD(&reset_ctl->reset_handlers); + INIT_WORK(&reset_ctl->reset_work, reset_ctl->async_reset); + /* Only mode2 is handled through reset control now */ + amdgpu_reset_add_handler(reset_ctl, &aldebaran_mode2_handler); + + adev->reset_cntl = reset_ctl; + + return 0; +} + +int aldebaran_reset_fini(struct amdgpu_device *adev) +{ + kfree(adev->reset_cntl); + adev->reset_cntl = NULL; + return 0; +} diff --git a/drivers/gpu/drm/amd/amdgpu/aldebaran.h b/drivers/gpu/drm/amd/amdgpu/aldebaran.h new file mode 100644 index 000000000000..a07db5454d49 --- /dev/null +++ b/drivers/gpu/drm/amd/amdgpu/aldebaran.h @@ -0,0 +1,32 @@ +/* + * Copyright 2021 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + */ + +#ifndef __ALDEBARAN_H__ +#define __ALDEBARAN_H__ + +#include "amdgpu.h" + +int aldebaran_reset_init(struct amdgpu_device *adev); +int aldebaran_reset_fini(struct amdgpu_device *adev); + +#endif diff --git a/drivers/gpu/drm/amd/amdgpu/aldebaran_reg_init.c b/drivers/gpu/drm/amd/amdgpu/aldebaran_reg_init.c new file mode 100644 index 000000000000..28e6c9ab8767 --- /dev/null +++ b/drivers/gpu/drm/amd/amdgpu/aldebaran_reg_init.c @@ -0,0 +1,54 @@ +/* + * Copyright 2020 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + */ +#include "amdgpu.h" +#include "soc15.h" + +#include "soc15_common.h" +#include "aldebaran_ip_offset.h" + +int aldebaran_reg_base_init(struct amdgpu_device *adev) +{ + /* HW has more IP blocks, only initialized the block needed by our driver */ + uint32_t i; + for (i = 0 ; i < MAX_INSTANCE ; ++i) { + adev->reg_offset[GC_HWIP][i] = (uint32_t *)(&(GC_BASE.instance[i])); + adev->reg_offset[HDP_HWIP][i] = (uint32_t *)(&(HDP_BASE.instance[i])); + adev->reg_offset[MMHUB_HWIP][i] = (uint32_t *)(&(MMHUB_BASE.instance[i])); + adev->reg_offset[ATHUB_HWIP][i] = (uint32_t *)(&(ATHUB_BASE.instance[i])); + adev->reg_offset[NBIO_HWIP][i] = (uint32_t *)(&(NBIO_BASE.instance[i])); + adev->reg_offset[MP0_HWIP][i] = (uint32_t *)(&(MP0_BASE.instance[i])); + adev->reg_offset[MP1_HWIP][i] = (uint32_t *)(&(MP1_BASE.instance[i])); + adev->reg_offset[DF_HWIP][i] = (uint32_t *)(&(DF_BASE.instance[i])); + adev->reg_offset[OSSSYS_HWIP][i] = (uint32_t *)(&(OSSSYS_BASE.instance[i])); + adev->reg_offset[SDMA0_HWIP][i] = (uint32_t *)(&(SDMA0_BASE.instance[i])); + adev->reg_offset[SDMA1_HWIP][i] = (uint32_t *)(&(SDMA1_BASE.instance[i])); + adev->reg_offset[SDMA2_HWIP][i] = (uint32_t *)(&(SDMA2_BASE.instance[i])); + adev->reg_offset[SDMA3_HWIP][i] = (uint32_t *)(&(SDMA3_BASE.instance[i])); + adev->reg_offset[SDMA4_HWIP][i] = (uint32_t *)(&(SDMA4_BASE.instance[i])); + adev->reg_offset[SMUIO_HWIP][i] = (uint32_t *)(&(SMUIO_BASE.instance[i])); + adev->reg_offset[THM_HWIP][i] = (uint32_t *)(&(THM_BASE.instance[i])); + adev->reg_offset[UMC_HWIP][i] = (uint32_t *)(&(UMC_BASE.instance[i])); + adev->reg_offset[VCN_HWIP][i] = (uint32_t *)(&(VCN_BASE.instance[i])); + } + return 0; +} diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu.h b/drivers/gpu/drm/amd/amdgpu/amdgpu.h index 29885febc0b0..dc3a69296321 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu.h @@ -107,7 +107,6 @@ #include "amdgpu_gfxhub.h" #include "amdgpu_df.h" #include "amdgpu_smuio.h" -#include "amdgpu_hdp.h" #define MAX_GPU_INSTANCE 16 @@ -124,6 +123,16 @@ struct amdgpu_mgpu_info uint32_t num_gpu; uint32_t num_dgpu; uint32_t num_apu; + + /* delayed reset_func for XGMI configuration if necessary */ + struct delayed_work delayed_reset_work; + bool pending_reset; +}; + +struct amdgpu_watchdog_timer +{ + bool timeout_fatal_disable; + uint32_t period; /* maxCycles = (1 << period), the number of cycles before a timeout */ }; #define AMDGPU_MAX_TIMEOUT_PARAM_LENGTH 256 @@ -177,7 +186,9 @@ extern int amdgpu_compute_multipipe; extern int amdgpu_gpu_recovery; extern int amdgpu_emu_mode; extern uint amdgpu_smu_memory_pool_size; +extern int amdgpu_smu_pptable_id; extern uint amdgpu_dc_feature_mask; +extern uint amdgpu_freesync_vid_mode; extern uint amdgpu_dc_debug_mask; extern uint amdgpu_dm_abm_level; extern int amdgpu_backlight; @@ -185,6 +196,7 @@ extern struct amdgpu_mgpu_info mgpu_info; extern int amdgpu_ras_enable; extern uint amdgpu_ras_mask; extern int amdgpu_bad_page_threshold; +extern struct amdgpu_watchdog_timer amdgpu_watchdog_timer; extern int amdgpu_async_gfx_ring; extern int amdgpu_mcbp; extern int amdgpu_discovery; @@ -258,6 +270,8 @@ struct amdgpu_bo_va_mapping; struct amdgpu_atif; struct kfd_vm_fault_info; struct amdgpu_hive_info; +struct amdgpu_reset_context; +struct amdgpu_reset_control; enum amdgpu_cp_irq { AMDGPU_CP_IRQ_GFX_ME0_PIPE0_EOP = 0, @@ -576,6 +590,7 @@ struct amdgpu_allowed_register_entry { }; enum amd_reset_method { + AMD_RESET_METHOD_NONE = -1, AMD_RESET_METHOD_LEGACY = 0, AMD_RESET_METHOD_MODE0, AMD_RESET_METHOD_MODE1, @@ -584,6 +599,19 @@ enum amd_reset_method { AMD_RESET_METHOD_PCI, }; +struct amdgpu_video_codec_info { + u32 codec_type; + u32 max_width; + u32 max_height; + u32 max_pixels_per_frame; + u32 max_level; +}; + +struct amdgpu_video_codecs { + const u32 codec_count; + const struct amdgpu_video_codec_info *codec_array; +}; + /* * ASIC specific functions. */ @@ -628,6 +656,9 @@ struct amdgpu_asic_funcs { void (*pre_asic_init)(struct amdgpu_device *adev); /* enter/exit umd stable pstate */ int (*update_umd_stable_pstate)(struct amdgpu_device *adev, bool enter); + /* query video codecs */ + int (*query_video_codecs)(struct amdgpu_device *adev, bool encode, + const struct amdgpu_video_codecs **codecs); }; /* @@ -792,12 +823,7 @@ struct amdgpu_device { bool accel_working; struct notifier_block acpi_nb; struct amdgpu_i2c_chan *i2c_bus[AMDGPU_MAX_I2C_BUS]; - struct amdgpu_debugfs debugfs[AMDGPU_DEBUGFS_MAX_COMPONENTS]; - unsigned debugfs_count; -#if defined(CONFIG_DEBUG_FS) - struct dentry *debugfs_preempt; - struct dentry *debugfs_regs[AMDGPU_DEBUGFS_MAX_COMPONENTS]; -#endif + struct debugfs_blob_wrapper debugfs_vbios_blob; struct amdgpu_atif *atif; struct amdgpu_atcs atcs; struct mutex srbm_mutex; @@ -853,8 +879,6 @@ struct amdgpu_device { spinlock_t audio_endpt_idx_lock; amdgpu_block_rreg_t audio_endpt_rreg; amdgpu_block_wreg_t audio_endpt_wreg; - void __iomem *rio_mem; - resource_size_t rio_mem_size; struct amdgpu_doorbell doorbell; /* clock/pll info */ @@ -897,6 +921,8 @@ struct amdgpu_device { struct amdgpu_irq_src vupdate_irq; struct amdgpu_irq_src pageflip_irq; struct amdgpu_irq_src hpd_irq; + struct amdgpu_irq_src dmub_trace_irq; + struct amdgpu_irq_src dmub_outbox_irq; /* rings */ u64 fence_context; @@ -1020,6 +1046,7 @@ struct amdgpu_device { int asic_reset_res; struct work_struct xgmi_reset_work; + struct list_head reset_list; long gfx_timeout; long sdma_timeout; @@ -1050,6 +1077,8 @@ struct amdgpu_device { bool in_pci_err_recovery; struct pci_saved_state *pci_state; + + struct amdgpu_reset_control *reset_cntl; }; static inline struct amdgpu_device *drm_to_adev(struct drm_device *ddev) @@ -1062,7 +1091,7 @@ static inline struct drm_device *adev_to_drm(struct amdgpu_device *adev) return &adev->ddev; } -static inline struct amdgpu_device *amdgpu_ttm_adev(struct ttm_bo_device *bdev) +static inline struct amdgpu_device *amdgpu_ttm_adev(struct ttm_device *bdev) { return container_of(bdev, struct amdgpu_device, mman.bdev); } @@ -1084,9 +1113,6 @@ void amdgpu_mm_wreg_mmio_rlc(struct amdgpu_device *adev, void amdgpu_mm_wreg8(struct amdgpu_device *adev, uint32_t offset, uint8_t value); uint8_t amdgpu_mm_rreg8(struct amdgpu_device *adev, uint32_t offset); -u32 amdgpu_io_rreg(struct amdgpu_device *adev, u32 reg); -void amdgpu_io_wreg(struct amdgpu_device *adev, u32 reg, u32 v); - u32 amdgpu_device_indirect_rreg(struct amdgpu_device *adev, u32 pcie_index, u32 pcie_data, u32 reg_addr); @@ -1103,6 +1129,12 @@ void amdgpu_device_indirect_wreg64(struct amdgpu_device *adev, bool amdgpu_device_asic_has_dc_support(enum amd_asic_type asic_type); bool amdgpu_device_has_dc_support(struct amdgpu_device *adev); +int amdgpu_device_pre_asic_reset(struct amdgpu_device *adev, + struct amdgpu_reset_context *reset_context); + +int amdgpu_do_asic_reset(struct list_head *device_list_handle, + struct amdgpu_reset_context *reset_context); + int emu_soc_asic_init(struct amdgpu_device *adev); /* @@ -1168,8 +1200,6 @@ int emu_soc_asic_init(struct amdgpu_device *adev); } while (0) #define DREG32_SYS(sqf, adev, reg) seq_printf((sqf), #reg " : 0x%08X\n", amdgpu_device_rreg((adev), (reg), false)) -#define RREG32_IO(reg) amdgpu_io_rreg(adev, (reg)) -#define WREG32_IO(reg, v) amdgpu_io_wreg(adev, (reg), (v)) #define REG_FIELD_SHIFT(reg, field) reg##__##field##__SHIFT #define REG_FIELD_MASK(reg, field) reg##__##field##_MASK @@ -1223,6 +1253,7 @@ int emu_soc_asic_init(struct amdgpu_device *adev); #define amdgpu_asic_pre_asic_init(adev) (adev)->asic_funcs->pre_asic_init((adev)) #define amdgpu_asic_update_umd_stable_pstate(adev, enter) \ ((adev)->asic_funcs->update_umd_stable_pstate ? (adev)->asic_funcs->update_umd_stable_pstate((adev), (enter)) : 0) +#define amdgpu_asic_query_video_codecs(adev, e, c) (adev)->asic_funcs->query_video_codecs((adev), (e), (c)) #define amdgpu_inc_vram_lost(adev) atomic_inc(&((adev)->vram_lost_counter)); @@ -1242,7 +1273,9 @@ void amdgpu_device_program_register_sequence(struct amdgpu_device *adev, const u32 *registers, const u32 array_size); +int amdgpu_device_mode1_reset(struct amdgpu_device *adev); bool amdgpu_device_supports_atpx(struct drm_device *dev); +bool amdgpu_device_supports_px(struct drm_device *dev); bool amdgpu_device_supports_boco(struct drm_device *dev); bool amdgpu_device_supports_baco(struct drm_device *dev); bool amdgpu_device_is_peer_accessible(struct amdgpu_device *adev, @@ -1356,6 +1389,13 @@ void amdgpu_pci_resume(struct pci_dev *pdev); bool amdgpu_device_cache_pci_state(struct pci_dev *pdev); bool amdgpu_device_load_pci_state(struct pci_dev *pdev); +bool amdgpu_device_skip_hw_access(struct amdgpu_device *adev); + +int amdgpu_device_set_cg_state(struct amdgpu_device *adev, + enum amd_clockgating_state state); +int amdgpu_device_set_pg_state(struct amdgpu_device *adev, + enum amd_powergating_state state); + #include "amdgpu_object.h" static inline bool amdgpu_is_tmz(struct amdgpu_device *adev) diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.c index c5343a5eecbe..5f6696a3c778 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.c @@ -44,7 +44,7 @@ int amdgpu_amdkfd_init(void) int ret; si_meminfo(&si); - amdgpu_amdkfd_total_mem_size = si.totalram - si.totalhigh; + amdgpu_amdkfd_total_mem_size = si.freeram - si.freehigh; amdgpu_amdkfd_total_mem_size *= si.mem_unit; ret = kgd2kfd_init(); @@ -165,7 +165,8 @@ void amdgpu_amdkfd_device_init(struct amdgpu_device *adev) adev->doorbell_index.last_non_cp; } - kgd2kfd_device_init(adev->kfd.dev, adev_to_drm(adev), &gpu_resources); + adev->kfd.init_complete = kgd2kfd_device_init(adev->kfd.dev, + adev_to_drm(adev), &gpu_resources); } } @@ -245,6 +246,7 @@ int amdgpu_amdkfd_alloc_gtt_mem(struct kgd_dev *kgd, size_t size, bp.flags = AMDGPU_GEM_CREATE_CPU_GTT_USWC; bp.type = ttm_bo_type_kernel; bp.resv = NULL; + bp.bo_ptr_size = sizeof(struct amdgpu_bo); if (cp_mqd_gfx9) bp.flags |= AMDGPU_GEM_CREATE_CP_MQD_GFX9; @@ -316,6 +318,7 @@ int amdgpu_amdkfd_alloc_gws(struct kgd_dev *kgd, size_t size, { struct amdgpu_device *adev = (struct amdgpu_device *)kgd; struct amdgpu_bo *bo = NULL; + struct amdgpu_bo_user *ubo; struct amdgpu_bo_param bp; int r; @@ -326,14 +329,16 @@ int amdgpu_amdkfd_alloc_gws(struct kgd_dev *kgd, size_t size, bp.flags = AMDGPU_GEM_CREATE_NO_CPU_ACCESS; bp.type = ttm_bo_type_device; bp.resv = NULL; + bp.bo_ptr_size = sizeof(struct amdgpu_bo); - r = amdgpu_bo_create(adev, &bp, &bo); + r = amdgpu_bo_create_user(adev, &bp, &ubo); if (r) { dev_err(adev->dev, "failed to allocate gws BO for amdkfd (%d)\n", r); return r; } + bo = &ubo->bo; *mem_obj = bo; return 0; } @@ -494,8 +499,6 @@ int amdgpu_amdkfd_get_dmabuf_info(struct kgd_dev *kgd, int dma_buf_fd, *dma_buf_kgd = (struct kgd_dev *)adev; if (bo_size) *bo_size = amdgpu_bo_size(bo); - if (metadata_size) - *metadata_size = bo->metadata_size; if (metadata_buffer) r = amdgpu_bo_get_metadata(bo, metadata_buffer, buffer_size, metadata_size, &metadata_flags); @@ -638,13 +641,6 @@ void amdgpu_amdkfd_set_compute_idle(struct kgd_dev *kgd, bool idle) { struct amdgpu_device *adev = (struct amdgpu_device *)kgd; - /* Temp workaround to fix the soft hang observed in certain compute - * applications if GFXOFF is enabled. - */ - if (adev->asic_type == CHIP_SIENNA_CICHLID) { - pr_debug("GFXOFF is %s\n", idle ? "enabled" : "disabled"); - amdgpu_gfx_off_ctrl(adev, idle); - } amdgpu_dpm_switch_power_profile(adev, PP_SMC_POWER_PROFILE_COMPUTE, !idle); diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.h index a81d9cacf9b8..5ffb07b02810 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.h @@ -80,6 +80,7 @@ struct amdgpu_amdkfd_fence { struct amdgpu_kfd_dev { struct kfd_dev *dev; uint64_t vram_used; + bool init_complete; }; enum kgd_engine_type { @@ -233,14 +234,10 @@ uint8_t amdgpu_amdkfd_get_xgmi_hops_count(struct kgd_dev *dst, struct kgd_dev *s }) /* GPUVM API */ -int amdgpu_amdkfd_gpuvm_create_process_vm(struct kgd_dev *kgd, u32 pasid, - void **vm, void **process_info, - struct dma_fence **ef); int amdgpu_amdkfd_gpuvm_acquire_process_vm(struct kgd_dev *kgd, struct file *filp, u32 pasid, void **vm, void **process_info, struct dma_fence **ef); -void amdgpu_amdkfd_gpuvm_destroy_process_vm(struct kgd_dev *kgd, void *vm); void amdgpu_amdkfd_gpuvm_release_process_vm(struct kgd_dev *kgd, void *vm); uint64_t amdgpu_amdkfd_gpuvm_get_process_page_dir(void *vm); int amdgpu_amdkfd_gpuvm_alloc_memory_of_gpu( diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_aldebaran.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_aldebaran.c new file mode 100644 index 000000000000..a5434b713856 --- /dev/null +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_aldebaran.c @@ -0,0 +1,47 @@ +/* + * Copyright 2020 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + */ +#include "amdgpu.h" +#include "amdgpu_amdkfd.h" +#include "amdgpu_amdkfd_arcturus.h" +#include "amdgpu_amdkfd_gfx_v9.h" + +const struct kfd2kgd_calls aldebaran_kfd2kgd = { + .program_sh_mem_settings = kgd_gfx_v9_program_sh_mem_settings, + .set_pasid_vmid_mapping = kgd_gfx_v9_set_pasid_vmid_mapping, + .init_interrupts = kgd_gfx_v9_init_interrupts, + .hqd_load = kgd_gfx_v9_hqd_load, + .hiq_mqd_load = kgd_gfx_v9_hiq_mqd_load, + .hqd_sdma_load = kgd_arcturus_hqd_sdma_load, + .hqd_dump = kgd_gfx_v9_hqd_dump, + .hqd_sdma_dump = kgd_arcturus_hqd_sdma_dump, + .hqd_is_occupied = kgd_gfx_v9_hqd_is_occupied, + .hqd_sdma_is_occupied = kgd_arcturus_hqd_sdma_is_occupied, + .hqd_destroy = kgd_gfx_v9_hqd_destroy, + .hqd_sdma_destroy = kgd_arcturus_hqd_sdma_destroy, + .address_watch_disable = kgd_gfx_v9_address_watch_disable, + .address_watch_execute = kgd_gfx_v9_address_watch_execute, + .wave_control_execute = kgd_gfx_v9_wave_control_execute, + .address_watch_get_offset = kgd_gfx_v9_address_watch_get_offset, + .get_atc_vmid_pasid_mapping_info = + kgd_gfx_v9_get_atc_vmid_pasid_mapping_info, + .set_vm_context_page_table_base = kgd_gfx_v9_set_vm_context_page_table_base, +}; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_arcturus.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_arcturus.c index 604757a1e440..9ef9f3ddad48 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_arcturus.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_arcturus.c @@ -122,7 +122,7 @@ static uint32_t get_sdma_rlc_reg_offset(struct amdgpu_device *adev, return sdma_rlc_reg_offset; } -static int kgd_hqd_sdma_load(struct kgd_dev *kgd, void *mqd, +int kgd_arcturus_hqd_sdma_load(struct kgd_dev *kgd, void *mqd, uint32_t __user *wptr, struct mm_struct *mm) { struct amdgpu_device *adev = get_amdgpu_device(kgd); @@ -192,7 +192,7 @@ static int kgd_hqd_sdma_load(struct kgd_dev *kgd, void *mqd, return 0; } -static int kgd_hqd_sdma_dump(struct kgd_dev *kgd, +int kgd_arcturus_hqd_sdma_dump(struct kgd_dev *kgd, uint32_t engine_id, uint32_t queue_id, uint32_t (**dump)[2], uint32_t *n_regs) { @@ -224,7 +224,7 @@ static int kgd_hqd_sdma_dump(struct kgd_dev *kgd, return 0; } -static bool kgd_hqd_sdma_is_occupied(struct kgd_dev *kgd, void *mqd) +bool kgd_arcturus_hqd_sdma_is_occupied(struct kgd_dev *kgd, void *mqd) { struct amdgpu_device *adev = get_amdgpu_device(kgd); struct v9_sdma_mqd *m; @@ -243,7 +243,7 @@ static bool kgd_hqd_sdma_is_occupied(struct kgd_dev *kgd, void *mqd) return false; } -static int kgd_hqd_sdma_destroy(struct kgd_dev *kgd, void *mqd, +int kgd_arcturus_hqd_sdma_destroy(struct kgd_dev *kgd, void *mqd, unsigned int utimeout) { struct amdgpu_device *adev = get_amdgpu_device(kgd); @@ -289,13 +289,13 @@ const struct kfd2kgd_calls arcturus_kfd2kgd = { .init_interrupts = kgd_gfx_v9_init_interrupts, .hqd_load = kgd_gfx_v9_hqd_load, .hiq_mqd_load = kgd_gfx_v9_hiq_mqd_load, - .hqd_sdma_load = kgd_hqd_sdma_load, + .hqd_sdma_load = kgd_arcturus_hqd_sdma_load, .hqd_dump = kgd_gfx_v9_hqd_dump, - .hqd_sdma_dump = kgd_hqd_sdma_dump, + .hqd_sdma_dump = kgd_arcturus_hqd_sdma_dump, .hqd_is_occupied = kgd_gfx_v9_hqd_is_occupied, - .hqd_sdma_is_occupied = kgd_hqd_sdma_is_occupied, + .hqd_sdma_is_occupied = kgd_arcturus_hqd_sdma_is_occupied, .hqd_destroy = kgd_gfx_v9_hqd_destroy, - .hqd_sdma_destroy = kgd_hqd_sdma_destroy, + .hqd_sdma_destroy = kgd_arcturus_hqd_sdma_destroy, .address_watch_disable = kgd_gfx_v9_address_watch_disable, .address_watch_execute = kgd_gfx_v9_address_watch_execute, .wave_control_execute = kgd_gfx_v9_wave_control_execute, diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_arcturus.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_arcturus.h new file mode 100644 index 000000000000..ce08131b7b5f --- /dev/null +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_arcturus.h @@ -0,0 +1,30 @@ +/* + * Copyright 2020 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + */ + +int kgd_arcturus_hqd_sdma_load(struct kgd_dev *kgd, void *mqd, + uint32_t __user *wptr, struct mm_struct *mm); +int kgd_arcturus_hqd_sdma_dump(struct kgd_dev *kgd, + uint32_t engine_id, uint32_t queue_id, + uint32_t (**dump)[2], uint32_t *n_regs); +bool kgd_arcturus_hqd_sdma_is_occupied(struct kgd_dev *kgd, void *mqd); +int kgd_arcturus_hqd_sdma_destroy(struct kgd_dev *kgd, void *mqd, + unsigned int utimeout); diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_fence.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_fence.c index 3107b9575929..5af464933976 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_fence.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_fence.c @@ -40,13 +40,13 @@ static atomic_t fence_seq = ATOMIC_INIT(0); * All the BOs in a process share an eviction fence. When process X wants * to map VRAM memory but TTM can't find enough space, TTM will attempt to * evict BOs from its LRU list. TTM checks if the BO is valuable to evict - * by calling ttm_bo_driver->eviction_valuable(). + * by calling ttm_device_funcs->eviction_valuable(). * - * ttm_bo_driver->eviction_valuable() - will return false if the BO belongs + * ttm_device_funcs->eviction_valuable() - will return false if the BO belongs * to process X. Otherwise, it will return true to indicate BO can be * evicted by TTM. * - * If ttm_bo_driver->eviction_valuable returns true, then TTM will continue + * If ttm_device_funcs->eviction_valuable returns true, then TTM will continue * the evcition process for that BO by calling ttm_bo_evict --> amdgpu_bo_move * --> amdgpu_copy_buffer(). This sets up job in GPU scheduler. * diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gpuvm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gpuvm.c index ac0a432a9bf7..7d4118c8128a 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gpuvm.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gpuvm.c @@ -31,6 +31,7 @@ #include "amdgpu_amdkfd.h" #include "amdgpu_dma_buf.h" #include <uapi/linux/kfd_ioctl.h> +#include "amdgpu_xgmi.h" /* BO flag to indicate a KFD userptr BO */ #define AMDGPU_AMDKFD_USERPTR_BO (1ULL << 63) @@ -96,7 +97,7 @@ void amdgpu_amdkfd_gpuvm_init_mem_limits(void) uint64_t mem; si_meminfo(&si); - mem = si.totalram - si.totalhigh; + mem = si.freeram - si.freehigh; mem *= si.mem_unit; spin_lock_init(&kfd_mem_limit.mem_limit_lock); @@ -119,6 +120,16 @@ void amdgpu_amdkfd_gpuvm_init_mem_limits(void) */ #define ESTIMATE_PT_SIZE(mem_size) ((mem_size) >> 14) +static size_t amdgpu_amdkfd_acc_size(uint64_t size) +{ + size >>= PAGE_SHIFT; + size *= sizeof(dma_addr_t) + sizeof(void *); + + return __roundup_pow_of_two(sizeof(struct amdgpu_bo)) + + __roundup_pow_of_two(sizeof(struct ttm_tt)) + + PAGE_ALIGN(size); +} + static int amdgpu_amdkfd_reserve_mem_limit(struct amdgpu_device *adev, uint64_t size, u32 domain, bool sg) { @@ -127,8 +138,7 @@ static int amdgpu_amdkfd_reserve_mem_limit(struct amdgpu_device *adev, size_t acc_size, system_mem_needed, ttm_mem_needed, vram_needed; int ret = 0; - acc_size = ttm_bo_dma_acc_size(&adev->mman.bdev, size, - sizeof(struct amdgpu_bo)); + acc_size = amdgpu_amdkfd_acc_size(size); vram_needed = 0; if (domain == AMDGPU_GEM_DOMAIN_GTT) { @@ -175,8 +185,7 @@ static void unreserve_mem_limit(struct amdgpu_device *adev, { size_t acc_size; - acc_size = ttm_bo_dma_acc_size(&adev->mman.bdev, size, - sizeof(struct amdgpu_bo)); + acc_size = amdgpu_amdkfd_acc_size(size); spin_lock(&kfd_mem_limit.mem_limit_lock); if (domain == AMDGPU_GEM_DOMAIN_GTT) { @@ -404,7 +413,10 @@ static uint64_t get_pte_flags(struct amdgpu_device *adev, struct kgd_mem *mem) { struct amdgpu_device *bo_adev = amdgpu_ttm_adev(mem->bo->tbo.bdev); bool coherent = mem->alloc_flags & KFD_IOC_ALLOC_MEM_FLAGS_COHERENT; + bool uncached = mem->alloc_flags & KFD_IOC_ALLOC_MEM_FLAGS_UNCACHED; uint32_t mapping_flags; + uint64_t pte_flags; + bool snoop = false; mapping_flags = AMDGPU_VM_PAGE_READABLE; if (mem->alloc_flags & KFD_IOC_ALLOC_MEM_FLAGS_WRITABLE) @@ -425,12 +437,38 @@ static uint64_t get_pte_flags(struct amdgpu_device *adev, struct kgd_mem *mem) AMDGPU_VM_MTYPE_UC : AMDGPU_VM_MTYPE_NC; } break; + case CHIP_ALDEBARAN: + if (coherent && uncached) { + if (adev->gmc.xgmi.connected_to_cpu || + !(mem->alloc_flags & KFD_IOC_ALLOC_MEM_FLAGS_VRAM)) + snoop = true; + mapping_flags |= AMDGPU_VM_MTYPE_UC; + } else if (mem->alloc_flags & KFD_IOC_ALLOC_MEM_FLAGS_VRAM) { + if (bo_adev == adev) { + mapping_flags |= coherent ? + AMDGPU_VM_MTYPE_CC : AMDGPU_VM_MTYPE_RW; + if (adev->gmc.xgmi.connected_to_cpu) + snoop = true; + } else { + mapping_flags |= AMDGPU_VM_MTYPE_UC; + if (amdgpu_xgmi_same_hive(adev, bo_adev)) + snoop = true; + } + } else { + snoop = true; + mapping_flags |= coherent ? + AMDGPU_VM_MTYPE_UC : AMDGPU_VM_MTYPE_NC; + } + break; default: mapping_flags |= coherent ? AMDGPU_VM_MTYPE_UC : AMDGPU_VM_MTYPE_NC; } - return amdgpu_gem_va_map_flags(adev, mapping_flags); + pte_flags = amdgpu_gem_va_map_flags(adev, mapping_flags); + pte_flags |= snoop ? AMDGPU_PTE_SNOOPED : 0; + + return pte_flags; } /* add_bo_to_vm - Add a BO to a VM @@ -996,41 +1034,6 @@ create_evict_fence_fail: return ret; } -int amdgpu_amdkfd_gpuvm_create_process_vm(struct kgd_dev *kgd, u32 pasid, - void **vm, void **process_info, - struct dma_fence **ef) -{ - struct amdgpu_device *adev = get_amdgpu_device(kgd); - struct amdgpu_vm *new_vm; - int ret; - - new_vm = kzalloc(sizeof(*new_vm), GFP_KERNEL); - if (!new_vm) - return -ENOMEM; - - /* Initialize AMDGPU part of the VM */ - ret = amdgpu_vm_init(adev, new_vm, AMDGPU_VM_CONTEXT_COMPUTE, pasid); - if (ret) { - pr_err("Failed init vm ret %d\n", ret); - goto amdgpu_vm_init_fail; - } - - /* Initialize KFD part of the VM and process info */ - ret = init_kfd_vm(new_vm, process_info, ef); - if (ret) - goto init_kfd_vm_fail; - - *vm = (void *) new_vm; - - return 0; - -init_kfd_vm_fail: - amdgpu_vm_fini(adev, new_vm); -amdgpu_vm_init_fail: - kfree(new_vm); - return ret; -} - int amdgpu_amdkfd_gpuvm_acquire_process_vm(struct kgd_dev *kgd, struct file *filp, u32 pasid, void **vm, void **process_info, @@ -1097,21 +1100,6 @@ void amdgpu_amdkfd_gpuvm_destroy_cb(struct amdgpu_device *adev, } } -void amdgpu_amdkfd_gpuvm_destroy_process_vm(struct kgd_dev *kgd, void *vm) -{ - struct amdgpu_device *adev = get_amdgpu_device(kgd); - struct amdgpu_vm *avm = (struct amdgpu_vm *)vm; - - if (WARN_ON(!kgd || !vm)) - return; - - pr_debug("Destroying process vm %p\n", vm); - - /* Release the VM context */ - amdgpu_vm_fini(adev, avm); - kfree(vm); -} - void amdgpu_amdkfd_gpuvm_release_process_vm(struct kgd_dev *kgd, void *vm) { struct amdgpu_device *adev = get_amdgpu_device(kgd); diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_atombios.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_atombios.c index 86add0f4ea4d..494b2e1717d5 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_atombios.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_atombios.c @@ -1232,157 +1232,6 @@ int amdgpu_atombios_get_leakage_vddc_based_on_leakage_idx(struct amdgpu_device * return amdgpu_atombios_get_max_vddc(adev, VOLTAGE_TYPE_VDDC, leakage_idx, voltage); } -int amdgpu_atombios_get_leakage_id_from_vbios(struct amdgpu_device *adev, - u16 *leakage_id) -{ - union set_voltage args; - int index = GetIndexIntoMasterTable(COMMAND, SetVoltage); - u8 frev, crev; - - if (!amdgpu_atom_parse_cmd_header(adev->mode_info.atom_context, index, &frev, &crev)) - return -EINVAL; - - switch (crev) { - case 3: - case 4: - args.v3.ucVoltageType = 0; - args.v3.ucVoltageMode = ATOM_GET_LEAKAGE_ID; - args.v3.usVoltageLevel = 0; - - amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args); - - *leakage_id = le16_to_cpu(args.v3.usVoltageLevel); - break; - default: - DRM_ERROR("Unknown table version %d, %d\n", frev, crev); - return -EINVAL; - } - - return 0; -} - -int amdgpu_atombios_get_leakage_vddc_based_on_leakage_params(struct amdgpu_device *adev, - u16 *vddc, u16 *vddci, - u16 virtual_voltage_id, - u16 vbios_voltage_id) -{ - int index = GetIndexIntoMasterTable(DATA, ASIC_ProfilingInfo); - u8 frev, crev; - u16 data_offset, size; - int i, j; - ATOM_ASIC_PROFILING_INFO_V2_1 *profile; - u16 *leakage_bin, *vddc_id_buf, *vddc_buf, *vddci_id_buf, *vddci_buf; - - *vddc = 0; - *vddci = 0; - - if (!amdgpu_atom_parse_data_header(adev->mode_info.atom_context, index, &size, - &frev, &crev, &data_offset)) - return -EINVAL; - - profile = (ATOM_ASIC_PROFILING_INFO_V2_1 *) - (adev->mode_info.atom_context->bios + data_offset); - - switch (frev) { - case 1: - return -EINVAL; - case 2: - switch (crev) { - case 1: - if (size < sizeof(ATOM_ASIC_PROFILING_INFO_V2_1)) - return -EINVAL; - leakage_bin = (u16 *) - (adev->mode_info.atom_context->bios + data_offset + - le16_to_cpu(profile->usLeakageBinArrayOffset)); - vddc_id_buf = (u16 *) - (adev->mode_info.atom_context->bios + data_offset + - le16_to_cpu(profile->usElbVDDC_IdArrayOffset)); - vddc_buf = (u16 *) - (adev->mode_info.atom_context->bios + data_offset + - le16_to_cpu(profile->usElbVDDC_LevelArrayOffset)); - vddci_id_buf = (u16 *) - (adev->mode_info.atom_context->bios + data_offset + - le16_to_cpu(profile->usElbVDDCI_IdArrayOffset)); - vddci_buf = (u16 *) - (adev->mode_info.atom_context->bios + data_offset + - le16_to_cpu(profile->usElbVDDCI_LevelArrayOffset)); - - if (profile->ucElbVDDC_Num > 0) { - for (i = 0; i < profile->ucElbVDDC_Num; i++) { - if (vddc_id_buf[i] == virtual_voltage_id) { - for (j = 0; j < profile->ucLeakageBinNum; j++) { - if (vbios_voltage_id <= leakage_bin[j]) { - *vddc = vddc_buf[j * profile->ucElbVDDC_Num + i]; - break; - } - } - break; - } - } - } - if (profile->ucElbVDDCI_Num > 0) { - for (i = 0; i < profile->ucElbVDDCI_Num; i++) { - if (vddci_id_buf[i] == virtual_voltage_id) { - for (j = 0; j < profile->ucLeakageBinNum; j++) { - if (vbios_voltage_id <= leakage_bin[j]) { - *vddci = vddci_buf[j * profile->ucElbVDDCI_Num + i]; - break; - } - } - break; - } - } - } - break; - default: - DRM_ERROR("Unknown table version %d, %d\n", frev, crev); - return -EINVAL; - } - break; - default: - DRM_ERROR("Unknown table version %d, %d\n", frev, crev); - return -EINVAL; - } - - return 0; -} - -union get_voltage_info { - struct _GET_VOLTAGE_INFO_INPUT_PARAMETER_V1_2 in; - struct _GET_EVV_VOLTAGE_INFO_OUTPUT_PARAMETER_V1_2 evv_out; -}; - -int amdgpu_atombios_get_voltage_evv(struct amdgpu_device *adev, - u16 virtual_voltage_id, - u16 *voltage) -{ - int index = GetIndexIntoMasterTable(COMMAND, GetVoltageInfo); - u32 entry_id; - u32 count = adev->pm.dpm.dyn_state.vddc_dependency_on_sclk.count; - union get_voltage_info args; - - for (entry_id = 0; entry_id < count; entry_id++) { - if (adev->pm.dpm.dyn_state.vddc_dependency_on_sclk.entries[entry_id].v == - virtual_voltage_id) - break; - } - - if (entry_id >= count) - return -EINVAL; - - args.in.ucVoltageType = VOLTAGE_TYPE_VDDC; - args.in.ucVoltageMode = ATOM_GET_VOLTAGE_EVV_VOLTAGE; - args.in.usVoltageLevel = cpu_to_le16(virtual_voltage_id); - args.in.ulSCLKFreq = - cpu_to_le32(adev->pm.dpm.dyn_state.vddc_dependency_on_sclk.entries[entry_id].clk); - - amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args); - - *voltage = le16_to_cpu(args.evv_out.usVoltageLevel); - - return 0; -} - union voltage_object_info { struct _ATOM_VOLTAGE_OBJECT_INFO v1; struct _ATOM_VOLTAGE_OBJECT_INFO_V2 v2; @@ -1905,40 +1754,6 @@ static uint32_t cail_reg_read(struct card_info *info, uint32_t reg) return r; } -/** - * cail_ioreg_write - write IO register - * - * @info: atom card_info pointer - * @reg: IO register offset - * @val: value to write to the pll register - * - * Provides a IO register accessor for the atom interpreter (r4xx+). - */ -static void cail_ioreg_write(struct card_info *info, uint32_t reg, uint32_t val) -{ - struct amdgpu_device *adev = drm_to_adev(info->dev); - - WREG32_IO(reg, val); -} - -/** - * cail_ioreg_read - read IO register - * - * @info: atom card_info pointer - * @reg: IO register offset - * - * Provides an IO register accessor for the atom interpreter (r4xx+). - * Returns the value of the IO register. - */ -static uint32_t cail_ioreg_read(struct card_info *info, uint32_t reg) -{ - struct amdgpu_device *adev = drm_to_adev(info->dev); - uint32_t r; - - r = RREG32_IO(reg); - return r; -} - static ssize_t amdgpu_atombios_get_vbios_version(struct device *dev, struct device_attribute *attr, char *buf) @@ -1947,7 +1762,7 @@ static ssize_t amdgpu_atombios_get_vbios_version(struct device *dev, struct amdgpu_device *adev = drm_to_adev(ddev); struct atom_context *ctx = adev->mode_info.atom_context; - return snprintf(buf, PAGE_SIZE, "%s\n", ctx->vbios_version); + return sysfs_emit(buf, "%s\n", ctx->vbios_version); } static DEVICE_ATTR(vbios_version, 0444, amdgpu_atombios_get_vbios_version, @@ -1998,15 +1813,6 @@ int amdgpu_atombios_init(struct amdgpu_device *adev) atom_card_info->dev = adev_to_drm(adev); atom_card_info->reg_read = cail_reg_read; atom_card_info->reg_write = cail_reg_write; - /* needed for iio ops */ - if (adev->rio_mem) { - atom_card_info->ioreg_read = cail_ioreg_read; - atom_card_info->ioreg_write = cail_ioreg_write; - } else { - DRM_DEBUG("PCI I/O BAR is not found. Using MMIO to access ATOM BIOS\n"); - atom_card_info->ioreg_read = cail_reg_read; - atom_card_info->ioreg_write = cail_reg_write; - } atom_card_info->mc_read = cail_mc_read; atom_card_info->mc_write = cail_mc_write; atom_card_info->pll_read = cail_pll_read; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_atombios.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_atombios.h index 1321ec09c734..8cc0222dba19 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_atombios.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_atombios.h @@ -168,18 +168,6 @@ int amdgpu_atombios_get_memory_pll_dividers(struct amdgpu_device *adev, void amdgpu_atombios_set_engine_dram_timings(struct amdgpu_device *adev, u32 eng_clock, u32 mem_clock); -int amdgpu_atombios_get_leakage_id_from_vbios(struct amdgpu_device *adev, - u16 *leakage_id); - -int amdgpu_atombios_get_leakage_vddc_based_on_leakage_params(struct amdgpu_device *adev, - u16 *vddc, u16 *vddci, - u16 virtual_voltage_id, - u16 vbios_voltage_id); - -int amdgpu_atombios_get_voltage_evv(struct amdgpu_device *adev, - u16 virtual_voltage_id, - u16 *voltage); - bool amdgpu_atombios_is_voltage_gpio(struct amdgpu_device *adev, u8 voltage_type, u8 voltage_mode); diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_atomfirmware.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_atomfirmware.c index 6107ac91db25..60716b35444b 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_atomfirmware.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_atomfirmware.c @@ -117,12 +117,15 @@ union igp_info { union umc_info { struct atom_umc_info_v3_1 v31; + struct atom_umc_info_v3_2 v32; + struct atom_umc_info_v3_3 v33; }; union vram_info { struct atom_vram_info_header_v2_3 v23; struct atom_vram_info_header_v2_4 v24; struct atom_vram_info_header_v2_5 v25; + struct atom_vram_info_header_v2_6 v26; }; union vram_module { @@ -164,6 +167,7 @@ static int convert_atom_mem_type_to_vram_type(struct amdgpu_device *adev, vram_type = AMDGPU_VRAM_TYPE_GDDR5; break; case ATOM_DGPU_VRAM_TYPE_HBM2: + case ATOM_DGPU_VRAM_TYPE_HBM2E: vram_type = AMDGPU_VRAM_TYPE_HBM; break; case ATOM_DGPU_VRAM_TYPE_GDDR6: @@ -315,6 +319,26 @@ amdgpu_atomfirmware_get_vram_info(struct amdgpu_device *adev, if (vram_vendor) *vram_vendor = mem_vendor; break; + case 6: + if (module_id > vram_info->v26.vram_module_num) + module_id = 0; + vram_module = (union vram_module *)vram_info->v26.vram_module; + while (i < module_id) { + vram_module = (union vram_module *) + ((u8 *)vram_module + vram_module->v9.vram_module_size); + i++; + } + mem_type = vram_module->v9.memory_type; + if (vram_type) + *vram_type = convert_atom_mem_type_to_vram_type(adev, mem_type); + mem_channel_number = vram_module->v9.channel_num; + mem_channel_width = vram_module->v9.channel_width; + if (vram_width) + *vram_width = mem_channel_number * (1 << mem_channel_width); + mem_vendor = (vram_module->v9.vender_rev_id) & 0xF; + if (vram_vendor) + *vram_vendor = mem_vendor; + break; default: return -EINVAL; } @@ -337,19 +361,39 @@ bool amdgpu_atomfirmware_mem_ecc_supported(struct amdgpu_device *adev) union umc_info *umc_info; u8 frev, crev; bool ecc_default_enabled = false; + u8 umc_config; + u32 umc_config1; index = get_index_into_master_table(atom_master_list_of_data_tables_v2_1, umc_info); if (amdgpu_atom_parse_data_header(mode_info->atom_context, index, &size, &frev, &crev, &data_offset)) { - /* support umc_info 3.1+ */ - if ((frev == 3 && crev >= 1) || (frev > 3)) { + if (frev == 3) { umc_info = (union umc_info *) (mode_info->atom_context->bios + data_offset); - ecc_default_enabled = - (le32_to_cpu(umc_info->v31.umc_config) & - UMC_CONFIG__DEFAULT_MEM_ECC_ENABLE) ? true : false; + switch (crev) { + case 1: + umc_config = le32_to_cpu(umc_info->v31.umc_config); + ecc_default_enabled = + (umc_config & UMC_CONFIG__DEFAULT_MEM_ECC_ENABLE) ? true : false; + break; + case 2: + umc_config = le32_to_cpu(umc_info->v32.umc_config); + ecc_default_enabled = + (umc_config & UMC_CONFIG__DEFAULT_MEM_ECC_ENABLE) ? true : false; + break; + case 3: + umc_config = le32_to_cpu(umc_info->v33.umc_config); + umc_config1 = le32_to_cpu(umc_info->v33.umc_config1); + ecc_default_enabled = + ((umc_config & UMC_CONFIG__DEFAULT_MEM_ECC_ENABLE) || + (umc_config1 & UMC_CONFIG1__ENABLE_ECC_CAPABLE)) ? true : false; + break; + default: + /* unsupported crev */ + return false; + } } } @@ -479,7 +523,8 @@ int amdgpu_atomfirmware_get_clock_info(struct amdgpu_device *adev) } union gfx_info { - struct atom_gfx_info_v2_4 v24; + struct atom_gfx_info_v2_4 v24; + struct atom_gfx_info_v2_7 v27; }; int amdgpu_atomfirmware_get_gfx_info(struct amdgpu_device *adev) @@ -514,6 +559,22 @@ int amdgpu_atomfirmware_get_gfx_info(struct amdgpu_device *adev) adev->gfx.cu_info.max_scratch_slots_per_cu = gfx_info->v24.gc_max_scratch_slots_per_cu; adev->gfx.cu_info.lds_size = le16_to_cpu(gfx_info->v24.gc_lds_size); return 0; + case 7: + adev->gfx.config.max_shader_engines = gfx_info->v27.max_shader_engines; + adev->gfx.config.max_cu_per_sh = gfx_info->v27.max_cu_per_sh; + adev->gfx.config.max_sh_per_se = gfx_info->v27.max_sh_per_se; + adev->gfx.config.max_backends_per_se = gfx_info->v27.max_backends_per_se; + adev->gfx.config.max_texture_channel_caches = gfx_info->v27.max_texture_channel_caches; + adev->gfx.config.max_gprs = le16_to_cpu(gfx_info->v27.gc_num_gprs); + adev->gfx.config.max_gs_threads = gfx_info->v27.gc_num_max_gs_thds; + adev->gfx.config.gs_vgt_table_depth = gfx_info->v27.gc_gs_table_depth; + adev->gfx.config.gs_prim_buffer_depth = le16_to_cpu(gfx_info->v27.gc_gsprim_buff_depth); + adev->gfx.config.double_offchip_lds_buf = gfx_info->v27.gc_double_offchip_lds_buffer; + adev->gfx.cu_info.wave_front_size = le16_to_cpu(gfx_info->v27.gc_wave_size); + adev->gfx.cu_info.max_waves_per_simd = le16_to_cpu(gfx_info->v27.gc_max_waves_per_simd); + adev->gfx.cu_info.max_scratch_slots_per_cu = gfx_info->v27.gc_max_scratch_slots_per_cu; + adev->gfx.cu_info.lds_size = le16_to_cpu(gfx_info->v27.gc_lds_size); + return 0; default: return -EINVAL; } diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_benchmark.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_benchmark.c index d9b35df33806..313517f7cf10 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_benchmark.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_benchmark.c @@ -85,6 +85,8 @@ static void amdgpu_benchmark_move(struct amdgpu_device *adev, unsigned size, bp.flags = 0; bp.type = ttm_bo_type_kernel; bp.resv = NULL; + bp.bo_ptr_size = sizeof(struct amdgpu_bo); + n = AMDGPU_BENCHMARK_ITERATIONS; r = amdgpu_bo_create(adev, &bp, &sobj); if (r) { diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_bios.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_bios.c index cfb1a9a04477..27b19503773b 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_bios.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_bios.c @@ -97,6 +97,10 @@ static bool igp_read_bios_from_vram(struct amdgpu_device *adev) if (amdgpu_device_need_post(adev)) return false; + /* FB BAR not enabled */ + if (pci_resource_len(adev->pdev, 0) == 0) + return false; + adev->bios = NULL; vram_base = pci_resource_start(adev->pdev, 0); bios = ioremap_wc(vram_base, size); @@ -316,7 +320,7 @@ static bool amdgpu_atrm_get_bios(struct amdgpu_device *adev) adev->bios = kmalloc(size, GFP_KERNEL); if (!adev->bios) { - DRM_ERROR("Unable to allocate bios\n"); + dev_err(adev->dev, "Unable to allocate bios\n"); return false; } @@ -364,7 +368,7 @@ static bool amdgpu_acpi_vfct_bios(struct amdgpu_device *adev) return false; tbl_size = hdr->length; if (tbl_size < sizeof(UEFI_ACPI_VFCT)) { - DRM_ERROR("ACPI VFCT table present but broken (too short #1)\n"); + dev_info(adev->dev, "ACPI VFCT table present but broken (too short #1),skipping\n"); return false; } @@ -377,13 +381,13 @@ static bool amdgpu_acpi_vfct_bios(struct amdgpu_device *adev) offset += sizeof(VFCT_IMAGE_HEADER); if (offset > tbl_size) { - DRM_ERROR("ACPI VFCT image header truncated\n"); + dev_info(adev->dev, "ACPI VFCT image header truncated,skipping\n"); return false; } offset += vhdr->ImageLength; if (offset > tbl_size) { - DRM_ERROR("ACPI VFCT image truncated\n"); + dev_info(adev->dev, "ACPI VFCT image truncated,skipping\n"); return false; } @@ -406,7 +410,7 @@ static bool amdgpu_acpi_vfct_bios(struct amdgpu_device *adev) } } - DRM_ERROR("ACPI VFCT table present but broken (too short #2)\n"); + dev_info(adev->dev, "ACPI VFCT table present but broken (too short #2),skipping\n"); return false; } #else @@ -453,7 +457,7 @@ bool amdgpu_get_bios(struct amdgpu_device *adev) goto success; } - DRM_ERROR("Unable to locate a BIOS ROM\n"); + dev_err(adev->dev, "Unable to locate a BIOS ROM\n"); return false; success: diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c index 3e240b952e79..b5c766998045 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c @@ -117,7 +117,7 @@ static int amdgpu_cs_parser_init(struct amdgpu_cs_parser *p, union drm_amdgpu_cs if (cs->in.num_chunks == 0) return 0; - chunk_array = kmalloc_array(cs->in.num_chunks, sizeof(uint64_t), GFP_KERNEL); + chunk_array = kvmalloc_array(cs->in.num_chunks, sizeof(uint64_t), GFP_KERNEL); if (!chunk_array) return -ENOMEM; @@ -144,7 +144,7 @@ static int amdgpu_cs_parser_init(struct amdgpu_cs_parser *p, union drm_amdgpu_cs } p->nchunks = cs->in.num_chunks; - p->chunks = kmalloc_array(p->nchunks, sizeof(struct amdgpu_cs_chunk), + p->chunks = kvmalloc_array(p->nchunks, sizeof(struct amdgpu_cs_chunk), GFP_KERNEL); if (!p->chunks) { ret = -ENOMEM; @@ -238,7 +238,7 @@ static int amdgpu_cs_parser_init(struct amdgpu_cs_parser *p, union drm_amdgpu_cs if (p->uf_entry.tv.bo) p->job->uf_addr = uf_offset; - kfree(chunk_array); + kvfree(chunk_array); /* Use this opportunity to fill in task info for the vm */ amdgpu_vm_set_task_info(vm); @@ -250,11 +250,11 @@ free_all_kdata: free_partial_kdata: for (; i >= 0; i--) kvfree(p->chunks[i].kdata); - kfree(p->chunks); + kvfree(p->chunks); p->chunks = NULL; p->nchunks = 0; free_chunk: - kfree(chunk_array); + kvfree(chunk_array); return ret; } @@ -559,7 +559,7 @@ static int amdgpu_cs_parser_bos(struct amdgpu_cs_parser *p, sizeof(struct page *), GFP_KERNEL | __GFP_ZERO); if (!e->user_pages) { - DRM_ERROR("calloc failure\n"); + DRM_ERROR("kvmalloc_array failure\n"); return -ENOMEM; } @@ -706,7 +706,7 @@ static void amdgpu_cs_parser_fini(struct amdgpu_cs_parser *parser, int error, for (i = 0; i < parser->nchunks; i++) kvfree(parser->chunks[i].kdata); - kfree(parser->chunks); + kvfree(parser->chunks); if (parser->job) amdgpu_job_free(parser->job); if (parser->uf_entry.tv.bo) { diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_debugfs.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_debugfs.c index 43059ead733b..bcaf271b39bf 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_debugfs.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_debugfs.c @@ -28,7 +28,6 @@ #include <linux/uaccess.h> #include <linux/pm_runtime.h> #include <linux/poll.h> -#include <drm/drm_debugfs.h> #include "amdgpu.h" #include "amdgpu_pm.h" @@ -38,45 +37,6 @@ #include "amdgpu_securedisplay.h" #include "amdgpu_fw_attestation.h" -/** - * amdgpu_debugfs_add_files - Add simple debugfs entries - * - * @adev: Device to attach debugfs entries to - * @files: Array of function callbacks that respond to reads - * @nfiles: Number of callbacks to register - * - */ -int amdgpu_debugfs_add_files(struct amdgpu_device *adev, - const struct drm_info_list *files, - unsigned nfiles) -{ - unsigned i; - - for (i = 0; i < adev->debugfs_count; i++) { - if (adev->debugfs[i].files == files) { - /* Already registered */ - return 0; - } - } - - i = adev->debugfs_count + 1; - if (i > AMDGPU_DEBUGFS_MAX_COMPONENTS) { - DRM_ERROR("Reached maximum number of debugfs components.\n"); - DRM_ERROR("Report so we increase " - "AMDGPU_DEBUGFS_MAX_COMPONENTS.\n"); - return -EINVAL; - } - adev->debugfs[adev->debugfs_count].files = files; - adev->debugfs[adev->debugfs_count].num_files = nfiles; - adev->debugfs_count = i; -#if defined(CONFIG_DEBUG_FS) - drm_debugfs_create_files(files, nfiles, - adev_to_drm(adev)->primary->debugfs_root, - adev_to_drm(adev)->primary); -#endif - return 0; -} - int amdgpu_debugfs_wait_dump(struct amdgpu_device *adev) { #if defined(CONFIG_DEBUG_FS) @@ -1228,22 +1188,20 @@ int amdgpu_debugfs_regs_init(struct amdgpu_device *adev) adev, debugfs_regs[i]); if (!i && !IS_ERR_OR_NULL(ent)) i_size_write(ent->d_inode, adev->rmmio_size); - adev->debugfs_regs[i] = ent; } return 0; } -static int amdgpu_debugfs_test_ib(struct seq_file *m, void *data) +static int amdgpu_debugfs_test_ib_show(struct seq_file *m, void *unused) { - struct drm_info_node *node = (struct drm_info_node *) m->private; - struct drm_device *dev = node->minor->dev; - struct amdgpu_device *adev = drm_to_adev(dev); + struct amdgpu_device *adev = (struct amdgpu_device *)m->private; + struct drm_device *dev = adev_to_drm(adev); int r = 0, i; r = pm_runtime_get_sync(dev->dev); if (r < 0) { - pm_runtime_put_autosuspend(adev_to_drm(adev)->dev); + pm_runtime_put_autosuspend(dev->dev); return r; } @@ -1285,30 +1243,19 @@ static int amdgpu_debugfs_test_ib(struct seq_file *m, void *data) return 0; } -static int amdgpu_debugfs_get_vbios_dump(struct seq_file *m, void *data) +static int amdgpu_debugfs_evict_vram(void *data, u64 *val) { - struct drm_info_node *node = (struct drm_info_node *) m->private; - struct drm_device *dev = node->minor->dev; - struct amdgpu_device *adev = drm_to_adev(dev); - - seq_write(m, adev->bios, adev->bios_size); - return 0; -} - -static int amdgpu_debugfs_evict_vram(struct seq_file *m, void *data) -{ - struct drm_info_node *node = (struct drm_info_node *)m->private; - struct drm_device *dev = node->minor->dev; - struct amdgpu_device *adev = drm_to_adev(dev); + struct amdgpu_device *adev = (struct amdgpu_device *)data; + struct drm_device *dev = adev_to_drm(adev); int r; r = pm_runtime_get_sync(dev->dev); if (r < 0) { - pm_runtime_put_autosuspend(adev_to_drm(adev)->dev); + pm_runtime_put_autosuspend(dev->dev); return r; } - seq_printf(m, "(%d)\n", amdgpu_bo_evict_vram(adev)); + *val = amdgpu_bo_evict_vram(adev); pm_runtime_mark_last_busy(dev->dev); pm_runtime_put_autosuspend(dev->dev); @@ -1316,11 +1263,11 @@ static int amdgpu_debugfs_evict_vram(struct seq_file *m, void *data) return 0; } -static int amdgpu_debugfs_evict_gtt(struct seq_file *m, void *data) + +static int amdgpu_debugfs_evict_gtt(void *data, u64 *val) { - struct drm_info_node *node = (struct drm_info_node *)m->private; - struct drm_device *dev = node->minor->dev; - struct amdgpu_device *adev = drm_to_adev(dev); + struct amdgpu_device *adev = (struct amdgpu_device *)data; + struct drm_device *dev = adev_to_drm(adev); struct ttm_resource_manager *man; int r; @@ -1331,8 +1278,7 @@ static int amdgpu_debugfs_evict_gtt(struct seq_file *m, void *data) } man = ttm_manager_type(&adev->mman.bdev, TTM_PL_TT); - r = ttm_resource_manager_evict_all(&adev->mman.bdev, man); - seq_printf(m, "(%d)\n", r); + *val = ttm_resource_manager_evict_all(&adev->mman.bdev, man); pm_runtime_mark_last_busy(dev->dev); pm_runtime_put_autosuspend(dev->dev); @@ -1340,10 +1286,11 @@ static int amdgpu_debugfs_evict_gtt(struct seq_file *m, void *data) return 0; } -static int amdgpu_debugfs_vm_info(struct seq_file *m, void *data) + +static int amdgpu_debugfs_vm_info_show(struct seq_file *m, void *unused) { - struct drm_info_node *node = (struct drm_info_node *)m->private; - struct drm_device *dev = node->minor->dev; + struct amdgpu_device *adev = (struct amdgpu_device *)m->private; + struct drm_device *dev = adev_to_drm(adev); struct drm_file *file; int r; @@ -1369,13 +1316,12 @@ static int amdgpu_debugfs_vm_info(struct seq_file *m, void *data) return r; } -static const struct drm_info_list amdgpu_debugfs_list[] = { - {"amdgpu_vbios", amdgpu_debugfs_get_vbios_dump}, - {"amdgpu_test_ib", &amdgpu_debugfs_test_ib}, - {"amdgpu_evict_vram", &amdgpu_debugfs_evict_vram}, - {"amdgpu_evict_gtt", &amdgpu_debugfs_evict_gtt}, - {"amdgpu_vm_info", &amdgpu_debugfs_vm_info}, -}; +DEFINE_SHOW_ATTRIBUTE(amdgpu_debugfs_test_ib); +DEFINE_SHOW_ATTRIBUTE(amdgpu_debugfs_vm_info); +DEFINE_DEBUGFS_ATTRIBUTE(amdgpu_evict_vram_fops, amdgpu_debugfs_evict_vram, + NULL, "%lld\n"); +DEFINE_DEBUGFS_ATTRIBUTE(amdgpu_evict_gtt_fops, amdgpu_debugfs_evict_gtt, + NULL, "%lld\n"); static void amdgpu_ib_preempt_fences_swap(struct amdgpu_ring *ring, struct dma_fence **fences) @@ -1586,71 +1532,50 @@ static int amdgpu_debugfs_sclk_set(void *data, u64 val) return 0; } -DEFINE_SIMPLE_ATTRIBUTE(fops_ib_preempt, NULL, +DEFINE_DEBUGFS_ATTRIBUTE(fops_ib_preempt, NULL, amdgpu_debugfs_ib_preempt, "%llu\n"); -DEFINE_SIMPLE_ATTRIBUTE(fops_sclk_set, NULL, +DEFINE_DEBUGFS_ATTRIBUTE(fops_sclk_set, NULL, amdgpu_debugfs_sclk_set, "%llu\n"); int amdgpu_debugfs_init(struct amdgpu_device *adev) { + struct dentry *root = adev_to_drm(adev)->primary->debugfs_root; + struct dentry *ent; int r, i; - adev->debugfs_preempt = - debugfs_create_file("amdgpu_preempt_ib", 0600, - adev_to_drm(adev)->primary->debugfs_root, adev, - &fops_ib_preempt); - if (!(adev->debugfs_preempt)) { + + + ent = debugfs_create_file("amdgpu_preempt_ib", 0600, root, adev, + &fops_ib_preempt); + if (!ent) { DRM_ERROR("unable to create amdgpu_preempt_ib debugsfs file\n"); return -EIO; } - adev->smu.debugfs_sclk = - debugfs_create_file("amdgpu_force_sclk", 0200, - adev_to_drm(adev)->primary->debugfs_root, adev, - &fops_sclk_set); - if (!(adev->smu.debugfs_sclk)) { + ent = debugfs_create_file("amdgpu_force_sclk", 0200, root, adev, + &fops_sclk_set); + if (!ent) { DRM_ERROR("unable to create amdgpu_set_sclk debugsfs file\n"); return -EIO; } /* Register debugfs entries for amdgpu_ttm */ - r = amdgpu_ttm_debugfs_init(adev); - if (r) { - DRM_ERROR("Failed to init debugfs\n"); - return r; - } - - r = amdgpu_debugfs_pm_init(adev); - if (r) { - DRM_ERROR("Failed to register debugfs file for dpm!\n"); - return r; - } - - if (amdgpu_debugfs_sa_init(adev)) { - dev_err(adev->dev, "failed to register debugfs file for SA\n"); - } - - if (amdgpu_debugfs_fence_init(adev)) - dev_err(adev->dev, "fence debugfs file creation failed\n"); - - r = amdgpu_debugfs_gem_init(adev); - if (r) - DRM_ERROR("registering gem debugfs failed (%d).\n", r); + amdgpu_ttm_debugfs_init(adev); + amdgpu_debugfs_pm_init(adev); + amdgpu_debugfs_sa_init(adev); + amdgpu_debugfs_fence_init(adev); + amdgpu_debugfs_gem_init(adev); r = amdgpu_debugfs_regs_init(adev); if (r) DRM_ERROR("registering register debugfs failed (%d).\n", r); - r = amdgpu_debugfs_firmware_init(adev); - if (r) - DRM_ERROR("registering firmware debugfs failed (%d).\n", r); + amdgpu_debugfs_firmware_init(adev); #if defined(CONFIG_DRM_AMD_DC) - if (amdgpu_device_has_dc_support(adev)) { - if (dtn_debugfs_init(adev)) - DRM_ERROR("amdgpu: failed initialize dtn debugfs support.\n"); - } + if (amdgpu_device_has_dc_support(adev)) + dtn_debugfs_init(adev); #endif for (i = 0; i < AMDGPU_MAX_RINGS; ++i) { @@ -1665,17 +1590,26 @@ int amdgpu_debugfs_init(struct amdgpu_device *adev) } amdgpu_ras_debugfs_create_all(adev); - amdgpu_debugfs_autodump_init(adev); - amdgpu_rap_debugfs_init(adev); - amdgpu_securedisplay_debugfs_init(adev); - amdgpu_fw_attestation_debugfs_init(adev); - return amdgpu_debugfs_add_files(adev, amdgpu_debugfs_list, - ARRAY_SIZE(amdgpu_debugfs_list)); + debugfs_create_file("amdgpu_evict_vram", 0444, root, adev, + &amdgpu_evict_vram_fops); + debugfs_create_file("amdgpu_evict_gtt", 0444, root, adev, + &amdgpu_evict_gtt_fops); + debugfs_create_file("amdgpu_test_ib", 0444, root, adev, + &amdgpu_debugfs_test_ib_fops); + debugfs_create_file("amdgpu_vm_info", 0444, root, adev, + &amdgpu_debugfs_vm_info_fops); + + adev->debugfs_vbios_blob.data = adev->bios; + adev->debugfs_vbios_blob.size = adev->bios_size; + debugfs_create_blob("amdgpu_vbios", 0444, root, + &adev->debugfs_vbios_blob); + + return 0; } #else diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_debugfs.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_debugfs.h index 2803884d338d..141a8474e24f 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_debugfs.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_debugfs.h @@ -26,11 +26,6 @@ /* * Debugfs */ -struct amdgpu_debugfs { - const struct drm_info_list *files; - unsigned num_files; -}; - struct amdgpu_autodump { struct completion dumping; struct wait_queue_head gpu_hang; @@ -39,10 +34,7 @@ struct amdgpu_autodump { int amdgpu_debugfs_regs_init(struct amdgpu_device *adev); int amdgpu_debugfs_init(struct amdgpu_device *adev); void amdgpu_debugfs_fini(struct amdgpu_device *adev); -int amdgpu_debugfs_add_files(struct amdgpu_device *adev, - const struct drm_info_list *files, - unsigned nfiles); -int amdgpu_debugfs_fence_init(struct amdgpu_device *adev); -int amdgpu_debugfs_firmware_init(struct amdgpu_device *adev); -int amdgpu_debugfs_gem_init(struct amdgpu_device *adev); +void amdgpu_debugfs_fence_init(struct amdgpu_device *adev); +void amdgpu_debugfs_firmware_init(struct amdgpu_device *adev); +void amdgpu_debugfs_gem_init(struct amdgpu_device *adev); int amdgpu_debugfs_wait_dump(struct amdgpu_device *adev); diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c index 8a5a8ff5d362..7d3b54615147 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c @@ -65,6 +65,7 @@ #include "amdgpu_ras.h" #include "amdgpu_pmu.h" #include "amdgpu_fru_eeprom.h" +#include "amdgpu_reset.h" #include <linux/suspend.h> #include <drm/task_barrier.h> @@ -110,6 +111,7 @@ const char *amdgpu_asic_name[] = { "RAVEN", "ARCTURUS", "RENOIR", + "ALDEBARAN", "NAVI10", "NAVI14", "NAVI12", @@ -136,7 +138,7 @@ static ssize_t amdgpu_device_get_pcie_replay_count(struct device *dev, struct amdgpu_device *adev = drm_to_adev(ddev); uint64_t cnt = amdgpu_asic_get_pcie_replay_count(adev); - return snprintf(buf, PAGE_SIZE, "%llu\n", cnt); + return sysfs_emit(buf, "%llu\n", cnt); } static DEVICE_ATTR(pcie_replay_count, S_IRUGO, @@ -160,7 +162,7 @@ static ssize_t amdgpu_device_get_product_name(struct device *dev, struct drm_device *ddev = dev_get_drvdata(dev); struct amdgpu_device *adev = drm_to_adev(ddev); - return snprintf(buf, PAGE_SIZE, "%s\n", adev->product_name); + return sysfs_emit(buf, "%s\n", adev->product_name); } static DEVICE_ATTR(product_name, S_IRUGO, @@ -182,7 +184,7 @@ static ssize_t amdgpu_device_get_product_number(struct device *dev, struct drm_device *ddev = dev_get_drvdata(dev); struct amdgpu_device *adev = drm_to_adev(ddev); - return snprintf(buf, PAGE_SIZE, "%s\n", adev->product_number); + return sysfs_emit(buf, "%s\n", adev->product_number); } static DEVICE_ATTR(product_number, S_IRUGO, @@ -204,25 +206,25 @@ static ssize_t amdgpu_device_get_serial_number(struct device *dev, struct drm_device *ddev = dev_get_drvdata(dev); struct amdgpu_device *adev = drm_to_adev(ddev); - return snprintf(buf, PAGE_SIZE, "%s\n", adev->serial); + return sysfs_emit(buf, "%s\n", adev->serial); } static DEVICE_ATTR(serial_number, S_IRUGO, amdgpu_device_get_serial_number, NULL); /** - * amdgpu_device_supports_atpx - Is the device a dGPU with HG/PX power control + * amdgpu_device_supports_px - Is the device a dGPU with ATPX power control * * @dev: drm_device pointer * - * Returns true if the device is a dGPU with HG/PX power control, + * Returns true if the device is a dGPU with ATPX power control, * otherwise return false. */ -bool amdgpu_device_supports_atpx(struct drm_device *dev) +bool amdgpu_device_supports_px(struct drm_device *dev) { struct amdgpu_device *adev = drm_to_adev(dev); - if (adev->flags & AMD_IS_PX) + if ((adev->flags & AMD_IS_PX) && !amdgpu_is_atpx_hybrid()) return true; return false; } @@ -232,14 +234,15 @@ bool amdgpu_device_supports_atpx(struct drm_device *dev) * * @dev: drm_device pointer * - * Returns true if the device is a dGPU with HG/PX power control, + * Returns true if the device is a dGPU with ACPI power control, * otherwise return false. */ bool amdgpu_device_supports_boco(struct drm_device *dev) { struct amdgpu_device *adev = drm_to_adev(dev); - if (adev->has_pr3) + if (adev->has_pr3 || + ((adev->flags & AMD_IS_PX) && amdgpu_is_atpx_hybrid())) return true; return false; } @@ -325,6 +328,35 @@ void amdgpu_device_vram_access(struct amdgpu_device *adev, loff_t pos, /* * register access helper functions. */ + +/* Check if hw access should be skipped because of hotplug or device error */ +bool amdgpu_device_skip_hw_access(struct amdgpu_device *adev) +{ + if (adev->in_pci_err_recovery) + return true; + +#ifdef CONFIG_LOCKDEP + /* + * This is a bit complicated to understand, so worth a comment. What we assert + * here is that the GPU reset is not running on another thread in parallel. + * + * For this we trylock the read side of the reset semaphore, if that succeeds + * we know that the reset is not running in paralell. + * + * If the trylock fails we assert that we are either already holding the read + * side of the lock or are the reset thread itself and hold the write side of + * the lock. + */ + if (in_task()) { + if (down_read_trylock(&adev->reset_sem)) + up_read(&adev->reset_sem); + else + lockdep_assert_held(&adev->reset_sem); + } +#endif + return false; +} + /** * amdgpu_device_rreg - read a memory mapped IO or indirect register * @@ -339,7 +371,7 @@ uint32_t amdgpu_device_rreg(struct amdgpu_device *adev, { uint32_t ret; - if (adev->in_pci_err_recovery) + if (amdgpu_device_skip_hw_access(adev)) return 0; if ((reg * 4) < adev->rmmio_size) { @@ -376,7 +408,7 @@ uint32_t amdgpu_device_rreg(struct amdgpu_device *adev, */ uint8_t amdgpu_mm_rreg8(struct amdgpu_device *adev, uint32_t offset) { - if (adev->in_pci_err_recovery) + if (amdgpu_device_skip_hw_access(adev)) return 0; if (offset < adev->rmmio_size) @@ -401,7 +433,7 @@ uint8_t amdgpu_mm_rreg8(struct amdgpu_device *adev, uint32_t offset) */ void amdgpu_mm_wreg8(struct amdgpu_device *adev, uint32_t offset, uint8_t value) { - if (adev->in_pci_err_recovery) + if (amdgpu_device_skip_hw_access(adev)) return; if (offset < adev->rmmio_size) @@ -424,7 +456,7 @@ void amdgpu_device_wreg(struct amdgpu_device *adev, uint32_t reg, uint32_t v, uint32_t acc_flags) { - if (adev->in_pci_err_recovery) + if (amdgpu_device_skip_hw_access(adev)) return; if ((reg * 4) < adev->rmmio_size) { @@ -451,63 +483,20 @@ void amdgpu_device_wreg(struct amdgpu_device *adev, void amdgpu_mm_wreg_mmio_rlc(struct amdgpu_device *adev, uint32_t reg, uint32_t v) { - if (adev->in_pci_err_recovery) + if (amdgpu_device_skip_hw_access(adev)) return; if (amdgpu_sriov_fullaccess(adev) && adev->gfx.rlc.funcs && adev->gfx.rlc.funcs->is_rlcg_access_range) { if (adev->gfx.rlc.funcs->is_rlcg_access_range(adev, reg)) - return adev->gfx.rlc.funcs->rlcg_wreg(adev, reg, v); + return adev->gfx.rlc.funcs->rlcg_wreg(adev, reg, v, 0); } else { writel(v, ((void __iomem *)adev->rmmio) + (reg * 4)); } } /** - * amdgpu_io_rreg - read an IO register - * - * @adev: amdgpu_device pointer - * @reg: dword aligned register offset - * - * Returns the 32 bit value from the offset specified. - */ -u32 amdgpu_io_rreg(struct amdgpu_device *adev, u32 reg) -{ - if (adev->in_pci_err_recovery) - return 0; - - if ((reg * 4) < adev->rio_mem_size) - return ioread32(adev->rio_mem + (reg * 4)); - else { - iowrite32((reg * 4), adev->rio_mem + (mmMM_INDEX * 4)); - return ioread32(adev->rio_mem + (mmMM_DATA * 4)); - } -} - -/** - * amdgpu_io_wreg - write to an IO register - * - * @adev: amdgpu_device pointer - * @reg: dword aligned register offset - * @v: 32 bit value to write to the register - * - * Writes the value specified to the offset specified. - */ -void amdgpu_io_wreg(struct amdgpu_device *adev, u32 reg, u32 v) -{ - if (adev->in_pci_err_recovery) - return; - - if ((reg * 4) < adev->rio_mem_size) - iowrite32(v, adev->rio_mem + (reg * 4)); - else { - iowrite32((reg * 4), adev->rio_mem + (mmMM_INDEX * 4)); - iowrite32(v, adev->rio_mem + (mmMM_DATA * 4)); - } -} - -/** * amdgpu_mm_rdoorbell - read a doorbell dword * * @adev: amdgpu_device pointer @@ -518,7 +507,7 @@ void amdgpu_io_wreg(struct amdgpu_device *adev, u32 reg, u32 v) */ u32 amdgpu_mm_rdoorbell(struct amdgpu_device *adev, u32 index) { - if (adev->in_pci_err_recovery) + if (amdgpu_device_skip_hw_access(adev)) return 0; if (index < adev->doorbell.num_doorbells) { @@ -541,7 +530,7 @@ u32 amdgpu_mm_rdoorbell(struct amdgpu_device *adev, u32 index) */ void amdgpu_mm_wdoorbell(struct amdgpu_device *adev, u32 index, u32 v) { - if (adev->in_pci_err_recovery) + if (amdgpu_device_skip_hw_access(adev)) return; if (index < adev->doorbell.num_doorbells) { @@ -562,7 +551,7 @@ void amdgpu_mm_wdoorbell(struct amdgpu_device *adev, u32 index, u32 v) */ u64 amdgpu_mm_rdoorbell64(struct amdgpu_device *adev, u32 index) { - if (adev->in_pci_err_recovery) + if (amdgpu_device_skip_hw_access(adev)) return 0; if (index < adev->doorbell.num_doorbells) { @@ -585,7 +574,7 @@ u64 amdgpu_mm_rdoorbell64(struct amdgpu_device *adev, u32 index) */ void amdgpu_mm_wdoorbell64(struct amdgpu_device *adev, u32 index, u64 v) { - if (adev->in_pci_err_recovery) + if (amdgpu_device_skip_hw_access(adev)) return; if (index < adev->doorbell.num_doorbells) { @@ -1223,6 +1212,10 @@ bool amdgpu_device_need_post(struct amdgpu_device *adev) } } + /* Don't post if we need to reset whole hive on init */ + if (adev->gmc.xgmi.pending_reset) + return false; + if (adev->has_hw_reset) { adev->has_hw_reset = false; return true; @@ -1429,7 +1422,7 @@ static void amdgpu_switcheroo_set_state(struct pci_dev *pdev, struct drm_device *dev = pci_get_drvdata(pdev); int r; - if (amdgpu_device_supports_atpx(dev) && state == VGA_SWITCHEROO_OFF) + if (amdgpu_device_supports_px(dev) && state == VGA_SWITCHEROO_OFF) return; if (state == VGA_SWITCHEROO_ON) { @@ -1810,6 +1803,7 @@ static int amdgpu_device_parse_gpu_info_fw(struct amdgpu_device *adev) case CHIP_CARRIZO: case CHIP_STONEY: case CHIP_VEGA20: + case CHIP_ALDEBARAN: case CHIP_SIENNA_CICHLID: case CHIP_NAVY_FLOUNDER: case CHIP_DIMGREY_CAVEFISH: @@ -2010,6 +2004,7 @@ static int amdgpu_device_ip_early_init(struct amdgpu_device *adev) case CHIP_RAVEN: case CHIP_ARCTURUS: case CHIP_RENOIR: + case CHIP_ALDEBARAN: if (adev->flags & AMD_IS_APU) adev->family = AMDGPU_FAMILY_RV; else @@ -2045,6 +2040,8 @@ static int amdgpu_device_ip_early_init(struct amdgpu_device *adev) adev->pm.pp_feature = amdgpu_pp_feature_mask; if (amdgpu_sriov_vf(adev) || sched_policy == KFD_SCHED_POLICY_NO_HWS) adev->pm.pp_feature &= ~PP_GFXOFF_MASK; + if (amdgpu_sriov_vf(adev) && adev->asic_type == CHIP_SIENNA_CICHLID) + adev->pm.pp_feature &= ~PP_OVERDRIVE_MASK; for (i = 0; i < adev->num_ip_blocks; i++) { if ((amdgpu_ip_block_mask & (1 << i)) == 0) { @@ -2083,6 +2080,11 @@ static int amdgpu_device_ip_early_init(struct amdgpu_device *adev) amdgpu_vf_error_put(adev, AMDGIM_ERROR_VF_ATOMBIOS_INIT_FAIL, 0, 0); return r; } + + /*get pf2vf msg info at it's earliest time*/ + if (amdgpu_sriov_vf(adev)) + amdgpu_virt_init_data_exchange(adev); + } } @@ -2149,6 +2151,9 @@ static int amdgpu_device_fw_loading(struct amdgpu_device *adev) if (adev->ip_blocks[i].version->type != AMD_IP_BLOCK_TYPE_PSP) continue; + if (!adev->ip_blocks[i].status.sw) + continue; + /* no need to do the fw loading again if already done*/ if (adev->ip_blocks[i].status.hw == true) break; @@ -2289,7 +2294,10 @@ static int amdgpu_device_ip_init(struct amdgpu_device *adev) if (adev->gmc.xgmi.num_physical_nodes > 1) amdgpu_xgmi_add_device(adev); - amdgpu_amdkfd_device_init(adev); + + /* Don't init kfd if whole hive need to be reset during init */ + if (!adev->gmc.xgmi.pending_reset) + amdgpu_amdkfd_device_init(adev); amdgpu_fru_get_product_info(adev); @@ -2359,8 +2367,8 @@ static bool amdgpu_device_check_vram_lost(struct amdgpu_device *adev) * Returns 0 on success, negative error code on failure. */ -static int amdgpu_device_set_cg_state(struct amdgpu_device *adev, - enum amd_clockgating_state state) +int amdgpu_device_set_cg_state(struct amdgpu_device *adev, + enum amd_clockgating_state state) { int i, j, r; @@ -2395,7 +2403,8 @@ static int amdgpu_device_set_cg_state(struct amdgpu_device *adev, return 0; } -static int amdgpu_device_set_pg_state(struct amdgpu_device *adev, enum amd_powergating_state state) +int amdgpu_device_set_pg_state(struct amdgpu_device *adev, + enum amd_powergating_state state) { int i, j, r; @@ -2506,6 +2515,11 @@ static int amdgpu_device_ip_late_init(struct amdgpu_device *adev) if (r) DRM_ERROR("enable mgpu fan boost failed (%d).\n", r); + /* For XGMI + passthrough configuration on arcturus, enable light SBR */ + if (adev->asic_type == CHIP_ARCTURUS && + amdgpu_passthrough(adev) && + adev->gmc.xgmi.num_physical_nodes > 1) + smu_set_light_sbr(&adev->smu, true); if (adev->gmc.xgmi.num_physical_nodes > 1) { mutex_lock(&mgpu_info.mutex); @@ -2743,6 +2757,16 @@ static int amdgpu_device_ip_suspend_phase2(struct amdgpu_device *adev) continue; } + /* skip unnecessary suspend if we do not initialize them yet */ + if (adev->gmc.xgmi.pending_reset && + !(adev->ip_blocks[i].version->type == AMD_IP_BLOCK_TYPE_GMC || + adev->ip_blocks[i].version->type == AMD_IP_BLOCK_TYPE_SMC || + adev->ip_blocks[i].version->type == AMD_IP_BLOCK_TYPE_COMMON || + adev->ip_blocks[i].version->type == AMD_IP_BLOCK_TYPE_IH)) { + adev->ip_blocks[i].status.hw = false; + continue; + } + /* skip suspend of gfx and psp for S0ix * gfx is in gfxoff state, so on resume it will exit gfxoff just * like at runtime. PSP is also part of the always on hardware @@ -2772,7 +2796,6 @@ static int amdgpu_device_ip_suspend_phase2(struct amdgpu_device *adev) } } } - adev->ip_blocks[i].status.hw = false; } return 0; @@ -2793,8 +2816,10 @@ int amdgpu_device_ip_suspend(struct amdgpu_device *adev) { int r; - if (amdgpu_sriov_vf(adev)) + if (amdgpu_sriov_vf(adev)) { + amdgpu_virt_fini_data_exchange(adev); amdgpu_virt_request_full_gpu(adev, false); + } r = amdgpu_device_ip_suspend_phase1(adev); if (r) @@ -3117,8 +3142,9 @@ static void amdgpu_device_xgmi_reset_func(struct work_struct *__work) if (adev->asic_reset_res) goto fail; - if (adev->mmhub.funcs && adev->mmhub.funcs->reset_ras_error_count) - adev->mmhub.funcs->reset_ras_error_count(adev); + if (adev->mmhub.ras_funcs && + adev->mmhub.ras_funcs->reset_ras_error_count) + adev->mmhub.ras_funcs->reset_ras_error_count(adev); } else { task_barrier_full(&hive->tb); @@ -3228,7 +3254,7 @@ int amdgpu_device_init(struct amdgpu_device *adev, struct drm_device *ddev = adev_to_drm(adev); struct pci_dev *pdev = adev->pdev; int r, i; - bool atpx = false; + bool px = false; u32 max_MBps; adev->shutdown = false; @@ -3276,7 +3302,6 @@ int amdgpu_device_init(struct amdgpu_device *adev, /* mutex initialization are all done here so we * can recall function without having locking issues */ - atomic_set(&adev->irq.ih.lock, 0); mutex_init(&adev->firmware.mutex); mutex_init(&adev->pm.mutex); mutex_init(&adev->gfx.gpu_clock_mutex); @@ -3309,6 +3334,8 @@ int amdgpu_device_init(struct amdgpu_device *adev, INIT_LIST_HEAD(&adev->shadow_list); mutex_init(&adev->shadow_list_lock); + INIT_LIST_HEAD(&adev->reset_list); + INIT_DELAYED_WORK(&adev->delayed_init_work, amdgpu_device_delayed_init_work_handler); INIT_DELAYED_WORK(&adev->gfx.gfx_off_delay_work, @@ -3347,17 +3374,6 @@ int amdgpu_device_init(struct amdgpu_device *adev, DRM_INFO("register mmio base: 0x%08X\n", (uint32_t)adev->rmmio_base); DRM_INFO("register mmio size: %u\n", (unsigned)adev->rmmio_size); - /* io port mapping */ - for (i = 0; i < DEVICE_COUNT_RESOURCE; i++) { - if (pci_resource_flags(adev->pdev, i) & IORESOURCE_IO) { - adev->rio_mem_size = pci_resource_len(adev->pdev, i); - adev->rio_mem = pci_iomap(adev->pdev, i, adev->rio_mem_size); - break; - } - } - if (adev->rio_mem == NULL) - DRM_INFO("PCI I/O BAR is not found.\n"); - /* enable PCIE atomic ops */ r = pci_enable_atomic_ops_to_root(adev->pdev, PCI_EXP_DEVCAP2_ATOMIC_COMP32 | @@ -3394,29 +3410,14 @@ int amdgpu_device_init(struct amdgpu_device *adev, /* doorbell bar mapping and doorbell index init*/ amdgpu_device_doorbell_init(adev); - /* if we have > 1 VGA cards, then disable the amdgpu VGA resources */ - /* this will fail for cards that aren't VGA class devices, just - * ignore it */ - if ((adev->pdev->class >> 8) == PCI_CLASS_DISPLAY_VGA) - vga_client_register(adev->pdev, adev, NULL, amdgpu_device_vga_set_decode); - - if (amdgpu_device_supports_atpx(ddev)) - atpx = true; - if (amdgpu_has_atpx() && - (amdgpu_is_atpx_hybrid() || - amdgpu_has_atpx_dgpu_power_cntl()) && - !pci_is_thunderbolt_attached(adev->pdev)) - vga_switcheroo_register_client(adev->pdev, - &amdgpu_switcheroo_ops, atpx); - if (atpx) - vga_switcheroo_init_domain_pm_ops(adev->dev, &adev->vga_pm_domain); - if (amdgpu_emu_mode == 1) { /* post the asic on emulation mode */ emu_soc_asic_init(adev); goto fence_driver_init; } + amdgpu_reset_init(adev); + /* detect if we are with an SRIOV vbios */ amdgpu_device_detect_sriov_bios(adev); @@ -3424,10 +3425,28 @@ int amdgpu_device_init(struct amdgpu_device *adev, * E.g., driver was not cleanly unloaded previously, etc. */ if (!amdgpu_sriov_vf(adev) && amdgpu_asic_need_reset_on_init(adev)) { - r = amdgpu_asic_reset(adev); - if (r) { - dev_err(adev->dev, "asic reset on init failed\n"); - goto failed; + if (adev->gmc.xgmi.num_physical_nodes) { + dev_info(adev->dev, "Pending hive reset.\n"); + adev->gmc.xgmi.pending_reset = true; + /* Only need to init necessary block for SMU to handle the reset */ + for (i = 0; i < adev->num_ip_blocks; i++) { + if (!adev->ip_blocks[i].status.valid) + continue; + if (!(adev->ip_blocks[i].version->type == AMD_IP_BLOCK_TYPE_GMC || + adev->ip_blocks[i].version->type == AMD_IP_BLOCK_TYPE_COMMON || + adev->ip_blocks[i].version->type == AMD_IP_BLOCK_TYPE_IH || + adev->ip_blocks[i].version->type == AMD_IP_BLOCK_TYPE_SMC)) { + DRM_DEBUG("IP %s disabled for hw_init.\n", + adev->ip_blocks[i].version->funcs->name); + adev->ip_blocks[i].status.hw = true; + } + } + } else { + r = amdgpu_asic_reset(adev); + if (r) { + dev_err(adev->dev, "asic reset on init failed\n"); + goto failed; + } } } @@ -3493,11 +3512,11 @@ fence_driver_init: adev->virt.caps &= ~AMDGPU_SRIOV_CAPS_RUNTIME; adev->virt.ops = NULL; r = -EAGAIN; - goto failed; + goto release_ras_con; } dev_err(adev->dev, "amdgpu_device_ip_init failed\n"); amdgpu_vf_error_put(adev, AMDGIM_ERROR_VF_AMDGPU_INIT_FAIL, 0, 0); - goto failed; + goto release_ras_con; } dev_info(adev->dev, @@ -3558,19 +3577,19 @@ fence_driver_init: /* enable clockgating, etc. after ib tests, etc. since some blocks require * explicit gating rather than handling it automatically. */ - r = amdgpu_device_ip_late_init(adev); - if (r) { - dev_err(adev->dev, "amdgpu_device_ip_late_init failed\n"); - amdgpu_vf_error_put(adev, AMDGIM_ERROR_VF_AMDGPU_LATE_INIT_FAIL, 0, r); - goto failed; + if (!adev->gmc.xgmi.pending_reset) { + r = amdgpu_device_ip_late_init(adev); + if (r) { + dev_err(adev->dev, "amdgpu_device_ip_late_init failed\n"); + amdgpu_vf_error_put(adev, AMDGIM_ERROR_VF_AMDGPU_LATE_INIT_FAIL, 0, r); + goto release_ras_con; + } + /* must succeed. */ + amdgpu_ras_resume(adev); + queue_delayed_work(system_wq, &adev->delayed_init_work, + msecs_to_jiffies(AMDGPU_RESUME_MS)); } - /* must succeed. */ - amdgpu_ras_resume(adev); - - queue_delayed_work(system_wq, &adev->delayed_init_work, - msecs_to_jiffies(AMDGPU_RESUME_MS)); - if (amdgpu_sriov_vf(adev)) flush_delayed_work(&adev->delayed_init_work); @@ -3587,12 +3606,30 @@ fence_driver_init: if (amdgpu_device_cache_pci_state(adev->pdev)) pci_restore_state(pdev); + /* if we have > 1 VGA cards, then disable the amdgpu VGA resources */ + /* this will fail for cards that aren't VGA class devices, just + * ignore it */ + if ((adev->pdev->class >> 8) == PCI_CLASS_DISPLAY_VGA) + vga_client_register(adev->pdev, adev, NULL, amdgpu_device_vga_set_decode); + + if (amdgpu_device_supports_px(ddev)) { + px = true; + vga_switcheroo_register_client(adev->pdev, + &amdgpu_switcheroo_ops, px); + vga_switcheroo_init_domain_pm_ops(adev->dev, &adev->vga_pm_domain); + } + + if (adev->gmc.xgmi.pending_reset) + queue_delayed_work(system_wq, &mgpu_info.delayed_reset_work, + msecs_to_jiffies(AMDGPU_RESUME_MS)); + return 0; +release_ras_con: + amdgpu_release_ras_context(adev); + failed: amdgpu_vf_error_trans_all(adev); - if (atpx) - vga_switcheroo_fini_domain_pm_ops(adev->dev); failed_unmap: iounmap(adev->rmmio); @@ -3613,6 +3650,7 @@ void amdgpu_device_fini(struct amdgpu_device *adev) { dev_info(adev->dev, "amdgpu: finishing device.\n"); flush_delayed_work(&adev->delayed_init_work); + ttm_bo_lock_delayed_workqueue(&adev->mman.bdev); adev->shutdown = true; kfree(adev->pci_state); @@ -3641,6 +3679,9 @@ void amdgpu_device_fini(struct amdgpu_device *adev) release_firmware(adev->firmware.gpu_info_fw); adev->firmware.gpu_info_fw = NULL; adev->accel_working = false; + + amdgpu_reset_fini(adev); + /* free i2c buses */ if (!amdgpu_device_has_dc_support(adev)) amdgpu_i2c_fini(adev); @@ -3650,18 +3691,12 @@ void amdgpu_device_fini(struct amdgpu_device *adev) kfree(adev->bios); adev->bios = NULL; - if (amdgpu_has_atpx() && - (amdgpu_is_atpx_hybrid() || - amdgpu_has_atpx_dgpu_power_cntl()) && - !pci_is_thunderbolt_attached(adev->pdev)) + if (amdgpu_device_supports_px(adev_to_drm(adev))) { vga_switcheroo_unregister_client(adev->pdev); - if (amdgpu_device_supports_atpx(adev_to_drm(adev))) vga_switcheroo_fini_domain_pm_ops(adev->dev); + } if ((adev->pdev->class >> 8) == PCI_CLASS_DISPLAY_VGA) vga_client_register(adev->pdev, NULL, NULL, NULL); - if (adev->rio_mem) - pci_iounmap(adev->pdev, adev->rio_mem); - adev->rio_mem = NULL; iounmap(adev->rmmio); adev->rmmio = NULL; amdgpu_device_doorbell_fini(adev); @@ -4079,11 +4114,11 @@ static int amdgpu_device_reset_sriov(struct amdgpu_device *adev, amdgpu_amdkfd_post_reset(adev); error: - amdgpu_virt_release_full_gpu(adev, true); if (!r && adev->virt.gim_feature & AMDGIM_FEATURE_GIM_FLR_VRAMLOST) { amdgpu_inc_vram_lost(adev); r = amdgpu_device_recover_vram(adev); } + amdgpu_virt_release_full_gpu(adev, true); return r; } @@ -4160,6 +4195,8 @@ bool amdgpu_device_should_recover_gpu(struct amdgpu_device *adev) case CHIP_SIENNA_CICHLID: case CHIP_NAVY_FLOUNDER: case CHIP_DIMGREY_CAVEFISH: + case CHIP_VANGOGH: + case CHIP_ALDEBARAN: break; default: goto disabled; @@ -4173,15 +4210,60 @@ disabled: return false; } +int amdgpu_device_mode1_reset(struct amdgpu_device *adev) +{ + u32 i; + int ret = 0; + + amdgpu_atombios_scratch_regs_engine_hung(adev, true); + + dev_info(adev->dev, "GPU mode1 reset\n"); + + /* disable BM */ + pci_clear_master(adev->pdev); + + amdgpu_device_cache_pci_state(adev->pdev); + + if (amdgpu_dpm_is_mode1_reset_supported(adev)) { + dev_info(adev->dev, "GPU smu mode1 reset\n"); + ret = amdgpu_dpm_mode1_reset(adev); + } else { + dev_info(adev->dev, "GPU psp mode1 reset\n"); + ret = psp_gpu_reset(adev); + } -static int amdgpu_device_pre_asic_reset(struct amdgpu_device *adev, - struct amdgpu_job *job, - bool *need_full_reset_arg) + if (ret) + dev_err(adev->dev, "GPU mode1 reset failed\n"); + + amdgpu_device_load_pci_state(adev->pdev); + + /* wait for asic to come out of reset */ + for (i = 0; i < adev->usec_timeout; i++) { + u32 memsize = adev->nbio.funcs->get_memsize(adev); + + if (memsize != 0xffffffff) + break; + udelay(1); + } + + amdgpu_atombios_scratch_regs_engine_hung(adev, false); + return ret; +} + +int amdgpu_device_pre_asic_reset(struct amdgpu_device *adev, + struct amdgpu_reset_context *reset_context) { int i, r = 0; - bool need_full_reset = *need_full_reset_arg; + struct amdgpu_job *job = NULL; + bool need_full_reset = + test_bit(AMDGPU_NEED_FULL_RESET, &reset_context->flags); - amdgpu_debugfs_wait_dump(adev); + if (reset_context->reset_req_dev == adev) + job = reset_context->job; + + /* no need to dump if device is not in good state during probe period */ + if (!adev->gmc.xgmi.pending_reset) + amdgpu_debugfs_wait_dump(adev); if (amdgpu_sriov_vf(adev)) { /* stop the data exchange thread */ @@ -4202,6 +4284,13 @@ static int amdgpu_device_pre_asic_reset(struct amdgpu_device *adev, if(job) drm_sched_increase_karma(&job->base); + r = amdgpu_reset_prepare_hwcontext(adev, reset_context); + /* If reset handler not implemented, continue; otherwise return */ + if (r == -ENOSYS) + r = 0; + else + return r; + /* Don't suspend on bare metal if we are not going to HW reset the ASIC */ if (!amdgpu_sriov_vf(adev)) { @@ -4220,30 +4309,47 @@ static int amdgpu_device_pre_asic_reset(struct amdgpu_device *adev, if (need_full_reset) r = amdgpu_device_ip_suspend(adev); - - *need_full_reset_arg = need_full_reset; + if (need_full_reset) + set_bit(AMDGPU_NEED_FULL_RESET, &reset_context->flags); + else + clear_bit(AMDGPU_NEED_FULL_RESET, + &reset_context->flags); } return r; } -static int amdgpu_do_asic_reset(struct amdgpu_hive_info *hive, - struct list_head *device_list_handle, - bool *need_full_reset_arg, - bool skip_hw_reset) +int amdgpu_do_asic_reset(struct list_head *device_list_handle, + struct amdgpu_reset_context *reset_context) { struct amdgpu_device *tmp_adev = NULL; - bool need_full_reset = *need_full_reset_arg, vram_lost = false; + bool need_full_reset, skip_hw_reset, vram_lost = false; int r = 0; + /* Try reset handler method first */ + tmp_adev = list_first_entry(device_list_handle, struct amdgpu_device, + reset_list); + r = amdgpu_reset_perform_reset(tmp_adev, reset_context); + /* If reset handler not implemented, continue; otherwise return */ + if (r == -ENOSYS) + r = 0; + else + return r; + + /* Reset handler not implemented, use the default method */ + need_full_reset = + test_bit(AMDGPU_NEED_FULL_RESET, &reset_context->flags); + skip_hw_reset = test_bit(AMDGPU_SKIP_HW_RESET, &reset_context->flags); + /* - * ASIC reset has to be done on all HGMI hive nodes ASAP + * ASIC reset has to be done on all XGMI hive nodes ASAP * to allow proper links negotiation in FW (within 1 sec) */ if (!skip_hw_reset && need_full_reset) { - list_for_each_entry(tmp_adev, device_list_handle, gmc.xgmi.head) { + list_for_each_entry(tmp_adev, device_list_handle, reset_list) { /* For XGMI run all resets in parallel to speed up the process */ if (tmp_adev->gmc.xgmi.num_physical_nodes > 1) { + tmp_adev->gmc.xgmi.pending_reset = false; if (!queue_work(system_unbound_wq, &tmp_adev->xgmi_reset_work)) r = -EALREADY; } else @@ -4258,8 +4364,7 @@ static int amdgpu_do_asic_reset(struct amdgpu_hive_info *hive, /* For XGMI wait for all resets to complete before proceed */ if (!r) { - list_for_each_entry(tmp_adev, device_list_handle, - gmc.xgmi.head) { + list_for_each_entry(tmp_adev, device_list_handle, reset_list) { if (tmp_adev->gmc.xgmi.num_physical_nodes > 1) { flush_work(&tmp_adev->xgmi_reset_work); r = tmp_adev->asic_reset_res; @@ -4271,22 +4376,22 @@ static int amdgpu_do_asic_reset(struct amdgpu_hive_info *hive, } if (!r && amdgpu_ras_intr_triggered()) { - list_for_each_entry(tmp_adev, device_list_handle, gmc.xgmi.head) { - if (tmp_adev->mmhub.funcs && - tmp_adev->mmhub.funcs->reset_ras_error_count) - tmp_adev->mmhub.funcs->reset_ras_error_count(tmp_adev); + list_for_each_entry(tmp_adev, device_list_handle, reset_list) { + if (tmp_adev->mmhub.ras_funcs && + tmp_adev->mmhub.ras_funcs->reset_ras_error_count) + tmp_adev->mmhub.ras_funcs->reset_ras_error_count(tmp_adev); } amdgpu_ras_intr_cleared(); } - list_for_each_entry(tmp_adev, device_list_handle, gmc.xgmi.head) { + list_for_each_entry(tmp_adev, device_list_handle, reset_list) { if (need_full_reset) { /* post card */ - if (amdgpu_device_asic_init(tmp_adev)) + r = amdgpu_device_asic_init(tmp_adev); + if (r) { dev_warn(tmp_adev->dev, "asic atom init failed!"); - - if (!r) { + } else { dev_info(tmp_adev->dev, "GPU reset succeeded, trying to resume\n"); r = amdgpu_device_ip_resume_phase1(tmp_adev); if (r) @@ -4319,6 +4424,10 @@ static int amdgpu_do_asic_reset(struct amdgpu_hive_info *hive, */ amdgpu_register_gpu_instance(tmp_adev); + if (!reset_context->hive && + tmp_adev->gmc.xgmi.num_physical_nodes > 1) + amdgpu_xgmi_add_device(tmp_adev); + r = amdgpu_device_ip_late_init(tmp_adev); if (r) goto out; @@ -4335,7 +4444,7 @@ static int amdgpu_do_asic_reset(struct amdgpu_hive_info *hive, * bad_page_threshold value to fix this once * probing driver again. */ - if (!amdgpu_ras_check_err_threshold(tmp_adev)) { + if (!amdgpu_ras_eeprom_check_err_threshold(tmp_adev)) { /* must succeed. */ amdgpu_ras_resume(tmp_adev); } else { @@ -4344,8 +4453,10 @@ static int amdgpu_do_asic_reset(struct amdgpu_hive_info *hive, } /* Update PSP FW topology after reset */ - if (hive && tmp_adev->gmc.xgmi.num_physical_nodes > 1) - r = amdgpu_xgmi_update_topology(hive, tmp_adev); + if (reset_context->hive && + tmp_adev->gmc.xgmi.num_physical_nodes > 1) + r = amdgpu_xgmi_update_topology( + reset_context->hive, tmp_adev); } } @@ -4369,7 +4480,10 @@ out: } end: - *need_full_reset_arg = need_full_reset; + if (need_full_reset) + set_bit(AMDGPU_NEED_FULL_RESET, &reset_context->flags); + else + clear_bit(AMDGPU_NEED_FULL_RESET, &reset_context->flags); return r; } @@ -4385,7 +4499,6 @@ static bool amdgpu_device_lock_adev(struct amdgpu_device *adev, down_write(&adev->reset_sem); } - atomic_inc(&adev->gpu_reset_counter); switch (amdgpu_asic_reset_method(adev)) { case AMD_RESET_METHOD_MODE1: adev->mp1_state = PP_MP1_STATE_SHUTDOWN; @@ -4507,6 +4620,74 @@ static int amdgpu_device_suspend_display_audio(struct amdgpu_device *adev) return 0; } +void amdgpu_device_recheck_guilty_jobs( + struct amdgpu_device *adev, struct list_head *device_list_handle, + struct amdgpu_reset_context *reset_context) +{ + int i, r = 0; + + for (i = 0; i < AMDGPU_MAX_RINGS; ++i) { + struct amdgpu_ring *ring = adev->rings[i]; + int ret = 0; + struct drm_sched_job *s_job; + + if (!ring || !ring->sched.thread) + continue; + + s_job = list_first_entry_or_null(&ring->sched.pending_list, + struct drm_sched_job, list); + if (s_job == NULL) + continue; + + /* clear job's guilty and depend the folowing step to decide the real one */ + drm_sched_reset_karma(s_job); + drm_sched_resubmit_jobs_ext(&ring->sched, 1); + + ret = dma_fence_wait_timeout(s_job->s_fence->parent, false, ring->sched.timeout); + if (ret == 0) { /* timeout */ + DRM_ERROR("Found the real bad job! ring:%s, job_id:%llx\n", + ring->sched.name, s_job->id); + + /* set guilty */ + drm_sched_increase_karma(s_job); +retry: + /* do hw reset */ + if (amdgpu_sriov_vf(adev)) { + amdgpu_virt_fini_data_exchange(adev); + r = amdgpu_device_reset_sriov(adev, false); + if (r) + adev->asic_reset_res = r; + } else { + clear_bit(AMDGPU_SKIP_HW_RESET, + &reset_context->flags); + r = amdgpu_do_asic_reset(device_list_handle, + reset_context); + if (r && r == -EAGAIN) + goto retry; + } + + /* + * add reset counter so that the following + * resubmitted job could flush vmid + */ + atomic_inc(&adev->gpu_reset_counter); + continue; + } + + /* got the hw fence, signal finished fence */ + atomic_dec(ring->sched.score); + dma_fence_get(&s_job->s_fence->finished); + dma_fence_signal(&s_job->s_fence->finished); + dma_fence_put(&s_job->s_fence->finished); + + /* remove node from list and free the job */ + spin_lock(&ring->sched.job_list_lock); + list_del_init(&s_job->list); + spin_unlock(&ring->sched.job_list_lock); + ring->sched.ops->free_job(s_job); + } +} + /** * amdgpu_device_gpu_recover - reset the asic and recover scheduler * @@ -4522,13 +4703,16 @@ int amdgpu_device_gpu_recover(struct amdgpu_device *adev, struct amdgpu_job *job) { struct list_head device_list, *device_list_handle = NULL; - bool need_full_reset = false; bool job_signaled = false; struct amdgpu_hive_info *hive = NULL; struct amdgpu_device *tmp_adev = NULL; int i, r = 0; bool need_emergency_restart = false; bool audio_suspended = false; + int tmp_vram_lost_counter; + struct amdgpu_reset_context reset_context; + + memset(&reset_context, 0, sizeof(reset_context)); /* * Special case: RAS triggered and full reset isn't supported @@ -4569,6 +4753,12 @@ int amdgpu_device_gpu_recover(struct amdgpu_device *adev, mutex_lock(&hive->hive_lock); } + reset_context.method = AMD_RESET_METHOD_NONE; + reset_context.reset_req_dev = adev; + reset_context.job = job; + reset_context.hive = hive; + clear_bit(AMDGPU_NEED_FULL_RESET, &reset_context.flags); + /* * lock the device before we try to operate the linked list * if didn't get the device lock, don't touch the linked list since @@ -4592,16 +4782,18 @@ int amdgpu_device_gpu_recover(struct amdgpu_device *adev, */ INIT_LIST_HEAD(&device_list); if (adev->gmc.xgmi.num_physical_nodes > 1) { - if (!list_is_first(&adev->gmc.xgmi.head, &hive->device_list)) - list_rotate_to_front(&adev->gmc.xgmi.head, &hive->device_list); - device_list_handle = &hive->device_list; + list_for_each_entry(tmp_adev, &hive->device_list, gmc.xgmi.head) + list_add_tail(&tmp_adev->reset_list, &device_list); + if (!list_is_first(&adev->reset_list, &device_list)) + list_rotate_to_front(&adev->reset_list, &device_list); + device_list_handle = &device_list; } else { - list_add_tail(&adev->gmc.xgmi.head, &device_list); + list_add_tail(&adev->reset_list, &device_list); device_list_handle = &device_list; } /* block all schedulers and reset given job's ring */ - list_for_each_entry(tmp_adev, device_list_handle, gmc.xgmi.head) { + list_for_each_entry(tmp_adev, device_list_handle, reset_list) { /* * Try to put the audio codec into suspend state * before gpu reset started. @@ -4646,6 +4838,7 @@ int amdgpu_device_gpu_recover(struct amdgpu_device *adev, if (need_emergency_restart) amdgpu_job_stop_all_jobs_on_sched(&ring->sched); } + atomic_inc(&tmp_adev->gpu_reset_counter); } if (need_emergency_restart) @@ -4665,10 +4858,8 @@ int amdgpu_device_gpu_recover(struct amdgpu_device *adev, } retry: /* Rest of adevs pre asic reset from XGMI hive. */ - list_for_each_entry(tmp_adev, device_list_handle, gmc.xgmi.head) { - r = amdgpu_device_pre_asic_reset(tmp_adev, - (tmp_adev == adev) ? job : NULL, - &need_full_reset); + list_for_each_entry(tmp_adev, device_list_handle, reset_list) { + r = amdgpu_device_pre_asic_reset(tmp_adev, &reset_context); /*TODO Should we stop ?*/ if (r) { dev_err(tmp_adev->dev, "GPU pre asic reset failed with err, %d for drm dev, %s ", @@ -4677,6 +4868,7 @@ retry: /* Rest of adevs pre asic reset from XGMI hive. */ } } + tmp_vram_lost_counter = atomic_read(&((adev)->vram_lost_counter)); /* Actual ASIC resets if needed.*/ /* TODO Implement XGMI hive reset logic for SRIOV */ if (amdgpu_sriov_vf(adev)) { @@ -4684,7 +4876,7 @@ retry: /* Rest of adevs pre asic reset from XGMI hive. */ if (r) adev->asic_reset_res = r; } else { - r = amdgpu_do_asic_reset(hive, device_list_handle, &need_full_reset, false); + r = amdgpu_do_asic_reset(device_list_handle, &reset_context); if (r && r == -EAGAIN) goto retry; } @@ -4692,7 +4884,19 @@ retry: /* Rest of adevs pre asic reset from XGMI hive. */ skip_hw_reset: /* Post ASIC reset for all devs .*/ - list_for_each_entry(tmp_adev, device_list_handle, gmc.xgmi.head) { + list_for_each_entry(tmp_adev, device_list_handle, reset_list) { + + /* + * Sometimes a later bad compute job can block a good gfx job as gfx + * and compute ring share internal GC HW mutually. We add an additional + * guilty jobs recheck step to find the real guilty job, it synchronously + * submits and pends for the first job being signaled. If it gets timeout, + * we identify it as a real guilty job. + */ + if (amdgpu_gpu_recovery == 2 && + !(tmp_vram_lost_counter < atomic_read(&adev->vram_lost_counter))) + amdgpu_device_recheck_guilty_jobs( + tmp_adev, device_list_handle, &reset_context); for (i = 0; i < AMDGPU_MAX_RINGS; ++i) { struct amdgpu_ring *ring = tmp_adev->rings[i]; @@ -4723,10 +4927,17 @@ skip_hw_reset: } skip_sched_resume: - list_for_each_entry(tmp_adev, device_list_handle, gmc.xgmi.head) { - /*unlock kfd: SRIOV would do it separately */ + list_for_each_entry(tmp_adev, device_list_handle, reset_list) { + /* unlock kfd: SRIOV would do it separately */ if (!need_emergency_restart && !amdgpu_sriov_vf(tmp_adev)) amdgpu_amdkfd_post_reset(tmp_adev); + + /* kfd_post_reset will do nothing if kfd device is not initialized, + * need to bring up kfd here if it's not be initialized before + */ + if (!adev->kfd.init_complete) + amdgpu_amdkfd_device_init(adev); + if (audio_suspended) amdgpu_device_resume_display_audio(tmp_adev); amdgpu_device_unlock_adev(tmp_adev); @@ -4988,6 +5199,7 @@ pci_ers_result_t amdgpu_pci_error_detected(struct pci_dev *pdev, pci_channel_sta drm_sched_stop(&ring->sched, NULL); } + atomic_inc(&adev->gpu_reset_counter); return PCI_ERS_RESULT_NEED_RESET; case pci_channel_io_perm_failure: /* Permanent error, prepare for device removal */ @@ -5029,14 +5241,16 @@ pci_ers_result_t amdgpu_pci_slot_reset(struct pci_dev *pdev) struct drm_device *dev = pci_get_drvdata(pdev); struct amdgpu_device *adev = drm_to_adev(dev); int r, i; - bool need_full_reset = true; + struct amdgpu_reset_context reset_context; u32 memsize; struct list_head device_list; DRM_INFO("PCI error: slot reset callback!!\n"); + memset(&reset_context, 0, sizeof(reset_context)); + INIT_LIST_HEAD(&device_list); - list_add_tail(&adev->gmc.xgmi.head, &device_list); + list_add_tail(&adev->reset_list, &device_list); /* wait for asic to come out of reset */ msleep(500); @@ -5057,13 +5271,18 @@ pci_ers_result_t amdgpu_pci_slot_reset(struct pci_dev *pdev) goto out; } + reset_context.method = AMD_RESET_METHOD_NONE; + reset_context.reset_req_dev = adev; + set_bit(AMDGPU_NEED_FULL_RESET, &reset_context.flags); + set_bit(AMDGPU_SKIP_HW_RESET, &reset_context.flags); + adev->in_pci_err_recovery = true; - r = amdgpu_device_pre_asic_reset(adev, NULL, &need_full_reset); + r = amdgpu_device_pre_asic_reset(adev, &reset_context); adev->in_pci_err_recovery = false; if (r) goto out; - r = amdgpu_do_asic_reset(NULL, &device_list, &need_full_reset, true); + r = amdgpu_do_asic_reset(&device_list, &reset_context); out: if (!r) { diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_display.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_display.c index f753e04fee99..8a1fb8b6606e 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_display.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_display.c @@ -837,6 +837,174 @@ static int convert_tiling_flags_to_modifier(struct amdgpu_framebuffer *afb) return 0; } +static void get_block_dimensions(unsigned int block_log2, unsigned int cpp, + unsigned int *width, unsigned int *height) +{ + unsigned int cpp_log2 = ilog2(cpp); + unsigned int pixel_log2 = block_log2 - cpp_log2; + unsigned int width_log2 = (pixel_log2 + 1) / 2; + unsigned int height_log2 = pixel_log2 - width_log2; + + *width = 1 << width_log2; + *height = 1 << height_log2; +} + +static unsigned int get_dcc_block_size(uint64_t modifier, bool rb_aligned, + bool pipe_aligned) +{ + unsigned int ver = AMD_FMT_MOD_GET(TILE_VERSION, modifier); + + switch (ver) { + case AMD_FMT_MOD_TILE_VER_GFX9: { + /* + * TODO: for pipe aligned we may need to check the alignment of the + * total size of the surface, which may need to be bigger than the + * natural alignment due to some HW workarounds + */ + return max(10 + (rb_aligned ? (int)AMD_FMT_MOD_GET(RB, modifier) : 0), 12); + } + case AMD_FMT_MOD_TILE_VER_GFX10: + case AMD_FMT_MOD_TILE_VER_GFX10_RBPLUS: { + int pipes_log2 = AMD_FMT_MOD_GET(PIPE_XOR_BITS, modifier); + + if (ver == AMD_FMT_MOD_TILE_VER_GFX10_RBPLUS && pipes_log2 > 1 && + AMD_FMT_MOD_GET(PACKERS, modifier) == pipes_log2) + ++pipes_log2; + + return max(8 + (pipe_aligned ? pipes_log2 : 0), 12); + } + default: + return 0; + } +} + +static int amdgpu_display_verify_plane(struct amdgpu_framebuffer *rfb, int plane, + const struct drm_format_info *format, + unsigned int block_width, unsigned int block_height, + unsigned int block_size_log2) +{ + unsigned int width = rfb->base.width / + ((plane && plane < format->num_planes) ? format->hsub : 1); + unsigned int height = rfb->base.height / + ((plane && plane < format->num_planes) ? format->vsub : 1); + unsigned int cpp = plane < format->num_planes ? format->cpp[plane] : 1; + unsigned int block_pitch = block_width * cpp; + unsigned int min_pitch = ALIGN(width * cpp, block_pitch); + unsigned int block_size = 1 << block_size_log2; + uint64_t size; + + if (rfb->base.pitches[plane] % block_pitch) { + drm_dbg_kms(rfb->base.dev, + "pitch %d for plane %d is not a multiple of block pitch %d\n", + rfb->base.pitches[plane], plane, block_pitch); + return -EINVAL; + } + if (rfb->base.pitches[plane] < min_pitch) { + drm_dbg_kms(rfb->base.dev, + "pitch %d for plane %d is less than minimum pitch %d\n", + rfb->base.pitches[plane], plane, min_pitch); + return -EINVAL; + } + + /* Force at least natural alignment. */ + if (rfb->base.offsets[plane] % block_size) { + drm_dbg_kms(rfb->base.dev, + "offset 0x%x for plane %d is not a multiple of block pitch 0x%x\n", + rfb->base.offsets[plane], plane, block_size); + return -EINVAL; + } + + size = rfb->base.offsets[plane] + + (uint64_t)rfb->base.pitches[plane] / block_pitch * + block_size * DIV_ROUND_UP(height, block_height); + + if (rfb->base.obj[0]->size < size) { + drm_dbg_kms(rfb->base.dev, + "BO size 0x%zx is less than 0x%llx required for plane %d\n", + rfb->base.obj[0]->size, size, plane); + return -EINVAL; + } + + return 0; +} + + +static int amdgpu_display_verify_sizes(struct amdgpu_framebuffer *rfb) +{ + const struct drm_format_info *format_info = drm_format_info(rfb->base.format->format); + uint64_t modifier = rfb->base.modifier; + int ret; + unsigned int i, block_width, block_height, block_size_log2; + + if (!rfb->base.dev->mode_config.allow_fb_modifiers) + return 0; + + for (i = 0; i < format_info->num_planes; ++i) { + if (modifier == DRM_FORMAT_MOD_LINEAR) { + block_width = 256 / format_info->cpp[i]; + block_height = 1; + block_size_log2 = 8; + } else { + int swizzle = AMD_FMT_MOD_GET(TILE, modifier); + + switch ((swizzle & ~3) + 1) { + case DC_SW_256B_S: + block_size_log2 = 8; + break; + case DC_SW_4KB_S: + case DC_SW_4KB_S_X: + block_size_log2 = 12; + break; + case DC_SW_64KB_S: + case DC_SW_64KB_S_T: + case DC_SW_64KB_S_X: + block_size_log2 = 16; + break; + default: + drm_dbg_kms(rfb->base.dev, + "Swizzle mode with unknown block size: %d\n", swizzle); + return -EINVAL; + } + + get_block_dimensions(block_size_log2, format_info->cpp[i], + &block_width, &block_height); + } + + ret = amdgpu_display_verify_plane(rfb, i, format_info, + block_width, block_height, block_size_log2); + if (ret) + return ret; + } + + if (AMD_FMT_MOD_GET(DCC, modifier)) { + if (AMD_FMT_MOD_GET(DCC_RETILE, modifier)) { + block_size_log2 = get_dcc_block_size(modifier, false, false); + get_block_dimensions(block_size_log2 + 8, format_info->cpp[0], + &block_width, &block_height); + ret = amdgpu_display_verify_plane(rfb, i, format_info, + block_width, block_height, + block_size_log2); + if (ret) + return ret; + + ++i; + block_size_log2 = get_dcc_block_size(modifier, true, true); + } else { + bool pipe_aligned = AMD_FMT_MOD_GET(DCC_PIPE_ALIGN, modifier); + + block_size_log2 = get_dcc_block_size(modifier, true, pipe_aligned); + } + get_block_dimensions(block_size_log2 + 8, format_info->cpp[0], + &block_width, &block_height); + ret = amdgpu_display_verify_plane(rfb, i, format_info, + block_width, block_height, block_size_log2); + if (ret) + return ret; + } + + return 0; +} + static int amdgpu_display_get_fb_info(const struct amdgpu_framebuffer *amdgpu_fb, uint64_t *tiling_flags, bool *tmz_surface) { @@ -870,17 +1038,73 @@ static int amdgpu_display_get_fb_info(const struct amdgpu_framebuffer *amdgpu_fb return r; } +int amdgpu_display_gem_fb_init(struct drm_device *dev, + struct amdgpu_framebuffer *rfb, + const struct drm_mode_fb_cmd2 *mode_cmd, + struct drm_gem_object *obj) +{ + int ret; + + rfb->base.obj[0] = obj; + drm_helper_mode_fill_fb_struct(dev, &rfb->base, mode_cmd); + ret = drm_framebuffer_init(dev, &rfb->base, &amdgpu_fb_funcs); + if (ret) + goto err; + + ret = amdgpu_display_framebuffer_init(dev, rfb, mode_cmd, obj); + if (ret) + goto err; + + return 0; +err: + drm_err(dev, "Failed to init gem fb: %d\n", ret); + rfb->base.obj[0] = NULL; + return ret; +} + +int amdgpu_display_gem_fb_verify_and_init( + struct drm_device *dev, struct amdgpu_framebuffer *rfb, + struct drm_file *file_priv, const struct drm_mode_fb_cmd2 *mode_cmd, + struct drm_gem_object *obj) +{ + int ret; + + rfb->base.obj[0] = obj; + drm_helper_mode_fill_fb_struct(dev, &rfb->base, mode_cmd); + ret = drm_framebuffer_init(dev, &rfb->base, &amdgpu_fb_funcs); + if (ret) + goto err; + /* Verify that the modifier is supported. */ + if (!drm_any_plane_has_format(dev, mode_cmd->pixel_format, + mode_cmd->modifier[0])) { + struct drm_format_name_buf format_name; + drm_dbg_kms(dev, + "unsupported pixel format %s / modifier 0x%llx\n", + drm_get_format_name(mode_cmd->pixel_format, + &format_name), + mode_cmd->modifier[0]); + + ret = -EINVAL; + goto err; + } + + ret = amdgpu_display_framebuffer_init(dev, rfb, mode_cmd, obj); + if (ret) + goto err; + + return 0; +err: + drm_err(dev, "Failed to verify and init gem fb: %d\n", ret); + rfb->base.obj[0] = NULL; + return ret; +} + int amdgpu_display_framebuffer_init(struct drm_device *dev, struct amdgpu_framebuffer *rfb, const struct drm_mode_fb_cmd2 *mode_cmd, struct drm_gem_object *obj) { int ret, i; - rfb->base.obj[0] = obj; - drm_helper_mode_fill_fb_struct(dev, &rfb->base, mode_cmd); - ret = drm_framebuffer_init(dev, &rfb->base, &amdgpu_fb_funcs); - if (ret) - goto fail; /* * This needs to happen before modifier conversion as that might change @@ -891,13 +1115,13 @@ int amdgpu_display_framebuffer_init(struct drm_device *dev, drm_dbg_kms(dev, "Plane 0 and %d have different BOs: %u vs. %u\n", i, mode_cmd->handles[0], mode_cmd->handles[i]); ret = -EINVAL; - goto fail; + return ret; } } ret = amdgpu_display_get_fb_info(rfb, &rfb->tiling_flags, &rfb->tmz_surface); if (ret) - goto fail; + return ret; if (dev->mode_config.allow_fb_modifiers && !(rfb->base.flags & DRM_MODE_FB_MODIFIERS)) { @@ -905,20 +1129,20 @@ int amdgpu_display_framebuffer_init(struct drm_device *dev, if (ret) { drm_dbg_kms(dev, "Failed to convert tiling flags 0x%llX to a modifier", rfb->tiling_flags); - goto fail; + return ret; } } - for (i = 1; i < rfb->base.format->num_planes; ++i) { + ret = amdgpu_display_verify_sizes(rfb); + if (ret) + return ret; + + for (i = 0; i < rfb->base.format->num_planes; ++i) { + drm_gem_object_get(rfb->base.obj[0]); rfb->base.obj[i] = rfb->base.obj[0]; - drm_gem_object_get(rfb->base.obj[i]); } return 0; - -fail: - rfb->base.obj[0] = NULL; - return ret; } struct drm_framebuffer * @@ -944,6 +1168,7 @@ amdgpu_display_user_framebuffer_create(struct drm_device *dev, domains = amdgpu_display_supported_domains(drm_to_adev(dev), bo->flags); if (obj->import_attach && !(domains & AMDGPU_GEM_DOMAIN_GTT)) { drm_dbg_kms(dev, "Cannot create framebuffer from imported dma_buf\n"); + drm_gem_object_put(obj); return ERR_PTR(-EINVAL); } @@ -953,13 +1178,15 @@ amdgpu_display_user_framebuffer_create(struct drm_device *dev, return ERR_PTR(-ENOMEM); } - ret = amdgpu_display_framebuffer_init(dev, amdgpu_fb, mode_cmd, obj); + ret = amdgpu_display_gem_fb_verify_and_init(dev, amdgpu_fb, file_priv, + mode_cmd, obj); if (ret) { kfree(amdgpu_fb); drm_gem_object_put(obj); return ERR_PTR(ret); } + drm_gem_object_put(obj); return &amdgpu_fb->base; } @@ -1355,7 +1582,7 @@ int amdgpu_display_suspend_helper(struct amdgpu_device *adev) } } } - return r; + return 0; } int amdgpu_display_resume_helper(struct amdgpu_device *adev) diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_dma_buf.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_dma_buf.c index 47e0b48dc26f..baa980a477d9 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_dma_buf.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_dma_buf.c @@ -291,8 +291,8 @@ static struct sg_table *amdgpu_dma_buf_map(struct dma_buf_attachment *attach, break; case TTM_PL_VRAM: - r = amdgpu_vram_mgr_alloc_sgt(adev, &bo->tbo.mem, attach->dev, - dir, &sgt); + r = amdgpu_vram_mgr_alloc_sgt(adev, &bo->tbo.mem, 0, + bo->tbo.base.size, attach->dev, dir, &sgt); if (r) return ERR_PTR(r); break; @@ -321,17 +321,12 @@ static void amdgpu_dma_buf_unmap(struct dma_buf_attachment *attach, struct sg_table *sgt, enum dma_data_direction dir) { - struct dma_buf *dma_buf = attach->dmabuf; - struct drm_gem_object *obj = dma_buf->priv; - struct amdgpu_bo *bo = gem_to_amdgpu_bo(obj); - struct amdgpu_device *adev = amdgpu_ttm_adev(bo->tbo.bdev); - if (sgt->sgl->page_link) { dma_unmap_sgtable(attach->dev, sgt, dir, 0); sg_free_table(sgt); kfree(sgt); } else { - amdgpu_vram_mgr_free_sgt(adev, attach->dev, dir, sgt); + amdgpu_vram_mgr_free_sgt(attach->dev, dir, sgt); } } @@ -434,22 +429,22 @@ amdgpu_dma_buf_create_obj(struct drm_device *dev, struct dma_buf *dma_buf) { struct dma_resv *resv = dma_buf->resv; struct amdgpu_device *adev = drm_to_adev(dev); - struct amdgpu_bo *bo; - struct amdgpu_bo_param bp; struct drm_gem_object *gobj; + struct amdgpu_bo *bo; + uint64_t flags = 0; int ret; - memset(&bp, 0, sizeof(bp)); - bp.size = dma_buf->size; - bp.byte_align = PAGE_SIZE; - bp.domain = AMDGPU_GEM_DOMAIN_CPU; - bp.flags = 0; - bp.type = ttm_bo_type_sg; - bp.resv = resv; dma_resv_lock(resv, NULL); + + if (dma_buf->ops == &amdgpu_dmabuf_ops) { + struct amdgpu_bo *other = gem_to_amdgpu_bo(dma_buf->priv); + + flags |= other->flags & AMDGPU_GEM_CREATE_CPU_GTT_USWC; + } + ret = amdgpu_gem_object_create(adev, dma_buf->size, PAGE_SIZE, - AMDGPU_GEM_DOMAIN_CPU, - 0, ttm_bo_type_sg, resv, &gobj); + AMDGPU_GEM_DOMAIN_CPU, flags, + ttm_bo_type_sg, resv, &gobj); if (ret) goto error; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c index e92e7dea71da..f93883db2b46 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c @@ -36,6 +36,7 @@ #include <linux/vga_switcheroo.h> #include <drm/drm_probe_helper.h> #include <linux/mmu_notifier.h> +#include <linux/suspend.h> #include "amdgpu.h" #include "amdgpu_irq.h" @@ -45,6 +46,8 @@ #include "amdgpu_amdkfd.h" #include "amdgpu_ras.h" +#include "amdgpu_xgmi.h" +#include "amdgpu_reset.h" /* * KMS wrapper. @@ -90,9 +93,10 @@ * - 3.38.0 - Add AMDGPU_IB_FLAG_EMIT_MEM_SYNC * - 3.39.0 - DMABUF implicit sync does a full pipeline sync * - 3.40.0 - Add AMDGPU_IDS_FLAGS_TMZ + * - 3.41.0 - Add video codec query */ #define KMS_DRIVER_MAJOR 3 -#define KMS_DRIVER_MINOR 40 +#define KMS_DRIVER_MINOR 41 #define KMS_DRIVER_PATCHLEVEL 0 int amdgpu_vram_limit; @@ -145,6 +149,7 @@ int amdgpu_compute_multipipe = -1; int amdgpu_gpu_recovery = -1; /* auto */ int amdgpu_emu_mode; uint amdgpu_smu_memory_pool_size; +int amdgpu_smu_pptable_id = -1; /* * FBC (bit 0) disabled by default * MULTI_MON_PP_MCLK_SWITCH (bit 1) enabled by default @@ -162,16 +167,26 @@ int amdgpu_discovery = -1; int amdgpu_mes; int amdgpu_noretry = -1; int amdgpu_force_asic_type = -1; -int amdgpu_tmz; +int amdgpu_tmz = -1; /* auto */ +uint amdgpu_freesync_vid_mode; int amdgpu_reset_method = -1; /* auto */ int amdgpu_num_kcq = -1; +static void amdgpu_drv_delayed_reset_work_handler(struct work_struct *work); + struct amdgpu_mgpu_info mgpu_info = { .mutex = __MUTEX_INITIALIZER(mgpu_info.mutex), + .delayed_reset_work = __DELAYED_WORK_INITIALIZER( + mgpu_info.delayed_reset_work, + amdgpu_drv_delayed_reset_work_handler, 0), }; int amdgpu_ras_enable = -1; uint amdgpu_ras_mask = 0xffffffff; int amdgpu_bad_page_threshold = -1; +struct amdgpu_watchdog_timer amdgpu_watchdog_timer = { + .timeout_fatal_disable = false, + .period = 0x0, /* default to 0x0 (timeout disable) */ +}; /** * DOC: vramlimit (int) @@ -502,7 +517,7 @@ module_param_named(compute_multipipe, amdgpu_compute_multipipe, int, 0444); * DOC: gpu_recovery (int) * Set to enable GPU recovery mechanism (1 = enable, 0 = disable). The default is -1 (auto, disabled except SRIOV). */ -MODULE_PARM_DESC(gpu_recovery, "Enable GPU recovery mechanism, (1 = enable, 0 = disable, -1 = auto)"); +MODULE_PARM_DESC(gpu_recovery, "Enable GPU recovery mechanism, (2 = advanced tdr mode, 1 = enable, 0 = disable, -1 = auto)"); module_param_named(gpu_recovery, amdgpu_gpu_recovery, int, 0444); /** @@ -528,6 +543,20 @@ MODULE_PARM_DESC(ras_mask, "Mask of RAS features to enable (default 0xffffffff), module_param_named(ras_mask, amdgpu_ras_mask, uint, 0444); /** + * DOC: timeout_fatal_disable (bool) + * Disable Watchdog timeout fatal error event + */ +MODULE_PARM_DESC(timeout_fatal_disable, "disable watchdog timeout fatal error (false = default)"); +module_param_named(timeout_fatal_disable, amdgpu_watchdog_timer.timeout_fatal_disable, bool, 0644); + +/** + * DOC: timeout_period (uint) + * Modify the watchdog timeout max_cycles as (1 << period) + */ +MODULE_PARM_DESC(timeout_period, "watchdog timeout period (0 = timeout disabled, 1 ~ 0x23 = timeout maxcycles = (1 << period)"); +module_param_named(timeout_period, amdgpu_watchdog_timer.period, uint, 0644); + +/** * DOC: si_support (int) * Set SI support driver. This parameter works after set config CONFIG_DRM_AMDGPU_SI. For SI asic, when radeon driver is enabled, * set value 0 to use radeon driver, while set value 1 to use amdgpu driver. The default is using radeon driver when it available, @@ -748,6 +777,13 @@ bool no_system_mem_limit; module_param(no_system_mem_limit, bool, 0644); MODULE_PARM_DESC(no_system_mem_limit, "disable system memory limit (false = default)"); +/** + * DOC: no_queue_eviction_on_vm_fault (int) + * If set, process queues will not be evicted on gpuvm fault. This is to keep the wavefront context for debugging (0 = queue eviction, 1 = no queue eviction). The default is 0 (queue eviction). + */ +int amdgpu_no_queue_eviction_on_vm_fault = 0; +MODULE_PARM_DESC(no_queue_eviction_on_vm_fault, "No queue eviction on VM fault (0 = queue eviction, 1 = no queue eviction)"); +module_param_named(no_queue_eviction_on_vm_fault, amdgpu_no_queue_eviction_on_vm_fault, int, 0444); #endif /** @@ -792,10 +828,21 @@ module_param_named(backlight, amdgpu_backlight, bint, 0444); * * The default value: 0 (off). TODO: change to auto till it is completed. */ -MODULE_PARM_DESC(tmz, "Enable TMZ feature (-1 = auto, 0 = off (default), 1 = on)"); +MODULE_PARM_DESC(tmz, "Enable TMZ feature (-1 = auto (default), 0 = off, 1 = on)"); module_param_named(tmz, amdgpu_tmz, int, 0444); /** + * DOC: freesync_video (uint) + * Enabled the optimization to adjust front porch timing to achieve seamless mode change experience + * when setting a freesync supported mode for which full modeset is not needed. + * The default value: 0 (off). + */ +MODULE_PARM_DESC( + freesync_video, + "Enable freesync modesetting optimization feature (0 = off (default), 1 = on)"); +module_param_named(freesync_video, amdgpu_freesync_vid_mode, uint, 0444); + +/** * DOC: reset_method (int) * GPU reset method (-1 = auto (default), 0 = legacy, 1 = mode0, 2 = mode1, 3 = mode2, 4 = baco, 5 = pci) */ @@ -815,6 +862,15 @@ module_param_named(bad_page_threshold, amdgpu_bad_page_threshold, int, 0444); MODULE_PARM_DESC(num_kcq, "number of kernel compute queue user want to setup (8 if set to greater than 8 or less than 0, only affect gfx 8+)"); module_param_named(num_kcq, amdgpu_num_kcq, int, 0444); +/** + * DOC: smu_pptable_id (int) + * Used to override pptable id. id = 0 use VBIOS pptable. + * id > 0 use the soft pptable with specicfied id. + */ +MODULE_PARM_DESC(smu_pptable_id, + "specify pptable id to be used (-1 = auto(default) value, 0 = use pptable from vbios, > 0 = soft pptable id)"); +module_param_named(smu_pptable_id, amdgpu_smu_pptable_id, int, 0444); + static const struct pci_device_id pciidlist[] = { #ifdef CONFIG_DRM_AMDGPU_SI {0x1002, 0x6780, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_TAHITI}, @@ -1125,6 +1181,11 @@ static const struct pci_device_id pciidlist[] = { {0x1002, 0x73E2, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_DIMGREY_CAVEFISH}, {0x1002, 0x73FF, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_DIMGREY_CAVEFISH}, + /* Aldebaran */ + {0x1002, 0x7408, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_ALDEBARAN|AMD_EXP_HW_SUPPORT}, + {0x1002, 0x740C, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_ALDEBARAN|AMD_EXP_HW_SUPPORT}, + {0x1002, 0x740F, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_ALDEBARAN|AMD_EXP_HW_SUPPORT}, + {0, 0, 0} }; @@ -1279,6 +1340,98 @@ amdgpu_pci_shutdown(struct pci_dev *pdev) adev->mp1_state = PP_MP1_STATE_NONE; } +/** + * amdgpu_drv_delayed_reset_work_handler - work handler for reset + * + * @work: work_struct. + */ +static void amdgpu_drv_delayed_reset_work_handler(struct work_struct *work) +{ + struct list_head device_list; + struct amdgpu_device *adev; + int i, r; + struct amdgpu_reset_context reset_context; + + memset(&reset_context, 0, sizeof(reset_context)); + + mutex_lock(&mgpu_info.mutex); + if (mgpu_info.pending_reset == true) { + mutex_unlock(&mgpu_info.mutex); + return; + } + mgpu_info.pending_reset = true; + mutex_unlock(&mgpu_info.mutex); + + /* Use a common context, just need to make sure full reset is done */ + reset_context.method = AMD_RESET_METHOD_NONE; + set_bit(AMDGPU_NEED_FULL_RESET, &reset_context.flags); + + for (i = 0; i < mgpu_info.num_dgpu; i++) { + adev = mgpu_info.gpu_ins[i].adev; + reset_context.reset_req_dev = adev; + r = amdgpu_device_pre_asic_reset(adev, &reset_context); + if (r) { + dev_err(adev->dev, "GPU pre asic reset failed with err, %d for drm dev, %s ", + r, adev_to_drm(adev)->unique); + } + if (!queue_work(system_unbound_wq, &adev->xgmi_reset_work)) + r = -EALREADY; + } + for (i = 0; i < mgpu_info.num_dgpu; i++) { + adev = mgpu_info.gpu_ins[i].adev; + flush_work(&adev->xgmi_reset_work); + adev->gmc.xgmi.pending_reset = false; + } + + /* reset function will rebuild the xgmi hive info , clear it now */ + for (i = 0; i < mgpu_info.num_dgpu; i++) + amdgpu_xgmi_remove_device(mgpu_info.gpu_ins[i].adev); + + INIT_LIST_HEAD(&device_list); + + for (i = 0; i < mgpu_info.num_dgpu; i++) + list_add_tail(&mgpu_info.gpu_ins[i].adev->reset_list, &device_list); + + /* unregister the GPU first, reset function will add them back */ + list_for_each_entry(adev, &device_list, reset_list) + amdgpu_unregister_gpu_instance(adev); + + /* Use a common context, just need to make sure full reset is done */ + set_bit(AMDGPU_SKIP_HW_RESET, &reset_context.flags); + r = amdgpu_do_asic_reset(&device_list, &reset_context); + + if (r) { + DRM_ERROR("reinit gpus failure"); + return; + } + for (i = 0; i < mgpu_info.num_dgpu; i++) { + adev = mgpu_info.gpu_ins[i].adev; + if (!adev->kfd.init_complete) + amdgpu_amdkfd_device_init(adev); + amdgpu_ttm_set_buffer_funcs_status(adev, true); + } + return; +} + +static int amdgpu_pmops_prepare(struct device *dev) +{ + struct drm_device *drm_dev = dev_get_drvdata(dev); + + /* Return a positive number here so + * DPM_FLAG_SMART_SUSPEND works properly + */ + if (amdgpu_device_supports_boco(drm_dev)) + return pm_runtime_suspended(dev) && + pm_suspend_via_firmware(); + + return 0; +} + +static void amdgpu_pmops_complete(struct device *dev) +{ + /* nothing to do */ +} + static int amdgpu_pmops_suspend(struct device *dev) { struct drm_device *drm_dev = dev_get_drvdata(dev); @@ -1364,7 +1517,7 @@ static int amdgpu_pmops_runtime_suspend(struct device *dev) } adev->in_runpm = true; - if (amdgpu_device_supports_atpx(drm_dev)) + if (amdgpu_device_supports_px(drm_dev)) drm_dev->switch_power_state = DRM_SWITCH_POWER_CHANGING; ret = amdgpu_device_suspend(drm_dev, false); @@ -1373,16 +1526,14 @@ static int amdgpu_pmops_runtime_suspend(struct device *dev) return ret; } - if (amdgpu_device_supports_atpx(drm_dev)) { + if (amdgpu_device_supports_px(drm_dev)) { /* Only need to handle PCI state in the driver for ATPX * PCI core handles it for _PR3. */ - if (!amdgpu_is_atpx_hybrid()) { - amdgpu_device_cache_pci_state(pdev); - pci_disable_device(pdev); - pci_ignore_hotplug(pdev); - pci_set_power_state(pdev, PCI_D3cold); - } + amdgpu_device_cache_pci_state(pdev); + pci_disable_device(pdev); + pci_ignore_hotplug(pdev); + pci_set_power_state(pdev, PCI_D3cold); drm_dev->switch_power_state = DRM_SWITCH_POWER_DYNAMIC_OFF; } else if (amdgpu_device_supports_baco(drm_dev)) { amdgpu_device_baco_enter(drm_dev); @@ -1401,19 +1552,17 @@ static int amdgpu_pmops_runtime_resume(struct device *dev) if (!adev->runpm) return -EINVAL; - if (amdgpu_device_supports_atpx(drm_dev)) { + if (amdgpu_device_supports_px(drm_dev)) { drm_dev->switch_power_state = DRM_SWITCH_POWER_CHANGING; /* Only need to handle PCI state in the driver for ATPX * PCI core handles it for _PR3. */ - if (!amdgpu_is_atpx_hybrid()) { - pci_set_power_state(pdev, PCI_D0); - amdgpu_device_load_pci_state(pdev); - ret = pci_enable_device(pdev); - if (ret) - return ret; - } + pci_set_power_state(pdev, PCI_D0); + amdgpu_device_load_pci_state(pdev); + ret = pci_enable_device(pdev); + if (ret) + return ret; pci_set_master(pdev); } else if (amdgpu_device_supports_boco(drm_dev)) { /* Only need to handle PCI state in the driver for ATPX @@ -1424,7 +1573,10 @@ static int amdgpu_pmops_runtime_resume(struct device *dev) amdgpu_device_baco_exit(drm_dev); } ret = amdgpu_device_resume(drm_dev, false); - if (amdgpu_device_supports_atpx(drm_dev)) + if (ret) + return ret; + + if (amdgpu_device_supports_px(drm_dev)) drm_dev->switch_power_state = DRM_SWITCH_POWER_ON; adev->in_runpm = false; return 0; @@ -1505,6 +1657,8 @@ out: } static const struct dev_pm_ops amdgpu_pm_ops = { + .prepare = amdgpu_pmops_prepare, + .complete = amdgpu_pmops_complete, .suspend = amdgpu_pmops_suspend, .resume = amdgpu_pmops_resume, .freeze = amdgpu_pmops_freeze, diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_fb.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_fb.c index 24010cacf7d0..4f10c4529840 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_fb.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_fb.c @@ -205,7 +205,6 @@ static int amdgpufb_create(struct drm_fb_helper *helper, struct drm_gem_object *gobj = NULL; struct amdgpu_bo *abo = NULL; int ret; - unsigned long tmp; memset(&mode_cmd, 0, sizeof(mode_cmd)); mode_cmd.width = sizes->surface_width; @@ -232,8 +231,8 @@ static int amdgpufb_create(struct drm_fb_helper *helper, goto out; } - ret = amdgpu_display_framebuffer_init(adev_to_drm(adev), &rfbdev->rfb, - &mode_cmd, gobj); + ret = amdgpu_display_gem_fb_init(adev_to_drm(adev), &rfbdev->rfb, + &mode_cmd, gobj); if (ret) { DRM_ERROR("failed to initialize framebuffer %d\n", ret); goto out; @@ -246,8 +245,7 @@ static int amdgpufb_create(struct drm_fb_helper *helper, info->fbops = &amdgpufb_ops; - tmp = amdgpu_bo_gpu_offset(abo) - adev->gmc.vram_start; - info->fix.smem_start = adev->gmc.aper_base + tmp; + info->fix.smem_start = amdgpu_gmc_vram_cpu_pa(adev, abo); info->fix.smem_len = amdgpu_bo_size(abo); info->screen_base = amdgpu_bo_kptr(abo); info->screen_size = amdgpu_bo_size(abo); diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_fence.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_fence.c index d56f4023ebb3..47ea46859618 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_fence.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_fence.c @@ -36,8 +36,6 @@ #include <linux/firmware.h> #include <linux/pm_runtime.h> -#include <drm/drm_debugfs.h> - #include "amdgpu.h" #include "amdgpu_trace.h" @@ -441,7 +439,8 @@ int amdgpu_fence_driver_start_ring(struct amdgpu_ring *ring, * Helper function for amdgpu_fence_driver_init(). */ int amdgpu_fence_driver_init_ring(struct amdgpu_ring *ring, - unsigned num_hw_submission) + unsigned num_hw_submission, + atomic_t *sched_score) { struct amdgpu_device *adev = ring->adev; long timeout; @@ -469,30 +468,31 @@ int amdgpu_fence_driver_init_ring(struct amdgpu_ring *ring, return -ENOMEM; /* No need to setup the GPU scheduler for rings that don't need it */ - if (!ring->no_scheduler) { - switch (ring->funcs->type) { - case AMDGPU_RING_TYPE_GFX: - timeout = adev->gfx_timeout; - break; - case AMDGPU_RING_TYPE_COMPUTE: - timeout = adev->compute_timeout; - break; - case AMDGPU_RING_TYPE_SDMA: - timeout = adev->sdma_timeout; - break; - default: - timeout = adev->video_timeout; - break; - } + if (ring->no_scheduler) + return 0; - r = drm_sched_init(&ring->sched, &amdgpu_sched_ops, - num_hw_submission, amdgpu_job_hang_limit, - timeout, ring->name); - if (r) { - DRM_ERROR("Failed to create scheduler on ring %s.\n", - ring->name); - return r; - } + switch (ring->funcs->type) { + case AMDGPU_RING_TYPE_GFX: + timeout = adev->gfx_timeout; + break; + case AMDGPU_RING_TYPE_COMPUTE: + timeout = adev->compute_timeout; + break; + case AMDGPU_RING_TYPE_SDMA: + timeout = adev->sdma_timeout; + break; + default: + timeout = adev->video_timeout; + break; + } + + r = drm_sched_init(&ring->sched, &amdgpu_sched_ops, + num_hw_submission, amdgpu_job_hang_limit, + timeout, sched_score, ring->name); + if (r) { + DRM_ERROR("Failed to create scheduler on ring %s.\n", + ring->name); + return r; } return 0; @@ -533,6 +533,8 @@ void amdgpu_fence_driver_fini(struct amdgpu_device *adev) if (!ring || !ring->fence_drv.initialized) continue; + if (!ring->no_scheduler) + drm_sched_fini(&ring->sched); r = amdgpu_fence_wait_empty(ring); if (r) { /* no need to trigger GPU reset as we are unloading */ @@ -541,8 +543,7 @@ void amdgpu_fence_driver_fini(struct amdgpu_device *adev) if (ring->fence_drv.irq_src) amdgpu_irq_put(adev, ring->fence_drv.irq_src, ring->fence_drv.irq_type); - if (!ring->no_scheduler) - drm_sched_fini(&ring->sched); + del_timer_sync(&ring->fence_drv.fallback_timer); for (j = 0; j <= ring->fence_drv.num_fences_mask; ++j) dma_fence_put(ring->fence_drv.fences[j]); @@ -697,11 +698,9 @@ static const struct dma_fence_ops amdgpu_fence_ops = { * Fence debugfs */ #if defined(CONFIG_DEBUG_FS) -static int amdgpu_debugfs_fence_info(struct seq_file *m, void *data) +static int amdgpu_debugfs_fence_info_show(struct seq_file *m, void *unused) { - struct drm_info_node *node = (struct drm_info_node *)m->private; - struct drm_device *dev = node->minor->dev; - struct amdgpu_device *adev = drm_to_adev(dev); + struct amdgpu_device *adev = (struct amdgpu_device *)m->private; int i; for (i = 0; i < AMDGPU_MAX_RINGS; ++i) { @@ -746,11 +745,10 @@ static int amdgpu_debugfs_fence_info(struct seq_file *m, void *data) * * Manually trigger a gpu reset at the next fence wait. */ -static int amdgpu_debugfs_gpu_recover(struct seq_file *m, void *data) +static int gpu_recover_get(void *data, u64 *val) { - struct drm_info_node *node = (struct drm_info_node *) m->private; - struct drm_device *dev = node->minor->dev; - struct amdgpu_device *adev = drm_to_adev(dev); + struct amdgpu_device *adev = (struct amdgpu_device *)data; + struct drm_device *dev = adev_to_drm(adev); int r; r = pm_runtime_get_sync(dev->dev); @@ -759,8 +757,7 @@ static int amdgpu_debugfs_gpu_recover(struct seq_file *m, void *data) return 0; } - seq_printf(m, "gpu recover\n"); - amdgpu_device_gpu_recover(adev, NULL); + *val = amdgpu_device_gpu_recover(adev, NULL); pm_runtime_mark_last_busy(dev->dev); pm_runtime_put_autosuspend(dev->dev); @@ -768,26 +765,24 @@ static int amdgpu_debugfs_gpu_recover(struct seq_file *m, void *data) return 0; } -static const struct drm_info_list amdgpu_debugfs_fence_list[] = { - {"amdgpu_fence_info", &amdgpu_debugfs_fence_info, 0, NULL}, - {"amdgpu_gpu_recover", &amdgpu_debugfs_gpu_recover, 0, NULL} -}; +DEFINE_SHOW_ATTRIBUTE(amdgpu_debugfs_fence_info); +DEFINE_DEBUGFS_ATTRIBUTE(amdgpu_debugfs_gpu_recover_fops, gpu_recover_get, NULL, + "%lld\n"); -static const struct drm_info_list amdgpu_debugfs_fence_list_sriov[] = { - {"amdgpu_fence_info", &amdgpu_debugfs_fence_info, 0, NULL}, -}; #endif -int amdgpu_debugfs_fence_init(struct amdgpu_device *adev) +void amdgpu_debugfs_fence_init(struct amdgpu_device *adev) { #if defined(CONFIG_DEBUG_FS) - if (amdgpu_sriov_vf(adev)) - return amdgpu_debugfs_add_files(adev, amdgpu_debugfs_fence_list_sriov, - ARRAY_SIZE(amdgpu_debugfs_fence_list_sriov)); - return amdgpu_debugfs_add_files(adev, amdgpu_debugfs_fence_list, - ARRAY_SIZE(amdgpu_debugfs_fence_list)); -#else - return 0; + struct drm_minor *minor = adev_to_drm(adev)->primary; + struct dentry *root = minor->debugfs_root; + + debugfs_create_file("amdgpu_fence_info", 0444, root, adev, + &amdgpu_debugfs_fence_info_fops); + + if (!amdgpu_sriov_vf(adev)) + debugfs_create_file("amdgpu_gpu_recover", 0444, root, adev, + &amdgpu_debugfs_gpu_recover_fops); #endif } diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_gart.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_gart.c index 0db933026722..c5a9a4fb10d2 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_gart.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_gart.c @@ -71,7 +71,7 @@ */ static int amdgpu_gart_dummy_page_init(struct amdgpu_device *adev) { - struct page *dummy_page = ttm_bo_glob.dummy_read_page; + struct page *dummy_page = ttm_glob.dummy_read_page; if (adev->dummy_page_addr) return 0; @@ -126,6 +126,8 @@ int amdgpu_gart_table_vram_alloc(struct amdgpu_device *adev) AMDGPU_GEM_CREATE_VRAM_CONTIGUOUS; bp.type = ttm_bo_type_kernel; bp.resv = NULL; + bp.bo_ptr_size = sizeof(struct amdgpu_bo); + r = amdgpu_bo_create(adev, &bp, &adev->gart.bo); if (r) { return r; @@ -202,6 +204,7 @@ void amdgpu_gart_table_vram_free(struct amdgpu_device *adev) return; } amdgpu_bo_unref(&adev->gart.bo); + adev->gart.ptr = NULL; } /* @@ -236,9 +239,6 @@ int amdgpu_gart_unbind(struct amdgpu_device *adev, uint64_t offset, t = offset / AMDGPU_GPU_PAGE_SIZE; p = t / AMDGPU_GPU_PAGES_IN_CPU_PAGE; for (i = 0; i < pages; i++, p++) { -#ifdef CONFIG_DRM_AMDGPU_GART_DEBUGFS - adev->gart.pages[p] = NULL; -#endif page_base = adev->dummy_page_addr; if (!adev->gart.ptr) continue; @@ -312,9 +312,6 @@ int amdgpu_gart_bind(struct amdgpu_device *adev, uint64_t offset, int pages, struct page **pagelist, dma_addr_t *dma_addr, uint64_t flags) { -#ifdef CONFIG_DRM_AMDGPU_GART_DEBUGFS - unsigned t,p; -#endif int r, i; if (!adev->gart.ready) { @@ -322,13 +319,6 @@ int amdgpu_gart_bind(struct amdgpu_device *adev, uint64_t offset, return -EINVAL; } -#ifdef CONFIG_DRM_AMDGPU_GART_DEBUGFS - t = offset / AMDGPU_GPU_PAGE_SIZE; - p = t / AMDGPU_GPU_PAGES_IN_CPU_PAGE; - for (i = 0; i < pages; i++, p++) - adev->gart.pages[p] = pagelist ? pagelist[i] : NULL; -#endif - if (!adev->gart.ptr) return 0; @@ -373,14 +363,6 @@ int amdgpu_gart_init(struct amdgpu_device *adev) DRM_INFO("GART: num cpu pages %u, num gpu pages %u\n", adev->gart.num_cpu_pages, adev->gart.num_gpu_pages); -#ifdef CONFIG_DRM_AMDGPU_GART_DEBUGFS - /* Allocate pages table */ - adev->gart.pages = vzalloc(array_size(sizeof(void *), - adev->gart.num_cpu_pages)); - if (adev->gart.pages == NULL) - return -ENOMEM; -#endif - return 0; } @@ -393,9 +375,5 @@ int amdgpu_gart_init(struct amdgpu_device *adev) */ void amdgpu_gart_fini(struct amdgpu_device *adev) { -#ifdef CONFIG_DRM_AMDGPU_GART_DEBUGFS - vfree(adev->gart.pages); - adev->gart.pages = NULL; -#endif amdgpu_gart_dummy_page_fini(adev); } diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_gart.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_gart.h index afa2e2877d87..a25fe97b0196 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_gart.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_gart.h @@ -46,9 +46,6 @@ struct amdgpu_gart { unsigned num_gpu_pages; unsigned num_cpu_pages; unsigned table_size; -#ifdef CONFIG_DRM_AMDGPU_GART_DEBUGFS - struct page **pages; -#endif bool ready; /* Asic default pte flags */ diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c index b443907afcea..311bcdc59eda 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c @@ -32,7 +32,6 @@ #include <linux/dma-buf.h> #include <drm/amdgpu_drm.h> -#include <drm/drm_debugfs.h> #include <drm/drm_gem_ttm_helper.h> #include "amdgpu.h" @@ -59,6 +58,7 @@ int amdgpu_gem_object_create(struct amdgpu_device *adev, unsigned long size, struct drm_gem_object **obj) { struct amdgpu_bo *bo; + struct amdgpu_bo_user *ubo; struct amdgpu_bo_param bp; int r; @@ -72,10 +72,13 @@ int amdgpu_gem_object_create(struct amdgpu_device *adev, unsigned long size, bp.preferred_domain = initial_domain; bp.flags = flags; bp.domain = initial_domain; - r = amdgpu_bo_create(adev, &bp, &bo); + bp.bo_ptr_size = sizeof(struct amdgpu_bo); + + r = amdgpu_bo_create_user(adev, &bp, &ubo); if (r) return r; + bo = &ubo->bo; *obj = &bo->tbo.base; (*obj)->funcs = &amdgpu_gem_object_funcs; @@ -855,10 +858,10 @@ int amdgpu_mode_dumb_create(struct drm_file *file_priv, } #if defined(CONFIG_DEBUG_FS) -static int amdgpu_debugfs_gem_info(struct seq_file *m, void *data) +static int amdgpu_debugfs_gem_info_show(struct seq_file *m, void *unused) { - struct drm_info_node *node = (struct drm_info_node *)m->private; - struct drm_device *dev = node->minor->dev; + struct amdgpu_device *adev = (struct amdgpu_device *)m->private; + struct drm_device *dev = adev_to_drm(adev); struct drm_file *file; int r; @@ -896,16 +899,17 @@ static int amdgpu_debugfs_gem_info(struct seq_file *m, void *data) return 0; } -static const struct drm_info_list amdgpu_debugfs_gem_list[] = { - {"amdgpu_gem_info", &amdgpu_debugfs_gem_info, 0, NULL}, -}; +DEFINE_SHOW_ATTRIBUTE(amdgpu_debugfs_gem_info); + #endif -int amdgpu_debugfs_gem_init(struct amdgpu_device *adev) +void amdgpu_debugfs_gem_init(struct amdgpu_device *adev) { #if defined(CONFIG_DEBUG_FS) - return amdgpu_debugfs_add_files(adev, amdgpu_debugfs_gem_list, - ARRAY_SIZE(amdgpu_debugfs_gem_list)); + struct drm_minor *minor = adev_to_drm(adev)->primary; + struct dentry *root = minor->debugfs_root; + + debugfs_create_file("amdgpu_gem_info", 0444, root, adev, + &amdgpu_debugfs_gem_info_fops); #endif - return 0; } diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.c index 8e0a6c62322e..95d4f43a03df 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.c @@ -310,9 +310,8 @@ int amdgpu_gfx_kiq_init_ring(struct amdgpu_device *adev, ring->eop_gpu_addr = kiq->eop_gpu_addr; ring->no_scheduler = true; sprintf(ring->name, "kiq_%d.%d.%d", ring->me, ring->pipe, ring->queue); - r = amdgpu_ring_init(adev, ring, 1024, - irq, AMDGPU_CP_KIQ_IRQ_DRIVER0, - AMDGPU_RING_PRIO_DEFAULT); + r = amdgpu_ring_init(adev, ring, 1024, irq, AMDGPU_CP_KIQ_IRQ_DRIVER0, + AMDGPU_RING_PRIO_DEFAULT, NULL); if (r) dev_warn(adev->dev, "(%d) failed to init kiq ring\n", r); @@ -463,20 +462,25 @@ int amdgpu_gfx_disable_kcq(struct amdgpu_device *adev) { struct amdgpu_kiq *kiq = &adev->gfx.kiq; struct amdgpu_ring *kiq_ring = &kiq->ring; - int i; + int i, r; if (!kiq->pmf || !kiq->pmf->kiq_unmap_queues) return -EINVAL; + spin_lock(&adev->gfx.kiq.ring_lock); if (amdgpu_ring_alloc(kiq_ring, kiq->pmf->unmap_queues_size * - adev->gfx.num_compute_rings)) + adev->gfx.num_compute_rings)) { + spin_unlock(&adev->gfx.kiq.ring_lock); return -ENOMEM; + } for (i = 0; i < adev->gfx.num_compute_rings; i++) kiq->pmf->kiq_unmap_queues(kiq_ring, &adev->gfx.compute_ring[i], RESET_QUEUES, 0, 0); + r = amdgpu_ring_test_helper(kiq_ring); + spin_unlock(&adev->gfx.kiq.ring_lock); - return amdgpu_ring_test_helper(kiq_ring); + return r; } int amdgpu_queue_mask_bit_to_set_resource_bit(struct amdgpu_device *adev, @@ -519,12 +523,13 @@ int amdgpu_gfx_enable_kcq(struct amdgpu_device *adev) DRM_INFO("kiq ring mec %d pipe %d q %d\n", kiq_ring->me, kiq_ring->pipe, kiq_ring->queue); - + spin_lock(&adev->gfx.kiq.ring_lock); r = amdgpu_ring_alloc(kiq_ring, kiq->pmf->map_queues_size * adev->gfx.num_compute_rings + kiq->pmf->set_resources_size); if (r) { DRM_ERROR("Failed to lock KIQ (%d).\n", r); + spin_unlock(&adev->gfx.kiq.ring_lock); return r; } @@ -533,6 +538,7 @@ int amdgpu_gfx_enable_kcq(struct amdgpu_device *adev) kiq->pmf->kiq_map_queues(kiq_ring, &adev->gfx.compute_ring[i]); r = amdgpu_ring_test_helper(kiq_ring); + spin_unlock(&adev->gfx.kiq.ring_lock); if (r) DRM_ERROR("KCQ enable failed\n"); @@ -601,6 +607,7 @@ int amdgpu_gfx_ras_late_init(struct amdgpu_device *adev) struct ras_ih_if ih_info = { .cb = amdgpu_gfx_process_ras_data_cb, }; + struct ras_query_if info = { 0 }; if (!adev->gfx.ras_if) { adev->gfx.ras_if = kmalloc(sizeof(struct ras_common_if), GFP_KERNEL); @@ -612,13 +619,19 @@ int amdgpu_gfx_ras_late_init(struct amdgpu_device *adev) strcpy(adev->gfx.ras_if->name, "gfx"); } fs_info.head = ih_info.head = *adev->gfx.ras_if; - r = amdgpu_ras_late_init(adev, adev->gfx.ras_if, &fs_info, &ih_info); if (r) goto free; if (amdgpu_ras_is_supported(adev, adev->gfx.ras_if->block)) { + if (adev->gmc.xgmi.connected_to_cpu) { + info.head = *adev->gfx.ras_if; + amdgpu_ras_query_error_status(adev, &info); + } else { + amdgpu_ras_reset_error_status(adev, AMDGPU_RAS_BLOCK__GFX); + } + r = amdgpu_irq_get(adev, &adev->gfx.cp_ecc_error_irq, 0); if (r) goto late_fini; @@ -664,8 +677,9 @@ int amdgpu_gfx_process_ras_data_cb(struct amdgpu_device *adev, */ if (!amdgpu_ras_is_supported(adev, AMDGPU_RAS_BLOCK__GFX)) { kgd2kfd_set_sram_ecc_flag(adev->kfd.dev); - if (adev->gfx.funcs->query_ras_error_count) - adev->gfx.funcs->query_ras_error_count(adev, err_data); + if (adev->gfx.ras_funcs && + adev->gfx.ras_funcs->query_ras_error_count) + adev->gfx.ras_funcs->query_ras_error_count(adev, err_data); amdgpu_ras_reset_gpu(adev); } return AMDGPU_RAS_SUCCESS; @@ -698,7 +712,7 @@ uint32_t amdgpu_kiq_rreg(struct amdgpu_device *adev, uint32_t reg) struct amdgpu_kiq *kiq = &adev->gfx.kiq; struct amdgpu_ring *ring = &kiq->ring; - if (adev->in_pci_err_recovery) + if (amdgpu_device_skip_hw_access(adev)) return 0; BUG_ON(!ring->funcs->emit_rreg); @@ -765,7 +779,7 @@ void amdgpu_kiq_wreg(struct amdgpu_device *adev, uint32_t reg, uint32_t v) BUG_ON(!ring->funcs->emit_wreg); - if (adev->in_pci_err_recovery) + if (amdgpu_device_skip_hw_access(adev)) return; spin_lock_irqsave(&kiq->ring_lock, flags); @@ -829,14 +843,10 @@ int amdgpu_gfx_get_num_kcq(struct amdgpu_device *adev) void amdgpu_gfx_state_change_set(struct amdgpu_device *adev, enum gfx_change_state state) { - if (is_support_sw_smu(adev)) { - smu_gfx_state_change_set(&adev->smu, state); - } else { - mutex_lock(&adev->pm.mutex); - if (adev->powerplay.pp_funcs && - adev->powerplay.pp_funcs->gfx_state_change_set) - ((adev)->powerplay.pp_funcs->gfx_state_change_set( - (adev)->powerplay.pp_handle, state)); - mutex_unlock(&adev->pm.mutex); - } + mutex_lock(&adev->pm.mutex); + if (adev->powerplay.pp_funcs && + adev->powerplay.pp_funcs->gfx_state_change_set) + ((adev)->powerplay.pp_funcs->gfx_state_change_set( + (adev)->powerplay.pp_handle, state)); + mutex_unlock(&adev->pm.mutex); } diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.h index 72dbcd2bc6a6..d43fe2ed8116 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.h @@ -30,6 +30,7 @@ #include "clearstate_defs.h" #include "amdgpu_ring.h" #include "amdgpu_rlc.h" +#include "soc15.h" /* GFX current status */ #define AMDGPU_GFX_NORMAL_MODE 0x00000000L @@ -204,6 +205,19 @@ struct amdgpu_cu_info { uint32_t bitmap[4][4]; }; +struct amdgpu_gfx_ras_funcs { + int (*ras_late_init)(struct amdgpu_device *adev); + void (*ras_fini)(struct amdgpu_device *adev); + int (*ras_error_inject)(struct amdgpu_device *adev, + void *inject_if); + int (*query_ras_error_count)(struct amdgpu_device *adev, + void *ras_error_status); + void (*reset_ras_error_count)(struct amdgpu_device *adev); + void (*query_ras_error_status)(struct amdgpu_device *adev); + void (*reset_ras_error_status)(struct amdgpu_device *adev); + void (*enable_watchdog_timer)(struct amdgpu_device *adev); +}; + struct amdgpu_gfx_funcs { /* get the gpu clock counter */ uint64_t (*get_gpu_clock_counter)(struct amdgpu_device *adev); @@ -219,11 +233,7 @@ struct amdgpu_gfx_funcs { uint32_t *dst); void (*select_me_pipe_q)(struct amdgpu_device *adev, u32 me, u32 pipe, u32 queue, u32 vmid); - int (*ras_error_inject)(struct amdgpu_device *adev, void *inject_if); - int (*query_ras_error_count) (struct amdgpu_device *adev, void *ras_error_status); - void (*reset_ras_error_count) (struct amdgpu_device *adev); void (*init_spm_golden)(struct amdgpu_device *adev); - void (*query_ras_error_status) (struct amdgpu_device *adev); void (*update_perfmon_mgcg)(struct amdgpu_device *adev, bool enable); }; @@ -327,7 +337,8 @@ struct amdgpu_gfx { DECLARE_BITMAP (pipe_reserve_bitmap, AMDGPU_MAX_COMPUTE_QUEUES); /*ras */ - struct ras_common_if *ras_if; + struct ras_common_if *ras_if; + const struct amdgpu_gfx_ras_funcs *ras_funcs; }; #define amdgpu_gfx_get_gpu_clock_counter(adev) (adev)->gfx.funcs->get_gpu_clock_counter((adev)) diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_gmc.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_gmc.c index fe1a39ffda72..c39ed9eb0987 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_gmc.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_gmc.c @@ -32,6 +32,59 @@ #include "amdgpu_xgmi.h" /** + * amdgpu_gmc_pdb0_alloc - allocate vram for pdb0 + * + * @adev: amdgpu_device pointer + * + * Allocate video memory for pdb0 and map it for CPU access + * Returns 0 for success, error for failure. + */ +int amdgpu_gmc_pdb0_alloc(struct amdgpu_device *adev) +{ + int r; + struct amdgpu_bo_param bp; + u64 vram_size = adev->gmc.xgmi.node_segment_size * adev->gmc.xgmi.num_physical_nodes; + uint32_t pde0_page_shift = adev->gmc.vmid0_page_table_block_size + 21; + uint32_t npdes = (vram_size + (1ULL << pde0_page_shift) -1) >> pde0_page_shift; + + memset(&bp, 0, sizeof(bp)); + bp.size = PAGE_ALIGN((npdes + 1) * 8); + bp.byte_align = PAGE_SIZE; + bp.domain = AMDGPU_GEM_DOMAIN_VRAM; + bp.flags = AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED | + AMDGPU_GEM_CREATE_VRAM_CONTIGUOUS; + bp.type = ttm_bo_type_kernel; + bp.resv = NULL; + bp.bo_ptr_size = sizeof(struct amdgpu_bo); + + r = amdgpu_bo_create(adev, &bp, &adev->gmc.pdb0_bo); + if (r) + return r; + + r = amdgpu_bo_reserve(adev->gmc.pdb0_bo, false); + if (unlikely(r != 0)) + goto bo_reserve_failure; + + r = amdgpu_bo_pin(adev->gmc.pdb0_bo, AMDGPU_GEM_DOMAIN_VRAM); + if (r) + goto bo_pin_failure; + r = amdgpu_bo_kmap(adev->gmc.pdb0_bo, &adev->gmc.ptr_pdb0); + if (r) + goto bo_kmap_failure; + + amdgpu_bo_unreserve(adev->gmc.pdb0_bo); + return 0; + +bo_kmap_failure: + amdgpu_bo_unpin(adev->gmc.pdb0_bo); +bo_pin_failure: + amdgpu_bo_unreserve(adev->gmc.pdb0_bo); +bo_reserve_failure: + amdgpu_bo_unref(&adev->gmc.pdb0_bo); + return r; +} + +/** * amdgpu_gmc_get_pde_for_bo - get the PDE for a BO * * @bo: the BO to get the PDE for @@ -158,6 +211,39 @@ void amdgpu_gmc_vram_location(struct amdgpu_device *adev, struct amdgpu_gmc *mc, mc->vram_end, mc->real_vram_size >> 20); } +/** amdgpu_gmc_sysvm_location - place vram and gart in sysvm aperture + * + * @adev: amdgpu device structure holding all necessary information + * @mc: memory controller structure holding memory information + * + * This function is only used if use GART for FB translation. In such + * case, we use sysvm aperture (vmid0 page tables) for both vram + * and gart (aka system memory) access. + * + * GPUVM (and our organization of vmid0 page tables) require sysvm + * aperture to be placed at a location aligned with 8 times of native + * page size. For example, if vm_context0_cntl.page_table_block_size + * is 12, then native page size is 8G (2M*2^12), sysvm should start + * with a 64G aligned address. For simplicity, we just put sysvm at + * address 0. So vram start at address 0 and gart is right after vram. + */ +void amdgpu_gmc_sysvm_location(struct amdgpu_device *adev, struct amdgpu_gmc *mc) +{ + u64 hive_vram_start = 0; + u64 hive_vram_end = mc->xgmi.node_segment_size * mc->xgmi.num_physical_nodes - 1; + mc->vram_start = mc->xgmi.node_segment_size * mc->xgmi.physical_node_id; + mc->vram_end = mc->vram_start + mc->xgmi.node_segment_size - 1; + mc->gart_start = hive_vram_end + 1; + mc->gart_end = mc->gart_start + mc->gart_size - 1; + mc->fb_start = hive_vram_start; + mc->fb_end = hive_vram_end; + dev_info(adev->dev, "VRAM: %lluM 0x%016llX - 0x%016llX (%lluM used)\n", + mc->mc_vram_size >> 20, mc->vram_start, + mc->vram_end, mc->real_vram_size >> 20); + dev_info(adev->dev, "GART: %lluM 0x%016llX - 0x%016llX\n", + mc->gart_size >> 20, mc->gart_start, mc->gart_end); +} + /** * amdgpu_gmc_gart_location - try to find GART location * @@ -165,7 +251,6 @@ void amdgpu_gmc_vram_location(struct amdgpu_device *adev, struct amdgpu_gmc *mc, * @mc: memory controller structure holding memory information * * Function will place try to place GART before or after VRAM. - * * If GART size is bigger than space left then we ajust GART size. * Thus function will never fails. */ @@ -176,8 +261,6 @@ void amdgpu_gmc_gart_location(struct amdgpu_device *adev, struct amdgpu_gmc *mc) /*To avoid the hole, limit the max mc address to AMDGPU_GMC_HOLE_START*/ u64 max_mc_address = min(adev->gmc.mc_mask, AMDGPU_GMC_HOLE_START - 1); - mc->gart_size += adev->pm.smu_prv_buffer_size; - /* VCE doesn't like it when BOs cross a 4GB segment, so align * the GART base on a 4GB boundary as well. */ @@ -308,26 +391,46 @@ int amdgpu_gmc_ras_late_init(struct amdgpu_device *adev) { int r; - if (adev->umc.funcs && adev->umc.funcs->ras_late_init) { - r = adev->umc.funcs->ras_late_init(adev); + if (adev->umc.ras_funcs && + adev->umc.ras_funcs->ras_late_init) { + r = adev->umc.ras_funcs->ras_late_init(adev); if (r) return r; } - if (adev->mmhub.funcs && adev->mmhub.funcs->ras_late_init) { - r = adev->mmhub.funcs->ras_late_init(adev); + if (adev->mmhub.ras_funcs && + adev->mmhub.ras_funcs->ras_late_init) { + r = adev->mmhub.ras_funcs->ras_late_init(adev); if (r) return r; } - return amdgpu_xgmi_ras_late_init(adev); + if (!adev->gmc.xgmi.connected_to_cpu) + adev->gmc.xgmi.ras_funcs = &xgmi_ras_funcs; + + if (adev->gmc.xgmi.ras_funcs && + adev->gmc.xgmi.ras_funcs->ras_late_init) { + r = adev->gmc.xgmi.ras_funcs->ras_late_init(adev); + if (r) + return r; + } + + return 0; } void amdgpu_gmc_ras_fini(struct amdgpu_device *adev) { - amdgpu_umc_ras_fini(adev); - amdgpu_mmhub_ras_fini(adev); - amdgpu_xgmi_ras_fini(adev); + if (adev->umc.ras_funcs && + adev->umc.ras_funcs->ras_fini) + adev->umc.ras_funcs->ras_fini(adev); + + if (adev->mmhub.ras_funcs && + adev->mmhub.ras_funcs->ras_fini) + amdgpu_mmhub_ras_fini(adev); + + if (adev->gmc.xgmi.ras_funcs && + adev->gmc.xgmi.ras_funcs->ras_fini) + adev->gmc.xgmi.ras_funcs->ras_fini(adev); } /* @@ -385,6 +488,16 @@ void amdgpu_gmc_tmz_set(struct amdgpu_device *adev) switch (adev->asic_type) { case CHIP_RAVEN: case CHIP_RENOIR: + if (amdgpu_tmz == 0) { + adev->gmc.tmz_enabled = false; + dev_info(adev->dev, + "Trusted Memory Zone (TMZ) feature disabled (cmd line)\n"); + } else { + adev->gmc.tmz_enabled = true; + dev_info(adev->dev, + "Trusted Memory Zone (TMZ) feature enabled\n"); + } + break; case CHIP_NAVI10: case CHIP_NAVI14: case CHIP_NAVI12: @@ -423,6 +536,8 @@ void amdgpu_gmc_noretry_set(struct amdgpu_device *adev) switch (adev->asic_type) { case CHIP_VEGA10: case CHIP_VEGA20: + case CHIP_ARCTURUS: + case CHIP_ALDEBARAN: /* * noretry = 0 will cause kfd page fault tests fail * for some ASICs, so set default to 1 for these ASICs. @@ -518,3 +633,90 @@ void amdgpu_gmc_get_vbios_allocations(struct amdgpu_device *adev) adev->mman.stolen_extended_size = 0; } } + +/** + * amdgpu_gmc_init_pdb0 - initialize PDB0 + * + * @adev: amdgpu_device pointer + * + * This function is only used when GART page table is used + * for FB address translatioin. In such a case, we construct + * a 2-level system VM page table: PDB0->PTB, to cover both + * VRAM of the hive and system memory. + * + * PDB0 is static, initialized once on driver initialization. + * The first n entries of PDB0 are used as PTE by setting + * P bit to 1, pointing to VRAM. The n+1'th entry points + * to a big PTB covering system memory. + * + */ +void amdgpu_gmc_init_pdb0(struct amdgpu_device *adev) +{ + int i; + uint64_t flags = adev->gart.gart_pte_flags; //TODO it is UC. explore NC/RW? + /* Each PDE0 (used as PTE) covers (2^vmid0_page_table_block_size)*2M + */ + u64 vram_size = adev->gmc.xgmi.node_segment_size * adev->gmc.xgmi.num_physical_nodes; + u64 pde0_page_size = (1ULL<<adev->gmc.vmid0_page_table_block_size)<<21; + u64 vram_addr = adev->vm_manager.vram_base_offset - + adev->gmc.xgmi.physical_node_id * adev->gmc.xgmi.node_segment_size; + u64 vram_end = vram_addr + vram_size; + u64 gart_ptb_gpu_pa = amdgpu_gmc_vram_pa(adev, adev->gart.bo); + + flags |= AMDGPU_PTE_VALID | AMDGPU_PTE_READABLE; + flags |= AMDGPU_PTE_WRITEABLE; + flags |= AMDGPU_PTE_SNOOPED; + flags |= AMDGPU_PTE_FRAG((adev->gmc.vmid0_page_table_block_size + 9*1)); + flags |= AMDGPU_PDE_PTE; + + /* The first n PDE0 entries are used as PTE, + * pointing to vram + */ + for (i = 0; vram_addr < vram_end; i++, vram_addr += pde0_page_size) + amdgpu_gmc_set_pte_pde(adev, adev->gmc.ptr_pdb0, i, vram_addr, flags); + + /* The n+1'th PDE0 entry points to a huge + * PTB who has more than 512 entries each + * pointing to a 4K system page + */ + flags = AMDGPU_PTE_VALID; + flags |= AMDGPU_PDE_BFS(0) | AMDGPU_PTE_SNOOPED; + /* Requires gart_ptb_gpu_pa to be 4K aligned */ + amdgpu_gmc_set_pte_pde(adev, adev->gmc.ptr_pdb0, i, gart_ptb_gpu_pa, flags); +} + +/** + * amdgpu_gmc_vram_mc2pa - calculate vram buffer's physical address from MC + * address + * + * @adev: amdgpu_device pointer + * @mc_addr: MC address of buffer + */ +uint64_t amdgpu_gmc_vram_mc2pa(struct amdgpu_device *adev, uint64_t mc_addr) +{ + return mc_addr - adev->gmc.vram_start + adev->vm_manager.vram_base_offset; +} + +/** + * amdgpu_gmc_vram_pa - calculate vram buffer object's physical address from + * GPU's view + * + * @adev: amdgpu_device pointer + * @bo: amdgpu buffer object + */ +uint64_t amdgpu_gmc_vram_pa(struct amdgpu_device *adev, struct amdgpu_bo *bo) +{ + return amdgpu_gmc_vram_mc2pa(adev, amdgpu_bo_gpu_offset(bo)); +} + +/** + * amdgpu_gmc_vram_cpu_pa - calculate vram buffer object's physical address + * from CPU's view + * + * @adev: amdgpu_device pointer + * @bo: amdgpu buffer object + */ +uint64_t amdgpu_gmc_vram_cpu_pa(struct amdgpu_device *adev, struct amdgpu_bo *bo) +{ + return amdgpu_bo_gpu_offset(bo) - adev->gmc.vram_start + adev->gmc.aper_base; +} diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_gmc.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_gmc.h index aa0c83776ce0..9d11c02a3938 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_gmc.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_gmc.h @@ -135,6 +135,14 @@ struct amdgpu_gmc_funcs { unsigned int (*get_vbios_fb_size)(struct amdgpu_device *adev); }; +struct amdgpu_xgmi_ras_funcs { + int (*ras_late_init)(struct amdgpu_device *adev); + void (*ras_fini)(struct amdgpu_device *adev); + int (*query_ras_error_count)(struct amdgpu_device *adev, + void *ras_error_status); + void (*reset_ras_error_count)(struct amdgpu_device *adev); +}; + struct amdgpu_xgmi { /* from psp */ u64 node_id; @@ -149,6 +157,9 @@ struct amdgpu_xgmi { struct list_head head; bool supported; struct ras_common_if *ras_if; + bool connected_to_cpu; + bool pending_reset; + const struct amdgpu_xgmi_ras_funcs *ras_funcs; }; struct amdgpu_gmc { @@ -189,10 +200,13 @@ struct amdgpu_gmc { u64 gart_end; /* Frame buffer aperture of this GPU device. Different from * fb_start (see below), this only covers the local GPU device. - * Driver get fb_start from MC_VM_FB_LOCATION_BASE (set by vbios) - * and calculate vram_start of this local device by adding an - * offset inside the XGMI hive. - * Under VMID0, logical address == MC address + * If driver uses FB aperture to access FB, driver get fb_start from + * MC_VM_FB_LOCATION_BASE (set by vbios) and calculate vram_start + * of this local device by adding an offset inside the XGMI hive. + * If driver uses GART table for VMID0 FB access, driver finds a hole in + * VMID0's virtual address space to place the SYSVM aperture inside + * which the first part is vram and the second part is gart (covering + * system ram). */ u64 vram_start; u64 vram_end; @@ -240,6 +254,12 @@ struct amdgpu_gmc { struct amdgpu_xgmi xgmi; struct amdgpu_irq_src ecc_irq; int noretry; + + uint32_t vmid0_page_table_block_size; + uint32_t vmid0_page_table_depth; + struct amdgpu_bo *pdb0_bo; + /* CPU kmapped address of pdb0*/ + void *ptr_pdb0; }; #define amdgpu_gmc_flush_gpu_tlb(adev, vmid, vmhub, type) ((adev)->gmc.gmc_funcs->flush_gpu_tlb((adev), (vmid), (vmhub), (type))) @@ -281,6 +301,7 @@ static inline uint64_t amdgpu_gmc_sign_extend(uint64_t addr) return addr; } +int amdgpu_gmc_pdb0_alloc(struct amdgpu_device *adev); void amdgpu_gmc_get_pde_for_bo(struct amdgpu_bo *bo, int level, uint64_t *addr, uint64_t *flags); int amdgpu_gmc_set_pte_pde(struct amdgpu_device *adev, void *cpu_pt_addr, @@ -288,6 +309,7 @@ int amdgpu_gmc_set_pte_pde(struct amdgpu_device *adev, void *cpu_pt_addr, uint64_t flags); uint64_t amdgpu_gmc_pd_addr(struct amdgpu_bo *bo); uint64_t amdgpu_gmc_agp_addr(struct ttm_buffer_object *bo); +void amdgpu_gmc_sysvm_location(struct amdgpu_device *adev, struct amdgpu_gmc *mc); void amdgpu_gmc_vram_location(struct amdgpu_device *adev, struct amdgpu_gmc *mc, u64 base); void amdgpu_gmc_gart_location(struct amdgpu_device *adev, @@ -309,4 +331,8 @@ amdgpu_gmc_set_vm_fault_masks(struct amdgpu_device *adev, int hub_type, void amdgpu_gmc_get_vbios_allocations(struct amdgpu_device *adev); +void amdgpu_gmc_init_pdb0(struct amdgpu_device *adev); +uint64_t amdgpu_gmc_vram_mc2pa(struct amdgpu_device *adev, uint64_t mc_addr); +uint64_t amdgpu_gmc_vram_pa(struct amdgpu_device *adev, struct amdgpu_bo *bo); +uint64_t amdgpu_gmc_vram_cpu_pa(struct amdgpu_device *adev, struct amdgpu_bo *bo); #endif diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_gtt_mgr.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_gtt_mgr.c index 8980329cded0..540c01052b21 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_gtt_mgr.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_gtt_mgr.c @@ -49,8 +49,7 @@ static ssize_t amdgpu_mem_info_gtt_total_show(struct device *dev, struct amdgpu_device *adev = drm_to_adev(ddev); struct ttm_resource_manager *man = ttm_manager_type(&adev->mman.bdev, TTM_PL_TT); - return snprintf(buf, PAGE_SIZE, "%llu\n", - man->size * PAGE_SIZE); + return sysfs_emit(buf, "%llu\n", man->size * PAGE_SIZE); } /** @@ -68,8 +67,7 @@ static ssize_t amdgpu_mem_info_gtt_used_show(struct device *dev, struct amdgpu_device *adev = drm_to_adev(ddev); struct ttm_resource_manager *man = ttm_manager_type(&adev->mman.bdev, TTM_PL_TT); - return snprintf(buf, PAGE_SIZE, "%llu\n", - amdgpu_gtt_mgr_usage(man)); + return sysfs_emit(buf, "%llu\n", amdgpu_gtt_mgr_usage(man)); } static DEVICE_ATTR(mem_info_gtt_total, S_IRUGO, diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ib.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ib.c index 7645223ea0ef..a2fe2dac32c1 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ib.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ib.c @@ -30,7 +30,6 @@ #include <linux/slab.h> #include <drm/amdgpu_drm.h> -#include <drm/drm_debugfs.h> #include "amdgpu.h" #include "atom.h" @@ -77,6 +76,8 @@ int amdgpu_ib_get(struct amdgpu_device *adev, struct amdgpu_vm *vm, } ib->ptr = amdgpu_sa_bo_cpu_addr(ib->sa_bo); + /* flush the cache before commit the IB */ + ib->flags = AMDGPU_IB_FLAG_EMIT_MEM_SYNC; if (!vm) ib->gpu_addr = amdgpu_sa_bo_gpu_addr(ib->sa_bo); @@ -453,11 +454,9 @@ int amdgpu_ib_ring_tests(struct amdgpu_device *adev) */ #if defined(CONFIG_DEBUG_FS) -static int amdgpu_debugfs_sa_info(struct seq_file *m, void *data) +static int amdgpu_debugfs_sa_info_show(struct seq_file *m, void *unused) { - struct drm_info_node *node = (struct drm_info_node *) m->private; - struct drm_device *dev = node->minor->dev; - struct amdgpu_device *adev = drm_to_adev(dev); + struct amdgpu_device *adev = (struct amdgpu_device *)m->private; seq_printf(m, "--------------------- DELAYED --------------------- \n"); amdgpu_sa_bo_dump_debug_info(&adev->ib_pools[AMDGPU_IB_POOL_DELAYED], @@ -471,18 +470,18 @@ static int amdgpu_debugfs_sa_info(struct seq_file *m, void *data) return 0; } -static const struct drm_info_list amdgpu_debugfs_sa_list[] = { - {"amdgpu_sa_info", &amdgpu_debugfs_sa_info, 0, NULL}, -}; +DEFINE_SHOW_ATTRIBUTE(amdgpu_debugfs_sa_info); #endif -int amdgpu_debugfs_sa_init(struct amdgpu_device *adev) +void amdgpu_debugfs_sa_init(struct amdgpu_device *adev) { #if defined(CONFIG_DEBUG_FS) - return amdgpu_debugfs_add_files(adev, amdgpu_debugfs_sa_list, - ARRAY_SIZE(amdgpu_debugfs_sa_list)); -#else - return 0; + struct drm_minor *minor = adev_to_drm(adev)->primary; + struct dentry *root = minor->debugfs_root; + + debugfs_create_file("amdgpu_sa_info", 0444, root, adev, + &amdgpu_debugfs_sa_info_fops); + #endif } diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ids.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ids.c index 94b069630db3..b4971e90b98c 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ids.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ids.c @@ -215,7 +215,11 @@ static int amdgpu_vmid_grab_idle(struct amdgpu_vm *vm, /* Check if we have an idle VMID */ i = 0; list_for_each_entry((*idle), &id_mgr->ids_lru, list) { - fences[i] = amdgpu_sync_peek_fence(&(*idle)->active, ring); + /* Don't use per engine and per process VMID at the same time */ + struct amdgpu_ring *r = adev->vm_manager.concurrent_flush ? + NULL : ring; + + fences[i] = amdgpu_sync_peek_fence(&(*idle)->active, r); if (!fences[i]) break; ++i; @@ -281,7 +285,7 @@ static int amdgpu_vmid_grab_reserved(struct amdgpu_vm *vm, if (updates && (*id)->flushed_updates && updates->context == (*id)->flushed_updates->context && !dma_fence_is_later(updates, (*id)->flushed_updates)) - updates = NULL; + updates = NULL; if ((*id)->owner != vm->immediate.fence_context || job->vm_pd_addr != (*id)->pd_gpu_addr || @@ -290,6 +294,10 @@ static int amdgpu_vmid_grab_reserved(struct amdgpu_vm *vm, !dma_fence_is_signaled((*id)->last_flush))) { struct dma_fence *tmp; + /* Don't use per engine and per process VMID at the same time */ + if (adev->vm_manager.concurrent_flush) + ring = NULL; + /* to prevent one context starved by another context */ (*id)->pd_gpu_addr = 0; tmp = amdgpu_sync_peek_fence(&(*id)->active, ring); @@ -365,12 +373,7 @@ static int amdgpu_vmid_grab_used(struct amdgpu_vm *vm, if (updates && (!flushed || dma_fence_is_later(updates, flushed))) needs_flush = true; - /* Concurrent flushes are only possible starting with Vega10 and - * are broken on Navi10 and Navi14. - */ - if (needs_flush && (adev->asic_type < CHIP_VEGA10 || - adev->asic_type == CHIP_NAVI10 || - adev->asic_type == CHIP_NAVI14)) + if (needs_flush && !adev->vm_manager.concurrent_flush) continue; /* Good, we can use this VMID. Remember this submission as diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ih.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ih.c index dc852af4f3b7..faaa6aa2faaf 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ih.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ih.c @@ -99,6 +99,8 @@ int amdgpu_ih_ring_init(struct amdgpu_device *adev, struct amdgpu_ih_ring *ih, ih->rptr_addr = adev->wb.gpu_addr + rptr_offs * 4; ih->rptr_cpu = &adev->wb.wb[rptr_offs]; } + + init_waitqueue_head(&ih->wait_process); return 0; } @@ -160,6 +162,52 @@ void amdgpu_ih_ring_write(struct amdgpu_ih_ring *ih, const uint32_t *iv, } } +/* Waiter helper that checks current rptr matches or passes checkpoint wptr */ +static bool amdgpu_ih_has_checkpoint_processed(struct amdgpu_device *adev, + struct amdgpu_ih_ring *ih, + uint32_t checkpoint_wptr, + uint32_t *prev_rptr) +{ + uint32_t cur_rptr = ih->rptr | (*prev_rptr & ~ih->ptr_mask); + + /* rptr has wrapped. */ + if (cur_rptr < *prev_rptr) + cur_rptr += ih->ptr_mask + 1; + *prev_rptr = cur_rptr; + + return cur_rptr >= checkpoint_wptr; +} + +/** + * amdgpu_ih_wait_on_checkpoint_process - wait to process IVs up to checkpoint + * + * @adev: amdgpu_device pointer + * @ih: ih ring to process + * + * Used to ensure ring has processed IVs up to the checkpoint write pointer. + */ +int amdgpu_ih_wait_on_checkpoint_process(struct amdgpu_device *adev, + struct amdgpu_ih_ring *ih) +{ + uint32_t checkpoint_wptr, rptr; + + if (!ih->enabled || adev->shutdown) + return -ENODEV; + + checkpoint_wptr = amdgpu_ih_get_wptr(adev, ih); + /* Order wptr with rptr. */ + rmb(); + rptr = READ_ONCE(ih->rptr); + + /* wptr has wrapped. */ + if (rptr > checkpoint_wptr) + checkpoint_wptr += ih->ptr_mask + 1; + + return wait_event_interruptible(ih->wait_process, + amdgpu_ih_has_checkpoint_processed(adev, ih, + checkpoint_wptr, &rptr)); +} + /** * amdgpu_ih_process - interrupt handler * @@ -180,10 +228,6 @@ int amdgpu_ih_process(struct amdgpu_device *adev, struct amdgpu_ih_ring *ih) wptr = amdgpu_ih_get_wptr(adev, ih); restart_ih: - /* is somebody else already processing irqs? */ - if (atomic_xchg(&ih->lock, 1)) - return IRQ_NONE; - DRM_DEBUG("%s: rptr %d, wptr %d\n", __func__, ih->rptr, wptr); /* Order reading of wptr vs. reading of IH ring data */ @@ -195,7 +239,7 @@ restart_ih: } amdgpu_ih_set_rptr(adev, ih); - atomic_set(&ih->lock, 0); + wake_up_all(&ih->wait_process); /* make sure wptr hasn't changed while processing */ wptr = amdgpu_ih_get_wptr(adev, ih); diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ih.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_ih.h index 6ed4a85fc7c3..0649b59830a5 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ih.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ih.h @@ -64,8 +64,10 @@ struct amdgpu_ih_ring { bool enabled; unsigned rptr; - atomic_t lock; struct amdgpu_ih_regs ih_regs; + + /* For waiting on IH processing at checkpoint. */ + wait_queue_head_t wait_process; }; /* provided by the ih block */ @@ -87,6 +89,8 @@ int amdgpu_ih_ring_init(struct amdgpu_device *adev, struct amdgpu_ih_ring *ih, void amdgpu_ih_ring_fini(struct amdgpu_device *adev, struct amdgpu_ih_ring *ih); void amdgpu_ih_ring_write(struct amdgpu_ih_ring *ih, const uint32_t *iv, unsigned int num_dw); +int amdgpu_ih_wait_on_checkpoint_process(struct amdgpu_device *adev, + struct amdgpu_ih_ring *ih); int amdgpu_ih_process(struct amdgpu_device *adev, struct amdgpu_ih_ring *ih); void amdgpu_ih_decode_iv_helper(struct amdgpu_device *adev, struct amdgpu_ih_ring *ih, diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_irq.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_irq.c index afbbec82a289..90f50561b43a 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_irq.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_irq.c @@ -65,6 +65,41 @@ #define AMDGPU_WAIT_IDLE_TIMEOUT 200 +const char *soc15_ih_clientid_name[] = { + "IH", + "SDMA2 or ACP", + "ATHUB", + "BIF", + "SDMA3 or DCE", + "SDMA4 or ISP", + "VMC1 or PCIE0", + "RLC", + "SDMA0", + "SDMA1", + "SE0SH", + "SE1SH", + "SE2SH", + "SE3SH", + "VCN1 or UVD1", + "THM", + "VCN or UVD", + "SDMA5 or VCE0", + "VMC", + "SDMA6 or XDMA", + "GRBM_CP", + "ATS", + "ROM_SMUIO", + "DF", + "SDMA7 or VCE1", + "PWR", + "reserved", + "UTCL2", + "EA", + "UTCL2LOG", + "MP0", + "MP1" +}; + /** * amdgpu_hotplug_work_func - work handler for display hotplug event * @@ -164,13 +199,13 @@ irqreturn_t amdgpu_irq_handler(int irq, void *arg) * ack the interrupt if it is there */ if (amdgpu_ras_is_supported(adev, AMDGPU_RAS_BLOCK__PCIE_BIF)) { - if (adev->nbio.funcs && - adev->nbio.funcs->handle_ras_controller_intr_no_bifring) - adev->nbio.funcs->handle_ras_controller_intr_no_bifring(adev); + if (adev->nbio.ras_funcs && + adev->nbio.ras_funcs->handle_ras_controller_intr_no_bifring) + adev->nbio.ras_funcs->handle_ras_controller_intr_no_bifring(adev); - if (adev->nbio.funcs && - adev->nbio.funcs->handle_ras_err_event_athub_intr_no_bifring) - adev->nbio.funcs->handle_ras_err_event_athub_intr_no_bifring(adev); + if (adev->nbio.ras_funcs && + adev->nbio.ras_funcs->handle_ras_err_event_athub_intr_no_bifring) + adev->nbio.ras_funcs->handle_ras_err_event_athub_intr_no_bifring(adev); } return ret; @@ -347,11 +382,6 @@ void amdgpu_irq_fini(struct amdgpu_device *adev) kfree(src->enabled_types); src->enabled_types = NULL; - if (src->data) { - kfree(src->data); - kfree(src); - adev->irq.client[i].sources[j] = NULL; - } } kfree(adev->irq.client[i].sources); adev->irq.client[i].sources = NULL; @@ -535,7 +565,7 @@ void amdgpu_irq_gpu_reset_resume_helper(struct amdgpu_device *adev) for (j = 0; j < AMDGPU_MAX_IRQ_SRC_ID; ++j) { struct amdgpu_irq_src *src = adev->irq.client[i].sources[j]; - if (!src) + if (!src || !src->funcs || !src->funcs->set) continue; for (k = 0; k < src->num_types; k++) amdgpu_irq_update(adev, src, k); diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_irq.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_irq.h index ac527e5deae6..cf6116648322 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_irq.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_irq.h @@ -62,7 +62,6 @@ struct amdgpu_irq_src { unsigned num_types; atomic_t *enabled_types; const struct amdgpu_irq_src_funcs *funcs; - void *data; }; struct amdgpu_irq_client { diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_job.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_job.c index ff48101bab55..759b34799221 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_job.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_job.c @@ -28,7 +28,7 @@ #include "amdgpu.h" #include "amdgpu_trace.h" -static void amdgpu_job_timedout(struct drm_sched_job *s_job) +static enum drm_gpu_sched_stat amdgpu_job_timedout(struct drm_sched_job *s_job) { struct amdgpu_ring *ring = to_amdgpu_ring(s_job->sched); struct amdgpu_job *job = to_amdgpu_job(s_job); @@ -41,7 +41,7 @@ static void amdgpu_job_timedout(struct drm_sched_job *s_job) amdgpu_ring_soft_recovery(ring, job->vmid, s_job->s_fence->parent)) { DRM_ERROR("ring %s timeout, but soft recovered\n", s_job->sched->name); - return; + return DRM_GPU_SCHED_STAT_NOMINAL; } amdgpu_vm_get_task_info(ring->adev, job->pasid, &ti); @@ -53,10 +53,12 @@ static void amdgpu_job_timedout(struct drm_sched_job *s_job) if (amdgpu_device_should_recover_gpu(ring->adev)) { amdgpu_device_gpu_recover(ring->adev, job); + return DRM_GPU_SCHED_STAT_NOMINAL; } else { drm_sched_suspend_timeout(&ring->sched); if (amdgpu_sriov_vf(adev)) adev->virt.tdr_debug = true; + return DRM_GPU_SCHED_STAT_NOMINAL; } } diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c index a4e2cf7cada1..39ee88d29cca 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c @@ -27,7 +27,6 @@ */ #include "amdgpu.h" -#include <drm/drm_debugfs.h> #include <drm/amdgpu_drm.h> #include "amdgpu_uvd.h" #include "amdgpu_vce.h" @@ -160,7 +159,7 @@ int amdgpu_driver_load_kms(struct amdgpu_device *adev, unsigned long flags) goto out; } - if (amdgpu_device_supports_atpx(dev) && + if (amdgpu_device_supports_px(dev) && (amdgpu_runtime_pm != 0)) { /* enable runpm by default for atpx */ adev->runpm = true; dev_info(adev->dev, "Using ATPX for runtime pm\n"); @@ -201,9 +200,13 @@ int amdgpu_driver_load_kms(struct amdgpu_device *adev, unsigned long flags) if (adev->runpm) { /* only need to skip on ATPX */ - if (amdgpu_device_supports_atpx(dev) && - !amdgpu_is_atpx_hybrid()) + if (amdgpu_device_supports_px(dev)) dev_pm_set_driver_flags(dev->dev, DPM_FLAG_NO_DIRECT_COMPLETE); + /* we want direct complete for BOCO */ + if (amdgpu_device_supports_boco(dev)) + dev_pm_set_driver_flags(dev->dev, DPM_FLAG_SMART_PREPARE | + DPM_FLAG_SMART_SUSPEND | + DPM_FLAG_MAY_SKIP_RESUME); pm_runtime_use_autosuspend(dev->dev); pm_runtime_set_autosuspend_delay(dev->dev, 5000); pm_runtime_allow(dev->dev); @@ -287,22 +290,30 @@ static int amdgpu_firmware_info(struct drm_amdgpu_info_firmware *fw_info, break; case AMDGPU_INFO_FW_TA: switch (query_fw->index) { - case 0: + case TA_FW_TYPE_PSP_XGMI: fw_info->ver = adev->psp.ta_fw_version; fw_info->feature = adev->psp.ta_xgmi_ucode_version; break; - case 1: + case TA_FW_TYPE_PSP_RAS: fw_info->ver = adev->psp.ta_fw_version; fw_info->feature = adev->psp.ta_ras_ucode_version; break; - case 2: + case TA_FW_TYPE_PSP_HDCP: fw_info->ver = adev->psp.ta_fw_version; fw_info->feature = adev->psp.ta_hdcp_ucode_version; break; - case 3: + case TA_FW_TYPE_PSP_DTM: fw_info->ver = adev->psp.ta_fw_version; fw_info->feature = adev->psp.ta_dtm_ucode_version; break; + case TA_FW_TYPE_PSP_RAP: + fw_info->ver = adev->psp.ta_fw_version; + fw_info->feature = adev->psp.ta_rap_ucode_version; + break; + case TA_FW_TYPE_PSP_SECUREDISPLAY: + fw_info->ver = adev->psp.ta_fw_version; + fw_info->feature = adev->psp.ta_securedisplay_ucode_version; + break; default: return -EINVAL; } @@ -981,6 +992,63 @@ int amdgpu_info_ioctl(struct drm_device *dev, void *data, struct drm_file *filp) min_t(u64, size, sizeof(ras_mask))) ? -EFAULT : 0; } + case AMDGPU_INFO_VIDEO_CAPS: { + const struct amdgpu_video_codecs *codecs; + struct drm_amdgpu_info_video_caps *caps; + int r; + + switch (info->video_cap.type) { + case AMDGPU_INFO_VIDEO_CAPS_DECODE: + r = amdgpu_asic_query_video_codecs(adev, false, &codecs); + if (r) + return -EINVAL; + break; + case AMDGPU_INFO_VIDEO_CAPS_ENCODE: + r = amdgpu_asic_query_video_codecs(adev, true, &codecs); + if (r) + return -EINVAL; + break; + default: + DRM_DEBUG_KMS("Invalid request %d\n", + info->video_cap.type); + return -EINVAL; + } + + caps = kzalloc(sizeof(*caps), GFP_KERNEL); + if (!caps) + return -ENOMEM; + + for (i = 0; i < codecs->codec_count; i++) { + int idx = codecs->codec_array[i].codec_type; + + switch (idx) { + case AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_MPEG2: + case AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_MPEG4: + case AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_VC1: + case AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_MPEG4_AVC: + case AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_HEVC: + case AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_JPEG: + case AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_VP9: + case AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_AV1: + caps->codec_info[idx].valid = 1; + caps->codec_info[idx].max_width = + codecs->codec_array[i].max_width; + caps->codec_info[idx].max_height = + codecs->codec_array[i].max_height; + caps->codec_info[idx].max_pixels_per_frame = + codecs->codec_array[i].max_pixels_per_frame; + caps->codec_info[idx].max_level = + codecs->codec_array[i].max_level; + break; + default: + break; + } + } + r = copy_to_user(out, caps, + min((size_t)size, sizeof(*caps))) ? -EFAULT : 0; + kfree(caps); + return r; + } default: DRM_DEBUG_KMS("Invalid request %d\n", info->query); return -EINVAL; @@ -1262,16 +1330,25 @@ void amdgpu_disable_vblank_kms(struct drm_crtc *crtc) */ #if defined(CONFIG_DEBUG_FS) -static int amdgpu_debugfs_firmware_info(struct seq_file *m, void *data) +static int amdgpu_debugfs_firmware_info_show(struct seq_file *m, void *unused) { - struct drm_info_node *node = (struct drm_info_node *) m->private; - struct drm_device *dev = node->minor->dev; - struct amdgpu_device *adev = drm_to_adev(dev); + struct amdgpu_device *adev = (struct amdgpu_device *)m->private; struct drm_amdgpu_info_firmware fw_info; struct drm_amdgpu_query_fw query_fw; struct atom_context *ctx = adev->mode_info.atom_context; int ret, i; + static const char *ta_fw_name[TA_FW_TYPE_MAX_INDEX] = { +#define TA_FW_NAME(type) [TA_FW_TYPE_PSP_##type] = #type + TA_FW_NAME(XGMI), + TA_FW_NAME(RAS), + TA_FW_NAME(HDCP), + TA_FW_NAME(DTM), + TA_FW_NAME(RAP), + TA_FW_NAME(SECUREDISPLAY), +#undef TA_FW_NAME + }; + /* VCE */ query_fw.fw_type = AMDGPU_INFO_FW_VCE; ret = amdgpu_firmware_info(&fw_info, &query_fw, adev); @@ -1389,31 +1466,14 @@ static int amdgpu_debugfs_firmware_info(struct seq_file *m, void *data) fw_info.feature, fw_info.ver); query_fw.fw_type = AMDGPU_INFO_FW_TA; - for (i = 0; i < 4; i++) { + for (i = TA_FW_TYPE_PSP_XGMI; i < TA_FW_TYPE_MAX_INDEX; i++) { query_fw.index = i; ret = amdgpu_firmware_info(&fw_info, &query_fw, adev); if (ret) continue; - switch (query_fw.index) { - case 0: - seq_printf(m, "TA %s feature version: 0x%08x, firmware version: 0x%08x\n", - "RAS", fw_info.feature, fw_info.ver); - break; - case 1: - seq_printf(m, "TA %s feature version: 0x%08x, firmware version: 0x%08x\n", - "XGMI", fw_info.feature, fw_info.ver); - break; - case 2: - seq_printf(m, "TA %s feature version: 0x%08x, firmware version: 0x%08x\n", - "HDCP", fw_info.feature, fw_info.ver); - break; - case 3: - seq_printf(m, "TA %s feature version: 0x%08x, firmware version: 0x%08x\n", - "DTM", fw_info.feature, fw_info.ver); - break; - default: - return -EINVAL; - } + + seq_printf(m, "TA %s feature version: 0x%08x, firmware version: 0x%08x\n", + ta_fw_name[i], fw_info.feature, fw_info.ver); } /* SMC */ @@ -1472,17 +1532,18 @@ static int amdgpu_debugfs_firmware_info(struct seq_file *m, void *data) return 0; } -static const struct drm_info_list amdgpu_firmware_info_list[] = { - {"amdgpu_firmware_info", amdgpu_debugfs_firmware_info, 0, NULL}, -}; +DEFINE_SHOW_ATTRIBUTE(amdgpu_debugfs_firmware_info); + #endif -int amdgpu_debugfs_firmware_init(struct amdgpu_device *adev) +void amdgpu_debugfs_firmware_init(struct amdgpu_device *adev) { #if defined(CONFIG_DEBUG_FS) - return amdgpu_debugfs_add_files(adev, amdgpu_firmware_info_list, - ARRAY_SIZE(amdgpu_firmware_info_list)); -#else - return 0; + struct drm_minor *minor = adev_to_drm(adev)->primary; + struct dentry *root = minor->debugfs_root; + + debugfs_create_file("amdgpu_firmware_info", 0444, root, + adev, &amdgpu_debugfs_firmware_info_fops); + #endif } diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_mmhub.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_mmhub.h index 1ae9bdae7311..11aa29933c1f 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_mmhub.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_mmhub.h @@ -21,12 +21,16 @@ #ifndef __AMDGPU_MMHUB_H__ #define __AMDGPU_MMHUB_H__ -struct amdgpu_mmhub_funcs { - void (*ras_init)(struct amdgpu_device *adev); +struct amdgpu_mmhub_ras_funcs { int (*ras_late_init)(struct amdgpu_device *adev); + void (*ras_fini)(struct amdgpu_device *adev); void (*query_ras_error_count)(struct amdgpu_device *adev, - void *ras_error_status); + void *ras_error_status); + void (*query_ras_error_status)(struct amdgpu_device *adev); void (*reset_ras_error_count)(struct amdgpu_device *adev); +}; + +struct amdgpu_mmhub_funcs { u64 (*get_fb_location)(struct amdgpu_device *adev); void (*init)(struct amdgpu_device *adev); int (*gart_enable)(struct amdgpu_device *adev); @@ -40,12 +44,12 @@ struct amdgpu_mmhub_funcs { uint64_t page_table_base); void (*update_power_gating)(struct amdgpu_device *adev, bool enable); - void (*query_ras_error_status)(struct amdgpu_device *adev); }; struct amdgpu_mmhub { struct ras_common_if *ras_if; const struct amdgpu_mmhub_funcs *funcs; + const struct amdgpu_mmhub_ras_funcs *ras_funcs; }; int amdgpu_mmhub_ras_late_init(struct amdgpu_device *adev); diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_mode.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_mode.h index 319cb19e1b99..cb0b581bbce7 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_mode.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_mode.h @@ -602,6 +602,14 @@ int amdgpu_display_get_crtc_scanoutpos(struct drm_device *dev, int *hpos, ktime_t *stime, ktime_t *etime, const struct drm_display_mode *mode); +int amdgpu_display_gem_fb_init(struct drm_device *dev, + struct amdgpu_framebuffer *rfb, + const struct drm_mode_fb_cmd2 *mode_cmd, + struct drm_gem_object *obj); +int amdgpu_display_gem_fb_verify_and_init( + struct drm_device *dev, struct amdgpu_framebuffer *rfb, + struct drm_file *file_priv, const struct drm_mode_fb_cmd2 *mode_cmd, + struct drm_gem_object *obj); int amdgpu_display_framebuffer_init(struct drm_device *dev, struct amdgpu_framebuffer *rfb, const struct drm_mode_fb_cmd2 *mode_cmd, diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_nbio.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_nbio.h index 7c11bce4514b..25ee53545837 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_nbio.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_nbio.h @@ -47,6 +47,17 @@ struct nbio_hdp_flush_reg { u32 ref_and_mask_sdma7; }; +struct amdgpu_nbio_ras_funcs { + void (*handle_ras_controller_intr_no_bifring)(struct amdgpu_device *adev); + void (*handle_ras_err_event_athub_intr_no_bifring)(struct amdgpu_device *adev); + int (*init_ras_controller_interrupt)(struct amdgpu_device *adev); + int (*init_ras_err_event_athub_interrupt)(struct amdgpu_device *adev); + void (*query_ras_error_count)(struct amdgpu_device *adev, + void *ras_error_status); + int (*ras_late_init)(struct amdgpu_device *adev); + void (*ras_fini)(struct amdgpu_device *adev); +}; + struct amdgpu_nbio_funcs { const struct nbio_hdp_flush_reg *hdp_flush_reg; u32 (*get_hdp_flush_req_offset)(struct amdgpu_device *adev); @@ -79,13 +90,6 @@ struct amdgpu_nbio_funcs { void (*ih_control)(struct amdgpu_device *adev); void (*init_registers)(struct amdgpu_device *adev); void (*remap_hdp_registers)(struct amdgpu_device *adev); - void (*handle_ras_controller_intr_no_bifring)(struct amdgpu_device *adev); - void (*handle_ras_err_event_athub_intr_no_bifring)(struct amdgpu_device *adev); - int (*init_ras_controller_interrupt)(struct amdgpu_device *adev); - int (*init_ras_err_event_athub_interrupt)(struct amdgpu_device *adev); - void (*query_ras_error_count)(struct amdgpu_device *adev, - void *ras_error_status); - int (*ras_late_init)(struct amdgpu_device *adev); void (*enable_aspm)(struct amdgpu_device *adev, bool enable); void (*program_aspm)(struct amdgpu_device *adev); @@ -97,6 +101,7 @@ struct amdgpu_nbio { struct amdgpu_irq_src ras_err_event_athub_irq; struct ras_common_if *ras_if; const struct amdgpu_nbio_funcs *funcs; + const struct amdgpu_nbio_ras_funcs *ras_funcs; }; int amdgpu_nbio_ras_late_init(struct amdgpu_device *adev); diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_object.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_object.c index 072050429a2f..1345f7eba011 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_object.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_object.c @@ -77,6 +77,7 @@ static void amdgpu_bo_destroy(struct ttm_buffer_object *tbo) { struct amdgpu_device *adev = amdgpu_ttm_adev(tbo->bdev); struct amdgpu_bo *bo = ttm_to_amdgpu_bo(tbo); + struct amdgpu_bo_user *ubo; if (bo->tbo.pin_count > 0) amdgpu_bo_subtract_pin_size(bo); @@ -94,7 +95,11 @@ static void amdgpu_bo_destroy(struct ttm_buffer_object *tbo) } amdgpu_bo_unref(&bo->parent); - kfree(bo->metadata); + if (bo->tbo.type == ttm_bo_type_device) { + ubo = to_amdgpu_bo_user(bo); + kfree(ubo->metadata); + } + kfree(bo); } @@ -248,6 +253,7 @@ int amdgpu_bo_create_reserved(struct amdgpu_device *adev, bp.flags |= AMDGPU_GEM_CREATE_VRAM_CONTIGUOUS; bp.type = ttm_bo_type_kernel; bp.resv = NULL; + bp.bo_ptr_size = sizeof(struct amdgpu_bo); if (!*bo_ptr) { r = amdgpu_bo_create(adev, &bp, bo_ptr); @@ -523,7 +529,6 @@ static int amdgpu_bo_do_create(struct amdgpu_device *adev, }; struct amdgpu_bo *bo; unsigned long page_align, size = bp->size; - size_t acc_size; int r; /* Note that GDS/GWS/OA allocates 1 page per byte/resource. */ @@ -544,12 +549,10 @@ static int amdgpu_bo_do_create(struct amdgpu_device *adev, if (!amdgpu_bo_validate_size(adev, size, bp->domain)) return -ENOMEM; - *bo_ptr = NULL; + BUG_ON(bp->bo_ptr_size < sizeof(struct amdgpu_bo)); - acc_size = ttm_bo_dma_acc_size(&adev->mman.bdev, size, - sizeof(struct amdgpu_bo)); - - bo = kzalloc(sizeof(struct amdgpu_bo), GFP_KERNEL); + *bo_ptr = NULL; + bo = kzalloc(bp->bo_ptr_size, GFP_KERNEL); if (bo == NULL) return -ENOMEM; drm_gem_private_object_init(adev_to_drm(adev), &bo->tbo.base, size); @@ -577,8 +580,8 @@ static int amdgpu_bo_do_create(struct amdgpu_device *adev, bo->tbo.priority = 1; r = ttm_bo_init_reserved(&adev->mman.bdev, &bo->tbo, size, bp->type, - &bo->placement, page_align, &ctx, acc_size, - NULL, bp->resv, &amdgpu_bo_destroy); + &bo->placement, page_align, &ctx, NULL, + bp->resv, &amdgpu_bo_destroy); if (unlikely(r != 0)) return r; @@ -639,6 +642,7 @@ static int amdgpu_bo_create_shadow(struct amdgpu_device *adev, AMDGPU_GEM_CREATE_SHADOW; bp.type = ttm_bo_type_kernel; bp.resv = bo->tbo.base.resv; + bp.bo_ptr_size = sizeof(struct amdgpu_bo); r = amdgpu_bo_do_create(adev, &bp, &bo->shadow); if (!r) { @@ -673,6 +677,7 @@ int amdgpu_bo_create(struct amdgpu_device *adev, int r; bp->flags = bp->flags & ~AMDGPU_GEM_CREATE_SHADOW; + r = amdgpu_bo_do_create(adev, bp, bo_ptr); if (r) return r; @@ -695,6 +700,34 @@ int amdgpu_bo_create(struct amdgpu_device *adev, } /** + * amdgpu_bo_create_user - create an &amdgpu_bo_user buffer object + * @adev: amdgpu device object + * @bp: parameters to be used for the buffer object + * @ubo_ptr: pointer to the buffer object pointer + * + * Create a BO to be used by user application; + * + * Returns: + * 0 for success or a negative error code on failure. + */ + +int amdgpu_bo_create_user(struct amdgpu_device *adev, + struct amdgpu_bo_param *bp, + struct amdgpu_bo_user **ubo_ptr) +{ + struct amdgpu_bo *bo_ptr; + int r; + + bp->flags = bp->flags & ~AMDGPU_GEM_CREATE_SHADOW; + bp->bo_ptr_size = sizeof(struct amdgpu_bo_user); + r = amdgpu_bo_do_create(adev, bp, &bo_ptr); + if (r) + return r; + + *ubo_ptr = to_amdgpu_bo_user(bo_ptr); + return r; +} +/** * amdgpu_bo_validate - validate an &amdgpu_bo buffer object * @bo: pointer to the buffer object * @@ -1062,13 +1095,17 @@ static const char *amdgpu_vram_names[] = { */ int amdgpu_bo_init(struct amdgpu_device *adev) { - /* reserve PAT memory space to WC for VRAM */ - arch_io_reserve_memtype_wc(adev->gmc.aper_base, - adev->gmc.aper_size); + /* On A+A platform, VRAM can be mapped as WB */ + if (!adev->gmc.xgmi.connected_to_cpu) { + /* reserve PAT memory space to WC for VRAM */ + arch_io_reserve_memtype_wc(adev->gmc.aper_base, + adev->gmc.aper_size); + + /* Add an MTRR for the VRAM */ + adev->gmc.vram_mtrr = arch_phys_wc_add(adev->gmc.aper_base, + adev->gmc.aper_size); + } - /* Add an MTRR for the VRAM */ - adev->gmc.vram_mtrr = arch_phys_wc_add(adev->gmc.aper_base, - adev->gmc.aper_size); DRM_INFO("Detected VRAM RAM=%lluM, BAR=%lluM\n", adev->gmc.mc_vram_size >> 20, (unsigned long long)adev->gmc.aper_size >> 20); @@ -1086,27 +1123,10 @@ int amdgpu_bo_init(struct amdgpu_device *adev) void amdgpu_bo_fini(struct amdgpu_device *adev) { amdgpu_ttm_fini(adev); - arch_phys_wc_del(adev->gmc.vram_mtrr); - arch_io_free_memtype_wc(adev->gmc.aper_base, adev->gmc.aper_size); -} - -/** - * amdgpu_bo_fbdev_mmap - mmap fbdev memory - * @bo: &amdgpu_bo buffer object - * @vma: vma as input from the fbdev mmap method - * - * Calls ttm_fbdev_mmap() to mmap fbdev memory if it is backed by a bo. - * - * Returns: - * 0 for success or a negative error code on failure. - */ -int amdgpu_bo_fbdev_mmap(struct amdgpu_bo *bo, - struct vm_area_struct *vma) -{ - if (vma->vm_pgoff != 0) - return -EACCES; - - return ttm_bo_mmap_obj(vma, &bo->tbo); + if (!adev->gmc.xgmi.connected_to_cpu) { + arch_phys_wc_del(adev->gmc.vram_mtrr); + arch_io_free_memtype_wc(adev->gmc.aper_base, adev->gmc.aper_size); + } } /** @@ -1123,12 +1143,15 @@ int amdgpu_bo_fbdev_mmap(struct amdgpu_bo *bo, int amdgpu_bo_set_tiling_flags(struct amdgpu_bo *bo, u64 tiling_flags) { struct amdgpu_device *adev = amdgpu_ttm_adev(bo->tbo.bdev); + struct amdgpu_bo_user *ubo; + BUG_ON(bo->tbo.type == ttm_bo_type_kernel); if (adev->family <= AMDGPU_FAMILY_CZ && AMDGPU_TILING_GET(tiling_flags, TILE_SPLIT) > 6) return -EINVAL; - bo->tiling_flags = tiling_flags; + ubo = to_amdgpu_bo_user(bo); + ubo->tiling_flags = tiling_flags; return 0; } @@ -1142,10 +1165,14 @@ int amdgpu_bo_set_tiling_flags(struct amdgpu_bo *bo, u64 tiling_flags) */ void amdgpu_bo_get_tiling_flags(struct amdgpu_bo *bo, u64 *tiling_flags) { + struct amdgpu_bo_user *ubo; + + BUG_ON(bo->tbo.type == ttm_bo_type_kernel); dma_resv_assert_held(bo->tbo.base.resv); + ubo = to_amdgpu_bo_user(bo); if (tiling_flags) - *tiling_flags = bo->tiling_flags; + *tiling_flags = ubo->tiling_flags; } /** @@ -1164,13 +1191,16 @@ void amdgpu_bo_get_tiling_flags(struct amdgpu_bo *bo, u64 *tiling_flags) int amdgpu_bo_set_metadata (struct amdgpu_bo *bo, void *metadata, uint32_t metadata_size, uint64_t flags) { + struct amdgpu_bo_user *ubo; void *buffer; + BUG_ON(bo->tbo.type == ttm_bo_type_kernel); + ubo = to_amdgpu_bo_user(bo); if (!metadata_size) { - if (bo->metadata_size) { - kfree(bo->metadata); - bo->metadata = NULL; - bo->metadata_size = 0; + if (ubo->metadata_size) { + kfree(ubo->metadata); + ubo->metadata = NULL; + ubo->metadata_size = 0; } return 0; } @@ -1182,10 +1212,10 @@ int amdgpu_bo_set_metadata (struct amdgpu_bo *bo, void *metadata, if (buffer == NULL) return -ENOMEM; - kfree(bo->metadata); - bo->metadata_flags = flags; - bo->metadata = buffer; - bo->metadata_size = metadata_size; + kfree(ubo->metadata); + ubo->metadata_flags = flags; + ubo->metadata = buffer; + ubo->metadata_size = metadata_size; return 0; } @@ -1209,21 +1239,25 @@ int amdgpu_bo_get_metadata(struct amdgpu_bo *bo, void *buffer, size_t buffer_size, uint32_t *metadata_size, uint64_t *flags) { + struct amdgpu_bo_user *ubo; + if (!buffer && !metadata_size) return -EINVAL; + BUG_ON(bo->tbo.type == ttm_bo_type_kernel); + ubo = to_amdgpu_bo_user(bo); if (buffer) { - if (buffer_size < bo->metadata_size) + if (buffer_size < ubo->metadata_size) return -EINVAL; - if (bo->metadata_size) - memcpy(buffer, bo->metadata, bo->metadata_size); + if (ubo->metadata_size) + memcpy(buffer, ubo->metadata, ubo->metadata_size); } if (metadata_size) - *metadata_size = bo->metadata_size; + *metadata_size = ubo->metadata_size; if (flags) - *flags = bo->metadata_flags; + *flags = ubo->metadata_flags; return 0; } diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_object.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_object.h index 9ac37569823f..2d1fefbe1e99 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_object.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_object.h @@ -37,9 +37,12 @@ #define AMDGPU_BO_INVALID_OFFSET LONG_MAX #define AMDGPU_BO_MAX_PLACEMENTS 3 +#define to_amdgpu_bo_user(abo) container_of((abo), struct amdgpu_bo_user, bo) + struct amdgpu_bo_param { unsigned long size; int byte_align; + u32 bo_ptr_size; u32 domain; u32 preferred_domain; u64 flags; @@ -89,10 +92,6 @@ struct amdgpu_bo { struct ttm_buffer_object tbo; struct ttm_bo_kmap_obj kmap; u64 flags; - u64 tiling_flags; - u64 metadata_flags; - void *metadata; - u32 metadata_size; unsigned prime_shared_count; /* per VM structure for page tables and with virtual addresses */ struct amdgpu_vm_bo_base *vm_bo; @@ -100,7 +99,6 @@ struct amdgpu_bo { struct amdgpu_bo *parent; struct amdgpu_bo *shadow; - struct amdgpu_mn *mn; #ifdef CONFIG_MMU_NOTIFIER @@ -112,6 +110,15 @@ struct amdgpu_bo { struct kgd_mem *kfd_bo; }; +struct amdgpu_bo_user { + struct amdgpu_bo bo; + u64 tiling_flags; + u64 metadata_flags; + void *metadata; + u32 metadata_size; + +}; + static inline struct amdgpu_bo *ttm_to_amdgpu_bo(struct ttm_buffer_object *tbo) { return container_of(tbo, struct amdgpu_bo, tbo); @@ -255,6 +262,9 @@ int amdgpu_bo_create_kernel(struct amdgpu_device *adev, int amdgpu_bo_create_kernel_at(struct amdgpu_device *adev, uint64_t offset, uint64_t size, uint32_t domain, struct amdgpu_bo **bo_ptr, void **cpu_addr); +int amdgpu_bo_create_user(struct amdgpu_device *adev, + struct amdgpu_bo_param *bp, + struct amdgpu_bo_user **ubo_ptr); void amdgpu_bo_free_kernel(struct amdgpu_bo **bo, u64 *gpu_addr, void **cpu_addr); int amdgpu_bo_kmap(struct amdgpu_bo *bo, void **ptr); @@ -269,8 +279,6 @@ void amdgpu_bo_unpin(struct amdgpu_bo *bo); int amdgpu_bo_evict_vram(struct amdgpu_device *adev); int amdgpu_bo_init(struct amdgpu_device *adev); void amdgpu_bo_fini(struct amdgpu_device *adev); -int amdgpu_bo_fbdev_mmap(struct amdgpu_bo *bo, - struct vm_area_struct *vma); int amdgpu_bo_set_tiling_flags(struct amdgpu_bo *bo, u64 tiling_flags); void amdgpu_bo_get_tiling_flags(struct amdgpu_bo *bo, u64 *tiling_flags); int amdgpu_bo_set_metadata (struct amdgpu_bo *bo, void *metadata, @@ -329,7 +337,7 @@ void amdgpu_sa_bo_dump_debug_info(struct amdgpu_sa_manager *sa_manager, struct seq_file *m); u64 amdgpu_bo_print_info(int id, struct amdgpu_bo *bo, struct seq_file *m); #endif -int amdgpu_debugfs_sa_init(struct amdgpu_device *adev); +void amdgpu_debugfs_sa_init(struct amdgpu_device *adev); bool amdgpu_bo_support_uswc(u64 bo_flags); diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_pmu.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_pmu.c index 19c0a3655228..82e9ecf84352 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_pmu.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_pmu.c @@ -519,8 +519,10 @@ static int init_pmu_entry_by_type_and_add(struct amdgpu_pmu_entry *pmu_entry, pmu_entry->pmu.attr_groups = kmemdup(attr_groups, sizeof(attr_groups), GFP_KERNEL); - if (!pmu_entry->pmu.attr_groups) + if (!pmu_entry->pmu.attr_groups) { + ret = -ENOMEM; goto err_attr_group; + } snprintf(pmu_name, PMU_NAME_SIZE, "%s_%d", pmu_entry->pmu_file_prefix, adev_to_drm(pmu_entry->adev)->primary->index); diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_psp.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_psp.c index 839917eb7bc3..a09483beb968 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_psp.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_psp.c @@ -34,6 +34,7 @@ #include "psp_v10_0.h" #include "psp_v11_0.h" #include "psp_v12_0.h" +#include "psp_v13_0.h" #include "amdgpu_ras.h" #include "amdgpu_securedisplay.h" @@ -56,7 +57,7 @@ static int psp_load_smu_fw(struct psp_context *psp); * - Load XGMI/RAS/HDCP/DTM TA if any * * This new sequence is required for - * - Arcturus + * - Arcturus and onwards * - Navi12 and onwards */ static void psp_check_pmfw_centralized_cstate_management(struct psp_context *psp) @@ -71,7 +72,7 @@ static void psp_check_pmfw_centralized_cstate_management(struct psp_context *psp if (adev->flags & AMD_IS_APU) return; - if ((adev->asic_type == CHIP_ARCTURUS) || + if ((adev->asic_type >= CHIP_ARCTURUS) || (adev->asic_type >= CHIP_NAVI12)) psp->pmfw_centralized_cstate_management = true; } @@ -109,6 +110,9 @@ static int psp_early_init(void *handle) case CHIP_RENOIR: psp_v12_0_set_psp_funcs(psp); break; + case CHIP_ALDEBARAN: + psp_v13_0_set_psp_funcs(psp); + break; default: return -EINVAL; } @@ -324,8 +328,12 @@ psp_cmd_submit_buf(struct psp_context *psp, static void psp_prep_tmr_cmd_buf(struct psp_context *psp, struct psp_gfx_cmd_resp *cmd, - uint64_t tmr_mc, uint32_t size) + uint64_t tmr_mc, struct amdgpu_bo *tmr_bo) { + struct amdgpu_device *adev = psp->adev; + uint32_t size = amdgpu_bo_size(tmr_bo); + uint64_t tmr_pa = amdgpu_gmc_vram_pa(adev, tmr_bo); + if (amdgpu_sriov_vf(psp->adev)) cmd->cmd_id = GFX_CMD_ID_SETUP_VMR; else @@ -333,6 +341,9 @@ static void psp_prep_tmr_cmd_buf(struct psp_context *psp, cmd->cmd.cmd_setup_tmr.buf_phy_addr_lo = lower_32_bits(tmr_mc); cmd->cmd.cmd_setup_tmr.buf_phy_addr_hi = upper_32_bits(tmr_mc); cmd->cmd.cmd_setup_tmr.buf_size = size; + cmd->cmd.cmd_setup_tmr.bitfield.virt_phy_addr = 1; + cmd->cmd.cmd_setup_tmr.system_phy_addr_lo = lower_32_bits(tmr_pa); + cmd->cmd.cmd_setup_tmr.system_phy_addr_hi = upper_32_bits(tmr_pa); } static void psp_prep_load_toc_cmd_buf(struct psp_gfx_cmd_resp *cmd, @@ -383,7 +394,7 @@ static int psp_tmr_init(struct psp_context *psp) * Note: this memory need be reserved till the driver * uninitializes. */ - tmr_size = PSP_TMR_SIZE; + tmr_size = PSP_TMR_SIZE(psp->adev); /* For ASICs support RLC autoload, psp will parse the toc * and calculate the total size of TMR needed */ @@ -399,7 +410,7 @@ static int psp_tmr_init(struct psp_context *psp) } pptr = amdgpu_sriov_vf(psp->adev) ? &tmr_buf : NULL; - ret = amdgpu_bo_create_kernel(psp->adev, tmr_size, PSP_TMR_SIZE, + ret = amdgpu_bo_create_kernel(psp->adev, tmr_size, PSP_TMR_SIZE(psp->adev), AMDGPU_GEM_DOMAIN_VRAM, &psp->tmr_bo, &psp->tmr_mc_addr, pptr); @@ -452,8 +463,7 @@ static int psp_tmr_load(struct psp_context *psp) if (!cmd) return -ENOMEM; - psp_prep_tmr_cmd_buf(psp, cmd, psp->tmr_mc_addr, - amdgpu_bo_size(psp->tmr_bo)); + psp_prep_tmr_cmd_buf(psp, cmd, psp->tmr_mc_addr, psp->tmr_bo); DRM_INFO("reserve 0x%lx from 0x%llx for PSP TMR\n", amdgpu_bo_size(psp->tmr_bo), psp->tmr_mc_addr); @@ -542,6 +552,46 @@ int psp_get_fw_attestation_records_addr(struct psp_context *psp, return ret; } +static int psp_boot_config_set(struct amdgpu_device *adev) +{ + struct psp_context *psp = &adev->psp; + struct psp_gfx_cmd_resp *cmd = psp->cmd; + + if (adev->asic_type != CHIP_SIENNA_CICHLID || amdgpu_sriov_vf(adev)) + return 0; + + memset(cmd, 0, sizeof(struct psp_gfx_cmd_resp)); + + cmd->cmd_id = GFX_CMD_ID_BOOT_CFG; + cmd->cmd.boot_cfg.sub_cmd = BOOTCFG_CMD_SET; + cmd->cmd.boot_cfg.boot_config = BOOT_CONFIG_GECC; + cmd->cmd.boot_cfg.boot_config_valid = BOOT_CONFIG_GECC; + + return psp_cmd_submit_buf(psp, NULL, cmd, psp->fence_buf_mc_addr); +} + +static int psp_rl_load(struct amdgpu_device *adev) +{ + struct psp_context *psp = &adev->psp; + struct psp_gfx_cmd_resp *cmd = psp->cmd; + + if (psp->rl_bin_size == 0) + return 0; + + memset(psp->fw_pri_buf, 0, PSP_1_MEG); + memcpy(psp->fw_pri_buf, psp->rl_start_addr, psp->rl_bin_size); + + memset(cmd, 0, sizeof(struct psp_gfx_cmd_resp)); + + cmd->cmd_id = GFX_CMD_ID_LOAD_IP_FW; + cmd->cmd.cmd_load_ip_fw.fw_phy_addr_lo = lower_32_bits(psp->fw_pri_mc_addr); + cmd->cmd.cmd_load_ip_fw.fw_phy_addr_hi = upper_32_bits(psp->fw_pri_mc_addr); + cmd->cmd.cmd_load_ip_fw.fw_size = psp->rl_bin_size; + cmd->cmd.cmd_load_ip_fw.fw_type = GFX_FW_TYPE_REG_LIST; + + return psp_cmd_submit_buf(psp, NULL, cmd, psp->fence_buf_mc_addr); +} + static void psp_prep_asd_load_cmd_buf(struct psp_gfx_cmd_resp *cmd, uint64_t asd_mc, uint32_t size) { @@ -755,8 +805,9 @@ static int psp_xgmi_unload(struct psp_context *psp) struct psp_gfx_cmd_resp *cmd; struct amdgpu_device *adev = psp->adev; - /* XGMI TA unload currently is not supported on Arcturus */ - if (adev->asic_type == CHIP_ARCTURUS) + /* XGMI TA unload currently is not supported on Arcturus/Aldebaran A+A */ + if (adev->asic_type == CHIP_ARCTURUS || + (adev->asic_type == CHIP_ALDEBARAN && adev->gmc.xgmi.connected_to_cpu)) return 0; /* @@ -1561,6 +1612,7 @@ static int psp_rap_unload(struct psp_context *psp) static int psp_rap_initialize(struct psp_context *psp) { int ret; + enum ta_rap_status status = TA_RAP_STATUS__SUCCESS; /* * TODO: bypass the initialize in sriov for now @@ -1584,8 +1636,8 @@ static int psp_rap_initialize(struct psp_context *psp) if (ret) return ret; - ret = psp_rap_invoke(psp, TA_CMD_RAP__INITIALIZE); - if (ret != TA_RAP_STATUS__SUCCESS) { + ret = psp_rap_invoke(psp, TA_CMD_RAP__INITIALIZE, &status); + if (ret || status != TA_RAP_STATUS__SUCCESS) { psp_rap_unload(psp); amdgpu_bo_free_kernel(&psp->rap_context.rap_shared_bo, @@ -1594,8 +1646,10 @@ static int psp_rap_initialize(struct psp_context *psp) psp->rap_context.rap_initialized = false; - dev_warn(psp->adev->dev, "RAP TA initialize fail.\n"); - return -EINVAL; + dev_warn(psp->adev->dev, "RAP TA initialize fail (%d) status %d.\n", + ret, status); + + return ret; } return 0; @@ -1620,13 +1674,13 @@ static int psp_rap_terminate(struct psp_context *psp) return ret; } -int psp_rap_invoke(struct psp_context *psp, uint32_t ta_cmd_id) +int psp_rap_invoke(struct psp_context *psp, uint32_t ta_cmd_id, enum ta_rap_status *status) { struct ta_rap_shared_memory *rap_cmd; - int ret; + int ret = 0; if (!psp->rap_context.rap_initialized) - return -EINVAL; + return 0; if (ta_cmd_id != TA_CMD_RAP__INITIALIZE && ta_cmd_id != TA_CMD_RAP__VALIDATE_L0) @@ -1642,14 +1696,16 @@ int psp_rap_invoke(struct psp_context *psp, uint32_t ta_cmd_id) rap_cmd->validation_method_id = METHOD_A; ret = psp_ta_invoke(psp, rap_cmd->cmd_id, psp->rap_context.session_id); - if (ret) { - mutex_unlock(&psp->rap_context.mutex); - return ret; - } + if (ret) + goto out_unlock; + + if (status) + *status = rap_cmd->rap_status; +out_unlock: mutex_unlock(&psp->rap_context.mutex); - return rap_cmd->rap_status; + return ret; } // RAP end @@ -1870,6 +1926,11 @@ static int psp_hw_start(struct psp_context *psp) return ret; } + ret = psp_boot_config_set(adev); + if (ret) { + DRM_WARN("PSP set boot config@\n"); + } + ret = psp_tmr_init(psp); if (ret) { DRM_ERROR("PSP tmr init failed!\n"); @@ -2104,9 +2165,13 @@ static int psp_load_smu_fw(struct psp_context *psp) if (!ucode->fw || amdgpu_sriov_vf(psp->adev)) return 0; - - if (amdgpu_in_reset(adev) && ras && ras->supported && - adev->asic_type == CHIP_ARCTURUS) { + if ((amdgpu_in_reset(adev) && + ras && ras->supported && + (adev->asic_type == CHIP_ARCTURUS || + adev->asic_type == CHIP_VEGA20)) || + (adev->in_runpm && + adev->asic_type >= CHIP_NAVI10 && + adev->asic_type <= CHIP_NAVI12)) { ret = amdgpu_dpm_set_mp1_state(adev, PP_MP1_STATE_UNLOAD); if (ret) { DRM_WARN("Failed to set MP1 state prepare for reload\n"); @@ -2159,6 +2224,22 @@ static bool fw_load_skip_check(struct psp_context *psp, return false; } +int psp_load_fw_list(struct psp_context *psp, + struct amdgpu_firmware_info **ucode_list, int ucode_count) +{ + int ret = 0, i; + struct amdgpu_firmware_info *ucode; + + for (i = 0; i < ucode_count; ++i) { + ucode = ucode_list[i]; + psp_print_fw_hdr(psp, ucode); + ret = psp_execute_np_fw_load(psp, ucode); + if (ret) + return ret; + } + return ret; +} + static int psp_np_fw_load(struct psp_context *psp) { int i, ret; @@ -2276,6 +2357,12 @@ skip_memalloc: return ret; } + ret = psp_rl_load(adev); + if (ret) { + DRM_ERROR("PSP load RL failed!\n"); + return ret; + } + if (psp->adev->psp.ta_fw) { ret = psp_ras_initialize(psp); if (ret) @@ -2751,6 +2838,9 @@ int psp_init_sos_microcode(struct psp_context *psp, adev->psp.spl_bin_size = le32_to_cpu(sos_hdr_v1_3->spl_size_bytes); adev->psp.spl_start_addr = (uint8_t *)adev->psp.sys_start_addr + le32_to_cpu(sos_hdr_v1_3->spl_offset_bytes); + adev->psp.rl_bin_size = le32_to_cpu(sos_hdr_v1_3->rl_size_bytes); + adev->psp.rl_start_addr = (uint8_t *)adev->psp.sys_start_addr + + le32_to_cpu(sos_hdr_v1_3->rl_offset_bytes); } break; default: @@ -2916,7 +3006,7 @@ static ssize_t psp_usbc_pd_fw_sysfs_read(struct device *dev, return ret; } - return snprintf(buf, PAGE_SIZE, "%x\n", fw_ver); + return sysfs_emit(buf, "%x\n", fw_ver); } static ssize_t psp_usbc_pd_fw_sysfs_write(struct device *dev, @@ -3052,3 +3142,11 @@ const struct amdgpu_ip_block_version psp_v12_0_ip_block = .rev = 0, .funcs = &psp_ip_funcs, }; + +const struct amdgpu_ip_block_version psp_v13_0_ip_block = { + .type = AMD_IP_BLOCK_TYPE_PSP, + .major = 13, + .minor = 0, + .rev = 0, + .funcs = &psp_ip_funcs, +}; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_psp.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_psp.h index cb50ba445f8c..46a5328e00e0 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_psp.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_psp.h @@ -37,7 +37,7 @@ #define PSP_XGMI_SHARED_MEM_SIZE 0x4000 #define PSP_RAS_SHARED_MEM_SIZE 0x4000 #define PSP_1_MEG 0x100000 -#define PSP_TMR_SIZE 0x400000 +#define PSP_TMR_SIZE(adev) ((adev)->asic_type == CHIP_ALDEBARAN ? 0x800000 : 0x400000) #define PSP_HDCP_SHARED_MEM_SIZE 0x4000 #define PSP_DTM_SHARED_MEM_SIZE 0x4000 #define PSP_RAP_SHARED_MEM_SIZE 0x4000 @@ -248,11 +248,13 @@ struct psp_context uint32_t toc_bin_size; uint32_t kdb_bin_size; uint32_t spl_bin_size; + uint32_t rl_bin_size; uint8_t *sys_start_addr; uint8_t *sos_start_addr; uint8_t *toc_start_addr; uint8_t *kdb_start_addr; uint8_t *spl_start_addr; + uint8_t *rl_start_addr; /* tmr buffer */ struct amdgpu_bo *tmr_bo; @@ -365,11 +367,13 @@ struct amdgpu_psp_funcs { extern const struct amd_ip_funcs psp_ip_funcs; extern const struct amdgpu_ip_block_version psp_v3_1_ip_block; -extern int psp_wait_for(struct psp_context *psp, uint32_t reg_index, - uint32_t field_val, uint32_t mask, bool check_changed); - extern const struct amdgpu_ip_block_version psp_v10_0_ip_block; +extern const struct amdgpu_ip_block_version psp_v11_0_ip_block; extern const struct amdgpu_ip_block_version psp_v12_0_ip_block; +extern const struct amdgpu_ip_block_version psp_v13_0_ip_block; + +extern int psp_wait_for(struct psp_context *psp, uint32_t reg_index, + uint32_t field_val, uint32_t mask, bool check_changed); int psp_gpu_reset(struct amdgpu_device *adev); int psp_update_vcn_sram(struct amdgpu_device *adev, int inst_idx, @@ -395,12 +399,11 @@ int psp_ras_trigger_error(struct psp_context *psp, int psp_hdcp_invoke(struct psp_context *psp, uint32_t ta_cmd_id); int psp_dtm_invoke(struct psp_context *psp, uint32_t ta_cmd_id); -int psp_rap_invoke(struct psp_context *psp, uint32_t ta_cmd_id); +int psp_rap_invoke(struct psp_context *psp, uint32_t ta_cmd_id, enum ta_rap_status *status); int psp_securedisplay_invoke(struct psp_context *psp, uint32_t ta_cmd_id); int psp_rlc_autoload_start(struct psp_context *psp); -extern const struct amdgpu_ip_block_version psp_v11_0_ip_block; int psp_reg_program(struct psp_context *psp, enum psp_reg_prog_id reg, uint32_t value); int psp_ring_cmd_submit(struct psp_context *psp, @@ -417,4 +420,7 @@ int psp_init_ta_microcode(struct psp_context *psp, const char *chip_name); int psp_get_fw_attestation_records_addr(struct psp_context *psp, uint64_t *output_ptr); + +int psp_load_fw_list(struct psp_context *psp, + struct amdgpu_firmware_info **ucode_list, int ucode_count); #endif diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_rap.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_rap.c index 8da5356c36f1..51909bf8798c 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_rap.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_rap.c @@ -48,6 +48,7 @@ static ssize_t amdgpu_rap_debugfs_write(struct file *f, const char __user *buf, struct ta_rap_cmd_output_data *rap_cmd_output; struct drm_device *dev = adev_to_drm(adev); uint32_t op; + enum ta_rap_status status; int ret; if (*pos || size != 2) @@ -70,9 +71,8 @@ static ssize_t amdgpu_rap_debugfs_write(struct file *f, const char __user *buf, switch (op) { case 2: - ret = psp_rap_invoke(&adev->psp, op); - - if (ret == TA_RAP_STATUS__SUCCESS) { + ret = psp_rap_invoke(&adev->psp, op, &status); + if (!ret && status == TA_RAP_STATUS__SUCCESS) { dev_info(adev->dev, "RAP L0 validate test success.\n"); } else { rap_shared_mem = (struct ta_rap_shared_memory *) @@ -97,6 +97,7 @@ static ssize_t amdgpu_rap_debugfs_write(struct file *f, const char __user *buf, default: dev_info(adev->dev, "Unsupported op id: %d, ", op); dev_info(adev->dev, "Only support op 2(L0 validate test).\n"); + break; } amdgpu_gfx_off_ctrl(adev, true); diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c index 1fb2a91ad30a..b0d2fc9454ca 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c @@ -99,6 +99,49 @@ static bool amdgpu_ras_get_error_query_ready(struct amdgpu_device *adev) return false; } +static int amdgpu_reserve_page_direct(struct amdgpu_device *adev, uint64_t address) +{ + struct ras_err_data err_data = {0, 0, 0, NULL}; + struct eeprom_table_record err_rec; + + if ((address >= adev->gmc.mc_vram_size) || + (address >= RAS_UMC_INJECT_ADDR_LIMIT)) { + dev_warn(adev->dev, + "RAS WARN: input address 0x%llx is invalid.\n", + address); + return -EINVAL; + } + + if (amdgpu_ras_check_bad_page(adev, address)) { + dev_warn(adev->dev, + "RAS WARN: 0x%llx has already been marked as bad page!\n", + address); + return 0; + } + + memset(&err_rec, 0x0, sizeof(struct eeprom_table_record)); + + err_rec.address = address; + err_rec.retired_page = address >> AMDGPU_GPU_PAGE_SHIFT; + err_rec.ts = (uint64_t)ktime_get_real_seconds(); + err_rec.err_type = AMDGPU_RAS_EEPROM_ERR_NON_RECOVERABLE; + + err_data.err_addr = &err_rec; + err_data.err_addr_cnt = 1; + + if (amdgpu_bad_page_threshold != 0) { + amdgpu_ras_add_bad_pages(adev, err_data.err_addr, + err_data.err_addr_cnt); + amdgpu_ras_save_bad_pages(adev); + } + + dev_warn(adev->dev, "WARNING: THIS IS ONLY FOR TEST PURPOSES AND WILL CORRUPT RAS EEPROM\n"); + dev_warn(adev->dev, "Clear EEPROM:\n"); + dev_warn(adev->dev, " echo 1 > /sys/kernel/debug/dri/0/ras/ras_eeprom_reset\n"); + + return 0; +} + static ssize_t amdgpu_ras_debugfs_read(struct file *f, char __user *buf, size_t size, loff_t *pos) { @@ -109,7 +152,7 @@ static ssize_t amdgpu_ras_debugfs_read(struct file *f, char __user *buf, ssize_t s; char val[128]; - if (amdgpu_ras_error_query(obj->adev, &info)) + if (amdgpu_ras_query_error_status(obj->adev, &info)) return -EINVAL; s = snprintf(val, sizeof(val), "%s: %lu\n%s: %lu\n", @@ -178,11 +221,24 @@ static int amdgpu_ras_debugfs_ctrl_parse_data(struct file *f, op = 1; else if (sscanf(str, "inject %32s %8s", block_name, err) == 2) op = 2; + else if (strstr(str, "retire_page") != NULL) + op = 3; else if (str[0] && str[1] && str[2] && str[3]) /* ascii string, but commands are not matched. */ return -EINVAL; if (op != -1) { + if (op == 3) { + if (sscanf(str, "%*s 0x%llx", &address) != 1 && + sscanf(str, "%*s %llu", &address) != 1) + return -EINVAL; + + data->op = op; + data->inject.address = address; + + return 0; + } + if (amdgpu_ras_find_block_id_by_name(block_name, &block_id)) return -EINVAL; @@ -198,11 +254,11 @@ static int amdgpu_ras_debugfs_ctrl_parse_data(struct file *f, data->op = op; if (op == 2) { - if (sscanf(str, "%*s %*s %*s %u %llu %llu", - &sub_block, &address, &value) != 3) - if (sscanf(str, "%*s %*s %*s 0x%x 0x%llx 0x%llx", - &sub_block, &address, &value) != 3) - return -EINVAL; + if (sscanf(str, "%*s %*s %*s 0x%x 0x%llx 0x%llx", + &sub_block, &address, &value) != 3 && + sscanf(str, "%*s %*s %*s %u %llu %llu", + &sub_block, &address, &value) != 3) + return -EINVAL; data->head.sub_block_index = sub_block; data->inject.address = address; data->inject.value = value; @@ -221,7 +277,7 @@ static int amdgpu_ras_debugfs_ctrl_parse_data(struct file *f, /** * DOC: AMDGPU RAS debugfs control interface * - * It accepts struct ras_debug_if who has two members. + * The control interface accepts struct ras_debug_if which has two members. * * First member: ras_debug_if::head or ras_debug_if::inject. * @@ -246,32 +302,33 @@ static int amdgpu_ras_debugfs_ctrl_parse_data(struct file *f, * * How to use the interface? * - * Programs + * In a program * - * Copy the struct ras_debug_if in your codes and initialize it. - * Write the struct to the control node. + * Copy the struct ras_debug_if in your code and initialize it. + * Write the struct to the control interface. * - * Shells + * From shell * * .. code-block:: bash * - * echo op block [error [sub_block address value]] > .../ras/ras_ctrl + * echo "disable <block>" > /sys/kernel/debug/dri/<N>/ras/ras_ctrl + * echo "enable <block> <error>" > /sys/kernel/debug/dri/<N>/ras/ras_ctrl + * echo "inject <block> <error> <sub-block> <address> <value> > /sys/kernel/debug/dri/<N>/ras/ras_ctrl * - * Parameters: + * Where N, is the card which you want to affect. * - * op: disable, enable, inject - * disable: only block is needed - * enable: block and error are needed - * inject: error, address, value are needed - * block: umc, sdma, gfx, ......... + * "disable" requires only the block. + * "enable" requires the block and error type. + * "inject" requires the block, error type, address, and value. + * The block is one of: umc, sdma, gfx, etc. * see ras_block_string[] for details - * error: ue, ce - * ue: multi_uncorrectable - * ce: single_correctable - * sub_block: - * sub block index, pass 0 if there is no sub block + * The error type is one of: ue, ce, where, + * ue is multi-uncorrectable + * ce is single-correctable + * The sub-block is a the sub-block index, pass 0 if there is no sub-block. + * The address and value are hexadecimal numbers, leading 0x is optional. * - * here are some examples for bash commands: + * For instance, * * .. code-block:: bash * @@ -279,17 +336,17 @@ static int amdgpu_ras_debugfs_ctrl_parse_data(struct file *f, * echo inject umc ce 0 0 0 > /sys/kernel/debug/dri/0/ras/ras_ctrl * echo disable umc > /sys/kernel/debug/dri/0/ras/ras_ctrl * - * How to check the result? + * How to check the result of the operation? * - * For disable/enable, please check ras features at + * To check disable/enable, see "ras" features at, * /sys/class/drm/card[0/1/2...]/device/ras/features * - * For inject, please check corresponding err count at - * /sys/class/drm/card[0/1/2...]/device/ras/[gfx/sdma/...]_err_count + * To check inject, see the corresponding error count at, + * /sys/class/drm/card[0/1/2...]/device/ras/[gfx|sdma|umc|...]_err_count * * .. note:: * Operations are only allowed on blocks which are supported. - * Please check ras mask at /sys/module/amdgpu/parameters/ras_mask + * Check the "ras" mask at /sys/module/amdgpu/parameters/ras_mask * to see which blocks support RAS on a particular asic. * */ @@ -310,6 +367,14 @@ static ssize_t amdgpu_ras_debugfs_ctrl_write(struct file *f, const char __user * if (ret) return -EINVAL; + if (data.op == 3) { + ret = amdgpu_reserve_page_direct(adev, data.inject.address); + if (!ret) + return size; + else + return ret; + } + if (!amdgpu_ras_is_supported(adev, data.head.block)) return -EINVAL; @@ -431,15 +496,19 @@ static ssize_t amdgpu_ras_sysfs_read(struct device *dev, }; if (!amdgpu_ras_get_error_query_ready(obj->adev)) - return snprintf(buf, PAGE_SIZE, - "Query currently inaccessible\n"); + return sysfs_emit(buf, "Query currently inaccessible\n"); - if (amdgpu_ras_error_query(obj->adev, &info)) + if (amdgpu_ras_query_error_status(obj->adev, &info)) return -EINVAL; - return snprintf(buf, PAGE_SIZE, "%s: %lu\n%s: %lu\n", - "ue", info.ue_count, - "ce", info.ce_count); + + if (obj->adev->asic_type == CHIP_ALDEBARAN) { + if (amdgpu_ras_reset_error_status(obj->adev, info.head.block)) + DRM_WARN("Failed to reset error counter and error status"); + } + + return sysfs_emit(buf, "%s: %lu\n%s: %lu\n", "ue", info.ue_count, + "ce", info.ce_count); } /* obj begin */ @@ -449,11 +518,10 @@ static ssize_t amdgpu_ras_sysfs_read(struct device *dev, static inline void put_obj(struct ras_manager *obj) { - if (obj && --obj->use == 0) + if (obj && (--obj->use == 0)) list_del(&obj->node); - if (obj && obj->use < 0) { - DRM_ERROR("RAS ERROR: Unbalance obj(%s) use\n", obj->head.name); - } + if (obj && (obj->use < 0)) + DRM_ERROR("RAS ERROR: Unbalance obj(%s) use\n", obj->head.name); } /* make one obj and return it. */ @@ -463,7 +531,7 @@ static struct ras_manager *amdgpu_ras_create_obj(struct amdgpu_device *adev, struct amdgpu_ras *con = amdgpu_ras_get_context(adev); struct ras_manager *obj; - if (!con) + if (!adev->ras_features || !con) return NULL; if (head->block >= AMDGPU_RAS_BLOCK_COUNT) @@ -490,7 +558,7 @@ struct ras_manager *amdgpu_ras_find_obj(struct amdgpu_device *adev, struct ras_manager *obj; int i; - if (!con) + if (!adev->ras_features || !con) return NULL; if (head) { @@ -590,7 +658,11 @@ static int __amdgpu_ras_feature_enable(struct amdgpu_device *adev, con->features |= BIT(head->block); } else { if (obj && amdgpu_ras_is_feature_enabled(adev, head)) { - con->features &= ~BIT(head->block); + /* skip clean gfx ras context feature for VEGA20 Gaming. + * will clean later + */ + if (!(!adev->ras_features && con->features & BIT(AMDGPU_RAS_BLOCK__GFX))) + con->features &= ~BIT(head->block); put_obj(obj); } } @@ -693,6 +765,10 @@ int amdgpu_ras_feature_enable_on_boot(struct amdgpu_device *adev, if (ret) return ret; + /* gfx block ras dsiable cmd must send to ras-ta */ + if (head->block == AMDGPU_RAS_BLOCK__GFX) + con->features |= BIT(head->block); + ret = amdgpu_ras_feature_enable(adev, head, 0); } } else @@ -757,8 +833,8 @@ static int amdgpu_ras_enable_all_features(struct amdgpu_device *adev, /* feature ctl end */ /* query/inject/cure begin */ -int amdgpu_ras_error_query(struct amdgpu_device *adev, - struct ras_query_if *info) +int amdgpu_ras_query_error_status(struct amdgpu_device *adev, + struct ras_query_if *info) { struct ras_manager *obj = amdgpu_ras_find_obj(adev, &info->head); struct ras_err_data err_data = {0, 0, 0, NULL}; @@ -769,13 +845,15 @@ int amdgpu_ras_error_query(struct amdgpu_device *adev, switch (info->head.block) { case AMDGPU_RAS_BLOCK__UMC: - if (adev->umc.funcs->query_ras_error_count) - adev->umc.funcs->query_ras_error_count(adev, &err_data); + if (adev->umc.ras_funcs && + adev->umc.ras_funcs->query_ras_error_count) + adev->umc.ras_funcs->query_ras_error_count(adev, &err_data); /* umc query_ras_error_address is also responsible for clearing * error status */ - if (adev->umc.funcs->query_ras_error_address) - adev->umc.funcs->query_ras_error_address(adev, &err_data); + if (adev->umc.ras_funcs && + adev->umc.ras_funcs->query_ras_error_address) + adev->umc.ras_funcs->query_ras_error_address(adev, &err_data); break; case AMDGPU_RAS_BLOCK__SDMA: if (adev->sdma.funcs->query_ras_error_count) { @@ -785,19 +863,32 @@ int amdgpu_ras_error_query(struct amdgpu_device *adev, } break; case AMDGPU_RAS_BLOCK__GFX: - if (adev->gfx.funcs->query_ras_error_count) - adev->gfx.funcs->query_ras_error_count(adev, &err_data); + if (adev->gfx.ras_funcs && + adev->gfx.ras_funcs->query_ras_error_count) + adev->gfx.ras_funcs->query_ras_error_count(adev, &err_data); + + if (adev->gfx.ras_funcs && + adev->gfx.ras_funcs->query_ras_error_status) + adev->gfx.ras_funcs->query_ras_error_status(adev); break; case AMDGPU_RAS_BLOCK__MMHUB: - if (adev->mmhub.funcs->query_ras_error_count) - adev->mmhub.funcs->query_ras_error_count(adev, &err_data); + if (adev->mmhub.ras_funcs && + adev->mmhub.ras_funcs->query_ras_error_count) + adev->mmhub.ras_funcs->query_ras_error_count(adev, &err_data); + + if (adev->mmhub.ras_funcs && + adev->mmhub.ras_funcs->query_ras_error_status) + adev->mmhub.ras_funcs->query_ras_error_status(adev); break; case AMDGPU_RAS_BLOCK__PCIE_BIF: - if (adev->nbio.funcs->query_ras_error_count) - adev->nbio.funcs->query_ras_error_count(adev, &err_data); + if (adev->nbio.ras_funcs && + adev->nbio.ras_funcs->query_ras_error_count) + adev->nbio.ras_funcs->query_ras_error_count(adev, &err_data); break; case AMDGPU_RAS_BLOCK__XGMI_WAFL: - amdgpu_xgmi_query_ras_error_count(adev, &err_data); + if (adev->gmc.xgmi.ras_funcs && + adev->gmc.xgmi.ras_funcs->query_ras_error_count) + adev->gmc.xgmi.ras_funcs->query_ras_error_count(adev, &err_data); break; default: break; @@ -826,6 +917,38 @@ int amdgpu_ras_error_query(struct amdgpu_device *adev, return 0; } +int amdgpu_ras_reset_error_status(struct amdgpu_device *adev, + enum amdgpu_ras_block block) +{ + if (!amdgpu_ras_is_supported(adev, block)) + return -EINVAL; + + switch (block) { + case AMDGPU_RAS_BLOCK__GFX: + if (adev->gfx.ras_funcs && + adev->gfx.ras_funcs->reset_ras_error_count) + adev->gfx.ras_funcs->reset_ras_error_count(adev); + + if (adev->gfx.ras_funcs && + adev->gfx.ras_funcs->reset_ras_error_status) + adev->gfx.ras_funcs->reset_ras_error_status(adev); + break; + case AMDGPU_RAS_BLOCK__MMHUB: + if (adev->mmhub.ras_funcs && + adev->mmhub.ras_funcs->reset_ras_error_count) + adev->mmhub.ras_funcs->reset_ras_error_count(adev); + break; + case AMDGPU_RAS_BLOCK__SDMA: + if (adev->sdma.funcs->reset_ras_error_count) + adev->sdma.funcs->reset_ras_error_count(adev); + break; + default: + break; + } + + return 0; +} + /* Trigger XGMI/WAFL error */ static int amdgpu_ras_error_inject_xgmi(struct amdgpu_device *adev, struct ta_ras_trigger_error_input *block_info) @@ -878,12 +1001,14 @@ int amdgpu_ras_error_inject(struct amdgpu_device *adev, switch (info->head.block) { case AMDGPU_RAS_BLOCK__GFX: - if (adev->gfx.funcs->ras_error_inject) - ret = adev->gfx.funcs->ras_error_inject(adev, info); + if (adev->gfx.ras_funcs && + adev->gfx.ras_funcs->ras_error_inject) + ret = adev->gfx.ras_funcs->ras_error_inject(adev, info); else ret = -EINVAL; break; case AMDGPU_RAS_BLOCK__UMC: + case AMDGPU_RAS_BLOCK__SDMA: case AMDGPU_RAS_BLOCK__MMHUB: case AMDGPU_RAS_BLOCK__PCIE_BIF: ret = psp_ras_trigger_error(&adev->psp, &block_info); @@ -913,7 +1038,7 @@ unsigned long amdgpu_ras_query_error_count(struct amdgpu_device *adev, struct ras_manager *obj; struct ras_err_data data = {0, 0}; - if (!con) + if (!adev->ras_features || !con) return 0; list_for_each_entry(obj, &con->head, node) { @@ -921,7 +1046,7 @@ unsigned long amdgpu_ras_query_error_count(struct amdgpu_device *adev, .head = obj->head, }; - if (amdgpu_ras_error_query(adev, &info)) + if (amdgpu_ras_query_error_status(adev, &info)) return 0; data.ce_count += info.ce_count; @@ -1137,16 +1262,19 @@ static int amdgpu_ras_sysfs_remove_all(struct amdgpu_device *adev) * */ /* debugfs begin */ -static void amdgpu_ras_debugfs_create_ctrl_node(struct amdgpu_device *adev) +static struct dentry *amdgpu_ras_debugfs_create_ctrl_node(struct amdgpu_device *adev) { struct amdgpu_ras *con = amdgpu_ras_get_context(adev); + struct dentry *dir; struct drm_minor *minor = adev_to_drm(adev)->primary; - con->dir = debugfs_create_dir(RAS_FS_NAME, minor->debugfs_root); - debugfs_create_file("ras_ctrl", S_IWUGO | S_IRUGO, con->dir, - adev, &amdgpu_ras_debugfs_ctrl_ops); - debugfs_create_file("ras_eeprom_reset", S_IWUGO | S_IRUGO, con->dir, - adev, &amdgpu_ras_debugfs_eeprom_ops); + dir = debugfs_create_dir(RAS_FS_NAME, minor->debugfs_root); + debugfs_create_file("ras_ctrl", S_IWUGO | S_IRUGO, dir, adev, + &amdgpu_ras_debugfs_ctrl_ops); + debugfs_create_file("ras_eeprom_reset", S_IWUGO | S_IRUGO, dir, adev, + &amdgpu_ras_debugfs_eeprom_ops); + debugfs_create_u32("bad_page_cnt_threshold", 0444, dir, + &con->bad_page_cnt_threshold); /* * After one uncorrectable error happens, usually GPU recovery will @@ -1156,24 +1284,24 @@ static void amdgpu_ras_debugfs_create_ctrl_node(struct amdgpu_device *adev) * ERREVENT_ATHUB_INTERRUPT generated. Normal GPU recovery routine * will never be called. */ - debugfs_create_bool("auto_reboot", S_IWUGO | S_IRUGO, con->dir, - &con->reboot); + debugfs_create_bool("auto_reboot", S_IWUGO | S_IRUGO, dir, &con->reboot); /* * User could set this not to clean up hardware's error count register * of RAS IPs during ras recovery. */ - debugfs_create_bool("disable_ras_err_cnt_harvest", 0644, - con->dir, &con->disable_ras_err_cnt_harvest); + debugfs_create_bool("disable_ras_err_cnt_harvest", 0644, dir, + &con->disable_ras_err_cnt_harvest); + return dir; } static void amdgpu_ras_debugfs_create(struct amdgpu_device *adev, - struct ras_fs_if *head) + struct ras_fs_if *head, + struct dentry *dir) { - struct amdgpu_ras *con = amdgpu_ras_get_context(adev); struct ras_manager *obj = amdgpu_ras_find_obj(adev, &head->head); - if (!obj || obj->ent) + if (!obj || !dir) return; get_obj(obj); @@ -1182,14 +1310,14 @@ static void amdgpu_ras_debugfs_create(struct amdgpu_device *adev, head->debugfs_name, sizeof(obj->fs_data.debugfs_name)); - obj->ent = debugfs_create_file(obj->fs_data.debugfs_name, - S_IWUGO | S_IRUGO, con->dir, obj, - &amdgpu_ras_debugfs_ops); + debugfs_create_file(obj->fs_data.debugfs_name, S_IWUGO | S_IRUGO, dir, + obj, &amdgpu_ras_debugfs_ops); } void amdgpu_ras_debugfs_create_all(struct amdgpu_device *adev) { struct amdgpu_ras *con = amdgpu_ras_get_context(adev); + struct dentry *dir; struct ras_manager *obj; struct ras_fs_if fs_info; @@ -1200,7 +1328,7 @@ void amdgpu_ras_debugfs_create_all(struct amdgpu_device *adev) if (!IS_ENABLED(CONFIG_DEBUG_FS) || !con) return; - amdgpu_ras_debugfs_create_ctrl_node(adev); + dir = amdgpu_ras_debugfs_create_ctrl_node(adev); list_for_each_entry(obj, &con->head, node) { if (amdgpu_ras_is_supported(adev, obj->head.block) && @@ -1208,34 +1336,11 @@ void amdgpu_ras_debugfs_create_all(struct amdgpu_device *adev) sprintf(fs_info.debugfs_name, "%s_err_inject", ras_block_str(obj->head.block)); fs_info.head = obj->head; - amdgpu_ras_debugfs_create(adev, &fs_info); + amdgpu_ras_debugfs_create(adev, &fs_info, dir); } } } -static void amdgpu_ras_debugfs_remove(struct amdgpu_device *adev, - struct ras_common_if *head) -{ - struct ras_manager *obj = amdgpu_ras_find_obj(adev, head); - - if (!obj || !obj->ent) - return; - - obj->ent = NULL; - put_obj(obj); -} - -static void amdgpu_ras_debugfs_remove_all(struct amdgpu_device *adev) -{ - struct amdgpu_ras *con = amdgpu_ras_get_context(adev); - struct ras_manager *obj, *tmp; - - list_for_each_entry_safe(obj, tmp, &con->head, node) { - amdgpu_ras_debugfs_remove(adev, &obj->head); - } - - con->dir = NULL; -} /* debugfs end */ /* ras fs */ @@ -1282,8 +1387,17 @@ static int amdgpu_ras_fs_init(struct amdgpu_device *adev) static int amdgpu_ras_fs_fini(struct amdgpu_device *adev) { - if (IS_ENABLED(CONFIG_DEBUG_FS)) - amdgpu_ras_debugfs_remove_all(adev); + struct amdgpu_ras *con = amdgpu_ras_get_context(adev); + struct ras_manager *con_obj, *ip_obj, *tmp; + + if (IS_ENABLED(CONFIG_DEBUG_FS)) { + list_for_each_entry_safe(con_obj, tmp, &con->head, node) { + ip_obj = amdgpu_ras_find_obj(adev, &con_obj->head); + if (ip_obj) + put_obj(ip_obj); + } + } + amdgpu_ras_sysfs_remove_all(adev); return 0; } @@ -1447,7 +1561,7 @@ static void amdgpu_ras_log_on_err_counter(struct amdgpu_device *adev) struct amdgpu_ras *con = amdgpu_ras_get_context(adev); struct ras_manager *obj; - if (!con) + if (!adev->ras_features || !con) return; list_for_each_entry(obj, &con->head, node) { @@ -1464,7 +1578,7 @@ static void amdgpu_ras_log_on_err_counter(struct amdgpu_device *adev) if (info.head.block == AMDGPU_RAS_BLOCK__PCIE_BIF) continue; - amdgpu_ras_error_query(adev, &info); + amdgpu_ras_query_error_status(adev, &info); } } @@ -1478,12 +1592,14 @@ static void amdgpu_ras_error_status_query(struct amdgpu_device *adev, */ switch (info->head.block) { case AMDGPU_RAS_BLOCK__GFX: - if (adev->gfx.funcs->query_ras_error_status) - adev->gfx.funcs->query_ras_error_status(adev); + if (adev->gfx.ras_funcs && + adev->gfx.ras_funcs->query_ras_error_status) + adev->gfx.ras_funcs->query_ras_error_status(adev); break; case AMDGPU_RAS_BLOCK__MMHUB: - if (adev->mmhub.funcs->query_ras_error_status) - adev->mmhub.funcs->query_ras_error_status(adev); + if (adev->mmhub.ras_funcs && + adev->mmhub.ras_funcs->query_ras_error_status) + adev->mmhub.ras_funcs->query_ras_error_status(adev); break; default: break; @@ -1495,7 +1611,7 @@ static void amdgpu_ras_query_err_status(struct amdgpu_device *adev) struct amdgpu_ras *con = amdgpu_ras_get_context(adev); struct ras_manager *obj; - if (!con) + if (!adev->ras_features || !con) return; list_for_each_entry(obj, &con->head, node) { @@ -1809,7 +1925,7 @@ int amdgpu_ras_recovery_init(struct amdgpu_device *adev) bool exc_err_limit = false; int ret; - if (con) + if (adev->ras_features && con) data = &con->eh_data; else return 0; @@ -1828,6 +1944,12 @@ int amdgpu_ras_recovery_init(struct amdgpu_device *adev) max_eeprom_records_len = amdgpu_ras_eeprom_get_record_max_length(); amdgpu_ras_validate_threshold(adev, max_eeprom_records_len); + /* Todo: During test the SMU might fail to read the eeprom through I2C + * when the GPU is pending on XGMI reset during probe time + * (Mostly after second bus reset), skip it now + */ + if (adev->gmc.xgmi.pending_reset) + return 0; ret = amdgpu_ras_eeprom_init(&con->eeprom_control, &exc_err_limit); /* * This calling fails when exc_err_limit is true or @@ -1897,15 +2019,13 @@ int amdgpu_ras_request_reset_on_boot(struct amdgpu_device *adev, return 0; } -static int amdgpu_ras_check_asic_type(struct amdgpu_device *adev) +static bool amdgpu_ras_asic_supported(struct amdgpu_device *adev) { - if (adev->asic_type != CHIP_VEGA10 && - adev->asic_type != CHIP_VEGA20 && - adev->asic_type != CHIP_ARCTURUS && - adev->asic_type != CHIP_SIENNA_CICHLID) - return 1; - else - return 0; + return adev->asic_type == CHIP_VEGA10 || + adev->asic_type == CHIP_VEGA20 || + adev->asic_type == CHIP_ARCTURUS || + adev->asic_type == CHIP_ALDEBARAN || + adev->asic_type == CHIP_SIENNA_CICHLID; } /* @@ -1924,22 +2044,32 @@ static void amdgpu_ras_check_supported(struct amdgpu_device *adev, *supported = 0; if (amdgpu_sriov_vf(adev) || !adev->is_atom_fw || - amdgpu_ras_check_asic_type(adev)) + !amdgpu_ras_asic_supported(adev)) return; - if (amdgpu_atomfirmware_mem_ecc_supported(adev)) { - dev_info(adev->dev, "HBM ECC is active.\n"); - *hw_supported |= (1 << AMDGPU_RAS_BLOCK__UMC | - 1 << AMDGPU_RAS_BLOCK__DF); - } else - dev_info(adev->dev, "HBM ECC is not presented.\n"); + if (!adev->gmc.xgmi.connected_to_cpu) { + if (amdgpu_atomfirmware_mem_ecc_supported(adev)) { + dev_info(adev->dev, "MEM ECC is active.\n"); + *hw_supported |= (1 << AMDGPU_RAS_BLOCK__UMC | + 1 << AMDGPU_RAS_BLOCK__DF); + } else { + dev_info(adev->dev, "MEM ECC is not presented.\n"); + } - if (amdgpu_atomfirmware_sram_ecc_supported(adev)) { - dev_info(adev->dev, "SRAM ECC is active.\n"); - *hw_supported |= ~(1 << AMDGPU_RAS_BLOCK__UMC | - 1 << AMDGPU_RAS_BLOCK__DF); - } else - dev_info(adev->dev, "SRAM ECC is not presented.\n"); + if (amdgpu_atomfirmware_sram_ecc_supported(adev)) { + dev_info(adev->dev, "SRAM ECC is active.\n"); + *hw_supported |= ~(1 << AMDGPU_RAS_BLOCK__UMC | + 1 << AMDGPU_RAS_BLOCK__DF); + } else { + dev_info(adev->dev, "SRAM ECC is not presented.\n"); + } + } else { + /* driver only manages a few IP blocks RAS feature + * when GPU is connected cpu through XGMI */ + *hw_supported |= (1 << AMDGPU_RAS_BLOCK__GFX | + 1 << AMDGPU_RAS_BLOCK__SDMA | + 1 << AMDGPU_RAS_BLOCK__MMHUB); + } /* hw_supported needs to be aligned with RAS block mask. */ *hw_supported &= AMDGPU_RAS_BLOCK_MASK; @@ -1970,6 +2100,15 @@ int amdgpu_ras_init(struct amdgpu_device *adev) amdgpu_ras_check_supported(adev, &con->hw_supported, &con->supported); if (!con->hw_supported || (adev->asic_type == CHIP_VEGA10)) { + /* set gfx block ras context feature for VEGA20 Gaming + * send ras disable cmd to ras ta during ras late init. + */ + if (!adev->ras_features && adev->asic_type == CHIP_VEGA20) { + con->features |= BIT(AMDGPU_RAS_BLOCK__GFX); + + return 0; + } + r = 0; goto release_con; } @@ -1979,14 +2118,31 @@ int amdgpu_ras_init(struct amdgpu_device *adev) /* Might need get this flag from vbios. */ con->flags = RAS_DEFAULT_FLAGS; - if (adev->nbio.funcs->init_ras_controller_interrupt) { - r = adev->nbio.funcs->init_ras_controller_interrupt(adev); + /* initialize nbio ras function ahead of any other + * ras functions so hardware fatal error interrupt + * can be enabled as early as possible */ + switch (adev->asic_type) { + case CHIP_VEGA20: + case CHIP_ARCTURUS: + case CHIP_ALDEBARAN: + if (!adev->gmc.xgmi.connected_to_cpu) + adev->nbio.ras_funcs = &nbio_v7_4_ras_funcs; + break; + default: + /* nbio ras is not available */ + break; + } + + if (adev->nbio.ras_funcs && + adev->nbio.ras_funcs->init_ras_controller_interrupt) { + r = adev->nbio.ras_funcs->init_ras_controller_interrupt(adev); if (r) goto release_con; } - if (adev->nbio.funcs->init_ras_err_event_athub_interrupt) { - r = adev->nbio.funcs->init_ras_err_event_athub_interrupt(adev); + if (adev->nbio.ras_funcs && + adev->nbio.ras_funcs->init_ras_err_event_athub_interrupt) { + r = adev->nbio.ras_funcs->init_ras_err_event_athub_interrupt(adev); if (r) goto release_con; } @@ -2007,6 +2163,32 @@ release_con: return r; } +static int amdgpu_persistent_edc_harvesting_supported(struct amdgpu_device *adev) +{ + if (adev->gmc.xgmi.connected_to_cpu) + return 1; + return 0; +} + +static int amdgpu_persistent_edc_harvesting(struct amdgpu_device *adev, + struct ras_common_if *ras_block) +{ + struct ras_query_if info = { + .head = *ras_block, + }; + + if (!amdgpu_persistent_edc_harvesting_supported(adev)) + return 0; + + if (amdgpu_ras_query_error_status(adev, &info) != 0) + DRM_WARN("RAS init harvest failure"); + + if (amdgpu_ras_reset_error_status(adev, ras_block->block) != 0) + DRM_WARN("RAS init harvest reset failure"); + + return 0; +} + /* helper function to handle common stuff in ip late init phase */ int amdgpu_ras_late_init(struct amdgpu_device *adev, struct ras_common_if *ras_block, @@ -2036,6 +2218,9 @@ int amdgpu_ras_late_init(struct amdgpu_device *adev, return r; } + /* check for errors on warm reset edc persisant supported ASIC */ + amdgpu_persistent_edc_harvesting(adev, ras_block); + /* in resume phase, no need to create ras fs node */ if (adev->in_suspend || amdgpu_in_reset(adev)) return 0; @@ -2083,8 +2268,12 @@ void amdgpu_ras_resume(struct amdgpu_device *adev) struct amdgpu_ras *con = amdgpu_ras_get_context(adev); struct ras_manager *obj, *tmp; - if (!con) + if (!adev->ras_features || !con) { + /* clean ras context for VEGA20 Gaming after send ras disable cmd */ + amdgpu_release_ras_context(adev); + return; + } if (con->flags & AMDGPU_RAS_FLAG_INIT_BY_VBIOS) { /* Set up all other IPs which are not implemented. There is a @@ -2125,7 +2314,7 @@ void amdgpu_ras_suspend(struct amdgpu_device *adev) { struct amdgpu_ras *con = amdgpu_ras_get_context(adev); - if (!con) + if (!adev->ras_features || !con) return; amdgpu_ras_disable_all_features(adev, 0); @@ -2139,7 +2328,7 @@ int amdgpu_ras_pre_fini(struct amdgpu_device *adev) { struct amdgpu_ras *con = amdgpu_ras_get_context(adev); - if (!con) + if (!adev->ras_features || !con) return 0; /* Need disable ras on all IPs here before ip [hw/sw]fini */ @@ -2152,7 +2341,7 @@ int amdgpu_ras_fini(struct amdgpu_device *adev) { struct amdgpu_ras *con = amdgpu_ras_get_context(adev); - if (!con) + if (!adev->ras_features || !con) return 0; amdgpu_ras_fs_fini(adev); @@ -2196,18 +2385,16 @@ bool amdgpu_ras_need_emergency_restart(struct amdgpu_device *adev) return false; } -bool amdgpu_ras_check_err_threshold(struct amdgpu_device *adev) +void amdgpu_release_ras_context(struct amdgpu_device *adev) { struct amdgpu_ras *con = amdgpu_ras_get_context(adev); - bool exc_err_limit = false; - if (con && (amdgpu_bad_page_threshold != 0)) - amdgpu_ras_eeprom_check_err_threshold(&con->eeprom_control, - &exc_err_limit); + if (!con) + return; - /* - * We are only interested in variable exc_err_limit, - * as it says if GPU is in bad state or not. - */ - return exc_err_limit; + if (!adev->ras_features && con->features & BIT(AMDGPU_RAS_BLOCK__GFX)) { + con->features &= ~BIT(AMDGPU_RAS_BLOCK__GFX); + amdgpu_ras_set_context(adev, NULL); + kfree(con); + } } diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.h index 762f5e46c007..60df268a0c66 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.h @@ -318,8 +318,6 @@ struct amdgpu_ras { uint32_t supported; uint32_t features; struct list_head head; - /* debugfs */ - struct dentry *dir; /* sysfs */ struct device_attribute features_attr; struct bin_attribute badpages_attr; @@ -395,8 +393,6 @@ struct ras_manager { struct list_head node; /* the device */ struct amdgpu_device *adev; - /* debugfs */ - struct dentry *ent; /* sysfs */ struct device_attribute sysfs_attr; int attr_inuse; @@ -495,8 +491,6 @@ void amdgpu_ras_suspend(struct amdgpu_device *adev); unsigned long amdgpu_ras_query_error_count(struct amdgpu_device *adev, bool is_ce); -bool amdgpu_ras_check_err_threshold(struct amdgpu_device *adev); - /* error handling functions */ int amdgpu_ras_add_bad_pages(struct amdgpu_device *adev, struct eeprom_table_record *bps, int pages); @@ -594,9 +588,12 @@ int amdgpu_ras_sysfs_remove(struct amdgpu_device *adev, void amdgpu_ras_debugfs_create_all(struct amdgpu_device *adev); -int amdgpu_ras_error_query(struct amdgpu_device *adev, +int amdgpu_ras_query_error_status(struct amdgpu_device *adev, struct ras_query_if *info); +int amdgpu_ras_reset_error_status(struct amdgpu_device *adev, + enum amdgpu_ras_block block); + int amdgpu_ras_error_inject(struct amdgpu_device *adev, struct ras_inject_if *info); @@ -629,4 +626,6 @@ void amdgpu_ras_global_ras_isr(struct amdgpu_device *adev); void amdgpu_ras_set_error_query_ready(struct amdgpu_device *adev, bool ready); bool amdgpu_ras_need_emergency_restart(struct amdgpu_device *adev); + +void amdgpu_release_ras_context(struct amdgpu_device *adev); #endif diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras_eeprom.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras_eeprom.c index 19d9aa76cfbf..f40c871da0c6 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras_eeprom.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras_eeprom.c @@ -31,6 +31,7 @@ #define EEPROM_I2C_TARGET_ADDR_ARCTURUS 0xA8 #define EEPROM_I2C_TARGET_ADDR_ARCTURUS_D342 0xA0 #define EEPROM_I2C_TARGET_ADDR_SIENNA_CICHLID 0xA0 +#define EEPROM_I2C_TARGET_ADDR_ALDEBARAN 0xA0 /* * The 2 macros bellow represent the actual size in bytes that @@ -64,7 +65,8 @@ static bool __is_ras_eeprom_supported(struct amdgpu_device *adev) { if ((adev->asic_type == CHIP_VEGA20) || (adev->asic_type == CHIP_ARCTURUS) || - (adev->asic_type == CHIP_SIENNA_CICHLID)) + (adev->asic_type == CHIP_SIENNA_CICHLID) || + (adev->asic_type == CHIP_ALDEBARAN)) return true; return false; @@ -106,6 +108,10 @@ static bool __get_eeprom_i2c_addr(struct amdgpu_device *adev, *i2c_addr = EEPROM_I2C_TARGET_ADDR_SIENNA_CICHLID; break; + case CHIP_ALDEBARAN: + *i2c_addr = EEPROM_I2C_TARGET_ADDR_ALDEBARAN; + break; + default: return false; } @@ -434,47 +440,28 @@ static uint32_t __correct_eeprom_dest_address(uint32_t curr_address) return curr_address; } -int amdgpu_ras_eeprom_check_err_threshold( - struct amdgpu_ras_eeprom_control *control, - bool *exceed_err_limit) +bool amdgpu_ras_eeprom_check_err_threshold(struct amdgpu_device *adev) { - struct amdgpu_device *adev = to_amdgpu_device(control); - unsigned char buff[EEPROM_ADDRESS_SIZE + - EEPROM_TABLE_HEADER_SIZE] = { 0 }; - struct amdgpu_ras_eeprom_table_header *hdr = &control->tbl_hdr; - struct i2c_msg msg = { - .addr = control->i2c_address, - .flags = I2C_M_RD, - .len = EEPROM_ADDRESS_SIZE + EEPROM_TABLE_HEADER_SIZE, - .buf = buff, - }; - int ret; - - *exceed_err_limit = false; + struct amdgpu_ras *con = amdgpu_ras_get_context(adev); if (!__is_ras_eeprom_supported(adev)) - return 0; - - /* read EEPROM table header */ - mutex_lock(&control->tbl_mutex); - ret = i2c_transfer(&adev->pm.smu_i2c, &msg, 1); - if (ret < 1) { - dev_err(adev->dev, "Failed to read EEPROM table header.\n"); - goto err; - } + return false; - __decode_table_header_from_buff(hdr, &buff[2]); + /* skip check eeprom table for VEGA20 Gaming */ + if (!con) + return false; + else + if (!(con->features & BIT(AMDGPU_RAS_BLOCK__UMC))) + return false; - if (hdr->header == EEPROM_TABLE_HDR_BAD) { + if (con->eeprom_control.tbl_hdr.header == EEPROM_TABLE_HDR_BAD) { dev_warn(adev->dev, "This GPU is in BAD status."); dev_warn(adev->dev, "Please retire it or setting one bigger " "threshold value when reloading driver.\n"); - *exceed_err_limit = true; + return true; } -err: - mutex_unlock(&control->tbl_mutex); - return 0; + return false; } int amdgpu_ras_eeprom_process_recods(struct amdgpu_ras_eeprom_control *control, diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras_eeprom.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras_eeprom.h index c7a5e5c7c61e..178721170974 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras_eeprom.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras_eeprom.h @@ -80,9 +80,7 @@ int amdgpu_ras_eeprom_init(struct amdgpu_ras_eeprom_control *control, bool *exceed_err_limit); int amdgpu_ras_eeprom_reset_table(struct amdgpu_ras_eeprom_control *control); -int amdgpu_ras_eeprom_check_err_threshold( - struct amdgpu_ras_eeprom_control *control, - bool *exceed_err_limit); +bool amdgpu_ras_eeprom_check_err_threshold(struct amdgpu_device *adev); int amdgpu_ras_eeprom_process_recods(struct amdgpu_ras_eeprom_control *control, struct eeprom_table_record *records, diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_res_cursor.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_res_cursor.h new file mode 100644 index 000000000000..40f2adf305bc --- /dev/null +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_res_cursor.h @@ -0,0 +1,105 @@ +// SPDX-License-Identifier: GPL-2.0 OR MIT +/* + * Copyright 2020 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + * Authors: Christian König + */ + +#ifndef __AMDGPU_RES_CURSOR_H__ +#define __AMDGPU_RES_CURSOR_H__ + +#include <drm/drm_mm.h> +#include <drm/ttm/ttm_resource.h> + +/* state back for walking over vram_mgr and gtt_mgr allocations */ +struct amdgpu_res_cursor { + uint64_t start; + uint64_t size; + uint64_t remaining; + struct drm_mm_node *node; +}; + +/** + * amdgpu_res_first - initialize a amdgpu_res_cursor + * + * @res: TTM resource object to walk + * @start: Start of the range + * @size: Size of the range + * @cur: cursor object to initialize + * + * Start walking over the range of allocations between @start and @size. + */ +static inline void amdgpu_res_first(struct ttm_resource *res, + uint64_t start, uint64_t size, + struct amdgpu_res_cursor *cur) +{ + struct drm_mm_node *node; + + if (!res || !res->mm_node) { + cur->start = start; + cur->size = size; + cur->remaining = size; + cur->node = NULL; + return; + } + + BUG_ON(start + size > res->num_pages << PAGE_SHIFT); + + node = res->mm_node; + while (start >= node->size << PAGE_SHIFT) + start -= node++->size << PAGE_SHIFT; + + cur->start = (node->start << PAGE_SHIFT) + start; + cur->size = min((node->size << PAGE_SHIFT) - start, size); + cur->remaining = size; + cur->node = node; +} + +/** + * amdgpu_res_next - advance the cursor + * + * @cur: the cursor to advance + * @size: number of bytes to move forward + * + * Move the cursor @size bytes forwrad, walking to the next node if necessary. + */ +static inline void amdgpu_res_next(struct amdgpu_res_cursor *cur, uint64_t size) +{ + struct drm_mm_node *node = cur->node; + + BUG_ON(size > cur->remaining); + + cur->remaining -= size; + if (!cur->remaining) + return; + + cur->size -= size; + if (cur->size) { + cur->start += size; + return; + } + + cur->node = ++node; + cur->start = node->start << PAGE_SHIFT; + cur->size = min(node->size << PAGE_SHIFT, cur->remaining); +} + +#endif diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_reset.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_reset.c new file mode 100644 index 000000000000..02afd4115675 --- /dev/null +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_reset.c @@ -0,0 +1,98 @@ +/* + * Copyright 2021 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + */ + +#include "amdgpu_reset.h" +#include "aldebaran.h" + +int amdgpu_reset_add_handler(struct amdgpu_reset_control *reset_ctl, + struct amdgpu_reset_handler *handler) +{ + /* TODO: Check if handler exists? */ + list_add_tail(&handler->handler_list, &reset_ctl->reset_handlers); + return 0; +} + +int amdgpu_reset_init(struct amdgpu_device *adev) +{ + int ret = 0; + + switch (adev->asic_type) { + case CHIP_ALDEBARAN: + ret = aldebaran_reset_init(adev); + break; + default: + break; + } + + return ret; +} + +int amdgpu_reset_fini(struct amdgpu_device *adev) +{ + int ret = 0; + + switch (adev->asic_type) { + case CHIP_ALDEBARAN: + ret = aldebaran_reset_fini(adev); + break; + default: + break; + } + + return ret; +} + +int amdgpu_reset_prepare_hwcontext(struct amdgpu_device *adev, + struct amdgpu_reset_context *reset_context) +{ + struct amdgpu_reset_handler *reset_handler = NULL; + + if (adev->reset_cntl && adev->reset_cntl->get_reset_handler) + reset_handler = adev->reset_cntl->get_reset_handler( + adev->reset_cntl, reset_context); + if (!reset_handler) + return -ENOSYS; + + return reset_handler->prepare_hwcontext(adev->reset_cntl, + reset_context); +} + +int amdgpu_reset_perform_reset(struct amdgpu_device *adev, + struct amdgpu_reset_context *reset_context) +{ + int ret; + struct amdgpu_reset_handler *reset_handler = NULL; + + if (adev->reset_cntl) + reset_handler = adev->reset_cntl->get_reset_handler( + adev->reset_cntl, reset_context); + if (!reset_handler) + return -ENOSYS; + + ret = reset_handler->perform_reset(adev->reset_cntl, reset_context); + if (ret) + return ret; + + return reset_handler->restore_hwcontext(adev->reset_cntl, + reset_context); +} diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_reset.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_reset.h new file mode 100644 index 000000000000..e00d38d9160a --- /dev/null +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_reset.h @@ -0,0 +1,85 @@ +/* + * Copyright 2021 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + */ + +#ifndef __AMDGPU_RESET_H__ +#define __AMDGPU_RESET_H__ + +#include "amdgpu.h" + +enum AMDGPU_RESET_FLAGS { + + AMDGPU_NEED_FULL_RESET = 0, + AMDGPU_SKIP_HW_RESET = 1, +}; + +struct amdgpu_reset_context { + enum amd_reset_method method; + struct amdgpu_device *reset_req_dev; + struct amdgpu_job *job; + struct amdgpu_hive_info *hive; + unsigned long flags; +}; + +struct amdgpu_reset_handler { + enum amd_reset_method reset_method; + struct list_head handler_list; + int (*prepare_env)(struct amdgpu_reset_control *reset_ctl, + struct amdgpu_reset_context *context); + int (*prepare_hwcontext)(struct amdgpu_reset_control *reset_ctl, + struct amdgpu_reset_context *context); + int (*perform_reset)(struct amdgpu_reset_control *reset_ctl, + struct amdgpu_reset_context *context); + int (*restore_hwcontext)(struct amdgpu_reset_control *reset_ctl, + struct amdgpu_reset_context *context); + int (*restore_env)(struct amdgpu_reset_control *reset_ctl, + struct amdgpu_reset_context *context); + + int (*do_reset)(struct amdgpu_device *adev); +}; + +struct amdgpu_reset_control { + void *handle; + struct work_struct reset_work; + struct mutex reset_lock; + struct list_head reset_handlers; + atomic_t in_reset; + enum amd_reset_method active_reset; + struct amdgpu_reset_handler *(*get_reset_handler)( + struct amdgpu_reset_control *reset_ctl, + struct amdgpu_reset_context *context); + void (*async_reset)(struct work_struct *work); +}; + +int amdgpu_reset_init(struct amdgpu_device *adev); +int amdgpu_reset_fini(struct amdgpu_device *adev); + +int amdgpu_reset_prepare_hwcontext(struct amdgpu_device *adev, + struct amdgpu_reset_context *reset_context); + +int amdgpu_reset_perform_reset(struct amdgpu_device *adev, + struct amdgpu_reset_context *reset_context); + +int amdgpu_reset_add_handler(struct amdgpu_reset_control *reset_ctl, + struct amdgpu_reset_handler *handler); + +#endif diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ring.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ring.c index b644c78475fd..688624ebe421 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ring.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ring.c @@ -164,7 +164,8 @@ void amdgpu_ring_undo(struct amdgpu_ring *ring) */ int amdgpu_ring_init(struct amdgpu_device *adev, struct amdgpu_ring *ring, unsigned int max_dw, struct amdgpu_irq_src *irq_src, - unsigned int irq_type, unsigned int hw_prio) + unsigned int irq_type, unsigned int hw_prio, + atomic_t *sched_score) { int r; int sched_hw_submission = amdgpu_sched_hw_submission; @@ -189,7 +190,8 @@ int amdgpu_ring_init(struct amdgpu_device *adev, struct amdgpu_ring *ring, ring->adev = adev; ring->idx = adev->num_rings++; adev->rings[ring->idx] = ring; - r = amdgpu_fence_driver_init_ring(ring, sched_hw_submission); + r = amdgpu_fence_driver_init_ring(ring, sched_hw_submission, + sched_score); if (r) return r; } diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ring.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_ring.h index 56acec1075ac..ca1622835296 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ring.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ring.h @@ -111,7 +111,8 @@ void amdgpu_fence_driver_fini(struct amdgpu_device *adev); void amdgpu_fence_driver_force_completion(struct amdgpu_ring *ring); int amdgpu_fence_driver_init_ring(struct amdgpu_ring *ring, - unsigned num_hw_submission); + unsigned num_hw_submission, + atomic_t *sched_score); int amdgpu_fence_driver_start_ring(struct amdgpu_ring *ring, struct amdgpu_irq_src *irq_src, unsigned irq_type); @@ -282,7 +283,8 @@ void amdgpu_ring_commit(struct amdgpu_ring *ring); void amdgpu_ring_undo(struct amdgpu_ring *ring); int amdgpu_ring_init(struct amdgpu_device *adev, struct amdgpu_ring *ring, unsigned int ring_size, struct amdgpu_irq_src *irq_src, - unsigned int irq_type, unsigned int prio); + unsigned int irq_type, unsigned int prio, + atomic_t *sched_score); void amdgpu_ring_fini(struct amdgpu_ring *ring); void amdgpu_ring_emit_reg_write_reg_wait_helper(struct amdgpu_ring *ring, uint32_t reg0, uint32_t val0, diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_rlc.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_rlc.h index aeaaae713c59..4fc2ce8ce8ab 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_rlc.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_rlc.h @@ -127,7 +127,8 @@ struct amdgpu_rlc_funcs { void (*reset)(struct amdgpu_device *adev); void (*start)(struct amdgpu_device *adev); void (*update_spm_vmid)(struct amdgpu_device *adev, unsigned vmid); - void (*rlcg_wreg)(struct amdgpu_device *adev, u32 offset, u32 v); + void (*rlcg_wreg)(struct amdgpu_device *adev, u32 offset, u32 v, u32 flag); + u32 (*rlcg_rreg)(struct amdgpu_device *adev, u32 offset, u32 flag); bool (*is_rlcg_access_range)(struct amdgpu_device *adev, uint32_t reg); }; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_sdma.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_sdma.h index e5b8fb8e75c5..f8fb755e3aa6 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_sdma.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_sdma.h @@ -64,6 +64,11 @@ struct amdgpu_sdma { struct amdgpu_irq_src trap_irq; struct amdgpu_irq_src illegal_inst_irq; struct amdgpu_irq_src ecc_irq; + struct amdgpu_irq_src vm_hole_irq; + struct amdgpu_irq_src doorbell_invalid_irq; + struct amdgpu_irq_src pool_timeout_irq; + struct amdgpu_irq_src srbm_write_irq; + int num_instances; uint32_t srbm_soft_reset; bool has_page_queue; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_securedisplay.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_securedisplay.c index 834440ab9ff7..5369c8dd0764 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_securedisplay.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_securedisplay.c @@ -69,6 +69,9 @@ void psp_securedisplay_parse_resp_status(struct psp_context *psp, case TA_SECUREDISPLAY_STATUS__READ_CRC_ERROR: dev_err(psp->adev->dev, "Secure display: Failed to Read CRC"); break; + case TA_SECUREDISPLAY_STATUS__I2C_INIT_ERROR: + dev_err(psp->adev->dev, "Secure display: Failed to initialize I2C."); + break; default: dev_err(psp->adev->dev, "Secure display: Failed to parse status: %d\n", status); } @@ -92,9 +95,7 @@ static ssize_t amdgpu_securedisplay_debugfs_write(struct file *f, const char __u struct drm_device *dev = adev_to_drm(adev); uint32_t phy_id; uint32_t op; - int i; char str[64]; - char i2c_output[256]; int ret; if (*pos || size > sizeof(str) - 1) @@ -136,11 +137,9 @@ static ssize_t amdgpu_securedisplay_debugfs_write(struct file *f, const char __u ret = psp_securedisplay_invoke(psp, TA_SECUREDISPLAY_COMMAND__SEND_ROI_CRC); if (!ret) { if (securedisplay_cmd->status == TA_SECUREDISPLAY_STATUS__SUCCESS) { - memset(i2c_output, 0, sizeof(i2c_output)); - for (i = 0; i < TA_SECUREDISPLAY_I2C_BUFFER_SIZE; i++) - sprintf(i2c_output, "%s 0x%X", i2c_output, - securedisplay_cmd->securedisplay_out_message.send_roi_crc.i2c_buf[i]); - dev_info(adev->dev, "SECUREDISPLAY: I2C buffer out put is :%s\n", i2c_output); + dev_info(adev->dev, "SECUREDISPLAY: I2C buffer out put is: %*ph\n", + TA_SECUREDISPLAY_I2C_BUFFER_SIZE, + securedisplay_cmd->securedisplay_out_message.send_roi_crc.i2c_buf); } else { psp_securedisplay_parse_resp_status(psp, securedisplay_cmd->status); } diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_smuio.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_smuio.h index 03009157aec8..b860ec913ac5 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_smuio.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_smuio.h @@ -28,6 +28,8 @@ struct amdgpu_smuio_funcs { u32 (*get_rom_data_offset)(struct amdgpu_device *adev); void (*update_rom_clock_gating)(struct amdgpu_device *adev, bool enable); void (*get_clock_gating_state)(struct amdgpu_device *adev, u32 *flags); + u32 (*get_die_id)(struct amdgpu_device *adev); + bool (*is_host_gpu_xgmi_supported)(struct amdgpu_device *adev); }; struct amdgpu_smuio { diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_test.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_test.c index 7b230bcbf2c6..909d830b513e 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_test.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_test.c @@ -62,6 +62,7 @@ static void amdgpu_do_test_moves(struct amdgpu_device *adev) bp.flags = 0; bp.type = ttm_bo_type_kernel; bp.resv = NULL; + bp.bo_ptr_size = sizeof(struct amdgpu_bo); r = amdgpu_bo_create(adev, &bp, &vram_obj); if (r) { diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c index 5efa331e3ee8..3bef0432cac2 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c @@ -47,7 +47,6 @@ #include <drm/ttm/ttm_bo_driver.h> #include <drm/ttm/ttm_placement.h> -#include <drm/drm_debugfs.h> #include <drm/amdgpu_drm.h> #include "amdgpu.h" @@ -57,14 +56,15 @@ #include "amdgpu_sdma.h" #include "amdgpu_ras.h" #include "amdgpu_atomfirmware.h" +#include "amdgpu_res_cursor.h" #include "bif/bif_4_1_d.h" #define AMDGPU_TTM_VRAM_MAX_DW_READ (size_t)128 -static int amdgpu_ttm_backend_bind(struct ttm_bo_device *bdev, +static int amdgpu_ttm_backend_bind(struct ttm_device *bdev, struct ttm_tt *ttm, struct ttm_resource *bo_mem); -static void amdgpu_ttm_backend_unbind(struct ttm_bo_device *bdev, +static void amdgpu_ttm_backend_unbind(struct ttm_device *bdev, struct ttm_tt *ttm); static int amdgpu_ttm_init_on_chip(struct amdgpu_device *adev, @@ -179,54 +179,11 @@ static int amdgpu_verify_access(struct ttm_buffer_object *bo, struct file *filp) } /** - * amdgpu_mm_node_addr - Compute the GPU relative offset of a GTT buffer. - * - * @bo: The bo to assign the memory to. - * @mm_node: Memory manager node for drm allocator. - * @mem: The region where the bo resides. - * - */ -static uint64_t amdgpu_mm_node_addr(struct ttm_buffer_object *bo, - struct drm_mm_node *mm_node, - struct ttm_resource *mem) -{ - uint64_t addr = 0; - - if (mm_node->start != AMDGPU_BO_INVALID_OFFSET) { - addr = mm_node->start << PAGE_SHIFT; - addr += amdgpu_ttm_domain_start(amdgpu_ttm_adev(bo->bdev), - mem->mem_type); - } - return addr; -} - -/** - * amdgpu_find_mm_node - Helper function finds the drm_mm_node corresponding to - * @offset. It also modifies the offset to be within the drm_mm_node returned - * - * @mem: The region where the bo resides. - * @offset: The offset that drm_mm_node is used for finding. - * - */ -static struct drm_mm_node *amdgpu_find_mm_node(struct ttm_resource *mem, - uint64_t *offset) -{ - struct drm_mm_node *mm_node = mem->mm_node; - - while (*offset >= (mm_node->size << PAGE_SHIFT)) { - *offset -= (mm_node->size << PAGE_SHIFT); - ++mm_node; - } - return mm_node; -} - -/** * amdgpu_ttm_map_buffer - Map memory into the GART windows * @bo: buffer object to map * @mem: memory object to map - * @mm_node: drm_mm node object to map + * @mm_cur: range to map * @num_pages: number of pages to map - * @offset: offset into @mm_node where to start * @window: which GART window to use * @ring: DMA ring to use for the copy * @tmz: if we should setup a TMZ enabled mapping @@ -237,10 +194,10 @@ static struct drm_mm_node *amdgpu_find_mm_node(struct ttm_resource *mem, */ static int amdgpu_ttm_map_buffer(struct ttm_buffer_object *bo, struct ttm_resource *mem, - struct drm_mm_node *mm_node, - unsigned num_pages, uint64_t offset, - unsigned window, struct amdgpu_ring *ring, - bool tmz, uint64_t *addr) + struct amdgpu_res_cursor *mm_cur, + unsigned num_pages, unsigned window, + struct amdgpu_ring *ring, bool tmz, + uint64_t *addr) { struct amdgpu_device *adev = ring->adev; struct amdgpu_job *job; @@ -257,14 +214,15 @@ static int amdgpu_ttm_map_buffer(struct ttm_buffer_object *bo, /* Map only what can't be accessed directly */ if (!tmz && mem->start != AMDGPU_BO_INVALID_OFFSET) { - *addr = amdgpu_mm_node_addr(bo, mm_node, mem) + offset; + *addr = amdgpu_ttm_domain_start(adev, mem->mem_type) + + mm_cur->start; return 0; } *addr = adev->gmc.gart_start; *addr += (u64)window * AMDGPU_GTT_MAX_TRANSFER_SIZE * AMDGPU_GPU_PAGE_SIZE; - *addr += offset & ~PAGE_MASK; + *addr += mm_cur->start & ~PAGE_MASK; num_dw = ALIGN(adev->mman.buffer_funcs->copy_num_dw, 8); num_bytes = num_pages * 8; @@ -292,17 +250,17 @@ static int amdgpu_ttm_map_buffer(struct ttm_buffer_object *bo, cpu_addr = &job->ibs[0].ptr[num_dw]; if (mem->mem_type == TTM_PL_TT) { - dma_addr_t *dma_address; + dma_addr_t *dma_addr; - dma_address = &bo->ttm->dma_address[offset >> PAGE_SHIFT]; - r = amdgpu_gart_map(adev, 0, num_pages, dma_address, flags, + dma_addr = &bo->ttm->dma_address[mm_cur->start >> PAGE_SHIFT]; + r = amdgpu_gart_map(adev, 0, num_pages, dma_addr, flags, cpu_addr); if (r) goto error_free; } else { dma_addr_t dma_address; - dma_address = (mm_node->start << PAGE_SHIFT) + offset; + dma_address = mm_cur->start; dma_address += adev->vm_manager.vram_base_offset; for (i = 0; i < num_pages; ++i) { @@ -354,9 +312,8 @@ int amdgpu_ttm_copy_mem_to_mem(struct amdgpu_device *adev, const uint32_t GTT_MAX_BYTES = (AMDGPU_GTT_MAX_TRANSFER_SIZE * AMDGPU_GPU_PAGE_SIZE); - uint64_t src_node_size, dst_node_size, src_offset, dst_offset; struct amdgpu_ring *ring = adev->mman.buffer_funcs_ring; - struct drm_mm_node *src_mm, *dst_mm; + struct amdgpu_res_cursor src_mm, dst_mm; struct dma_fence *fence = NULL; int r = 0; @@ -365,29 +322,13 @@ int amdgpu_ttm_copy_mem_to_mem(struct amdgpu_device *adev, return -EINVAL; } - src_offset = src->offset; - if (src->mem->mm_node) { - src_mm = amdgpu_find_mm_node(src->mem, &src_offset); - src_node_size = (src_mm->size << PAGE_SHIFT) - src_offset; - } else { - src_mm = NULL; - src_node_size = ULLONG_MAX; - } - - dst_offset = dst->offset; - if (dst->mem->mm_node) { - dst_mm = amdgpu_find_mm_node(dst->mem, &dst_offset); - dst_node_size = (dst_mm->size << PAGE_SHIFT) - dst_offset; - } else { - dst_mm = NULL; - dst_node_size = ULLONG_MAX; - } + amdgpu_res_first(src->mem, src->offset, size, &src_mm); + amdgpu_res_first(dst->mem, dst->offset, size, &dst_mm); mutex_lock(&adev->mman.gtt_window_lock); - - while (size) { - uint32_t src_page_offset = src_offset & ~PAGE_MASK; - uint32_t dst_page_offset = dst_offset & ~PAGE_MASK; + while (src_mm.remaining) { + uint32_t src_page_offset = src_mm.start & ~PAGE_MASK; + uint32_t dst_page_offset = dst_mm.start & ~PAGE_MASK; struct dma_fence *next; uint32_t cur_size; uint64_t from, to; @@ -396,19 +337,19 @@ int amdgpu_ttm_copy_mem_to_mem(struct amdgpu_device *adev, * begins at an offset, then adjust the size accordingly */ cur_size = max(src_page_offset, dst_page_offset); - cur_size = min(min3(src_node_size, dst_node_size, size), + cur_size = min(min3(src_mm.size, dst_mm.size, size), (uint64_t)(GTT_MAX_BYTES - cur_size)); /* Map src to window 0 and dst to window 1. */ - r = amdgpu_ttm_map_buffer(src->bo, src->mem, src_mm, + r = amdgpu_ttm_map_buffer(src->bo, src->mem, &src_mm, PFN_UP(cur_size + src_page_offset), - src_offset, 0, ring, tmz, &from); + 0, ring, tmz, &from); if (r) goto error; - r = amdgpu_ttm_map_buffer(dst->bo, dst->mem, dst_mm, + r = amdgpu_ttm_map_buffer(dst->bo, dst->mem, &dst_mm, PFN_UP(cur_size + dst_page_offset), - dst_offset, 1, ring, tmz, &to); + 1, ring, tmz, &to); if (r) goto error; @@ -420,27 +361,8 @@ int amdgpu_ttm_copy_mem_to_mem(struct amdgpu_device *adev, dma_fence_put(fence); fence = next; - size -= cur_size; - if (!size) - break; - - src_node_size -= cur_size; - if (!src_node_size) { - ++src_mm; - src_node_size = src_mm->size << PAGE_SHIFT; - src_offset = 0; - } else { - src_offset += cur_size; - } - - dst_node_size -= cur_size; - if (!dst_node_size) { - ++dst_mm; - dst_node_size = dst_mm->size << PAGE_SHIFT; - dst_offset = 0; - } else { - dst_offset += cur_size; - } + amdgpu_res_next(&src_mm, cur_size); + amdgpu_res_next(&dst_mm, cur_size); } error: mutex_unlock(&adev->mman.gtt_window_lock); @@ -519,7 +441,8 @@ error: static bool amdgpu_mem_visible(struct amdgpu_device *adev, struct ttm_resource *mem) { - struct drm_mm_node *nodes = mem->mm_node; + uint64_t mem_size = (u64)mem->num_pages << PAGE_SHIFT; + struct amdgpu_res_cursor cursor; if (mem->mem_type == TTM_PL_SYSTEM || mem->mem_type == TTM_PL_TT) @@ -527,12 +450,13 @@ static bool amdgpu_mem_visible(struct amdgpu_device *adev, if (mem->mem_type != TTM_PL_VRAM) return false; + amdgpu_res_first(mem, 0, mem_size, &cursor); + /* ttm_resource_ioremap only supports contiguous memory */ - if (nodes->size != mem->num_pages) + if (cursor.size != mem_size) return false; - return ((nodes->start + nodes->size) << PAGE_SHIFT) - <= adev->gmc.visible_vram_size; + return cursor.start + cursor.size <= adev->gmc.visible_vram_size; } /* @@ -646,7 +570,7 @@ out: * * Called by ttm_mem_io_reserve() ultimately via ttm_bo_vm_fault() */ -static int amdgpu_ttm_io_mem_reserve(struct ttm_bo_device *bdev, struct ttm_resource *mem) +static int amdgpu_ttm_io_mem_reserve(struct ttm_device *bdev, struct ttm_resource *mem) { struct amdgpu_device *adev = amdgpu_ttm_adev(bdev); struct drm_mm_node *mm_node = mem->mm_node; @@ -674,7 +598,10 @@ static int amdgpu_ttm_io_mem_reserve(struct ttm_bo_device *bdev, struct ttm_reso mem->bus.offset += adev->gmc.aper_base; mem->bus.is_iomem = true; - mem->bus.caching = ttm_write_combined; + if (adev->gmc.xgmi.connected_to_cpu) + mem->bus.caching = ttm_cached; + else + mem->bus.caching = ttm_write_combined; break; default: return -EINVAL; @@ -686,12 +613,10 @@ static unsigned long amdgpu_ttm_io_mem_pfn(struct ttm_buffer_object *bo, unsigned long page_offset) { struct amdgpu_device *adev = amdgpu_ttm_adev(bo->bdev); - uint64_t offset = (page_offset << PAGE_SHIFT); - struct drm_mm_node *mm; + struct amdgpu_res_cursor cursor; - mm = amdgpu_find_mm_node(&bo->mem, &offset); - offset += adev->gmc.aper_base; - return mm->start + (offset >> PAGE_SHIFT); + amdgpu_res_first(&bo->mem, (u64)page_offset << PAGE_SHIFT, 0, &cursor); + return (adev->gmc.aper_base + cursor.start) >> PAGE_SHIFT; } /** @@ -893,16 +818,15 @@ void amdgpu_ttm_tt_set_user_pages(struct ttm_tt *ttm, struct page **pages) * * Called by amdgpu_ttm_backend_bind() **/ -static int amdgpu_ttm_tt_pin_userptr(struct ttm_bo_device *bdev, +static int amdgpu_ttm_tt_pin_userptr(struct ttm_device *bdev, struct ttm_tt *ttm) { struct amdgpu_device *adev = amdgpu_ttm_adev(bdev); struct amdgpu_ttm_tt *gtt = (void *)ttm; - int r; - int write = !(gtt->userflags & AMDGPU_GEM_USERPTR_READONLY); enum dma_data_direction direction = write ? DMA_BIDIRECTIONAL : DMA_TO_DEVICE; + int r; /* Allocate an SG array and squash pages into it */ r = sg_alloc_table_from_pages(ttm->sg, ttm->pages, ttm->num_pages, 0, @@ -931,18 +855,17 @@ release_sg: /* * amdgpu_ttm_tt_unpin_userptr - Unpin and unmap userptr pages */ -static void amdgpu_ttm_tt_unpin_userptr(struct ttm_bo_device *bdev, +static void amdgpu_ttm_tt_unpin_userptr(struct ttm_device *bdev, struct ttm_tt *ttm) { struct amdgpu_device *adev = amdgpu_ttm_adev(bdev); struct amdgpu_ttm_tt *gtt = (void *)ttm; - int write = !(gtt->userflags & AMDGPU_GEM_USERPTR_READONLY); enum dma_data_direction direction = write ? DMA_BIDIRECTIONAL : DMA_TO_DEVICE; /* double check that we don't free the table twice */ - if (!ttm->sg->sgl) + if (!ttm->sg || !ttm->sg->sgl) return; /* unmap the pages mapped to the device */ @@ -1015,7 +938,7 @@ gart_bind_fail: * Called by ttm_tt_bind() on behalf of ttm_bo_handle_move_mem(). * This handles binding GTT memory to the device address space. */ -static int amdgpu_ttm_backend_bind(struct ttm_bo_device *bdev, +static int amdgpu_ttm_backend_bind(struct ttm_device *bdev, struct ttm_tt *ttm, struct ttm_resource *bo_mem) { @@ -1155,20 +1078,20 @@ int amdgpu_ttm_recover_gart(struct ttm_buffer_object *tbo) * Called by ttm_tt_unbind() on behalf of ttm_bo_move_ttm() and * ttm_tt_destroy(). */ -static void amdgpu_ttm_backend_unbind(struct ttm_bo_device *bdev, +static void amdgpu_ttm_backend_unbind(struct ttm_device *bdev, struct ttm_tt *ttm) { struct amdgpu_device *adev = amdgpu_ttm_adev(bdev); struct amdgpu_ttm_tt *gtt = (void *)ttm; int r; - if (!gtt->bound) - return; - /* if the pages have userptr pinning then clear that first */ if (gtt->userptr) amdgpu_ttm_tt_unpin_userptr(bdev, ttm); + if (!gtt->bound) + return; + if (gtt->offset == AMDGPU_BO_INVALID_OFFSET) return; @@ -1180,7 +1103,7 @@ static void amdgpu_ttm_backend_unbind(struct ttm_bo_device *bdev, gtt->bound = false; } -static void amdgpu_ttm_backend_destroy(struct ttm_bo_device *bdev, +static void amdgpu_ttm_backend_destroy(struct ttm_device *bdev, struct ttm_tt *ttm) { struct amdgpu_ttm_tt *gtt = (void *)ttm; @@ -1234,7 +1157,7 @@ static struct ttm_tt *amdgpu_ttm_tt_create(struct ttm_buffer_object *bo, * Map the pages of a ttm_tt object to an address space visible * to the underlying device. */ -static int amdgpu_ttm_tt_populate(struct ttm_bo_device *bdev, +static int amdgpu_ttm_tt_populate(struct ttm_device *bdev, struct ttm_tt *ttm, struct ttm_operation_ctx *ctx) { @@ -1278,7 +1201,7 @@ static int amdgpu_ttm_tt_populate(struct ttm_bo_device *bdev, * Unmaps pages of a ttm_tt object from the device address space and * unpopulates the page array backing it. */ -static void amdgpu_ttm_tt_unpopulate(struct ttm_bo_device *bdev, +static void amdgpu_ttm_tt_unpopulate(struct ttm_device *bdev, struct ttm_tt *ttm) { struct amdgpu_ttm_tt *gtt = (void *)ttm; @@ -1430,6 +1353,10 @@ uint64_t amdgpu_ttm_tt_pde_flags(struct ttm_tt *ttm, struct ttm_resource *mem) flags |= AMDGPU_PTE_SNOOPED; } + if (mem && mem->mem_type == TTM_PL_VRAM && + mem->bus.caching == ttm_cached) + flags |= AMDGPU_PTE_SNOOPED; + return flags; } @@ -1469,7 +1396,7 @@ static bool amdgpu_ttm_bo_eviction_valuable(struct ttm_buffer_object *bo, const struct ttm_place *place) { unsigned long num_pages = bo->mem.num_pages; - struct drm_mm_node *node = bo->mem.mm_node; + struct amdgpu_res_cursor cursor; struct dma_resv_list *flist; struct dma_fence *f; int i; @@ -1501,13 +1428,15 @@ static bool amdgpu_ttm_bo_eviction_valuable(struct ttm_buffer_object *bo, case TTM_PL_VRAM: /* Check each drm MM node individually */ - while (num_pages) { - if (place->fpfn < (node->start + node->size) && - !(place->lpfn && place->lpfn <= node->start)) + amdgpu_res_first(&bo->mem, 0, (u64)num_pages << PAGE_SHIFT, + &cursor); + while (cursor.remaining) { + if (place->fpfn < PFN_DOWN(cursor.start + cursor.size) + && !(place->lpfn && + place->lpfn <= PFN_DOWN(cursor.start))) return true; - num_pages -= node->size; - ++node; + amdgpu_res_next(&cursor, cursor.size); } return false; @@ -1531,41 +1460,36 @@ static bool amdgpu_ttm_bo_eviction_valuable(struct ttm_buffer_object *bo, * access for debugging purposes. */ static int amdgpu_ttm_access_memory(struct ttm_buffer_object *bo, - unsigned long offset, - void *buf, int len, int write) + unsigned long offset, void *buf, int len, + int write) { struct amdgpu_bo *abo = ttm_to_amdgpu_bo(bo); struct amdgpu_device *adev = amdgpu_ttm_adev(abo->tbo.bdev); - struct drm_mm_node *nodes; + struct amdgpu_res_cursor cursor; + unsigned long flags; uint32_t value = 0; int ret = 0; - uint64_t pos; - unsigned long flags; if (bo->mem.mem_type != TTM_PL_VRAM) return -EIO; - pos = offset; - nodes = amdgpu_find_mm_node(&abo->tbo.mem, &pos); - pos += (nodes->start << PAGE_SHIFT); - - while (len && pos < adev->gmc.mc_vram_size) { - uint64_t aligned_pos = pos & ~(uint64_t)3; - uint64_t bytes = 4 - (pos & 3); - uint32_t shift = (pos & 3) * 8; + amdgpu_res_first(&bo->mem, offset, len, &cursor); + while (cursor.remaining) { + uint64_t aligned_pos = cursor.start & ~(uint64_t)3; + uint64_t bytes = 4 - (cursor.start & 3); + uint32_t shift = (cursor.start & 3) * 8; uint32_t mask = 0xffffffff << shift; - if (len < bytes) { - mask &= 0xffffffff >> (bytes - len) * 8; - bytes = len; + if (cursor.size < bytes) { + mask &= 0xffffffff >> (bytes - cursor.size) * 8; + bytes = cursor.size; } if (mask != 0xffffffff) { spin_lock_irqsave(&adev->mmio_idx_lock, flags); WREG32_NO_KIQ(mmMM_INDEX, ((uint32_t)aligned_pos) | 0x80000000); WREG32_NO_KIQ(mmMM_INDEX_HI, aligned_pos >> 31); - if (!write || mask != 0xffffffff) - value = RREG32_NO_KIQ(mmMM_DATA); + value = RREG32_NO_KIQ(mmMM_DATA); if (write) { value &= ~mask; value |= (*(uint32_t *)buf << shift) & mask; @@ -1577,21 +1501,15 @@ static int amdgpu_ttm_access_memory(struct ttm_buffer_object *bo, memcpy(buf, &value, bytes); } } else { - bytes = (nodes->start + nodes->size) << PAGE_SHIFT; - bytes = min(bytes - pos, (uint64_t)len & ~0x3ull); - - amdgpu_device_vram_access(adev, pos, (uint32_t *)buf, - bytes, write); + bytes = cursor.size & ~0x3ULL; + amdgpu_device_vram_access(adev, cursor.start, + (uint32_t *)buf, bytes, + write); } ret += bytes; buf = (uint8_t *)buf + bytes; - pos += bytes; - len -= bytes; - if (pos >= (nodes->start + nodes->size) << PAGE_SHIFT) { - ++nodes; - pos = (nodes->start << PAGE_SHIFT); - } + amdgpu_res_next(&cursor, bytes); } return ret; @@ -1603,7 +1521,7 @@ amdgpu_bo_delete_mem_notify(struct ttm_buffer_object *bo) amdgpu_bo_move_notify(bo, false, NULL); } -static struct ttm_bo_driver amdgpu_bo_driver = { +static struct ttm_device_funcs amdgpu_bo_driver = { .ttm_tt_create = &amdgpu_ttm_tt_create, .ttm_tt_populate = &amdgpu_ttm_tt_populate, .ttm_tt_unpopulate = &amdgpu_ttm_tt_unpopulate, @@ -1696,7 +1614,7 @@ static void amdgpu_ttm_training_data_block_init(struct amdgpu_device *adev) (adev->gmc.mc_vram_size - GDDR6_MEM_TRAINING_OFFSET); ctx->train_data_size = GDDR6_MEM_TRAINING_DATA_SIZE_IN_BYTES; - + DRM_DEBUG("train_data_size:%llx,p2c_train_data_offset:%llx,c2p_train_data_offset:%llx.\n", ctx->train_data_size, ctx->p2c_train_data_offset, @@ -1785,7 +1703,7 @@ int amdgpu_ttm_init(struct amdgpu_device *adev) mutex_init(&adev->mman.gtt_window_lock); /* No others user of address space so set it to 0 */ - r = ttm_bo_device_init(&adev->mman.bdev, &amdgpu_bo_driver, adev->dev, + r = ttm_device_init(&adev->mman.bdev, &amdgpu_bo_driver, adev->dev, adev_to_drm(adev)->anon_inode->i_mapping, adev_to_drm(adev)->vma_offset_manager, adev->need_swiotlb, @@ -1812,8 +1730,15 @@ int amdgpu_ttm_init(struct amdgpu_device *adev) /* Change the size here instead of the init above so only lpfn is affected */ amdgpu_ttm_set_buffer_funcs_status(adev, false); #ifdef CONFIG_64BIT - adev->mman.aper_base_kaddr = ioremap_wc(adev->gmc.aper_base, - adev->gmc.visible_vram_size); +#ifdef CONFIG_X86 + if (adev->gmc.xgmi.connected_to_cpu) + adev->mman.aper_base_kaddr = ioremap_cache(adev->gmc.aper_base, + adev->gmc.visible_vram_size); + + else +#endif + adev->mman.aper_base_kaddr = ioremap_wc(adev->gmc.aper_base, + adev->gmc.visible_vram_size); #endif /* @@ -1926,7 +1851,7 @@ void amdgpu_ttm_fini(struct amdgpu_device *adev) ttm_range_man_fini(&adev->mman.bdev, AMDGPU_PL_GDS); ttm_range_man_fini(&adev->mman.bdev, AMDGPU_PL_GWS); ttm_range_man_fini(&adev->mman.bdev, AMDGPU_PL_OA); - ttm_bo_device_release(&adev->mman.bdev); + ttm_device_fini(&adev->mman.bdev); adev->mman.initialized = false; DRM_INFO("amdgpu: ttm finalized\n"); } @@ -2002,7 +1927,7 @@ unlock: return ret; } -static struct vm_operations_struct amdgpu_ttm_vm_ops = { +static const struct vm_operations_struct amdgpu_ttm_vm_ops = { .fault = amdgpu_ttm_fault, .open = ttm_bo_vm_open, .close = ttm_bo_vm_close, @@ -2053,7 +1978,8 @@ int amdgpu_copy_buffer(struct amdgpu_ring *ring, uint64_t src_offset, return r; if (vm_needs_flush) { - job->vm_pd_addr = amdgpu_gmc_pd_addr(adev->gart.bo); + job->vm_pd_addr = amdgpu_gmc_pd_addr(adev->gmc.pdb0_bo ? + adev->gmc.pdb0_bo : adev->gart.bo); job->vm_needs_flush = true; } if (resv) { @@ -2104,9 +2030,9 @@ int amdgpu_fill_buffer(struct amdgpu_bo *bo, uint32_t max_bytes = adev->mman.buffer_funcs->fill_max_bytes; struct amdgpu_ring *ring = adev->mman.buffer_funcs_ring; - struct drm_mm_node *mm_node; - unsigned long num_pages; + struct amdgpu_res_cursor cursor; unsigned int num_loops, num_dw; + uint64_t num_bytes; struct amdgpu_job *job; int r; @@ -2122,15 +2048,13 @@ int amdgpu_fill_buffer(struct amdgpu_bo *bo, return r; } - num_pages = bo->tbo.mem.num_pages; - mm_node = bo->tbo.mem.mm_node; + num_bytes = bo->tbo.mem.num_pages << PAGE_SHIFT; num_loops = 0; - while (num_pages) { - uint64_t byte_count = mm_node->size << PAGE_SHIFT; - num_loops += DIV_ROUND_UP_ULL(byte_count, max_bytes); - num_pages -= mm_node->size; - ++mm_node; + amdgpu_res_first(&bo->tbo.mem, 0, num_bytes, &cursor); + while (cursor.remaining) { + num_loops += DIV_ROUND_UP_ULL(cursor.size, max_bytes); + amdgpu_res_next(&cursor, cursor.size); } num_dw = num_loops * adev->mman.buffer_funcs->fill_num_dw; @@ -2152,27 +2076,16 @@ int amdgpu_fill_buffer(struct amdgpu_bo *bo, } } - num_pages = bo->tbo.mem.num_pages; - mm_node = bo->tbo.mem.mm_node; + amdgpu_res_first(&bo->tbo.mem, 0, num_bytes, &cursor); + while (cursor.remaining) { + uint32_t cur_size = min_t(uint64_t, cursor.size, max_bytes); + uint64_t dst_addr = cursor.start; - while (num_pages) { - uint64_t byte_count = mm_node->size << PAGE_SHIFT; - uint64_t dst_addr; + dst_addr += amdgpu_ttm_domain_start(adev, bo->tbo.mem.mem_type); + amdgpu_emit_fill_buffer(adev, &job->ibs[0], src_data, dst_addr, + cur_size); - dst_addr = amdgpu_mm_node_addr(&bo->tbo, mm_node, &bo->tbo.mem); - while (byte_count) { - uint32_t cur_size_in_bytes = min_t(uint64_t, byte_count, - max_bytes); - - amdgpu_emit_fill_buffer(adev, &job->ibs[0], src_data, - dst_addr, cur_size_in_bytes); - - dst_addr += cur_size_in_bytes; - byte_count -= cur_size_in_bytes; - } - - num_pages -= mm_node->size; - ++mm_node; + amdgpu_res_next(&cursor, cur_size); } amdgpu_ring_pad_ib(ring, &job->ibs[0]); @@ -2191,36 +2104,74 @@ error_free: #if defined(CONFIG_DEBUG_FS) -static int amdgpu_mm_dump_table(struct seq_file *m, void *data) +static int amdgpu_mm_vram_table_show(struct seq_file *m, void *unused) { - struct drm_info_node *node = (struct drm_info_node *)m->private; - unsigned ttm_pl = (uintptr_t)node->info_ent->data; - struct drm_device *dev = node->minor->dev; - struct amdgpu_device *adev = drm_to_adev(dev); - struct ttm_resource_manager *man = ttm_manager_type(&adev->mman.bdev, ttm_pl); + struct amdgpu_device *adev = (struct amdgpu_device *)m->private; + struct ttm_resource_manager *man = ttm_manager_type(&adev->mman.bdev, + TTM_PL_VRAM); struct drm_printer p = drm_seq_file_printer(m); man->func->debug(man, &p); return 0; } -static int amdgpu_ttm_pool_debugfs(struct seq_file *m, void *data) +static int amdgpu_ttm_page_pool_show(struct seq_file *m, void *unused) { - struct drm_info_node *node = (struct drm_info_node *)m->private; - struct drm_device *dev = node->minor->dev; - struct amdgpu_device *adev = drm_to_adev(dev); + struct amdgpu_device *adev = (struct amdgpu_device *)m->private; return ttm_pool_debugfs(&adev->mman.bdev.pool, m); } -static const struct drm_info_list amdgpu_ttm_debugfs_list[] = { - {"amdgpu_vram_mm", amdgpu_mm_dump_table, 0, (void *)TTM_PL_VRAM}, - {"amdgpu_gtt_mm", amdgpu_mm_dump_table, 0, (void *)TTM_PL_TT}, - {"amdgpu_gds_mm", amdgpu_mm_dump_table, 0, (void *)AMDGPU_PL_GDS}, - {"amdgpu_gws_mm", amdgpu_mm_dump_table, 0, (void *)AMDGPU_PL_GWS}, - {"amdgpu_oa_mm", amdgpu_mm_dump_table, 0, (void *)AMDGPU_PL_OA}, - {"ttm_page_pool", amdgpu_ttm_pool_debugfs, 0, NULL}, -}; +static int amdgpu_mm_tt_table_show(struct seq_file *m, void *unused) +{ + struct amdgpu_device *adev = (struct amdgpu_device *)m->private; + struct ttm_resource_manager *man = ttm_manager_type(&adev->mman.bdev, + TTM_PL_TT); + struct drm_printer p = drm_seq_file_printer(m); + + man->func->debug(man, &p); + return 0; +} + +static int amdgpu_mm_gds_table_show(struct seq_file *m, void *unused) +{ + struct amdgpu_device *adev = (struct amdgpu_device *)m->private; + struct ttm_resource_manager *man = ttm_manager_type(&adev->mman.bdev, + AMDGPU_PL_GDS); + struct drm_printer p = drm_seq_file_printer(m); + + man->func->debug(man, &p); + return 0; +} + +static int amdgpu_mm_gws_table_show(struct seq_file *m, void *unused) +{ + struct amdgpu_device *adev = (struct amdgpu_device *)m->private; + struct ttm_resource_manager *man = ttm_manager_type(&adev->mman.bdev, + AMDGPU_PL_GWS); + struct drm_printer p = drm_seq_file_printer(m); + + man->func->debug(man, &p); + return 0; +} + +static int amdgpu_mm_oa_table_show(struct seq_file *m, void *unused) +{ + struct amdgpu_device *adev = (struct amdgpu_device *)m->private; + struct ttm_resource_manager *man = ttm_manager_type(&adev->mman.bdev, + AMDGPU_PL_OA); + struct drm_printer p = drm_seq_file_printer(m); + + man->func->debug(man, &p); + return 0; +} + +DEFINE_SHOW_ATTRIBUTE(amdgpu_mm_vram_table); +DEFINE_SHOW_ATTRIBUTE(amdgpu_mm_tt_table); +DEFINE_SHOW_ATTRIBUTE(amdgpu_mm_gds_table); +DEFINE_SHOW_ATTRIBUTE(amdgpu_mm_gws_table); +DEFINE_SHOW_ATTRIBUTE(amdgpu_mm_oa_table); +DEFINE_SHOW_ATTRIBUTE(amdgpu_ttm_page_pool); /* * amdgpu_ttm_vram_read - Linear read access to VRAM @@ -2308,58 +2259,6 @@ static const struct file_operations amdgpu_ttm_vram_fops = { .llseek = default_llseek, }; -#ifdef CONFIG_DRM_AMDGPU_GART_DEBUGFS - -/* - * amdgpu_ttm_gtt_read - Linear read access to GTT memory - */ -static ssize_t amdgpu_ttm_gtt_read(struct file *f, char __user *buf, - size_t size, loff_t *pos) -{ - struct amdgpu_device *adev = file_inode(f)->i_private; - ssize_t result = 0; - int r; - - while (size) { - loff_t p = *pos / PAGE_SIZE; - unsigned off = *pos & ~PAGE_MASK; - size_t cur_size = min_t(size_t, size, PAGE_SIZE - off); - struct page *page; - void *ptr; - - if (p >= adev->gart.num_cpu_pages) - return result; - - page = adev->gart.pages[p]; - if (page) { - ptr = kmap(page); - ptr += off; - - r = copy_to_user(buf, ptr, cur_size); - kunmap(adev->gart.pages[p]); - } else - r = clear_user(buf, cur_size); - - if (r) - return -EFAULT; - - result += cur_size; - buf += cur_size; - *pos += cur_size; - size -= cur_size; - } - - return result; -} - -static const struct file_operations amdgpu_ttm_gtt_fops = { - .owner = THIS_MODULE, - .read = amdgpu_ttm_gtt_read, - .llseek = default_llseek -}; - -#endif - /* * amdgpu_iomem_read - Virtual read access to GPU mapped memory * @@ -2474,46 +2373,29 @@ static const struct file_operations amdgpu_ttm_iomem_fops = { .llseek = default_llseek }; -static const struct { - char *name; - const struct file_operations *fops; - int domain; -} ttm_debugfs_entries[] = { - { "amdgpu_vram", &amdgpu_ttm_vram_fops, TTM_PL_VRAM }, -#ifdef CONFIG_DRM_AMDGPU_GART_DEBUGFS - { "amdgpu_gtt", &amdgpu_ttm_gtt_fops, TTM_PL_TT }, #endif - { "amdgpu_iomem", &amdgpu_ttm_iomem_fops, TTM_PL_SYSTEM }, -}; -#endif - -int amdgpu_ttm_debugfs_init(struct amdgpu_device *adev) +void amdgpu_ttm_debugfs_init(struct amdgpu_device *adev) { #if defined(CONFIG_DEBUG_FS) - unsigned count; - struct drm_minor *minor = adev_to_drm(adev)->primary; - struct dentry *ent, *root = minor->debugfs_root; - - for (count = 0; count < ARRAY_SIZE(ttm_debugfs_entries); count++) { - ent = debugfs_create_file( - ttm_debugfs_entries[count].name, - S_IFREG | S_IRUGO, root, - adev, - ttm_debugfs_entries[count].fops); - if (IS_ERR(ent)) - return PTR_ERR(ent); - if (ttm_debugfs_entries[count].domain == TTM_PL_VRAM) - i_size_write(ent->d_inode, adev->gmc.mc_vram_size); - else if (ttm_debugfs_entries[count].domain == TTM_PL_TT) - i_size_write(ent->d_inode, adev->gmc.gart_size); - adev->mman.debugfs_entries[count] = ent; - } - - count = ARRAY_SIZE(amdgpu_ttm_debugfs_list); - return amdgpu_debugfs_add_files(adev, amdgpu_ttm_debugfs_list, count); -#else - return 0; + struct dentry *root = minor->debugfs_root; + + debugfs_create_file_size("amdgpu_vram", 0444, root, adev, + &amdgpu_ttm_vram_fops, adev->gmc.mc_vram_size); + debugfs_create_file("amdgpu_iomem", 0444, root, adev, + &amdgpu_ttm_iomem_fops); + debugfs_create_file("amdgpu_vram_mm", 0444, root, adev, + &amdgpu_mm_vram_table_fops); + debugfs_create_file("amdgpu_gtt_mm", 0444, root, adev, + &amdgpu_mm_tt_table_fops); + debugfs_create_file("amdgpu_gds_mm", 0444, root, adev, + &amdgpu_mm_gds_table_fops); + debugfs_create_file("amdgpu_gws_mm", 0444, root, adev, + &amdgpu_mm_gws_table_fops); + debugfs_create_file("amdgpu_oa_mm", 0444, root, adev, + &amdgpu_mm_oa_table_fops); + debugfs_create_file("ttm_page_pool", 0444, root, adev, + &amdgpu_ttm_page_pool_fops); #endif } diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.h index d2987536d7cd..9e38475e0f8d 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.h @@ -60,14 +60,10 @@ struct amdgpu_gtt_mgr { }; struct amdgpu_mman { - struct ttm_bo_device bdev; + struct ttm_device bdev; bool initialized; void __iomem *aper_base_kaddr; -#if defined(CONFIG_DEBUG_FS) - struct dentry *debugfs_entries[8]; -#endif - /* buffer handling */ const struct amdgpu_buffer_funcs *buffer_funcs; struct amdgpu_ring *buffer_funcs_ring; @@ -116,11 +112,11 @@ int amdgpu_gtt_mgr_recover(struct ttm_resource_manager *man); u64 amdgpu_vram_mgr_bo_visible_size(struct amdgpu_bo *bo); int amdgpu_vram_mgr_alloc_sgt(struct amdgpu_device *adev, struct ttm_resource *mem, + u64 offset, u64 size, struct device *dev, enum dma_data_direction dir, struct sg_table **sgt); -void amdgpu_vram_mgr_free_sgt(struct amdgpu_device *adev, - struct device *dev, +void amdgpu_vram_mgr_free_sgt(struct device *dev, enum dma_data_direction dir, struct sg_table *sgt); uint64_t amdgpu_vram_mgr_usage(struct ttm_resource_manager *man); @@ -186,6 +182,6 @@ uint64_t amdgpu_ttm_tt_pde_flags(struct ttm_tt *ttm, struct ttm_resource *mem); uint64_t amdgpu_ttm_tt_pte_flags(struct amdgpu_device *adev, struct ttm_tt *ttm, struct ttm_resource *mem); -int amdgpu_ttm_debugfs_init(struct amdgpu_device *adev); +void amdgpu_ttm_debugfs_init(struct amdgpu_device *adev); #endif diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ucode.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ucode.c index 1beb08af347f..9733224117e3 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ucode.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ucode.c @@ -402,6 +402,7 @@ amdgpu_ucode_get_load_type(struct amdgpu_device *adev, int load_type) case CHIP_NAVY_FLOUNDER: case CHIP_VANGOGH: case CHIP_DIMGREY_CAVEFISH: + case CHIP_ALDEBARAN: if (!load_type) return AMDGPU_FW_LOAD_DIRECT; else diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ucode.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_ucode.h index 46449e70348b..2c42874f7784 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ucode.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ucode.h @@ -105,6 +105,9 @@ struct psp_firmware_header_v1_3 { uint32_t spl_header_version; uint32_t spl_offset_bytes; uint32_t spl_size_bytes; + uint32_t rl_header_version; + uint32_t rl_offset_bytes; + uint32_t rl_size_bytes; }; /* version_major=1, version_minor=0 */ @@ -136,6 +139,7 @@ enum ta_fw_type { TA_FW_TYPE_PSP_DTM, TA_FW_TYPE_PSP_RAP, TA_FW_TYPE_PSP_SECUREDISPLAY, + TA_FW_TYPE_MAX_INDEX, }; struct ta_fw_bin_desc { diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_umc.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_umc.c index a2975c8092a9..ea6f99be070b 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_umc.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_umc.c @@ -60,8 +60,9 @@ int amdgpu_umc_ras_late_init(struct amdgpu_device *adev) } /* ras init of specific umc version */ - if (adev->umc.funcs && adev->umc.funcs->err_cnt_init) - adev->umc.funcs->err_cnt_init(adev); + if (adev->umc.ras_funcs && + adev->umc.ras_funcs->err_cnt_init) + adev->umc.ras_funcs->err_cnt_init(adev); return 0; @@ -95,12 +96,12 @@ int amdgpu_umc_process_ras_data_cb(struct amdgpu_device *adev, struct ras_err_data *err_data = (struct ras_err_data *)ras_error_status; kgd2kfd_set_sram_ecc_flag(adev->kfd.dev); - if (adev->umc.funcs && - adev->umc.funcs->query_ras_error_count) - adev->umc.funcs->query_ras_error_count(adev, ras_error_status); + if (adev->umc.ras_funcs && + adev->umc.ras_funcs->query_ras_error_count) + adev->umc.ras_funcs->query_ras_error_count(adev, ras_error_status); - if (adev->umc.funcs && - adev->umc.funcs->query_ras_error_address && + if (adev->umc.ras_funcs && + adev->umc.ras_funcs->query_ras_error_address && adev->umc.max_ras_err_cnt_per_query) { err_data->err_addr = kcalloc(adev->umc.max_ras_err_cnt_per_query, @@ -116,7 +117,7 @@ int amdgpu_umc_process_ras_data_cb(struct amdgpu_device *adev, /* umc query_ras_error_address is also responsible for clearing * error status */ - adev->umc.funcs->query_ras_error_address(adev, ras_error_status); + adev->umc.ras_funcs->query_ras_error_address(adev, ras_error_status); } /* only uncorrectable error needs gpu reset */ diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_umc.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_umc.h index 183814493658..bbcccf53080d 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_umc.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_umc.h @@ -35,13 +35,17 @@ #define LOOP_UMC_CH_INST(ch_inst) for ((ch_inst) = 0; (ch_inst) < adev->umc.channel_inst_num; (ch_inst)++) #define LOOP_UMC_INST_AND_CH(umc_inst, ch_inst) LOOP_UMC_INST((umc_inst)) LOOP_UMC_CH_INST((ch_inst)) -struct amdgpu_umc_funcs { +struct amdgpu_umc_ras_funcs { void (*err_cnt_init)(struct amdgpu_device *adev); int (*ras_late_init)(struct amdgpu_device *adev); + void (*ras_fini)(struct amdgpu_device *adev); void (*query_ras_error_count)(struct amdgpu_device *adev, - void *ras_error_status); + void *ras_error_status); void (*query_ras_error_address)(struct amdgpu_device *adev, void *ras_error_status); +}; + +struct amdgpu_umc_funcs { void (*init_registers)(struct amdgpu_device *adev); }; @@ -59,6 +63,7 @@ struct amdgpu_umc { struct ras_common_if *ras_if; const struct amdgpu_umc_funcs *funcs; + const struct amdgpu_umc_ras_funcs *ras_funcs; }; int amdgpu_umc_ras_late_init(struct amdgpu_device *adev); diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_uvd.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_uvd.c index e2ed4689118a..c6dbc0801604 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_uvd.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_uvd.c @@ -259,7 +259,7 @@ int amdgpu_uvd_sw_init(struct amdgpu_device *adev) if ((adev->asic_type == CHIP_POLARIS10 || adev->asic_type == CHIP_POLARIS11) && (adev->uvd.fw_version < FW_1_66_16)) - DRM_ERROR("POLARIS10/11 UVD firmware version %hu.%hu is too old.\n", + DRM_ERROR("POLARIS10/11 UVD firmware version %u.%u is too old.\n", version_major, version_minor); } else { unsigned int enc_major, enc_minor, dec_minor; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vcn.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_vcn.c index 99b82f3c2617..201645963ba5 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vcn.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vcn.c @@ -47,6 +47,7 @@ #define FIRMWARE_NAVY_FLOUNDER "amdgpu/navy_flounder_vcn.bin" #define FIRMWARE_VANGOGH "amdgpu/vangogh_vcn.bin" #define FIRMWARE_DIMGREY_CAVEFISH "amdgpu/dimgrey_cavefish_vcn.bin" +#define FIRMWARE_ALDEBARAN "amdgpu/aldebaran_vcn.bin" MODULE_FIRMWARE(FIRMWARE_RAVEN); MODULE_FIRMWARE(FIRMWARE_PICASSO); @@ -54,6 +55,7 @@ MODULE_FIRMWARE(FIRMWARE_RAVEN2); MODULE_FIRMWARE(FIRMWARE_ARCTURUS); MODULE_FIRMWARE(FIRMWARE_RENOIR); MODULE_FIRMWARE(FIRMWARE_GREEN_SARDINE); +MODULE_FIRMWARE(FIRMWARE_ALDEBARAN); MODULE_FIRMWARE(FIRMWARE_NAVI10); MODULE_FIRMWARE(FIRMWARE_NAVI14); MODULE_FIRMWARE(FIRMWARE_NAVI12); @@ -104,6 +106,12 @@ int amdgpu_vcn_sw_init(struct amdgpu_device *adev) (adev->pg_flags & AMD_PG_SUPPORT_VCN_DPG)) adev->vcn.indirect_sram = true; break; + case CHIP_ALDEBARAN: + fw_name = FIRMWARE_ALDEBARAN; + if ((adev->firmware.load_type == AMDGPU_FW_LOAD_PSP) && + (adev->pg_flags & AMD_PG_SUPPORT_VCN_DPG)) + adev->vcn.indirect_sram = true; + break; case CHIP_NAVI10: fw_name = FIRMWARE_NAVI10; if ((adev->firmware.load_type == AMDGPU_FW_LOAD_PSP) && diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vcn.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_vcn.h index 13aa417f6be7..bc76cab67697 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vcn.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vcn.h @@ -155,6 +155,7 @@ } \ } while (0) +#define AMDGPU_VCN_FW_SHARED_FLAG_0_RB (1 << 6) #define AMDGPU_VCN_MULTI_QUEUE_FLAG (1 << 8) #define AMDGPU_VCN_SW_RING_FLAG (1 << 9) @@ -211,6 +212,7 @@ struct amdgpu_vcn_inst { void *saved_bo; struct amdgpu_ring ring_dec; struct amdgpu_ring ring_enc[AMDGPU_VCN_MAX_ENC_RINGS]; + atomic_t sched_score; struct amdgpu_irq_src irq; struct amdgpu_vcn_reg external; struct amdgpu_bo *dpg_sram_bo; @@ -243,6 +245,12 @@ struct amdgpu_vcn { int inst_idx, struct dpg_pause_state *new_state); }; +struct amdgpu_fw_shared_rb_ptrs_struct { + /* to WA DPG R/W ptr issues.*/ + uint32_t rptr; + uint32_t wptr; +}; + struct amdgpu_fw_shared_multi_queue { uint8_t decode_queue_mode; uint8_t encode_generalpurpose_queue_mode; @@ -258,10 +266,12 @@ struct amdgpu_fw_shared_sw_ring { struct amdgpu_fw_shared { uint32_t present_flag_0; - uint8_t pad[53]; + uint8_t pad[44]; + struct amdgpu_fw_shared_rb_ptrs_struct rb; + uint8_t pad1[1]; struct amdgpu_fw_shared_multi_queue multi_queue; struct amdgpu_fw_shared_sw_ring sw_ring; -} __attribute__((__packed__)); +}; struct amdgpu_vcn_decode_buffer { uint32_t valid_buf_flag; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_virt.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_virt.c index 5da04d45b637..0c9c5255aa42 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_virt.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_virt.c @@ -466,6 +466,8 @@ static int amdgpu_virt_read_pf2vf_data(struct amdgpu_device *adev) ((struct amd_sriov_msg_pf2vf_info *)pf2vf_info)->vf2pf_update_interval_ms; adev->virt.gim_feature = ((struct amd_sriov_msg_pf2vf_info *)pf2vf_info)->feature_flags.all; + adev->virt.reg_access = + ((struct amd_sriov_msg_pf2vf_info *)pf2vf_info)->reg_access_flags.all; break; default: @@ -617,6 +619,14 @@ void amdgpu_virt_init_data_exchange(struct amdgpu_device *adev) if (adev->virt.ras_init_done) amdgpu_virt_add_bad_page(adev, bp_block_offset, bp_block_size); } + } else if (adev->bios != NULL) { + adev->virt.fw_reserve.p_pf2vf = + (struct amd_sriov_msg_pf2vf_info_header *) + (adev->bios + (AMD_SRIOV_MSG_PF2VF_OFFSET_KB << 10)); + + amdgpu_virt_read_pf2vf_data(adev); + + return; } if (adev->virt.vf2pf_update_interval_ms != 0) { @@ -640,6 +650,7 @@ void amdgpu_detect_virtualization(struct amdgpu_device *adev) case CHIP_NAVI12: case CHIP_SIENNA_CICHLID: case CHIP_ARCTURUS: + case CHIP_ALDEBARAN: reg = RREG32(mmRCC_IOV_FUNC_IDENTIFIER); break; default: /* other chip doesn't support SRIOV */ diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_virt.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_virt.h index 8dd624c20f89..383d4bdc3fb5 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_virt.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_virt.h @@ -104,6 +104,17 @@ enum AMDGIM_FEATURE_FLAG { AMDGIM_FEATURE_GIM_MM_BW_MGR = 0x8, /* PP ONE VF MODE in GIM */ AMDGIM_FEATURE_PP_ONE_VF = (1 << 4), + /* Indirect Reg Access enabled */ + AMDGIM_FEATURE_INDIRECT_REG_ACCESS = (1 << 5), +}; + +enum AMDGIM_REG_ACCESS_FLAG { + /* Use PSP to program IH_RB_CNTL */ + AMDGIM_FEATURE_IH_REG_PSP_EN = (1 << 0), + /* Use RLC to program MMHUB regs */ + AMDGIM_FEATURE_MMHUB_REG_RLC_EN = (1 << 1), + /* Use RLC to program GC regs */ + AMDGIM_FEATURE_GC_REG_RLC_EN = (1 << 2), }; struct amdgim_pf2vf_info_v1 { @@ -217,6 +228,7 @@ struct amdgpu_virt { bool tdr_debug; struct amdgpu_virt_ras_err_handler_data *virt_eh_data; bool ras_init_done; + uint32_t reg_access; /* vf2pf message */ struct delayed_work vf2pf_work; @@ -238,6 +250,22 @@ struct amdgpu_virt { #define amdgpu_sriov_fullaccess(adev) \ (amdgpu_sriov_vf((adev)) && !amdgpu_sriov_runtime((adev))) +#define amdgpu_sriov_reg_indirect_en(adev) \ +(amdgpu_sriov_vf((adev)) && \ + ((adev)->virt.gim_feature & (AMDGIM_FEATURE_INDIRECT_REG_ACCESS))) + +#define amdgpu_sriov_reg_indirect_ih(adev) \ +(amdgpu_sriov_vf((adev)) && \ + ((adev)->virt.reg_access & (AMDGIM_FEATURE_IH_REG_PSP_EN))) + +#define amdgpu_sriov_reg_indirect_mmhub(adev) \ +(amdgpu_sriov_vf((adev)) && \ + ((adev)->virt.reg_access & (AMDGIM_FEATURE_MMHUB_REG_RLC_EN))) + +#define amdgpu_sriov_reg_indirect_gc(adev) \ +(amdgpu_sriov_vf((adev)) && \ + ((adev)->virt.reg_access & (AMDGIM_FEATURE_GC_REG_RLC_EN))) + #define amdgpu_passthrough(adev) \ ((adev)->virt.caps & AMDGPU_PASSTHROUGH_MODE) diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c index 7d2c8b169827..9acee4a5b2ba 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c @@ -92,13 +92,13 @@ struct amdgpu_prt_cb { static inline void amdgpu_vm_eviction_lock(struct amdgpu_vm *vm) { mutex_lock(&vm->eviction_lock); - vm->saved_flags = memalloc_nofs_save(); + vm->saved_flags = memalloc_noreclaim_save(); } static inline int amdgpu_vm_eviction_trylock(struct amdgpu_vm *vm) { if (mutex_trylock(&vm->eviction_lock)) { - vm->saved_flags = memalloc_nofs_save(); + vm->saved_flags = memalloc_noreclaim_save(); return 1; } return 0; @@ -106,7 +106,7 @@ static inline int amdgpu_vm_eviction_trylock(struct amdgpu_vm *vm) static inline void amdgpu_vm_eviction_unlock(struct amdgpu_vm *vm) { - memalloc_nofs_restore(vm->saved_flags); + memalloc_noreclaim_restore(vm->saved_flags); mutex_unlock(&vm->eviction_lock); } @@ -638,15 +638,15 @@ void amdgpu_vm_move_to_lru_tail(struct amdgpu_device *adev, struct amdgpu_vm_bo_base *bo_base; if (vm->bulk_moveable) { - spin_lock(&ttm_bo_glob.lru_lock); + spin_lock(&adev->mman.bdev.lru_lock); ttm_bo_bulk_move_lru_tail(&vm->lru_bulk_move); - spin_unlock(&ttm_bo_glob.lru_lock); + spin_unlock(&adev->mman.bdev.lru_lock); return; } memset(&vm->lru_bulk_move, 0, sizeof(vm->lru_bulk_move)); - spin_lock(&ttm_bo_glob.lru_lock); + spin_lock(&adev->mman.bdev.lru_lock); list_for_each_entry(bo_base, &vm->idle, vm_status) { struct amdgpu_bo *bo = bo_base->bo; @@ -660,7 +660,7 @@ void amdgpu_vm_move_to_lru_tail(struct amdgpu_device *adev, &bo->shadow->tbo.mem, &vm->lru_bulk_move); } - spin_unlock(&ttm_bo_glob.lru_lock); + spin_unlock(&adev->mman.bdev.lru_lock); vm->bulk_moveable = true; } @@ -869,6 +869,7 @@ static void amdgpu_vm_bo_param(struct amdgpu_device *adev, struct amdgpu_vm *vm, bp->domain = amdgpu_bo_get_preferred_pin_domain(adev, bp->domain); bp->flags = AMDGPU_GEM_CREATE_VRAM_CONTIGUOUS | AMDGPU_GEM_CREATE_CPU_GTT_USWC; + bp->bo_ptr_size = sizeof(struct amdgpu_bo); if (vm->use_cpu_for_update) bp->flags |= AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED; else if (!vm->root.base.bo || vm->root.base.bo->shadow) @@ -3147,6 +3148,12 @@ void amdgpu_vm_manager_init(struct amdgpu_device *adev) { unsigned i; + /* Concurrent flushes are only possible starting with Vega10 and + * are broken on Navi10 and Navi14. + */ + adev->vm_manager.concurrent_flush = !(adev->asic_type < CHIP_VEGA10 || + adev->asic_type == CHIP_NAVI10 || + adev->asic_type == CHIP_NAVI14); amdgpu_vmid_mgr_init(adev); adev->vm_manager.fence_context = @@ -3300,7 +3307,7 @@ bool amdgpu_vm_handle_fault(struct amdgpu_device *adev, u32 pasid, struct amdgpu_bo *root; uint64_t value, flags; struct amdgpu_vm *vm; - long r; + int r; spin_lock(&adev->vm_manager.pasid_lock); vm = idr_find(&adev->vm_manager.pasid_idr, pasid); @@ -3349,6 +3356,12 @@ bool amdgpu_vm_handle_fault(struct amdgpu_device *adev, u32 pasid, value = 0; } + r = dma_resv_reserve_shared(root->tbo.base.resv, 1); + if (r) { + pr_debug("failed %d to reserve fence slot\n", r); + goto error_unlock; + } + r = amdgpu_vm_bo_update_mapping(adev, adev, vm, true, false, NULL, addr, addr, flags, value, NULL, NULL, NULL); @@ -3360,7 +3373,7 @@ bool amdgpu_vm_handle_fault(struct amdgpu_device *adev, u32 pasid, error_unlock: amdgpu_bo_unreserve(root); if (r < 0) - DRM_ERROR("Can't handle page fault (%ld)\n", r); + DRM_ERROR("Can't handle page fault (%d)\n", r); error_unref: amdgpu_bo_unref(&root); diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.h index 976a12e5a8b9..4e140288159c 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.h @@ -331,6 +331,7 @@ struct amdgpu_vm_manager { /* Handling of VMIDs */ struct amdgpu_vmid_mgr id_mgr[AMDGPU_MAX_VMHUBS]; unsigned int first_kfd_vmid; + bool concurrent_flush; /* Handling of VM fences */ u64 fence_context; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vram_mgr.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_vram_mgr.c index c89b66bb70e2..bce105e2973e 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vram_mgr.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vram_mgr.c @@ -25,6 +25,7 @@ #include <linux/dma-mapping.h> #include "amdgpu.h" #include "amdgpu_vm.h" +#include "amdgpu_res_cursor.h" #include "amdgpu_atomfirmware.h" #include "atom.h" @@ -52,7 +53,7 @@ static ssize_t amdgpu_mem_info_vram_total_show(struct device *dev, struct drm_device *ddev = dev_get_drvdata(dev); struct amdgpu_device *adev = drm_to_adev(ddev); - return snprintf(buf, PAGE_SIZE, "%llu\n", adev->gmc.real_vram_size); + return sysfs_emit(buf, "%llu\n", adev->gmc.real_vram_size); } /** @@ -69,7 +70,7 @@ static ssize_t amdgpu_mem_info_vis_vram_total_show(struct device *dev, struct drm_device *ddev = dev_get_drvdata(dev); struct amdgpu_device *adev = drm_to_adev(ddev); - return snprintf(buf, PAGE_SIZE, "%llu\n", adev->gmc.visible_vram_size); + return sysfs_emit(buf, "%llu\n", adev->gmc.visible_vram_size); } /** @@ -87,8 +88,7 @@ static ssize_t amdgpu_mem_info_vram_used_show(struct device *dev, struct amdgpu_device *adev = drm_to_adev(ddev); struct ttm_resource_manager *man = ttm_manager_type(&adev->mman.bdev, TTM_PL_VRAM); - return snprintf(buf, PAGE_SIZE, "%llu\n", - amdgpu_vram_mgr_usage(man)); + return sysfs_emit(buf, "%llu\n", amdgpu_vram_mgr_usage(man)); } /** @@ -106,8 +106,7 @@ static ssize_t amdgpu_mem_info_vis_vram_used_show(struct device *dev, struct amdgpu_device *adev = drm_to_adev(ddev); struct ttm_resource_manager *man = ttm_manager_type(&adev->mman.bdev, TTM_PL_VRAM); - return snprintf(buf, PAGE_SIZE, "%llu\n", - amdgpu_vram_mgr_vis_usage(man)); + return sysfs_emit(buf, "%llu\n", amdgpu_vram_mgr_vis_usage(man)); } static ssize_t amdgpu_mem_info_vram_vendor(struct device *dev, @@ -119,27 +118,27 @@ static ssize_t amdgpu_mem_info_vram_vendor(struct device *dev, switch (adev->gmc.vram_vendor) { case SAMSUNG: - return snprintf(buf, PAGE_SIZE, "samsung\n"); + return sysfs_emit(buf, "samsung\n"); case INFINEON: - return snprintf(buf, PAGE_SIZE, "infineon\n"); + return sysfs_emit(buf, "infineon\n"); case ELPIDA: - return snprintf(buf, PAGE_SIZE, "elpida\n"); + return sysfs_emit(buf, "elpida\n"); case ETRON: - return snprintf(buf, PAGE_SIZE, "etron\n"); + return sysfs_emit(buf, "etron\n"); case NANYA: - return snprintf(buf, PAGE_SIZE, "nanya\n"); + return sysfs_emit(buf, "nanya\n"); case HYNIX: - return snprintf(buf, PAGE_SIZE, "hynix\n"); + return sysfs_emit(buf, "hynix\n"); case MOSEL: - return snprintf(buf, PAGE_SIZE, "mosel\n"); + return sysfs_emit(buf, "mosel\n"); case WINBOND: - return snprintf(buf, PAGE_SIZE, "winbond\n"); + return sysfs_emit(buf, "winbond\n"); case ESMT: - return snprintf(buf, PAGE_SIZE, "esmt\n"); + return sysfs_emit(buf, "esmt\n"); case MICRON: - return snprintf(buf, PAGE_SIZE, "micron\n"); + return sysfs_emit(buf, "micron\n"); default: - return snprintf(buf, PAGE_SIZE, "unknown\n"); + return sysfs_emit(buf, "unknown\n"); } } @@ -567,6 +566,8 @@ static void amdgpu_vram_mgr_del(struct ttm_resource_manager *man, * * @adev: amdgpu device pointer * @mem: TTM memory object + * @offset: byte offset from the base of VRAM BO + * @length: number of bytes to export in sg_table * @dev: the other device * @dir: dma direction * @sgt: resulting sg table @@ -575,39 +576,47 @@ static void amdgpu_vram_mgr_del(struct ttm_resource_manager *man, */ int amdgpu_vram_mgr_alloc_sgt(struct amdgpu_device *adev, struct ttm_resource *mem, + u64 offset, u64 length, struct device *dev, enum dma_data_direction dir, struct sg_table **sgt) { - struct drm_mm_node *node; + struct amdgpu_res_cursor cursor; struct scatterlist *sg; int num_entries = 0; - unsigned int pages; int i, r; *sgt = kmalloc(sizeof(**sgt), GFP_KERNEL); if (!*sgt) return -ENOMEM; - for (pages = mem->num_pages, node = mem->mm_node; - pages; pages -= node->size, ++node) - ++num_entries; + /* Determine the number of DRM_MM nodes to export */ + amdgpu_res_first(mem, offset, length, &cursor); + while (cursor.remaining) { + num_entries++; + amdgpu_res_next(&cursor, cursor.size); + } r = sg_alloc_table(*sgt, num_entries, GFP_KERNEL); if (r) goto error_free; + /* Initialize scatterlist nodes of sg_table */ for_each_sgtable_sg((*sgt), sg, i) sg->length = 0; - node = mem->mm_node; + /* + * Walk down DRM_MM nodes to populate scatterlist nodes + * @note: Use iterator api to get first the DRM_MM node + * and the number of bytes from it. Access the following + * DRM_MM node(s) if more buffer needs to exported + */ + amdgpu_res_first(mem, offset, length, &cursor); for_each_sgtable_sg((*sgt), sg, i) { - phys_addr_t phys = (node->start << PAGE_SHIFT) + - adev->gmc.aper_base; - size_t size = node->size << PAGE_SHIFT; + phys_addr_t phys = cursor.start + adev->gmc.aper_base; + size_t size = cursor.size; dma_addr_t addr; - ++node; addr = dma_map_resource(dev, phys, size, dir, DMA_ATTR_SKIP_CPU_SYNC); r = dma_mapping_error(dev, addr); @@ -617,7 +626,10 @@ int amdgpu_vram_mgr_alloc_sgt(struct amdgpu_device *adev, sg_set_page(sg, NULL, size, 0); sg_dma_address(sg) = addr; sg_dma_len(sg) = size; + + amdgpu_res_next(&cursor, cursor.size); } + return 0; error_unmap: @@ -639,15 +651,13 @@ error_free: /** * amdgpu_vram_mgr_free_sgt - allocate and fill a sg table * - * @adev: amdgpu device pointer * @dev: device pointer * @dir: data direction of resource to unmap * @sgt: sg table to free * * Free a previously allocate sg table. */ -void amdgpu_vram_mgr_free_sgt(struct amdgpu_device *adev, - struct device *dev, +void amdgpu_vram_mgr_free_sgt(struct device *dev, enum dma_data_direction dir, struct sg_table *sgt) { diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_xgmi.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_xgmi.c index 659b385b27b5..8567d5d77346 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_xgmi.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_xgmi.c @@ -24,7 +24,6 @@ #include <linux/list.h> #include "amdgpu.h" #include "amdgpu_xgmi.h" -#include "amdgpu_smu.h" #include "amdgpu_ras.h" #include "soc15.h" #include "df/df_3_6_offset.h" @@ -217,7 +216,7 @@ static ssize_t amdgpu_xgmi_show_device_id(struct device *dev, struct drm_device *ddev = dev_get_drvdata(dev); struct amdgpu_device *adev = drm_to_adev(ddev); - return snprintf(buf, PAGE_SIZE, "%llu\n", adev->gmc.xgmi.node_id); + return sysfs_emit(buf, "%llu\n", adev->gmc.xgmi.node_id); } @@ -246,7 +245,7 @@ static ssize_t amdgpu_xgmi_show_error(struct device *dev, adev->df.funcs->set_fica(adev, ficaa_pie_status_in, 0, 0); - return snprintf(buf, PAGE_SIZE, "%u\n", error_count); + return sysfs_emit(buf, "%u\n", error_count); } @@ -468,15 +467,22 @@ int amdgpu_xgmi_update_topology(struct amdgpu_hive_info *hive, struct amdgpu_dev } +/* + * NOTE psp_xgmi_node_info.num_hops layout is as follows: + * num_hops[7:6] = link type (0 = xGMI2, 1 = xGMI3, 2/3 = reserved) + * num_hops[5:3] = reserved + * num_hops[2:0] = number of hops + */ int amdgpu_xgmi_get_hops_count(struct amdgpu_device *adev, struct amdgpu_device *peer_adev) { struct psp_xgmi_topology_info *top = &adev->psp.xgmi_context.top_info; + uint8_t num_hops_mask = 0x7; int i; for (i = 0 ; i < top->num_nodes; ++i) if (top->nodes[i].node_id == peer_adev->gmc.xgmi.node_id) - return top->nodes[i].num_hops; + return top->nodes[i].num_hops & num_hops_mask; return -EINVAL; } @@ -492,7 +498,8 @@ int amdgpu_xgmi_add_device(struct amdgpu_device *adev) if (!adev->gmc.xgmi.supported) return 0; - if (amdgpu_device_ip_get_ip_block(adev, AMD_IP_BLOCK_TYPE_PSP)) { + if (!adev->gmc.xgmi.pending_reset && + amdgpu_device_ip_get_ip_block(adev, AMD_IP_BLOCK_TYPE_PSP)) { ret = psp_xgmi_initialize(&adev->psp); if (ret) { dev_err(adev->dev, @@ -538,7 +545,8 @@ int amdgpu_xgmi_add_device(struct amdgpu_device *adev) task_barrier_add_task(&hive->tb); - if (amdgpu_device_ip_get_ip_block(adev, AMD_IP_BLOCK_TYPE_PSP)) { + if (!adev->gmc.xgmi.pending_reset && + amdgpu_device_ip_get_ip_block(adev, AMD_IP_BLOCK_TYPE_PSP)) { list_for_each_entry(tmp_adev, &hive->device_list, gmc.xgmi.head) { /* update node list for other device in the hive */ if (tmp_adev != adev) { @@ -567,7 +575,7 @@ int amdgpu_xgmi_add_device(struct amdgpu_device *adev) } } - if (!ret) + if (!ret && !adev->gmc.xgmi.pending_reset) ret = amdgpu_xgmi_sysfs_add_dev_info(adev, hive); exit_unlock: @@ -620,7 +628,7 @@ int amdgpu_xgmi_remove_device(struct amdgpu_device *adev) return psp_xgmi_terminate(&adev->psp); } -int amdgpu_xgmi_ras_late_init(struct amdgpu_device *adev) +static int amdgpu_xgmi_ras_late_init(struct amdgpu_device *adev) { int r; struct ras_ih_if ih_info = { @@ -634,7 +642,7 @@ int amdgpu_xgmi_ras_late_init(struct amdgpu_device *adev) adev->gmc.xgmi.num_physical_nodes == 0) return 0; - amdgpu_xgmi_reset_ras_error_count(adev); + adev->gmc.xgmi.ras_funcs->reset_ras_error_count(adev); if (!adev->gmc.xgmi.ras_if) { adev->gmc.xgmi.ras_if = kmalloc(sizeof(struct ras_common_if), GFP_KERNEL); @@ -656,7 +664,7 @@ int amdgpu_xgmi_ras_late_init(struct amdgpu_device *adev) return r; } -void amdgpu_xgmi_ras_fini(struct amdgpu_device *adev) +static void amdgpu_xgmi_ras_fini(struct amdgpu_device *adev) { if (amdgpu_ras_is_supported(adev, AMDGPU_RAS_BLOCK__XGMI_WAFL) && adev->gmc.xgmi.ras_if) { @@ -683,7 +691,7 @@ static void pcs_clear_status(struct amdgpu_device *adev, uint32_t pcs_status_reg WREG32_PCIE(pcs_status_reg, 0); } -void amdgpu_xgmi_reset_ras_error_count(struct amdgpu_device *adev) +static void amdgpu_xgmi_reset_ras_error_count(struct amdgpu_device *adev) { uint32_t i; @@ -743,8 +751,8 @@ static int amdgpu_xgmi_query_pcs_error_status(struct amdgpu_device *adev, return 0; } -int amdgpu_xgmi_query_ras_error_count(struct amdgpu_device *adev, - void *ras_error_status) +static int amdgpu_xgmi_query_ras_error_count(struct amdgpu_device *adev, + void *ras_error_status) { struct ras_err_data *err_data = (struct ras_err_data *)ras_error_status; int i; @@ -793,10 +801,17 @@ int amdgpu_xgmi_query_ras_error_count(struct amdgpu_device *adev, break; } - amdgpu_xgmi_reset_ras_error_count(adev); + adev->gmc.xgmi.ras_funcs->reset_ras_error_count(adev); err_data->ue_count += ue_cnt; err_data->ce_count += ce_cnt; return 0; } + +const struct amdgpu_xgmi_ras_funcs xgmi_ras_funcs = { + .ras_late_init = amdgpu_xgmi_ras_late_init, + .ras_fini = amdgpu_xgmi_ras_fini, + .query_ras_error_count = amdgpu_xgmi_query_ras_error_count, + .reset_ras_error_count = amdgpu_xgmi_reset_ras_error_count, +}; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_xgmi.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_xgmi.h index 148560d63554..12969c0830d5 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_xgmi.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_xgmi.h @@ -50,6 +50,7 @@ struct amdgpu_pcs_ras_field { uint32_t pcs_err_shift; }; +extern const struct amdgpu_xgmi_ras_funcs xgmi_ras_funcs; struct amdgpu_hive_info *amdgpu_get_xgmi_hive(struct amdgpu_device *adev); void amdgpu_put_xgmi_hive(struct amdgpu_hive_info *hive); int amdgpu_xgmi_update_topology(struct amdgpu_hive_info *hive, struct amdgpu_device *adev); @@ -58,14 +59,8 @@ int amdgpu_xgmi_remove_device(struct amdgpu_device *adev); int amdgpu_xgmi_set_pstate(struct amdgpu_device *adev, int pstate); int amdgpu_xgmi_get_hops_count(struct amdgpu_device *adev, struct amdgpu_device *peer_adev); -int amdgpu_xgmi_ras_late_init(struct amdgpu_device *adev); -void amdgpu_xgmi_ras_fini(struct amdgpu_device *adev); uint64_t amdgpu_xgmi_get_relative_phy_addr(struct amdgpu_device *adev, uint64_t addr); -int amdgpu_xgmi_query_ras_error_count(struct amdgpu_device *adev, - void *ras_error_status); -void amdgpu_xgmi_reset_ras_error_count(struct amdgpu_device *adev); - static inline bool amdgpu_xgmi_same_hive(struct amdgpu_device *adev, struct amdgpu_device *bo_adev) { diff --git a/drivers/gpu/drm/amd/amdgpu/amdgv_sriovmsg.h b/drivers/gpu/drm/amd/amdgpu/amdgv_sriovmsg.h index 5355827ed0ae..1a8f6d4baab2 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgv_sriovmsg.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgv_sriovmsg.h @@ -90,11 +90,22 @@ union amd_sriov_msg_feature_flags { uint32_t host_flr_vramlost : 1; uint32_t mm_bw_management : 1; uint32_t pp_one_vf_mode : 1; - uint32_t reserved : 27; + uint32_t reg_indirect_acc : 1; + uint32_t reserved : 26; } flags; uint32_t all; }; +union amd_sriov_reg_access_flags { + struct { + uint32_t vf_reg_access_ih : 1; + uint32_t vf_reg_access_mmhub : 1; + uint32_t vf_reg_access_gc : 1; + uint32_t reserved : 29; + } flags; + uint32_t all; +}; + union amd_sriov_msg_os_info { struct { uint32_t windows : 1; @@ -149,8 +160,10 @@ struct amd_sriov_msg_pf2vf_info { /* identification in ROCm SMI */ uint64_t uuid; uint32_t fcn_idx; + /* flags which indicate the register access method VF should use */ + union amd_sriov_reg_access_flags reg_access_flags; /* reserved */ - uint32_t reserved[256-26]; + uint32_t reserved[256-27]; }; struct amd_sriov_msg_vf2pf_info_header { diff --git a/drivers/gpu/drm/amd/amdgpu/athub_v2_1.c b/drivers/gpu/drm/amd/amdgpu/athub_v2_1.c index 7b1b18350bf9..2ac4988ea0ff 100644 --- a/drivers/gpu/drm/amd/amdgpu/athub_v2_1.c +++ b/drivers/gpu/drm/amd/amdgpu/athub_v2_1.c @@ -74,10 +74,8 @@ int athub_v2_1_set_clockgating(struct amdgpu_device *adev, case CHIP_SIENNA_CICHLID: case CHIP_NAVY_FLOUNDER: case CHIP_DIMGREY_CAVEFISH: - athub_v2_1_update_medium_grain_clock_gating(adev, - state == AMD_CG_STATE_GATE ? true : false); - athub_v2_1_update_medium_grain_light_sleep(adev, - state == AMD_CG_STATE_GATE ? true : false); + athub_v2_1_update_medium_grain_clock_gating(adev, state == AMD_CG_STATE_GATE); + athub_v2_1_update_medium_grain_light_sleep(adev, state == AMD_CG_STATE_GATE); break; default: break; diff --git a/drivers/gpu/drm/amd/amdgpu/atom.c b/drivers/gpu/drm/amd/amdgpu/atom.c index 515890f4f5a0..3dcb8b32f48b 100644 --- a/drivers/gpu/drm/amd/amdgpu/atom.c +++ b/drivers/gpu/drm/amd/amdgpu/atom.c @@ -114,11 +114,11 @@ static uint32_t atom_iio_execute(struct atom_context *ctx, int base, base++; break; case ATOM_IIO_READ: - temp = ctx->card->ioreg_read(ctx->card, CU16(base + 1)); + temp = ctx->card->reg_read(ctx->card, CU16(base + 1)); base += 3; break; case ATOM_IIO_WRITE: - ctx->card->ioreg_write(ctx->card, CU16(base + 1), temp); + ctx->card->reg_write(ctx->card, CU16(base + 1), temp); base += 3; break; case ATOM_IIO_CLEAR: diff --git a/drivers/gpu/drm/amd/amdgpu/atom.h b/drivers/gpu/drm/amd/amdgpu/atom.h index 4205bbe5d8d7..d279759cab47 100644 --- a/drivers/gpu/drm/amd/amdgpu/atom.h +++ b/drivers/gpu/drm/amd/amdgpu/atom.h @@ -116,8 +116,6 @@ struct card_info { struct drm_device *dev; void (* reg_write)(struct card_info *, uint32_t, uint32_t); /* filled by driver */ uint32_t (* reg_read)(struct card_info *, uint32_t); /* filled by driver */ - void (* ioreg_write)(struct card_info *, uint32_t, uint32_t); /* filled by driver */ - uint32_t (* ioreg_read)(struct card_info *, uint32_t); /* filled by driver */ void (* mc_write)(struct card_info *, uint32_t, uint32_t); /* filled by driver */ uint32_t (* mc_read)(struct card_info *, uint32_t); /* filled by driver */ void (* pll_write)(struct card_info *, uint32_t, uint32_t); /* filled by driver */ diff --git a/drivers/gpu/drm/amd/amdgpu/cik.c b/drivers/gpu/drm/amd/amdgpu/cik.c index 4d6832cc7fb0..c0fcc41ee574 100644 --- a/drivers/gpu/drm/amd/amdgpu/cik.c +++ b/drivers/gpu/drm/amd/amdgpu/cik.c @@ -26,6 +26,8 @@ #include <linux/module.h> #include <linux/pci.h> +#include <drm/amdgpu_drm.h> + #include "amdgpu.h" #include "amdgpu_atombios.h" #include "amdgpu_ih.h" @@ -70,6 +72,80 @@ #include "amdgpu_amdkfd.h" #include "dce_virtual.h" +static const struct amdgpu_video_codec_info cik_video_codecs_encode_array[] = +{ + { + .codec_type = AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_MPEG4_AVC, + .max_width = 2048, + .max_height = 1152, + .max_pixels_per_frame = 2048 * 1152, + .max_level = 0, + }, +}; + +static const struct amdgpu_video_codecs cik_video_codecs_encode = +{ + .codec_count = ARRAY_SIZE(cik_video_codecs_encode_array), + .codec_array = cik_video_codecs_encode_array, +}; + +static const struct amdgpu_video_codec_info cik_video_codecs_decode_array[] = +{ + { + .codec_type = AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_MPEG2, + .max_width = 2048, + .max_height = 1152, + .max_pixels_per_frame = 2048 * 1152, + .max_level = 3, + }, + { + .codec_type = AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_MPEG4, + .max_width = 2048, + .max_height = 1152, + .max_pixels_per_frame = 2048 * 1152, + .max_level = 5, + }, + { + .codec_type = AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_MPEG4_AVC, + .max_width = 2048, + .max_height = 1152, + .max_pixels_per_frame = 2048 * 1152, + .max_level = 41, + }, + { + .codec_type = AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_VC1, + .max_width = 2048, + .max_height = 1152, + .max_pixels_per_frame = 2048 * 1152, + .max_level = 4, + }, +}; + +static const struct amdgpu_video_codecs cik_video_codecs_decode = +{ + .codec_count = ARRAY_SIZE(cik_video_codecs_decode_array), + .codec_array = cik_video_codecs_decode_array, +}; + +static int cik_query_video_codecs(struct amdgpu_device *adev, bool encode, + const struct amdgpu_video_codecs **codecs) +{ + switch (adev->asic_type) { + case CHIP_BONAIRE: + case CHIP_HAWAII: + case CHIP_KAVERI: + case CHIP_KABINI: + case CHIP_MULLINS: + if (encode) + *codecs = &cik_video_codecs_encode; + else + *codecs = &cik_video_codecs_decode; + return 0; + default: + return -EINVAL; + } +} + /* * Indirect registers accessor */ @@ -1933,6 +2009,7 @@ static const struct amdgpu_asic_funcs cik_asic_funcs = .get_pcie_replay_count = &cik_get_pcie_replay_count, .supports_baco = &cik_asic_supports_baco, .pre_asic_init = &cik_pre_asic_init, + .query_video_codecs = &cik_query_video_codecs, }; static int cik_common_early_init(void *handle) diff --git a/drivers/gpu/drm/amd/amdgpu/cik_sdma.c b/drivers/gpu/drm/amd/amdgpu/cik_sdma.c index 43b978144b79..c4bb8eed246d 100644 --- a/drivers/gpu/drm/amd/amdgpu/cik_sdma.c +++ b/drivers/gpu/drm/amd/amdgpu/cik_sdma.c @@ -984,10 +984,9 @@ static int cik_sdma_sw_init(void *handle) sprintf(ring->name, "sdma%d", i); r = amdgpu_ring_init(adev, ring, 1024, &adev->sdma.trap_irq, - (i == 0) ? - AMDGPU_SDMA_IRQ_INSTANCE0 : + (i == 0) ? AMDGPU_SDMA_IRQ_INSTANCE0 : AMDGPU_SDMA_IRQ_INSTANCE1, - AMDGPU_RING_PRIO_DEFAULT); + AMDGPU_RING_PRIO_DEFAULT, NULL); if (r) return r; } diff --git a/drivers/gpu/drm/amd/amdgpu/dce_v10_0.c b/drivers/gpu/drm/amd/amdgpu/dce_v10_0.c index 19abb740a169..d1570a462a51 100644 --- a/drivers/gpu/drm/amd/amdgpu/dce_v10_0.c +++ b/drivers/gpu/drm/amd/amdgpu/dce_v10_0.c @@ -1862,7 +1862,6 @@ static int dce_v10_0_crtc_do_set_base(struct drm_crtc *crtc, u32 tmp, viewport_w, viewport_h; int r; bool bypass_lut = false; - struct drm_format_name_buf format_name; /* no fb bound */ if (!atomic && !crtc->primary->fb) { @@ -1981,8 +1980,8 @@ static int dce_v10_0_crtc_do_set_base(struct drm_crtc *crtc, #endif break; default: - DRM_ERROR("Unsupported screen format %s\n", - drm_get_format_name(target_fb->format->format, &format_name)); + DRM_ERROR("Unsupported screen format %p4cc\n", + &target_fb->format->format); return -EINVAL; } diff --git a/drivers/gpu/drm/amd/amdgpu/dce_v11_0.c b/drivers/gpu/drm/amd/amdgpu/dce_v11_0.c index 320ec35bfd37..18a7b3bd633b 100644 --- a/drivers/gpu/drm/amd/amdgpu/dce_v11_0.c +++ b/drivers/gpu/drm/amd/amdgpu/dce_v11_0.c @@ -1904,7 +1904,6 @@ static int dce_v11_0_crtc_do_set_base(struct drm_crtc *crtc, u32 tmp, viewport_w, viewport_h; int r; bool bypass_lut = false; - struct drm_format_name_buf format_name; /* no fb bound */ if (!atomic && !crtc->primary->fb) { @@ -2023,8 +2022,8 @@ static int dce_v11_0_crtc_do_set_base(struct drm_crtc *crtc, #endif break; default: - DRM_ERROR("Unsupported screen format %s\n", - drm_get_format_name(target_fb->format->format, &format_name)); + DRM_ERROR("Unsupported screen format %p4cc\n", + &target_fb->format->format); return -EINVAL; } diff --git a/drivers/gpu/drm/amd/amdgpu/dce_v6_0.c b/drivers/gpu/drm/amd/amdgpu/dce_v6_0.c index 13322000ebd6..dbcb09cf83e6 100644 --- a/drivers/gpu/drm/amd/amdgpu/dce_v6_0.c +++ b/drivers/gpu/drm/amd/amdgpu/dce_v6_0.c @@ -1820,7 +1820,6 @@ static int dce_v6_0_crtc_do_set_base(struct drm_crtc *crtc, u32 viewport_w, viewport_h; int r; bool bypass_lut = false; - struct drm_format_name_buf format_name; /* no fb bound */ if (!atomic && !crtc->primary->fb) { @@ -1929,8 +1928,8 @@ static int dce_v6_0_crtc_do_set_base(struct drm_crtc *crtc, #endif break; default: - DRM_ERROR("Unsupported screen format %s\n", - drm_get_format_name(target_fb->format->format, &format_name)); + DRM_ERROR("Unsupported screen format %p4cc\n", + &target_fb->format->format); return -EINVAL; } diff --git a/drivers/gpu/drm/amd/amdgpu/dce_v8_0.c b/drivers/gpu/drm/amd/amdgpu/dce_v8_0.c index 04ebf02e5b8c..b200b9e722d9 100644 --- a/drivers/gpu/drm/amd/amdgpu/dce_v8_0.c +++ b/drivers/gpu/drm/amd/amdgpu/dce_v8_0.c @@ -1791,7 +1791,6 @@ static int dce_v8_0_crtc_do_set_base(struct drm_crtc *crtc, u32 viewport_w, viewport_h; int r; bool bypass_lut = false; - struct drm_format_name_buf format_name; /* no fb bound */ if (!atomic && !crtc->primary->fb) { @@ -1902,8 +1901,8 @@ static int dce_v8_0_crtc_do_set_base(struct drm_crtc *crtc, #endif break; default: - DRM_ERROR("Unsupported screen format %s\n", - drm_get_format_name(target_fb->format->format, &format_name)); + DRM_ERROR("Unsupported screen format %p4cc\n", + &target_fb->format->format); return -EINVAL; } diff --git a/drivers/gpu/drm/amd/amdgpu/df_v3_6.c b/drivers/gpu/drm/amd/amdgpu/df_v3_6.c index 6b4b30a8dce5..0d8459d63bac 100644 --- a/drivers/gpu/drm/amd/amdgpu/df_v3_6.c +++ b/drivers/gpu/drm/amd/amdgpu/df_v3_6.c @@ -205,7 +205,7 @@ static ssize_t df_v3_6_get_df_cntr_avail(struct device *dev, count++; } - return snprintf(buf, PAGE_SIZE, "%i\n", count); + return sysfs_emit(buf, "%i\n", count); } /* device attr for available perfmon counters */ @@ -568,6 +568,8 @@ static int df_v3_6_pmc_stop(struct amdgpu_device *adev, uint64_t config, if (ret) return ret; + df_v3_6_perfmon_wreg(adev, lo_base_addr, lo_val, + hi_base_addr, hi_val); if (is_remove) { df_v3_6_reset_perfmon_cntr(adev, config, counter_idx); diff --git a/drivers/gpu/drm/amd/amdgpu/dimgrey_cavefish_reg_init.c b/drivers/gpu/drm/amd/amdgpu/dimgrey_cavefish_reg_init.c index e9f177e9e3cf..e9f177e9e3cf 100755..100644 --- a/drivers/gpu/drm/amd/amdgpu/dimgrey_cavefish_reg_init.c +++ b/drivers/gpu/drm/amd/amdgpu/dimgrey_cavefish_reg_init.c diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v10_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v10_0.c index 45d1172b7bff..2408ed4c7d84 100644 --- a/drivers/gpu/drm/amd/amdgpu/gfx_v10_0.c +++ b/drivers/gpu/drm/amd/amdgpu/gfx_v10_0.c @@ -29,7 +29,6 @@ #include "amdgpu.h" #include "amdgpu_gfx.h" #include "amdgpu_psp.h" -#include "amdgpu_smu.h" #include "nv.h" #include "nvd.h" @@ -174,6 +173,11 @@ #define mmGC_THROTTLE_CTRL_Sienna_Cichlid 0x2030 #define mmGC_THROTTLE_CTRL_Sienna_Cichlid_BASE_IDX 0 +#define GFX_RLCG_GC_WRITE_OLD (0x8 << 28) +#define GFX_RLCG_GC_WRITE (0x0 << 28) +#define GFX_RLCG_GC_READ (0x1 << 28) +#define GFX_RLCG_MMHUB_WRITE (0x2 << 28) + MODULE_FIRMWARE("amdgpu/navi10_ce.bin"); MODULE_FIRMWARE("amdgpu/navi10_pfp.bin"); MODULE_FIRMWARE("amdgpu/navi10_me.bin"); @@ -1419,38 +1423,127 @@ static const struct soc15_reg_golden golden_settings_gc_10_1_2[] = SOC15_REG_GOLDEN_VALUE(GC, 0, mmUTCL1_CTRL, 0xffffffff, 0x00800000) }; -static void gfx_v10_rlcg_wreg(struct amdgpu_device *adev, u32 offset, u32 v) +static bool gfx_v10_is_rlcg_rw(struct amdgpu_device *adev, u32 offset, uint32_t *flag, bool write) +{ + /* always programed by rlcg, only for gc */ + if (offset == SOC15_REG_OFFSET(GC, 0, mmRLC_CSIB_ADDR_HI) || + offset == SOC15_REG_OFFSET(GC, 0, mmRLC_CSIB_ADDR_LO) || + offset == SOC15_REG_OFFSET(GC, 0, mmRLC_CSIB_LENGTH) || + offset == SOC15_REG_OFFSET(GC, 0, mmGRBM_GFX_CNTL) || + offset == SOC15_REG_OFFSET(GC, 0, mmGRBM_GFX_INDEX) || + offset == SOC15_REG_OFFSET(GC, 0, mmCP_ME_CNTL)) { + if (!amdgpu_sriov_reg_indirect_gc(adev)) + *flag = GFX_RLCG_GC_WRITE_OLD; + else + *flag = write ? GFX_RLCG_GC_WRITE : GFX_RLCG_GC_READ; + + return true; + } + + /* currently support gc read/write, mmhub write */ + if (offset >= SOC15_REG_OFFSET(GC, 0, mmSDMA0_DEC_START) && + offset <= SOC15_REG_OFFSET(GC, 0, mmRLC_GTS_OFFSET_MSB)) { + if (amdgpu_sriov_reg_indirect_gc(adev)) + *flag = write ? GFX_RLCG_GC_WRITE : GFX_RLCG_GC_READ; + else + return false; + } else { + if (amdgpu_sriov_reg_indirect_mmhub(adev)) + *flag = GFX_RLCG_MMHUB_WRITE; + else + return false; + } + + return true; +} + +static u32 gfx_v10_rlcg_rw(struct amdgpu_device *adev, u32 offset, u32 v, uint32_t flag) { static void *scratch_reg0; static void *scratch_reg1; + static void *scratch_reg2; + static void *scratch_reg3; static void *spare_int; + static uint32_t grbm_cntl; + static uint32_t grbm_idx; uint32_t i = 0; uint32_t retries = 50000; + u32 ret = 0; + + scratch_reg0 = adev->rmmio + + (adev->reg_offset[GC_HWIP][0][mmSCRATCH_REG0_BASE_IDX] + mmSCRATCH_REG0) * 4; + scratch_reg1 = adev->rmmio + + (adev->reg_offset[GC_HWIP][0][mmSCRATCH_REG1_BASE_IDX] + mmSCRATCH_REG1) * 4; + scratch_reg2 = adev->rmmio + + (adev->reg_offset[GC_HWIP][0][mmSCRATCH_REG0_BASE_IDX] + mmSCRATCH_REG2) * 4; + scratch_reg3 = adev->rmmio + + (adev->reg_offset[GC_HWIP][0][mmSCRATCH_REG1_BASE_IDX] + mmSCRATCH_REG3) * 4; + spare_int = adev->rmmio + + (adev->reg_offset[GC_HWIP][0][mmRLC_SPARE_INT_BASE_IDX] + mmRLC_SPARE_INT) * 4; + + grbm_cntl = adev->reg_offset[GC_HWIP][0][mmGRBM_GFX_CNTL_BASE_IDX] + mmGRBM_GFX_CNTL; + grbm_idx = adev->reg_offset[GC_HWIP][0][mmGRBM_GFX_INDEX_BASE_IDX] + mmGRBM_GFX_INDEX; + + if (offset == grbm_cntl || offset == grbm_idx) { + if (offset == grbm_cntl) + writel(v, scratch_reg2); + else if (offset == grbm_idx) + writel(v, scratch_reg3); + + writel(v, ((void __iomem *)adev->rmmio) + (offset * 4)); + } else { + writel(v, scratch_reg0); + writel(offset | flag, scratch_reg1); + writel(1, spare_int); + for (i = 0; i < retries; i++) { + u32 tmp; + + tmp = readl(scratch_reg1); + if (!(tmp & flag)) + break; - scratch_reg0 = adev->rmmio + (adev->reg_offset[GC_HWIP][0][mmSCRATCH_REG0_BASE_IDX] + mmSCRATCH_REG0)*4; - scratch_reg1 = adev->rmmio + (adev->reg_offset[GC_HWIP][0][mmSCRATCH_REG1_BASE_IDX] + mmSCRATCH_REG1)*4; - spare_int = adev->rmmio + (adev->reg_offset[GC_HWIP][0][mmRLC_SPARE_INT_BASE_IDX] + mmRLC_SPARE_INT)*4; + udelay(10); + } - if (amdgpu_sriov_runtime(adev)) { - pr_err("shouldn't call rlcg write register during runtime\n"); - return; + if (i >= retries) + pr_err("timeout: rlcg program reg:0x%05x failed !\n", offset); } - writel(v, scratch_reg0); - writel(offset | 0x80000000, scratch_reg1); - writel(1, spare_int); - for (i = 0; i < retries; i++) { - u32 tmp; + ret = readl(scratch_reg0); - tmp = readl(scratch_reg1); - if (!(tmp & 0x80000000)) - break; + return ret; +} + +static void gfx_v10_rlcg_wreg(struct amdgpu_device *adev, u32 offset, u32 value, u32 flag) +{ + uint32_t rlcg_flag; - udelay(10); + if (amdgpu_sriov_fullaccess(adev) && + gfx_v10_is_rlcg_rw(adev, offset, &rlcg_flag, 1)) { + gfx_v10_rlcg_rw(adev, offset, value, rlcg_flag); + + return; } + if (flag & AMDGPU_REGS_NO_KIQ) + WREG32_NO_KIQ(offset, value); + else + WREG32(offset, value); +} - if (i >= retries) - pr_err("timeout: rlcg program reg:0x%05x failed !\n", offset); +static u32 gfx_v10_rlcg_rreg(struct amdgpu_device *adev, u32 offset, u32 flag) +{ + uint32_t rlcg_flag; + + if (amdgpu_sriov_fullaccess(adev) && + gfx_v10_is_rlcg_rw(adev, offset, &rlcg_flag, 0)) + return gfx_v10_rlcg_rw(adev, offset, 0, rlcg_flag); + + if (flag & AMDGPU_REGS_NO_KIQ) + return RREG32_NO_KIQ(offset); + else + return RREG32(offset); + + return 0; } static const struct soc15_reg_golden golden_settings_gc_10_1_nv14[] = @@ -3280,7 +3373,7 @@ static const struct soc15_reg_golden golden_settings_gc_10_3_4[] = SOC15_REG_GOLDEN_VALUE(GC, 0, mmCPF_GCR_CNTL, 0x0007ffff, 0x0000c000), SOC15_REG_GOLDEN_VALUE(GC, 0, mmDB_DEBUG3, 0x00000280, 0x00000280), SOC15_REG_GOLDEN_VALUE(GC, 0, mmDB_DEBUG4, 0x07800000, 0x00800000), - SOC15_REG_GOLDEN_VALUE(GC, 0, mmGCR_GENERAL_CNTL, 0x00001d00, 0x00000500), + SOC15_REG_GOLDEN_VALUE(GC, 0, mmGCR_GENERAL_CNTL_Sienna_Cichlid, 0x00001d00, 0x00000500), SOC15_REG_GOLDEN_VALUE(GC, 0, mmGE_PC_CNTL, 0x003c0000, 0x00280400), SOC15_REG_GOLDEN_VALUE(GC, 0, mmGL2A_ADDR_MATCH_MASK, 0xffffffff, 0xffffffcf), SOC15_REG_GOLDEN_VALUE(GC, 0, mmGL2C_ADDR_MATCH_MASK, 0xffffffff, 0xffffffcf), @@ -4459,9 +4552,8 @@ static int gfx_v10_0_gfx_ring_init(struct amdgpu_device *adev, int ring_id, sprintf(ring->name, "gfx_%d.%d.%d", ring->me, ring->pipe, ring->queue); irq_type = AMDGPU_CP_IRQ_GFX_ME0_PIPE0_EOP + ring->pipe; - r = amdgpu_ring_init(adev, ring, 1024, - &adev->gfx.eop_irq, irq_type, - AMDGPU_RING_PRIO_DEFAULT); + r = amdgpu_ring_init(adev, ring, 1024, &adev->gfx.eop_irq, irq_type, + AMDGPU_RING_PRIO_DEFAULT, NULL); if (r) return r; return 0; @@ -4495,8 +4587,8 @@ static int gfx_v10_0_compute_ring_init(struct amdgpu_device *adev, int ring_id, hw_prio = amdgpu_gfx_is_high_priority_compute_queue(adev, ring) ? AMDGPU_GFX_PIPE_PRIO_HIGH : AMDGPU_GFX_PIPE_PRIO_NORMAL; /* type-2 packets are deprecated on MEC, use type-3 instead */ - r = amdgpu_ring_init(adev, ring, 1024, - &adev->gfx.eop_irq, irq_type, hw_prio); + r = amdgpu_ring_init(adev, ring, 1024, &adev->gfx.eop_irq, irq_type, + hw_prio, NULL); if (r) return r; @@ -7172,16 +7264,10 @@ static int gfx_v10_0_hw_init(void *handle) * loaded firstly, so in direct type, it has to load smc ucode * here before rlc. */ - if (adev->smu.ppt_funcs != NULL && !(adev->flags & AMD_IS_APU)) { - r = smu_load_microcode(&adev->smu); + if (!(adev->flags & AMD_IS_APU)) { + r = amdgpu_pm_load_smu_firmware(adev, NULL); if (r) return r; - - r = smu_check_fw_status(&adev->smu); - if (r) { - pr_err("SMC firmware status is not correct\n"); - return r; - } } gfx_v10_0_disable_gpa_mode(adev); } @@ -7892,6 +7978,7 @@ static const struct amdgpu_rlc_funcs gfx_v10_0_rlc_funcs_sriov = { .start = gfx_v10_0_rlc_start, .update_spm_vmid = gfx_v10_0_update_spm_vmid, .rlcg_wreg = gfx_v10_rlcg_wreg, + .rlcg_rreg = gfx_v10_rlcg_rreg, .is_rlcg_access_range = gfx_v10_0_is_rlcg_access_range, }; diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v6_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v6_0.c index ca74638dec9b..3a8d52a54873 100644 --- a/drivers/gpu/drm/amd/amdgpu/gfx_v6_0.c +++ b/drivers/gpu/drm/amd/amdgpu/gfx_v6_0.c @@ -3114,7 +3114,7 @@ static int gfx_v6_0_sw_init(void *handle) r = amdgpu_ring_init(adev, ring, 1024, &adev->gfx.eop_irq, AMDGPU_CP_IRQ_GFX_ME0_PIPE0_EOP, - AMDGPU_RING_PRIO_DEFAULT); + AMDGPU_RING_PRIO_DEFAULT, NULL); if (r) return r; } @@ -3137,7 +3137,7 @@ static int gfx_v6_0_sw_init(void *handle) irq_type = AMDGPU_CP_IRQ_COMPUTE_MEC1_PIPE0_EOP + ring->pipe; r = amdgpu_ring_init(adev, ring, 1024, &adev->gfx.eop_irq, irq_type, - AMDGPU_RING_PRIO_DEFAULT); + AMDGPU_RING_PRIO_DEFAULT, NULL); if (r) return r; } diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v7_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v7_0.c index a368724c3dfc..c35fdd2ef2d4 100644 --- a/drivers/gpu/drm/amd/amdgpu/gfx_v7_0.c +++ b/drivers/gpu/drm/amd/amdgpu/gfx_v7_0.c @@ -1877,7 +1877,7 @@ static void gfx_v7_0_init_compute_vmid(struct amdgpu_device *adev) mutex_unlock(&adev->srbm_mutex); /* Initialize all compute VMIDs to have no GDS, GWS, or OA - acccess. These should be enabled by FW for target VMIDs. */ + access. These should be enabled by FW for target VMIDs. */ for (i = adev->vm_manager.first_kfd_vmid; i < AMDGPU_NUM_VMID; i++) { WREG32(amdgpu_gds_reg_offset[i].mem_base, 0); WREG32(amdgpu_gds_reg_offset[i].mem_size, 0); @@ -2058,7 +2058,7 @@ static void gfx_v7_0_constants_init(struct amdgpu_device *adev) * @adev: amdgpu_device pointer * * Set up the number and offset of the CP scratch registers. - * NOTE: use of CP scratch registers is a legacy inferface and + * NOTE: use of CP scratch registers is a legacy interface and * is not used by default on newer asics (r6xx+). On newer asics, * memory buffers are used for fences rather than scratch regs. */ @@ -2172,7 +2172,7 @@ static void gfx_v7_0_ring_emit_vgt_flush(struct amdgpu_ring *ring) * @seq: sequence number * @flags: fence related flags * - * Emits a fence sequnce number on the gfx ring and flushes + * Emits a fence sequence number on the gfx ring and flushes * GPU caches. */ static void gfx_v7_0_ring_emit_fence_gfx(struct amdgpu_ring *ring, u64 addr, @@ -2215,7 +2215,7 @@ static void gfx_v7_0_ring_emit_fence_gfx(struct amdgpu_ring *ring, u64 addr, * @seq: sequence number * @flags: fence related flags * - * Emits a fence sequnce number on the compute ring and flushes + * Emits a fence sequence number on the compute ring and flushes * GPU caches. */ static void gfx_v7_0_ring_emit_fence_compute(struct amdgpu_ring *ring, @@ -2245,14 +2245,14 @@ static void gfx_v7_0_ring_emit_fence_compute(struct amdgpu_ring *ring, * gfx_v7_0_ring_emit_ib - emit an IB (Indirect Buffer) on the ring * * @ring: amdgpu_ring structure holding ring information - * @job: job to retrive vmid from + * @job: job to retrieve vmid from * @ib: amdgpu indirect buffer object * @flags: options (AMDGPU_HAVE_CTX_SWITCH) * * Emits an DE (drawing engine) or CE (constant engine) IB * on the gfx ring. IBs are usually generated by userspace * acceleration drivers and submitted to the kernel for - * sheduling on the ring. This function schedules the IB + * scheduling on the ring. This function schedules the IB * on the gfx ring for execution by the GPU. */ static void gfx_v7_0_ring_emit_ib_gfx(struct amdgpu_ring *ring, @@ -2402,7 +2402,7 @@ err1: /* * CP. - * On CIK, gfx and compute now have independant command processors. + * On CIK, gfx and compute now have independent command processors. * * GFX * Gfx consists of a single ring and can process both gfx jobs and @@ -2630,7 +2630,7 @@ static int gfx_v7_0_cp_gfx_resume(struct amdgpu_device *adev) ring->wptr = 0; WREG32(mmCP_RB0_WPTR, lower_32_bits(ring->wptr)); - /* set the wb address wether it's enabled or not */ + /* set the wb address whether it's enabled or not */ rptr_addr = adev->wb.gpu_addr + (ring->rptr_offs * 4); WREG32(mmCP_RB0_RPTR_ADDR, lower_32_bits(rptr_addr)); WREG32(mmCP_RB0_RPTR_ADDR_HI, upper_32_bits(rptr_addr) & 0xFF); @@ -2985,7 +2985,7 @@ static void gfx_v7_0_mqd_init(struct amdgpu_device *adev, mqd->cp_hqd_pq_wptr_poll_addr_lo = wb_gpu_addr & 0xfffffffc; mqd->cp_hqd_pq_wptr_poll_addr_hi = upper_32_bits(wb_gpu_addr) & 0xffff; - /* set the wb address wether it's enabled or not */ + /* set the wb address whether it's enabled or not */ wb_gpu_addr = adev->wb.gpu_addr + (ring->rptr_offs * 4); mqd->cp_hqd_pq_rptr_report_addr_lo = wb_gpu_addr & 0xfffffffc; mqd->cp_hqd_pq_rptr_report_addr_hi = @@ -3198,7 +3198,7 @@ static int gfx_v7_0_cp_resume(struct amdgpu_device *adev) /** * gfx_v7_0_ring_emit_vm_flush - cik vm flush using the CP * - * @ring: the ring to emmit the commands to + * @ring: the ring to emit the commands to * * Sync the command pipeline with the PFP. E.g. wait for everything * to be completed. @@ -3220,7 +3220,7 @@ static void gfx_v7_0_ring_emit_pipeline_sync(struct amdgpu_ring *ring) amdgpu_ring_write(ring, 4); /* poll interval */ if (usepfp) { - /* synce CE with ME to prevent CE fetch CEIB before context switch done */ + /* sync CE with ME to prevent CE fetch CEIB before context switch done */ amdgpu_ring_write(ring, PACKET3(PACKET3_SWITCH_BUFFER, 0)); amdgpu_ring_write(ring, 0); amdgpu_ring_write(ring, PACKET3(PACKET3_SWITCH_BUFFER, 0)); @@ -4438,7 +4438,7 @@ static int gfx_v7_0_compute_ring_init(struct amdgpu_device *adev, int ring_id, /* type-2 packets are deprecated on MEC, use type-3 instead */ r = amdgpu_ring_init(adev, ring, 1024, &adev->gfx.eop_irq, irq_type, - AMDGPU_RING_PRIO_DEFAULT); + AMDGPU_RING_PRIO_DEFAULT, NULL); if (r) return r; @@ -4512,7 +4512,7 @@ static int gfx_v7_0_sw_init(void *handle) r = amdgpu_ring_init(adev, ring, 1024, &adev->gfx.eop_irq, AMDGPU_CP_IRQ_GFX_ME0_PIPE0_EOP, - AMDGPU_RING_PRIO_DEFAULT); + AMDGPU_RING_PRIO_DEFAULT, NULL); if (r) return r; } diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c index 84d2eaa38101..c26e06059466 100644 --- a/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c +++ b/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c @@ -1927,8 +1927,8 @@ static int gfx_v8_0_compute_ring_init(struct amdgpu_device *adev, int ring_id, hw_prio = amdgpu_gfx_is_high_priority_compute_queue(adev, ring) ? AMDGPU_GFX_PIPE_PRIO_HIGH : AMDGPU_RING_PRIO_DEFAULT; /* type-2 packets are deprecated on MEC, use type-3 instead */ - r = amdgpu_ring_init(adev, ring, 1024, - &adev->gfx.eop_irq, irq_type, hw_prio); + r = amdgpu_ring_init(adev, ring, 1024, &adev->gfx.eop_irq, irq_type, + hw_prio, NULL); if (r) return r; @@ -2033,7 +2033,7 @@ static int gfx_v8_0_sw_init(void *handle) r = amdgpu_ring_init(adev, ring, 1024, &adev->gfx.eop_irq, AMDGPU_CP_IRQ_GFX_ME0_PIPE0_EOP, - AMDGPU_RING_PRIO_DEFAULT); + AMDGPU_RING_PRIO_DEFAULT, NULL); if (r) return r; } @@ -6718,7 +6718,8 @@ static int gfx_v8_0_cp_ecc_error_irq(struct amdgpu_device *adev, return 0; } -static void gfx_v8_0_parse_sq_irq(struct amdgpu_device *adev, unsigned ih_data) +static void gfx_v8_0_parse_sq_irq(struct amdgpu_device *adev, unsigned ih_data, + bool from_wq) { u32 enc, se_id, sh_id, cu_id; char type[20]; @@ -6756,7 +6757,7 @@ static void gfx_v8_0_parse_sq_irq(struct amdgpu_device *adev, unsigned ih_data) * or from BH in which case we can access SQ_EDC_INFO * instance */ - if (in_task()) { + if (from_wq) { mutex_lock(&adev->grbm_idx_mutex); gfx_v8_0_select_se_sh(adev, se_id, sh_id, cu_id); @@ -6794,7 +6795,7 @@ static void gfx_v8_0_sq_irq_work_func(struct work_struct *work) struct amdgpu_device *adev = container_of(work, struct amdgpu_device, gfx.sq_work.work); struct sq_work *sq_work = container_of(work, struct sq_work, work); - gfx_v8_0_parse_sq_irq(adev, sq_work->ih_data); + gfx_v8_0_parse_sq_irq(adev, sq_work->ih_data, true); } static int gfx_v8_0_sq_irq(struct amdgpu_device *adev, @@ -6809,7 +6810,7 @@ static int gfx_v8_0_sq_irq(struct amdgpu_device *adev, * just print whatever info is possible directly from the ISR. */ if (work_pending(&adev->gfx.sq_work.work)) { - gfx_v8_0_parse_sq_irq(adev, ih_data); + gfx_v8_0_parse_sq_irq(adev, ih_data, false); } else { adev->gfx.sq_work.ih_data = ih_data; schedule_work(&adev->gfx.sq_work.work); diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c index 65db88bb6cbc..a078a38c2cee 100644 --- a/drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c +++ b/drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c @@ -49,6 +49,7 @@ #include "gfx_v9_4.h" #include "gfx_v9_0.h" +#include "gfx_v9_4_2.h" #include "asic_reg/pwr/pwr_10_0_offset.h" #include "asic_reg/pwr/pwr_10_0_sh_mask.h" @@ -107,14 +108,12 @@ MODULE_FIRMWARE("amdgpu/raven2_rlc.bin"); MODULE_FIRMWARE("amdgpu/raven_kicker_rlc.bin"); MODULE_FIRMWARE("amdgpu/arcturus_mec.bin"); -MODULE_FIRMWARE("amdgpu/arcturus_mec2.bin"); MODULE_FIRMWARE("amdgpu/arcturus_rlc.bin"); MODULE_FIRMWARE("amdgpu/renoir_ce.bin"); MODULE_FIRMWARE("amdgpu/renoir_pfp.bin"); MODULE_FIRMWARE("amdgpu/renoir_me.bin"); MODULE_FIRMWARE("amdgpu/renoir_mec.bin"); -MODULE_FIRMWARE("amdgpu/renoir_mec2.bin"); MODULE_FIRMWARE("amdgpu/renoir_rlc.bin"); MODULE_FIRMWARE("amdgpu/green_sardine_ce.bin"); @@ -124,6 +123,10 @@ MODULE_FIRMWARE("amdgpu/green_sardine_mec.bin"); MODULE_FIRMWARE("amdgpu/green_sardine_mec2.bin"); MODULE_FIRMWARE("amdgpu/green_sardine_rlc.bin"); +MODULE_FIRMWARE("amdgpu/aldebaran_mec.bin"); +MODULE_FIRMWARE("amdgpu/aldebaran_mec2.bin"); +MODULE_FIRMWARE("amdgpu/aldebaran_rlc.bin"); + #define mmTCP_CHAN_STEER_0_ARCT 0x0b03 #define mmTCP_CHAN_STEER_0_ARCT_BASE_IDX 0 #define mmTCP_CHAN_STEER_1_ARCT 0x0b04 @@ -731,7 +734,7 @@ static const u32 GFX_RLC_SRM_INDEX_CNTL_DATA_OFFSETS[] = mmRLC_SRM_INDEX_CNTL_DATA_7 - mmRLC_SRM_INDEX_CNTL_DATA_0, }; -static void gfx_v9_0_rlcg_wreg(struct amdgpu_device *adev, u32 offset, u32 v) +static void gfx_v9_0_rlcg_rw(struct amdgpu_device *adev, u32 offset, u32 v, u32 flag) { static void *scratch_reg0; static void *scratch_reg1; @@ -784,6 +787,20 @@ static void gfx_v9_0_rlcg_wreg(struct amdgpu_device *adev, u32 offset, u32 v) } +static void gfx_v9_0_rlcg_wreg(struct amdgpu_device *adev, u32 offset, u32 v, u32 flag) +{ + if (amdgpu_sriov_fullaccess(adev)) { + gfx_v9_0_rlcg_rw(adev, offset, v, flag); + + return; + } + + if (flag & AMDGPU_REGS_NO_KIQ) + WREG32_NO_KIQ(offset, v); + else + WREG32(offset, v); +} + #define VEGA10_GB_ADDR_CONFIG_GOLDEN 0x2a114042 #define VEGA12_GB_ADDR_CONFIG_GOLDEN 0x24104041 #define RAVEN_GB_ADDR_CONFIG_GOLDEN 0x24000042 @@ -981,11 +998,16 @@ static void gfx_v9_0_init_golden_registers(struct amdgpu_device *adev) golden_settings_gc_9_1_rn, ARRAY_SIZE(golden_settings_gc_9_1_rn)); return; /* for renoir, don't need common goldensetting */ + case CHIP_ALDEBARAN: + gfx_v9_4_2_init_golden_registers(adev, + adev->smuio.funcs->get_die_id(adev)); + break; default: break; } - if (adev->asic_type != CHIP_ARCTURUS) + if ((adev->asic_type != CHIP_ARCTURUS) && + (adev->asic_type != CHIP_ALDEBARAN)) soc15_program_register_sequence(adev, golden_settings_gc_9_x_common, (const u32)ARRAY_SIZE(golden_settings_gc_9_x_common)); } @@ -1517,6 +1539,16 @@ out: return err; } +static bool gfx_v9_0_load_mec2_fw_bin_support(struct amdgpu_device *adev) +{ + if (adev->asic_type == CHIP_ALDEBARAN || + adev->asic_type == CHIP_ARCTURUS || + adev->asic_type == CHIP_RENOIR) + return false; + + return true; +} + static int gfx_v9_0_init_cp_compute_microcode(struct amdgpu_device *adev, const char *chip_name) { @@ -1538,21 +1570,26 @@ static int gfx_v9_0_init_cp_compute_microcode(struct amdgpu_device *adev, adev->gfx.mec_feature_version = le32_to_cpu(cp_hdr->ucode_feature_version); - snprintf(fw_name, sizeof(fw_name), "amdgpu/%s_mec2.bin", chip_name); - err = request_firmware(&adev->gfx.mec2_fw, fw_name, adev->dev); - if (!err) { - err = amdgpu_ucode_validate(adev->gfx.mec2_fw); - if (err) - goto out; - cp_hdr = (const struct gfx_firmware_header_v1_0 *) - adev->gfx.mec2_fw->data; - adev->gfx.mec2_fw_version = - le32_to_cpu(cp_hdr->header.ucode_version); - adev->gfx.mec2_feature_version = - le32_to_cpu(cp_hdr->ucode_feature_version); + if (gfx_v9_0_load_mec2_fw_bin_support(adev)) { + snprintf(fw_name, sizeof(fw_name), "amdgpu/%s_mec2.bin", chip_name); + err = request_firmware(&adev->gfx.mec2_fw, fw_name, adev->dev); + if (!err) { + err = amdgpu_ucode_validate(adev->gfx.mec2_fw); + if (err) + goto out; + cp_hdr = (const struct gfx_firmware_header_v1_0 *) + adev->gfx.mec2_fw->data; + adev->gfx.mec2_fw_version = + le32_to_cpu(cp_hdr->header.ucode_version); + adev->gfx.mec2_feature_version = + le32_to_cpu(cp_hdr->ucode_feature_version); + } else { + err = 0; + adev->gfx.mec2_fw = NULL; + } } else { - err = 0; - adev->gfx.mec2_fw = NULL; + adev->gfx.mec2_fw_version = adev->gfx.mec_fw_version; + adev->gfx.mec2_feature_version = adev->gfx.mec_feature_version; } if (adev->firmware.load_type == AMDGPU_FW_LOAD_PSP) { @@ -1581,8 +1618,7 @@ static int gfx_v9_0_init_cp_compute_microcode(struct amdgpu_device *adev, /* TODO: Determine if MEC2 JT FW loading can be removed for all GFX V9 asic and above */ - if (adev->asic_type != CHIP_ARCTURUS && - adev->asic_type != CHIP_RENOIR) { + if (gfx_v9_0_load_mec2_fw_bin_support(adev)) { info = &adev->firmware.ucode[AMDGPU_UCODE_ID_CP_MEC2_JT]; info->ucode_id = AMDGPU_UCODE_ID_CP_MEC2_JT; info->fw = adev->gfx.mec2_fw; @@ -1642,6 +1678,9 @@ static int gfx_v9_0_init_microcode(struct amdgpu_device *adev) else chip_name = "green_sardine"; break; + case CHIP_ALDEBARAN: + chip_name = "aldebaran"; + break; default: BUG(); } @@ -1882,7 +1921,10 @@ static void gfx_v9_0_enable_lbpw(struct amdgpu_device *adev, bool enable) static int gfx_v9_0_cp_jump_table_num(struct amdgpu_device *adev) { - return 5; + if (gfx_v9_0_load_mec2_fw_bin_support(adev)) + return 5; + else + return 4; } static int gfx_v9_0_rlc_init(struct amdgpu_device *adev) @@ -2064,30 +2106,22 @@ static void gfx_v9_0_select_me_pipe_q(struct amdgpu_device *adev, } static const struct amdgpu_gfx_funcs gfx_v9_0_gfx_funcs = { - .get_gpu_clock_counter = &gfx_v9_0_get_gpu_clock_counter, - .select_se_sh = &gfx_v9_0_select_se_sh, - .read_wave_data = &gfx_v9_0_read_wave_data, - .read_wave_sgprs = &gfx_v9_0_read_wave_sgprs, - .read_wave_vgprs = &gfx_v9_0_read_wave_vgprs, - .select_me_pipe_q = &gfx_v9_0_select_me_pipe_q, + .get_gpu_clock_counter = &gfx_v9_0_get_gpu_clock_counter, + .select_se_sh = &gfx_v9_0_select_se_sh, + .read_wave_data = &gfx_v9_0_read_wave_data, + .read_wave_sgprs = &gfx_v9_0_read_wave_sgprs, + .read_wave_vgprs = &gfx_v9_0_read_wave_vgprs, + .select_me_pipe_q = &gfx_v9_0_select_me_pipe_q, +}; + +static const struct amdgpu_gfx_ras_funcs gfx_v9_0_ras_funcs = { + .ras_late_init = amdgpu_gfx_ras_late_init, + .ras_fini = amdgpu_gfx_ras_fini, .ras_error_inject = &gfx_v9_0_ras_error_inject, .query_ras_error_count = &gfx_v9_0_query_ras_error_count, .reset_ras_error_count = &gfx_v9_0_reset_ras_error_count, }; -static const struct amdgpu_gfx_funcs gfx_v9_4_gfx_funcs = { - .get_gpu_clock_counter = &gfx_v9_0_get_gpu_clock_counter, - .select_se_sh = &gfx_v9_0_select_se_sh, - .read_wave_data = &gfx_v9_0_read_wave_data, - .read_wave_sgprs = &gfx_v9_0_read_wave_sgprs, - .read_wave_vgprs = &gfx_v9_0_read_wave_vgprs, - .select_me_pipe_q = &gfx_v9_0_select_me_pipe_q, - .ras_error_inject = &gfx_v9_4_ras_error_inject, - .query_ras_error_count = &gfx_v9_4_query_ras_error_count, - .reset_ras_error_count = &gfx_v9_4_reset_ras_error_count, - .query_ras_error_status = &gfx_v9_4_query_ras_error_status, -}; - static int gfx_v9_0_gpu_early_init(struct amdgpu_device *adev) { u32 gb_addr_config; @@ -2114,6 +2148,7 @@ static int gfx_v9_0_gpu_early_init(struct amdgpu_device *adev) DRM_INFO("fix gfx.config for vega12\n"); break; case CHIP_VEGA20: + adev->gfx.ras_funcs = &gfx_v9_0_ras_funcs; adev->gfx.config.max_hw_contexts = 8; adev->gfx.config.sc_prim_fifo_size_frontend = 0x20; adev->gfx.config.sc_prim_fifo_size_backend = 0x100; @@ -2139,7 +2174,7 @@ static int gfx_v9_0_gpu_early_init(struct amdgpu_device *adev) gb_addr_config = RAVEN_GB_ADDR_CONFIG_GOLDEN; break; case CHIP_ARCTURUS: - adev->gfx.funcs = &gfx_v9_4_gfx_funcs; + adev->gfx.ras_funcs = &gfx_v9_4_ras_funcs; adev->gfx.config.max_hw_contexts = 8; adev->gfx.config.sc_prim_fifo_size_frontend = 0x20; adev->gfx.config.sc_prim_fifo_size_backend = 0x100; @@ -2159,6 +2194,21 @@ static int gfx_v9_0_gpu_early_init(struct amdgpu_device *adev) gb_addr_config &= ~0xf3e777ff; gb_addr_config |= 0x22010042; break; + case CHIP_ALDEBARAN: + adev->gfx.ras_funcs = &gfx_v9_4_2_ras_funcs; + adev->gfx.config.max_hw_contexts = 8; + adev->gfx.config.sc_prim_fifo_size_frontend = 0x20; + adev->gfx.config.sc_prim_fifo_size_backend = 0x100; + adev->gfx.config.sc_hiz_tile_fifo_size = 0x30; + adev->gfx.config.sc_earlyz_tile_fifo_size = 0x4C0; + gb_addr_config = RREG32_SOC15(GC, 0, mmGB_ADDR_CONFIG); + gb_addr_config &= ~0xf3e777ff; + gb_addr_config |= 0x22014042; + /* check vbios table if gpu info is not available */ + err = amdgpu_atomfirmware_get_gfx_info(adev); + if (err) + return err; + break; default: BUG(); break; @@ -2231,8 +2281,8 @@ static int gfx_v9_0_compute_ring_init(struct amdgpu_device *adev, int ring_id, hw_prio = amdgpu_gfx_is_high_priority_compute_queue(adev, ring) ? AMDGPU_GFX_PIPE_PRIO_HIGH : AMDGPU_GFX_PIPE_PRIO_NORMAL; /* type-2 packets are deprecated on MEC, use type-3 instead */ - return amdgpu_ring_init(adev, ring, 1024, - &adev->gfx.eop_irq, irq_type, hw_prio); + return amdgpu_ring_init(adev, ring, 1024, &adev->gfx.eop_irq, irq_type, + hw_prio, NULL); } static int gfx_v9_0_sw_init(void *handle) @@ -2249,6 +2299,7 @@ static int gfx_v9_0_sw_init(void *handle) case CHIP_RAVEN: case CHIP_ARCTURUS: case CHIP_RENOIR: + case CHIP_ALDEBARAN: adev->gfx.mec.num_mec = 2; break; default: @@ -2320,10 +2371,9 @@ static int gfx_v9_0_sw_init(void *handle) sprintf(ring->name, "gfx_%d", i); ring->use_doorbell = true; ring->doorbell_index = adev->doorbell_index.gfx_ring0 << 1; - r = amdgpu_ring_init(adev, ring, 1024, - &adev->gfx.eop_irq, + r = amdgpu_ring_init(adev, ring, 1024, &adev->gfx.eop_irq, AMDGPU_CP_IRQ_GFX_ME0_PIPE0_EOP, - AMDGPU_RING_PRIO_DEFAULT); + AMDGPU_RING_PRIO_DEFAULT, NULL); if (r) return r; } @@ -2378,7 +2428,9 @@ static int gfx_v9_0_sw_fini(void *handle) int i; struct amdgpu_device *adev = (struct amdgpu_device *)handle; - amdgpu_gfx_ras_fini(adev); + if (adev->gfx.ras_funcs && + adev->gfx.ras_funcs->ras_fini) + adev->gfx.ras_funcs->ras_fini(adev); for (i = 0; i < adev->gfx.num_gfx_rings; i++) amdgpu_ring_fini(&adev->gfx.gfx_ring[i]); @@ -2634,17 +2686,15 @@ static void gfx_v9_0_enable_gui_idle_interrupt(struct amdgpu_device *adev, { u32 tmp; - /* don't toggle interrupts that are only applicable - * to me0 pipe0 on AISCs that have me0 removed */ - if (!adev->gfx.num_gfx_rings) - return; + /* These interrupts should be enabled to drive DS clock */ tmp= RREG32_SOC15(GC, 0, mmCP_INT_CNTL_RING0); tmp = REG_SET_FIELD(tmp, CP_INT_CNTL_RING0, CNTX_BUSY_INT_ENABLE, enable ? 1 : 0); tmp = REG_SET_FIELD(tmp, CP_INT_CNTL_RING0, CNTX_EMPTY_INT_ENABLE, enable ? 1 : 0); tmp = REG_SET_FIELD(tmp, CP_INT_CNTL_RING0, CMP_BUSY_INT_ENABLE, enable ? 1 : 0); - tmp = REG_SET_FIELD(tmp, CP_INT_CNTL_RING0, GFX_IDLE_INT_ENABLE, enable ? 1 : 0); + if(adev->gfx.num_gfx_rings) + tmp = REG_SET_FIELD(tmp, CP_INT_CNTL_RING0, GFX_IDLE_INT_ENABLE, enable ? 1 : 0); WREG32_SOC15(GC, 0, mmCP_INT_CNTL_RING0, tmp); } @@ -3696,11 +3746,18 @@ static int gfx_v9_0_kiq_init_queue(struct amdgpu_ring *ring) struct amdgpu_device *adev = ring->adev; struct v9_mqd *mqd = ring->mqd_ptr; int mqd_idx = AMDGPU_MAX_COMPUTE_RINGS; + struct v9_mqd *tmp_mqd; gfx_v9_0_kiq_setting(ring); - if (amdgpu_in_reset(adev)) { /* for GPU_RESET case */ - /* reset MQD to a clean status */ + /* GPU could be in bad state during probe, driver trigger the reset + * after load the SMU, in this case , the mqd is not be initialized. + * driver need to re-init the mqd. + * check mqd->cp_hqd_pq_control since this value should not be 0 + */ + tmp_mqd = (struct v9_mqd *)adev->gfx.mec.mqd_backup[mqd_idx]; + if (amdgpu_in_reset(adev) && tmp_mqd->cp_hqd_pq_control){ + /* for GPU_RESET case , reset MQD to a clean status */ if (adev->gfx.mec.mqd_backup[mqd_idx]) memcpy(mqd, adev->gfx.mec.mqd_backup[mqd_idx], sizeof(struct v9_mqd_allocation)); @@ -3736,8 +3793,15 @@ static int gfx_v9_0_kcq_init_queue(struct amdgpu_ring *ring) struct amdgpu_device *adev = ring->adev; struct v9_mqd *mqd = ring->mqd_ptr; int mqd_idx = ring - &adev->gfx.compute_ring[0]; + struct v9_mqd *tmp_mqd; - if (!amdgpu_in_reset(adev) && !adev->in_suspend) { + /* Same as above kiq init, driver need to re-init the mqd if mqd->cp_hqd_pq_control + * is not be initialized before + */ + tmp_mqd = (struct v9_mqd *)adev->gfx.mec.mqd_backup[mqd_idx]; + + if (!tmp_mqd->cp_hqd_pq_control || + (!amdgpu_in_reset(adev) && !adev->in_suspend)) { memset((void *)mqd, 0, sizeof(struct v9_mqd_allocation)); ((struct v9_mqd_allocation *)mqd)->dynamic_cu_mask = 0xFFFFFFFF; ((struct v9_mqd_allocation *)mqd)->dynamic_rb_mask = 0xFFFFFFFF; @@ -3913,6 +3977,9 @@ static int gfx_v9_0_hw_init(void *handle) if (r) return r; + if (adev->asic_type == CHIP_ALDEBARAN) + gfx_v9_4_2_set_power_brake_sequence(adev); + return r; } @@ -3954,8 +4021,14 @@ static int gfx_v9_0_hw_fini(void *handle) } gfx_v9_0_cp_enable(adev, false); - adev->gfx.rlc.funcs->stop(adev); + /* Skip suspend with A+A reset */ + if (adev->gmc.xgmi.connected_to_cpu && amdgpu_in_reset(adev)) { + dev_dbg(adev->dev, "Device in reset. Skipping RLC halt\n"); + return 0; + } + + adev->gfx.rlc.funcs->stop(adev); return 0; } @@ -4101,7 +4174,7 @@ static uint64_t gfx_v9_0_kiq_read_clock(struct amdgpu_device *adev) * * also don't wait anymore for IRQ context * */ - if (r < 1 && (amdgpu_in_reset(adev) || in_interrupt())) + if (r < 1 && (amdgpu_in_reset(adev))) goto failed_kiq_read; might_sleep(); @@ -4486,7 +4559,8 @@ static int gfx_v9_0_do_edc_gpr_workarounds(struct amdgpu_device *adev) if (!ring->sched.ready) return 0; - if (adev->asic_type == CHIP_ARCTURUS) { + if (adev->asic_type == CHIP_ARCTURUS || + adev->asic_type == CHIP_ALDEBARAN) { vgpr_init_shader_ptr = vgpr_init_compute_shader_arcturus; vgpr_init_shader_size = sizeof(vgpr_init_compute_shader_arcturus); vgpr_init_regs_ptr = vgpr_init_regs_arcturus; @@ -4636,7 +4710,8 @@ static int gfx_v9_0_early_init(void *handle) { struct amdgpu_device *adev = (struct amdgpu_device *)handle; - if (adev->asic_type == CHIP_ARCTURUS) + if (adev->asic_type == CHIP_ARCTURUS || + adev->asic_type == CHIP_ALDEBARAN) adev->gfx.num_gfx_rings = 0; else adev->gfx.num_gfx_rings = GFX9_NUM_GFX_RINGS; @@ -4662,7 +4737,8 @@ static int gfx_v9_0_ecc_late_init(void *handle) * to GDS in suspend/resume sequence on several cards. So just * limit this operation in cold boot sequence. */ - if (!adev->in_suspend) { + if ((!adev->in_suspend) && + (adev->gds.gds_size)) { r = gfx_v9_0_do_edc_gds_workarounds(adev); if (r) return r; @@ -4673,13 +4749,16 @@ static int gfx_v9_0_ecc_late_init(void *handle) if (r) return r; - if (adev->gfx.funcs && - adev->gfx.funcs->reset_ras_error_count) - adev->gfx.funcs->reset_ras_error_count(adev); + if (adev->gfx.ras_funcs && + adev->gfx.ras_funcs->ras_late_init) { + r = adev->gfx.ras_funcs->ras_late_init(adev); + if (r) + return r; + } - r = amdgpu_gfx_ras_late_init(adev); - if (r) - return r; + if (adev->gfx.ras_funcs && + adev->gfx.ras_funcs->enable_watchdog_timer) + adev->gfx.ras_funcs->enable_watchdog_timer(adev); return 0; } @@ -4858,7 +4937,7 @@ static void gfx_v9_0_update_3d_clock_gating(struct amdgpu_device *adev, { uint32_t data, def; - if (adev->asic_type == CHIP_ARCTURUS) + if (!adev->gfx.num_gfx_rings) return; amdgpu_gfx_rlc_enter_safe_mode(adev); @@ -5105,6 +5184,7 @@ static int gfx_v9_0_set_clockgating_state(void *handle, case CHIP_RAVEN: case CHIP_ARCTURUS: case CHIP_RENOIR: + case CHIP_ALDEBARAN: gfx_v9_0_update_gfx_clock_gating(adev, state == AMD_CG_STATE_GATE); break; @@ -6924,6 +7004,7 @@ static void gfx_v9_0_set_rlc_funcs(struct amdgpu_device *adev) case CHIP_RAVEN: case CHIP_ARCTURUS: case CHIP_RENOIR: + case CHIP_ALDEBARAN: adev->gfx.rlc.funcs = &gfx_v9_0_rlc_funcs; break; default: @@ -6944,6 +7025,12 @@ static void gfx_v9_0_set_gds_init(struct amdgpu_device *adev) case CHIP_ARCTURUS: adev->gds.gds_size = 0x1000; break; + case CHIP_ALDEBARAN: + /* aldebaran removed all the GDS internal memory, + * only support GWS opcode in kernel, like barrier + * semaphore.etc */ + adev->gds.gds_size = 0; + break; default: adev->gds.gds_size = 0x10000; break; @@ -6966,6 +7053,10 @@ static void gfx_v9_0_set_gds_init(struct amdgpu_device *adev) case CHIP_ARCTURUS: adev->gds.gds_compute_max_wave_id = 0xfff; break; + case CHIP_ALDEBARAN: + /* deprecated for Aldebaran, no usage at all */ + adev->gds.gds_compute_max_wave_id = 0; + break; default: /* this really depends on the chip */ adev->gds.gds_compute_max_wave_id = 0x7ff; diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v9_4.c b/drivers/gpu/drm/amd/amdgpu/gfx_v9_4.c index bc699d680ce8..b4789dfc2bb9 100644 --- a/drivers/gpu/drm/amd/amdgpu/gfx_v9_4.c +++ b/drivers/gpu/drm/amd/amdgpu/gfx_v9_4.c @@ -863,8 +863,8 @@ static int gfx_v9_4_ras_error_count(struct amdgpu_device *adev, return 0; } -int gfx_v9_4_query_ras_error_count(struct amdgpu_device *adev, - void *ras_error_status) +static int gfx_v9_4_query_ras_error_count(struct amdgpu_device *adev, + void *ras_error_status) { struct ras_err_data *err_data = (struct ras_err_data *)ras_error_status; uint32_t sec_count = 0, ded_count = 0; @@ -906,7 +906,7 @@ int gfx_v9_4_query_ras_error_count(struct amdgpu_device *adev, return 0; } -void gfx_v9_4_reset_ras_error_count(struct amdgpu_device *adev) +static void gfx_v9_4_reset_ras_error_count(struct amdgpu_device *adev) { int i, j, k; @@ -971,7 +971,8 @@ void gfx_v9_4_reset_ras_error_count(struct amdgpu_device *adev) WREG32_SOC15(GC, 0, mmATC_L2_CACHE_4K_DSM_INDEX, 255); } -int gfx_v9_4_ras_error_inject(struct amdgpu_device *adev, void *inject_if) +static int gfx_v9_4_ras_error_inject(struct amdgpu_device *adev, + void *inject_if) { struct ras_inject_if *info = (struct ras_inject_if *)inject_if; int ret; @@ -993,10 +994,10 @@ int gfx_v9_4_ras_error_inject(struct amdgpu_device *adev, void *inject_if) return ret; } -static const struct soc15_reg_entry gfx_v9_4_rdrsp_status_regs = +static const struct soc15_reg_entry gfx_v9_4_ea_err_status_regs = { SOC15_REG_ENTRY(GC, 0, mmGCEA_ERR_STATUS), 0, 1, 32 }; -void gfx_v9_4_query_ras_error_status(struct amdgpu_device *adev) +static void gfx_v9_4_query_ras_error_status(struct amdgpu_device *adev) { uint32_t i, j; uint32_t reg_value; @@ -1006,18 +1007,33 @@ void gfx_v9_4_query_ras_error_status(struct amdgpu_device *adev) mutex_lock(&adev->grbm_idx_mutex); - for (i = 0; i < gfx_v9_4_rdrsp_status_regs.se_num; i++) { - for (j = 0; j < gfx_v9_4_rdrsp_status_regs.instance; + for (i = 0; i < gfx_v9_4_ea_err_status_regs.se_num; i++) { + for (j = 0; j < gfx_v9_4_ea_err_status_regs.instance; j++) { gfx_v9_4_select_se_sh(adev, i, 0, j); reg_value = RREG32(SOC15_REG_ENTRY_OFFSET( - gfx_v9_4_rdrsp_status_regs)); - if (reg_value) + gfx_v9_4_ea_err_status_regs)); + if (REG_GET_FIELD(reg_value, GCEA_ERR_STATUS, SDP_RDRSP_STATUS) || + REG_GET_FIELD(reg_value, GCEA_ERR_STATUS, SDP_WRRSP_STATUS) || + REG_GET_FIELD(reg_value, GCEA_ERR_STATUS, SDP_RDRSP_DATAPARITY_ERROR)) { + /* SDP read/write error/parity error in FUE_IS_FATAL mode + * can cause system fatal error in arcturas. Harvest the error + * status before GPU reset */ dev_warn(adev->dev, "GCEA err detected at instance: %d, status: 0x%x!\n", j, reg_value); + } } } gfx_v9_4_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff); mutex_unlock(&adev->grbm_idx_mutex); } + +const struct amdgpu_gfx_ras_funcs gfx_v9_4_ras_funcs = { + .ras_late_init = amdgpu_gfx_ras_late_init, + .ras_fini = amdgpu_gfx_ras_fini, + .ras_error_inject = &gfx_v9_4_ras_error_inject, + .query_ras_error_count = &gfx_v9_4_query_ras_error_count, + .reset_ras_error_count = &gfx_v9_4_reset_ras_error_count, + .query_ras_error_status = &gfx_v9_4_query_ras_error_status, +}; diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v9_4.h b/drivers/gpu/drm/amd/amdgpu/gfx_v9_4.h index 875f18473a98..bdd16b568021 100644 --- a/drivers/gpu/drm/amd/amdgpu/gfx_v9_4.h +++ b/drivers/gpu/drm/amd/amdgpu/gfx_v9_4.h @@ -24,16 +24,6 @@ #ifndef __GFX_V9_4_H__ #define __GFX_V9_4_H__ -void gfx_v9_4_clear_ras_edc_counter(struct amdgpu_device *adev); - -int gfx_v9_4_query_ras_error_count(struct amdgpu_device *adev, - void *ras_error_status); - -int gfx_v9_4_ras_error_inject(struct amdgpu_device *adev, - void *inject_if); - -void gfx_v9_4_reset_ras_error_count(struct amdgpu_device *adev); - -void gfx_v9_4_query_ras_error_status(struct amdgpu_device *adev); +extern const struct amdgpu_gfx_ras_funcs gfx_v9_4_ras_funcs; #endif /* __GFX_V9_4_H__ */ diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v9_4_2.c b/drivers/gpu/drm/amd/amdgpu/gfx_v9_4_2.c new file mode 100644 index 000000000000..a30c7c10cd9a --- /dev/null +++ b/drivers/gpu/drm/amd/amdgpu/gfx_v9_4_2.c @@ -0,0 +1,1301 @@ +/* + * Copyright 2020 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + */ +#include "amdgpu.h" +#include "soc15.h" + +#include "gc/gc_9_4_2_offset.h" +#include "gc/gc_9_4_2_sh_mask.h" +#include "gfx_v9_0.h" + +#include "gfx_v9_4_2.h" +#include "amdgpu_ras.h" +#include "amdgpu_gfx.h" + +enum gfx_v9_4_2_utc_type { + VML2_MEM, + VML2_WALKER_MEM, + UTCL2_MEM, + ATC_L2_CACHE_2M, + ATC_L2_CACHE_32K, + ATC_L2_CACHE_4K +}; + +struct gfx_v9_4_2_utc_block { + enum gfx_v9_4_2_utc_type type; + uint32_t num_banks; + uint32_t num_ways; + uint32_t num_mem_blocks; + struct soc15_reg idx_reg; + struct soc15_reg data_reg; + uint32_t sec_count_mask; + uint32_t sec_count_shift; + uint32_t ded_count_mask; + uint32_t ded_count_shift; + uint32_t clear; +}; + +static const struct soc15_reg_golden golden_settings_gc_9_4_2_alde_die_0[] = { + SOC15_REG_GOLDEN_VALUE(GC, 0, regTCP_CHAN_STEER_0, 0x3fffffff, 0x141dc920), + SOC15_REG_GOLDEN_VALUE(GC, 0, regTCP_CHAN_STEER_1, 0x3fffffff, 0x3b458b93), + SOC15_REG_GOLDEN_VALUE(GC, 0, regTCP_CHAN_STEER_2, 0x3fffffff, 0x1a4f5583), + SOC15_REG_GOLDEN_VALUE(GC, 0, regTCP_CHAN_STEER_3, 0x3fffffff, 0x317717f6), + SOC15_REG_GOLDEN_VALUE(GC, 0, regTCP_CHAN_STEER_4, 0x3fffffff, 0x107cc1e6), + SOC15_REG_GOLDEN_VALUE(GC, 0, regTCP_CHAN_STEER_5, 0x3ff, 0x351), +}; + +static const struct soc15_reg_golden golden_settings_gc_9_4_2_alde_die_1[] = { + SOC15_REG_GOLDEN_VALUE(GC, 0, regTCP_CHAN_STEER_0, 0x3fffffff, 0x2591aa38), + SOC15_REG_GOLDEN_VALUE(GC, 0, regTCP_CHAN_STEER_1, 0x3fffffff, 0xac9e88b), + SOC15_REG_GOLDEN_VALUE(GC, 0, regTCP_CHAN_STEER_2, 0x3fffffff, 0x2bc3369b), + SOC15_REG_GOLDEN_VALUE(GC, 0, regTCP_CHAN_STEER_3, 0x3fffffff, 0xfb74ee), + SOC15_REG_GOLDEN_VALUE(GC, 0, regTCP_CHAN_STEER_4, 0x3fffffff, 0x21f0a2fe), + SOC15_REG_GOLDEN_VALUE(GC, 0, regTCP_CHAN_STEER_5, 0x3ff, 0x49), +}; + +static const struct soc15_reg_golden golden_settings_gc_9_4_2_alde[] = { + SOC15_REG_GOLDEN_VALUE(GC, 0, regGB_ADDR_CONFIG, 0xffff77ff, 0x2a114042), + SOC15_REG_GOLDEN_VALUE(GC, 0, regTA_CNTL_AUX, 0xfffffeef, 0x10b0000), + SOC15_REG_GOLDEN_VALUE(GC, 0, regTCP_UTCL1_CNTL1, 0xffffffff, 0x30800400), + SOC15_REG_GOLDEN_VALUE(GC, 0, regTCI_CNTL_3, 0xff, 0x20), +}; + +static void gfx_v9_4_2_query_sq_timeout_status(struct amdgpu_device *adev); +static void gfx_v9_4_2_reset_sq_timeout_status(struct amdgpu_device *adev); + +void gfx_v9_4_2_init_golden_registers(struct amdgpu_device *adev, + uint32_t die_id) +{ + soc15_program_register_sequence(adev, + golden_settings_gc_9_4_2_alde, + ARRAY_SIZE(golden_settings_gc_9_4_2_alde)); + + /* apply golden settings per die */ + switch (die_id) { + case 0: + soc15_program_register_sequence(adev, + golden_settings_gc_9_4_2_alde_die_0, + ARRAY_SIZE(golden_settings_gc_9_4_2_alde_die_0)); + break; + case 1: + soc15_program_register_sequence(adev, + golden_settings_gc_9_4_2_alde_die_1, + ARRAY_SIZE(golden_settings_gc_9_4_2_alde_die_1)); + break; + default: + dev_warn(adev->dev, + "invalid die id %d, ignore channel fabricid remap settings\n", + die_id); + break; + } + + return; +} + +void gfx_v9_4_2_debug_trap_config_init(struct amdgpu_device *adev, + uint32_t first_vmid, + uint32_t last_vmid) +{ + uint32_t data; + int i; + + mutex_lock(&adev->srbm_mutex); + + for (i = first_vmid; i < last_vmid; i++) { + data = 0; + soc15_grbm_select(adev, 0, 0, 0, i); + data = REG_SET_FIELD(data, SPI_GDBG_PER_VMID_CNTL, TRAP_EN, 1); + data = REG_SET_FIELD(data, SPI_GDBG_PER_VMID_CNTL, EXCP_EN, 0); + data = REG_SET_FIELD(data, SPI_GDBG_PER_VMID_CNTL, EXCP_REPLACE, + 0); + WREG32(SOC15_REG_OFFSET(GC, 0, regSPI_GDBG_PER_VMID_CNTL), data); + } + + soc15_grbm_select(adev, 0, 0, 0, 0); + mutex_unlock(&adev->srbm_mutex); +} + +void gfx_v9_4_2_set_power_brake_sequence(struct amdgpu_device *adev) +{ + u32 tmp; + + gfx_v9_0_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff); + + tmp = 0; + tmp = REG_SET_FIELD(tmp, GC_THROTTLE_CTRL, PATTERN_MODE, 1); + WREG32_SOC15(GC, 0, regGC_THROTTLE_CTRL, tmp); + + tmp = 0; + tmp = REG_SET_FIELD(tmp, GC_THROTTLE_CTRL1, PWRBRK_STALL_EN, 1); + WREG32_SOC15(GC, 0, regGC_THROTTLE_CTRL1, tmp); + + WREG32_SOC15(GC, 0, regDIDT_IND_INDEX, ixDIDT_SQ_THROTTLE_CTRL); + tmp = 0; + tmp = REG_SET_FIELD(tmp, DIDT_SQ_THROTTLE_CTRL, PWRBRK_STALL_EN, 1); + WREG32_SOC15(GC, 0, regDIDT_IND_DATA, tmp); + + WREG32_SOC15(GC, 0, regGC_CAC_IND_INDEX, ixPWRBRK_STALL_PATTERN_CTRL); + tmp = 0; + tmp = REG_SET_FIELD(tmp, PWRBRK_STALL_PATTERN_CTRL, PWRBRK_END_STEP, 0x12); + WREG32_SOC15(GC, 0, regGC_CAC_IND_DATA, tmp); +} + +static const struct soc15_reg_entry gfx_v9_4_2_edc_counter_regs[] = { + /* CPF */ + { SOC15_REG_ENTRY(GC, 0, regCPF_EDC_ROQ_CNT), 0, 1, 1 }, + { SOC15_REG_ENTRY(GC, 0, regCPF_EDC_TAG_CNT), 0, 1, 1 }, + /* CPC */ + { SOC15_REG_ENTRY(GC, 0, regCPC_EDC_SCRATCH_CNT), 0, 1, 1 }, + { SOC15_REG_ENTRY(GC, 0, regCPC_EDC_UCODE_CNT), 0, 1, 1 }, + { SOC15_REG_ENTRY(GC, 0, regDC_EDC_STATE_CNT), 0, 1, 1 }, + { SOC15_REG_ENTRY(GC, 0, regDC_EDC_CSINVOC_CNT), 0, 1, 1 }, + { SOC15_REG_ENTRY(GC, 0, regDC_EDC_RESTORE_CNT), 0, 1, 1 }, + /* GDS */ + { SOC15_REG_ENTRY(GC, 0, regGDS_EDC_CNT), 0, 1, 1 }, + { SOC15_REG_ENTRY(GC, 0, regGDS_EDC_GRBM_CNT), 0, 1, 1 }, + { SOC15_REG_ENTRY(GC, 0, regGDS_EDC_OA_DED), 0, 1, 1 }, + { SOC15_REG_ENTRY(GC, 0, regGDS_EDC_OA_PHY_CNT), 0, 1, 1 }, + { SOC15_REG_ENTRY(GC, 0, regGDS_EDC_OA_PIPE_CNT), 0, 1, 1 }, + /* RLC */ + { SOC15_REG_ENTRY(GC, 0, regRLC_EDC_CNT), 0, 1, 1 }, + { SOC15_REG_ENTRY(GC, 0, regRLC_EDC_CNT2), 0, 1, 1 }, + /* SPI */ + { SOC15_REG_ENTRY(GC, 0, regSPI_EDC_CNT), 0, 8, 1 }, + /* SQC */ + { SOC15_REG_ENTRY(GC, 0, regSQC_EDC_CNT), 0, 8, 7 }, + { SOC15_REG_ENTRY(GC, 0, regSQC_EDC_CNT2), 0, 8, 7 }, + { SOC15_REG_ENTRY(GC, 0, regSQC_EDC_CNT3), 0, 8, 7 }, + { SOC15_REG_ENTRY(GC, 0, regSQC_EDC_PARITY_CNT3), 0, 8, 7 }, + /* SQ */ + { SOC15_REG_ENTRY(GC, 0, regSQ_EDC_CNT), 0, 8, 14 }, + /* TCP */ + { SOC15_REG_ENTRY(GC, 0, regTCP_EDC_CNT_NEW), 0, 8, 14 }, + /* TCI */ + { SOC15_REG_ENTRY(GC, 0, regTCI_EDC_CNT), 0, 1, 69 }, + /* TCC */ + { SOC15_REG_ENTRY(GC, 0, regTCC_EDC_CNT), 0, 1, 16 }, + { SOC15_REG_ENTRY(GC, 0, regTCC_EDC_CNT2), 0, 1, 16 }, + /* TCA */ + { SOC15_REG_ENTRY(GC, 0, regTCA_EDC_CNT), 0, 1, 2 }, + /* TCX */ + { SOC15_REG_ENTRY(GC, 0, regTCX_EDC_CNT), 0, 1, 2 }, + { SOC15_REG_ENTRY(GC, 0, regTCX_EDC_CNT2), 0, 1, 2 }, + /* TD */ + { SOC15_REG_ENTRY(GC, 0, regTD_EDC_CNT), 0, 8, 14 }, + /* TA */ + { SOC15_REG_ENTRY(GC, 0, regTA_EDC_CNT), 0, 8, 14 }, + /* GCEA */ + { SOC15_REG_ENTRY(GC, 0, regGCEA_EDC_CNT), 0, 1, 16 }, + { SOC15_REG_ENTRY(GC, 0, regGCEA_EDC_CNT2), 0, 1, 16 }, + { SOC15_REG_ENTRY(GC, 0, regGCEA_EDC_CNT3), 0, 1, 16 }, +}; + +static void gfx_v9_4_2_select_se_sh(struct amdgpu_device *adev, u32 se_num, + u32 sh_num, u32 instance) +{ + u32 data; + + if (instance == 0xffffffff) + data = REG_SET_FIELD(0, GRBM_GFX_INDEX, + INSTANCE_BROADCAST_WRITES, 1); + else + data = REG_SET_FIELD(0, GRBM_GFX_INDEX, INSTANCE_INDEX, + instance); + + if (se_num == 0xffffffff) + data = REG_SET_FIELD(data, GRBM_GFX_INDEX, SE_BROADCAST_WRITES, + 1); + else + data = REG_SET_FIELD(data, GRBM_GFX_INDEX, SE_INDEX, se_num); + + if (sh_num == 0xffffffff) + data = REG_SET_FIELD(data, GRBM_GFX_INDEX, SH_BROADCAST_WRITES, + 1); + else + data = REG_SET_FIELD(data, GRBM_GFX_INDEX, SH_INDEX, sh_num); + + WREG32_SOC15_RLC_SHADOW_EX(reg, GC, 0, regGRBM_GFX_INDEX, data); +} + +static const struct soc15_ras_field_entry gfx_v9_4_2_ras_fields[] = { + /* CPF */ + { "CPF_ROQ_ME2", SOC15_REG_ENTRY(GC, 0, regCPF_EDC_ROQ_CNT), + SOC15_REG_FIELD(CPF_EDC_ROQ_CNT, SEC_COUNT_ME2), + SOC15_REG_FIELD(CPF_EDC_ROQ_CNT, DED_COUNT_ME2) }, + { "CPF_ROQ_ME1", SOC15_REG_ENTRY(GC, 0, regCPF_EDC_ROQ_CNT), + SOC15_REG_FIELD(CPF_EDC_ROQ_CNT, SEC_COUNT_ME1), + SOC15_REG_FIELD(CPF_EDC_ROQ_CNT, DED_COUNT_ME1) }, + { "CPF_TCIU_TAG", SOC15_REG_ENTRY(GC, 0, regCPF_EDC_TAG_CNT), + SOC15_REG_FIELD(CPF_EDC_TAG_CNT, SEC_COUNT), + SOC15_REG_FIELD(CPF_EDC_TAG_CNT, DED_COUNT) }, + + /* CPC */ + { "CPC_SCRATCH", SOC15_REG_ENTRY(GC, 0, regCPC_EDC_SCRATCH_CNT), + SOC15_REG_FIELD(CPC_EDC_SCRATCH_CNT, SEC_COUNT), + SOC15_REG_FIELD(CPC_EDC_SCRATCH_CNT, DED_COUNT) }, + { "CPC_UCODE", SOC15_REG_ENTRY(GC, 0, regCPC_EDC_UCODE_CNT), + SOC15_REG_FIELD(CPC_EDC_UCODE_CNT, SEC_COUNT), + SOC15_REG_FIELD(CPC_EDC_UCODE_CNT, DED_COUNT) }, + { "CPC_DC_STATE_RAM_ME1", SOC15_REG_ENTRY(GC, 0, regDC_EDC_STATE_CNT), + SOC15_REG_FIELD(DC_EDC_STATE_CNT, SEC_COUNT_ME1), + SOC15_REG_FIELD(DC_EDC_STATE_CNT, DED_COUNT_ME1) }, + { "CPC_DC_CSINVOC_RAM_ME1", + SOC15_REG_ENTRY(GC, 0, regDC_EDC_CSINVOC_CNT), + SOC15_REG_FIELD(DC_EDC_CSINVOC_CNT, SEC_COUNT_ME1), + SOC15_REG_FIELD(DC_EDC_CSINVOC_CNT, DED_COUNT_ME1) }, + { "CPC_DC_RESTORE_RAM_ME1", + SOC15_REG_ENTRY(GC, 0, regDC_EDC_RESTORE_CNT), + SOC15_REG_FIELD(DC_EDC_RESTORE_CNT, SEC_COUNT_ME1), + SOC15_REG_FIELD(DC_EDC_RESTORE_CNT, DED_COUNT_ME1) }, + { "CPC_DC_CSINVOC_RAM1_ME1", + SOC15_REG_ENTRY(GC, 0, regDC_EDC_CSINVOC_CNT), + SOC15_REG_FIELD(DC_EDC_CSINVOC_CNT, SEC_COUNT1_ME1), + SOC15_REG_FIELD(DC_EDC_CSINVOC_CNT, DED_COUNT1_ME1) }, + { "CPC_DC_RESTORE_RAM1_ME1", + SOC15_REG_ENTRY(GC, 0, regDC_EDC_RESTORE_CNT), + SOC15_REG_FIELD(DC_EDC_RESTORE_CNT, SEC_COUNT1_ME1), + SOC15_REG_FIELD(DC_EDC_RESTORE_CNT, DED_COUNT1_ME1) }, + + /* GDS */ + { "GDS_GRBM", SOC15_REG_ENTRY(GC, 0, regGDS_EDC_GRBM_CNT), + SOC15_REG_FIELD(GDS_EDC_GRBM_CNT, SEC), + SOC15_REG_FIELD(GDS_EDC_GRBM_CNT, DED) }, + { "GDS_MEM", SOC15_REG_ENTRY(GC, 0, regGDS_EDC_CNT), + SOC15_REG_FIELD(GDS_EDC_CNT, GDS_MEM_SEC), + SOC15_REG_FIELD(GDS_EDC_CNT, GDS_MEM_DED) }, + { "GDS_PHY_CMD_RAM_MEM", SOC15_REG_ENTRY(GC, 0, regGDS_EDC_OA_PHY_CNT), + SOC15_REG_FIELD(GDS_EDC_OA_PHY_CNT, PHY_CMD_RAM_MEM_SEC), + SOC15_REG_FIELD(GDS_EDC_OA_PHY_CNT, PHY_CMD_RAM_MEM_DED) }, + { "GDS_PHY_DATA_RAM_MEM", SOC15_REG_ENTRY(GC, 0, regGDS_EDC_OA_PHY_CNT), + SOC15_REG_FIELD(GDS_EDC_OA_PHY_CNT, PHY_DATA_RAM_MEM_SEC), + SOC15_REG_FIELD(GDS_EDC_OA_PHY_CNT, PHY_DATA_RAM_MEM_DED) }, + { "GDS_ME0_CS_PIPE_MEM", SOC15_REG_ENTRY(GC, 0, regGDS_EDC_OA_PHY_CNT), + SOC15_REG_FIELD(GDS_EDC_OA_PHY_CNT, ME0_CS_PIPE_MEM_SEC), + SOC15_REG_FIELD(GDS_EDC_OA_PHY_CNT, ME0_CS_PIPE_MEM_DED) }, + { "GDS_ME1_PIPE0_PIPE_MEM", + SOC15_REG_ENTRY(GC, 0, regGDS_EDC_OA_PIPE_CNT), + SOC15_REG_FIELD(GDS_EDC_OA_PIPE_CNT, ME1_PIPE0_PIPE_MEM_SEC), + SOC15_REG_FIELD(GDS_EDC_OA_PIPE_CNT, ME1_PIPE0_PIPE_MEM_DED) }, + { "GDS_ME1_PIPE1_PIPE_MEM", + SOC15_REG_ENTRY(GC, 0, regGDS_EDC_OA_PIPE_CNT), + SOC15_REG_FIELD(GDS_EDC_OA_PIPE_CNT, ME1_PIPE1_PIPE_MEM_SEC), + SOC15_REG_FIELD(GDS_EDC_OA_PIPE_CNT, ME1_PIPE1_PIPE_MEM_DED) }, + { "GDS_ME1_PIPE2_PIPE_MEM", + SOC15_REG_ENTRY(GC, 0, regGDS_EDC_OA_PIPE_CNT), + SOC15_REG_FIELD(GDS_EDC_OA_PIPE_CNT, ME1_PIPE2_PIPE_MEM_SEC), + SOC15_REG_FIELD(GDS_EDC_OA_PIPE_CNT, ME1_PIPE2_PIPE_MEM_DED) }, + { "GDS_ME1_PIPE3_PIPE_MEM", + SOC15_REG_ENTRY(GC, 0, regGDS_EDC_OA_PIPE_CNT), + SOC15_REG_FIELD(GDS_EDC_OA_PIPE_CNT, ME1_PIPE3_PIPE_MEM_SEC), + SOC15_REG_FIELD(GDS_EDC_OA_PIPE_CNT, ME1_PIPE3_PIPE_MEM_DED) }, + { "GDS_ME0_GFXHP3D_PIX_DED", + SOC15_REG_ENTRY(GC, 0, regGDS_EDC_OA_DED), 0, 0, + SOC15_REG_FIELD(GDS_EDC_OA_DED, ME0_GFXHP3D_PIX_DED) }, + { "GDS_ME0_GFXHP3D_VTX_DED", + SOC15_REG_ENTRY(GC, 0, regGDS_EDC_OA_DED), 0, 0, + SOC15_REG_FIELD(GDS_EDC_OA_DED, ME0_GFXHP3D_VTX_DED) }, + { "GDS_ME0_CS_DED", + SOC15_REG_ENTRY(GC, 0, regGDS_EDC_OA_DED), 0, 0, + SOC15_REG_FIELD(GDS_EDC_OA_DED, ME0_CS_DED) }, + { "GDS_ME0_GFXHP3D_GS_DED", + SOC15_REG_ENTRY(GC, 0, regGDS_EDC_OA_DED), 0, 0, + SOC15_REG_FIELD(GDS_EDC_OA_DED, ME0_GFXHP3D_GS_DED) }, + { "GDS_ME1_PIPE0_DED", + SOC15_REG_ENTRY(GC, 0, regGDS_EDC_OA_DED), 0, 0, + SOC15_REG_FIELD(GDS_EDC_OA_DED, ME1_PIPE0_DED) }, + { "GDS_ME1_PIPE1_DED", + SOC15_REG_ENTRY(GC, 0, regGDS_EDC_OA_DED), 0, 0, + SOC15_REG_FIELD(GDS_EDC_OA_DED, ME1_PIPE1_DED) }, + { "GDS_ME1_PIPE2_DED", + SOC15_REG_ENTRY(GC, 0, regGDS_EDC_OA_DED), 0, 0, + SOC15_REG_FIELD(GDS_EDC_OA_DED, ME1_PIPE2_DED) }, + { "GDS_ME1_PIPE3_DED", + SOC15_REG_ENTRY(GC, 0, regGDS_EDC_OA_DED), 0, 0, + SOC15_REG_FIELD(GDS_EDC_OA_DED, ME1_PIPE3_DED) }, + { "GDS_ME2_PIPE0_DED", + SOC15_REG_ENTRY(GC, 0, regGDS_EDC_OA_DED), 0, 0, + SOC15_REG_FIELD(GDS_EDC_OA_DED, ME2_PIPE0_DED) }, + { "GDS_ME2_PIPE1_DED", + SOC15_REG_ENTRY(GC, 0, regGDS_EDC_OA_DED), 0, 0, + SOC15_REG_FIELD(GDS_EDC_OA_DED, ME2_PIPE1_DED) }, + { "GDS_ME2_PIPE2_DED", + SOC15_REG_ENTRY(GC, 0, regGDS_EDC_OA_DED), 0, 0, + SOC15_REG_FIELD(GDS_EDC_OA_DED, ME2_PIPE2_DED) }, + { "GDS_ME2_PIPE3_DED", + SOC15_REG_ENTRY(GC, 0, regGDS_EDC_OA_DED), 0, 0, + SOC15_REG_FIELD(GDS_EDC_OA_DED, ME2_PIPE3_DED) }, + + /* RLC */ + { "RLCG_INSTR_RAM", SOC15_REG_ENTRY(GC, 0, regRLC_EDC_CNT), + SOC15_REG_FIELD(RLC_EDC_CNT, RLCG_INSTR_RAM_SEC_COUNT), + SOC15_REG_FIELD(RLC_EDC_CNT, RLCG_INSTR_RAM_DED_COUNT) }, + { "RLCG_SCRATCH_RAM", SOC15_REG_ENTRY(GC, 0, regRLC_EDC_CNT), + SOC15_REG_FIELD(RLC_EDC_CNT, RLCG_SCRATCH_RAM_SEC_COUNT), + SOC15_REG_FIELD(RLC_EDC_CNT, RLCG_SCRATCH_RAM_DED_COUNT) }, + { "RLCV_INSTR_RAM", SOC15_REG_ENTRY(GC, 0, regRLC_EDC_CNT), + SOC15_REG_FIELD(RLC_EDC_CNT, RLCV_INSTR_RAM_SEC_COUNT), + SOC15_REG_FIELD(RLC_EDC_CNT, RLCV_INSTR_RAM_DED_COUNT) }, + { "RLCV_SCRATCH_RAM", SOC15_REG_ENTRY(GC, 0, regRLC_EDC_CNT), + SOC15_REG_FIELD(RLC_EDC_CNT, RLCV_SCRATCH_RAM_SEC_COUNT), + SOC15_REG_FIELD(RLC_EDC_CNT, RLCV_SCRATCH_RAM_DED_COUNT) }, + { "RLC_TCTAG_RAM", SOC15_REG_ENTRY(GC, 0, regRLC_EDC_CNT), + SOC15_REG_FIELD(RLC_EDC_CNT, RLC_TCTAG_RAM_SEC_COUNT), + SOC15_REG_FIELD(RLC_EDC_CNT, RLC_TCTAG_RAM_DED_COUNT) }, + { "RLC_SPM_SCRATCH_RAM", SOC15_REG_ENTRY(GC, 0, regRLC_EDC_CNT), + SOC15_REG_FIELD(RLC_EDC_CNT, RLC_SPM_SCRATCH_RAM_SEC_COUNT), + SOC15_REG_FIELD(RLC_EDC_CNT, RLC_SPM_SCRATCH_RAM_DED_COUNT) }, + { "RLC_SRM_DATA_RAM", SOC15_REG_ENTRY(GC, 0, regRLC_EDC_CNT), + SOC15_REG_FIELD(RLC_EDC_CNT, RLC_SRM_DATA_RAM_SEC_COUNT), + SOC15_REG_FIELD(RLC_EDC_CNT, RLC_SRM_DATA_RAM_DED_COUNT) }, + { "RLC_SRM_ADDR_RAM", SOC15_REG_ENTRY(GC, 0, regRLC_EDC_CNT), + SOC15_REG_FIELD(RLC_EDC_CNT, RLC_SRM_ADDR_RAM_SEC_COUNT), + SOC15_REG_FIELD(RLC_EDC_CNT, RLC_SRM_ADDR_RAM_DED_COUNT) }, + { "RLC_SPM_SE0_SCRATCH_RAM", SOC15_REG_ENTRY(GC, 0, regRLC_EDC_CNT2), + SOC15_REG_FIELD(RLC_EDC_CNT2, RLC_SPM_SE0_SCRATCH_RAM_SEC_COUNT), + SOC15_REG_FIELD(RLC_EDC_CNT2, RLC_SPM_SE0_SCRATCH_RAM_DED_COUNT) }, + { "RLC_SPM_SE1_SCRATCH_RAM", SOC15_REG_ENTRY(GC, 0, regRLC_EDC_CNT2), + SOC15_REG_FIELD(RLC_EDC_CNT2, RLC_SPM_SE1_SCRATCH_RAM_SEC_COUNT), + SOC15_REG_FIELD(RLC_EDC_CNT2, RLC_SPM_SE1_SCRATCH_RAM_DED_COUNT) }, + { "RLC_SPM_SE2_SCRATCH_RAM", SOC15_REG_ENTRY(GC, 0, regRLC_EDC_CNT2), + SOC15_REG_FIELD(RLC_EDC_CNT2, RLC_SPM_SE2_SCRATCH_RAM_SEC_COUNT), + SOC15_REG_FIELD(RLC_EDC_CNT2, RLC_SPM_SE2_SCRATCH_RAM_DED_COUNT) }, + { "RLC_SPM_SE3_SCRATCH_RAM", SOC15_REG_ENTRY(GC, 0, regRLC_EDC_CNT2), + SOC15_REG_FIELD(RLC_EDC_CNT2, RLC_SPM_SE3_SCRATCH_RAM_SEC_COUNT), + SOC15_REG_FIELD(RLC_EDC_CNT2, RLC_SPM_SE3_SCRATCH_RAM_DED_COUNT) }, + { "RLC_SPM_SE4_SCRATCH_RAM", SOC15_REG_ENTRY(GC, 0, regRLC_EDC_CNT2), + SOC15_REG_FIELD(RLC_EDC_CNT2, RLC_SPM_SE4_SCRATCH_RAM_SEC_COUNT), + SOC15_REG_FIELD(RLC_EDC_CNT2, RLC_SPM_SE4_SCRATCH_RAM_DED_COUNT) }, + { "RLC_SPM_SE5_SCRATCH_RAM", SOC15_REG_ENTRY(GC, 0, regRLC_EDC_CNT2), + SOC15_REG_FIELD(RLC_EDC_CNT2, RLC_SPM_SE5_SCRATCH_RAM_SEC_COUNT), + SOC15_REG_FIELD(RLC_EDC_CNT2, RLC_SPM_SE5_SCRATCH_RAM_DED_COUNT) }, + { "RLC_SPM_SE6_SCRATCH_RAM", SOC15_REG_ENTRY(GC, 0, regRLC_EDC_CNT2), + SOC15_REG_FIELD(RLC_EDC_CNT2, RLC_SPM_SE6_SCRATCH_RAM_SEC_COUNT), + SOC15_REG_FIELD(RLC_EDC_CNT2, RLC_SPM_SE6_SCRATCH_RAM_DED_COUNT) }, + { "RLC_SPM_SE7_SCRATCH_RAM", SOC15_REG_ENTRY(GC, 0, regRLC_EDC_CNT2), + SOC15_REG_FIELD(RLC_EDC_CNT2, RLC_SPM_SE7_SCRATCH_RAM_SEC_COUNT), + SOC15_REG_FIELD(RLC_EDC_CNT2, RLC_SPM_SE7_SCRATCH_RAM_DED_COUNT) }, + + /* SPI */ + { "SPI_SR_MEM", SOC15_REG_ENTRY(GC, 0, regSPI_EDC_CNT), + SOC15_REG_FIELD(SPI_EDC_CNT, SPI_SR_MEM_SEC_COUNT), + SOC15_REG_FIELD(SPI_EDC_CNT, SPI_SR_MEM_DED_COUNT) }, + { "SPI_GDS_EXPREQ", SOC15_REG_ENTRY(GC, 0, regSPI_EDC_CNT), + SOC15_REG_FIELD(SPI_EDC_CNT, SPI_GDS_EXPREQ_SEC_COUNT), + SOC15_REG_FIELD(SPI_EDC_CNT, SPI_GDS_EXPREQ_DED_COUNT) }, + { "SPI_WB_GRANT_30", SOC15_REG_ENTRY(GC, 0, regSPI_EDC_CNT), + SOC15_REG_FIELD(SPI_EDC_CNT, SPI_WB_GRANT_30_SEC_COUNT), + SOC15_REG_FIELD(SPI_EDC_CNT, SPI_WB_GRANT_30_DED_COUNT) }, + { "SPI_LIFE_CNT", SOC15_REG_ENTRY(GC, 0, regSPI_EDC_CNT), + SOC15_REG_FIELD(SPI_EDC_CNT, SPI_LIFE_CNT_SEC_COUNT), + SOC15_REG_FIELD(SPI_EDC_CNT, SPI_LIFE_CNT_DED_COUNT) }, + + /* SQC - regSQC_EDC_CNT */ + { "SQC_DATA_CU0_WRITE_DATA_BUF", SOC15_REG_ENTRY(GC, 0, regSQC_EDC_CNT), + SOC15_REG_FIELD(SQC_EDC_CNT, DATA_CU0_WRITE_DATA_BUF_SEC_COUNT), + SOC15_REG_FIELD(SQC_EDC_CNT, DATA_CU0_WRITE_DATA_BUF_DED_COUNT) }, + { "SQC_DATA_CU0_UTCL1_LFIFO", SOC15_REG_ENTRY(GC, 0, regSQC_EDC_CNT), + SOC15_REG_FIELD(SQC_EDC_CNT, DATA_CU0_UTCL1_LFIFO_SEC_COUNT), + SOC15_REG_FIELD(SQC_EDC_CNT, DATA_CU0_UTCL1_LFIFO_DED_COUNT) }, + { "SQC_DATA_CU1_WRITE_DATA_BUF", SOC15_REG_ENTRY(GC, 0, regSQC_EDC_CNT), + SOC15_REG_FIELD(SQC_EDC_CNT, DATA_CU1_WRITE_DATA_BUF_SEC_COUNT), + SOC15_REG_FIELD(SQC_EDC_CNT, DATA_CU1_WRITE_DATA_BUF_DED_COUNT) }, + { "SQC_DATA_CU1_UTCL1_LFIFO", SOC15_REG_ENTRY(GC, 0, regSQC_EDC_CNT), + SOC15_REG_FIELD(SQC_EDC_CNT, DATA_CU1_UTCL1_LFIFO_SEC_COUNT), + SOC15_REG_FIELD(SQC_EDC_CNT, DATA_CU1_UTCL1_LFIFO_DED_COUNT) }, + { "SQC_DATA_CU2_WRITE_DATA_BUF", SOC15_REG_ENTRY(GC, 0, regSQC_EDC_CNT), + SOC15_REG_FIELD(SQC_EDC_CNT, DATA_CU2_WRITE_DATA_BUF_SEC_COUNT), + SOC15_REG_FIELD(SQC_EDC_CNT, DATA_CU2_WRITE_DATA_BUF_DED_COUNT) }, + { "SQC_DATA_CU2_UTCL1_LFIFO", SOC15_REG_ENTRY(GC, 0, regSQC_EDC_CNT), + SOC15_REG_FIELD(SQC_EDC_CNT, DATA_CU2_UTCL1_LFIFO_SEC_COUNT), + SOC15_REG_FIELD(SQC_EDC_CNT, DATA_CU2_UTCL1_LFIFO_DED_COUNT) }, + { "SQC_DATA_CU3_WRITE_DATA_BUF", SOC15_REG_ENTRY(GC, 0, regSQC_EDC_CNT), + SOC15_REG_FIELD(SQC_EDC_CNT, DATA_CU3_WRITE_DATA_BUF_SEC_COUNT), + SOC15_REG_FIELD(SQC_EDC_CNT, DATA_CU3_WRITE_DATA_BUF_DED_COUNT) }, + { "SQC_DATA_CU3_UTCL1_LFIFO", SOC15_REG_ENTRY(GC, 0, regSQC_EDC_CNT), + SOC15_REG_FIELD(SQC_EDC_CNT, DATA_CU3_UTCL1_LFIFO_SEC_COUNT), + SOC15_REG_FIELD(SQC_EDC_CNT, DATA_CU3_UTCL1_LFIFO_DED_COUNT) }, + + /* SQC - regSQC_EDC_CNT2 */ + { "SQC_INST_BANKA_TAG_RAM", SOC15_REG_ENTRY(GC, 0, regSQC_EDC_CNT2), + SOC15_REG_FIELD(SQC_EDC_CNT2, INST_BANKA_TAG_RAM_SEC_COUNT), + SOC15_REG_FIELD(SQC_EDC_CNT2, INST_BANKA_TAG_RAM_DED_COUNT) }, + { "SQC_INST_BANKA_BANK_RAM", SOC15_REG_ENTRY(GC, 0, regSQC_EDC_CNT2), + SOC15_REG_FIELD(SQC_EDC_CNT2, INST_BANKA_BANK_RAM_SEC_COUNT), + SOC15_REG_FIELD(SQC_EDC_CNT2, INST_BANKA_BANK_RAM_DED_COUNT) }, + { "SQC_DATA_BANKA_TAG_RAM", SOC15_REG_ENTRY(GC, 0, regSQC_EDC_CNT2), + SOC15_REG_FIELD(SQC_EDC_CNT2, DATA_BANKA_TAG_RAM_SEC_COUNT), + SOC15_REG_FIELD(SQC_EDC_CNT2, DATA_BANKA_TAG_RAM_DED_COUNT) }, + { "SQC_DATA_BANKA_BANK_RAM", SOC15_REG_ENTRY(GC, 0, regSQC_EDC_CNT2), + SOC15_REG_FIELD(SQC_EDC_CNT2, DATA_BANKA_BANK_RAM_SEC_COUNT), + SOC15_REG_FIELD(SQC_EDC_CNT2, DATA_BANKA_BANK_RAM_DED_COUNT) }, + { "SQC_INST_UTCL1_LFIFO", SOC15_REG_ENTRY(GC, 0, regSQC_EDC_CNT2), + SOC15_REG_FIELD(SQC_EDC_CNT2, INST_UTCL1_LFIFO_SEC_COUNT), + SOC15_REG_FIELD(SQC_EDC_CNT2, INST_UTCL1_LFIFO_DED_COUNT) }, + { "SQC_DATA_BANKA_DIRTY_BIT_RAM", SOC15_REG_ENTRY(GC, 0, regSQC_EDC_CNT2), + SOC15_REG_FIELD(SQC_EDC_CNT2, DATA_BANKA_DIRTY_BIT_RAM_SEC_COUNT), + SOC15_REG_FIELD(SQC_EDC_CNT2, DATA_BANKA_DIRTY_BIT_RAM_DED_COUNT) }, + + /* SQC - regSQC_EDC_CNT3 */ + { "SQC_INST_BANKB_TAG_RAM", SOC15_REG_ENTRY(GC, 0, regSQC_EDC_CNT3), + SOC15_REG_FIELD(SQC_EDC_CNT3, INST_BANKB_TAG_RAM_SEC_COUNT), + SOC15_REG_FIELD(SQC_EDC_CNT3, INST_BANKB_TAG_RAM_DED_COUNT) }, + { "SQC_INST_BANKB_BANK_RAM", SOC15_REG_ENTRY(GC, 0, regSQC_EDC_CNT3), + SOC15_REG_FIELD(SQC_EDC_CNT3, INST_BANKB_BANK_RAM_SEC_COUNT), + SOC15_REG_FIELD(SQC_EDC_CNT3, INST_BANKB_BANK_RAM_DED_COUNT) }, + { "SQC_DATA_BANKB_TAG_RAM", SOC15_REG_ENTRY(GC, 0, regSQC_EDC_CNT3), + SOC15_REG_FIELD(SQC_EDC_CNT3, DATA_BANKB_TAG_RAM_SEC_COUNT), + SOC15_REG_FIELD(SQC_EDC_CNT3, DATA_BANKB_TAG_RAM_DED_COUNT) }, + { "SQC_DATA_BANKB_BANK_RAM", SOC15_REG_ENTRY(GC, 0, regSQC_EDC_CNT3), + SOC15_REG_FIELD(SQC_EDC_CNT3, DATA_BANKB_BANK_RAM_SEC_COUNT), + SOC15_REG_FIELD(SQC_EDC_CNT3, DATA_BANKB_BANK_RAM_DED_COUNT) }, + { "SQC_DATA_BANKB_DIRTY_BIT_RAM", SOC15_REG_ENTRY(GC, 0, regSQC_EDC_CNT3), + SOC15_REG_FIELD(SQC_EDC_CNT3, DATA_BANKB_DIRTY_BIT_RAM_SEC_COUNT), + SOC15_REG_FIELD(SQC_EDC_CNT3, DATA_BANKB_DIRTY_BIT_RAM_DED_COUNT) }, + + /* SQC - regSQC_EDC_PARITY_CNT3 */ + { "SQC_INST_BANKA_UTCL1_MISS_FIFO", SOC15_REG_ENTRY(GC, 0, regSQC_EDC_PARITY_CNT3), + SOC15_REG_FIELD(SQC_EDC_PARITY_CNT3, INST_BANKA_UTCL1_MISS_FIFO_SEC_COUNT), + SOC15_REG_FIELD(SQC_EDC_PARITY_CNT3, INST_BANKA_UTCL1_MISS_FIFO_DED_COUNT) }, + { "SQC_INST_BANKA_MISS_FIFO", SOC15_REG_ENTRY(GC, 0, regSQC_EDC_PARITY_CNT3), + SOC15_REG_FIELD(SQC_EDC_PARITY_CNT3, INST_BANKA_MISS_FIFO_SEC_COUNT), + SOC15_REG_FIELD(SQC_EDC_PARITY_CNT3, INST_BANKA_MISS_FIFO_DED_COUNT) }, + { "SQC_DATA_BANKA_HIT_FIFO", SOC15_REG_ENTRY(GC, 0, regSQC_EDC_PARITY_CNT3), + SOC15_REG_FIELD(SQC_EDC_PARITY_CNT3, DATA_BANKA_HIT_FIFO_SEC_COUNT), + SOC15_REG_FIELD(SQC_EDC_PARITY_CNT3, DATA_BANKA_HIT_FIFO_DED_COUNT) }, + { "SQC_DATA_BANKA_MISS_FIFO", SOC15_REG_ENTRY(GC, 0, regSQC_EDC_PARITY_CNT3), + SOC15_REG_FIELD(SQC_EDC_PARITY_CNT3, DATA_BANKA_MISS_FIFO_SEC_COUNT), + SOC15_REG_FIELD(SQC_EDC_PARITY_CNT3, DATA_BANKA_MISS_FIFO_DED_COUNT) }, + { "SQC_INST_BANKB_UTCL1_MISS_FIFO", SOC15_REG_ENTRY(GC, 0, regSQC_EDC_PARITY_CNT3), + SOC15_REG_FIELD(SQC_EDC_PARITY_CNT3, INST_BANKB_UTCL1_MISS_FIFO_SEC_COUNT), + SOC15_REG_FIELD(SQC_EDC_PARITY_CNT3, INST_BANKB_UTCL1_MISS_FIFO_DED_COUNT) }, + { "SQC_INST_BANKB_MISS_FIFO", SOC15_REG_ENTRY(GC, 0, regSQC_EDC_PARITY_CNT3), + SOC15_REG_FIELD(SQC_EDC_PARITY_CNT3, INST_BANKB_MISS_FIFO_SEC_COUNT), + SOC15_REG_FIELD(SQC_EDC_PARITY_CNT3, INST_BANKB_MISS_FIFO_DED_COUNT) }, + { "SQC_DATA_BANKB_HIT_FIFO", SOC15_REG_ENTRY(GC, 0, regSQC_EDC_PARITY_CNT3), + SOC15_REG_FIELD(SQC_EDC_PARITY_CNT3, DATA_BANKB_HIT_FIFO_SEC_COUNT), + SOC15_REG_FIELD(SQC_EDC_PARITY_CNT3, DATA_BANKB_HIT_FIFO_DED_COUNT) }, + { "SQC_DATA_BANKB_MISS_FIFO", SOC15_REG_ENTRY(GC, 0, regSQC_EDC_PARITY_CNT3), + SOC15_REG_FIELD(SQC_EDC_PARITY_CNT3, DATA_BANKB_MISS_FIFO_SEC_COUNT), + SOC15_REG_FIELD(SQC_EDC_PARITY_CNT3, DATA_BANKB_MISS_FIFO_DED_COUNT) }, + + /* SQ */ + { "SQ_LDS_D", SOC15_REG_ENTRY(GC, 0, regSQ_EDC_CNT), + SOC15_REG_FIELD(SQ_EDC_CNT, LDS_D_SEC_COUNT), + SOC15_REG_FIELD(SQ_EDC_CNT, LDS_D_DED_COUNT) }, + { "SQ_LDS_I", SOC15_REG_ENTRY(GC, 0, regSQ_EDC_CNT), + SOC15_REG_FIELD(SQ_EDC_CNT, LDS_I_SEC_COUNT), + SOC15_REG_FIELD(SQ_EDC_CNT, LDS_I_DED_COUNT) }, + { "SQ_SGPR", SOC15_REG_ENTRY(GC, 0, regSQ_EDC_CNT), + SOC15_REG_FIELD(SQ_EDC_CNT, SGPR_SEC_COUNT), + SOC15_REG_FIELD(SQ_EDC_CNT, SGPR_DED_COUNT) }, + { "SQ_VGPR0", SOC15_REG_ENTRY(GC, 0, regSQ_EDC_CNT), + SOC15_REG_FIELD(SQ_EDC_CNT, VGPR0_SEC_COUNT), + SOC15_REG_FIELD(SQ_EDC_CNT, VGPR0_DED_COUNT) }, + { "SQ_VGPR1", SOC15_REG_ENTRY(GC, 0, regSQ_EDC_CNT), + SOC15_REG_FIELD(SQ_EDC_CNT, VGPR1_SEC_COUNT), + SOC15_REG_FIELD(SQ_EDC_CNT, VGPR1_DED_COUNT) }, + { "SQ_VGPR2", SOC15_REG_ENTRY(GC, 0, regSQ_EDC_CNT), + SOC15_REG_FIELD(SQ_EDC_CNT, VGPR2_SEC_COUNT), + SOC15_REG_FIELD(SQ_EDC_CNT, VGPR2_DED_COUNT) }, + { "SQ_VGPR3", SOC15_REG_ENTRY(GC, 0, regSQ_EDC_CNT), + SOC15_REG_FIELD(SQ_EDC_CNT, VGPR3_SEC_COUNT), + SOC15_REG_FIELD(SQ_EDC_CNT, VGPR3_DED_COUNT) }, + + /* TCP */ + { "TCP_CACHE_RAM", SOC15_REG_ENTRY(GC, 0, regTCP_EDC_CNT_NEW), + SOC15_REG_FIELD(TCP_EDC_CNT_NEW, CACHE_RAM_SEC_COUNT), + SOC15_REG_FIELD(TCP_EDC_CNT_NEW, CACHE_RAM_DED_COUNT) }, + { "TCP_LFIFO_RAM", SOC15_REG_ENTRY(GC, 0, regTCP_EDC_CNT_NEW), + SOC15_REG_FIELD(TCP_EDC_CNT_NEW, LFIFO_RAM_SEC_COUNT), + SOC15_REG_FIELD(TCP_EDC_CNT_NEW, LFIFO_RAM_DED_COUNT) }, + { "TCP_CMD_FIFO", SOC15_REG_ENTRY(GC, 0, regTCP_EDC_CNT_NEW), + SOC15_REG_FIELD(TCP_EDC_CNT_NEW, CMD_FIFO_SEC_COUNT), + SOC15_REG_FIELD(TCP_EDC_CNT_NEW, CMD_FIFO_DED_COUNT) }, + { "TCP_VM_FIFO", SOC15_REG_ENTRY(GC, 0, regTCP_EDC_CNT_NEW), + SOC15_REG_FIELD(TCP_EDC_CNT_NEW, VM_FIFO_SEC_COUNT), + SOC15_REG_FIELD(TCP_EDC_CNT_NEW, VM_FIFO_DED_COUNT) }, + { "TCP_DB_RAM", SOC15_REG_ENTRY(GC, 0, regTCP_EDC_CNT_NEW), + SOC15_REG_FIELD(TCP_EDC_CNT_NEW, DB_RAM_SEC_COUNT), + SOC15_REG_FIELD(TCP_EDC_CNT_NEW, DB_RAM_DED_COUNT) }, + { "TCP_UTCL1_LFIFO0", SOC15_REG_ENTRY(GC, 0, regTCP_EDC_CNT_NEW), + SOC15_REG_FIELD(TCP_EDC_CNT_NEW, UTCL1_LFIFO0_SEC_COUNT), + SOC15_REG_FIELD(TCP_EDC_CNT_NEW, UTCL1_LFIFO0_DED_COUNT) }, + { "TCP_UTCL1_LFIFO1", SOC15_REG_ENTRY(GC, 0, regTCP_EDC_CNT_NEW), + SOC15_REG_FIELD(TCP_EDC_CNT_NEW, UTCL1_LFIFO1_SEC_COUNT), + SOC15_REG_FIELD(TCP_EDC_CNT_NEW, UTCL1_LFIFO1_DED_COUNT) }, + + /* TCI */ + { "TCI_WRITE_RAM", SOC15_REG_ENTRY(GC, 0, regTCI_EDC_CNT), + SOC15_REG_FIELD(TCI_EDC_CNT, WRITE_RAM_SEC_COUNT), + SOC15_REG_FIELD(TCI_EDC_CNT, WRITE_RAM_DED_COUNT) }, + + /* TCC */ + { "TCC_CACHE_DATA", SOC15_REG_ENTRY(GC, 0, regTCC_EDC_CNT), + SOC15_REG_FIELD(TCC_EDC_CNT, CACHE_DATA_SEC_COUNT), + SOC15_REG_FIELD(TCC_EDC_CNT, CACHE_DATA_DED_COUNT) }, + { "TCC_CACHE_DIRTY", SOC15_REG_ENTRY(GC, 0, regTCC_EDC_CNT), + SOC15_REG_FIELD(TCC_EDC_CNT, CACHE_DIRTY_SEC_COUNT), + SOC15_REG_FIELD(TCC_EDC_CNT, CACHE_DIRTY_DED_COUNT) }, + { "TCC_HIGH_RATE_TAG", SOC15_REG_ENTRY(GC, 0, regTCC_EDC_CNT), + SOC15_REG_FIELD(TCC_EDC_CNT, HIGH_RATE_TAG_SEC_COUNT), + SOC15_REG_FIELD(TCC_EDC_CNT, HIGH_RATE_TAG_DED_COUNT) }, + { "TCC_LOW_RATE_TAG", SOC15_REG_ENTRY(GC, 0, regTCC_EDC_CNT), + SOC15_REG_FIELD(TCC_EDC_CNT, LOW_RATE_TAG_SEC_COUNT), + SOC15_REG_FIELD(TCC_EDC_CNT, LOW_RATE_TAG_DED_COUNT) }, + { "TCC_SRC_FIFO", SOC15_REG_ENTRY(GC, 0, regTCC_EDC_CNT), + SOC15_REG_FIELD(TCC_EDC_CNT, SRC_FIFO_SEC_COUNT), + SOC15_REG_FIELD(TCC_EDC_CNT, SRC_FIFO_DED_COUNT) }, + { "TCC_LATENCY_FIFO", SOC15_REG_ENTRY(GC, 0, regTCC_EDC_CNT), + SOC15_REG_FIELD(TCC_EDC_CNT, LATENCY_FIFO_SEC_COUNT), + SOC15_REG_FIELD(TCC_EDC_CNT, LATENCY_FIFO_DED_COUNT) }, + { "TCC_LATENCY_FIFO_NEXT_RAM", SOC15_REG_ENTRY(GC, 0, regTCC_EDC_CNT), + SOC15_REG_FIELD(TCC_EDC_CNT, LATENCY_FIFO_NEXT_RAM_SEC_COUNT), + SOC15_REG_FIELD(TCC_EDC_CNT, LATENCY_FIFO_NEXT_RAM_DED_COUNT) }, + { "TCC_CACHE_TAG_PROBE_FIFO", SOC15_REG_ENTRY(GC, 0, regTCC_EDC_CNT2), + SOC15_REG_FIELD(TCC_EDC_CNT2, CACHE_TAG_PROBE_FIFO_SEC_COUNT), + SOC15_REG_FIELD(TCC_EDC_CNT2, CACHE_TAG_PROBE_FIFO_DED_COUNT) }, + { "TCC_UC_ATOMIC_FIFO", SOC15_REG_ENTRY(GC, 0, regTCC_EDC_CNT2), + SOC15_REG_FIELD(TCC_EDC_CNT2, UC_ATOMIC_FIFO_SEC_COUNT), + SOC15_REG_FIELD(TCC_EDC_CNT2, UC_ATOMIC_FIFO_DED_COUNT) }, + { "TCC_WRITE_CACHE_READ", SOC15_REG_ENTRY(GC, 0, regTCC_EDC_CNT2), + SOC15_REG_FIELD(TCC_EDC_CNT2, WRITE_CACHE_READ_SEC_COUNT), + SOC15_REG_FIELD(TCC_EDC_CNT2, WRITE_CACHE_READ_DED_COUNT) }, + { "TCC_RETURN_CONTROL", SOC15_REG_ENTRY(GC, 0, regTCC_EDC_CNT2), + SOC15_REG_FIELD(TCC_EDC_CNT2, RETURN_CONTROL_SEC_COUNT), + SOC15_REG_FIELD(TCC_EDC_CNT2, RETURN_CONTROL_DED_COUNT) }, + { "TCC_IN_USE_TRANSFER", SOC15_REG_ENTRY(GC, 0, regTCC_EDC_CNT2), + SOC15_REG_FIELD(TCC_EDC_CNT2, IN_USE_TRANSFER_SEC_COUNT), + SOC15_REG_FIELD(TCC_EDC_CNT2, IN_USE_TRANSFER_DED_COUNT) }, + { "TCC_IN_USE_DEC", SOC15_REG_ENTRY(GC, 0, regTCC_EDC_CNT2), + SOC15_REG_FIELD(TCC_EDC_CNT2, IN_USE_DEC_SEC_COUNT), + SOC15_REG_FIELD(TCC_EDC_CNT2, IN_USE_DEC_DED_COUNT) }, + { "TCC_WRITE_RETURN", SOC15_REG_ENTRY(GC, 0, regTCC_EDC_CNT2), + SOC15_REG_FIELD(TCC_EDC_CNT2, WRITE_RETURN_SEC_COUNT), + SOC15_REG_FIELD(TCC_EDC_CNT2, WRITE_RETURN_DED_COUNT) }, + { "TCC_RETURN_DATA", SOC15_REG_ENTRY(GC, 0, regTCC_EDC_CNT2), + SOC15_REG_FIELD(TCC_EDC_CNT2, RETURN_DATA_SEC_COUNT), + SOC15_REG_FIELD(TCC_EDC_CNT2, RETURN_DATA_DED_COUNT) }, + + /* TCA */ + { "TCA_HOLE_FIFO", SOC15_REG_ENTRY(GC, 0, regTCA_EDC_CNT), + SOC15_REG_FIELD(TCA_EDC_CNT, HOLE_FIFO_SEC_COUNT), + SOC15_REG_FIELD(TCA_EDC_CNT, HOLE_FIFO_DED_COUNT) }, + { "TCA_REQ_FIFO", SOC15_REG_ENTRY(GC, 0, regTCA_EDC_CNT), + SOC15_REG_FIELD(TCA_EDC_CNT, REQ_FIFO_SEC_COUNT), + SOC15_REG_FIELD(TCA_EDC_CNT, REQ_FIFO_DED_COUNT) }, + + /* TCX */ + { "TCX_GROUP0", SOC15_REG_ENTRY(GC, 0, regTCX_EDC_CNT), + SOC15_REG_FIELD(TCX_EDC_CNT, GROUP0_SEC_COUNT), + SOC15_REG_FIELD(TCX_EDC_CNT, GROUP0_DED_COUNT) }, + { "TCX_GROUP1", SOC15_REG_ENTRY(GC, 0, regTCX_EDC_CNT), + SOC15_REG_FIELD(TCX_EDC_CNT, GROUP1_SEC_COUNT), + SOC15_REG_FIELD(TCX_EDC_CNT, GROUP1_DED_COUNT) }, + { "TCX_GROUP2", SOC15_REG_ENTRY(GC, 0, regTCX_EDC_CNT), + SOC15_REG_FIELD(TCX_EDC_CNT, GROUP2_SEC_COUNT), + SOC15_REG_FIELD(TCX_EDC_CNT, GROUP2_DED_COUNT) }, + { "TCX_GROUP3", SOC15_REG_ENTRY(GC, 0, regTCX_EDC_CNT), + SOC15_REG_FIELD(TCX_EDC_CNT, GROUP3_SEC_COUNT), + SOC15_REG_FIELD(TCX_EDC_CNT, GROUP3_DED_COUNT) }, + { "TCX_GROUP4", SOC15_REG_ENTRY(GC, 0, regTCX_EDC_CNT), + SOC15_REG_FIELD(TCX_EDC_CNT, GROUP4_SEC_COUNT), + SOC15_REG_FIELD(TCX_EDC_CNT, GROUP4_DED_COUNT) }, + { "TCX_GROUP5", SOC15_REG_ENTRY(GC, 0, regTCX_EDC_CNT), + SOC15_REG_FIELD(TCX_EDC_CNT, GROUP5_SED_COUNT), 0, 0 }, + { "TCX_GROUP6", SOC15_REG_ENTRY(GC, 0, regTCX_EDC_CNT), + SOC15_REG_FIELD(TCX_EDC_CNT, GROUP6_SED_COUNT), 0, 0 }, + { "TCX_GROUP7", SOC15_REG_ENTRY(GC, 0, regTCX_EDC_CNT), + SOC15_REG_FIELD(TCX_EDC_CNT, GROUP7_SED_COUNT), 0, 0 }, + { "TCX_GROUP8", SOC15_REG_ENTRY(GC, 0, regTCX_EDC_CNT), + SOC15_REG_FIELD(TCX_EDC_CNT, GROUP8_SED_COUNT), 0, 0 }, + { "TCX_GROUP9", SOC15_REG_ENTRY(GC, 0, regTCX_EDC_CNT), + SOC15_REG_FIELD(TCX_EDC_CNT, GROUP9_SED_COUNT), 0, 0 }, + { "TCX_GROUP10", SOC15_REG_ENTRY(GC, 0, regTCX_EDC_CNT), + SOC15_REG_FIELD(TCX_EDC_CNT, GROUP10_SED_COUNT), 0, 0 }, + { "TCX_GROUP11", SOC15_REG_ENTRY(GC, 0, regTCX_EDC_CNT2), + SOC15_REG_FIELD(TCX_EDC_CNT2, GROUP11_SED_COUNT), 0, 0 }, + { "TCX_GROUP12", SOC15_REG_ENTRY(GC, 0, regTCX_EDC_CNT2), + SOC15_REG_FIELD(TCX_EDC_CNT2, GROUP12_SED_COUNT), 0, 0 }, + { "TCX_GROUP13", SOC15_REG_ENTRY(GC, 0, regTCX_EDC_CNT2), + SOC15_REG_FIELD(TCX_EDC_CNT2, GROUP13_SED_COUNT), 0, 0 }, + { "TCX_GROUP14", SOC15_REG_ENTRY(GC, 0, regTCX_EDC_CNT2), + SOC15_REG_FIELD(TCX_EDC_CNT2, GROUP14_SED_COUNT), 0, 0 }, + + /* TD */ + { "TD_SS_FIFO_LO", SOC15_REG_ENTRY(GC, 0, regTD_EDC_CNT), + SOC15_REG_FIELD(TD_EDC_CNT, SS_FIFO_LO_SEC_COUNT), + SOC15_REG_FIELD(TD_EDC_CNT, SS_FIFO_LO_DED_COUNT) }, + { "TD_SS_FIFO_HI", SOC15_REG_ENTRY(GC, 0, regTD_EDC_CNT), + SOC15_REG_FIELD(TD_EDC_CNT, SS_FIFO_HI_SEC_COUNT), + SOC15_REG_FIELD(TD_EDC_CNT, SS_FIFO_HI_DED_COUNT) }, + { "TD_CS_FIFO", SOC15_REG_ENTRY(GC, 0, regTD_EDC_CNT), + SOC15_REG_FIELD(TD_EDC_CNT, CS_FIFO_SEC_COUNT), + SOC15_REG_FIELD(TD_EDC_CNT, CS_FIFO_DED_COUNT) }, + + /* TA */ + { "TA_FS_DFIFO", SOC15_REG_ENTRY(GC, 0, regTA_EDC_CNT), + SOC15_REG_FIELD(TA_EDC_CNT, TA_FS_DFIFO_SEC_COUNT), + SOC15_REG_FIELD(TA_EDC_CNT, TA_FS_DFIFO_DED_COUNT) }, + { "TA_FS_AFIFO_LO", SOC15_REG_ENTRY(GC, 0, regTA_EDC_CNT), + SOC15_REG_FIELD(TA_EDC_CNT, TA_FS_AFIFO_LO_SEC_COUNT), + SOC15_REG_FIELD(TA_EDC_CNT, TA_FS_AFIFO_LO_DED_COUNT) }, + { "TA_FL_LFIFO", SOC15_REG_ENTRY(GC, 0, regTA_EDC_CNT), + SOC15_REG_FIELD(TA_EDC_CNT, TA_FL_LFIFO_SEC_COUNT), + SOC15_REG_FIELD(TA_EDC_CNT, TA_FL_LFIFO_DED_COUNT) }, + { "TA_FX_LFIFO", SOC15_REG_ENTRY(GC, 0, regTA_EDC_CNT), + SOC15_REG_FIELD(TA_EDC_CNT, TA_FX_LFIFO_SEC_COUNT), + SOC15_REG_FIELD(TA_EDC_CNT, TA_FX_LFIFO_DED_COUNT) }, + { "TA_FS_CFIFO", SOC15_REG_ENTRY(GC, 0, regTA_EDC_CNT), + SOC15_REG_FIELD(TA_EDC_CNT, TA_FS_CFIFO_SEC_COUNT), + SOC15_REG_FIELD(TA_EDC_CNT, TA_FS_CFIFO_DED_COUNT) }, + { "TA_FS_AFIFO_HI", SOC15_REG_ENTRY(GC, 0, regTA_EDC_CNT), + SOC15_REG_FIELD(TA_EDC_CNT, TA_FS_AFIFO_HI_SEC_COUNT), + SOC15_REG_FIELD(TA_EDC_CNT, TA_FS_AFIFO_HI_DED_COUNT) }, + + /* EA - regGCEA_EDC_CNT */ + { "EA_DRAMRD_CMDMEM", SOC15_REG_ENTRY(GC, 0, regGCEA_EDC_CNT), + SOC15_REG_FIELD(GCEA_EDC_CNT, DRAMRD_CMDMEM_SEC_COUNT), + SOC15_REG_FIELD(GCEA_EDC_CNT, DRAMRD_CMDMEM_DED_COUNT) }, + { "EA_DRAMWR_CMDMEM", SOC15_REG_ENTRY(GC, 0, regGCEA_EDC_CNT), + SOC15_REG_FIELD(GCEA_EDC_CNT, DRAMWR_CMDMEM_SEC_COUNT), + SOC15_REG_FIELD(GCEA_EDC_CNT, DRAMWR_CMDMEM_DED_COUNT) }, + { "EA_DRAMWR_DATAMEM", SOC15_REG_ENTRY(GC, 0, regGCEA_EDC_CNT), + SOC15_REG_FIELD(GCEA_EDC_CNT, DRAMWR_DATAMEM_SEC_COUNT), + SOC15_REG_FIELD(GCEA_EDC_CNT, DRAMWR_DATAMEM_DED_COUNT) }, + { "EA_RRET_TAGMEM", SOC15_REG_ENTRY(GC, 0, regGCEA_EDC_CNT), + SOC15_REG_FIELD(GCEA_EDC_CNT, RRET_TAGMEM_SEC_COUNT), + SOC15_REG_FIELD(GCEA_EDC_CNT, RRET_TAGMEM_DED_COUNT) }, + { "EA_WRET_TAGMEM", SOC15_REG_ENTRY(GC, 0, regGCEA_EDC_CNT), + SOC15_REG_FIELD(GCEA_EDC_CNT, WRET_TAGMEM_SEC_COUNT), + SOC15_REG_FIELD(GCEA_EDC_CNT, WRET_TAGMEM_DED_COUNT) }, + { "EA_IOWR_DATAMEM", SOC15_REG_ENTRY(GC, 0, regGCEA_EDC_CNT), + SOC15_REG_FIELD(GCEA_EDC_CNT, IOWR_DATAMEM_SEC_COUNT), + SOC15_REG_FIELD(GCEA_EDC_CNT, IOWR_DATAMEM_DED_COUNT) }, + { "EA_DRAMRD_PAGEMEM", SOC15_REG_ENTRY(GC, 0, regGCEA_EDC_CNT), + SOC15_REG_FIELD(GCEA_EDC_CNT, DRAMRD_PAGEMEM_SED_COUNT), 0, 0 }, + { "EA_DRAMWR_PAGEMEM", SOC15_REG_ENTRY(GC, 0, regGCEA_EDC_CNT), + SOC15_REG_FIELD(GCEA_EDC_CNT, DRAMWR_PAGEMEM_SED_COUNT), 0, 0 }, + { "EA_IORD_CMDMEM", SOC15_REG_ENTRY(GC, 0, regGCEA_EDC_CNT), + SOC15_REG_FIELD(GCEA_EDC_CNT, IORD_CMDMEM_SED_COUNT), 0, 0 }, + { "EA_IOWR_CMDMEM", SOC15_REG_ENTRY(GC, 0, regGCEA_EDC_CNT), + SOC15_REG_FIELD(GCEA_EDC_CNT, IOWR_CMDMEM_SED_COUNT), 0, 0 }, + + /* EA - regGCEA_EDC_CNT2 */ + { "EA_GMIRD_CMDMEM", SOC15_REG_ENTRY(GC, 0, regGCEA_EDC_CNT2), + SOC15_REG_FIELD(GCEA_EDC_CNT2, GMIRD_CMDMEM_SEC_COUNT), + SOC15_REG_FIELD(GCEA_EDC_CNT2, GMIRD_CMDMEM_DED_COUNT) }, + { "EA_GMIWR_CMDMEM", SOC15_REG_ENTRY(GC, 0, regGCEA_EDC_CNT2), + SOC15_REG_FIELD(GCEA_EDC_CNT2, GMIWR_CMDMEM_SEC_COUNT), + SOC15_REG_FIELD(GCEA_EDC_CNT2, GMIWR_CMDMEM_DED_COUNT) }, + { "EA_GMIWR_DATAMEM", SOC15_REG_ENTRY(GC, 0, regGCEA_EDC_CNT2), + SOC15_REG_FIELD(GCEA_EDC_CNT2, GMIWR_DATAMEM_SEC_COUNT), + SOC15_REG_FIELD(GCEA_EDC_CNT2, GMIWR_DATAMEM_DED_COUNT) }, + { "EA_GMIRD_PAGEMEM", SOC15_REG_ENTRY(GC, 0, regGCEA_EDC_CNT2), + SOC15_REG_FIELD(GCEA_EDC_CNT2, GMIRD_PAGEMEM_SED_COUNT), 0, 0 }, + { "EA_GMIWR_PAGEMEM", SOC15_REG_ENTRY(GC, 0, regGCEA_EDC_CNT2), + SOC15_REG_FIELD(GCEA_EDC_CNT2, GMIWR_PAGEMEM_SED_COUNT), 0, 0 }, + { "EA_MAM_D0MEM", SOC15_REG_ENTRY(GC, 0, regGCEA_EDC_CNT2), + SOC15_REG_FIELD(GCEA_EDC_CNT2, MAM_D0MEM_SED_COUNT), + SOC15_REG_FIELD(GCEA_EDC_CNT2, MAM_D0MEM_DED_COUNT) }, + { "EA_MAM_D1MEM", SOC15_REG_ENTRY(GC, 0, regGCEA_EDC_CNT2), + SOC15_REG_FIELD(GCEA_EDC_CNT2, MAM_D1MEM_SED_COUNT), + SOC15_REG_FIELD(GCEA_EDC_CNT2, MAM_D1MEM_DED_COUNT) }, + { "EA_MAM_D2MEM", SOC15_REG_ENTRY(GC, 0, regGCEA_EDC_CNT2), + SOC15_REG_FIELD(GCEA_EDC_CNT2, MAM_D2MEM_SED_COUNT), + SOC15_REG_FIELD(GCEA_EDC_CNT2, MAM_D2MEM_DED_COUNT) }, + { "EA_MAM_D3MEM", SOC15_REG_ENTRY(GC, 0, regGCEA_EDC_CNT2), + SOC15_REG_FIELD(GCEA_EDC_CNT2, MAM_D3MEM_SED_COUNT), + SOC15_REG_FIELD(GCEA_EDC_CNT2, MAM_D3MEM_DED_COUNT) }, + + /* EA - regGCEA_EDC_CNT3 */ + { "EA_DRAMRD_PAGEMEM", SOC15_REG_ENTRY(GC, 0, regGCEA_EDC_CNT3), 0, 0, + SOC15_REG_FIELD(GCEA_EDC_CNT3, DRAMRD_PAGEMEM_DED_COUNT) }, + { "EA_DRAMWR_PAGEMEM", SOC15_REG_ENTRY(GC, 0, regGCEA_EDC_CNT3), 0, 0, + SOC15_REG_FIELD(GCEA_EDC_CNT3, DRAMWR_PAGEMEM_DED_COUNT) }, + { "EA_IORD_CMDMEM", SOC15_REG_ENTRY(GC, 0, regGCEA_EDC_CNT3), 0, 0, + SOC15_REG_FIELD(GCEA_EDC_CNT3, IORD_CMDMEM_DED_COUNT) }, + { "EA_IOWR_CMDMEM", SOC15_REG_ENTRY(GC, 0, regGCEA_EDC_CNT3), 0, 0, + SOC15_REG_FIELD(GCEA_EDC_CNT3, IOWR_CMDMEM_DED_COUNT) }, + { "EA_GMIRD_PAGEMEM", SOC15_REG_ENTRY(GC, 0, regGCEA_EDC_CNT3), 0, 0, + SOC15_REG_FIELD(GCEA_EDC_CNT3, GMIRD_PAGEMEM_DED_COUNT) }, + { "EA_GMIWR_PAGEMEM", SOC15_REG_ENTRY(GC, 0, regGCEA_EDC_CNT3), 0, 0, + SOC15_REG_FIELD(GCEA_EDC_CNT3, GMIWR_PAGEMEM_DED_COUNT) }, + { "EA_MAM_A0MEM", SOC15_REG_ENTRY(GC, 0, regGCEA_EDC_CNT3), + SOC15_REG_FIELD(GCEA_EDC_CNT3, MAM_A0MEM_SEC_COUNT), + SOC15_REG_FIELD(GCEA_EDC_CNT3, MAM_A0MEM_DED_COUNT) }, + { "EA_MAM_A1MEM", SOC15_REG_ENTRY(GC, 0, regGCEA_EDC_CNT3), + SOC15_REG_FIELD(GCEA_EDC_CNT3, MAM_A1MEM_SEC_COUNT), + SOC15_REG_FIELD(GCEA_EDC_CNT3, MAM_A1MEM_DED_COUNT) }, + { "EA_MAM_A2MEM", SOC15_REG_ENTRY(GC, 0, regGCEA_EDC_CNT3), + SOC15_REG_FIELD(GCEA_EDC_CNT3, MAM_A2MEM_SEC_COUNT), + SOC15_REG_FIELD(GCEA_EDC_CNT3, MAM_A2MEM_DED_COUNT) }, + { "EA_MAM_A3MEM", SOC15_REG_ENTRY(GC, 0, regGCEA_EDC_CNT3), + SOC15_REG_FIELD(GCEA_EDC_CNT3, MAM_A3MEM_SEC_COUNT), + SOC15_REG_FIELD(GCEA_EDC_CNT3, MAM_A3MEM_DED_COUNT) }, + { "EA_MAM_AFMEM", SOC15_REG_ENTRY(GC, 0, regGCEA_EDC_CNT3), + SOC15_REG_FIELD(GCEA_EDC_CNT3, MAM_AFMEM_SEC_COUNT), + SOC15_REG_FIELD(GCEA_EDC_CNT3, MAM_AFMEM_DED_COUNT) }, +}; + +static const char * const vml2_walker_mems[] = { + "UTC_VML2_CACHE_PDE0_MEM0", + "UTC_VML2_CACHE_PDE0_MEM1", + "UTC_VML2_CACHE_PDE1_MEM0", + "UTC_VML2_CACHE_PDE1_MEM1", + "UTC_VML2_CACHE_PDE2_MEM0", + "UTC_VML2_CACHE_PDE2_MEM1", + "UTC_VML2_RDIF_ARADDRS", + "UTC_VML2_RDIF_LOG_FIFO", + "UTC_VML2_QUEUE_REQ", + "UTC_VML2_QUEUE_RET", +}; + +static struct gfx_v9_4_2_utc_block gfx_v9_4_2_utc_blocks[] = { + { VML2_MEM, 8, 2, 2, + { SOC15_REG_ENTRY(GC, 0, regVML2_MEM_ECC_INDEX) }, + { SOC15_REG_ENTRY(GC, 0, regVML2_MEM_ECC_CNTL) }, + SOC15_REG_FIELD(VML2_MEM_ECC_CNTL, SEC_COUNT), + SOC15_REG_FIELD(VML2_MEM_ECC_CNTL, DED_COUNT), + REG_SET_FIELD(0, VML2_MEM_ECC_CNTL, WRITE_COUNTERS, 1) }, + { VML2_WALKER_MEM, ARRAY_SIZE(vml2_walker_mems), 1, 1, + { SOC15_REG_ENTRY(GC, 0, regVML2_WALKER_MEM_ECC_INDEX) }, + { SOC15_REG_ENTRY(GC, 0, regVML2_WALKER_MEM_ECC_CNTL) }, + SOC15_REG_FIELD(VML2_WALKER_MEM_ECC_CNTL, SEC_COUNT), + SOC15_REG_FIELD(VML2_WALKER_MEM_ECC_CNTL, DED_COUNT), + REG_SET_FIELD(0, VML2_WALKER_MEM_ECC_CNTL, WRITE_COUNTERS, 1) }, + { UTCL2_MEM, 18, 1, 2, + { SOC15_REG_ENTRY(GC, 0, regUTCL2_MEM_ECC_INDEX) }, + { SOC15_REG_ENTRY(GC, 0, regUTCL2_MEM_ECC_CNTL) }, + SOC15_REG_FIELD(UTCL2_MEM_ECC_CNTL, SEC_COUNT), + SOC15_REG_FIELD(UTCL2_MEM_ECC_CNTL, DED_COUNT), + REG_SET_FIELD(0, UTCL2_MEM_ECC_CNTL, WRITE_COUNTERS, 1) }, + { ATC_L2_CACHE_2M, 8, 2, 1, + { SOC15_REG_ENTRY(GC, 0, regATC_L2_CACHE_2M_DSM_INDEX) }, + { SOC15_REG_ENTRY(GC, 0, regATC_L2_CACHE_2M_DSM_CNTL) }, + SOC15_REG_FIELD(ATC_L2_CACHE_2M_DSM_CNTL, SEC_COUNT), + SOC15_REG_FIELD(ATC_L2_CACHE_2M_DSM_CNTL, DED_COUNT), + REG_SET_FIELD(0, ATC_L2_CACHE_2M_DSM_CNTL, WRITE_COUNTERS, 1) }, + { ATC_L2_CACHE_32K, 8, 2, 2, + { SOC15_REG_ENTRY(GC, 0, regATC_L2_CACHE_32K_DSM_INDEX) }, + { SOC15_REG_ENTRY(GC, 0, regATC_L2_CACHE_32K_DSM_CNTL) }, + SOC15_REG_FIELD(ATC_L2_CACHE_32K_DSM_CNTL, SEC_COUNT), + SOC15_REG_FIELD(ATC_L2_CACHE_32K_DSM_CNTL, DED_COUNT), + REG_SET_FIELD(0, ATC_L2_CACHE_32K_DSM_CNTL, WRITE_COUNTERS, 1) }, + { ATC_L2_CACHE_4K, 8, 2, 8, + { SOC15_REG_ENTRY(GC, 0, regATC_L2_CACHE_4K_DSM_INDEX) }, + { SOC15_REG_ENTRY(GC, 0, regATC_L2_CACHE_4K_DSM_CNTL) }, + SOC15_REG_FIELD(ATC_L2_CACHE_4K_DSM_CNTL, SEC_COUNT), + SOC15_REG_FIELD(ATC_L2_CACHE_4K_DSM_CNTL, DED_COUNT), + REG_SET_FIELD(0, ATC_L2_CACHE_4K_DSM_CNTL, WRITE_COUNTERS, 1) }, +}; + +static const struct soc15_reg_entry gfx_v9_4_2_ea_err_status_regs = + { SOC15_REG_ENTRY(GC, 0, regGCEA_ERR_STATUS), 0, 1, 16 }; + +static int gfx_v9_4_2_get_reg_error_count(struct amdgpu_device *adev, + const struct soc15_reg_entry *reg, + uint32_t se_id, uint32_t inst_id, + uint32_t value, uint32_t *sec_count, + uint32_t *ded_count) +{ + uint32_t i; + uint32_t sec_cnt, ded_cnt; + + for (i = 0; i < ARRAY_SIZE(gfx_v9_4_2_ras_fields); i++) { + if (gfx_v9_4_2_ras_fields[i].reg_offset != reg->reg_offset || + gfx_v9_4_2_ras_fields[i].seg != reg->seg || + gfx_v9_4_2_ras_fields[i].inst != reg->inst) + continue; + + sec_cnt = SOC15_RAS_REG_FIELD_VAL( + value, gfx_v9_4_2_ras_fields[i], sec); + if (sec_cnt) { + dev_info(adev->dev, + "GFX SubBlock %s, Instance[%d][%d], SEC %d\n", + gfx_v9_4_2_ras_fields[i].name, se_id, inst_id, + sec_cnt); + *sec_count += sec_cnt; + } + + ded_cnt = SOC15_RAS_REG_FIELD_VAL( + value, gfx_v9_4_2_ras_fields[i], ded); + if (ded_cnt) { + dev_info(adev->dev, + "GFX SubBlock %s, Instance[%d][%d], DED %d\n", + gfx_v9_4_2_ras_fields[i].name, se_id, inst_id, + ded_cnt); + *ded_count += ded_cnt; + } + } + + return 0; +} + +static int gfx_v9_4_2_query_sram_edc_count(struct amdgpu_device *adev, + uint32_t *sec_count, uint32_t *ded_count) +{ + uint32_t i, j, k, data; + uint32_t sec_cnt = 0, ded_cnt = 0; + + if (sec_count && ded_count) { + *sec_count = 0; + *ded_count = 0; + } + + mutex_lock(&adev->grbm_idx_mutex); + + for (i = 0; i < ARRAY_SIZE(gfx_v9_4_2_edc_counter_regs); i++) { + for (j = 0; j < gfx_v9_4_2_edc_counter_regs[i].se_num; j++) { + for (k = 0; k < gfx_v9_4_2_edc_counter_regs[i].instance; + k++) { + gfx_v9_4_2_select_se_sh(adev, j, 0, k); + + /* if sec/ded_count is null, just clear counter */ + if (!sec_count || !ded_count) { + WREG32(SOC15_REG_ENTRY_OFFSET( + gfx_v9_4_2_edc_counter_regs[i]), 0); + continue; + } + + data = RREG32(SOC15_REG_ENTRY_OFFSET( + gfx_v9_4_2_edc_counter_regs[i])); + + if (!data) + continue; + + gfx_v9_4_2_get_reg_error_count(adev, + &gfx_v9_4_2_edc_counter_regs[i], + j, k, data, &sec_cnt, &ded_cnt); + + /* clear counter after read */ + WREG32(SOC15_REG_ENTRY_OFFSET( + gfx_v9_4_2_edc_counter_regs[i]), 0); + } + } + } + + if (sec_count && ded_count) { + *sec_count += sec_cnt; + *ded_count += ded_cnt; + } + + gfx_v9_4_2_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff); + mutex_unlock(&adev->grbm_idx_mutex); + + return 0; +} + +static void gfx_v9_4_2_log_utc_edc_count(struct amdgpu_device *adev, + struct gfx_v9_4_2_utc_block *blk, + uint32_t instance, uint32_t sec_cnt, + uint32_t ded_cnt) +{ + uint32_t bank, way, mem; + static const char *vml2_way_str[] = { "BIGK", "4K" }; + static const char *utcl2_rounter_str[] = { "VMC", "APT" }; + + mem = instance % blk->num_mem_blocks; + way = (instance / blk->num_mem_blocks) % blk->num_ways; + bank = instance / (blk->num_mem_blocks * blk->num_ways); + + switch (blk->type) { + case VML2_MEM: + dev_info( + adev->dev, + "GFX SubBlock UTC_VML2_BANK_CACHE_%d_%s_MEM%d, SED %d, DED %d\n", + bank, vml2_way_str[way], mem, sec_cnt, ded_cnt); + break; + case VML2_WALKER_MEM: + dev_info(adev->dev, "GFX SubBlock %s, SED %d, DED %d\n", + vml2_walker_mems[bank], sec_cnt, ded_cnt); + break; + case UTCL2_MEM: + dev_info( + adev->dev, + "GFX SubBlock UTCL2_ROUTER_IFIF%d_GROUP0_%s, SED %d, DED %d\n", + bank, utcl2_rounter_str[mem], sec_cnt, ded_cnt); + break; + case ATC_L2_CACHE_2M: + dev_info( + adev->dev, + "GFX SubBlock UTC_ATCL2_CACHE_2M_BANK%d_WAY%d_MEM, SED %d, DED %d\n", + bank, way, sec_cnt, ded_cnt); + break; + case ATC_L2_CACHE_32K: + dev_info( + adev->dev, + "GFX SubBlock UTC_ATCL2_CACHE_32K_BANK%d_WAY%d_MEM%d, SED %d, DED %d\n", + bank, way, mem, sec_cnt, ded_cnt); + break; + case ATC_L2_CACHE_4K: + dev_info( + adev->dev, + "GFX SubBlock UTC_ATCL2_CACHE_4K_BANK%d_WAY%d_MEM%d, SED %d, DED %d\n", + bank, way, mem, sec_cnt, ded_cnt); + break; + } +} + +static int gfx_v9_4_2_query_utc_edc_count(struct amdgpu_device *adev, + uint32_t *sec_count, + uint32_t *ded_count) +{ + uint32_t i, j, data; + uint32_t sec_cnt, ded_cnt; + uint32_t num_instances; + struct gfx_v9_4_2_utc_block *blk; + + if (sec_count && ded_count) { + *sec_count = 0; + *ded_count = 0; + } + + for (i = 0; i < ARRAY_SIZE(gfx_v9_4_2_utc_blocks); i++) { + blk = &gfx_v9_4_2_utc_blocks[i]; + num_instances = + blk->num_banks * blk->num_ways * blk->num_mem_blocks; + for (j = 0; j < num_instances; j++) { + WREG32(SOC15_REG_ENTRY_OFFSET(blk->idx_reg), j); + + /* if sec/ded_count is NULL, just clear counter */ + if (!sec_count || !ded_count) { + WREG32(SOC15_REG_ENTRY_OFFSET(blk->data_reg), + blk->clear); + continue; + } + + data = RREG32(SOC15_REG_ENTRY_OFFSET(blk->data_reg)); + if (!data) + continue; + + sec_cnt = SOC15_RAS_REG_FIELD_VAL(data, *blk, sec); + *sec_count += sec_cnt; + ded_cnt = SOC15_RAS_REG_FIELD_VAL(data, *blk, ded); + *ded_count += ded_cnt; + + /* clear counter after read */ + WREG32(SOC15_REG_ENTRY_OFFSET(blk->data_reg), + blk->clear); + + /* print the edc count */ + if (sec_cnt || ded_cnt) + gfx_v9_4_2_log_utc_edc_count(adev, blk, j, sec_cnt, + ded_cnt); + } + } + + return 0; +} + +int gfx_v9_4_2_query_ras_error_count(struct amdgpu_device *adev, + void *ras_error_status) +{ + struct ras_err_data *err_data = (struct ras_err_data *)ras_error_status; + uint32_t sec_count = 0, ded_count = 0; + + if (!amdgpu_ras_is_supported(adev, AMDGPU_RAS_BLOCK__GFX)) + return -EINVAL; + + err_data->ue_count = 0; + err_data->ce_count = 0; + + gfx_v9_4_2_query_sram_edc_count(adev, &sec_count, &ded_count); + err_data->ce_count += sec_count; + err_data->ue_count += ded_count; + + gfx_v9_4_2_query_utc_edc_count(adev, &sec_count, &ded_count); + err_data->ce_count += sec_count; + err_data->ue_count += ded_count; + + return 0; +} + +static void gfx_v9_4_2_reset_utc_err_status(struct amdgpu_device *adev) +{ + WREG32_SOC15(GC, 0, regUTCL2_MEM_ECC_STATUS, 0x3); + WREG32_SOC15(GC, 0, regVML2_MEM_ECC_STATUS, 0x3); + WREG32_SOC15(GC, 0, regVML2_WALKER_MEM_ECC_STATUS, 0x3); +} + +static void gfx_v9_4_2_reset_ea_err_status(struct amdgpu_device *adev) +{ + uint32_t i, j; + + mutex_lock(&adev->grbm_idx_mutex); + for (i = 0; i < gfx_v9_4_2_ea_err_status_regs.se_num; i++) { + for (j = 0; j < gfx_v9_4_2_ea_err_status_regs.instance; + j++) { + gfx_v9_4_2_select_se_sh(adev, i, 0, j); + WREG32(SOC15_REG_ENTRY_OFFSET(gfx_v9_4_2_ea_err_status_regs), 0x10); + } + } + gfx_v9_4_2_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff); + mutex_unlock(&adev->grbm_idx_mutex); +} + +void gfx_v9_4_2_reset_ras_error_count(struct amdgpu_device *adev) +{ + if (!amdgpu_ras_is_supported(adev, AMDGPU_RAS_BLOCK__GFX)) + return; + + gfx_v9_4_2_query_sram_edc_count(adev, NULL, NULL); + gfx_v9_4_2_query_utc_edc_count(adev, NULL, NULL); +} + +int gfx_v9_4_2_ras_error_inject(struct amdgpu_device *adev, void *inject_if) +{ + struct ras_inject_if *info = (struct ras_inject_if *)inject_if; + int ret; + struct ta_ras_trigger_error_input block_info = { 0 }; + + if (!amdgpu_ras_is_supported(adev, AMDGPU_RAS_BLOCK__GFX)) + return -EINVAL; + + block_info.block_id = amdgpu_ras_block_to_ta(info->head.block); + block_info.sub_block_index = info->head.sub_block_index; + block_info.inject_error_type = amdgpu_ras_error_to_ta(info->head.type); + block_info.address = info->address; + block_info.value = info->value; + + mutex_lock(&adev->grbm_idx_mutex); + ret = psp_ras_trigger_error(&adev->psp, &block_info); + mutex_unlock(&adev->grbm_idx_mutex); + + return ret; +} + +static void gfx_v9_4_2_query_ea_err_status(struct amdgpu_device *adev) +{ + uint32_t i, j; + uint32_t reg_value; + + mutex_lock(&adev->grbm_idx_mutex); + + for (i = 0; i < gfx_v9_4_2_ea_err_status_regs.se_num; i++) { + for (j = 0; j < gfx_v9_4_2_ea_err_status_regs.instance; + j++) { + gfx_v9_4_2_select_se_sh(adev, i, 0, j); + reg_value = RREG32(SOC15_REG_ENTRY_OFFSET( + gfx_v9_4_2_ea_err_status_regs)); + if (REG_GET_FIELD(reg_value, GCEA_ERR_STATUS, SDP_RDRSP_STATUS) || + REG_GET_FIELD(reg_value, GCEA_ERR_STATUS, SDP_WRRSP_STATUS) || + REG_GET_FIELD(reg_value, GCEA_ERR_STATUS, SDP_RDRSP_DATAPARITY_ERROR)) { + dev_warn(adev->dev, "GCEA err detected at instance: %d, status: 0x%x!\n", + j, reg_value); + } + /* clear after read */ + WREG32(SOC15_REG_ENTRY_OFFSET(gfx_v9_4_2_ea_err_status_regs), 0x10); + } + } + + gfx_v9_4_2_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff); + mutex_unlock(&adev->grbm_idx_mutex); +} + +static void gfx_v9_4_2_query_utc_err_status(struct amdgpu_device *adev) +{ + uint32_t data; + + data = RREG32_SOC15(GC, 0, regUTCL2_MEM_ECC_STATUS); + if (data) { + dev_warn(adev->dev, "GFX UTCL2 Mem Ecc Status: 0x%x!\n", data); + WREG32_SOC15(GC, 0, regUTCL2_MEM_ECC_STATUS, 0x3); + } + + data = RREG32_SOC15(GC, 0, regVML2_MEM_ECC_STATUS); + if (data) { + dev_warn(adev->dev, "GFX VML2 Mem Ecc Status: 0x%x!\n", data); + WREG32_SOC15(GC, 0, regVML2_MEM_ECC_STATUS, 0x3); + } + + data = RREG32_SOC15(GC, 0, regVML2_WALKER_MEM_ECC_STATUS); + if (data) { + dev_warn(adev->dev, "GFX VML2 Walker Mem Ecc Status: 0x%x!\n", data); + WREG32_SOC15(GC, 0, regVML2_WALKER_MEM_ECC_STATUS, 0x3); + } +} + +void gfx_v9_4_2_query_ras_error_status(struct amdgpu_device *adev) +{ + if (!amdgpu_ras_is_supported(adev, AMDGPU_RAS_BLOCK__GFX)) + return; + + gfx_v9_4_2_query_ea_err_status(adev); + gfx_v9_4_2_query_utc_err_status(adev); + gfx_v9_4_2_query_sq_timeout_status(adev); +} + +void gfx_v9_4_2_reset_ras_error_status(struct amdgpu_device *adev) +{ + if (!amdgpu_ras_is_supported(adev, AMDGPU_RAS_BLOCK__GFX)) + return; + + gfx_v9_4_2_reset_utc_err_status(adev); + gfx_v9_4_2_reset_ea_err_status(adev); + gfx_v9_4_2_reset_sq_timeout_status(adev); +} + +void gfx_v9_4_2_enable_watchdog_timer(struct amdgpu_device *adev) +{ + uint32_t i; + uint32_t data; + + data = REG_SET_FIELD(0, SQ_TIMEOUT_CONFIG, TIMEOUT_FATAL_DISABLE, + amdgpu_watchdog_timer.timeout_fatal_disable ? 1 : + 0); + + if (amdgpu_watchdog_timer.timeout_fatal_disable && + (amdgpu_watchdog_timer.period < 1 || + amdgpu_watchdog_timer.period > 0x23)) { + dev_warn(adev->dev, "Watchdog period range is 1 to 0x23\n"); + amdgpu_watchdog_timer.period = 0x23; + } + data = REG_SET_FIELD(data, SQ_TIMEOUT_CONFIG, PERIOD_SEL, + amdgpu_watchdog_timer.period); + + mutex_lock(&adev->grbm_idx_mutex); + for (i = 0; i < adev->gfx.config.max_shader_engines; i++) { + gfx_v9_4_2_select_se_sh(adev, i, 0xffffffff, 0xffffffff); + WREG32_SOC15(GC, 0, regSQ_TIMEOUT_CONFIG, data); + } + gfx_v9_4_2_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff); + mutex_unlock(&adev->grbm_idx_mutex); +} + +static uint32_t wave_read_ind(struct amdgpu_device *adev, uint32_t simd, uint32_t wave, uint32_t address) +{ + WREG32_SOC15_RLC_EX(reg, GC, 0, regSQ_IND_INDEX, + (wave << SQ_IND_INDEX__WAVE_ID__SHIFT) | + (simd << SQ_IND_INDEX__SIMD_ID__SHIFT) | + (address << SQ_IND_INDEX__INDEX__SHIFT) | + (SQ_IND_INDEX__FORCE_READ_MASK)); + return RREG32_SOC15(GC, 0, regSQ_IND_DATA); +} + +static void gfx_v9_4_2_log_cu_timeout_status(struct amdgpu_device *adev, + uint32_t status) +{ + struct amdgpu_cu_info *cu_info = &adev->gfx.cu_info; + uint32_t i, simd, wave; + uint32_t wave_status; + uint32_t wave_pc_lo, wave_pc_hi; + uint32_t wave_exec_lo, wave_exec_hi; + uint32_t wave_inst_dw0, wave_inst_dw1; + uint32_t wave_ib_sts; + + for (i = 0; i < 32; i++) { + if (!((i << 1) & status)) + continue; + + simd = i / cu_info->max_waves_per_simd; + wave = i % cu_info->max_waves_per_simd; + + wave_status = wave_read_ind(adev, simd, wave, ixSQ_WAVE_STATUS); + wave_pc_lo = wave_read_ind(adev, simd, wave, ixSQ_WAVE_PC_LO); + wave_pc_hi = wave_read_ind(adev, simd, wave, ixSQ_WAVE_PC_HI); + wave_exec_lo = + wave_read_ind(adev, simd, wave, ixSQ_WAVE_EXEC_LO); + wave_exec_hi = + wave_read_ind(adev, simd, wave, ixSQ_WAVE_EXEC_HI); + wave_inst_dw0 = + wave_read_ind(adev, simd, wave, ixSQ_WAVE_INST_DW0); + wave_inst_dw1 = + wave_read_ind(adev, simd, wave, ixSQ_WAVE_INST_DW1); + wave_ib_sts = wave_read_ind(adev, simd, wave, ixSQ_WAVE_IB_STS); + + dev_info( + adev->dev, + "\t SIMD %d, Wave %d: status 0x%x, pc 0x%llx, exec 0x%llx, inst 0x%llx, ib_sts 0x%x\n", + simd, wave, wave_status, + ((uint64_t)wave_pc_hi << 32 | wave_pc_lo), + ((uint64_t)wave_exec_hi << 32 | wave_exec_lo), + ((uint64_t)wave_inst_dw1 << 32 | wave_inst_dw0), + wave_ib_sts); + } +} + +static void gfx_v9_4_2_query_sq_timeout_status(struct amdgpu_device *adev) +{ + uint32_t se_idx, sh_idx, cu_idx; + uint32_t status; + + mutex_lock(&adev->grbm_idx_mutex); + for (se_idx = 0; se_idx < adev->gfx.config.max_shader_engines; + se_idx++) { + for (sh_idx = 0; sh_idx < adev->gfx.config.max_sh_per_se; + sh_idx++) { + for (cu_idx = 0; + cu_idx < adev->gfx.config.max_cu_per_sh; + cu_idx++) { + gfx_v9_4_2_select_se_sh(adev, se_idx, sh_idx, + cu_idx); + status = RREG32_SOC15(GC, 0, + regSQ_TIMEOUT_STATUS); + if (status != 0) { + dev_info( + adev->dev, + "GFX Watchdog Timeout: SE %d, SH %d, CU %d\n", + se_idx, sh_idx, cu_idx); + gfx_v9_4_2_log_cu_timeout_status( + adev, status); + } + /* clear old status */ + WREG32_SOC15(GC, 0, regSQ_TIMEOUT_STATUS, 0); + } + } + } + gfx_v9_4_2_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff); + mutex_unlock(&adev->grbm_idx_mutex); +} + +static void gfx_v9_4_2_reset_sq_timeout_status(struct amdgpu_device *adev) +{ + uint32_t se_idx, sh_idx, cu_idx; + + mutex_lock(&adev->grbm_idx_mutex); + for (se_idx = 0; se_idx < adev->gfx.config.max_shader_engines; + se_idx++) { + for (sh_idx = 0; sh_idx < adev->gfx.config.max_sh_per_se; + sh_idx++) { + for (cu_idx = 0; + cu_idx < adev->gfx.config.max_cu_per_sh; + cu_idx++) { + gfx_v9_4_2_select_se_sh(adev, se_idx, sh_idx, + cu_idx); + WREG32_SOC15(GC, 0, regSQ_TIMEOUT_STATUS, 0); + } + } + } + gfx_v9_4_2_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff); + mutex_unlock(&adev->grbm_idx_mutex); +} + +const struct amdgpu_gfx_ras_funcs gfx_v9_4_2_ras_funcs = { + .ras_late_init = amdgpu_gfx_ras_late_init, + .ras_fini = amdgpu_gfx_ras_fini, + .ras_error_inject = &gfx_v9_4_2_ras_error_inject, + .query_ras_error_count = &gfx_v9_4_2_query_ras_error_count, + .reset_ras_error_count = &gfx_v9_4_2_reset_ras_error_count, + .query_ras_error_status = &gfx_v9_4_2_query_ras_error_status, + .reset_ras_error_status = &gfx_v9_4_2_reset_ras_error_status, + .enable_watchdog_timer = &gfx_v9_4_2_enable_watchdog_timer, +}; diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v9_4_2.h b/drivers/gpu/drm/amd/amdgpu/gfx_v9_4_2.h new file mode 100644 index 000000000000..81c5833b6b9f --- /dev/null +++ b/drivers/gpu/drm/amd/amdgpu/gfx_v9_4_2.h @@ -0,0 +1,35 @@ +/* + * Copyright 2020 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + */ + +#ifndef __GFX_V9_4_2_H__ +#define __GFX_V9_4_2_H__ + +void gfx_v9_4_2_debug_trap_config_init(struct amdgpu_device *adev, + uint32_t first_vmid, uint32_t last_vmid); +void gfx_v9_4_2_init_golden_registers(struct amdgpu_device *adev, + uint32_t die_id); +void gfx_v9_4_2_set_power_brake_sequence(struct amdgpu_device *adev); + +extern const struct amdgpu_gfx_ras_funcs gfx_v9_4_2_ras_funcs; + +#endif /* __GFX_V9_4_2_H__ */ diff --git a/drivers/gpu/drm/amd/amdgpu/gfxhub_v1_0.c b/drivers/gpu/drm/amd/amdgpu/gfxhub_v1_0.c index 6ddd53ba8b77..1e4678cb98f0 100644 --- a/drivers/gpu/drm/amd/amdgpu/gfxhub_v1_0.c +++ b/drivers/gpu/drm/amd/amdgpu/gfxhub_v1_0.c @@ -53,19 +53,39 @@ static void gfxhub_v1_0_setup_vm_pt_regs(struct amdgpu_device *adev, static void gfxhub_v1_0_init_gart_aperture_regs(struct amdgpu_device *adev) { - uint64_t pt_base = amdgpu_gmc_pd_addr(adev->gart.bo); + uint64_t pt_base; - gfxhub_v1_0_setup_vm_pt_regs(adev, 0, pt_base); + if (adev->gmc.pdb0_bo) + pt_base = amdgpu_gmc_pd_addr(adev->gmc.pdb0_bo); + else + pt_base = amdgpu_gmc_pd_addr(adev->gart.bo); - WREG32_SOC15(GC, 0, mmVM_CONTEXT0_PAGE_TABLE_START_ADDR_LO32, - (u32)(adev->gmc.gart_start >> 12)); - WREG32_SOC15(GC, 0, mmVM_CONTEXT0_PAGE_TABLE_START_ADDR_HI32, - (u32)(adev->gmc.gart_start >> 44)); + gfxhub_v1_0_setup_vm_pt_regs(adev, 0, pt_base); - WREG32_SOC15(GC, 0, mmVM_CONTEXT0_PAGE_TABLE_END_ADDR_LO32, - (u32)(adev->gmc.gart_end >> 12)); - WREG32_SOC15(GC, 0, mmVM_CONTEXT0_PAGE_TABLE_END_ADDR_HI32, - (u32)(adev->gmc.gart_end >> 44)); + /* If use GART for FB translation, vmid0 page table covers both + * vram and system memory (gart) + */ + if (adev->gmc.pdb0_bo) { + WREG32_SOC15(GC, 0, mmVM_CONTEXT0_PAGE_TABLE_START_ADDR_LO32, + (u32)(adev->gmc.fb_start >> 12)); + WREG32_SOC15(GC, 0, mmVM_CONTEXT0_PAGE_TABLE_START_ADDR_HI32, + (u32)(adev->gmc.fb_start >> 44)); + + WREG32_SOC15(GC, 0, mmVM_CONTEXT0_PAGE_TABLE_END_ADDR_LO32, + (u32)(adev->gmc.gart_end >> 12)); + WREG32_SOC15(GC, 0, mmVM_CONTEXT0_PAGE_TABLE_END_ADDR_HI32, + (u32)(adev->gmc.gart_end >> 44)); + } else { + WREG32_SOC15(GC, 0, mmVM_CONTEXT0_PAGE_TABLE_START_ADDR_LO32, + (u32)(adev->gmc.gart_start >> 12)); + WREG32_SOC15(GC, 0, mmVM_CONTEXT0_PAGE_TABLE_START_ADDR_HI32, + (u32)(adev->gmc.gart_start >> 44)); + + WREG32_SOC15(GC, 0, mmVM_CONTEXT0_PAGE_TABLE_END_ADDR_LO32, + (u32)(adev->gmc.gart_end >> 12)); + WREG32_SOC15(GC, 0, mmVM_CONTEXT0_PAGE_TABLE_END_ADDR_HI32, + (u32)(adev->gmc.gart_end >> 44)); + } } static void gfxhub_v1_0_init_system_aperture_regs(struct amdgpu_device *adev) @@ -100,8 +120,7 @@ static void gfxhub_v1_0_init_system_aperture_regs(struct amdgpu_device *adev) max(adev->gmc.fb_end, adev->gmc.agp_end) >> 18); /* Set default page address. */ - value = adev->vram_scratch.gpu_addr - adev->gmc.vram_start + - adev->vm_manager.vram_base_offset; + value = amdgpu_gmc_vram_mc2pa(adev, adev->vram_scratch.gpu_addr); WREG32_SOC15(GC, 0, mmMC_VM_SYSTEM_APERTURE_DEFAULT_ADDR_LSB, (u32)(value >> 12)); WREG32_SOC15(GC, 0, mmMC_VM_SYSTEM_APERTURE_DEFAULT_ADDR_MSB, @@ -116,6 +135,18 @@ static void gfxhub_v1_0_init_system_aperture_regs(struct amdgpu_device *adev) WREG32_FIELD15(GC, 0, VM_L2_PROTECTION_FAULT_CNTL2, ACTIVE_PAGE_MIGRATION_PTE_READ_RETRY, 1); } + + /* In the case squeezing vram into GART aperture, we don't use + * FB aperture and AGP aperture. Disable them. + */ + if (adev->gmc.pdb0_bo) { + WREG32_SOC15(GC, 0, mmMC_VM_FB_LOCATION_TOP, 0); + WREG32_SOC15(GC, 0, mmMC_VM_FB_LOCATION_BASE, 0x00FFFFFF); + WREG32_SOC15(GC, 0, mmMC_VM_AGP_TOP, 0); + WREG32_SOC15(GC, 0, mmMC_VM_AGP_BOT, 0xFFFFFF); + WREG32_SOC15(GC, 0, mmMC_VM_SYSTEM_APERTURE_LOW_ADDR, 0x3FFFFFFF); + WREG32_SOC15(GC, 0, mmMC_VM_SYSTEM_APERTURE_HIGH_ADDR, 0); + } } static void gfxhub_v1_0_init_tlb_regs(struct amdgpu_device *adev) @@ -173,8 +204,13 @@ static void gfxhub_v1_0_init_cache_regs(struct amdgpu_device *adev) WREG32_SOC15_RLC(GC, 0, mmVM_L2_CNTL3, tmp); tmp = mmVM_L2_CNTL4_DEFAULT; - tmp = REG_SET_FIELD(tmp, VM_L2_CNTL4, VMC_TAP_PDE_REQUEST_PHYSICAL, 0); - tmp = REG_SET_FIELD(tmp, VM_L2_CNTL4, VMC_TAP_PTE_REQUEST_PHYSICAL, 0); + if (adev->gmc.xgmi.connected_to_cpu) { + tmp = REG_SET_FIELD(tmp, VM_L2_CNTL4, VMC_TAP_PDE_REQUEST_PHYSICAL, 1); + tmp = REG_SET_FIELD(tmp, VM_L2_CNTL4, VMC_TAP_PTE_REQUEST_PHYSICAL, 1); + } else { + tmp = REG_SET_FIELD(tmp, VM_L2_CNTL4, VMC_TAP_PDE_REQUEST_PHYSICAL, 0); + tmp = REG_SET_FIELD(tmp, VM_L2_CNTL4, VMC_TAP_PTE_REQUEST_PHYSICAL, 0); + } WREG32_SOC15_RLC(GC, 0, mmVM_L2_CNTL4, tmp); } @@ -184,7 +220,10 @@ static void gfxhub_v1_0_enable_system_domain(struct amdgpu_device *adev) tmp = RREG32_SOC15(GC, 0, mmVM_CONTEXT0_CNTL); tmp = REG_SET_FIELD(tmp, VM_CONTEXT0_CNTL, ENABLE_CONTEXT, 1); - tmp = REG_SET_FIELD(tmp, VM_CONTEXT0_CNTL, PAGE_TABLE_DEPTH, 0); + tmp = REG_SET_FIELD(tmp, VM_CONTEXT0_CNTL, PAGE_TABLE_DEPTH, + adev->gmc.vmid0_page_table_depth); + tmp = REG_SET_FIELD(tmp, VM_CONTEXT0_CNTL, PAGE_TABLE_BLOCK_SIZE, + adev->gmc.vmid0_page_table_block_size); tmp = REG_SET_FIELD(tmp, VM_CONTEXT0_CNTL, RETRY_PERMISSION_OR_INVALID_PAGE_FAULT, 0); WREG32_SOC15(GC, 0, mmVM_CONTEXT0_CNTL, tmp); diff --git a/drivers/gpu/drm/amd/amdgpu/gfxhub_v1_1.c b/drivers/gpu/drm/amd/amdgpu/gfxhub_v1_1.c index c0ab71df0d90..8fca72ebd11c 100644 --- a/drivers/gpu/drm/amd/amdgpu/gfxhub_v1_1.c +++ b/drivers/gpu/drm/amd/amdgpu/gfxhub_v1_1.c @@ -28,13 +28,42 @@ #include "soc15_common.h" +#define mmMC_VM_XGMI_LFB_CNTL_ALDE 0x0978 +#define mmMC_VM_XGMI_LFB_CNTL_ALDE_BASE_IDX 0 +#define mmMC_VM_XGMI_LFB_SIZE_ALDE 0x0979 +#define mmMC_VM_XGMI_LFB_SIZE_ALDE_BASE_IDX 0 +//MC_VM_XGMI_LFB_CNTL +#define MC_VM_XGMI_LFB_CNTL_ALDE__PF_LFB_REGION__SHIFT 0x0 +#define MC_VM_XGMI_LFB_CNTL_ALDE__PF_MAX_REGION__SHIFT 0x4 +#define MC_VM_XGMI_LFB_CNTL_ALDE__PF_LFB_REGION_MASK 0x0000000FL +#define MC_VM_XGMI_LFB_CNTL_ALDE__PF_MAX_REGION_MASK 0x000000F0L +//MC_VM_XGMI_LFB_SIZE +#define MC_VM_XGMI_LFB_SIZE_ALDE__PF_LFB_SIZE__SHIFT 0x0 +#define MC_VM_XGMI_LFB_SIZE_ALDE__PF_LFB_SIZE_MASK 0x0001FFFFL + int gfxhub_v1_1_get_xgmi_info(struct amdgpu_device *adev) { - u32 xgmi_lfb_cntl = RREG32_SOC15(GC, 0, mmMC_VM_XGMI_LFB_CNTL); - u32 max_region = + u32 max_num_physical_nodes; + u32 max_physical_node_id; + u32 xgmi_lfb_cntl; + u32 max_region; + u64 seg_size; + + if (adev->asic_type == CHIP_ALDEBARAN) { + xgmi_lfb_cntl = RREG32_SOC15(GC, 0, mmMC_VM_XGMI_LFB_CNTL_ALDE); + seg_size = REG_GET_FIELD( + RREG32_SOC15(GC, 0, mmMC_VM_XGMI_LFB_SIZE_ALDE), + MC_VM_XGMI_LFB_SIZE, PF_LFB_SIZE) << 24; + } else { + xgmi_lfb_cntl = RREG32_SOC15(GC, 0, mmMC_VM_XGMI_LFB_CNTL); + seg_size = REG_GET_FIELD( + RREG32_SOC15(GC, 0, mmMC_VM_XGMI_LFB_SIZE), + MC_VM_XGMI_LFB_SIZE, PF_LFB_SIZE) << 24; + } + + max_region = REG_GET_FIELD(xgmi_lfb_cntl, MC_VM_XGMI_LFB_CNTL, PF_MAX_REGION); - u32 max_num_physical_nodes = 0; - u32 max_physical_node_id = 0; + switch (adev->asic_type) { case CHIP_VEGA20: @@ -45,23 +74,30 @@ int gfxhub_v1_1_get_xgmi_info(struct amdgpu_device *adev) max_num_physical_nodes = 8; max_physical_node_id = 7; break; + case CHIP_ALDEBARAN: + /* just using duplicates for Aldebaran support, revisit later */ + max_num_physical_nodes = 8; + max_physical_node_id = 7; + break; default: return -EINVAL; } /* PF_MAX_REGION=0 means xgmi is disabled */ - if (max_region) { + if (max_region || adev->gmc.xgmi.connected_to_cpu) { adev->gmc.xgmi.num_physical_nodes = max_region + 1; + if (adev->gmc.xgmi.num_physical_nodes > max_num_physical_nodes) return -EINVAL; adev->gmc.xgmi.physical_node_id = - REG_GET_FIELD(xgmi_lfb_cntl, MC_VM_XGMI_LFB_CNTL, PF_LFB_REGION); + REG_GET_FIELD(xgmi_lfb_cntl, MC_VM_XGMI_LFB_CNTL, + PF_LFB_REGION); + if (adev->gmc.xgmi.physical_node_id > max_physical_node_id) return -EINVAL; - adev->gmc.xgmi.node_segment_size = REG_GET_FIELD( - RREG32_SOC15(GC, 0, mmMC_VM_XGMI_LFB_SIZE), - MC_VM_XGMI_LFB_SIZE, PF_LFB_SIZE) << 24; + + adev->gmc.xgmi.node_segment_size = seg_size; } return 0; diff --git a/drivers/gpu/drm/amd/amdgpu/gfxhub_v2_0.c b/drivers/gpu/drm/amd/amdgpu/gfxhub_v2_0.c index 2aecc6a243e8..14c1c1a297dd 100644 --- a/drivers/gpu/drm/amd/amdgpu/gfxhub_v2_0.c +++ b/drivers/gpu/drm/amd/amdgpu/gfxhub_v2_0.c @@ -165,8 +165,7 @@ static void gfxhub_v2_0_init_system_aperture_regs(struct amdgpu_device *adev) max(adev->gmc.fb_end, adev->gmc.agp_end) >> 18); /* Set default page address. */ - value = adev->vram_scratch.gpu_addr - adev->gmc.vram_start - + adev->vm_manager.vram_base_offset; + value = amdgpu_gmc_vram_mc2pa(adev, adev->vram_scratch.gpu_addr); WREG32_SOC15(GC, 0, mmGCMC_VM_SYSTEM_APERTURE_DEFAULT_ADDR_LSB, (u32)(value >> 12)); WREG32_SOC15(GC, 0, mmGCMC_VM_SYSTEM_APERTURE_DEFAULT_ADDR_MSB, diff --git a/drivers/gpu/drm/amd/amdgpu/gfxhub_v2_1.c b/drivers/gpu/drm/amd/amdgpu/gfxhub_v2_1.c index 410fd3a1a388..41807817de7d 100644 --- a/drivers/gpu/drm/amd/amdgpu/gfxhub_v2_1.c +++ b/drivers/gpu/drm/amd/amdgpu/gfxhub_v2_1.c @@ -164,8 +164,7 @@ static void gfxhub_v2_1_init_system_aperture_regs(struct amdgpu_device *adev) max(adev->gmc.fb_end, adev->gmc.agp_end) >> 18); /* Set default page address. */ - value = adev->vram_scratch.gpu_addr - adev->gmc.vram_start - + adev->vm_manager.vram_base_offset; + value = amdgpu_gmc_vram_mc2pa(adev, adev->vram_scratch.gpu_addr); WREG32_SOC15(GC, 0, mmGCMC_VM_SYSTEM_APERTURE_DEFAULT_ADDR_LSB, (u32)(value >> 12)); WREG32_SOC15(GC, 0, mmGCMC_VM_SYSTEM_APERTURE_DEFAULT_ADDR_MSB, diff --git a/drivers/gpu/drm/amd/amdgpu/gmc_v10_0.c b/drivers/gpu/drm/amd/amdgpu/gmc_v10_0.c index 3b7c6c31fce1..498b28a35f5b 100644 --- a/drivers/gpu/drm/amd/amdgpu/gmc_v10_0.c +++ b/drivers/gpu/drm/amd/amdgpu/gmc_v10_0.c @@ -113,7 +113,7 @@ static int gmc_v10_0_process_interrupt(struct amdgpu_device *adev, /* Delegate it to a different ring if the hardware hasn't * already done it. */ - if (in_interrupt()) { + if (entry->ih == &adev->irq.ih) { amdgpu_irq_delegate(adev, entry, 8); return 1; } @@ -152,8 +152,9 @@ static int gmc_v10_0_process_interrupt(struct amdgpu_device *adev, entry->src_id, entry->ring_id, entry->vmid, entry->pasid, task_info.process_name, task_info.tgid, task_info.task_name, task_info.pid); - dev_err(adev->dev, " in page starting at address 0x%012llx from client %d\n", - addr, entry->client_id); + dev_err(adev->dev, " in page starting at address 0x%016llx from client 0x%x (%s)\n", + addr, entry->client_id, + soc15_ih_clientid_name[entry->client_id]); if (!amdgpu_sriov_vf(adev)) hub->vmhub_funcs->print_l2_protection_fault_status(adev, @@ -567,8 +568,7 @@ static void gmc_v10_0_get_vm_pde(struct amdgpu_device *adev, int level, uint64_t *addr, uint64_t *flags) { if (!(*flags & AMDGPU_PDE_PTE) && !(*flags & AMDGPU_PTE_SYSTEM)) - *addr = adev->vm_manager.vram_base_offset + *addr - - adev->gmc.vram_start; + *addr = amdgpu_gmc_vram_mc2pa(adev, *addr); BUG_ON(*addr & 0xFFFF00000000003FULL); if (!adev->gmc.translate_further) @@ -654,7 +654,7 @@ static void gmc_v10_0_set_umc_funcs(struct amdgpu_device *adev) adev->umc.umc_inst_num = UMC_V8_7_UMC_INSTANCE_NUM; adev->umc.channel_offs = UMC_V8_7_PER_CHANNEL_OFFSET_SIENNA; adev->umc.channel_idx_tbl = &umc_v8_7_channel_idx_tbl[0][0]; - adev->umc.funcs = &umc_v8_7_funcs; + adev->umc.ras_funcs = &umc_v8_7_ras_funcs; break; default: break; diff --git a/drivers/gpu/drm/amd/amdgpu/gmc_v6_0.c b/drivers/gpu/drm/amd/amdgpu/gmc_v6_0.c index f5b69484c45a..405d6ad09022 100644 --- a/drivers/gpu/drm/amd/amdgpu/gmc_v6_0.c +++ b/drivers/gpu/drm/amd/amdgpu/gmc_v6_0.c @@ -346,6 +346,7 @@ static int gmc_v6_0_mc_init(struct amdgpu_device *adev) adev->gmc.gart_size = (u64)amdgpu_gart_size << 20; } + adev->gmc.gart_size += adev->pm.smu_prv_buffer_size; gmc_v6_0_vram_gtt_location(adev, &adev->gmc); return 0; diff --git a/drivers/gpu/drm/amd/amdgpu/gmc_v7_0.c b/drivers/gpu/drm/amd/amdgpu/gmc_v7_0.c index dee2b34effb6..210ada2289ec 100644 --- a/drivers/gpu/drm/amd/amdgpu/gmc_v7_0.c +++ b/drivers/gpu/drm/amd/amdgpu/gmc_v7_0.c @@ -414,6 +414,7 @@ static int gmc_v7_0_mc_init(struct amdgpu_device *adev) adev->gmc.gart_size = (u64)amdgpu_gart_size << 20; } + adev->gmc.gart_size += adev->pm.smu_prv_buffer_size; gmc_v7_0_vram_gtt_location(adev, &adev->gmc); return 0; diff --git a/drivers/gpu/drm/amd/amdgpu/gmc_v8_0.c b/drivers/gpu/drm/amd/amdgpu/gmc_v8_0.c index 2d832fc23119..e4f27b3f28fb 100644 --- a/drivers/gpu/drm/amd/amdgpu/gmc_v8_0.c +++ b/drivers/gpu/drm/amd/amdgpu/gmc_v8_0.c @@ -59,6 +59,7 @@ MODULE_FIRMWARE("amdgpu/tonga_mc.bin"); MODULE_FIRMWARE("amdgpu/polaris11_mc.bin"); MODULE_FIRMWARE("amdgpu/polaris10_mc.bin"); MODULE_FIRMWARE("amdgpu/polaris12_mc.bin"); +MODULE_FIRMWARE("amdgpu/polaris12_32_mc.bin"); MODULE_FIRMWARE("amdgpu/polaris11_k_mc.bin"); MODULE_FIRMWARE("amdgpu/polaris10_k_mc.bin"); MODULE_FIRMWARE("amdgpu/polaris12_k_mc.bin"); @@ -243,10 +244,16 @@ static int gmc_v8_0_init_microcode(struct amdgpu_device *adev) chip_name = "polaris10"; break; case CHIP_POLARIS12: - if (ASICID_IS_P23(adev->pdev->device, adev->pdev->revision)) + if (ASICID_IS_P23(adev->pdev->device, adev->pdev->revision)) { chip_name = "polaris12_k"; - else - chip_name = "polaris12"; + } else { + WREG32(mmMC_SEQ_IO_DEBUG_INDEX, ixMC_IO_DEBUG_UP_159); + /* Polaris12 32bit ASIC needs a special MC firmware */ + if (RREG32(mmMC_SEQ_IO_DEBUG_DATA) == 0x05b4dc40) + chip_name = "polaris12_32"; + else + chip_name = "polaris12"; + } break; case CHIP_FIJI: case CHIP_CARRIZO: @@ -599,6 +606,7 @@ static int gmc_v8_0_mc_init(struct amdgpu_device *adev) adev->gmc.gart_size = (u64)amdgpu_gart_size << 20; } + adev->gmc.gart_size += adev->pm.smu_prv_buffer_size; gmc_v8_0_vram_gtt_location(adev, &adev->gmc); return 0; diff --git a/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c b/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c index 3686e777c76c..455bb91060d0 100644 --- a/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c +++ b/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c @@ -50,6 +50,7 @@ #include "athub_v1_0.h" #include "gfxhub_v1_1.h" #include "mmhub_v9_4.h" +#include "mmhub_v1_7.h" #include "umc_v6_1.h" #include "umc_v6_0.h" @@ -279,6 +280,47 @@ static const char *mmhub_client_ids_arcturus[][2] = { [384][1] = "OSS", }; +static const char *mmhub_client_ids_aldebaran[][2] = { + [2][0] = "MP1", + [3][0] = "MP0", + [32+1][0] = "DBGU_IO0", + [32+2][0] = "DBGU_IO2", + [32+4][0] = "MPIO", + [96+11][0] = "JPEG0", + [96+12][0] = "VCN0", + [96+13][0] = "VCNU0", + [128+11][0] = "JPEG1", + [128+12][0] = "VCN1", + [128+13][0] = "VCNU1", + [160+1][0] = "XDP", + [160+14][0] = "HDP", + [256+0][0] = "SDMA0", + [256+1][0] = "SDMA1", + [256+2][0] = "SDMA2", + [256+3][0] = "SDMA3", + [256+4][0] = "SDMA4", + [384+0][0] = "OSS", + [2][1] = "MP1", + [3][1] = "MP0", + [32+1][1] = "DBGU_IO0", + [32+2][1] = "DBGU_IO2", + [32+4][1] = "MPIO", + [96+11][1] = "JPEG0", + [96+12][1] = "VCN0", + [96+13][1] = "VCNU0", + [128+11][1] = "JPEG1", + [128+12][1] = "VCN1", + [128+13][1] = "VCNU1", + [160+1][1] = "XDP", + [160+14][1] = "HDP", + [256+0][1] = "SDMA0", + [256+1][1] = "SDMA1", + [256+2][1] = "SDMA2", + [256+3][1] = "SDMA3", + [256+4][1] = "SDMA4", + [384+0][1] = "OSS", +}; + static const struct soc15_reg_golden golden_settings_mmhub_1_0_0[] = { SOC15_REG_GOLDEN_VALUE(MMHUB, 0, mmDAGB1_WRCLI2, 0x00000007, 0xfe5fe0fa), @@ -484,7 +526,7 @@ static int gmc_v9_0_process_interrupt(struct amdgpu_device *adev, /* Delegate it to a different ring if the hardware hasn't * already done it. */ - if (in_interrupt()) { + if (entry->ih == &adev->irq.ih) { amdgpu_irq_delegate(adev, entry, 8); return 1; } @@ -520,8 +562,9 @@ static int gmc_v9_0_process_interrupt(struct amdgpu_device *adev, entry->src_id, entry->ring_id, entry->vmid, entry->pasid, task_info.process_name, task_info.tgid, task_info.task_name, task_info.pid); - dev_err(adev->dev, " in page starting at address 0x%012llx from client %d\n", - addr, entry->client_id); + dev_err(adev->dev, " in page starting at address 0x%016llx from IH client 0x%x (%s)\n", + addr, entry->client_id, + soc15_ih_clientid_name[entry->client_id]); if (amdgpu_sriov_vf(adev)) return 0; @@ -531,7 +574,8 @@ static int gmc_v9_0_process_interrupt(struct amdgpu_device *adev, * be updated to avoid reading an incorrect value due to * the new fast GRBM interface. */ - if (entry->vmid_src == AMDGPU_GFXHUB_0) + if ((entry->vmid_src == AMDGPU_GFXHUB_0) && + (adev->asic_type < CHIP_ALDEBARAN)) RREG32(hub->vm_l2_pro_fault_status); status = RREG32(hub->vm_l2_pro_fault_status); @@ -568,6 +612,9 @@ static int gmc_v9_0_process_interrupt(struct amdgpu_device *adev, case CHIP_RENOIR: mmhub_cid = mmhub_client_ids_renoir[cid][rw]; break; + case CHIP_ALDEBARAN: + mmhub_cid = mmhub_client_ids_aldebaran[cid][rw]; + break; default: mmhub_cid = NULL; break; @@ -607,7 +654,8 @@ static void gmc_v9_0_set_irq_funcs(struct amdgpu_device *adev) adev->gmc.vm_fault.num_types = 1; adev->gmc.vm_fault.funcs = &gmc_v9_0_irq_funcs; - if (!amdgpu_sriov_vf(adev)) { + if (!amdgpu_sriov_vf(adev) && + !adev->gmc.xgmi.connected_to_cpu) { adev->gmc.ecc_irq.num_types = 1; adev->gmc.ecc_irq.funcs = &gmc_v9_0_ecc_funcs; } @@ -642,6 +690,9 @@ static uint32_t gmc_v9_0_get_invalidate_req(unsigned int vmid, static bool gmc_v9_0_use_invalidate_semaphore(struct amdgpu_device *adev, uint32_t vmhub) { + if (adev->asic_type == CHIP_ALDEBARAN) + return false; + return ((vmhub == AMDGPU_MMHUB_0 || vmhub == AMDGPU_MMHUB_1) && (!amdgpu_sriov_vf(adev)) && @@ -752,7 +803,8 @@ static void gmc_v9_0_flush_gpu_tlb(struct amdgpu_device *adev, uint32_t vmid, * be cleared to avoid a false ACK due to the new fast * GRBM interface. */ - if (vmhub == AMDGPU_GFXHUB_0) + if ((vmhub == AMDGPU_GFXHUB_0) && + (adev->asic_type < CHIP_ALDEBARAN)) RREG32_NO_KIQ(hub->vm_inv_eng0_req + hub->eng_distance * eng); @@ -998,8 +1050,7 @@ static void gmc_v9_0_get_vm_pde(struct amdgpu_device *adev, int level, uint64_t *addr, uint64_t *flags) { if (!(*flags & AMDGPU_PDE_PTE) && !(*flags & AMDGPU_PTE_SYSTEM)) - *addr = adev->vm_manager.vram_base_offset + *addr - - adev->gmc.vram_start; + *addr = amdgpu_gmc_vram_mc2pa(adev, *addr); BUG_ON(*addr & 0xFFFF00000000003FULL); if (!adev->gmc.translate_further) @@ -1033,10 +1084,14 @@ static void gmc_v9_0_get_vm_pte(struct amdgpu_device *adev, *flags &= ~AMDGPU_PTE_VALID; } - if (adev->asic_type == CHIP_ARCTURUS && + if ((adev->asic_type == CHIP_ARCTURUS || + adev->asic_type == CHIP_ALDEBARAN) && !(*flags & AMDGPU_PTE_SYSTEM) && mapping->bo_va->is_xgmi) *flags |= AMDGPU_PTE_SNOOPED; + + if (adev->asic_type == CHIP_ALDEBARAN) + *flags |= mapping->flags & AMDGPU_PTE_SNOOPED; } static unsigned gmc_v9_0_get_vbios_fb_size(struct amdgpu_device *adev) @@ -1102,7 +1157,7 @@ static void gmc_v9_0_set_umc_funcs(struct amdgpu_device *adev) adev->umc.umc_inst_num = UMC_V6_1_UMC_INSTANCE_NUM; adev->umc.channel_offs = UMC_V6_1_PER_CHANNEL_OFFSET_VG20; adev->umc.channel_idx_tbl = &umc_v6_1_channel_idx_tbl[0][0]; - adev->umc.funcs = &umc_v6_1_funcs; + adev->umc.ras_funcs = &umc_v6_1_ras_funcs; break; case CHIP_ARCTURUS: adev->umc.max_ras_err_cnt_per_query = UMC_V6_1_TOTAL_CHANNEL_NUM; @@ -1110,7 +1165,7 @@ static void gmc_v9_0_set_umc_funcs(struct amdgpu_device *adev) adev->umc.umc_inst_num = UMC_V6_1_UMC_INSTANCE_NUM; adev->umc.channel_offs = UMC_V6_1_PER_CHANNEL_OFFSET_ARCT; adev->umc.channel_idx_tbl = &umc_v6_1_channel_idx_tbl[0][0]; - adev->umc.funcs = &umc_v6_1_funcs; + adev->umc.ras_funcs = &umc_v6_1_ras_funcs; break; default: break; @@ -1123,12 +1178,33 @@ static void gmc_v9_0_set_mmhub_funcs(struct amdgpu_device *adev) case CHIP_ARCTURUS: adev->mmhub.funcs = &mmhub_v9_4_funcs; break; + case CHIP_ALDEBARAN: + adev->mmhub.funcs = &mmhub_v1_7_funcs; + break; default: adev->mmhub.funcs = &mmhub_v1_0_funcs; break; } } +static void gmc_v9_0_set_mmhub_ras_funcs(struct amdgpu_device *adev) +{ + switch (adev->asic_type) { + case CHIP_VEGA20: + adev->mmhub.ras_funcs = &mmhub_v1_0_ras_funcs; + break; + case CHIP_ARCTURUS: + adev->mmhub.ras_funcs = &mmhub_v9_4_ras_funcs; + break; + case CHIP_ALDEBARAN: + adev->mmhub.ras_funcs = &mmhub_v1_7_ras_funcs; + break; + default: + /* mmhub ras is not available */ + break; + } +} + static void gmc_v9_0_set_gfxhub_funcs(struct amdgpu_device *adev) { adev->gfxhub.funcs = &gfxhub_v1_0_funcs; @@ -1138,10 +1214,21 @@ static int gmc_v9_0_early_init(void *handle) { struct amdgpu_device *adev = (struct amdgpu_device *)handle; + if (adev->asic_type == CHIP_VEGA20 || + adev->asic_type == CHIP_ARCTURUS) + adev->gmc.xgmi.supported = true; + + if (adev->asic_type == CHIP_ALDEBARAN) { + adev->gmc.xgmi.supported = true; + adev->gmc.xgmi.connected_to_cpu = + adev->smuio.funcs->is_host_gpu_xgmi_supported(adev); + } + gmc_v9_0_set_gmc_funcs(adev); gmc_v9_0_set_irq_funcs(adev); gmc_v9_0_set_umc_funcs(adev); gmc_v9_0_set_mmhub_funcs(adev); + gmc_v9_0_set_mmhub_ras_funcs(adev); gmc_v9_0_set_gfxhub_funcs(adev); adev->gmc.shared_aperture_start = 0x2000000000000000ULL; @@ -1174,8 +1261,9 @@ static int gmc_v9_0_late_init(void *handle) } } - if (adev->mmhub.funcs && adev->mmhub.funcs->reset_ras_error_count) - adev->mmhub.funcs->reset_ras_error_count(adev); + if (adev->mmhub.ras_funcs && + adev->mmhub.ras_funcs->reset_ras_error_count) + adev->mmhub.ras_funcs->reset_ras_error_count(adev); r = amdgpu_gmc_ras_late_init(adev); if (r) @@ -1194,9 +1282,13 @@ static void gmc_v9_0_vram_gtt_location(struct amdgpu_device *adev, /* add the xgmi offset of the physical node */ base += adev->gmc.xgmi.physical_node_id * adev->gmc.xgmi.node_segment_size; - amdgpu_gmc_vram_location(adev, mc, base); - amdgpu_gmc_gart_location(adev, mc); - amdgpu_gmc_agp_location(adev, mc); + if (adev->gmc.xgmi.connected_to_cpu) { + amdgpu_gmc_sysvm_location(adev, mc); + } else { + amdgpu_gmc_vram_location(adev, mc, base); + amdgpu_gmc_gart_location(adev, mc); + amdgpu_gmc_agp_location(adev, mc); + } /* base offset of vram pages */ adev->vm_manager.vram_base_offset = adev->gfxhub.funcs->get_mc_fb_offset(adev); @@ -1223,7 +1315,8 @@ static int gmc_v9_0_mc_init(struct amdgpu_device *adev) adev->nbio.funcs->get_memsize(adev) * 1024ULL * 1024ULL; adev->gmc.real_vram_size = adev->gmc.mc_vram_size; - if (!(adev->flags & AMD_IS_APU)) { + if (!(adev->flags & AMD_IS_APU) && + !adev->gmc.xgmi.connected_to_cpu) { r = amdgpu_device_resize_fb_bar(adev); if (r) return r; @@ -1232,10 +1325,28 @@ static int gmc_v9_0_mc_init(struct amdgpu_device *adev) adev->gmc.aper_size = pci_resource_len(adev->pdev, 0); #ifdef CONFIG_X86_64 - if (adev->flags & AMD_IS_APU) { - adev->gmc.aper_base = adev->gfxhub.funcs->get_mc_fb_offset(adev); + /* + * AMD Accelerated Processing Platform (APP) supporting GPU-HOST xgmi + * interface can use VRAM through here as it appears system reserved + * memory in host address space. + * + * For APUs, VRAM is just the stolen system memory and can be accessed + * directly. + * + * Otherwise, use the legacy Host Data Path (HDP) through PCIe BAR. + */ + + /* check whether both host-gpu and gpu-gpu xgmi links exist */ + if ((adev->flags & AMD_IS_APU) || + (adev->gmc.xgmi.supported && + adev->gmc.xgmi.connected_to_cpu)) { + adev->gmc.aper_base = + adev->gfxhub.funcs->get_mc_fb_offset(adev) + + adev->gmc.xgmi.physical_node_id * + adev->gmc.xgmi.node_segment_size; adev->gmc.aper_size = adev->gmc.real_vram_size; } + #endif /* In case the PCI BAR is larger than the actual amount of vram */ adev->gmc.visible_vram_size = adev->gmc.aper_size; @@ -1249,6 +1360,7 @@ static int gmc_v9_0_mc_init(struct amdgpu_device *adev) case CHIP_VEGA12: /* all engines support GPUVM */ case CHIP_VEGA20: case CHIP_ARCTURUS: + case CHIP_ALDEBARAN: default: adev->gmc.gart_size = 512ULL << 20; break; @@ -1261,6 +1373,8 @@ static int gmc_v9_0_mc_init(struct amdgpu_device *adev) adev->gmc.gart_size = (u64)amdgpu_gart_size << 20; } + adev->gmc.gart_size += adev->pm.smu_prv_buffer_size; + gmc_v9_0_vram_gtt_location(adev, &adev->gmc); return 0; @@ -1274,6 +1388,15 @@ static int gmc_v9_0_gart_init(struct amdgpu_device *adev) WARN(1, "VEGA10 PCIE GART already initialized\n"); return 0; } + + if (adev->gmc.xgmi.connected_to_cpu) { + adev->gmc.vmid0_page_table_depth = 1; + adev->gmc.vmid0_page_table_block_size = 12; + } else { + adev->gmc.vmid0_page_table_depth = 0; + adev->gmc.vmid0_page_table_block_size = 0; + } + /* Initialize common gart structure */ r = amdgpu_gart_init(adev); if (r) @@ -1281,7 +1404,16 @@ static int gmc_v9_0_gart_init(struct amdgpu_device *adev) adev->gart.table_size = adev->gart.num_gpu_pages * 8; adev->gart.gart_pte_flags = AMDGPU_PTE_MTYPE_VG10(MTYPE_UC) | AMDGPU_PTE_EXECUTABLE; - return amdgpu_gart_table_vram_alloc(adev); + + r = amdgpu_gart_table_vram_alloc(adev); + if (r) + return r; + + if (adev->gmc.xgmi.connected_to_cpu) { + r = amdgpu_gmc_pdb0_alloc(adev); + } + + return r; } /** @@ -1352,6 +1484,7 @@ static int gmc_v9_0_sw_init(void *handle) case CHIP_VEGA12: case CHIP_VEGA20: case CHIP_RENOIR: + case CHIP_ALDEBARAN: adev->num_vmhubs = 2; @@ -1395,7 +1528,8 @@ static int gmc_v9_0_sw_init(void *handle) if (r) return r; - if (!amdgpu_sriov_vf(adev)) { + if (!amdgpu_sriov_vf(adev) && + !adev->gmc.xgmi.connected_to_cpu) { /* interrupt sent to DF. */ r = amdgpu_irq_add_id(adev, SOC15_IH_CLIENTID_DF, 0, &adev->gmc.ecc_irq); @@ -1448,7 +1582,8 @@ static int gmc_v9_0_sw_init(void *handle) * for video processing. */ adev->vm_manager.first_kfd_vmid = - adev->asic_type == CHIP_ARCTURUS ? 3 : 8; + (adev->asic_type == CHIP_ARCTURUS || + adev->asic_type == CHIP_ALDEBARAN) ? 3 : 8; amdgpu_vm_manager_init(adev); @@ -1465,6 +1600,7 @@ static int gmc_v9_0_sw_fini(void *handle) amdgpu_gem_force_release(adev); amdgpu_vm_manager_fini(adev); amdgpu_gart_table_vram_free(adev); + amdgpu_bo_unref(&adev->gmc.pdb0_bo); amdgpu_bo_fini(adev); amdgpu_gart_fini(adev); @@ -1525,10 +1661,14 @@ static int gmc_v9_0_gart_enable(struct amdgpu_device *adev) { int r; + if (adev->gmc.xgmi.connected_to_cpu) + amdgpu_gmc_init_pdb0(adev); + if (adev->gart.bo == NULL) { dev_err(adev->dev, "No VRAM object for PCIE GART.\n"); return -EINVAL; } + r = amdgpu_gart_table_vram_pin(adev); if (r) return r; @@ -1541,9 +1681,14 @@ static int gmc_v9_0_gart_enable(struct amdgpu_device *adev) if (r) return r; - DRM_INFO("PCIE GART of %uM enabled (table at 0x%016llX).\n", - (unsigned)(adev->gmc.gart_size >> 20), - (unsigned long long)amdgpu_bo_gpu_offset(adev->gart.bo)); + DRM_INFO("PCIE GART of %uM enabled.\n", + (unsigned)(adev->gmc.gart_size >> 20)); + if (adev->gmc.pdb0_bo) + DRM_INFO("PDB0 located at 0x%016llX\n", + (unsigned long long)amdgpu_bo_gpu_offset(adev->gmc.pdb0_bo)); + DRM_INFO("PTB located at 0x%016llX\n", + (unsigned long long)amdgpu_bo_gpu_offset(adev->gart.bo)); + adev->gart.ready = true; return 0; } diff --git a/drivers/gpu/drm/amd/amdgpu/hdp_v4_0.c b/drivers/gpu/drm/amd/amdgpu/hdp_v4_0.c index e46621fed5b9..edbd35d293eb 100644 --- a/drivers/gpu/drm/amd/amdgpu/hdp_v4_0.c +++ b/drivers/gpu/drm/amd/amdgpu/hdp_v4_0.c @@ -49,6 +49,9 @@ static void hdp_v4_0_flush_hdp(struct amdgpu_device *adev, static void hdp_v4_0_invalidate_hdp(struct amdgpu_device *adev, struct amdgpu_ring *ring) { + if (adev->asic_type == CHIP_ALDEBARAN) + return; + if (!ring || !ring->funcs->emit_wreg) WREG32_SOC15_NO_KIQ(HDP, 0, mmHDP_READ_CACHE_INVALIDATE, 1); else diff --git a/drivers/gpu/drm/amd/amdgpu/jpeg_v1_0.c b/drivers/gpu/drm/amd/amdgpu/jpeg_v1_0.c index 7332a320ede8..9360204da7fb 100644 --- a/drivers/gpu/drm/amd/amdgpu/jpeg_v1_0.c +++ b/drivers/gpu/drm/amd/amdgpu/jpeg_v1_0.c @@ -487,7 +487,7 @@ int jpeg_v1_0_sw_init(void *handle) ring = &adev->jpeg.inst->ring_dec; sprintf(ring->name, "jpeg_dec"); r = amdgpu_ring_init(adev, ring, 512, &adev->jpeg.inst->irq, - 0, AMDGPU_RING_PRIO_DEFAULT); + 0, AMDGPU_RING_PRIO_DEFAULT, NULL); if (r) return r; diff --git a/drivers/gpu/drm/amd/amdgpu/jpeg_v2_0.c b/drivers/gpu/drm/amd/amdgpu/jpeg_v2_0.c index 3b22953aa62e..de5abceced0d 100644 --- a/drivers/gpu/drm/amd/amdgpu/jpeg_v2_0.c +++ b/drivers/gpu/drm/amd/amdgpu/jpeg_v2_0.c @@ -108,7 +108,7 @@ static int jpeg_v2_0_sw_init(void *handle) ring->doorbell_index = (adev->doorbell_index.vcn.vcn_ring0_1 << 1) + 1; sprintf(ring->name, "jpeg_dec"); r = amdgpu_ring_init(adev, ring, 512, &adev->jpeg.inst->irq, - 0, AMDGPU_RING_PRIO_DEFAULT); + 0, AMDGPU_RING_PRIO_DEFAULT, NULL); if (r) return r; diff --git a/drivers/gpu/drm/amd/amdgpu/jpeg_v2_5.c b/drivers/gpu/drm/amd/amdgpu/jpeg_v2_5.c index c6724a0e0c43..83531997aeba 100644 --- a/drivers/gpu/drm/amd/amdgpu/jpeg_v2_5.c +++ b/drivers/gpu/drm/amd/amdgpu/jpeg_v2_5.c @@ -115,7 +115,7 @@ static int jpeg_v2_5_sw_init(void *handle) ring->doorbell_index = (adev->doorbell_index.vcn.vcn_ring0_1 << 1) + 1 + 8 * i; sprintf(ring->name, "jpeg_dec_%d", i); r = amdgpu_ring_init(adev, ring, 512, &adev->jpeg.inst[i].irq, - 0, AMDGPU_RING_PRIO_DEFAULT); + 0, AMDGPU_RING_PRIO_DEFAULT, NULL); if (r) return r; @@ -565,6 +565,26 @@ static const struct amd_ip_funcs jpeg_v2_5_ip_funcs = { .set_powergating_state = jpeg_v2_5_set_powergating_state, }; +static const struct amd_ip_funcs jpeg_v2_6_ip_funcs = { + .name = "jpeg_v2_6", + .early_init = jpeg_v2_5_early_init, + .late_init = NULL, + .sw_init = jpeg_v2_5_sw_init, + .sw_fini = jpeg_v2_5_sw_fini, + .hw_init = jpeg_v2_5_hw_init, + .hw_fini = jpeg_v2_5_hw_fini, + .suspend = jpeg_v2_5_suspend, + .resume = jpeg_v2_5_resume, + .is_idle = jpeg_v2_5_is_idle, + .wait_for_idle = jpeg_v2_5_wait_for_idle, + .check_soft_reset = NULL, + .pre_soft_reset = NULL, + .soft_reset = NULL, + .post_soft_reset = NULL, + .set_clockgating_state = jpeg_v2_5_set_clockgating_state, + .set_powergating_state = jpeg_v2_5_set_powergating_state, +}; + static const struct amdgpu_ring_funcs jpeg_v2_5_dec_ring_vm_funcs = { .type = AMDGPU_RING_TYPE_VCN_JPEG, .align_mask = 0xf, @@ -595,6 +615,36 @@ static const struct amdgpu_ring_funcs jpeg_v2_5_dec_ring_vm_funcs = { .emit_reg_write_reg_wait = amdgpu_ring_emit_reg_write_reg_wait_helper, }; +static const struct amdgpu_ring_funcs jpeg_v2_6_dec_ring_vm_funcs = { + .type = AMDGPU_RING_TYPE_VCN_JPEG, + .align_mask = 0xf, + .vmhub = AMDGPU_MMHUB_0, + .get_rptr = jpeg_v2_5_dec_ring_get_rptr, + .get_wptr = jpeg_v2_5_dec_ring_get_wptr, + .set_wptr = jpeg_v2_5_dec_ring_set_wptr, + .emit_frame_size = + SOC15_FLUSH_GPU_TLB_NUM_WREG * 6 + + SOC15_FLUSH_GPU_TLB_NUM_REG_WAIT * 8 + + 8 + /* jpeg_v2_5_dec_ring_emit_vm_flush */ + 18 + 18 + /* jpeg_v2_5_dec_ring_emit_fence x2 vm fence */ + 8 + 16, + .emit_ib_size = 22, /* jpeg_v2_5_dec_ring_emit_ib */ + .emit_ib = jpeg_v2_0_dec_ring_emit_ib, + .emit_fence = jpeg_v2_0_dec_ring_emit_fence, + .emit_vm_flush = jpeg_v2_0_dec_ring_emit_vm_flush, + .test_ring = amdgpu_jpeg_dec_ring_test_ring, + .test_ib = amdgpu_jpeg_dec_ring_test_ib, + .insert_nop = jpeg_v2_0_dec_ring_nop, + .insert_start = jpeg_v2_0_dec_ring_insert_start, + .insert_end = jpeg_v2_0_dec_ring_insert_end, + .pad_ib = amdgpu_ring_generic_pad_ib, + .begin_use = amdgpu_jpeg_ring_begin_use, + .end_use = amdgpu_jpeg_ring_end_use, + .emit_wreg = jpeg_v2_0_dec_ring_emit_wreg, + .emit_reg_wait = jpeg_v2_0_dec_ring_emit_reg_wait, + .emit_reg_write_reg_wait = amdgpu_ring_emit_reg_write_reg_wait_helper, +}; + static void jpeg_v2_5_set_dec_ring_funcs(struct amdgpu_device *adev) { int i; @@ -602,8 +652,10 @@ static void jpeg_v2_5_set_dec_ring_funcs(struct amdgpu_device *adev) for (i = 0; i < adev->jpeg.num_jpeg_inst; ++i) { if (adev->jpeg.harvest_config & (1 << i)) continue; - - adev->jpeg.inst[i].ring_dec.funcs = &jpeg_v2_5_dec_ring_vm_funcs; + if (adev->asic_type == CHIP_ARCTURUS) + adev->jpeg.inst[i].ring_dec.funcs = &jpeg_v2_5_dec_ring_vm_funcs; + else /* CHIP_ALDEBARAN */ + adev->jpeg.inst[i].ring_dec.funcs = &jpeg_v2_6_dec_ring_vm_funcs; adev->jpeg.inst[i].ring_dec.me = i; DRM_INFO("JPEG(%d) JPEG decode is enabled in VM mode\n", i); } @@ -635,3 +687,12 @@ const struct amdgpu_ip_block_version jpeg_v2_5_ip_block = .rev = 0, .funcs = &jpeg_v2_5_ip_funcs, }; + +const struct amdgpu_ip_block_version jpeg_v2_6_ip_block = +{ + .type = AMD_IP_BLOCK_TYPE_JPEG, + .major = 2, + .minor = 6, + .rev = 0, + .funcs = &jpeg_v2_6_ip_funcs, +}; diff --git a/drivers/gpu/drm/amd/amdgpu/jpeg_v2_5.h b/drivers/gpu/drm/amd/amdgpu/jpeg_v2_5.h index 2b4087c02620..3b0aa29b9879 100644 --- a/drivers/gpu/drm/amd/amdgpu/jpeg_v2_5.h +++ b/drivers/gpu/drm/amd/amdgpu/jpeg_v2_5.h @@ -25,5 +25,6 @@ #define __JPEG_V2_5_H__ extern const struct amdgpu_ip_block_version jpeg_v2_5_ip_block; +extern const struct amdgpu_ip_block_version jpeg_v2_6_ip_block; #endif /* __JPEG_V2_5_H__ */ diff --git a/drivers/gpu/drm/amd/amdgpu/jpeg_v3_0.c b/drivers/gpu/drm/amd/amdgpu/jpeg_v3_0.c index e8fbb2a0de34..de5dfcfb3859 100644 --- a/drivers/gpu/drm/amd/amdgpu/jpeg_v3_0.c +++ b/drivers/gpu/drm/amd/amdgpu/jpeg_v3_0.c @@ -94,7 +94,7 @@ static int jpeg_v3_0_sw_init(void *handle) ring->doorbell_index = (adev->doorbell_index.vcn.vcn_ring0_1 << 1) + 1; sprintf(ring->name, "jpeg_dec"); r = amdgpu_ring_init(adev, ring, 512, &adev->jpeg.inst->irq, 0, - AMDGPU_RING_PRIO_DEFAULT); + AMDGPU_RING_PRIO_DEFAULT, NULL); if (r) return r; diff --git a/drivers/gpu/drm/amd/amdgpu/mes_v10_1.c b/drivers/gpu/drm/amd/amdgpu/mes_v10_1.c index 7f30629f21a2..a7ec4ac89da5 100644 --- a/drivers/gpu/drm/amd/amdgpu/mes_v10_1.c +++ b/drivers/gpu/drm/amd/amdgpu/mes_v10_1.c @@ -848,7 +848,8 @@ static int mes_v10_1_ring_init(struct amdgpu_device *adev) ring->no_scheduler = true; sprintf(ring->name, "mes_%d.%d.%d", ring->me, ring->pipe, ring->queue); - return amdgpu_ring_init(adev, ring, 1024, NULL, 0, AMDGPU_RING_PRIO_DEFAULT); + return amdgpu_ring_init(adev, ring, 1024, NULL, 0, + AMDGPU_RING_PRIO_DEFAULT, NULL); } static int mes_v10_1_mqd_sw_init(struct amdgpu_device *adev) diff --git a/drivers/gpu/drm/amd/amdgpu/mmhub_v1_0.c b/drivers/gpu/drm/amd/amdgpu/mmhub_v1_0.c index d7b39c07de20..a99953833820 100644 --- a/drivers/gpu/drm/amd/amdgpu/mmhub_v1_0.c +++ b/drivers/gpu/drm/amd/amdgpu/mmhub_v1_0.c @@ -114,8 +114,7 @@ static void mmhub_v1_0_init_system_aperture_regs(struct amdgpu_device *adev) return; /* Set default page address. */ - value = adev->vram_scratch.gpu_addr - adev->gmc.vram_start + - adev->vm_manager.vram_base_offset; + value = amdgpu_gmc_vram_mc2pa(adev, adev->vram_scratch.gpu_addr); WREG32_SOC15(MMHUB, 0, mmMC_VM_SYSTEM_APERTURE_DEFAULT_ADDR_LSB, (u32)(value >> 12)); WREG32_SOC15(MMHUB, 0, mmMC_VM_SYSTEM_APERTURE_DEFAULT_ADDR_MSB, @@ -776,10 +775,14 @@ static void mmhub_v1_0_reset_ras_error_count(struct amdgpu_device *adev) } } -const struct amdgpu_mmhub_funcs mmhub_v1_0_funcs = { +const struct amdgpu_mmhub_ras_funcs mmhub_v1_0_ras_funcs = { .ras_late_init = amdgpu_mmhub_ras_late_init, + .ras_fini = amdgpu_mmhub_ras_fini, .query_ras_error_count = mmhub_v1_0_query_ras_error_count, .reset_ras_error_count = mmhub_v1_0_reset_ras_error_count, +}; + +const struct amdgpu_mmhub_funcs mmhub_v1_0_funcs = { .get_fb_location = mmhub_v1_0_get_fb_location, .init = mmhub_v1_0_init, .gart_enable = mmhub_v1_0_gart_enable, diff --git a/drivers/gpu/drm/amd/amdgpu/mmhub_v1_0.h b/drivers/gpu/drm/amd/amdgpu/mmhub_v1_0.h index d77f5b65a618..4661b094e007 100644 --- a/drivers/gpu/drm/amd/amdgpu/mmhub_v1_0.h +++ b/drivers/gpu/drm/amd/amdgpu/mmhub_v1_0.h @@ -24,5 +24,6 @@ #define __MMHUB_V1_0_H__ extern const struct amdgpu_mmhub_funcs mmhub_v1_0_funcs; +extern const struct amdgpu_mmhub_ras_funcs mmhub_v1_0_ras_funcs; #endif diff --git a/drivers/gpu/drm/amd/amdgpu/mmhub_v1_7.c b/drivers/gpu/drm/amd/amdgpu/mmhub_v1_7.c new file mode 100644 index 000000000000..0103a5ab28e6 --- /dev/null +++ b/drivers/gpu/drm/amd/amdgpu/mmhub_v1_7.c @@ -0,0 +1,1333 @@ +/* + * Copyright 2019 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + */ +#include "amdgpu.h" +#include "amdgpu_ras.h" +#include "mmhub_v1_7.h" + +#include "mmhub/mmhub_1_7_offset.h" +#include "mmhub/mmhub_1_7_sh_mask.h" +#include "vega10_enum.h" + +#include "soc15_common.h" +#include "soc15.h" + +#define regVM_L2_CNTL3_DEFAULT 0x80100007 +#define regVM_L2_CNTL4_DEFAULT 0x000000c1 + +static u64 mmhub_v1_7_get_fb_location(struct amdgpu_device *adev) +{ + u64 base = RREG32_SOC15(MMHUB, 0, regMC_VM_FB_LOCATION_BASE); + u64 top = RREG32_SOC15(MMHUB, 0, regMC_VM_FB_LOCATION_TOP); + + base &= MC_VM_FB_LOCATION_BASE__FB_BASE_MASK; + base <<= 24; + + top &= MC_VM_FB_LOCATION_TOP__FB_TOP_MASK; + top <<= 24; + + adev->gmc.fb_start = base; + adev->gmc.fb_end = top; + + return base; +} + +static void mmhub_v1_7_setup_vm_pt_regs(struct amdgpu_device *adev, uint32_t vmid, + uint64_t page_table_base) +{ + struct amdgpu_vmhub *hub = &adev->vmhub[AMDGPU_MMHUB_0]; + + WREG32_SOC15_OFFSET(MMHUB, 0, regVM_CONTEXT0_PAGE_TABLE_BASE_ADDR_LO32, + hub->ctx_addr_distance * vmid, lower_32_bits(page_table_base)); + + WREG32_SOC15_OFFSET(MMHUB, 0, regVM_CONTEXT0_PAGE_TABLE_BASE_ADDR_HI32, + hub->ctx_addr_distance * vmid, upper_32_bits(page_table_base)); +} + +static void mmhub_v1_7_init_gart_aperture_regs(struct amdgpu_device *adev) +{ + uint64_t pt_base; + + if (adev->gmc.pdb0_bo) + pt_base = amdgpu_gmc_pd_addr(adev->gmc.pdb0_bo); + else + pt_base = amdgpu_gmc_pd_addr(adev->gart.bo); + + mmhub_v1_7_setup_vm_pt_regs(adev, 0, pt_base); + + /* If use GART for FB translation, vmid0 page table covers both + * vram and system memory (gart) + */ + if (adev->gmc.pdb0_bo) { + WREG32_SOC15(MMHUB, 0, regVM_CONTEXT0_PAGE_TABLE_START_ADDR_LO32, + (u32)(adev->gmc.fb_start >> 12)); + WREG32_SOC15(MMHUB, 0, regVM_CONTEXT0_PAGE_TABLE_START_ADDR_HI32, + (u32)(adev->gmc.fb_start >> 44)); + + WREG32_SOC15(MMHUB, 0, regVM_CONTEXT0_PAGE_TABLE_END_ADDR_LO32, + (u32)(adev->gmc.gart_end >> 12)); + WREG32_SOC15(MMHUB, 0, regVM_CONTEXT0_PAGE_TABLE_END_ADDR_HI32, + (u32)(adev->gmc.gart_end >> 44)); + + } else { + WREG32_SOC15(MMHUB, 0, regVM_CONTEXT0_PAGE_TABLE_START_ADDR_LO32, + (u32)(adev->gmc.gart_start >> 12)); + WREG32_SOC15(MMHUB, 0, regVM_CONTEXT0_PAGE_TABLE_START_ADDR_HI32, + (u32)(adev->gmc.gart_start >> 44)); + + WREG32_SOC15(MMHUB, 0, regVM_CONTEXT0_PAGE_TABLE_END_ADDR_LO32, + (u32)(adev->gmc.gart_end >> 12)); + WREG32_SOC15(MMHUB, 0, regVM_CONTEXT0_PAGE_TABLE_END_ADDR_HI32, + (u32)(adev->gmc.gart_end >> 44)); + } +} + +static void mmhub_v1_7_init_system_aperture_regs(struct amdgpu_device *adev) +{ + uint64_t value; + uint32_t tmp; + + /* Program the AGP BAR */ + WREG32_SOC15(MMHUB, 0, regMC_VM_AGP_BASE, 0); + WREG32_SOC15(MMHUB, 0, regMC_VM_AGP_BOT, adev->gmc.agp_start >> 24); + WREG32_SOC15(MMHUB, 0, regMC_VM_AGP_TOP, adev->gmc.agp_end >> 24); + + /* Program the system aperture low logical page number. */ + WREG32_SOC15(MMHUB, 0, regMC_VM_SYSTEM_APERTURE_LOW_ADDR, + min(adev->gmc.fb_start, adev->gmc.agp_start) >> 18); + + WREG32_SOC15(MMHUB, 0, regMC_VM_SYSTEM_APERTURE_HIGH_ADDR, + max(adev->gmc.fb_end, adev->gmc.agp_end) >> 18); + + /* In the case squeezing vram into GART aperture, we don't use + * FB aperture and AGP aperture. Disable them. + */ + if (adev->gmc.pdb0_bo) { + WREG32_SOC15(MMHUB, 0, regMC_VM_AGP_BOT, 0xFFFFFF); + WREG32_SOC15(MMHUB, 0, regMC_VM_AGP_TOP, 0); + WREG32_SOC15(MMHUB, 0, regMC_VM_FB_LOCATION_TOP, 0); + WREG32_SOC15(MMHUB, 0, regMC_VM_FB_LOCATION_BASE, 0x00FFFFFF); + WREG32_SOC15(MMHUB, 0, regMC_VM_SYSTEM_APERTURE_LOW_ADDR, 0x3FFFFFFF); + WREG32_SOC15(MMHUB, 0, regMC_VM_SYSTEM_APERTURE_HIGH_ADDR, 0); + } + if (amdgpu_sriov_vf(adev)) + return; + + /* Set default page address. */ + value = amdgpu_gmc_vram_mc2pa(adev, adev->vram_scratch.gpu_addr); + WREG32_SOC15(MMHUB, 0, regMC_VM_SYSTEM_APERTURE_DEFAULT_ADDR_LSB, + (u32)(value >> 12)); + WREG32_SOC15(MMHUB, 0, regMC_VM_SYSTEM_APERTURE_DEFAULT_ADDR_MSB, + (u32)(value >> 44)); + + /* Program "protection fault". */ + WREG32_SOC15(MMHUB, 0, regVM_L2_PROTECTION_FAULT_DEFAULT_ADDR_LO32, + (u32)(adev->dummy_page_addr >> 12)); + WREG32_SOC15(MMHUB, 0, regVM_L2_PROTECTION_FAULT_DEFAULT_ADDR_HI32, + (u32)((u64)adev->dummy_page_addr >> 44)); + + tmp = RREG32_SOC15(MMHUB, 0, regVM_L2_PROTECTION_FAULT_CNTL2); + tmp = REG_SET_FIELD(tmp, VM_L2_PROTECTION_FAULT_CNTL2, + ACTIVE_PAGE_MIGRATION_PTE_READ_RETRY, 1); + WREG32_SOC15(MMHUB, 0, regVM_L2_PROTECTION_FAULT_CNTL2, tmp); +} + +static void mmhub_v1_7_init_tlb_regs(struct amdgpu_device *adev) +{ + uint32_t tmp; + + /* Setup TLB control */ + tmp = RREG32_SOC15(MMHUB, 0, regMC_VM_MX_L1_TLB_CNTL); + + tmp = REG_SET_FIELD(tmp, MC_VM_MX_L1_TLB_CNTL, ENABLE_L1_TLB, 1); + tmp = REG_SET_FIELD(tmp, MC_VM_MX_L1_TLB_CNTL, SYSTEM_ACCESS_MODE, 3); + tmp = REG_SET_FIELD(tmp, MC_VM_MX_L1_TLB_CNTL, + ENABLE_ADVANCED_DRIVER_MODEL, 1); + tmp = REG_SET_FIELD(tmp, MC_VM_MX_L1_TLB_CNTL, + SYSTEM_APERTURE_UNMAPPED_ACCESS, 0); + tmp = REG_SET_FIELD(tmp, MC_VM_MX_L1_TLB_CNTL, ECO_BITS, 0); + tmp = REG_SET_FIELD(tmp, MC_VM_MX_L1_TLB_CNTL, + MTYPE, MTYPE_UC);/* XXX for emulation. */ + tmp = REG_SET_FIELD(tmp, MC_VM_MX_L1_TLB_CNTL, ATC_EN, 1); + + WREG32_SOC15(MMHUB, 0, regMC_VM_MX_L1_TLB_CNTL, tmp); +} + +static void mmhub_v1_7_init_cache_regs(struct amdgpu_device *adev) +{ + uint32_t tmp; + + if (amdgpu_sriov_vf(adev)) + return; + + /* Setup L2 cache */ + tmp = RREG32_SOC15(MMHUB, 0, regVM_L2_CNTL); + tmp = REG_SET_FIELD(tmp, VM_L2_CNTL, ENABLE_L2_CACHE, 1); + tmp = REG_SET_FIELD(tmp, VM_L2_CNTL, ENABLE_L2_FRAGMENT_PROCESSING, 1); + /* XXX for emulation, Refer to closed source code.*/ + tmp = REG_SET_FIELD(tmp, VM_L2_CNTL, L2_PDE0_CACHE_TAG_GENERATION_MODE, + 0); + tmp = REG_SET_FIELD(tmp, VM_L2_CNTL, PDE_FAULT_CLASSIFICATION, 0); + tmp = REG_SET_FIELD(tmp, VM_L2_CNTL, CONTEXT1_IDENTITY_ACCESS_MODE, 1); + tmp = REG_SET_FIELD(tmp, VM_L2_CNTL, IDENTITY_MODE_FRAGMENT_SIZE, 0); + WREG32_SOC15(MMHUB, 0, regVM_L2_CNTL, tmp); + + tmp = RREG32_SOC15(MMHUB, 0, regVM_L2_CNTL2); + tmp = REG_SET_FIELD(tmp, VM_L2_CNTL2, INVALIDATE_ALL_L1_TLBS, 1); + tmp = REG_SET_FIELD(tmp, VM_L2_CNTL2, INVALIDATE_L2_CACHE, 1); + WREG32_SOC15(MMHUB, 0, regVM_L2_CNTL2, tmp); + + tmp = regVM_L2_CNTL3_DEFAULT; + if (adev->gmc.translate_further) { + tmp = REG_SET_FIELD(tmp, VM_L2_CNTL3, BANK_SELECT, 12); + tmp = REG_SET_FIELD(tmp, VM_L2_CNTL3, + L2_CACHE_BIGK_FRAGMENT_SIZE, 9); + } else { + tmp = REG_SET_FIELD(tmp, VM_L2_CNTL3, BANK_SELECT, 9); + tmp = REG_SET_FIELD(tmp, VM_L2_CNTL3, + L2_CACHE_BIGK_FRAGMENT_SIZE, 6); + } + WREG32_SOC15(MMHUB, 0, regVM_L2_CNTL3, tmp); + + tmp = regVM_L2_CNTL4_DEFAULT; + if (adev->gmc.xgmi.connected_to_cpu) { + tmp = REG_SET_FIELD(tmp, VM_L2_CNTL4, + VMC_TAP_PDE_REQUEST_PHYSICAL, 1); + tmp = REG_SET_FIELD(tmp, VM_L2_CNTL4, + VMC_TAP_PTE_REQUEST_PHYSICAL, 1); + } else { + tmp = REG_SET_FIELD(tmp, VM_L2_CNTL4, + VMC_TAP_PDE_REQUEST_PHYSICAL, 0); + tmp = REG_SET_FIELD(tmp, VM_L2_CNTL4, + VMC_TAP_PTE_REQUEST_PHYSICAL, 0); + } + WREG32_SOC15(MMHUB, 0, regVM_L2_CNTL4, tmp); +} + +static void mmhub_v1_7_enable_system_domain(struct amdgpu_device *adev) +{ + uint32_t tmp; + + tmp = RREG32_SOC15(MMHUB, 0, regVM_CONTEXT0_CNTL); + tmp = REG_SET_FIELD(tmp, VM_CONTEXT0_CNTL, ENABLE_CONTEXT, 1); + tmp = REG_SET_FIELD(tmp, VM_CONTEXT0_CNTL, PAGE_TABLE_DEPTH, + adev->gmc.vmid0_page_table_depth); + tmp = REG_SET_FIELD(tmp, VM_CONTEXT0_CNTL, PAGE_TABLE_BLOCK_SIZE, + adev->gmc.vmid0_page_table_block_size); + tmp = REG_SET_FIELD(tmp, VM_CONTEXT0_CNTL, + RETRY_PERMISSION_OR_INVALID_PAGE_FAULT, 0); + WREG32_SOC15(MMHUB, 0, regVM_CONTEXT0_CNTL, tmp); +} + +static void mmhub_v1_7_disable_identity_aperture(struct amdgpu_device *adev) +{ + if (amdgpu_sriov_vf(adev)) + return; + + WREG32_SOC15(MMHUB, 0, regVM_L2_CONTEXT1_IDENTITY_APERTURE_LOW_ADDR_LO32, + 0XFFFFFFFF); + WREG32_SOC15(MMHUB, 0, regVM_L2_CONTEXT1_IDENTITY_APERTURE_LOW_ADDR_HI32, + 0x0000000F); + + WREG32_SOC15(MMHUB, 0, + regVM_L2_CONTEXT1_IDENTITY_APERTURE_HIGH_ADDR_LO32, 0); + WREG32_SOC15(MMHUB, 0, + regVM_L2_CONTEXT1_IDENTITY_APERTURE_HIGH_ADDR_HI32, 0); + + WREG32_SOC15(MMHUB, 0, regVM_L2_CONTEXT_IDENTITY_PHYSICAL_OFFSET_LO32, + 0); + WREG32_SOC15(MMHUB, 0, regVM_L2_CONTEXT_IDENTITY_PHYSICAL_OFFSET_HI32, + 0); +} + +static void mmhub_v1_7_setup_vmid_config(struct amdgpu_device *adev) +{ + struct amdgpu_vmhub *hub = &adev->vmhub[AMDGPU_MMHUB_0]; + unsigned num_level, block_size; + uint32_t tmp; + int i; + + num_level = adev->vm_manager.num_level; + block_size = adev->vm_manager.block_size; + if (adev->gmc.translate_further) + num_level -= 1; + else + block_size -= 9; + + for (i = 0; i <= 14; i++) { + tmp = RREG32_SOC15_OFFSET(MMHUB, 0, regVM_CONTEXT1_CNTL, i); + tmp = REG_SET_FIELD(tmp, VM_CONTEXT1_CNTL, ENABLE_CONTEXT, 1); + tmp = REG_SET_FIELD(tmp, VM_CONTEXT1_CNTL, PAGE_TABLE_DEPTH, + num_level); + tmp = REG_SET_FIELD(tmp, VM_CONTEXT1_CNTL, + RANGE_PROTECTION_FAULT_ENABLE_DEFAULT, 1); + tmp = REG_SET_FIELD(tmp, VM_CONTEXT1_CNTL, + DUMMY_PAGE_PROTECTION_FAULT_ENABLE_DEFAULT, + 1); + tmp = REG_SET_FIELD(tmp, VM_CONTEXT1_CNTL, + PDE0_PROTECTION_FAULT_ENABLE_DEFAULT, 1); + tmp = REG_SET_FIELD(tmp, VM_CONTEXT1_CNTL, + VALID_PROTECTION_FAULT_ENABLE_DEFAULT, 1); + tmp = REG_SET_FIELD(tmp, VM_CONTEXT1_CNTL, + READ_PROTECTION_FAULT_ENABLE_DEFAULT, 1); + tmp = REG_SET_FIELD(tmp, VM_CONTEXT1_CNTL, + WRITE_PROTECTION_FAULT_ENABLE_DEFAULT, 1); + tmp = REG_SET_FIELD(tmp, VM_CONTEXT1_CNTL, + EXECUTE_PROTECTION_FAULT_ENABLE_DEFAULT, 1); + tmp = REG_SET_FIELD(tmp, VM_CONTEXT1_CNTL, + PAGE_TABLE_BLOCK_SIZE, + block_size); + /* Send no-retry XNACK on fault to suppress VM fault storm. */ + tmp = REG_SET_FIELD(tmp, VM_CONTEXT1_CNTL, + RETRY_PERMISSION_OR_INVALID_PAGE_FAULT, + !adev->gmc.noretry); + WREG32_SOC15_OFFSET(MMHUB, 0, regVM_CONTEXT1_CNTL, + i * hub->ctx_distance, tmp); + WREG32_SOC15_OFFSET(MMHUB, 0, regVM_CONTEXT1_PAGE_TABLE_START_ADDR_LO32, + i * hub->ctx_addr_distance, 0); + WREG32_SOC15_OFFSET(MMHUB, 0, regVM_CONTEXT1_PAGE_TABLE_START_ADDR_HI32, + i * hub->ctx_addr_distance, 0); + WREG32_SOC15_OFFSET(MMHUB, 0, regVM_CONTEXT1_PAGE_TABLE_END_ADDR_LO32, + i * hub->ctx_addr_distance, + lower_32_bits(adev->vm_manager.max_pfn - 1)); + WREG32_SOC15_OFFSET(MMHUB, 0, regVM_CONTEXT1_PAGE_TABLE_END_ADDR_HI32, + i * hub->ctx_addr_distance, + upper_32_bits(adev->vm_manager.max_pfn - 1)); + } +} + +static void mmhub_v1_7_program_invalidation(struct amdgpu_device *adev) +{ + struct amdgpu_vmhub *hub = &adev->vmhub[AMDGPU_MMHUB_0]; + unsigned i; + + for (i = 0; i < 18; ++i) { + WREG32_SOC15_OFFSET(MMHUB, 0, regVM_INVALIDATE_ENG0_ADDR_RANGE_LO32, + i * hub->eng_addr_distance, 0xffffffff); + WREG32_SOC15_OFFSET(MMHUB, 0, regVM_INVALIDATE_ENG0_ADDR_RANGE_HI32, + i * hub->eng_addr_distance, 0x1f); + } +} + +static int mmhub_v1_7_gart_enable(struct amdgpu_device *adev) +{ + if (amdgpu_sriov_vf(adev)) { + /* + * MC_VM_FB_LOCATION_BASE/TOP is NULL for VF, becuase they are + * VF copy registers so vbios post doesn't program them, for + * SRIOV driver need to program them + */ + WREG32_SOC15(MMHUB, 0, regMC_VM_FB_LOCATION_BASE, + adev->gmc.vram_start >> 24); + WREG32_SOC15(MMHUB, 0, regMC_VM_FB_LOCATION_TOP, + adev->gmc.vram_end >> 24); + } + + /* GART Enable. */ + mmhub_v1_7_init_gart_aperture_regs(adev); + mmhub_v1_7_init_system_aperture_regs(adev); + mmhub_v1_7_init_tlb_regs(adev); + mmhub_v1_7_init_cache_regs(adev); + + mmhub_v1_7_enable_system_domain(adev); + mmhub_v1_7_disable_identity_aperture(adev); + mmhub_v1_7_setup_vmid_config(adev); + mmhub_v1_7_program_invalidation(adev); + + return 0; +} + +static void mmhub_v1_7_gart_disable(struct amdgpu_device *adev) +{ + struct amdgpu_vmhub *hub = &adev->vmhub[AMDGPU_MMHUB_0]; + u32 tmp; + u32 i; + + /* Disable all tables */ + for (i = 0; i < 16; i++) + WREG32_SOC15_OFFSET(MMHUB, 0, regVM_CONTEXT0_CNTL, + i * hub->ctx_distance, 0); + + /* Setup TLB control */ + tmp = RREG32_SOC15(MMHUB, 0, regMC_VM_MX_L1_TLB_CNTL); + tmp = REG_SET_FIELD(tmp, MC_VM_MX_L1_TLB_CNTL, ENABLE_L1_TLB, 0); + tmp = REG_SET_FIELD(tmp, + MC_VM_MX_L1_TLB_CNTL, + ENABLE_ADVANCED_DRIVER_MODEL, + 0); + WREG32_SOC15(MMHUB, 0, regMC_VM_MX_L1_TLB_CNTL, tmp); + + if (!amdgpu_sriov_vf(adev)) { + /* Setup L2 cache */ + tmp = RREG32_SOC15(MMHUB, 0, regVM_L2_CNTL); + tmp = REG_SET_FIELD(tmp, VM_L2_CNTL, ENABLE_L2_CACHE, 0); + WREG32_SOC15(MMHUB, 0, regVM_L2_CNTL, tmp); + WREG32_SOC15(MMHUB, 0, regVM_L2_CNTL3, 0); + } +} + +/** + * mmhub_v1_7_set_fault_enable_default - update GART/VM fault handling + * + * @adev: amdgpu_device pointer + * @value: true redirects VM faults to the default page + */ +static void mmhub_v1_7_set_fault_enable_default(struct amdgpu_device *adev, bool value) +{ + u32 tmp; + + if (amdgpu_sriov_vf(adev)) + return; + + tmp = RREG32_SOC15(MMHUB, 0, regVM_L2_PROTECTION_FAULT_CNTL); + tmp = REG_SET_FIELD(tmp, VM_L2_PROTECTION_FAULT_CNTL, + RANGE_PROTECTION_FAULT_ENABLE_DEFAULT, value); + tmp = REG_SET_FIELD(tmp, VM_L2_PROTECTION_FAULT_CNTL, + PDE0_PROTECTION_FAULT_ENABLE_DEFAULT, value); + tmp = REG_SET_FIELD(tmp, VM_L2_PROTECTION_FAULT_CNTL, + PDE1_PROTECTION_FAULT_ENABLE_DEFAULT, value); + tmp = REG_SET_FIELD(tmp, VM_L2_PROTECTION_FAULT_CNTL, + PDE2_PROTECTION_FAULT_ENABLE_DEFAULT, value); + tmp = REG_SET_FIELD(tmp, + VM_L2_PROTECTION_FAULT_CNTL, + TRANSLATE_FURTHER_PROTECTION_FAULT_ENABLE_DEFAULT, + value); + tmp = REG_SET_FIELD(tmp, VM_L2_PROTECTION_FAULT_CNTL, + NACK_PROTECTION_FAULT_ENABLE_DEFAULT, value); + tmp = REG_SET_FIELD(tmp, VM_L2_PROTECTION_FAULT_CNTL, + DUMMY_PAGE_PROTECTION_FAULT_ENABLE_DEFAULT, value); + tmp = REG_SET_FIELD(tmp, VM_L2_PROTECTION_FAULT_CNTL, + VALID_PROTECTION_FAULT_ENABLE_DEFAULT, value); + tmp = REG_SET_FIELD(tmp, VM_L2_PROTECTION_FAULT_CNTL, + READ_PROTECTION_FAULT_ENABLE_DEFAULT, value); + tmp = REG_SET_FIELD(tmp, VM_L2_PROTECTION_FAULT_CNTL, + WRITE_PROTECTION_FAULT_ENABLE_DEFAULT, value); + tmp = REG_SET_FIELD(tmp, VM_L2_PROTECTION_FAULT_CNTL, + EXECUTE_PROTECTION_FAULT_ENABLE_DEFAULT, value); + if (!value) { + tmp = REG_SET_FIELD(tmp, VM_L2_PROTECTION_FAULT_CNTL, + CRASH_ON_NO_RETRY_FAULT, 1); + tmp = REG_SET_FIELD(tmp, VM_L2_PROTECTION_FAULT_CNTL, + CRASH_ON_RETRY_FAULT, 1); + } + + WREG32_SOC15(MMHUB, 0, regVM_L2_PROTECTION_FAULT_CNTL, tmp); +} + +static void mmhub_v1_7_init(struct amdgpu_device *adev) +{ + struct amdgpu_vmhub *hub = &adev->vmhub[AMDGPU_MMHUB_0]; + + hub->ctx0_ptb_addr_lo32 = + SOC15_REG_OFFSET(MMHUB, 0, + regVM_CONTEXT0_PAGE_TABLE_BASE_ADDR_LO32); + hub->ctx0_ptb_addr_hi32 = + SOC15_REG_OFFSET(MMHUB, 0, + regVM_CONTEXT0_PAGE_TABLE_BASE_ADDR_HI32); + hub->vm_inv_eng0_req = + SOC15_REG_OFFSET(MMHUB, 0, regVM_INVALIDATE_ENG0_REQ); + hub->vm_inv_eng0_ack = + SOC15_REG_OFFSET(MMHUB, 0, regVM_INVALIDATE_ENG0_ACK); + hub->vm_context0_cntl = + SOC15_REG_OFFSET(MMHUB, 0, regVM_CONTEXT0_CNTL); + hub->vm_l2_pro_fault_status = + SOC15_REG_OFFSET(MMHUB, 0, regVM_L2_PROTECTION_FAULT_STATUS); + hub->vm_l2_pro_fault_cntl = + SOC15_REG_OFFSET(MMHUB, 0, regVM_L2_PROTECTION_FAULT_CNTL); + + hub->ctx_distance = regVM_CONTEXT1_CNTL - regVM_CONTEXT0_CNTL; + hub->ctx_addr_distance = regVM_CONTEXT1_PAGE_TABLE_BASE_ADDR_LO32 - + regVM_CONTEXT0_PAGE_TABLE_BASE_ADDR_LO32; + hub->eng_distance = regVM_INVALIDATE_ENG1_REQ - regVM_INVALIDATE_ENG0_REQ; + hub->eng_addr_distance = regVM_INVALIDATE_ENG1_ADDR_RANGE_LO32 - + regVM_INVALIDATE_ENG0_ADDR_RANGE_LO32; + +} + +static void mmhub_v1_7_update_medium_grain_clock_gating(struct amdgpu_device *adev, + bool enable) +{ + uint32_t def, data, def1, data1, def2 = 0, data2 = 0; + + def = data = RREG32_SOC15(MMHUB, 0, regATC_L2_MISC_CG); + + def1 = data1 = RREG32_SOC15(MMHUB, 0, regDAGB0_CNTL_MISC2); + def2 = data2 = RREG32_SOC15(MMHUB, 0, regDAGB1_CNTL_MISC2); + + if (enable) { + data |= ATC_L2_MISC_CG__ENABLE_MASK; + + data1 &= ~(DAGB0_CNTL_MISC2__DISABLE_WRREQ_CG_MASK | + DAGB0_CNTL_MISC2__DISABLE_WRRET_CG_MASK | + DAGB0_CNTL_MISC2__DISABLE_RDREQ_CG_MASK | + DAGB0_CNTL_MISC2__DISABLE_RDRET_CG_MASK | + DAGB0_CNTL_MISC2__DISABLE_TLBWR_CG_MASK | + DAGB0_CNTL_MISC2__DISABLE_TLBRD_CG_MASK); + + data2 &= ~(DAGB1_CNTL_MISC2__DISABLE_WRREQ_CG_MASK | + DAGB1_CNTL_MISC2__DISABLE_WRRET_CG_MASK | + DAGB1_CNTL_MISC2__DISABLE_RDREQ_CG_MASK | + DAGB1_CNTL_MISC2__DISABLE_RDRET_CG_MASK | + DAGB1_CNTL_MISC2__DISABLE_TLBWR_CG_MASK | + DAGB1_CNTL_MISC2__DISABLE_TLBRD_CG_MASK); + } else { + data &= ~ATC_L2_MISC_CG__ENABLE_MASK; + + data1 |= (DAGB0_CNTL_MISC2__DISABLE_WRREQ_CG_MASK | + DAGB0_CNTL_MISC2__DISABLE_WRRET_CG_MASK | + DAGB0_CNTL_MISC2__DISABLE_RDREQ_CG_MASK | + DAGB0_CNTL_MISC2__DISABLE_RDRET_CG_MASK | + DAGB0_CNTL_MISC2__DISABLE_TLBWR_CG_MASK | + DAGB0_CNTL_MISC2__DISABLE_TLBRD_CG_MASK); + + data2 |= (DAGB1_CNTL_MISC2__DISABLE_WRREQ_CG_MASK | + DAGB1_CNTL_MISC2__DISABLE_WRRET_CG_MASK | + DAGB1_CNTL_MISC2__DISABLE_RDREQ_CG_MASK | + DAGB1_CNTL_MISC2__DISABLE_RDRET_CG_MASK | + DAGB1_CNTL_MISC2__DISABLE_TLBWR_CG_MASK | + DAGB1_CNTL_MISC2__DISABLE_TLBRD_CG_MASK); + } + + if (def != data) + WREG32_SOC15(MMHUB, 0, regATC_L2_MISC_CG, data); + + if (def1 != data1) + WREG32_SOC15(MMHUB, 0, regDAGB0_CNTL_MISC2, data1); + + if (def2 != data2) + WREG32_SOC15(MMHUB, 0, regDAGB1_CNTL_MISC2, data2); +} + +static void mmhub_v1_7_update_medium_grain_light_sleep(struct amdgpu_device *adev, + bool enable) +{ + uint32_t def, data; + + def = data = RREG32_SOC15(MMHUB, 0, regATC_L2_MISC_CG); + + if (enable) + data |= ATC_L2_MISC_CG__MEM_LS_ENABLE_MASK; + else + data &= ~ATC_L2_MISC_CG__MEM_LS_ENABLE_MASK; + + if (def != data) + WREG32_SOC15(MMHUB, 0, regATC_L2_MISC_CG, data); +} + +static int mmhub_v1_7_set_clockgating(struct amdgpu_device *adev, + enum amd_clockgating_state state) +{ + if (amdgpu_sriov_vf(adev)) + return 0; + + /* Change state only if MCCG support is enabled through driver */ + if (adev->cg_flags & AMD_CG_SUPPORT_MC_MGCG) + mmhub_v1_7_update_medium_grain_clock_gating(adev, + state == AMD_CG_STATE_GATE); + + /* Change state only if LS support is enabled through driver */ + if (adev->cg_flags & AMD_CG_SUPPORT_MC_LS) + mmhub_v1_7_update_medium_grain_light_sleep(adev, + state == AMD_CG_STATE_GATE); + + return 0; +} + +static void mmhub_v1_7_get_clockgating(struct amdgpu_device *adev, u32 *flags) +{ + int data, data1; + + if (amdgpu_sriov_vf(adev)) + *flags = 0; + + data = RREG32_SOC15(MMHUB, 0, regATC_L2_MISC_CG); + + data1 = RREG32_SOC15(MMHUB, 0, regDAGB0_CNTL_MISC2); + + /* AMD_CG_SUPPORT_MC_MGCG */ + if ((data & ATC_L2_MISC_CG__ENABLE_MASK) && + !(data1 & (DAGB0_CNTL_MISC2__DISABLE_WRREQ_CG_MASK | + DAGB0_CNTL_MISC2__DISABLE_WRRET_CG_MASK | + DAGB0_CNTL_MISC2__DISABLE_RDREQ_CG_MASK | + DAGB0_CNTL_MISC2__DISABLE_RDRET_CG_MASK | + DAGB0_CNTL_MISC2__DISABLE_TLBWR_CG_MASK | + DAGB0_CNTL_MISC2__DISABLE_TLBRD_CG_MASK))) + *flags |= AMD_CG_SUPPORT_MC_MGCG; + + /* AMD_CG_SUPPORT_MC_LS */ + if (data & ATC_L2_MISC_CG__MEM_LS_ENABLE_MASK) + *flags |= AMD_CG_SUPPORT_MC_LS; +} + +static const struct soc15_ras_field_entry mmhub_v1_7_ras_fields[] = { + /* MMHUB Range 0 */ + { "MMEA0_DRAMRD_CMDMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA0_EDC_CNT), + SOC15_REG_FIELD(MMEA0_EDC_CNT, DRAMRD_CMDMEM_SEC_COUNT), + SOC15_REG_FIELD(MMEA0_EDC_CNT, DRAMRD_CMDMEM_DED_COUNT), + }, + { "MMEA0_DRAMWR_CMDMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA0_EDC_CNT), + SOC15_REG_FIELD(MMEA0_EDC_CNT, DRAMWR_CMDMEM_SEC_COUNT), + SOC15_REG_FIELD(MMEA0_EDC_CNT, DRAMWR_CMDMEM_DED_COUNT), + }, + { "MMEA0_DRAMWR_DATAMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA0_EDC_CNT), + SOC15_REG_FIELD(MMEA0_EDC_CNT, DRAMWR_DATAMEM_SEC_COUNT), + SOC15_REG_FIELD(MMEA0_EDC_CNT, DRAMWR_DATAMEM_DED_COUNT), + }, + { "MMEA0_RRET_TAGMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA0_EDC_CNT), + SOC15_REG_FIELD(MMEA0_EDC_CNT, RRET_TAGMEM_SEC_COUNT), + SOC15_REG_FIELD(MMEA0_EDC_CNT, RRET_TAGMEM_DED_COUNT), + }, + { "MMEA0_WRET_TAGMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA0_EDC_CNT), + SOC15_REG_FIELD(MMEA0_EDC_CNT, WRET_TAGMEM_SEC_COUNT), + SOC15_REG_FIELD(MMEA0_EDC_CNT, WRET_TAGMEM_DED_COUNT), + }, + { "MMEA0_IOWR_DATAMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA0_EDC_CNT), + SOC15_REG_FIELD(MMEA0_EDC_CNT, IOWR_DATAMEM_SEC_COUNT), + SOC15_REG_FIELD(MMEA0_EDC_CNT, IOWR_DATAMEM_DED_COUNT), + }, + { "MMEA0_DRAMRD_PAGEMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA0_EDC_CNT), + SOC15_REG_FIELD(MMEA0_EDC_CNT, DRAMRD_PAGEMEM_SED_COUNT), + 0, 0, + }, + { "MMEA0_DRAMWR_PAGEMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA0_EDC_CNT), + SOC15_REG_FIELD(MMEA0_EDC_CNT, DRAMWR_PAGEMEM_SED_COUNT), + 0, 0, + }, + { "MMEA0_IORD_CMDMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA0_EDC_CNT), + SOC15_REG_FIELD(MMEA0_EDC_CNT, IORD_CMDMEM_SED_COUNT), + 0, 0, + }, + { "MMEA0_IOWR_CMDMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA0_EDC_CNT), + SOC15_REG_FIELD(MMEA0_EDC_CNT, IOWR_CMDMEM_SED_COUNT), + 0, 0, + }, + { "MMEA0_GMIRD_CMDMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA0_EDC_CNT2), + SOC15_REG_FIELD(MMEA0_EDC_CNT2, GMIRD_CMDMEM_SEC_COUNT), + SOC15_REG_FIELD(MMEA0_EDC_CNT2, GMIRD_CMDMEM_DED_COUNT), + }, + { "MMEA0_GMIWR_CMDMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA0_EDC_CNT2), + SOC15_REG_FIELD(MMEA0_EDC_CNT2, GMIWR_CMDMEM_SEC_COUNT), + SOC15_REG_FIELD(MMEA0_EDC_CNT2, GMIWR_CMDMEM_DED_COUNT), + }, + { "MMEA0_GMIWR_DATAMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA0_EDC_CNT2), + SOC15_REG_FIELD(MMEA0_EDC_CNT2, GMIWR_DATAMEM_SEC_COUNT), + SOC15_REG_FIELD(MMEA0_EDC_CNT2, GMIWR_DATAMEM_DED_COUNT), + }, + { "MMEA0_GMIRD_PAGEMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA0_EDC_CNT2), + SOC15_REG_FIELD(MMEA0_EDC_CNT2, GMIRD_PAGEMEM_SED_COUNT), + 0, 0, + }, + { "MMEA0_GMIWR_PAGEMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA0_EDC_CNT2), + SOC15_REG_FIELD(MMEA0_EDC_CNT2, GMIWR_PAGEMEM_SED_COUNT), + 0, 0, + }, + { "MMEA0_MAM_D0MEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA0_EDC_CNT2), + SOC15_REG_FIELD(MMEA0_EDC_CNT2, MAM_D0MEM_SED_COUNT), + SOC15_REG_FIELD(MMEA0_EDC_CNT2, MAM_D0MEM_DED_COUNT), + }, + { "MMEA0_MAM_D1MEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA0_EDC_CNT2), + SOC15_REG_FIELD(MMEA0_EDC_CNT2, MAM_D1MEM_SED_COUNT), + SOC15_REG_FIELD(MMEA0_EDC_CNT2, MAM_D1MEM_DED_COUNT), + }, + { "MMEA0_MAM_D2MEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA0_EDC_CNT2), + SOC15_REG_FIELD(MMEA0_EDC_CNT2, MAM_D2MEM_SED_COUNT), + SOC15_REG_FIELD(MMEA0_EDC_CNT2, MAM_D2MEM_DED_COUNT), + }, + { "MMEA0_MAM_D3MEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA0_EDC_CNT2), + SOC15_REG_FIELD(MMEA0_EDC_CNT2, MAM_D3MEM_SED_COUNT), + SOC15_REG_FIELD(MMEA0_EDC_CNT2, MAM_D3MEM_DED_COUNT), + }, + { "MMEA0_DRAMRD_PAGEMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA0_EDC_CNT3), + 0, 0, + SOC15_REG_FIELD(MMEA0_EDC_CNT3, DRAMRD_PAGEMEM_DED_COUNT), + }, + { "MMEA0_DRAMWR_PAGEMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA0_EDC_CNT3), + 0, 0, + SOC15_REG_FIELD(MMEA0_EDC_CNT3, DRAMWR_PAGEMEM_DED_COUNT), + }, + { "MMEA0_IORD_CMDMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA0_EDC_CNT3), + 0, 0, + SOC15_REG_FIELD(MMEA0_EDC_CNT3, IORD_CMDMEM_DED_COUNT), + }, + { "MMEA0_IOWR_CMDMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA0_EDC_CNT3), + 0, 0, + SOC15_REG_FIELD(MMEA0_EDC_CNT3, IOWR_CMDMEM_DED_COUNT), + }, + { "MMEA0_GMIRD_PAGEMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA0_EDC_CNT3), + 0, 0, + SOC15_REG_FIELD(MMEA0_EDC_CNT3, GMIRD_PAGEMEM_DED_COUNT), + }, + { "MMEA0_GMIWR_PAGEMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA0_EDC_CNT3), + 0, 0, + SOC15_REG_FIELD(MMEA0_EDC_CNT3, GMIWR_PAGEMEM_DED_COUNT), + }, + + /* MMHUB Range 1 */ + { "MMEA1_DRAMRD_CMDMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA1_EDC_CNT), + SOC15_REG_FIELD(MMEA1_EDC_CNT, DRAMRD_CMDMEM_SEC_COUNT), + SOC15_REG_FIELD(MMEA1_EDC_CNT, DRAMRD_CMDMEM_DED_COUNT), + }, + { "MMEA1_DRAMWR_CMDMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA1_EDC_CNT), + SOC15_REG_FIELD(MMEA1_EDC_CNT, DRAMWR_CMDMEM_SEC_COUNT), + SOC15_REG_FIELD(MMEA1_EDC_CNT, DRAMWR_CMDMEM_DED_COUNT), + }, + { "MMEA1_DRAMWR_DATAMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA1_EDC_CNT), + SOC15_REG_FIELD(MMEA1_EDC_CNT, DRAMWR_DATAMEM_SEC_COUNT), + SOC15_REG_FIELD(MMEA1_EDC_CNT, DRAMWR_DATAMEM_DED_COUNT), + }, + { "MMEA1_RRET_TAGMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA1_EDC_CNT), + SOC15_REG_FIELD(MMEA1_EDC_CNT, RRET_TAGMEM_SEC_COUNT), + SOC15_REG_FIELD(MMEA1_EDC_CNT, RRET_TAGMEM_DED_COUNT), + }, + { "MMEA1_WRET_TAGMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA1_EDC_CNT), + SOC15_REG_FIELD(MMEA1_EDC_CNT, WRET_TAGMEM_SEC_COUNT), + SOC15_REG_FIELD(MMEA1_EDC_CNT, WRET_TAGMEM_DED_COUNT), + }, + { "MMEA1_IOWR_DATAMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA1_EDC_CNT), + SOC15_REG_FIELD(MMEA1_EDC_CNT, IOWR_DATAMEM_SEC_COUNT), + SOC15_REG_FIELD(MMEA1_EDC_CNT, IOWR_DATAMEM_DED_COUNT), + }, + { "MMEA1_DRAMRD_PAGEMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA1_EDC_CNT), + SOC15_REG_FIELD(MMEA1_EDC_CNT, DRAMRD_PAGEMEM_SED_COUNT), + 0, 0, + }, + { "MMEA1_DRAMWR_PAGEMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA1_EDC_CNT), + SOC15_REG_FIELD(MMEA1_EDC_CNT, DRAMWR_PAGEMEM_SED_COUNT), + 0, 0, + }, + { "MMEA1_IORD_CMDMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA1_EDC_CNT), + SOC15_REG_FIELD(MMEA1_EDC_CNT, IORD_CMDMEM_SED_COUNT), + 0, 0, + }, + { "MMEA1_IOWR_CMDMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA1_EDC_CNT), + SOC15_REG_FIELD(MMEA1_EDC_CNT, IOWR_CMDMEM_SED_COUNT), + 0, 0, + }, + { "MMEA1_GMIRD_CMDMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA1_EDC_CNT2), + SOC15_REG_FIELD(MMEA1_EDC_CNT2, GMIRD_CMDMEM_SEC_COUNT), + SOC15_REG_FIELD(MMEA1_EDC_CNT2, GMIRD_CMDMEM_DED_COUNT), + }, + { "MMEA1_GMIWR_CMDMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA1_EDC_CNT2), + SOC15_REG_FIELD(MMEA1_EDC_CNT2, GMIWR_CMDMEM_SEC_COUNT), + SOC15_REG_FIELD(MMEA1_EDC_CNT2, GMIWR_CMDMEM_DED_COUNT), + }, + { "MMEA1_GMIWR_DATAMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA1_EDC_CNT2), + SOC15_REG_FIELD(MMEA1_EDC_CNT2, GMIWR_DATAMEM_SEC_COUNT), + SOC15_REG_FIELD(MMEA1_EDC_CNT2, GMIWR_DATAMEM_DED_COUNT), + }, + { "MMEA1_GMIRD_PAGEMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA1_EDC_CNT2), + SOC15_REG_FIELD(MMEA1_EDC_CNT2, GMIRD_PAGEMEM_SED_COUNT), + 0, 0, + }, + { "MMEA1_GMIWR_PAGEMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA1_EDC_CNT2), + SOC15_REG_FIELD(MMEA1_EDC_CNT2, GMIWR_PAGEMEM_SED_COUNT), + 0, 0, + }, + { "MMEA1_MAM_D0MEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA1_EDC_CNT2), + SOC15_REG_FIELD(MMEA1_EDC_CNT2, MAM_D0MEM_SED_COUNT), + SOC15_REG_FIELD(MMEA1_EDC_CNT2, MAM_D0MEM_DED_COUNT), + }, + { "MMEA1_MAM_D1MEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA1_EDC_CNT2), + SOC15_REG_FIELD(MMEA1_EDC_CNT2, MAM_D1MEM_SED_COUNT), + SOC15_REG_FIELD(MMEA1_EDC_CNT2, MAM_D1MEM_DED_COUNT), + }, + { "MMEA1_MAM_D2MEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA1_EDC_CNT2), + SOC15_REG_FIELD(MMEA1_EDC_CNT2, MAM_D2MEM_SED_COUNT), + SOC15_REG_FIELD(MMEA1_EDC_CNT2, MAM_D2MEM_DED_COUNT), + }, + { "MMEA1_MAM_D3MEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA1_EDC_CNT2), + SOC15_REG_FIELD(MMEA1_EDC_CNT2, MAM_D3MEM_SED_COUNT), + SOC15_REG_FIELD(MMEA1_EDC_CNT2, MAM_D3MEM_DED_COUNT), + }, + { "MMEA1_DRAMRD_PAGEMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA1_EDC_CNT3), + 0, 0, + SOC15_REG_FIELD(MMEA1_EDC_CNT3, DRAMRD_PAGEMEM_DED_COUNT), + }, + { "MMEA1_DRAMWR_PAGEMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA1_EDC_CNT3), + 0, 0, + SOC15_REG_FIELD(MMEA1_EDC_CNT3, DRAMWR_PAGEMEM_DED_COUNT), + }, + { "MMEA1_IORD_CMDMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA1_EDC_CNT3), + 0, 0, + SOC15_REG_FIELD(MMEA1_EDC_CNT3, IORD_CMDMEM_DED_COUNT), + }, + { "MMEA1_IOWR_CMDMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA1_EDC_CNT3), + 0, 0, + SOC15_REG_FIELD(MMEA1_EDC_CNT3, IOWR_CMDMEM_DED_COUNT), + }, + { "MMEA1_GMIRD_PAGEMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA1_EDC_CNT3), + 0, 0, + SOC15_REG_FIELD(MMEA1_EDC_CNT3, GMIRD_PAGEMEM_DED_COUNT), + }, + { "MMEA1_GMIWR_PAGEMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA1_EDC_CNT3), + 0, 0, + SOC15_REG_FIELD(MMEA1_EDC_CNT3, GMIWR_PAGEMEM_DED_COUNT), + }, + + /* MMHAB Range 2*/ + { "MMEA2_DRAMRD_CMDMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA2_EDC_CNT), + SOC15_REG_FIELD(MMEA2_EDC_CNT, DRAMRD_CMDMEM_SEC_COUNT), + SOC15_REG_FIELD(MMEA2_EDC_CNT, DRAMRD_CMDMEM_DED_COUNT), + }, + { "MMEA2_DRAMWR_CMDMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA2_EDC_CNT), + SOC15_REG_FIELD(MMEA2_EDC_CNT, DRAMWR_CMDMEM_SEC_COUNT), + SOC15_REG_FIELD(MMEA2_EDC_CNT, DRAMWR_CMDMEM_DED_COUNT), + }, + { "MMEA2_DRAMWR_DATAMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA2_EDC_CNT), + SOC15_REG_FIELD(MMEA2_EDC_CNT, DRAMWR_DATAMEM_SEC_COUNT), + SOC15_REG_FIELD(MMEA2_EDC_CNT, DRAMWR_DATAMEM_DED_COUNT), + }, + { "MMEA2_RRET_TAGMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA2_EDC_CNT), + SOC15_REG_FIELD(MMEA2_EDC_CNT, RRET_TAGMEM_SEC_COUNT), + SOC15_REG_FIELD(MMEA2_EDC_CNT, RRET_TAGMEM_DED_COUNT), + }, + { "MMEA2_WRET_TAGMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA2_EDC_CNT), + SOC15_REG_FIELD(MMEA2_EDC_CNT, WRET_TAGMEM_SEC_COUNT), + SOC15_REG_FIELD(MMEA2_EDC_CNT, WRET_TAGMEM_DED_COUNT), + }, + { "MMEA2_IOWR_DATAMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA2_EDC_CNT), + SOC15_REG_FIELD(MMEA2_EDC_CNT, IOWR_DATAMEM_SEC_COUNT), + SOC15_REG_FIELD(MMEA2_EDC_CNT, IOWR_DATAMEM_DED_COUNT), + }, + { "MMEA2_DRAMRD_PAGEMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA2_EDC_CNT), + SOC15_REG_FIELD(MMEA2_EDC_CNT, DRAMRD_PAGEMEM_SED_COUNT), + 0, 0, + }, + { "MMEA2_DRAMWR_PAGEMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA2_EDC_CNT), + SOC15_REG_FIELD(MMEA2_EDC_CNT, DRAMWR_PAGEMEM_SED_COUNT), + 0, 0, + }, + { "MMEA2_IORD_CMDMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA2_EDC_CNT), + SOC15_REG_FIELD(MMEA2_EDC_CNT, IORD_CMDMEM_SED_COUNT), + 0, 0, + }, + { "MMEA2_IOWR_CMDMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA2_EDC_CNT), + SOC15_REG_FIELD(MMEA2_EDC_CNT, IOWR_CMDMEM_SED_COUNT), + 0, 0, + }, + { "MMEA2_GMIRD_CMDMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA2_EDC_CNT2), + SOC15_REG_FIELD(MMEA2_EDC_CNT2, GMIRD_CMDMEM_SEC_COUNT), + SOC15_REG_FIELD(MMEA2_EDC_CNT2, GMIRD_CMDMEM_DED_COUNT), + }, + { "MMEA2_GMIWR_CMDMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA2_EDC_CNT2), + SOC15_REG_FIELD(MMEA2_EDC_CNT2, GMIWR_CMDMEM_SEC_COUNT), + SOC15_REG_FIELD(MMEA2_EDC_CNT2, GMIWR_CMDMEM_DED_COUNT), + }, + { "MMEA2_GMIWR_DATAMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA2_EDC_CNT2), + SOC15_REG_FIELD(MMEA2_EDC_CNT2, GMIWR_DATAMEM_SEC_COUNT), + SOC15_REG_FIELD(MMEA2_EDC_CNT2, GMIWR_DATAMEM_DED_COUNT), + }, + { "MMEA2_GMIRD_PAGEMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA2_EDC_CNT2), + SOC15_REG_FIELD(MMEA2_EDC_CNT2, GMIRD_PAGEMEM_SED_COUNT), + 0, 0, + }, + { "MMEA2_GMIWR_PAGEMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA2_EDC_CNT2), + SOC15_REG_FIELD(MMEA2_EDC_CNT2, GMIWR_PAGEMEM_SED_COUNT), + 0, 0, + }, + { "MMEA2_MAM_D0MEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA2_EDC_CNT2), + SOC15_REG_FIELD(MMEA0_EDC_CNT2, MAM_D0MEM_SED_COUNT), + SOC15_REG_FIELD(MMEA0_EDC_CNT2, MAM_D0MEM_DED_COUNT), + }, + { "MMEA2_MAM_D1MEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA2_EDC_CNT2), + SOC15_REG_FIELD(MMEA2_EDC_CNT2, MAM_D1MEM_SED_COUNT), + SOC15_REG_FIELD(MMEA2_EDC_CNT2, MAM_D1MEM_DED_COUNT), + }, + { "MMEA2_MAM_D2MEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA2_EDC_CNT2), + SOC15_REG_FIELD(MMEA2_EDC_CNT2, MAM_D2MEM_SED_COUNT), + SOC15_REG_FIELD(MMEA2_EDC_CNT2, MAM_D2MEM_DED_COUNT), + }, + { "MMEA2_MAM_D3MEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA2_EDC_CNT2), + SOC15_REG_FIELD(MMEA2_EDC_CNT2, MAM_D3MEM_SED_COUNT), + SOC15_REG_FIELD(MMEA2_EDC_CNT2, MAM_D3MEM_DED_COUNT), + }, + { "MMEA2_DRAMRD_PAGEMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA2_EDC_CNT3), + 0, 0, + SOC15_REG_FIELD(MMEA2_EDC_CNT3, DRAMRD_PAGEMEM_DED_COUNT), + }, + { "MMEA2_DRAMWR_PAGEMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA2_EDC_CNT3), + 0, 0, + SOC15_REG_FIELD(MMEA2_EDC_CNT3, DRAMWR_PAGEMEM_DED_COUNT), + }, + { "MMEA2_IORD_CMDMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA2_EDC_CNT3), + 0, 0, + SOC15_REG_FIELD(MMEA2_EDC_CNT3, IORD_CMDMEM_DED_COUNT), + }, + { "MMEA2_IOWR_CMDMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA2_EDC_CNT3), + 0, 0, + SOC15_REG_FIELD(MMEA2_EDC_CNT3, IOWR_CMDMEM_DED_COUNT), + }, + { "MMEA2_GMIRD_PAGEMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA2_EDC_CNT3), + 0, 0, + SOC15_REG_FIELD(MMEA2_EDC_CNT3, GMIRD_PAGEMEM_DED_COUNT), + }, + { "MMEA2_GMIWR_PAGEMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA2_EDC_CNT3), + 0, 0, + SOC15_REG_FIELD(MMEA2_EDC_CNT3, GMIWR_PAGEMEM_DED_COUNT), + }, + + /* MMHUB Rang 3 */ + { "MMEA3_DRAMRD_CMDMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA3_EDC_CNT), + SOC15_REG_FIELD(MMEA3_EDC_CNT, DRAMRD_CMDMEM_SEC_COUNT), + SOC15_REG_FIELD(MMEA3_EDC_CNT, DRAMRD_CMDMEM_DED_COUNT), + }, + { "MMEA3_DRAMWR_CMDMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA3_EDC_CNT), + SOC15_REG_FIELD(MMEA3_EDC_CNT, DRAMWR_CMDMEM_SEC_COUNT), + SOC15_REG_FIELD(MMEA3_EDC_CNT, DRAMWR_CMDMEM_DED_COUNT), + }, + { "MMEA3_DRAMWR_DATAMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA3_EDC_CNT), + SOC15_REG_FIELD(MMEA3_EDC_CNT, DRAMWR_DATAMEM_SEC_COUNT), + SOC15_REG_FIELD(MMEA3_EDC_CNT, DRAMWR_DATAMEM_DED_COUNT), + }, + { "MMEA3_RRET_TAGMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA3_EDC_CNT), + SOC15_REG_FIELD(MMEA3_EDC_CNT, RRET_TAGMEM_SEC_COUNT), + SOC15_REG_FIELD(MMEA3_EDC_CNT, RRET_TAGMEM_DED_COUNT), + }, + { "MMEA3_WRET_TAGMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA3_EDC_CNT), + SOC15_REG_FIELD(MMEA3_EDC_CNT, WRET_TAGMEM_SEC_COUNT), + SOC15_REG_FIELD(MMEA3_EDC_CNT, WRET_TAGMEM_DED_COUNT), + }, + { "MMEA3_IOWR_DATAMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA3_EDC_CNT), + SOC15_REG_FIELD(MMEA3_EDC_CNT, IOWR_DATAMEM_SEC_COUNT), + SOC15_REG_FIELD(MMEA3_EDC_CNT, IOWR_DATAMEM_DED_COUNT), + }, + { "MMEA3_DRAMRD_PAGEMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA3_EDC_CNT), + SOC15_REG_FIELD(MMEA3_EDC_CNT, DRAMRD_PAGEMEM_SED_COUNT), + 0, 0, + }, + { "MMEA3_DRAMWR_PAGEMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA3_EDC_CNT), + SOC15_REG_FIELD(MMEA3_EDC_CNT, DRAMWR_PAGEMEM_SED_COUNT), + 0, 0, + }, + { "MMEA3_IORD_CMDMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA3_EDC_CNT), + SOC15_REG_FIELD(MMEA3_EDC_CNT, IORD_CMDMEM_SED_COUNT), + 0, 0, + }, + { "MMEA3_IOWR_CMDMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA3_EDC_CNT), + SOC15_REG_FIELD(MMEA3_EDC_CNT, IOWR_CMDMEM_SED_COUNT), + 0, 0, + }, + { "MMEA3_GMIRD_CMDMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA3_EDC_CNT2), + SOC15_REG_FIELD(MMEA3_EDC_CNT2, GMIRD_CMDMEM_SEC_COUNT), + SOC15_REG_FIELD(MMEA3_EDC_CNT2, GMIRD_CMDMEM_DED_COUNT), + }, + { "MMEA3_GMIWR_CMDMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA3_EDC_CNT2), + SOC15_REG_FIELD(MMEA3_EDC_CNT2, GMIWR_CMDMEM_SEC_COUNT), + SOC15_REG_FIELD(MMEA3_EDC_CNT2, GMIWR_CMDMEM_DED_COUNT), + }, + { "MMEA3_GMIWR_DATAMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA3_EDC_CNT2), + SOC15_REG_FIELD(MMEA3_EDC_CNT2, GMIWR_DATAMEM_SEC_COUNT), + SOC15_REG_FIELD(MMEA3_EDC_CNT2, GMIWR_DATAMEM_DED_COUNT), + }, + { "MMEA3_GMIRD_PAGEMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA3_EDC_CNT2), + SOC15_REG_FIELD(MMEA3_EDC_CNT2, GMIRD_PAGEMEM_SED_COUNT), + 0, 0, + }, + { "MMEA3_GMIWR_PAGEMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA3_EDC_CNT2), + SOC15_REG_FIELD(MMEA3_EDC_CNT2, GMIWR_PAGEMEM_SED_COUNT), + 0, 0, + }, + { "MMEA3_MAM_D0MEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA3_EDC_CNT2), + SOC15_REG_FIELD(MMEA3_EDC_CNT2, MAM_D0MEM_SED_COUNT), + SOC15_REG_FIELD(MMEA3_EDC_CNT2, MAM_D0MEM_DED_COUNT), + }, + { "MMEA3_MAM_D1MEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA3_EDC_CNT2), + SOC15_REG_FIELD(MMEA3_EDC_CNT2, MAM_D1MEM_SED_COUNT), + SOC15_REG_FIELD(MMEA3_EDC_CNT2, MAM_D1MEM_DED_COUNT), + }, + { "MMEA3_MAM_D2MEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA3_EDC_CNT2), + SOC15_REG_FIELD(MMEA3_EDC_CNT2, MAM_D2MEM_SED_COUNT), + SOC15_REG_FIELD(MMEA3_EDC_CNT2, MAM_D2MEM_DED_COUNT), + }, + { "MMEA3_MAM_D3MEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA3_EDC_CNT2), + SOC15_REG_FIELD(MMEA3_EDC_CNT2, MAM_D3MEM_SED_COUNT), + SOC15_REG_FIELD(MMEA3_EDC_CNT2, MAM_D3MEM_DED_COUNT), + }, + { "MMEA3_DRAMRD_PAGEMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA3_EDC_CNT3), + 0, 0, + SOC15_REG_FIELD(MMEA3_EDC_CNT3, DRAMRD_PAGEMEM_DED_COUNT), + }, + { "MMEA3_DRAMWR_PAGEMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA3_EDC_CNT3), + 0, 0, + SOC15_REG_FIELD(MMEA3_EDC_CNT3, DRAMWR_PAGEMEM_DED_COUNT), + }, + { "MMEA3_IORD_CMDMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA3_EDC_CNT3), + 0, 0, + SOC15_REG_FIELD(MMEA3_EDC_CNT3, IORD_CMDMEM_DED_COUNT), + }, + { "MMEA3_IOWR_CMDMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA3_EDC_CNT3), + 0, 0, + SOC15_REG_FIELD(MMEA3_EDC_CNT3, IOWR_CMDMEM_DED_COUNT), + }, + { "MMEA3_GMIRD_PAGEMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA3_EDC_CNT3), + 0, 0, + SOC15_REG_FIELD(MMEA3_EDC_CNT3, GMIRD_PAGEMEM_DED_COUNT), + }, + { "MMEA3_GMIWR_PAGEMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA3_EDC_CNT3), + 0, 0, + SOC15_REG_FIELD(MMEA3_EDC_CNT3, GMIWR_PAGEMEM_DED_COUNT), + }, + + /* MMHUB Range 4 */ + { "MMEA4_DRAMRD_CMDMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA4_EDC_CNT), + SOC15_REG_FIELD(MMEA4_EDC_CNT, DRAMRD_CMDMEM_SEC_COUNT), + SOC15_REG_FIELD(MMEA4_EDC_CNT, DRAMRD_CMDMEM_DED_COUNT), + }, + { "MMEA4_DRAMWR_CMDMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA4_EDC_CNT), + SOC15_REG_FIELD(MMEA4_EDC_CNT, DRAMWR_CMDMEM_SEC_COUNT), + SOC15_REG_FIELD(MMEA4_EDC_CNT, DRAMWR_CMDMEM_DED_COUNT), + }, + { "MMEA4_DRAMWR_DATAMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA4_EDC_CNT), + SOC15_REG_FIELD(MMEA4_EDC_CNT, DRAMWR_DATAMEM_SEC_COUNT), + SOC15_REG_FIELD(MMEA4_EDC_CNT, DRAMWR_DATAMEM_DED_COUNT), + }, + { "MMEA4_RRET_TAGMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA4_EDC_CNT), + SOC15_REG_FIELD(MMEA4_EDC_CNT, RRET_TAGMEM_SEC_COUNT), + SOC15_REG_FIELD(MMEA4_EDC_CNT, RRET_TAGMEM_DED_COUNT), + }, + { "MMEA4_WRET_TAGMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA4_EDC_CNT), + SOC15_REG_FIELD(MMEA4_EDC_CNT, WRET_TAGMEM_SEC_COUNT), + SOC15_REG_FIELD(MMEA4_EDC_CNT, WRET_TAGMEM_DED_COUNT), + }, + { "MMEA4_IOWR_DATAMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA4_EDC_CNT), + SOC15_REG_FIELD(MMEA4_EDC_CNT, IOWR_DATAMEM_SEC_COUNT), + SOC15_REG_FIELD(MMEA4_EDC_CNT, IOWR_DATAMEM_DED_COUNT), + }, + { "MMEA4_DRAMRD_PAGEMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA4_EDC_CNT), + SOC15_REG_FIELD(MMEA4_EDC_CNT, DRAMRD_PAGEMEM_SED_COUNT), + 0, 0, + }, + { "MMEA4_DRAMWR_PAGEMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA4_EDC_CNT), + SOC15_REG_FIELD(MMEA4_EDC_CNT, DRAMWR_PAGEMEM_SED_COUNT), + 0, 0, + }, + { "MMEA4_IORD_CMDMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA4_EDC_CNT), + SOC15_REG_FIELD(MMEA4_EDC_CNT, IORD_CMDMEM_SED_COUNT), + 0, 0, + }, + { "MMEA4_IOWR_CMDMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA4_EDC_CNT), + SOC15_REG_FIELD(MMEA4_EDC_CNT, IOWR_CMDMEM_SED_COUNT), + 0, 0, + }, + { "MMEA4_GMIRD_CMDMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA4_EDC_CNT2), + SOC15_REG_FIELD(MMEA4_EDC_CNT2, GMIRD_CMDMEM_SEC_COUNT), + SOC15_REG_FIELD(MMEA4_EDC_CNT2, GMIRD_CMDMEM_DED_COUNT), + }, + { "MMEA4_GMIWR_CMDMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA4_EDC_CNT2), + SOC15_REG_FIELD(MMEA4_EDC_CNT2, GMIWR_CMDMEM_SEC_COUNT), + SOC15_REG_FIELD(MMEA4_EDC_CNT2, GMIWR_CMDMEM_DED_COUNT), + }, + { "MMEA4_GMIWR_DATAMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA4_EDC_CNT2), + SOC15_REG_FIELD(MMEA4_EDC_CNT2, GMIWR_DATAMEM_SEC_COUNT), + SOC15_REG_FIELD(MMEA4_EDC_CNT2, GMIWR_DATAMEM_DED_COUNT), + }, + { "MMEA4_GMIRD_PAGEMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA4_EDC_CNT2), + SOC15_REG_FIELD(MMEA4_EDC_CNT2, GMIRD_PAGEMEM_SED_COUNT), + 0, 0, + }, + { "MMEA4_GMIWR_PAGEMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA4_EDC_CNT2), + SOC15_REG_FIELD(MMEA4_EDC_CNT2, GMIWR_PAGEMEM_SED_COUNT), + 0, 0, + }, + { "MMEA4_MAM_D0MEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA4_EDC_CNT2), + SOC15_REG_FIELD(MMEA4_EDC_CNT2, MAM_D0MEM_SED_COUNT), + SOC15_REG_FIELD(MMEA4_EDC_CNT2, MAM_D0MEM_DED_COUNT), + }, + { "MMEA4_MAM_D1MEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA4_EDC_CNT2), + SOC15_REG_FIELD(MMEA4_EDC_CNT2, MAM_D1MEM_SED_COUNT), + SOC15_REG_FIELD(MMEA4_EDC_CNT2, MAM_D1MEM_DED_COUNT), + }, + { "MMEA4_MAM_D2MEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA4_EDC_CNT2), + SOC15_REG_FIELD(MMEA4_EDC_CNT2, MAM_D2MEM_SED_COUNT), + SOC15_REG_FIELD(MMEA4_EDC_CNT2, MAM_D2MEM_DED_COUNT), + }, + { "MMEA4_MAM_D3MEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA4_EDC_CNT2), + SOC15_REG_FIELD(MMEA4_EDC_CNT2, MAM_D3MEM_SED_COUNT), + SOC15_REG_FIELD(MMEA4_EDC_CNT2, MAM_D3MEM_DED_COUNT), + }, + { "MMEA4_DRAMRD_PAGEMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA4_EDC_CNT3), + 0, 0, + SOC15_REG_FIELD(MMEA4_EDC_CNT3, DRAMRD_PAGEMEM_DED_COUNT), + }, + { "MMEA4_DRAMWR_PAGEMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA4_EDC_CNT3), + 0, 0, + SOC15_REG_FIELD(MMEA4_EDC_CNT3, DRAMWR_PAGEMEM_DED_COUNT), + }, + { "MMEA4_IORD_CMDMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA4_EDC_CNT3), + 0, 0, + SOC15_REG_FIELD(MMEA4_EDC_CNT3, IORD_CMDMEM_DED_COUNT), + }, + { "MMEA4_IOWR_CMDMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA4_EDC_CNT3), + 0, 0, + SOC15_REG_FIELD(MMEA4_EDC_CNT3, IOWR_CMDMEM_DED_COUNT), + }, + { "MMEA4_GMIRD_PAGEMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA4_EDC_CNT3), + 0, 0, + SOC15_REG_FIELD(MMEA4_EDC_CNT3, GMIRD_PAGEMEM_DED_COUNT), + }, + { "MMEA4_GMIWR_PAGEMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA4_EDC_CNT3), + 0, 0, + SOC15_REG_FIELD(MMEA4_EDC_CNT3, GMIWR_PAGEMEM_DED_COUNT), + }, + + /* MMHUAB Range 5 */ + { "MMEA5_DRAMRD_CMDMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA5_EDC_CNT), + SOC15_REG_FIELD(MMEA5_EDC_CNT, DRAMRD_CMDMEM_SEC_COUNT), + SOC15_REG_FIELD(MMEA5_EDC_CNT, DRAMRD_CMDMEM_DED_COUNT), + }, + { "MMEA5_DRAMWR_CMDMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA5_EDC_CNT), + SOC15_REG_FIELD(MMEA5_EDC_CNT, DRAMWR_CMDMEM_SEC_COUNT), + SOC15_REG_FIELD(MMEA5_EDC_CNT, DRAMWR_CMDMEM_DED_COUNT), + }, + { "MMEA5_DRAMWR_DATAMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA5_EDC_CNT), + SOC15_REG_FIELD(MMEA5_EDC_CNT, DRAMWR_DATAMEM_SEC_COUNT), + SOC15_REG_FIELD(MMEA5_EDC_CNT, DRAMWR_DATAMEM_DED_COUNT), + }, + { "MMEA5_RRET_TAGMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA5_EDC_CNT), + SOC15_REG_FIELD(MMEA5_EDC_CNT, RRET_TAGMEM_SEC_COUNT), + SOC15_REG_FIELD(MMEA5_EDC_CNT, RRET_TAGMEM_DED_COUNT), + }, + { "MMEA5_WRET_TAGMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA5_EDC_CNT), + SOC15_REG_FIELD(MMEA5_EDC_CNT, WRET_TAGMEM_SEC_COUNT), + SOC15_REG_FIELD(MMEA5_EDC_CNT, WRET_TAGMEM_DED_COUNT), + }, + { "MMEA5_IOWR_DATAMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA5_EDC_CNT), + SOC15_REG_FIELD(MMEA5_EDC_CNT, IOWR_DATAMEM_SEC_COUNT), + SOC15_REG_FIELD(MMEA5_EDC_CNT, IOWR_DATAMEM_DED_COUNT), + }, + { "MMEA5_DRAMRD_PAGEMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA5_EDC_CNT), + SOC15_REG_FIELD(MMEA5_EDC_CNT, DRAMRD_PAGEMEM_SED_COUNT), + 0, 0, + }, + { "MMEA5_DRAMWR_PAGEMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA5_EDC_CNT), + SOC15_REG_FIELD(MMEA5_EDC_CNT, DRAMWR_PAGEMEM_SED_COUNT), + 0, 0, + }, + { "MMEA5_IORD_CMDMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA5_EDC_CNT), + SOC15_REG_FIELD(MMEA5_EDC_CNT, IORD_CMDMEM_SED_COUNT), + 0, 0, + }, + { "MMEA5_IOWR_CMDMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA5_EDC_CNT), + SOC15_REG_FIELD(MMEA5_EDC_CNT, IOWR_CMDMEM_SED_COUNT), + 0, 0, + }, + { "MMEA5_GMIRD_CMDMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA5_EDC_CNT2), + SOC15_REG_FIELD(MMEA5_EDC_CNT2, GMIRD_CMDMEM_SEC_COUNT), + SOC15_REG_FIELD(MMEA5_EDC_CNT2, GMIRD_CMDMEM_DED_COUNT), + }, + { "MMEA5_GMIWR_CMDMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA5_EDC_CNT2), + SOC15_REG_FIELD(MMEA5_EDC_CNT2, GMIWR_CMDMEM_SEC_COUNT), + SOC15_REG_FIELD(MMEA5_EDC_CNT2, GMIWR_CMDMEM_DED_COUNT), + }, + { "MMEA5_GMIWR_DATAMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA5_EDC_CNT2), + SOC15_REG_FIELD(MMEA5_EDC_CNT2, GMIWR_DATAMEM_SEC_COUNT), + SOC15_REG_FIELD(MMEA5_EDC_CNT2, GMIWR_DATAMEM_DED_COUNT), + }, + { "MMEA5_GMIRD_PAGEMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA5_EDC_CNT2), + SOC15_REG_FIELD(MMEA5_EDC_CNT2, GMIRD_PAGEMEM_SED_COUNT), + 0, 0, + }, + { "MMEA5_GMIWR_PAGEMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA5_EDC_CNT2), + SOC15_REG_FIELD(MMEA5_EDC_CNT2, GMIWR_PAGEMEM_SED_COUNT), + 0, 0, + }, + { "MMEA5_MAM_D0MEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA5_EDC_CNT2), + SOC15_REG_FIELD(MMEA5_EDC_CNT2, MAM_D0MEM_SED_COUNT), + SOC15_REG_FIELD(MMEA5_EDC_CNT2, MAM_D0MEM_DED_COUNT), + }, + { "MMEA5_MAM_D1MEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA5_EDC_CNT2), + SOC15_REG_FIELD(MMEA5_EDC_CNT2, MAM_D1MEM_SED_COUNT), + SOC15_REG_FIELD(MMEA5_EDC_CNT2, MAM_D1MEM_DED_COUNT), + }, + { "MMEA5_MAM_D2MEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA5_EDC_CNT2), + SOC15_REG_FIELD(MMEA5_EDC_CNT2, MAM_D2MEM_SED_COUNT), + SOC15_REG_FIELD(MMEA5_EDC_CNT2, MAM_D2MEM_DED_COUNT), + }, + { "MMEA5_MAM_D3MEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA5_EDC_CNT2), + SOC15_REG_FIELD(MMEA5_EDC_CNT2, MAM_D3MEM_SED_COUNT), + SOC15_REG_FIELD(MMEA5_EDC_CNT2, MAM_D3MEM_DED_COUNT), + }, + { "MMEA5_DRAMRD_PAGEMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA5_EDC_CNT3), + 0, 0, + SOC15_REG_FIELD(MMEA5_EDC_CNT3, DRAMRD_PAGEMEM_DED_COUNT), + }, + { "MMEA5_DRAMWR_PAGEMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA5_EDC_CNT3), + 0, 0, + SOC15_REG_FIELD(MMEA5_EDC_CNT3, DRAMWR_PAGEMEM_DED_COUNT), + }, + { "MMEA5_IORD_CMDMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA5_EDC_CNT3), + 0, 0, + SOC15_REG_FIELD(MMEA5_EDC_CNT3, IORD_CMDMEM_DED_COUNT), + }, + { "MMEA5_IOWR_CMDMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA5_EDC_CNT3), + 0, 0, + SOC15_REG_FIELD(MMEA5_EDC_CNT3, IOWR_CMDMEM_DED_COUNT), + }, + { "MMEA5_GMIRD_PAGEMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA5_EDC_CNT3), + 0, 0, + SOC15_REG_FIELD(MMEA5_EDC_CNT3, GMIRD_PAGEMEM_DED_COUNT), + }, + { "MMEA5_GMIWR_PAGEMEM", SOC15_REG_ENTRY(MMHUB, 0, regMMEA5_EDC_CNT3), + 0, 0, + SOC15_REG_FIELD(MMEA5_EDC_CNT3, GMIWR_PAGEMEM_DED_COUNT), + }, +}; + +static const struct soc15_reg_entry mmhub_v1_7_edc_cnt_regs[] = { + { SOC15_REG_ENTRY(MMHUB, 0, regMMEA0_EDC_CNT), 0, 0, 0 }, + { SOC15_REG_ENTRY(MMHUB, 0, regMMEA0_EDC_CNT2), 0, 0, 0 }, + { SOC15_REG_ENTRY(MMHUB, 0, regMMEA0_EDC_CNT3), 0, 0, 0 }, + { SOC15_REG_ENTRY(MMHUB, 0, regMMEA1_EDC_CNT), 0, 0, 0 }, + { SOC15_REG_ENTRY(MMHUB, 0, regMMEA1_EDC_CNT2), 0, 0, 0 }, + { SOC15_REG_ENTRY(MMHUB, 0, regMMEA1_EDC_CNT3), 0, 0, 0 }, + { SOC15_REG_ENTRY(MMHUB, 0, regMMEA2_EDC_CNT), 0, 0, 0 }, + { SOC15_REG_ENTRY(MMHUB, 0, regMMEA2_EDC_CNT2), 0, 0, 0 }, + { SOC15_REG_ENTRY(MMHUB, 0, regMMEA2_EDC_CNT3), 0, 0, 0 }, + { SOC15_REG_ENTRY(MMHUB, 0, regMMEA3_EDC_CNT), 0, 0, 0 }, + { SOC15_REG_ENTRY(MMHUB, 0, regMMEA3_EDC_CNT2), 0, 0, 0 }, + { SOC15_REG_ENTRY(MMHUB, 0, regMMEA3_EDC_CNT3), 0, 0, 0 }, + { SOC15_REG_ENTRY(MMHUB, 0, regMMEA4_EDC_CNT), 0, 0, 0 }, + { SOC15_REG_ENTRY(MMHUB, 0, regMMEA4_EDC_CNT2), 0, 0, 0 }, + { SOC15_REG_ENTRY(MMHUB, 0, regMMEA4_EDC_CNT3), 0, 0, 0 }, + { SOC15_REG_ENTRY(MMHUB, 0, regMMEA5_EDC_CNT), 0, 0, 0 }, + { SOC15_REG_ENTRY(MMHUB, 0, regMMEA5_EDC_CNT2), 0, 0, 0 }, + { SOC15_REG_ENTRY(MMHUB, 0, regMMEA5_EDC_CNT3), 0, 0, 0 }, +}; + +static int mmhub_v1_7_get_ras_error_count(struct amdgpu_device *adev, + const struct soc15_reg_entry *reg, + uint32_t value, + uint32_t *sec_count, + uint32_t *ded_count) +{ + uint32_t i; + uint32_t sec_cnt, ded_cnt; + + for (i = 0; i < ARRAY_SIZE(mmhub_v1_7_ras_fields); i++) { + if(mmhub_v1_7_ras_fields[i].reg_offset != reg->reg_offset) + continue; + + sec_cnt = (value & + mmhub_v1_7_ras_fields[i].sec_count_mask) >> + mmhub_v1_7_ras_fields[i].sec_count_shift; + if (sec_cnt) { + dev_info(adev->dev, "MMHUB SubBlock %s, SEC %d\n", + mmhub_v1_7_ras_fields[i].name, + sec_cnt); + *sec_count += sec_cnt; + } + + ded_cnt = (value & + mmhub_v1_7_ras_fields[i].ded_count_mask) >> + mmhub_v1_7_ras_fields[i].ded_count_shift; + if (ded_cnt) { + dev_info(adev->dev, "MMHUB SubBlock %s, DED %d\n", + mmhub_v1_7_ras_fields[i].name, + ded_cnt); + *ded_count += ded_cnt; + } + } + + return 0; +} + +static void mmhub_v1_7_query_ras_error_count(struct amdgpu_device *adev, + void *ras_error_status) +{ + struct ras_err_data *err_data = (struct ras_err_data *)ras_error_status; + uint32_t sec_count = 0, ded_count = 0; + uint32_t i; + uint32_t reg_value; + + err_data->ue_count = 0; + err_data->ce_count = 0; + + for (i = 0; i < ARRAY_SIZE(mmhub_v1_7_edc_cnt_regs); i++) { + reg_value = + RREG32(SOC15_REG_ENTRY_OFFSET(mmhub_v1_7_edc_cnt_regs[i])); + if (reg_value) + mmhub_v1_7_get_ras_error_count(adev, &mmhub_v1_7_edc_cnt_regs[i], + reg_value, &sec_count, &ded_count); + } + + err_data->ce_count += sec_count; + err_data->ue_count += ded_count; +} + +static void mmhub_v1_7_reset_ras_error_count(struct amdgpu_device *adev) +{ + uint32_t i; + + /* write 0 to reset the edc counters */ + if (amdgpu_ras_is_supported(adev, AMDGPU_RAS_BLOCK__MMHUB)) { + for (i = 0; i < ARRAY_SIZE(mmhub_v1_7_edc_cnt_regs); i++) + WREG32(SOC15_REG_ENTRY_OFFSET(mmhub_v1_7_edc_cnt_regs[i]), 0); + } +} + +static const struct soc15_reg_entry mmhub_v1_7_ea_err_status_regs[] = { + { SOC15_REG_ENTRY(MMHUB, 0, regMMEA0_ERR_STATUS), 0, 0, 0 }, + { SOC15_REG_ENTRY(MMHUB, 0, regMMEA1_ERR_STATUS), 0, 0, 0 }, + { SOC15_REG_ENTRY(MMHUB, 0, regMMEA2_ERR_STATUS), 0, 0, 0 }, + { SOC15_REG_ENTRY(MMHUB, 0, regMMEA3_ERR_STATUS), 0, 0, 0 }, + { SOC15_REG_ENTRY(MMHUB, 0, regMMEA4_ERR_STATUS), 0, 0, 0 }, + { SOC15_REG_ENTRY(MMHUB, 0, regMMEA5_ERR_STATUS), 0, 0, 0 }, +}; + +static void mmhub_v1_7_query_ras_error_status(struct amdgpu_device *adev) +{ + int i; + uint32_t reg_value; + + if (!amdgpu_ras_is_supported(adev, AMDGPU_RAS_BLOCK__MMHUB)) + return; + + for (i = 0; i < ARRAY_SIZE(mmhub_v1_7_ea_err_status_regs); i++) { + reg_value = + RREG32(SOC15_REG_ENTRY_OFFSET(mmhub_v1_7_ea_err_status_regs[i])); + if (REG_GET_FIELD(reg_value, MMEA0_ERR_STATUS, SDP_RDRSP_STATUS) || + REG_GET_FIELD(reg_value, MMEA0_ERR_STATUS, SDP_WRRSP_STATUS) || + REG_GET_FIELD(reg_value, MMEA0_ERR_STATUS, SDP_RDRSP_DATAPARITY_ERROR)) { + dev_warn(adev->dev, "MMHUB EA err detected at instance: %d, status: 0x%x!\n", + i, reg_value); + } + } +} + +const struct amdgpu_mmhub_ras_funcs mmhub_v1_7_ras_funcs = { + .ras_late_init = amdgpu_mmhub_ras_late_init, + .ras_fini = amdgpu_mmhub_ras_fini, + .query_ras_error_count = mmhub_v1_7_query_ras_error_count, + .reset_ras_error_count = mmhub_v1_7_reset_ras_error_count, + .query_ras_error_status = mmhub_v1_7_query_ras_error_status, +}; + +const struct amdgpu_mmhub_funcs mmhub_v1_7_funcs = { + .get_fb_location = mmhub_v1_7_get_fb_location, + .init = mmhub_v1_7_init, + .gart_enable = mmhub_v1_7_gart_enable, + .set_fault_enable_default = mmhub_v1_7_set_fault_enable_default, + .gart_disable = mmhub_v1_7_gart_disable, + .set_clockgating = mmhub_v1_7_set_clockgating, + .get_clockgating = mmhub_v1_7_get_clockgating, + .setup_vm_pt_regs = mmhub_v1_7_setup_vm_pt_regs, +}; diff --git a/drivers/gpu/drm/amd/amdgpu/mmhub_v1_7.h b/drivers/gpu/drm/amd/amdgpu/mmhub_v1_7.h new file mode 100644 index 000000000000..a7f9dfc24697 --- /dev/null +++ b/drivers/gpu/drm/amd/amdgpu/mmhub_v1_7.h @@ -0,0 +1,29 @@ +/* + * Copyright 2016 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + */ +#ifndef __MMHUB_V1_7_H__ +#define __MMHUB_V1_7_H__ + +extern const struct amdgpu_mmhub_funcs mmhub_v1_7_funcs; +extern const struct amdgpu_mmhub_ras_funcs mmhub_v1_7_ras_funcs; + +#endif diff --git a/drivers/gpu/drm/amd/amdgpu/mmhub_v2_0.c b/drivers/gpu/drm/amd/amdgpu/mmhub_v2_0.c index f107385faba2..ac76081b91d5 100644 --- a/drivers/gpu/drm/amd/amdgpu/mmhub_v2_0.c +++ b/drivers/gpu/drm/amd/amdgpu/mmhub_v2_0.c @@ -210,8 +210,7 @@ static void mmhub_v2_0_init_system_aperture_regs(struct amdgpu_device *adev) } /* Set default page address. */ - value = adev->vram_scratch.gpu_addr - adev->gmc.vram_start + - adev->vm_manager.vram_base_offset; + value = amdgpu_gmc_vram_mc2pa(adev, adev->vram_scratch.gpu_addr); WREG32_SOC15(MMHUB, 0, mmMMMC_VM_SYSTEM_APERTURE_DEFAULT_ADDR_LSB, (u32)(value >> 12)); WREG32_SOC15(MMHUB, 0, mmMMMC_VM_SYSTEM_APERTURE_DEFAULT_ADDR_MSB, @@ -689,7 +688,6 @@ static void mmhub_v2_0_get_clockgating(struct amdgpu_device *adev, u32 *flags) } const struct amdgpu_mmhub_funcs mmhub_v2_0_funcs = { - .ras_late_init = amdgpu_mmhub_ras_late_init, .init = mmhub_v2_0_init, .gart_enable = mmhub_v2_0_gart_enable, .set_fault_enable_default = mmhub_v2_0_set_fault_enable_default, diff --git a/drivers/gpu/drm/amd/amdgpu/mmhub_v2_3.c b/drivers/gpu/drm/amd/amdgpu/mmhub_v2_3.c index ab9be5ad5a5f..a9899335d0b1 100644 --- a/drivers/gpu/drm/amd/amdgpu/mmhub_v2_3.c +++ b/drivers/gpu/drm/amd/amdgpu/mmhub_v2_3.c @@ -162,8 +162,7 @@ static void mmhub_v2_3_init_system_aperture_regs(struct amdgpu_device *adev) max(adev->gmc.fb_end, adev->gmc.agp_end) >> 18); /* Set default page address. */ - value = adev->vram_scratch.gpu_addr - adev->gmc.vram_start + - adev->vm_manager.vram_base_offset; + value = amdgpu_gmc_vram_mc2pa(adev, adev->vram_scratch.gpu_addr); WREG32_SOC15(MMHUB, 0, mmMMMC_VM_SYSTEM_APERTURE_DEFAULT_ADDR_LSB, (u32)(value >> 12)); WREG32_SOC15(MMHUB, 0, mmMMMC_VM_SYSTEM_APERTURE_DEFAULT_ADDR_MSB, @@ -616,7 +615,6 @@ static void mmhub_v2_3_get_clockgating(struct amdgpu_device *adev, u32 *flags) } const struct amdgpu_mmhub_funcs mmhub_v2_3_funcs = { - .ras_late_init = amdgpu_mmhub_ras_late_init, .init = mmhub_v2_3_init, .gart_enable = mmhub_v2_3_gart_enable, .set_fault_enable_default = mmhub_v2_3_set_fault_enable_default, diff --git a/drivers/gpu/drm/amd/amdgpu/mmhub_v9_4.c b/drivers/gpu/drm/amd/amdgpu/mmhub_v9_4.c index 4a31737b6bb0..47c8dd9d1c78 100644 --- a/drivers/gpu/drm/amd/amdgpu/mmhub_v9_4.c +++ b/drivers/gpu/drm/amd/amdgpu/mmhub_v9_4.c @@ -136,8 +136,7 @@ static void mmhub_v9_4_init_system_aperture_regs(struct amdgpu_device *adev, max(adev->gmc.fb_end, adev->gmc.agp_end) >> 18); /* Set default page address. */ - value = adev->vram_scratch.gpu_addr - adev->gmc.vram_start + - adev->vm_manager.vram_base_offset; + value = amdgpu_gmc_vram_mc2pa(adev, adev->vram_scratch.gpu_addr); WREG32_SOC15_OFFSET( MMHUB, 0, mmVMSHAREDPF0_MC_VM_SYSTEM_APERTURE_DEFAULT_ADDR_LSB, @@ -1646,16 +1645,27 @@ static void mmhub_v9_4_query_ras_error_status(struct amdgpu_device *adev) for (i = 0; i < ARRAY_SIZE(mmhub_v9_4_err_status_regs); i++) { reg_value = RREG32(SOC15_REG_ENTRY_OFFSET(mmhub_v9_4_err_status_regs[i])); - if (reg_value) + if (REG_GET_FIELD(reg_value, MMEA0_ERR_STATUS, SDP_RDRSP_STATUS) || + REG_GET_FIELD(reg_value, MMEA0_ERR_STATUS, SDP_WRRSP_STATUS) || + REG_GET_FIELD(reg_value, MMEA0_ERR_STATUS, SDP_RDRSP_DATAPARITY_ERROR)) { + /* SDP read/write error/parity error in FUE_IS_FATAL mode + * can cause system fatal error in arcturas. Harvest the error + * status before GPU reset */ dev_warn(adev->dev, "MMHUB EA err detected at instance: %d, status: 0x%x!\n", i, reg_value); + } } } -const struct amdgpu_mmhub_funcs mmhub_v9_4_funcs = { +const struct amdgpu_mmhub_ras_funcs mmhub_v9_4_ras_funcs = { .ras_late_init = amdgpu_mmhub_ras_late_init, + .ras_fini = amdgpu_mmhub_ras_fini, .query_ras_error_count = mmhub_v9_4_query_ras_error_count, .reset_ras_error_count = mmhub_v9_4_reset_ras_error_count, + .query_ras_error_status = mmhub_v9_4_query_ras_error_status, +}; + +const struct amdgpu_mmhub_funcs mmhub_v9_4_funcs = { .get_fb_location = mmhub_v9_4_get_fb_location, .init = mmhub_v9_4_init, .gart_enable = mmhub_v9_4_gart_enable, @@ -1664,5 +1674,4 @@ const struct amdgpu_mmhub_funcs mmhub_v9_4_funcs = { .set_clockgating = mmhub_v9_4_set_clockgating, .get_clockgating = mmhub_v9_4_get_clockgating, .setup_vm_pt_regs = mmhub_v9_4_setup_vm_pt_regs, - .query_ras_error_status = mmhub_v9_4_query_ras_error_status, }; diff --git a/drivers/gpu/drm/amd/amdgpu/mmhub_v9_4.h b/drivers/gpu/drm/amd/amdgpu/mmhub_v9_4.h index 92404a8f66f3..90436efa92ef 100644 --- a/drivers/gpu/drm/amd/amdgpu/mmhub_v9_4.h +++ b/drivers/gpu/drm/amd/amdgpu/mmhub_v9_4.h @@ -24,5 +24,6 @@ #define __MMHUB_V9_4_H__ extern const struct amdgpu_mmhub_funcs mmhub_v9_4_funcs; +extern const struct amdgpu_mmhub_ras_funcs mmhub_v9_4_ras_funcs; #endif diff --git a/drivers/gpu/drm/amd/amdgpu/nbio_v6_1.c b/drivers/gpu/drm/amd/amdgpu/nbio_v6_1.c index 83ea063388fd..0d2d629e2d6a 100644 --- a/drivers/gpu/drm/amd/amdgpu/nbio_v6_1.c +++ b/drivers/gpu/drm/amd/amdgpu/nbio_v6_1.c @@ -31,6 +31,28 @@ #include "vega10_enum.h" #include <uapi/linux/kfd_ioctl.h> +#define smnPCIE_LC_CNTL 0x11140280 +#define smnPCIE_LC_CNTL3 0x111402d4 +#define smnPCIE_LC_CNTL6 0x111402ec +#define smnPCIE_LC_CNTL7 0x111402f0 +#define smnNBIF_MGCG_CTRL_LCLK 0x1013a05c +#define NBIF_MGCG_CTRL_LCLK__NBIF_MGCG_REG_DIS_LCLK_MASK 0x00001000L +#define RCC_BIF_STRAP3__STRAP_VLINK_ASPM_IDLE_TIMER_MASK 0x0000FFFFL +#define RCC_BIF_STRAP3__STRAP_VLINK_PM_L1_ENTRY_TIMER_MASK 0xFFFF0000L +#define smnRCC_EP_DEV0_0_EP_PCIE_TX_LTR_CNTL 0x10123530 +#define smnBIF_CFG_DEV0_EPF0_DEVICE_CNTL2 0x1014008c +#define smnBIF_CFG_DEV0_EPF0_PCIE_LTR_CAP 0x10140324 +#define smnPSWUSP0_PCIE_LC_CNTL2 0x111402c4 +#define smnRCC_BIF_STRAP2 0x10123488 +#define smnRCC_BIF_STRAP3 0x1012348c +#define smnRCC_BIF_STRAP5 0x10123494 +#define BIF_CFG_DEV0_EPF0_DEVICE_CNTL2__LTR_EN_MASK 0x0400L +#define RCC_BIF_STRAP5__STRAP_VLINK_LDN_ENTRY_TIMER_MASK 0x0000FFFFL +#define RCC_BIF_STRAP2__STRAP_LTR_IN_ASPML1_DIS_MASK 0x00004000L +#define RCC_BIF_STRAP3__STRAP_VLINK_ASPM_IDLE_TIMER__SHIFT 0x0 +#define RCC_BIF_STRAP3__STRAP_VLINK_PM_L1_ENTRY_TIMER__SHIFT 0x10 +#define RCC_BIF_STRAP5__STRAP_VLINK_LDN_ENTRY_TIMER__SHIFT 0x0 + static void nbio_v6_1_remap_hdp_registers(struct amdgpu_device *adev) { WREG32_SOC15(NBIO, 0, mmREMAP_HDP_MEM_FLUSH_CNTL, @@ -256,6 +278,111 @@ static void nbio_v6_1_init_registers(struct amdgpu_device *adev) WREG32_PCIE(smnPCIE_CI_CNTL, data); } +static void nbio_v6_1_program_ltr(struct amdgpu_device *adev) +{ + uint32_t def, data; + + WREG32_PCIE(smnRCC_EP_DEV0_0_EP_PCIE_TX_LTR_CNTL, 0x75EB); + + def = data = RREG32_PCIE(smnRCC_BIF_STRAP2); + data &= ~RCC_BIF_STRAP2__STRAP_LTR_IN_ASPML1_DIS_MASK; + if (def != data) + WREG32_PCIE(smnRCC_BIF_STRAP2, data); + + def = data = RREG32_PCIE(smnRCC_EP_DEV0_0_EP_PCIE_TX_LTR_CNTL); + data &= ~EP_PCIE_TX_LTR_CNTL__LTR_PRIV_MSG_DIS_IN_PM_NON_D0_MASK; + if (def != data) + WREG32_PCIE(smnRCC_EP_DEV0_0_EP_PCIE_TX_LTR_CNTL, data); + + def = data = RREG32_PCIE(smnBIF_CFG_DEV0_EPF0_DEVICE_CNTL2); + data |= BIF_CFG_DEV0_EPF0_DEVICE_CNTL2__LTR_EN_MASK; + if (def != data) + WREG32_PCIE(smnBIF_CFG_DEV0_EPF0_DEVICE_CNTL2, data); +} + +static void nbio_v6_1_program_aspm(struct amdgpu_device *adev) +{ + uint32_t def, data; + + def = data = RREG32_PCIE(smnPCIE_LC_CNTL); + data &= ~PCIE_LC_CNTL__LC_L1_INACTIVITY_MASK; + data &= ~PCIE_LC_CNTL__LC_L0S_INACTIVITY_MASK; + data |= PCIE_LC_CNTL__LC_PMI_TO_L1_DIS_MASK; + if (def != data) + WREG32_PCIE(smnPCIE_LC_CNTL, data); + + def = data = RREG32_PCIE(smnPCIE_LC_CNTL7); + data |= PCIE_LC_CNTL7__LC_NBIF_ASPM_INPUT_EN_MASK; + if (def != data) + WREG32_PCIE(smnPCIE_LC_CNTL7, data); + + def = data = RREG32_PCIE(smnNBIF_MGCG_CTRL_LCLK); + data |= NBIF_MGCG_CTRL_LCLK__NBIF_MGCG_REG_DIS_LCLK_MASK; + if (def != data) + WREG32_PCIE(smnNBIF_MGCG_CTRL_LCLK, data); + + def = data = RREG32_PCIE(smnPCIE_LC_CNTL3); + data |= PCIE_LC_CNTL3__LC_DSC_DONT_ENTER_L23_AFTER_PME_ACK_MASK; + if (def != data) + WREG32_PCIE(smnPCIE_LC_CNTL3, data); + + def = data = RREG32_PCIE(smnRCC_BIF_STRAP3); + data &= ~RCC_BIF_STRAP3__STRAP_VLINK_ASPM_IDLE_TIMER_MASK; + data &= ~RCC_BIF_STRAP3__STRAP_VLINK_PM_L1_ENTRY_TIMER_MASK; + if (def != data) + WREG32_PCIE(smnRCC_BIF_STRAP3, data); + + def = data = RREG32_PCIE(smnRCC_BIF_STRAP5); + data &= ~RCC_BIF_STRAP5__STRAP_VLINK_LDN_ENTRY_TIMER_MASK; + if (def != data) + WREG32_PCIE(smnRCC_BIF_STRAP5, data); + + def = data = RREG32_PCIE(smnBIF_CFG_DEV0_EPF0_DEVICE_CNTL2); + data &= ~BIF_CFG_DEV0_EPF0_DEVICE_CNTL2__LTR_EN_MASK; + if (def != data) + WREG32_PCIE(smnBIF_CFG_DEV0_EPF0_DEVICE_CNTL2, data); + + WREG32_PCIE(smnBIF_CFG_DEV0_EPF0_PCIE_LTR_CAP, 0x10011001); + + def = data = RREG32_PCIE(smnPSWUSP0_PCIE_LC_CNTL2); + data |= PSWUSP0_PCIE_LC_CNTL2__LC_ALLOW_PDWN_IN_L1_MASK | + PSWUSP0_PCIE_LC_CNTL2__LC_ALLOW_PDWN_IN_L23_MASK; + data &= ~PSWUSP0_PCIE_LC_CNTL2__LC_RCV_L0_TO_RCV_L0S_DIS_MASK; + if (def != data) + WREG32_PCIE(smnPSWUSP0_PCIE_LC_CNTL2, data); + + def = data = RREG32_PCIE(smnPCIE_LC_CNTL6); + data |= PCIE_LC_CNTL6__LC_L1_POWERDOWN_MASK | + PCIE_LC_CNTL6__LC_RX_L0S_STANDBY_EN_MASK; + if (def != data) + WREG32_PCIE(smnPCIE_LC_CNTL6, data); + + nbio_v6_1_program_ltr(adev); + + def = data = RREG32_PCIE(smnRCC_BIF_STRAP3); + data |= 0x5DE0 << RCC_BIF_STRAP3__STRAP_VLINK_ASPM_IDLE_TIMER__SHIFT; + data |= 0x0010 << RCC_BIF_STRAP3__STRAP_VLINK_PM_L1_ENTRY_TIMER__SHIFT; + if (def != data) + WREG32_PCIE(smnRCC_BIF_STRAP3, data); + + def = data = RREG32_PCIE(smnRCC_BIF_STRAP5); + data |= 0x0010 << RCC_BIF_STRAP5__STRAP_VLINK_LDN_ENTRY_TIMER__SHIFT; + if (def != data) + WREG32_PCIE(smnRCC_BIF_STRAP5, data); + + def = data = RREG32_PCIE(smnPCIE_LC_CNTL); + data &= ~PCIE_LC_CNTL__LC_L0S_INACTIVITY_MASK; + data |= 0x9 << PCIE_LC_CNTL__LC_L1_INACTIVITY__SHIFT; + data |= 0x1 << PCIE_LC_CNTL__LC_PMI_TO_L1_DIS__SHIFT; + if (def != data) + WREG32_PCIE(smnPCIE_LC_CNTL, data); + + def = data = RREG32_PCIE(smnPCIE_LC_CNTL3); + data &= ~PCIE_LC_CNTL3__LC_DSC_DONT_ENTER_L23_AFTER_PME_ACK_MASK; + if (def != data) + WREG32_PCIE(smnPCIE_LC_CNTL3, data); +} + const struct amdgpu_nbio_funcs nbio_v6_1_funcs = { .get_hdp_flush_req_offset = nbio_v6_1_get_hdp_flush_req_offset, .get_hdp_flush_done_offset = nbio_v6_1_get_hdp_flush_done_offset, @@ -274,4 +401,5 @@ const struct amdgpu_nbio_funcs nbio_v6_1_funcs = { .ih_control = nbio_v6_1_ih_control, .init_registers = nbio_v6_1_init_registers, .remap_hdp_registers = nbio_v6_1_remap_hdp_registers, + .program_aspm = nbio_v6_1_program_aspm, }; diff --git a/drivers/gpu/drm/amd/amdgpu/nbio_v7_4.c b/drivers/gpu/drm/amd/amdgpu/nbio_v7_4.c index 4bc1d1434065..cef929746739 100644 --- a/drivers/gpu/drm/amd/amdgpu/nbio_v7_4.c +++ b/drivers/gpu/drm/amd/amdgpu/nbio_v7_4.c @@ -31,7 +31,26 @@ #include "ivsrcid/nbio/irqsrcs_nbif_7_4.h" #include <uapi/linux/kfd_ioctl.h> +#define smnPCIE_LC_CNTL 0x11140280 +#define smnPCIE_LC_CNTL3 0x111402d4 +#define smnPCIE_LC_CNTL6 0x111402ec +#define smnPCIE_LC_CNTL7 0x111402f0 #define smnNBIF_MGCG_CTRL_LCLK 0x1013a21c +#define smnRCC_BIF_STRAP3 0x1012348c +#define RCC_BIF_STRAP3__STRAP_VLINK_ASPM_IDLE_TIMER_MASK 0x0000FFFFL +#define RCC_BIF_STRAP3__STRAP_VLINK_PM_L1_ENTRY_TIMER_MASK 0xFFFF0000L +#define smnRCC_BIF_STRAP5 0x10123494 +#define RCC_BIF_STRAP5__STRAP_VLINK_LDN_ENTRY_TIMER_MASK 0x0000FFFFL +#define smnBIF_CFG_DEV0_EPF0_DEVICE_CNTL2 0x1014008c +#define BIF_CFG_DEV0_EPF0_DEVICE_CNTL2__LTR_EN_MASK 0x0400L +#define smnBIF_CFG_DEV0_EPF0_PCIE_LTR_CAP 0x10140324 +#define smnPSWUSP0_PCIE_LC_CNTL2 0x111402c4 +#define smnRCC_EP_DEV0_0_EP_PCIE_TX_LTR_CNTL 0x10123538 +#define smnRCC_BIF_STRAP2 0x10123488 +#define RCC_BIF_STRAP2__STRAP_LTR_IN_ASPML1_DIS_MASK 0x00004000L +#define RCC_BIF_STRAP3__STRAP_VLINK_ASPM_IDLE_TIMER__SHIFT 0x0 +#define RCC_BIF_STRAP3__STRAP_VLINK_PM_L1_ENTRY_TIMER__SHIFT 0x10 +#define RCC_BIF_STRAP5__STRAP_VLINK_LDN_ENTRY_TIMER__SHIFT 0x0 /* * These are nbio v7_4_1 registers mask. Temporarily define these here since @@ -52,6 +71,20 @@ #define BIF_MMSCH1_DOORBELL_RANGE__OFFSET_MASK 0x00000FFCL #define BIF_MMSCH1_DOORBELL_RANGE__SIZE_MASK 0x001F0000L +#define BIF_MMSCH1_DOORBELL_RANGE__OFFSET_MASK 0x00000FFCL +#define BIF_MMSCH1_DOORBELL_RANGE__SIZE_MASK 0x001F0000L + +#define mmBIF_MMSCH1_DOORBELL_RANGE_ALDE 0x01d8 +#define mmBIF_MMSCH1_DOORBELL_RANGE_ALDE_BASE_IDX 2 +//BIF_MMSCH1_DOORBELL_ALDE_RANGE +#define BIF_MMSCH1_DOORBELL_RANGE_ALDE__OFFSET__SHIFT 0x2 +#define BIF_MMSCH1_DOORBELL_RANGE_ALDE__SIZE__SHIFT 0x10 +#define BIF_MMSCH1_DOORBELL_RANGE_ALDE__OFFSET_MASK 0x00000FFCL +#define BIF_MMSCH1_DOORBELL_RANGE_ALDE__SIZE_MASK 0x001F0000L + +#define mmRCC_DEV0_EPF0_STRAP0_ALDE 0x0015 +#define mmRCC_DEV0_EPF0_STRAP0_ALDE_BASE_IDX 2 + static void nbio_v7_4_query_ras_error_count(struct amdgpu_device *adev, void *ras_error_status); @@ -65,7 +98,12 @@ static void nbio_v7_4_remap_hdp_registers(struct amdgpu_device *adev) static u32 nbio_v7_4_get_rev_id(struct amdgpu_device *adev) { - u32 tmp = RREG32_SOC15(NBIO, 0, mmRCC_DEV0_EPF0_STRAP0); + u32 tmp; + + if (adev->asic_type == CHIP_ALDEBARAN) + tmp = RREG32_SOC15(NBIO, 0, mmRCC_DEV0_EPF0_STRAP0_ALDE); + else + tmp = RREG32_SOC15(NBIO, 0, mmRCC_DEV0_EPF0_STRAP0); tmp &= RCC_DEV0_EPF0_STRAP0__STRAP_ATI_REV_ID_DEV0_F0_MASK; tmp >>= RCC_DEV0_EPF0_STRAP0__STRAP_ATI_REV_ID_DEV0_F0__SHIFT; @@ -92,10 +130,10 @@ static void nbio_v7_4_sdma_doorbell_range(struct amdgpu_device *adev, int instan { u32 reg, doorbell_range; - if (instance < 2) + if (instance < 2) { reg = instance + SOC15_REG_OFFSET(NBIO, 0, mmBIF_SDMA0_DOORBELL_RANGE); - else + } else { /* * These registers address of SDMA2~7 is not consecutive * from SDMA0~1. Need plus 4 dwords offset. @@ -103,9 +141,19 @@ static void nbio_v7_4_sdma_doorbell_range(struct amdgpu_device *adev, int instan * BIF_SDMA0_DOORBELL_RANGE: 0x3bc0 * BIF_SDMA1_DOORBELL_RANGE: 0x3bc4 * BIF_SDMA2_DOORBELL_RANGE: 0x3bd8 ++ * BIF_SDMA4_DOORBELL_RANGE: ++ * ARCTURUS: 0x3be0 ++ * ALDEBARAN: 0x3be4 */ - reg = instance + 0x4 + - SOC15_REG_OFFSET(NBIO, 0, mmBIF_SDMA0_DOORBELL_RANGE); + if (adev->asic_type == CHIP_ALDEBARAN && instance == 4) + reg = instance + 0x4 + 0x1 + + SOC15_REG_OFFSET(NBIO, 0, + mmBIF_SDMA0_DOORBELL_RANGE); + else + reg = instance + 0x4 + + SOC15_REG_OFFSET(NBIO, 0, + mmBIF_SDMA0_DOORBELL_RANGE); + } doorbell_range = RREG32(reg); @@ -124,9 +172,12 @@ static void nbio_v7_4_vcn_doorbell_range(struct amdgpu_device *adev, bool use_do u32 reg; u32 doorbell_range; - if (instance) - reg = SOC15_REG_OFFSET(NBIO, 0, mmBIF_MMSCH1_DOORBELL_RANGE); - else + if (instance) { + if (adev->asic_type == CHIP_ALDEBARAN) + reg = SOC15_REG_OFFSET(NBIO, 0, mmBIF_MMSCH1_DOORBELL_RANGE_ALDE); + else + reg = SOC15_REG_OFFSET(NBIO, 0, mmBIF_MMSCH1_DOORBELL_RANGE); + } else reg = SOC15_REG_OFFSET(NBIO, 0, mmBIF_MMSCH0_DOORBELL_RANGE); doorbell_range = RREG32(reg); @@ -525,6 +576,121 @@ static void nbio_v7_4_enable_doorbell_interrupt(struct amdgpu_device *adev, DOORBELL_INTERRUPT_DISABLE, enable ? 0 : 1); } +const struct amdgpu_nbio_ras_funcs nbio_v7_4_ras_funcs = { + .handle_ras_controller_intr_no_bifring = nbio_v7_4_handle_ras_controller_intr_no_bifring, + .handle_ras_err_event_athub_intr_no_bifring = nbio_v7_4_handle_ras_err_event_athub_intr_no_bifring, + .init_ras_controller_interrupt = nbio_v7_4_init_ras_controller_interrupt, + .init_ras_err_event_athub_interrupt = nbio_v7_4_init_ras_err_event_athub_interrupt, + .query_ras_error_count = nbio_v7_4_query_ras_error_count, + .ras_late_init = amdgpu_nbio_ras_late_init, + .ras_fini = amdgpu_nbio_ras_fini, +}; + +static void nbio_v7_4_program_ltr(struct amdgpu_device *adev) +{ + uint32_t def, data; + + WREG32_PCIE(smnRCC_EP_DEV0_0_EP_PCIE_TX_LTR_CNTL, 0x75EB); + + def = data = RREG32_PCIE(smnRCC_BIF_STRAP2); + data &= ~RCC_BIF_STRAP2__STRAP_LTR_IN_ASPML1_DIS_MASK; + if (def != data) + WREG32_PCIE(smnRCC_BIF_STRAP2, data); + + def = data = RREG32_PCIE(smnRCC_EP_DEV0_0_EP_PCIE_TX_LTR_CNTL); + data &= ~EP_PCIE_TX_LTR_CNTL__LTR_PRIV_MSG_DIS_IN_PM_NON_D0_MASK; + if (def != data) + WREG32_PCIE(smnRCC_EP_DEV0_0_EP_PCIE_TX_LTR_CNTL, data); + + def = data = RREG32_PCIE(smnBIF_CFG_DEV0_EPF0_DEVICE_CNTL2); + data |= BIF_CFG_DEV0_EPF0_DEVICE_CNTL2__LTR_EN_MASK; + if (def != data) + WREG32_PCIE(smnBIF_CFG_DEV0_EPF0_DEVICE_CNTL2, data); +} + +static void nbio_v7_4_program_aspm(struct amdgpu_device *adev) +{ + uint32_t def, data; + + def = data = RREG32_PCIE(smnPCIE_LC_CNTL); + data &= ~PCIE_LC_CNTL__LC_L1_INACTIVITY_MASK; + data &= ~PCIE_LC_CNTL__LC_L0S_INACTIVITY_MASK; + data |= PCIE_LC_CNTL__LC_PMI_TO_L1_DIS_MASK; + if (def != data) + WREG32_PCIE(smnPCIE_LC_CNTL, data); + + def = data = RREG32_PCIE(smnPCIE_LC_CNTL7); + data |= PCIE_LC_CNTL7__LC_NBIF_ASPM_INPUT_EN_MASK; + if (def != data) + WREG32_PCIE(smnPCIE_LC_CNTL7, data); + + def = data = RREG32_PCIE(smnNBIF_MGCG_CTRL_LCLK); + data |= NBIF_MGCG_CTRL_LCLK__NBIF_MGCG_REG_DIS_LCLK_MASK; + if (def != data) + WREG32_PCIE(smnNBIF_MGCG_CTRL_LCLK, data); + + def = data = RREG32_PCIE(smnPCIE_LC_CNTL3); + data |= PCIE_LC_CNTL3__LC_DSC_DONT_ENTER_L23_AFTER_PME_ACK_MASK; + if (def != data) + WREG32_PCIE(smnPCIE_LC_CNTL3, data); + + def = data = RREG32_PCIE(smnRCC_BIF_STRAP3); + data &= ~RCC_BIF_STRAP3__STRAP_VLINK_ASPM_IDLE_TIMER_MASK; + data &= ~RCC_BIF_STRAP3__STRAP_VLINK_PM_L1_ENTRY_TIMER_MASK; + if (def != data) + WREG32_PCIE(smnRCC_BIF_STRAP3, data); + + def = data = RREG32_PCIE(smnRCC_BIF_STRAP5); + data &= ~RCC_BIF_STRAP5__STRAP_VLINK_LDN_ENTRY_TIMER_MASK; + if (def != data) + WREG32_PCIE(smnRCC_BIF_STRAP5, data); + + def = data = RREG32_PCIE(smnBIF_CFG_DEV0_EPF0_DEVICE_CNTL2); + data &= ~BIF_CFG_DEV0_EPF0_DEVICE_CNTL2__LTR_EN_MASK; + if (def != data) + WREG32_PCIE(smnBIF_CFG_DEV0_EPF0_DEVICE_CNTL2, data); + + WREG32_PCIE(smnBIF_CFG_DEV0_EPF0_PCIE_LTR_CAP, 0x10011001); + + def = data = RREG32_PCIE(smnPSWUSP0_PCIE_LC_CNTL2); + data |= PSWUSP0_PCIE_LC_CNTL2__LC_ALLOW_PDWN_IN_L1_MASK | + PSWUSP0_PCIE_LC_CNTL2__LC_ALLOW_PDWN_IN_L23_MASK; + data &= ~PSWUSP0_PCIE_LC_CNTL2__LC_RCV_L0_TO_RCV_L0S_DIS_MASK; + if (def != data) + WREG32_PCIE(smnPSWUSP0_PCIE_LC_CNTL2, data); + + def = data = RREG32_PCIE(smnPCIE_LC_CNTL6); + data |= PCIE_LC_CNTL6__LC_L1_POWERDOWN_MASK | + PCIE_LC_CNTL6__LC_RX_L0S_STANDBY_EN_MASK; + if (def != data) + WREG32_PCIE(smnPCIE_LC_CNTL6, data); + + nbio_v7_4_program_ltr(adev); + + def = data = RREG32_PCIE(smnRCC_BIF_STRAP3); + data |= 0x5DE0 << RCC_BIF_STRAP3__STRAP_VLINK_ASPM_IDLE_TIMER__SHIFT; + data |= 0x0010 << RCC_BIF_STRAP3__STRAP_VLINK_PM_L1_ENTRY_TIMER__SHIFT; + if (def != data) + WREG32_PCIE(smnRCC_BIF_STRAP3, data); + + def = data = RREG32_PCIE(smnRCC_BIF_STRAP5); + data |= 0x0010 << RCC_BIF_STRAP5__STRAP_VLINK_LDN_ENTRY_TIMER__SHIFT; + if (def != data) + WREG32_PCIE(smnRCC_BIF_STRAP5, data); + + def = data = RREG32_PCIE(smnPCIE_LC_CNTL); + data &= ~PCIE_LC_CNTL__LC_L0S_INACTIVITY_MASK; + data |= 0x9 << PCIE_LC_CNTL__LC_L1_INACTIVITY__SHIFT; + data |= 0x1 << PCIE_LC_CNTL__LC_PMI_TO_L1_DIS__SHIFT; + if (def != data) + WREG32_PCIE(smnPCIE_LC_CNTL, data); + + def = data = RREG32_PCIE(smnPCIE_LC_CNTL3); + data &= ~PCIE_LC_CNTL3__LC_DSC_DONT_ENTER_L23_AFTER_PME_ACK_MASK; + if (def != data) + WREG32_PCIE(smnPCIE_LC_CNTL3, data); +} + const struct amdgpu_nbio_funcs nbio_v7_4_funcs = { .get_hdp_flush_req_offset = nbio_v7_4_get_hdp_flush_req_offset, .get_hdp_flush_done_offset = nbio_v7_4_get_hdp_flush_done_offset, @@ -545,10 +711,5 @@ const struct amdgpu_nbio_funcs nbio_v7_4_funcs = { .ih_control = nbio_v7_4_ih_control, .init_registers = nbio_v7_4_init_registers, .remap_hdp_registers = nbio_v7_4_remap_hdp_registers, - .handle_ras_controller_intr_no_bifring = nbio_v7_4_handle_ras_controller_intr_no_bifring, - .handle_ras_err_event_athub_intr_no_bifring = nbio_v7_4_handle_ras_err_event_athub_intr_no_bifring, - .init_ras_controller_interrupt = nbio_v7_4_init_ras_controller_interrupt, - .init_ras_err_event_athub_interrupt = nbio_v7_4_init_ras_err_event_athub_interrupt, - .query_ras_error_count = nbio_v7_4_query_ras_error_count, - .ras_late_init = amdgpu_nbio_ras_late_init, + .program_aspm = nbio_v7_4_program_aspm, }; diff --git a/drivers/gpu/drm/amd/amdgpu/nbio_v7_4.h b/drivers/gpu/drm/amd/amdgpu/nbio_v7_4.h index b1ac82872752..b8216581ec8d 100644 --- a/drivers/gpu/drm/amd/amdgpu/nbio_v7_4.h +++ b/drivers/gpu/drm/amd/amdgpu/nbio_v7_4.h @@ -28,5 +28,6 @@ extern const struct nbio_hdp_flush_reg nbio_v7_4_hdp_flush_reg; extern const struct amdgpu_nbio_funcs nbio_v7_4_funcs; +extern const struct amdgpu_nbio_ras_funcs nbio_v7_4_ras_funcs; #endif diff --git a/drivers/gpu/drm/amd/amdgpu/nv.c b/drivers/gpu/drm/amd/amdgpu/nv.c index c625c5d8ed89..d54af7f8801b 100644 --- a/drivers/gpu/drm/amd/amdgpu/nv.c +++ b/drivers/gpu/drm/amd/amdgpu/nv.c @@ -25,6 +25,8 @@ #include <linux/module.h> #include <linux/pci.h> +#include <drm/amdgpu_drm.h> + #include "amdgpu.h" #include "amdgpu_atombios.h" #include "amdgpu_ih.h" @@ -32,7 +34,6 @@ #include "amdgpu_vce.h" #include "amdgpu_ucode.h" #include "amdgpu_psp.h" -#include "amdgpu_smu.h" #include "atom.h" #include "amd_pcie.h" @@ -65,6 +66,184 @@ static const struct amd_ip_funcs nv_common_ip_funcs; +/* Navi */ +static const struct amdgpu_video_codec_info nv_video_codecs_encode_array[] = +{ + { + .codec_type = AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_MPEG4_AVC, + .max_width = 4096, + .max_height = 2304, + .max_pixels_per_frame = 4096 * 2304, + .max_level = 0, + }, + { + .codec_type = AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_HEVC, + .max_width = 4096, + .max_height = 2304, + .max_pixels_per_frame = 4096 * 2304, + .max_level = 0, + }, +}; + +static const struct amdgpu_video_codecs nv_video_codecs_encode = +{ + .codec_count = ARRAY_SIZE(nv_video_codecs_encode_array), + .codec_array = nv_video_codecs_encode_array, +}; + +/* Navi1x */ +static const struct amdgpu_video_codec_info nv_video_codecs_decode_array[] = +{ + { + .codec_type = AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_MPEG2, + .max_width = 4096, + .max_height = 4096, + .max_pixels_per_frame = 4096 * 4096, + .max_level = 3, + }, + { + .codec_type = AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_MPEG4, + .max_width = 4096, + .max_height = 4096, + .max_pixels_per_frame = 4096 * 4096, + .max_level = 5, + }, + { + .codec_type = AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_MPEG4_AVC, + .max_width = 4096, + .max_height = 4096, + .max_pixels_per_frame = 4096 * 4096, + .max_level = 52, + }, + { + .codec_type = AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_VC1, + .max_width = 4096, + .max_height = 4096, + .max_pixels_per_frame = 4096 * 4096, + .max_level = 4, + }, + { + .codec_type = AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_HEVC, + .max_width = 8192, + .max_height = 4352, + .max_pixels_per_frame = 8192 * 4352, + .max_level = 186, + }, + { + .codec_type = AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_JPEG, + .max_width = 4096, + .max_height = 4096, + .max_pixels_per_frame = 4096 * 4096, + .max_level = 0, + }, + { + .codec_type = AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_VP9, + .max_width = 8192, + .max_height = 4352, + .max_pixels_per_frame = 8192 * 4352, + .max_level = 0, + }, +}; + +static const struct amdgpu_video_codecs nv_video_codecs_decode = +{ + .codec_count = ARRAY_SIZE(nv_video_codecs_decode_array), + .codec_array = nv_video_codecs_decode_array, +}; + +/* Sienna Cichlid */ +static const struct amdgpu_video_codec_info sc_video_codecs_decode_array[] = +{ + { + .codec_type = AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_MPEG2, + .max_width = 4096, + .max_height = 4096, + .max_pixels_per_frame = 4096 * 4096, + .max_level = 3, + }, + { + .codec_type = AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_MPEG4, + .max_width = 4096, + .max_height = 4096, + .max_pixels_per_frame = 4096 * 4096, + .max_level = 5, + }, + { + .codec_type = AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_MPEG4_AVC, + .max_width = 4096, + .max_height = 4096, + .max_pixels_per_frame = 4096 * 4096, + .max_level = 52, + }, + { + .codec_type = AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_VC1, + .max_width = 4096, + .max_height = 4096, + .max_pixels_per_frame = 4096 * 4096, + .max_level = 4, + }, + { + .codec_type = AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_HEVC, + .max_width = 8192, + .max_height = 4352, + .max_pixels_per_frame = 8192 * 4352, + .max_level = 186, + }, + { + .codec_type = AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_JPEG, + .max_width = 4096, + .max_height = 4096, + .max_pixels_per_frame = 4096 * 4096, + .max_level = 0, + }, + { + .codec_type = AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_VP9, + .max_width = 8192, + .max_height = 4352, + .max_pixels_per_frame = 8192 * 4352, + .max_level = 0, + }, + { + .codec_type = AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_AV1, + .max_width = 8192, + .max_height = 4352, + .max_pixels_per_frame = 8192 * 4352, + .max_level = 0, + }, +}; + +static const struct amdgpu_video_codecs sc_video_codecs_decode = +{ + .codec_count = ARRAY_SIZE(sc_video_codecs_decode_array), + .codec_array = sc_video_codecs_decode_array, +}; + +static int nv_query_video_codecs(struct amdgpu_device *adev, bool encode, + const struct amdgpu_video_codecs **codecs) +{ + switch (adev->asic_type) { + case CHIP_SIENNA_CICHLID: + case CHIP_NAVY_FLOUNDER: + case CHIP_DIMGREY_CAVEFISH: + case CHIP_VANGOGH: + if (encode) + *codecs = &nv_video_codecs_encode; + else + *codecs = &sc_video_codecs_decode; + return 0; + case CHIP_NAVI10: + case CHIP_NAVI14: + case CHIP_NAVI12: + if (encode) + *codecs = &nv_video_codecs_encode; + else + *codecs = &nv_video_codecs_decode; + return 0; + default: + return -EINVAL; + } +} + /* * Indirect registers accessor */ @@ -304,44 +483,6 @@ static int nv_read_register(struct amdgpu_device *adev, u32 se_num, return -EINVAL; } -static int nv_asic_mode1_reset(struct amdgpu_device *adev) -{ - u32 i; - int ret = 0; - - amdgpu_atombios_scratch_regs_engine_hung(adev, true); - - /* disable BM */ - pci_clear_master(adev->pdev); - - amdgpu_device_cache_pci_state(adev->pdev); - - if (amdgpu_dpm_is_mode1_reset_supported(adev)) { - dev_info(adev->dev, "GPU smu mode1 reset\n"); - ret = amdgpu_dpm_mode1_reset(adev); - } else { - dev_info(adev->dev, "GPU psp mode1 reset\n"); - ret = psp_gpu_reset(adev); - } - - if (ret) - dev_err(adev->dev, "GPU mode1 reset failed\n"); - amdgpu_device_load_pci_state(adev->pdev); - - /* wait for asic to come out of reset */ - for (i = 0; i < adev->usec_timeout; i++) { - u32 memsize = adev->nbio.funcs->get_memsize(adev); - - if (memsize != 0xffffffff) - break; - udelay(1); - } - - amdgpu_atombios_scratch_regs_engine_hung(adev, false); - - return ret; -} - static int nv_asic_mode2_reset(struct amdgpu_device *adev) { u32 i; @@ -374,21 +515,9 @@ static int nv_asic_mode2_reset(struct amdgpu_device *adev) return ret; } -static bool nv_asic_supports_baco(struct amdgpu_device *adev) -{ - struct smu_context *smu = &adev->smu; - - if (smu_baco_is_support(smu)) - return true; - else - return false; -} - static enum amd_reset_method nv_asic_reset_method(struct amdgpu_device *adev) { - struct smu_context *smu = &adev->smu; - if (amdgpu_reset_method == AMD_RESET_METHOD_MODE1 || amdgpu_reset_method == AMD_RESET_METHOD_MODE2 || amdgpu_reset_method == AMD_RESET_METHOD_BACO || @@ -407,7 +536,7 @@ nv_asic_reset_method(struct amdgpu_device *adev) case CHIP_DIMGREY_CAVEFISH: return AMD_RESET_METHOD_MODE1; default: - if (smu_baco_is_support(smu)) + if (amdgpu_dpm_is_baco_supported(adev)) return AMD_RESET_METHOD_BACO; else return AMD_RESET_METHOD_MODE1; @@ -417,11 +546,6 @@ nv_asic_reset_method(struct amdgpu_device *adev) static int nv_asic_reset(struct amdgpu_device *adev) { int ret = 0; - struct smu_context *smu = &adev->smu; - - /* skip reset on vangogh for now */ - if (adev->asic_type == CHIP_VANGOGH) - return 0; switch (nv_asic_reset_method(adev)) { case AMD_RESET_METHOD_PCI: @@ -430,13 +554,7 @@ static int nv_asic_reset(struct amdgpu_device *adev) break; case AMD_RESET_METHOD_BACO: dev_info(adev->dev, "BACO reset\n"); - - ret = smu_baco_enter(smu); - if (ret) - return ret; - ret = smu_baco_exit(smu); - if (ret) - return ret; + ret = amdgpu_dpm_baco_reset(adev); break; case AMD_RESET_METHOD_MODE2: dev_info(adev->dev, "MODE2 reset\n"); @@ -444,7 +562,7 @@ static int nv_asic_reset(struct amdgpu_device *adev) break; default: dev_info(adev->dev, "MODE1 reset\n"); - ret = nv_asic_mode1_reset(adev); + ret = amdgpu_device_mode1_reset(adev); break; } @@ -483,8 +601,7 @@ static void nv_program_aspm(struct amdgpu_device *adev) if (amdgpu_aspm != 1) return; - if ((adev->asic_type >= CHIP_SIENNA_CICHLID) && - !(adev->flags & AMD_IS_APU) && + if (!(adev->flags & AMD_IS_APU) && (adev->nbio.funcs->program_aspm)) adev->nbio.funcs->program_aspm(adev); @@ -816,12 +933,7 @@ static int nv_update_umd_stable_pstate(struct amdgpu_device *adev, if (adev->gfx.funcs->update_perfmon_mgcg) adev->gfx.funcs->update_perfmon_mgcg(adev, !enter); - /* - * The ASPM function is not fully enabled and verified on - * Navi yet. Temporarily skip this until ASPM enabled. - */ - if ((adev->asic_type >= CHIP_SIENNA_CICHLID) && - !(adev->flags & AMD_IS_APU) && + if (!(adev->flags & AMD_IS_APU) && (adev->nbio.funcs->enable_aspm)) adev->nbio.funcs->enable_aspm(adev, !enter); @@ -844,9 +956,10 @@ static const struct amdgpu_asic_funcs nv_asic_funcs = .need_full_reset = &nv_need_full_reset, .need_reset_on_init = &nv_need_reset_on_init, .get_pcie_replay_count = &nv_get_pcie_replay_count, - .supports_baco = &nv_asic_supports_baco, + .supports_baco = &amdgpu_dpm_is_baco_supported, .pre_asic_init = &nv_pre_asic_init, .update_umd_stable_pstate = &nv_update_umd_stable_pstate, + .query_video_codecs = &nv_query_video_codecs, }; static int nv_common_early_init(void *handle) diff --git a/drivers/gpu/drm/amd/amdgpu/psp_gfx_if.h b/drivers/gpu/drm/amd/amdgpu/psp_gfx_if.h index 3ba7bdfde65d..96064c343163 100644 --- a/drivers/gpu/drm/amd/amdgpu/psp_gfx_if.h +++ b/drivers/gpu/drm/amd/amdgpu/psp_gfx_if.h @@ -102,6 +102,21 @@ enum psp_gfx_cmd_id /* IDs upto 0x1F are reserved for older programs (Raven, Vega 10/12/20) */ GFX_CMD_ID_LOAD_TOC = 0x00000020, /* Load TOC and obtain TMR size */ GFX_CMD_ID_AUTOLOAD_RLC = 0x00000021, /* Indicates all graphics fw loaded, start RLC autoload */ + GFX_CMD_ID_BOOT_CFG = 0x00000022, /* Boot Config */ +}; + +/* PSP boot config sub-commands */ +enum psp_gfx_boot_config_cmd +{ + BOOTCFG_CMD_SET = 1, /* Set boot configuration settings */ + BOOTCFG_CMD_GET = 2, /* Get boot configuration settings */ + BOOTCFG_CMD_INVALIDATE = 3 /* Reset current boot configuration settings to VBIOS defaults */ +}; + +/* PSP boot config bitmask values */ +enum psp_gfx_boot_config +{ + BOOT_CONFIG_GECC = 0x1, }; /* Command to load Trusted Application binary into PSP OS. */ @@ -170,10 +185,19 @@ struct psp_gfx_cmd_setup_tmr uint32_t buf_phy_addr_lo; /* bits [31:0] of GPU Virtual address of TMR buffer (must be 4 KB aligned) */ uint32_t buf_phy_addr_hi; /* bits [63:32] of GPU Virtual address of TMR buffer */ uint32_t buf_size; /* buffer size in bytes (must be multiple of 4 KB) */ + union { + struct { + uint32_t sriov_enabled:1; /* whether the device runs under SR-IOV*/ + uint32_t virt_phy_addr:1; /* driver passes both virtual and physical address to PSP*/ + uint32_t reserved:30; + } bitfield; + uint32_t tmr_flags; + }; + uint32_t system_phy_addr_lo; /* bits [31:0] of system physical address of TMR buffer (must be 4 KB aligned) */ + uint32_t system_phy_addr_hi; /* bits [63:32] of system physical address of TMR buffer */ }; - /* FW types for GFX_CMD_ID_LOAD_IP_FW command. Limit 31. */ enum psp_gfx_fw_type { GFX_FW_TYPE_NONE = 0, /* */ @@ -235,6 +259,7 @@ enum psp_gfx_fw_type { GFX_FW_TYPE_SDMA6 = 56, /* SDMA6 MI */ GFX_FW_TYPE_SDMA7 = 57, /* SDMA7 MI */ GFX_FW_TYPE_VCN1 = 58, /* VCN1 MI */ + GFX_FW_TYPE_REG_LIST = 67, /* REG_LIST MI */ GFX_FW_TYPE_MAX }; @@ -272,6 +297,15 @@ struct psp_gfx_cmd_load_toc uint32_t toc_size; /* FW buffer size in bytes */ }; +/* Dynamic boot configuration */ +struct psp_gfx_cmd_boot_cfg +{ + uint32_t timestamp; /* calendar time as number of seconds */ + enum psp_gfx_boot_config_cmd sub_cmd; /* sub-command indicating how to process command data */ + uint32_t boot_config; /* dynamic boot configuration bitmask */ + uint32_t boot_config_valid; /* dynamic boot configuration valid bits bitmask */ +}; + /* All GFX ring buffer commands. */ union psp_gfx_commands { @@ -284,6 +318,7 @@ union psp_gfx_commands struct psp_gfx_cmd_reg_prog cmd_setup_reg_prog; struct psp_gfx_cmd_setup_tmr cmd_setup_vmr; struct psp_gfx_cmd_load_toc cmd_load_toc; + struct psp_gfx_cmd_boot_cfg boot_cfg; }; struct psp_gfx_uresp_reserved diff --git a/drivers/gpu/drm/amd/amdgpu/psp_v11_0.c b/drivers/gpu/drm/amd/amdgpu/psp_v11_0.c index c325d6f53a71..589410c32d09 100644 --- a/drivers/gpu/drm/amd/amdgpu/psp_v11_0.c +++ b/drivers/gpu/drm/amd/amdgpu/psp_v11_0.c @@ -598,7 +598,7 @@ static int psp_v11_0_memory_training_send_msg(struct psp_context *psp, int msg) } /* - * save and restore proces + * save and restore process */ static int psp_v11_0_memory_training(struct psp_context *psp, uint32_t ops) { @@ -661,9 +661,9 @@ static int psp_v11_0_memory_training(struct psp_context *psp, uint32_t ops) if (ops & PSP_MEM_TRAIN_SEND_LONG_MSG) { /* - * Long traing will encroach certain mount of bottom VRAM, - * saving the content of this bottom VRAM to system memory - * before training, and restoring it after training to avoid + * Long training will encroach a certain amount on the bottom of VRAM; + * save the content from the bottom of VRAM to system memory + * before training, and restore it after training to avoid * VRAM corruption. */ sz = GDDR6_MEM_TRAINING_ENCROACHED_SIZE; diff --git a/drivers/gpu/drm/amd/amdgpu/psp_v13_0.c b/drivers/gpu/drm/amd/amdgpu/psp_v13_0.c new file mode 100644 index 000000000000..fcdce46445d6 --- /dev/null +++ b/drivers/gpu/drm/amd/amdgpu/psp_v13_0.c @@ -0,0 +1,378 @@ +/* + * Copyright 2020 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + */ +#include "amdgpu.h" +#include "amdgpu_psp.h" +#include "amdgpu_ucode.h" +#include "soc15_common.h" +#include "psp_v13_0.h" + +#include "mp/mp_13_0_2_offset.h" +#include "mp/mp_13_0_2_sh_mask.h" + +MODULE_FIRMWARE("amdgpu/aldebaran_sos.bin"); +MODULE_FIRMWARE("amdgpu/aldebaran_ta.bin"); + +static int psp_v13_0_init_microcode(struct psp_context *psp) +{ + struct amdgpu_device *adev = psp->adev; + const char *chip_name; + int err = 0; + + switch (adev->asic_type) { + case CHIP_ALDEBARAN: + chip_name = "aldebaran"; + break; + default: + BUG(); + } + + err = psp_init_sos_microcode(psp, chip_name); + if (err) + return err; + + err = psp_init_ta_microcode(&adev->psp, chip_name); + + return err; +} + +static bool psp_v13_0_is_sos_alive(struct psp_context *psp) +{ + struct amdgpu_device *adev = psp->adev; + uint32_t sol_reg; + + sol_reg = RREG32_SOC15(MP0, 0, regMP0_SMN_C2PMSG_81); + + return sol_reg != 0x0; +} + +static int psp_v13_0_wait_for_bootloader(struct psp_context *psp) +{ + struct amdgpu_device *adev = psp->adev; + + int ret; + int retry_loop; + + for (retry_loop = 0; retry_loop < 10; retry_loop++) { + /* Wait for bootloader to signify that is + ready having bit 31 of C2PMSG_35 set to 1 */ + ret = psp_wait_for(psp, + SOC15_REG_OFFSET(MP0, 0, regMP0_SMN_C2PMSG_35), + 0x80000000, + 0x80000000, + false); + + if (ret == 0) + return 0; + } + + return ret; +} + +static int psp_v13_0_bootloader_load_kdb(struct psp_context *psp) +{ + int ret; + uint32_t psp_gfxdrv_command_reg = 0; + struct amdgpu_device *adev = psp->adev; + + /* Check tOS sign of life register to confirm sys driver and sOS + * are already been loaded. + */ + if (psp_v13_0_is_sos_alive(psp)) + return 0; + + ret = psp_v13_0_wait_for_bootloader(psp); + if (ret) + return ret; + + memset(psp->fw_pri_buf, 0, PSP_1_MEG); + + /* Copy PSP KDB binary to memory */ + memcpy(psp->fw_pri_buf, psp->kdb_start_addr, psp->kdb_bin_size); + + /* Provide the PSP KDB to bootloader */ + WREG32_SOC15(MP0, 0, regMP0_SMN_C2PMSG_36, + (uint32_t)(psp->fw_pri_mc_addr >> 20)); + psp_gfxdrv_command_reg = PSP_BL__LOAD_KEY_DATABASE; + WREG32_SOC15(MP0, 0, regMP0_SMN_C2PMSG_35, + psp_gfxdrv_command_reg); + + ret = psp_v13_0_wait_for_bootloader(psp); + + return ret; +} + +static int psp_v13_0_bootloader_load_sysdrv(struct psp_context *psp) +{ + int ret; + uint32_t psp_gfxdrv_command_reg = 0; + struct amdgpu_device *adev = psp->adev; + + /* Check sOS sign of life register to confirm sys driver and sOS + * are already been loaded. + */ + if (psp_v13_0_is_sos_alive(psp)) + return 0; + + ret = psp_v13_0_wait_for_bootloader(psp); + if (ret) + return ret; + + memset(psp->fw_pri_buf, 0, PSP_1_MEG); + + /* Copy PSP System Driver binary to memory */ + memcpy(psp->fw_pri_buf, psp->sys_start_addr, psp->sys_bin_size); + + /* Provide the sys driver to bootloader */ + WREG32_SOC15(MP0, 0, regMP0_SMN_C2PMSG_36, + (uint32_t)(psp->fw_pri_mc_addr >> 20)); + psp_gfxdrv_command_reg = PSP_BL__LOAD_SYSDRV; + WREG32_SOC15(MP0, 0, regMP0_SMN_C2PMSG_35, + psp_gfxdrv_command_reg); + + /* there might be handshake issue with hardware which needs delay */ + mdelay(20); + + ret = psp_v13_0_wait_for_bootloader(psp); + + return ret; +} + +static int psp_v13_0_bootloader_load_sos(struct psp_context *psp) +{ + int ret; + unsigned int psp_gfxdrv_command_reg = 0; + struct amdgpu_device *adev = psp->adev; + + /* Check sOS sign of life register to confirm sys driver and sOS + * are already been loaded. + */ + if (psp_v13_0_is_sos_alive(psp)) + return 0; + + ret = psp_v13_0_wait_for_bootloader(psp); + if (ret) + return ret; + + memset(psp->fw_pri_buf, 0, PSP_1_MEG); + + /* Copy Secure OS binary to PSP memory */ + memcpy(psp->fw_pri_buf, psp->sos_start_addr, psp->sos_bin_size); + + /* Provide the PSP secure OS to bootloader */ + WREG32_SOC15(MP0, 0, regMP0_SMN_C2PMSG_36, + (uint32_t)(psp->fw_pri_mc_addr >> 20)); + psp_gfxdrv_command_reg = PSP_BL__LOAD_SOSDRV; + WREG32_SOC15(MP0, 0, regMP0_SMN_C2PMSG_35, + psp_gfxdrv_command_reg); + + /* there might be handshake issue with hardware which needs delay */ + mdelay(20); + ret = psp_wait_for(psp, SOC15_REG_OFFSET(MP0, 0, regMP0_SMN_C2PMSG_81), + RREG32_SOC15(MP0, 0, regMP0_SMN_C2PMSG_81), + 0, true); + + return ret; +} + +static int psp_v13_0_ring_init(struct psp_context *psp, + enum psp_ring_type ring_type) +{ + int ret = 0; + struct psp_ring *ring; + struct amdgpu_device *adev = psp->adev; + + ring = &psp->km_ring; + + ring->ring_type = ring_type; + + /* allocate 4k Page of Local Frame Buffer memory for ring */ + ring->ring_size = 0x1000; + ret = amdgpu_bo_create_kernel(adev, ring->ring_size, PAGE_SIZE, + AMDGPU_GEM_DOMAIN_VRAM, + &adev->firmware.rbuf, + &ring->ring_mem_mc_addr, + (void **)&ring->ring_mem); + if (ret) { + ring->ring_size = 0; + return ret; + } + + return 0; +} + +static int psp_v13_0_ring_stop(struct psp_context *psp, + enum psp_ring_type ring_type) +{ + int ret = 0; + struct amdgpu_device *adev = psp->adev; + + if (amdgpu_sriov_vf(adev)) { + /* Write the ring destroy command*/ + WREG32_SOC15(MP0, 0, regMP0_SMN_C2PMSG_101, + GFX_CTRL_CMD_ID_DESTROY_GPCOM_RING); + /* there might be handshake issue with hardware which needs delay */ + mdelay(20); + /* Wait for response flag (bit 31) */ + ret = psp_wait_for(psp, SOC15_REG_OFFSET(MP0, 0, regMP0_SMN_C2PMSG_101), + 0x80000000, 0x80000000, false); + } else { + /* Write the ring destroy command*/ + WREG32_SOC15(MP0, 0, regMP0_SMN_C2PMSG_64, + GFX_CTRL_CMD_ID_DESTROY_RINGS); + /* there might be handshake issue with hardware which needs delay */ + mdelay(20); + /* Wait for response flag (bit 31) */ + ret = psp_wait_for(psp, SOC15_REG_OFFSET(MP0, 0, regMP0_SMN_C2PMSG_64), + 0x80000000, 0x80000000, false); + } + + return ret; +} + +static int psp_v13_0_ring_create(struct psp_context *psp, + enum psp_ring_type ring_type) +{ + int ret = 0; + unsigned int psp_ring_reg = 0; + struct psp_ring *ring = &psp->km_ring; + struct amdgpu_device *adev = psp->adev; + + if (amdgpu_sriov_vf(adev)) { + ret = psp_v13_0_ring_stop(psp, ring_type); + if (ret) { + DRM_ERROR("psp_v13_0_ring_stop_sriov failed!\n"); + return ret; + } + + /* Write low address of the ring to C2PMSG_102 */ + psp_ring_reg = lower_32_bits(ring->ring_mem_mc_addr); + WREG32_SOC15(MP0, 0, regMP0_SMN_C2PMSG_102, psp_ring_reg); + /* Write high address of the ring to C2PMSG_103 */ + psp_ring_reg = upper_32_bits(ring->ring_mem_mc_addr); + WREG32_SOC15(MP0, 0, regMP0_SMN_C2PMSG_103, psp_ring_reg); + + /* Write the ring initialization command to C2PMSG_101 */ + WREG32_SOC15(MP0, 0, regMP0_SMN_C2PMSG_101, + GFX_CTRL_CMD_ID_INIT_GPCOM_RING); + + /* there might be handshake issue with hardware which needs delay */ + mdelay(20); + + /* Wait for response flag (bit 31) in C2PMSG_101 */ + ret = psp_wait_for(psp, SOC15_REG_OFFSET(MP0, 0, regMP0_SMN_C2PMSG_101), + 0x80000000, 0x8000FFFF, false); + + } else { + /* Wait for sOS ready for ring creation */ + ret = psp_wait_for(psp, SOC15_REG_OFFSET(MP0, 0, regMP0_SMN_C2PMSG_64), + 0x80000000, 0x80000000, false); + if (ret) { + DRM_ERROR("Failed to wait for trust OS ready for ring creation\n"); + return ret; + } + + /* Write low address of the ring to C2PMSG_69 */ + psp_ring_reg = lower_32_bits(ring->ring_mem_mc_addr); + WREG32_SOC15(MP0, 0, regMP0_SMN_C2PMSG_69, psp_ring_reg); + /* Write high address of the ring to C2PMSG_70 */ + psp_ring_reg = upper_32_bits(ring->ring_mem_mc_addr); + WREG32_SOC15(MP0, 0, regMP0_SMN_C2PMSG_70, psp_ring_reg); + /* Write size of ring to C2PMSG_71 */ + psp_ring_reg = ring->ring_size; + WREG32_SOC15(MP0, 0, regMP0_SMN_C2PMSG_71, psp_ring_reg); + /* Write the ring initialization command to C2PMSG_64 */ + psp_ring_reg = ring_type; + psp_ring_reg = psp_ring_reg << 16; + WREG32_SOC15(MP0, 0, regMP0_SMN_C2PMSG_64, psp_ring_reg); + + /* there might be handshake issue with hardware which needs delay */ + mdelay(20); + + /* Wait for response flag (bit 31) in C2PMSG_64 */ + ret = psp_wait_for(psp, SOC15_REG_OFFSET(MP0, 0, regMP0_SMN_C2PMSG_64), + 0x80000000, 0x8000FFFF, false); + } + + return ret; +} + +static int psp_v13_0_ring_destroy(struct psp_context *psp, + enum psp_ring_type ring_type) +{ + int ret = 0; + struct psp_ring *ring = &psp->km_ring; + struct amdgpu_device *adev = psp->adev; + + ret = psp_v13_0_ring_stop(psp, ring_type); + if (ret) + DRM_ERROR("Fail to stop psp ring\n"); + + amdgpu_bo_free_kernel(&adev->firmware.rbuf, + &ring->ring_mem_mc_addr, + (void **)&ring->ring_mem); + + return ret; +} + +static uint32_t psp_v13_0_ring_get_wptr(struct psp_context *psp) +{ + uint32_t data; + struct amdgpu_device *adev = psp->adev; + + if (amdgpu_sriov_vf(adev)) + data = RREG32_SOC15(MP0, 0, regMP0_SMN_C2PMSG_102); + else + data = RREG32_SOC15(MP0, 0, regMP0_SMN_C2PMSG_67); + + return data; +} + +static void psp_v13_0_ring_set_wptr(struct psp_context *psp, uint32_t value) +{ + struct amdgpu_device *adev = psp->adev; + + if (amdgpu_sriov_vf(adev)) { + WREG32_SOC15(MP0, 0, regMP0_SMN_C2PMSG_102, value); + WREG32_SOC15(MP0, 0, regMP0_SMN_C2PMSG_101, + GFX_CTRL_CMD_ID_CONSUME_CMD); + } else + WREG32_SOC15(MP0, 0, regMP0_SMN_C2PMSG_67, value); +} + +static const struct psp_funcs psp_v13_0_funcs = { + .init_microcode = psp_v13_0_init_microcode, + .bootloader_load_kdb = psp_v13_0_bootloader_load_kdb, + .bootloader_load_sysdrv = psp_v13_0_bootloader_load_sysdrv, + .bootloader_load_sos = psp_v13_0_bootloader_load_sos, + .ring_init = psp_v13_0_ring_init, + .ring_create = psp_v13_0_ring_create, + .ring_stop = psp_v13_0_ring_stop, + .ring_destroy = psp_v13_0_ring_destroy, + .ring_get_wptr = psp_v13_0_ring_get_wptr, + .ring_set_wptr = psp_v13_0_ring_set_wptr, +}; + +void psp_v13_0_set_psp_funcs(struct psp_context *psp) +{ + psp->funcs = &psp_v13_0_funcs; +} diff --git a/drivers/gpu/drm/amd/amdgpu/psp_v13_0.h b/drivers/gpu/drm/amd/amdgpu/psp_v13_0.h new file mode 100644 index 000000000000..b2414a729ca1 --- /dev/null +++ b/drivers/gpu/drm/amd/amdgpu/psp_v13_0.h @@ -0,0 +1,30 @@ +/* + * Copyright 2020 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + */ +#ifndef __PSP_V13_0_H__ +#define __PSP_V13_0_H__ + +#include "amdgpu_psp.h" + +void psp_v13_0_set_psp_funcs(struct psp_context *psp); + +#endif diff --git a/drivers/gpu/drm/amd/amdgpu/sdma_v2_4.c b/drivers/gpu/drm/amd/amdgpu/sdma_v2_4.c index eb5dc6c5b46e..9f0dda040ec8 100644 --- a/drivers/gpu/drm/amd/amdgpu/sdma_v2_4.c +++ b/drivers/gpu/drm/amd/amdgpu/sdma_v2_4.c @@ -876,12 +876,10 @@ static int sdma_v2_4_sw_init(void *handle) ring->ring_obj = NULL; ring->use_doorbell = false; sprintf(ring->name, "sdma%d", i); - r = amdgpu_ring_init(adev, ring, 1024, - &adev->sdma.trap_irq, - (i == 0) ? - AMDGPU_SDMA_IRQ_INSTANCE0 : + r = amdgpu_ring_init(adev, ring, 1024, &adev->sdma.trap_irq, + (i == 0) ? AMDGPU_SDMA_IRQ_INSTANCE0 : AMDGPU_SDMA_IRQ_INSTANCE1, - AMDGPU_RING_PRIO_DEFAULT); + AMDGPU_RING_PRIO_DEFAULT, NULL); if (r) return r; } diff --git a/drivers/gpu/drm/amd/amdgpu/sdma_v3_0.c b/drivers/gpu/drm/amd/amdgpu/sdma_v3_0.c index ad308d8c6d30..135727b59c41 100644 --- a/drivers/gpu/drm/amd/amdgpu/sdma_v3_0.c +++ b/drivers/gpu/drm/amd/amdgpu/sdma_v3_0.c @@ -1160,12 +1160,10 @@ static int sdma_v3_0_sw_init(void *handle) } sprintf(ring->name, "sdma%d", i); - r = amdgpu_ring_init(adev, ring, 1024, - &adev->sdma.trap_irq, - (i == 0) ? - AMDGPU_SDMA_IRQ_INSTANCE0 : + r = amdgpu_ring_init(adev, ring, 1024, &adev->sdma.trap_irq, + (i == 0) ? AMDGPU_SDMA_IRQ_INSTANCE0 : AMDGPU_SDMA_IRQ_INSTANCE1, - AMDGPU_RING_PRIO_DEFAULT); + AMDGPU_RING_PRIO_DEFAULT, NULL); if (r) return r; } diff --git a/drivers/gpu/drm/amd/amdgpu/sdma_v4_0.c b/drivers/gpu/drm/amd/amdgpu/sdma_v4_0.c index c8c22c1d1e65..5715be6770ec 100644 --- a/drivers/gpu/drm/amd/amdgpu/sdma_v4_0.c +++ b/drivers/gpu/drm/amd/amdgpu/sdma_v4_0.c @@ -56,6 +56,7 @@ #include "ivsrcid/sdma1/irqsrcs_sdma1_4_0.h" #include "amdgpu_ras.h" +#include "sdma_v4_4.h" MODULE_FIRMWARE("amdgpu/vega10_sdma.bin"); MODULE_FIRMWARE("amdgpu/vega10_sdma1.bin"); @@ -69,6 +70,7 @@ MODULE_FIRMWARE("amdgpu/raven2_sdma.bin"); MODULE_FIRMWARE("amdgpu/arcturus_sdma.bin"); MODULE_FIRMWARE("amdgpu/renoir_sdma.bin"); MODULE_FIRMWARE("amdgpu/green_sardine_sdma.bin"); +MODULE_FIRMWARE("amdgpu/aldebaran_sdma.bin"); #define SDMA0_POWER_CNTL__ON_OFF_CONDITION_HOLD_TIME_MASK 0x000000F8L #define SDMA0_POWER_CNTL__ON_OFF_STATUS_DURATION_TIME_MASK 0xFC000000L @@ -259,6 +261,24 @@ static const struct soc15_reg_golden golden_settings_sdma_arct[] = SOC15_REG_GOLDEN_VALUE(SDMA7, 0, mmSDMA7_UTCL1_TIMEOUT, 0xffffffff, 0x00010001) }; +static const struct soc15_reg_golden golden_settings_sdma_aldebaran[] = { + SOC15_REG_GOLDEN_VALUE(SDMA0, 0, mmSDMA0_GB_ADDR_CONFIG, 0x0018773f, 0x00104002), + SOC15_REG_GOLDEN_VALUE(SDMA0, 0, mmSDMA0_GB_ADDR_CONFIG_READ, 0x0018773f, 0x00104002), + SOC15_REG_GOLDEN_VALUE(SDMA0, 0, mmSDMA0_UTCL1_TIMEOUT, 0xffffffff, 0x00010001), + SOC15_REG_GOLDEN_VALUE(SDMA1, 0, mmSDMA1_GB_ADDR_CONFIG, 0x0018773f, 0x00104002), + SOC15_REG_GOLDEN_VALUE(SDMA1, 0, mmSDMA1_GB_ADDR_CONFIG_READ, 0x0018773f, 0x00104002), + SOC15_REG_GOLDEN_VALUE(SDMA1, 0, mmSDMA1_UTCL1_TIMEOUT, 0xffffffff, 0x00010001), + SOC15_REG_GOLDEN_VALUE(SDMA2, 0, mmSDMA2_GB_ADDR_CONFIG, 0x0018773f, 0x00104002), + SOC15_REG_GOLDEN_VALUE(SDMA2, 0, mmSDMA2_GB_ADDR_CONFIG_READ, 0x0018773f, 0x00104002), + SOC15_REG_GOLDEN_VALUE(SDMA3, 0, mmSDMA2_UTCL1_TIMEOUT, 0xffffffff, 0x00010001), + SOC15_REG_GOLDEN_VALUE(SDMA3, 0, mmSDMA3_GB_ADDR_CONFIG, 0x0018773f, 0x00104002), + SOC15_REG_GOLDEN_VALUE(SDMA3, 0, mmSDMA3_GB_ADDR_CONFIG_READ, 0x0018773f, 0x00104002), + SOC15_REG_GOLDEN_VALUE(SDMA3, 0, mmSDMA3_UTCL1_TIMEOUT, 0xffffffff, 0x00010001), + SOC15_REG_GOLDEN_VALUE(SDMA4, 0, mmSDMA4_GB_ADDR_CONFIG, 0x0018773f, 0x00104002), + SOC15_REG_GOLDEN_VALUE(SDMA4, 0, mmSDMA4_GB_ADDR_CONFIG_READ, 0x0018773f, 0x00104002), + SOC15_REG_GOLDEN_VALUE(SDMA4, 0, mmSDMA4_UTCL1_TIMEOUT, 0xffffffff, 0x00010001), +}; + static const struct soc15_reg_golden golden_settings_sdma_4_3[] = { SOC15_REG_GOLDEN_VALUE(SDMA0, 0, mmSDMA0_CHICKEN_BITS, 0xfe931f07, 0x02831f07), SOC15_REG_GOLDEN_VALUE(SDMA0, 0, mmSDMA0_CLK_CTRL, 0xffffffff, 0x3f000100), @@ -482,6 +502,11 @@ static void sdma_v4_0_init_golden_registers(struct amdgpu_device *adev) golden_settings_sdma_arct, ARRAY_SIZE(golden_settings_sdma_arct)); break; + case CHIP_ALDEBARAN: + soc15_program_register_sequence(adev, + golden_settings_sdma_aldebaran, + ARRAY_SIZE(golden_settings_sdma_aldebaran)); + break; case CHIP_RAVEN: soc15_program_register_sequence(adev, golden_settings_sdma_4_1, @@ -564,7 +589,8 @@ static void sdma_v4_0_destroy_inst_ctx(struct amdgpu_device *adev) /* arcturus shares the same FW memory across all SDMA isntances */ - if (adev->asic_type == CHIP_ARCTURUS) + if (adev->asic_type == CHIP_ARCTURUS || + adev->asic_type == CHIP_ALDEBARAN) break; } @@ -621,6 +647,9 @@ static int sdma_v4_0_init_microcode(struct amdgpu_device *adev) else chip_name = "green_sardine"; break; + case CHIP_ALDEBARAN: + chip_name = "aldebaran"; + break; default: BUG(); } @@ -636,8 +665,9 @@ static int sdma_v4_0_init_microcode(struct amdgpu_device *adev) goto out; for (i = 1; i < adev->sdma.num_instances; i++) { - if (adev->asic_type == CHIP_ARCTURUS) { - /* Acturus will leverage the same FW memory + if (adev->asic_type == CHIP_ARCTURUS || + adev->asic_type == CHIP_ALDEBARAN) { + /* Acturus & Aldebaran will leverage the same FW memory for every SDMA instance */ memcpy((void *)&adev->sdma.instance[i], (void *)&adev->sdma.instance[0], @@ -1825,6 +1855,8 @@ static int sdma_v4_0_early_init(void *handle) adev->sdma.num_instances = 1; else if (adev->asic_type == CHIP_ARCTURUS) adev->sdma.num_instances = 8; + else if (adev->asic_type == CHIP_ALDEBARAN) + adev->sdma.num_instances = 5; else adev->sdma.num_instances = 2; @@ -1895,6 +1927,33 @@ static int sdma_v4_0_sw_init(void *handle) return r; } + /* SDMA VM_HOLE/DOORBELL_INV/POLL_TIMEOUT/SRBM_WRITE_PROTECTION event*/ + for (i = 0; i < adev->sdma.num_instances; i++) { + r = amdgpu_irq_add_id(adev, sdma_v4_0_seq_to_irq_id(i), + SDMA0_4_0__SRCID__SDMA_VM_HOLE, + &adev->sdma.vm_hole_irq); + if (r) + return r; + + r = amdgpu_irq_add_id(adev, sdma_v4_0_seq_to_irq_id(i), + SDMA0_4_0__SRCID__SDMA_DOORBELL_INVALID, + &adev->sdma.doorbell_invalid_irq); + if (r) + return r; + + r = amdgpu_irq_add_id(adev, sdma_v4_0_seq_to_irq_id(i), + SDMA0_4_0__SRCID__SDMA_POLL_TIMEOUT, + &adev->sdma.pool_timeout_irq); + if (r) + return r; + + r = amdgpu_irq_add_id(adev, sdma_v4_0_seq_to_irq_id(i), + SDMA0_4_0__SRCID__SDMA_SRBMWRITE, + &adev->sdma.srbm_write_irq); + if (r) + return r; + } + for (i = 0; i < adev->sdma.num_instances; i++) { ring = &adev->sdma.instance[i].ring; ring->ring_obj = NULL; @@ -1909,7 +1968,7 @@ static int sdma_v4_0_sw_init(void *handle) sprintf(ring->name, "sdma%d", i); r = amdgpu_ring_init(adev, ring, 1024, &adev->sdma.trap_irq, AMDGPU_SDMA_IRQ_INSTANCE0 + i, - AMDGPU_RING_PRIO_DEFAULT); + AMDGPU_RING_PRIO_DEFAULT, NULL); if (r) return r; @@ -1928,7 +1987,7 @@ static int sdma_v4_0_sw_init(void *handle) r = amdgpu_ring_init(adev, ring, 1024, &adev->sdma.trap_irq, AMDGPU_SDMA_IRQ_INSTANCE0 + i, - AMDGPU_RING_PRIO_DEFAULT); + AMDGPU_RING_PRIO_DEFAULT, NULL); if (r) return r; } @@ -2149,6 +2208,72 @@ static int sdma_v4_0_set_ecc_irq_state(struct amdgpu_device *adev, return 0; } +static int sdma_v4_0_print_iv_entry(struct amdgpu_device *adev, + struct amdgpu_iv_entry *entry) +{ + int instance; + struct amdgpu_task_info task_info; + u64 addr; + + instance = sdma_v4_0_irq_id_to_seq(entry->client_id); + if (instance < 0 || instance >= adev->sdma.num_instances) { + dev_err(adev->dev, "sdma instance invalid %d\n", instance); + return -EINVAL; + } + + addr = (u64)entry->src_data[0] << 12; + addr |= ((u64)entry->src_data[1] & 0xf) << 44; + + memset(&task_info, 0, sizeof(struct amdgpu_task_info)); + amdgpu_vm_get_task_info(adev, entry->pasid, &task_info); + + dev_info(adev->dev, + "[sdma%d] address:0x%016llx src_id:%u ring:%u vmid:%u " + "pasid:%u, for process %s pid %d thread %s pid %d\n", + instance, addr, entry->src_id, entry->ring_id, entry->vmid, + entry->pasid, task_info.process_name, task_info.tgid, + task_info.task_name, task_info.pid); + return 0; +} + +static int sdma_v4_0_process_vm_hole_irq(struct amdgpu_device *adev, + struct amdgpu_irq_src *source, + struct amdgpu_iv_entry *entry) +{ + dev_err(adev->dev, "MC or SEM address in VM hole\n"); + sdma_v4_0_print_iv_entry(adev, entry); + return 0; +} + +static int sdma_v4_0_process_doorbell_invalid_irq(struct amdgpu_device *adev, + struct amdgpu_irq_src *source, + struct amdgpu_iv_entry *entry) +{ + dev_err(adev->dev, "SDMA received a doorbell from BIF with byte_enable !=0xff\n"); + sdma_v4_0_print_iv_entry(adev, entry); + return 0; +} + +static int sdma_v4_0_process_pool_timeout_irq(struct amdgpu_device *adev, + struct amdgpu_irq_src *source, + struct amdgpu_iv_entry *entry) +{ + dev_err(adev->dev, + "Polling register/memory timeout executing POLL_REG/MEM with finite timer\n"); + sdma_v4_0_print_iv_entry(adev, entry); + return 0; +} + +static int sdma_v4_0_process_srbm_write_irq(struct amdgpu_device *adev, + struct amdgpu_irq_src *source, + struct amdgpu_iv_entry *entry) +{ + dev_err(adev->dev, + "SDMA gets an Register Write SRBM_WRITE command in non-privilege command buffer\n"); + sdma_v4_0_print_iv_entry(adev, entry); + return 0; +} + static void sdma_v4_0_update_medium_grain_clock_gating( struct amdgpu_device *adev, bool enable) @@ -2222,21 +2347,10 @@ static int sdma_v4_0_set_clockgating_state(void *handle, if (amdgpu_sriov_vf(adev)) return 0; - switch (adev->asic_type) { - case CHIP_VEGA10: - case CHIP_VEGA12: - case CHIP_VEGA20: - case CHIP_RAVEN: - case CHIP_ARCTURUS: - case CHIP_RENOIR: - sdma_v4_0_update_medium_grain_clock_gating(adev, - state == AMD_CG_STATE_GATE); - sdma_v4_0_update_medium_grain_light_sleep(adev, - state == AMD_CG_STATE_GATE); - break; - default: - break; - } + sdma_v4_0_update_medium_grain_clock_gating(adev, + state == AMD_CG_STATE_GATE); + sdma_v4_0_update_medium_grain_light_sleep(adev, + state == AMD_CG_STATE_GATE); return 0; } @@ -2249,7 +2363,7 @@ static int sdma_v4_0_set_powergating_state(void *handle, case CHIP_RAVEN: case CHIP_RENOIR: sdma_v4_1_update_power_gating(adev, - state == AMD_PG_STATE_GATE ? true : false); + state == AMD_PG_STATE_GATE); break; default: break; @@ -2465,7 +2579,21 @@ static const struct amdgpu_irq_src_funcs sdma_v4_0_ecc_irq_funcs = { .process = amdgpu_sdma_process_ecc_irq, }; +static const struct amdgpu_irq_src_funcs sdma_v4_0_vm_hole_irq_funcs = { + .process = sdma_v4_0_process_vm_hole_irq, +}; + +static const struct amdgpu_irq_src_funcs sdma_v4_0_doorbell_invalid_irq_funcs = { + .process = sdma_v4_0_process_doorbell_invalid_irq, +}; + +static const struct amdgpu_irq_src_funcs sdma_v4_0_pool_timeout_irq_funcs = { + .process = sdma_v4_0_process_pool_timeout_irq, +}; +static const struct amdgpu_irq_src_funcs sdma_v4_0_srbm_write_irq_funcs = { + .process = sdma_v4_0_process_srbm_write_irq, +}; static void sdma_v4_0_set_irq_funcs(struct amdgpu_device *adev) { @@ -2474,9 +2602,17 @@ static void sdma_v4_0_set_irq_funcs(struct amdgpu_device *adev) adev->sdma.trap_irq.num_types = AMDGPU_SDMA_IRQ_INSTANCE1; adev->sdma.ecc_irq.num_types = AMDGPU_SDMA_IRQ_INSTANCE1; break; + case 5: + adev->sdma.trap_irq.num_types = AMDGPU_SDMA_IRQ_INSTANCE5; + adev->sdma.ecc_irq.num_types = AMDGPU_SDMA_IRQ_INSTANCE5; + break; case 8: adev->sdma.trap_irq.num_types = AMDGPU_SDMA_IRQ_LAST; adev->sdma.ecc_irq.num_types = AMDGPU_SDMA_IRQ_LAST; + adev->sdma.vm_hole_irq.num_types = AMDGPU_SDMA_IRQ_INSTANCE5; + adev->sdma.doorbell_invalid_irq.num_types = AMDGPU_SDMA_IRQ_LAST; + adev->sdma.pool_timeout_irq.num_types = AMDGPU_SDMA_IRQ_LAST; + adev->sdma.srbm_write_irq.num_types = AMDGPU_SDMA_IRQ_LAST; break; case 2: default: @@ -2487,6 +2623,10 @@ static void sdma_v4_0_set_irq_funcs(struct amdgpu_device *adev) adev->sdma.trap_irq.funcs = &sdma_v4_0_trap_irq_funcs; adev->sdma.illegal_inst_irq.funcs = &sdma_v4_0_illegal_inst_irq_funcs; adev->sdma.ecc_irq.funcs = &sdma_v4_0_ecc_irq_funcs; + adev->sdma.vm_hole_irq.funcs = &sdma_v4_0_vm_hole_irq_funcs; + adev->sdma.doorbell_invalid_irq.funcs = &sdma_v4_0_doorbell_invalid_irq_funcs; + adev->sdma.pool_timeout_irq.funcs = &sdma_v4_0_pool_timeout_irq_funcs; + adev->sdma.srbm_write_irq.funcs = &sdma_v4_0_srbm_write_irq_funcs; } /** @@ -2655,6 +2795,9 @@ static void sdma_v4_0_set_ras_funcs(struct amdgpu_device *adev) case CHIP_ARCTURUS: adev->sdma.funcs = &sdma_v4_0_ras_funcs; break; + case CHIP_ALDEBARAN: + adev->sdma.funcs = &sdma_v4_4_ras_funcs; + break; default: break; } diff --git a/drivers/gpu/drm/amd/amdgpu/sdma_v4_4.c b/drivers/gpu/drm/amd/amdgpu/sdma_v4_4.c new file mode 100644 index 000000000000..bf95007f0843 --- /dev/null +++ b/drivers/gpu/drm/amd/amdgpu/sdma_v4_4.c @@ -0,0 +1,253 @@ +/* + * Copyright 2020 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + */ +#include "amdgpu.h" +#include "sdma/sdma_4_4_0_offset.h" +#include "sdma/sdma_4_4_0_sh_mask.h" +#include "soc15.h" +#include "amdgpu_ras.h" + +#define SDMA1_REG_OFFSET 0x600 +#define SDMA2_REG_OFFSET 0x1cda0 +#define SDMA3_REG_OFFSET 0x1d1a0 +#define SDMA4_REG_OFFSET 0x1d5a0 + +/* helper function that allow only use sdma0 register offset + * to calculate register offset for all the sdma instances */ +static uint32_t sdma_v4_4_get_reg_offset(struct amdgpu_device *adev, + uint32_t instance, + uint32_t offset) +{ + uint32_t sdma_base = adev->reg_offset[SDMA0_HWIP][0][0]; + + switch (instance) { + case 0: + return (sdma_base + offset); + case 1: + return (sdma_base + SDMA1_REG_OFFSET + offset); + case 2: + return (sdma_base + SDMA2_REG_OFFSET + offset); + case 3: + return (sdma_base + SDMA3_REG_OFFSET + offset); + case 4: + return (sdma_base + SDMA4_REG_OFFSET + offset); + default: + break; + } + return 0; +} + +static const struct soc15_ras_field_entry sdma_v4_4_ras_fields[] = { + { "SDMA_MBANK_DATA_BUF0_SED", SOC15_REG_ENTRY(SDMA0, 0, regSDMA0_EDC_COUNTER), + SOC15_REG_FIELD(SDMA0_EDC_COUNTER, SDMA_MBANK_DATA_BUF0_SED), + 0, 0, + }, + { "SDMA_MBANK_DATA_BUF1_SED", SOC15_REG_ENTRY(SDMA0, 0, regSDMA0_EDC_COUNTER), + SOC15_REG_FIELD(SDMA0_EDC_COUNTER, SDMA_MBANK_DATA_BUF1_SED), + 0, 0, + }, + { "SDMA_MBANK_DATA_BUF2_SED", SOC15_REG_ENTRY(SDMA0, 0, regSDMA0_EDC_COUNTER), + SOC15_REG_FIELD(SDMA0_EDC_COUNTER, SDMA_MBANK_DATA_BUF2_SED), + 0, 0, + }, + { "SDMA_MBANK_DATA_BUF3_SED", SOC15_REG_ENTRY(SDMA0, 0, regSDMA0_EDC_COUNTER), + SOC15_REG_FIELD(SDMA0_EDC_COUNTER, SDMA_MBANK_DATA_BUF3_SED), + 0, 0, + }, + { "SDMA_MBANK_DATA_BUF4_SED", SOC15_REG_ENTRY(SDMA0, 0, regSDMA0_EDC_COUNTER), + SOC15_REG_FIELD(SDMA0_EDC_COUNTER, SDMA_MBANK_DATA_BUF4_SED), + 0, 0, + }, + { "SDMA_MBANK_DATA_BUF5_SED", SOC15_REG_ENTRY(SDMA0, 0, regSDMA0_EDC_COUNTER), + SOC15_REG_FIELD(SDMA0_EDC_COUNTER, SDMA_MBANK_DATA_BUF5_SED), + 0, 0, + }, + { "SDMA_MBANK_DATA_BUF6_SED", SOC15_REG_ENTRY(SDMA0, 0, regSDMA0_EDC_COUNTER), + SOC15_REG_FIELD(SDMA0_EDC_COUNTER, SDMA_MBANK_DATA_BUF6_SED), + 0, 0, + }, + { "SDMA_MBANK_DATA_BUF7_SED", SOC15_REG_ENTRY(SDMA0, 0, regSDMA0_EDC_COUNTER), + SOC15_REG_FIELD(SDMA0_EDC_COUNTER, SDMA_MBANK_DATA_BUF7_SED), + 0, 0, + }, + { "SDMA_MBANK_DATA_BUF8_SED", SOC15_REG_ENTRY(SDMA0, 0, regSDMA0_EDC_COUNTER), + SOC15_REG_FIELD(SDMA0_EDC_COUNTER, SDMA_MBANK_DATA_BUF8_SED), + 0, 0, + }, + { "SDMA_MBANK_DATA_BUF9_SED", SOC15_REG_ENTRY(SDMA0, 0, regSDMA0_EDC_COUNTER), + SOC15_REG_FIELD(SDMA0_EDC_COUNTER, SDMA_MBANK_DATA_BUF9_SED), + 0, 0, + }, + { "SDMA_MBANK_DATA_BUF10_SED", SOC15_REG_ENTRY(SDMA0, 0, regSDMA0_EDC_COUNTER), + SOC15_REG_FIELD(SDMA0_EDC_COUNTER, SDMA_MBANK_DATA_BUF10_SED), + 0, 0, + }, + { "SDMA_MBANK_DATA_BUF11_SED", SOC15_REG_ENTRY(SDMA0, 0, regSDMA0_EDC_COUNTER), + SOC15_REG_FIELD(SDMA0_EDC_COUNTER, SDMA_MBANK_DATA_BUF11_SED), + 0, 0, + }, + { "SDMA_MBANK_DATA_BUF12_SED", SOC15_REG_ENTRY(SDMA0, 0, regSDMA0_EDC_COUNTER), + SOC15_REG_FIELD(SDMA0_EDC_COUNTER, SDMA_MBANK_DATA_BUF12_SED), + 0, 0, + }, + { "SDMA_MBANK_DATA_BUF13_SED", SOC15_REG_ENTRY(SDMA0, 0, regSDMA0_EDC_COUNTER), + SOC15_REG_FIELD(SDMA0_EDC_COUNTER, SDMA_MBANK_DATA_BUF13_SED), + 0, 0, + }, + { "SDMA_MBANK_DATA_BUF14_SED", SOC15_REG_ENTRY(SDMA0, 0, regSDMA0_EDC_COUNTER), + SOC15_REG_FIELD(SDMA0_EDC_COUNTER, SDMA_MBANK_DATA_BUF14_SED), + 0, 0, + }, + { "SDMA_MBANK_DATA_BUF15_SED", SOC15_REG_ENTRY(SDMA0, 0, regSDMA0_EDC_COUNTER), + SOC15_REG_FIELD(SDMA0_EDC_COUNTER, SDMA_MBANK_DATA_BUF15_SED), + 0, 0, + }, + { "SDMA_UCODE_BUF_SED", SOC15_REG_ENTRY(SDMA0, 0, regSDMA0_EDC_COUNTER2), + SOC15_REG_FIELD(SDMA0_EDC_COUNTER2, SDMA_UCODE_BUF_SED), + 0, 0, + }, + { "SDMA_RB_CMD_BUF_SED", SOC15_REG_ENTRY(SDMA0, 0, regSDMA0_EDC_COUNTER2), + SOC15_REG_FIELD(SDMA0_EDC_COUNTER2, SDMA_RB_CMD_BUF_SED), + 0, 0, + }, + { "SDMA_IB_CMD_BUF_SED", SOC15_REG_ENTRY(SDMA0, 0, regSDMA0_EDC_COUNTER2), + SOC15_REG_FIELD(SDMA0_EDC_COUNTER2, SDMA_IB_CMD_BUF_SED), + 0, 0, + }, + { "SDMA_UTCL1_RD_FIFO_SED", SOC15_REG_ENTRY(SDMA0, 0, regSDMA0_EDC_COUNTER2), + SOC15_REG_FIELD(SDMA0_EDC_COUNTER2, SDMA_UTCL1_RD_FIFO_SED), + 0, 0, + }, + { "SDMA_UTCL1_RDBST_FIFO_SED", SOC15_REG_ENTRY(SDMA0, 0, regSDMA0_EDC_COUNTER2), + SOC15_REG_FIELD(SDMA0_EDC_COUNTER2, SDMA_UTCL1_RDBST_FIFO_SED), + 0, 0, + }, + { "SDMA_DATA_LUT_FIFO_SED", SOC15_REG_ENTRY(SDMA0, 0, regSDMA0_EDC_COUNTER2), + SOC15_REG_FIELD(SDMA0_EDC_COUNTER2, SDMA_DATA_LUT_FIFO_SED), + 0, 0, + }, + { "SDMA_SPLIT_DATA_BUF_SED", SOC15_REG_ENTRY(SDMA0, 0, regSDMA0_EDC_COUNTER2), + SOC15_REG_FIELD(SDMA0_EDC_COUNTER2, SDMA_SPLIT_DATA_BUF_SED), + 0, 0, + }, + { "SDMA_MC_WR_ADDR_FIFO_SED", SOC15_REG_ENTRY(SDMA0, 0, regSDMA0_EDC_COUNTER2), + SOC15_REG_FIELD(SDMA0_EDC_COUNTER2, SDMA_MC_WR_ADDR_FIFO_SED), + 0, 0, + }, + { "SDMA_MC_RDRET_BUF_SED", SOC15_REG_ENTRY(SDMA0, 0, regSDMA0_EDC_COUNTER2), + SOC15_REG_FIELD(SDMA0_EDC_COUNTER2, SDMA_MC_WR_ADDR_FIFO_SED), + 0, 0, + }, +}; + +static void sdma_v4_4_get_ras_error_count(struct amdgpu_device *adev, + uint32_t reg_offset, + uint32_t value, + uint32_t instance, + uint32_t *sec_count) +{ + uint32_t i; + uint32_t sec_cnt; + + /* double bits error (multiple bits) error detection is not supported */ + for (i = 0; i < ARRAY_SIZE(sdma_v4_4_ras_fields); i++) { + if (sdma_v4_4_ras_fields[i].reg_offset != reg_offset) + continue; + + /* the SDMA_EDC_COUNTER register in each sdma instance + * shares the same sed shift_mask + * */ + sec_cnt = (value & + sdma_v4_4_ras_fields[i].sec_count_mask) >> + sdma_v4_4_ras_fields[i].sec_count_shift; + if (sec_cnt) { + dev_info(adev->dev, "Detected %s in SDMA%d, SED %d\n", + sdma_v4_4_ras_fields[i].name, + instance, sec_cnt); + *sec_count += sec_cnt; + } + } +} + +static int sdma_v4_4_query_ras_error_count(struct amdgpu_device *adev, + uint32_t instance, + void *ras_error_status) +{ + struct ras_err_data *err_data = (struct ras_err_data *)ras_error_status; + uint32_t sec_count = 0; + uint32_t reg_value = 0; + uint32_t reg_offset = 0; + + reg_offset = sdma_v4_4_get_reg_offset(adev, instance, regSDMA0_EDC_COUNTER); + reg_value = RREG32(reg_offset); + /* double bit error is not supported */ + if (reg_value) + sdma_v4_4_get_ras_error_count(adev, regSDMA0_EDC_COUNTER, reg_value, + instance, &sec_count); + + reg_offset = sdma_v4_4_get_reg_offset(adev, instance, regSDMA0_EDC_COUNTER2); + reg_value = RREG32(reg_offset); + /* double bit error is not supported */ + if (reg_value) + sdma_v4_4_get_ras_error_count(adev, regSDMA0_EDC_COUNTER2, reg_value, + instance, &sec_count); + + /* + * err_data->ue_count should be initialized to 0 + * before calling into this function + * + * SDMA RAS supports single bit uncorrectable error detection. + * So, increment uncorrectable error count. + */ + err_data->ue_count += sec_count; + + /* + * SDMA RAS does not support correctable errors. + * Set ce count to 0. + */ + err_data->ce_count = 0; + + return 0; +}; + +static void sdma_v4_4_reset_ras_error_count(struct amdgpu_device *adev) +{ + int i; + uint32_t reg_offset; + + /* write 0 to EDC_COUNTER reg to clear sdma edc counters */ + if (amdgpu_ras_is_supported(adev, AMDGPU_RAS_BLOCK__SDMA)) { + for (i = 0; i < adev->sdma.num_instances; i++) { + reg_offset = sdma_v4_4_get_reg_offset(adev, i, regSDMA0_EDC_COUNTER); + WREG32(reg_offset, 0); + reg_offset = sdma_v4_4_get_reg_offset(adev, i, regSDMA0_EDC_COUNTER2); + WREG32(reg_offset, 0); + } + } +} + +const struct amdgpu_sdma_ras_funcs sdma_v4_4_ras_funcs = { + .ras_late_init = amdgpu_sdma_ras_late_init, + .ras_fini = amdgpu_sdma_ras_fini, + .query_ras_error_count = sdma_v4_4_query_ras_error_count, + .reset_ras_error_count = sdma_v4_4_reset_ras_error_count, +}; diff --git a/drivers/gpu/drm/amd/amdgpu/sdma_v4_4.h b/drivers/gpu/drm/amd/amdgpu/sdma_v4_4.h new file mode 100644 index 000000000000..74a6e5b5e949 --- /dev/null +++ b/drivers/gpu/drm/amd/amdgpu/sdma_v4_4.h @@ -0,0 +1,28 @@ +/* + * Copyright 2020 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + */ +#ifndef __SDMA_V4_4_H__ +#define __SDMA_V4_4_H__ + +extern const struct amdgpu_sdma_ras_funcs sdma_v4_4_ras_funcs; + +#endif diff --git a/drivers/gpu/drm/amd/amdgpu/sdma_v5_0.c b/drivers/gpu/drm/amd/amdgpu/sdma_v5_0.c index d345e324837d..920fc6d4a127 100644 --- a/drivers/gpu/drm/amd/amdgpu/sdma_v5_0.c +++ b/drivers/gpu/drm/amd/amdgpu/sdma_v5_0.c @@ -1273,12 +1273,10 @@ static int sdma_v5_0_sw_init(void *handle) : (adev->doorbell_index.sdma_engine[1] << 1); // get DWORD offset sprintf(ring->name, "sdma%d", i); - r = amdgpu_ring_init(adev, ring, 1024, - &adev->sdma.trap_irq, - (i == 0) ? - AMDGPU_SDMA_IRQ_INSTANCE0 : + r = amdgpu_ring_init(adev, ring, 1024, &adev->sdma.trap_irq, + (i == 0) ? AMDGPU_SDMA_IRQ_INSTANCE0 : AMDGPU_SDMA_IRQ_INSTANCE1, - AMDGPU_RING_PRIO_DEFAULT); + AMDGPU_RING_PRIO_DEFAULT, NULL); if (r) return r; } diff --git a/drivers/gpu/drm/amd/amdgpu/sdma_v5_2.c b/drivers/gpu/drm/amd/amdgpu/sdma_v5_2.c index 690a5090475a..b1ad9e52b234 100644 --- a/drivers/gpu/drm/amd/amdgpu/sdma_v5_2.c +++ b/drivers/gpu/drm/amd/amdgpu/sdma_v5_2.c @@ -370,6 +370,33 @@ static void sdma_v5_2_ring_emit_ib(struct amdgpu_ring *ring, } /** + * sdma_v5_2_ring_emit_mem_sync - flush the IB by graphics cache rinse + * + * @ring: amdgpu ring pointer + * @job: job to retrieve vmid from + * @ib: IB object to schedule + * + * flush the IB by graphics cache rinse. + */ +static void sdma_v5_2_ring_emit_mem_sync(struct amdgpu_ring *ring) +{ + uint32_t gcr_cntl = + SDMA_GCR_GL2_INV | SDMA_GCR_GL2_WB | SDMA_GCR_GLM_INV | + SDMA_GCR_GL1_INV | SDMA_GCR_GLV_INV | SDMA_GCR_GLK_INV | + SDMA_GCR_GLI_INV(1); + + /* flush entire cache L0/L1/L2, this can be optimized by performance requirement */ + amdgpu_ring_write(ring, SDMA_PKT_HEADER_OP(SDMA_OP_GCR_REQ)); + amdgpu_ring_write(ring, SDMA_PKT_GCR_REQ_PAYLOAD1_BASE_VA_31_7(0)); + amdgpu_ring_write(ring, SDMA_PKT_GCR_REQ_PAYLOAD2_GCR_CONTROL_15_0(gcr_cntl) | + SDMA_PKT_GCR_REQ_PAYLOAD2_BASE_VA_47_32(0)); + amdgpu_ring_write(ring, SDMA_PKT_GCR_REQ_PAYLOAD3_LIMIT_VA_31_7(0) | + SDMA_PKT_GCR_REQ_PAYLOAD3_GCR_CONTROL_18_16(gcr_cntl >> 16)); + amdgpu_ring_write(ring, SDMA_PKT_GCR_REQ_PAYLOAD4_LIMIT_VA_47_32(0) | + SDMA_PKT_GCR_REQ_PAYLOAD4_VMID(0)); +} + +/** * sdma_v5_2_ring_emit_hdp_flush - emit an hdp flush on the DMA ring * * @ring: amdgpu ring pointer @@ -1283,10 +1310,9 @@ static int sdma_v5_2_sw_init(void *handle) (adev->doorbell_index.sdma_engine[i] << 1); //get DWORD offset sprintf(ring->name, "sdma%d", i); - r = amdgpu_ring_init(adev, ring, 1024, - &adev->sdma.trap_irq, + r = amdgpu_ring_init(adev, ring, 1024, &adev->sdma.trap_irq, AMDGPU_SDMA_IRQ_INSTANCE0 + i, - AMDGPU_RING_PRIO_DEFAULT); + AMDGPU_RING_PRIO_DEFAULT, NULL); if (r) return r; } @@ -1595,9 +1621,9 @@ static int sdma_v5_2_set_clockgating_state(void *handle, case CHIP_VANGOGH: case CHIP_DIMGREY_CAVEFISH: sdma_v5_2_update_medium_grain_clock_gating(adev, - state == AMD_CG_STATE_GATE ? true : false); + state == AMD_CG_STATE_GATE); sdma_v5_2_update_medium_grain_light_sleep(adev, - state == AMD_CG_STATE_GATE ? true : false); + state == AMD_CG_STATE_GATE); break; default: break; @@ -1664,6 +1690,7 @@ static const struct amdgpu_ring_funcs sdma_v5_2_ring_funcs = { 10 + 10 + 10, /* sdma_v5_2_ring_emit_fence x3 for user fence, vm fence */ .emit_ib_size = 7 + 6, /* sdma_v5_2_ring_emit_ib */ .emit_ib = sdma_v5_2_ring_emit_ib, + .emit_mem_sync = sdma_v5_2_ring_emit_mem_sync, .emit_fence = sdma_v5_2_ring_emit_fence, .emit_pipeline_sync = sdma_v5_2_ring_emit_pipeline_sync, .emit_vm_flush = sdma_v5_2_ring_emit_vm_flush, diff --git a/drivers/gpu/drm/amd/amdgpu/si.c b/drivers/gpu/drm/amd/amdgpu/si.c index 6b5cf7882a12..7cbc2bb03bc6 100644 --- a/drivers/gpu/drm/amd/amdgpu/si.c +++ b/drivers/gpu/drm/amd/amdgpu/si.c @@ -26,6 +26,8 @@ #include <linux/module.h> #include <linux/pci.h> +#include <drm/amdgpu_drm.h> + #include "amdgpu.h" #include "amdgpu_atombios.h" #include "amdgpu_ih.h" @@ -905,6 +907,114 @@ static const u32 hainan_mgcg_cgcg_init[] = 0x3630, 0xfffffff0, 0x00000100, }; +/* XXX: update when we support VCE */ +#if 0 +/* tahiti, pitcarin, verde */ +static const struct amdgpu_video_codec_info tahiti_video_codecs_encode_array[] = +{ + { + .codec_type = AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_MPEG4_AVC, + .max_width = 2048, + .max_height = 1152, + .max_pixels_per_frame = 2048 * 1152, + .max_level = 0, + }, +}; + +static const struct amdgpu_video_codecs tahiti_video_codecs_encode = +{ + .codec_count = ARRAY_SIZE(tahiti_video_codecs_encode_array), + .codec_array = tahiti_video_codecs_encode_array, +}; +#else +static const struct amdgpu_video_codecs tahiti_video_codecs_encode = +{ + .codec_count = 0, + .codec_array = NULL, +}; +#endif +/* oland and hainan don't support encode */ +static const struct amdgpu_video_codecs hainan_video_codecs_encode = +{ + .codec_count = 0, + .codec_array = NULL, +}; + +/* tahiti, pitcarin, verde, oland */ +static const struct amdgpu_video_codec_info tahiti_video_codecs_decode_array[] = +{ + { + .codec_type = AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_MPEG2, + .max_width = 2048, + .max_height = 1152, + .max_pixels_per_frame = 2048 * 1152, + .max_level = 3, + }, + { + .codec_type = AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_MPEG4, + .max_width = 2048, + .max_height = 1152, + .max_pixels_per_frame = 2048 * 1152, + .max_level = 5, + }, + { + .codec_type = AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_MPEG4_AVC, + .max_width = 2048, + .max_height = 1152, + .max_pixels_per_frame = 2048 * 1152, + .max_level = 41, + }, + { + .codec_type = AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_VC1, + .max_width = 2048, + .max_height = 1152, + .max_pixels_per_frame = 2048 * 1152, + .max_level = 4, + }, +}; + +static const struct amdgpu_video_codecs tahiti_video_codecs_decode = +{ + .codec_count = ARRAY_SIZE(tahiti_video_codecs_decode_array), + .codec_array = tahiti_video_codecs_decode_array, +}; + +/* hainan doesn't support decode */ +static const struct amdgpu_video_codecs hainan_video_codecs_decode = +{ + .codec_count = 0, + .codec_array = NULL, +}; + +static int si_query_video_codecs(struct amdgpu_device *adev, bool encode, + const struct amdgpu_video_codecs **codecs) +{ + switch (adev->asic_type) { + case CHIP_VERDE: + case CHIP_TAHITI: + case CHIP_PITCAIRN: + if (encode) + *codecs = &tahiti_video_codecs_encode; + else + *codecs = &tahiti_video_codecs_decode; + return 0; + case CHIP_OLAND: + if (encode) + *codecs = &hainan_video_codecs_encode; + else + *codecs = &tahiti_video_codecs_decode; + return 0; + case CHIP_HAINAN: + if (encode) + *codecs = &hainan_video_codecs_encode; + else + *codecs = &hainan_video_codecs_decode; + return 0; + default: + return -EINVAL; + } +} + static u32 si_pcie_rreg(struct amdgpu_device *adev, u32 reg) { unsigned long flags; @@ -1903,6 +2013,7 @@ static const struct amdgpu_asic_funcs si_asic_funcs = .get_pcie_replay_count = &si_get_pcie_replay_count, .supports_baco = &si_asic_supports_baco, .pre_asic_init = &si_pre_asic_init, + .query_video_codecs = &si_query_video_codecs, }; static uint32_t si_get_rev_id(struct amdgpu_device *adev) diff --git a/drivers/gpu/drm/amd/amdgpu/si_dma.c b/drivers/gpu/drm/amd/amdgpu/si_dma.c index 488497ad5e0c..cb703e307238 100644 --- a/drivers/gpu/drm/amd/amdgpu/si_dma.c +++ b/drivers/gpu/drm/amd/amdgpu/si_dma.c @@ -507,10 +507,9 @@ static int si_dma_sw_init(void *handle) sprintf(ring->name, "sdma%d", i); r = amdgpu_ring_init(adev, ring, 1024, &adev->sdma.trap_irq, - (i == 0) ? - AMDGPU_SDMA_IRQ_INSTANCE0 : + (i == 0) ? AMDGPU_SDMA_IRQ_INSTANCE0 : AMDGPU_SDMA_IRQ_INSTANCE1, - AMDGPU_RING_PRIO_DEFAULT); + AMDGPU_RING_PRIO_DEFAULT, NULL); if (r) return r; } diff --git a/drivers/gpu/drm/amd/amdgpu/smuio_v13_0.c b/drivers/gpu/drm/amd/amdgpu/smuio_v13_0.c new file mode 100644 index 000000000000..079b094c48ad --- /dev/null +++ b/drivers/gpu/drm/amd/amdgpu/smuio_v13_0.c @@ -0,0 +1,121 @@ +/* + * Copyright 2020 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + */ +#include "amdgpu.h" +#include "smuio_v13_0.h" +#include "smuio/smuio_13_0_2_offset.h" +#include "smuio/smuio_13_0_2_sh_mask.h" + +#define SMUIO_MCM_CONFIG__HOST_GPU_XGMI_MASK 0x00000001L + +static u32 smuio_v13_0_get_rom_index_offset(struct amdgpu_device *adev) +{ + return SOC15_REG_OFFSET(SMUIO, 0, regROM_INDEX); +} + +static u32 smuio_v13_0_get_rom_data_offset(struct amdgpu_device *adev) +{ + return SOC15_REG_OFFSET(SMUIO, 0, regROM_DATA); +} + +static void smuio_v13_0_update_rom_clock_gating(struct amdgpu_device *adev, bool enable) +{ + u32 def, data; + + /* enable/disable ROM CG is not supported on APU */ + if (adev->flags & AMD_IS_APU) + return; + + def = data = RREG32_SOC15(SMUIO, 0, regCGTT_ROM_CLK_CTRL0); + + if (enable && (adev->cg_flags & AMD_CG_SUPPORT_ROM_MGCG)) + data &= ~(CGTT_ROM_CLK_CTRL0__SOFT_OVERRIDE0_MASK | + CGTT_ROM_CLK_CTRL0__SOFT_OVERRIDE1_MASK); + else + data |= CGTT_ROM_CLK_CTRL0__SOFT_OVERRIDE0_MASK | + CGTT_ROM_CLK_CTRL0__SOFT_OVERRIDE1_MASK; + + if (def != data) + WREG32_SOC15(SMUIO, 0, regCGTT_ROM_CLK_CTRL0, data); +} + +static void smuio_v13_0_get_clock_gating_state(struct amdgpu_device *adev, u32 *flags) +{ + u32 data; + + /* CGTT_ROM_CLK_CTRL0 is not available for APU */ + if (adev->flags & AMD_IS_APU) + return; + + data = RREG32_SOC15(SMUIO, 0, regCGTT_ROM_CLK_CTRL0); + if (!(data & CGTT_ROM_CLK_CTRL0__SOFT_OVERRIDE0_MASK)) + *flags |= AMD_CG_SUPPORT_ROM_MGCG; +} + +/** + * smuio_v13_0_get_die_id - query die id from FCH. + * + * @adev: amdgpu device pointer + * + * Returns die id + */ +static u32 smuio_v13_0_get_die_id(struct amdgpu_device *adev) +{ + u32 data, die_id; + + data = RREG32_SOC15(SMUIO, 0, regSMUIO_MCM_CONFIG); + die_id = REG_GET_FIELD(data, SMUIO_MCM_CONFIG, DIE_ID); + + return die_id; +} + +/** + * smuio_v13_0_supports_host_gpu_xgmi - detect xgmi interface between cpu and gpu/s. + * + * @adev: amdgpu device pointer + * + * Returns true on success or false otherwise. + */ +static bool smuio_v13_0_is_host_gpu_xgmi_supported(struct amdgpu_device *adev) +{ + u32 data; + + data = RREG32_SOC15(SMUIO, 0, regSMUIO_MCM_CONFIG); + data = REG_GET_FIELD(data, SMUIO_MCM_CONFIG, TOPOLOGY_ID); + /* data[4:0] + * bit 0 == 0 host-gpu interface is PCIE + * bit 0 == 1 host-gpu interface is Alternate Protocal + * for AMD, this is XGMI + */ + data &= SMUIO_MCM_CONFIG__HOST_GPU_XGMI_MASK; + + return data ? true : false; +} + +const struct amdgpu_smuio_funcs smuio_v13_0_funcs = { + .get_rom_index_offset = smuio_v13_0_get_rom_index_offset, + .get_rom_data_offset = smuio_v13_0_get_rom_data_offset, + .get_die_id = smuio_v13_0_get_die_id, + .is_host_gpu_xgmi_supported = smuio_v13_0_is_host_gpu_xgmi_supported, + .update_rom_clock_gating = smuio_v13_0_update_rom_clock_gating, + .get_clock_gating_state = smuio_v13_0_get_clock_gating_state, +}; diff --git a/drivers/gpu/drm/amd/amdgpu/smuio_v13_0.h b/drivers/gpu/drm/amd/amdgpu/smuio_v13_0.h new file mode 100644 index 000000000000..a3bfe3e4fb46 --- /dev/null +++ b/drivers/gpu/drm/amd/amdgpu/smuio_v13_0.h @@ -0,0 +1,30 @@ +/* + * Copyright 2020 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + */ +#ifndef __SMUIO_V13_0_H__ +#define __SMUIO_V13_0_H__ + +#include "soc15_common.h" + +extern const struct amdgpu_smuio_funcs smuio_v13_0_funcs; + +#endif /* __SMUIO_V13_0_H__ */ diff --git a/drivers/gpu/drm/amd/amdgpu/soc15.c b/drivers/gpu/drm/amd/amdgpu/soc15.c index 1221aa6b40a9..d80e12b80c7e 100644 --- a/drivers/gpu/drm/amd/amdgpu/soc15.c +++ b/drivers/gpu/drm/amd/amdgpu/soc15.c @@ -25,6 +25,8 @@ #include <linux/module.h> #include <linux/pci.h> +#include <drm/amdgpu_drm.h> + #include "amdgpu.h" #include "amdgpu_atombios.h" #include "amdgpu_ih.h" @@ -71,9 +73,9 @@ #include "jpeg_v2_5.h" #include "smuio_v9_0.h" #include "smuio_v11_0.h" +#include "smuio_v13_0.h" #include "dce_virtual.h" #include "mxgpu_ai.h" -#include "amdgpu_smu.h" #include "amdgpu_ras.h" #include "amdgpu_xgmi.h" #include <uapi/linux/kfd_ioctl.h> @@ -83,6 +85,234 @@ #define mmMP0_MISC_LIGHT_SLEEP_CTRL 0x01ba #define mmMP0_MISC_LIGHT_SLEEP_CTRL_BASE_IDX 0 +/* Vega, Raven, Arcturus */ +static const struct amdgpu_video_codec_info vega_video_codecs_encode_array[] = +{ + { + .codec_type = AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_MPEG4_AVC, + .max_width = 4096, + .max_height = 2304, + .max_pixels_per_frame = 4096 * 2304, + .max_level = 0, + }, + { + .codec_type = AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_HEVC, + .max_width = 4096, + .max_height = 2304, + .max_pixels_per_frame = 4096 * 2304, + .max_level = 0, + }, +}; + +static const struct amdgpu_video_codecs vega_video_codecs_encode = +{ + .codec_count = ARRAY_SIZE(vega_video_codecs_encode_array), + .codec_array = vega_video_codecs_encode_array, +}; + +/* Vega */ +static const struct amdgpu_video_codec_info vega_video_codecs_decode_array[] = +{ + { + .codec_type = AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_MPEG2, + .max_width = 4096, + .max_height = 4096, + .max_pixels_per_frame = 4096 * 4096, + .max_level = 3, + }, + { + .codec_type = AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_MPEG4, + .max_width = 4096, + .max_height = 4096, + .max_pixels_per_frame = 4096 * 4096, + .max_level = 5, + }, + { + .codec_type = AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_MPEG4_AVC, + .max_width = 4096, + .max_height = 4096, + .max_pixels_per_frame = 4096 * 4096, + .max_level = 52, + }, + { + .codec_type = AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_VC1, + .max_width = 4096, + .max_height = 4096, + .max_pixels_per_frame = 4096 * 4096, + .max_level = 4, + }, + { + .codec_type = AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_HEVC, + .max_width = 4096, + .max_height = 4096, + .max_pixels_per_frame = 4096 * 4096, + .max_level = 186, + }, + { + .codec_type = AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_JPEG, + .max_width = 4096, + .max_height = 4096, + .max_pixels_per_frame = 4096 * 4096, + .max_level = 0, + }, +}; + +static const struct amdgpu_video_codecs vega_video_codecs_decode = +{ + .codec_count = ARRAY_SIZE(vega_video_codecs_decode_array), + .codec_array = vega_video_codecs_decode_array, +}; + +/* Raven */ +static const struct amdgpu_video_codec_info rv_video_codecs_decode_array[] = +{ + { + .codec_type = AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_MPEG2, + .max_width = 4096, + .max_height = 4096, + .max_pixels_per_frame = 4096 * 4096, + .max_level = 3, + }, + { + .codec_type = AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_MPEG4, + .max_width = 4096, + .max_height = 4096, + .max_pixels_per_frame = 4096 * 4096, + .max_level = 5, + }, + { + .codec_type = AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_MPEG4_AVC, + .max_width = 4096, + .max_height = 4096, + .max_pixels_per_frame = 4096 * 4096, + .max_level = 52, + }, + { + .codec_type = AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_VC1, + .max_width = 4096, + .max_height = 4096, + .max_pixels_per_frame = 4096 * 4096, + .max_level = 4, + }, + { + .codec_type = AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_HEVC, + .max_width = 4096, + .max_height = 4096, + .max_pixels_per_frame = 4096 * 4096, + .max_level = 186, + }, + { + .codec_type = AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_JPEG, + .max_width = 4096, + .max_height = 4096, + .max_pixels_per_frame = 4096 * 4096, + .max_level = 0, + }, + { + .codec_type = AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_VP9, + .max_width = 4096, + .max_height = 4096, + .max_pixels_per_frame = 4096 * 4096, + .max_level = 0, + }, +}; + +static const struct amdgpu_video_codecs rv_video_codecs_decode = +{ + .codec_count = ARRAY_SIZE(rv_video_codecs_decode_array), + .codec_array = rv_video_codecs_decode_array, +}; + +/* Renoir, Arcturus */ +static const struct amdgpu_video_codec_info rn_video_codecs_decode_array[] = +{ + { + .codec_type = AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_MPEG2, + .max_width = 4096, + .max_height = 4096, + .max_pixels_per_frame = 4096 * 4096, + .max_level = 3, + }, + { + .codec_type = AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_MPEG4, + .max_width = 4096, + .max_height = 4096, + .max_pixels_per_frame = 4096 * 4096, + .max_level = 5, + }, + { + .codec_type = AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_MPEG4_AVC, + .max_width = 4096, + .max_height = 4096, + .max_pixels_per_frame = 4096 * 4096, + .max_level = 52, + }, + { + .codec_type = AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_VC1, + .max_width = 4096, + .max_height = 4096, + .max_pixels_per_frame = 4096 * 4096, + .max_level = 4, + }, + { + .codec_type = AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_HEVC, + .max_width = 8192, + .max_height = 4352, + .max_pixels_per_frame = 4096 * 4096, + .max_level = 186, + }, + { + .codec_type = AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_JPEG, + .max_width = 4096, + .max_height = 4096, + .max_pixels_per_frame = 4096 * 4096, + .max_level = 0, + }, + { + .codec_type = AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_VP9, + .max_width = 8192, + .max_height = 4352, + .max_pixels_per_frame = 4096 * 4096, + .max_level = 0, + }, +}; + +static const struct amdgpu_video_codecs rn_video_codecs_decode = +{ + .codec_count = ARRAY_SIZE(rn_video_codecs_decode_array), + .codec_array = rn_video_codecs_decode_array, +}; + +static int soc15_query_video_codecs(struct amdgpu_device *adev, bool encode, + const struct amdgpu_video_codecs **codecs) +{ + switch (adev->asic_type) { + case CHIP_VEGA20: + case CHIP_VEGA10: + case CHIP_VEGA12: + if (encode) + *codecs = &vega_video_codecs_encode; + else + *codecs = &vega_video_codecs_decode; + return 0; + case CHIP_RAVEN: + if (encode) + *codecs = &vega_video_codecs_encode; + else + *codecs = &rv_video_codecs_decode; + return 0; + case CHIP_ARCTURUS: + case CHIP_RENOIR: + if (encode) + *codecs = &vega_video_codecs_encode; + else + *codecs = &rn_video_codecs_decode; + return 0; + default: + return -EINVAL; + } +} + /* * Indirect registers accessor */ @@ -419,40 +649,6 @@ void soc15_program_register_sequence(struct amdgpu_device *adev, } -static int soc15_asic_mode1_reset(struct amdgpu_device *adev) -{ - u32 i; - int ret = 0; - - amdgpu_atombios_scratch_regs_engine_hung(adev, true); - - dev_info(adev->dev, "GPU mode1 reset\n"); - - /* disable BM */ - pci_clear_master(adev->pdev); - - amdgpu_device_cache_pci_state(adev->pdev); - - ret = psp_gpu_reset(adev); - if (ret) - dev_err(adev->dev, "GPU mode1 reset failed\n"); - - amdgpu_device_load_pci_state(adev->pdev); - - /* wait for asic to come out of reset */ - for (i = 0; i < adev->usec_timeout; i++) { - u32 memsize = adev->nbio.funcs->get_memsize(adev); - - if (memsize != 0xffffffff) - break; - udelay(1); - } - - amdgpu_atombios_scratch_regs_engine_hung(adev, false); - - return ret; -} - static int soc15_asic_baco_reset(struct amdgpu_device *adev) { struct amdgpu_ras *ras = amdgpu_ras_get_context(adev); @@ -477,13 +673,21 @@ static enum amd_reset_method soc15_asic_reset_method(struct amdgpu_device *adev) { bool baco_reset = false; + bool connected_to_cpu = false; struct amdgpu_ras *ras = amdgpu_ras_get_context(adev); + if (adev->gmc.xgmi.supported && adev->gmc.xgmi.connected_to_cpu) + connected_to_cpu = true; + if (amdgpu_reset_method == AMD_RESET_METHOD_MODE1 || amdgpu_reset_method == AMD_RESET_METHOD_MODE2 || amdgpu_reset_method == AMD_RESET_METHOD_BACO || - amdgpu_reset_method == AMD_RESET_METHOD_PCI) - return amdgpu_reset_method; + amdgpu_reset_method == AMD_RESET_METHOD_PCI) { + /* If connected to cpu, driver only support mode2 */ + if (connected_to_cpu) + return AMD_RESET_METHOD_MODE2; + return amdgpu_reset_method; + } if (amdgpu_reset_method != -1) dev_warn(adev->dev, "Specified reset method:%d isn't supported, using AUTO instead.\n", @@ -509,6 +713,14 @@ soc15_asic_reset_method(struct amdgpu_device *adev) if ((ras && ras->supported) && adev->pm.fw_version <= 0x283400) baco_reset = false; break; + case CHIP_ALDEBARAN: + /* + * 1.connected to cpu: driver issue mode2 reset + * 2.discret gpu: driver issue mode1 reset + */ + if (connected_to_cpu) + return AMD_RESET_METHOD_MODE2; + break; default: break; } @@ -538,7 +750,7 @@ static int soc15_asic_reset(struct amdgpu_device *adev) return amdgpu_dpm_mode2_reset(adev); default: dev_info(adev->dev, "MODE1 reset\n"); - return soc15_asic_mode1_reset(adev); + return amdgpu_device_mode1_reset(adev); } } @@ -604,11 +816,12 @@ static void soc15_pcie_gen3_enable(struct amdgpu_device *adev) static void soc15_program_aspm(struct amdgpu_device *adev) { - - if (amdgpu_aspm == 0) + if (amdgpu_aspm != 1) return; - /* todo */ + if (!(adev->flags & AMD_IS_APU) && + (adev->nbio.funcs->program_aspm)) + adev->nbio.funcs->program_aspm(adev); } static void soc15_enable_doorbell_aperture(struct amdgpu_device *adev, @@ -661,6 +874,9 @@ static void soc15_reg_base_init(struct amdgpu_device *adev) case CHIP_ARCTURUS: arct_reg_base_init(adev); break; + case CHIP_ALDEBARAN: + aldebaran_reg_base_init(adev); + break; default: DRM_ERROR("Unsupported asic type: %d!\n", adev->asic_type); break; @@ -683,14 +899,12 @@ int soc15_set_ip_blocks(struct amdgpu_device *adev) if (!amdgpu_sriov_vf(adev)) soc15_reg_base_init(adev); - if (adev->asic_type == CHIP_VEGA20 || adev->asic_type == CHIP_ARCTURUS) - adev->gmc.xgmi.supported = true; - if (adev->flags & AMD_IS_APU) { adev->nbio.funcs = &nbio_v7_0_funcs; adev->nbio.hdp_flush_reg = &nbio_v7_0_hdp_flush_reg; } else if (adev->asic_type == CHIP_VEGA20 || - adev->asic_type == CHIP_ARCTURUS) { + adev->asic_type == CHIP_ARCTURUS || + adev->asic_type == CHIP_ALDEBARAN) { adev->nbio.funcs = &nbio_v7_4_funcs; adev->nbio.hdp_flush_reg = &nbio_v7_4_hdp_flush_reg; } else { @@ -699,7 +913,9 @@ int soc15_set_ip_blocks(struct amdgpu_device *adev) } adev->hdp.funcs = &hdp_v4_0_funcs; - if (adev->asic_type == CHIP_VEGA20 || adev->asic_type == CHIP_ARCTURUS) + if (adev->asic_type == CHIP_VEGA20 || + adev->asic_type == CHIP_ARCTURUS || + adev->asic_type == CHIP_ALDEBARAN) adev->df.funcs = &df_v3_6_funcs; else adev->df.funcs = &df_v1_7_funcs; @@ -707,6 +923,8 @@ int soc15_set_ip_blocks(struct amdgpu_device *adev) if (adev->asic_type == CHIP_VEGA20 || adev->asic_type == CHIP_ARCTURUS) adev->smuio.funcs = &smuio_v11_0_funcs; + else if (adev->asic_type == CHIP_ALDEBARAN) + adev->smuio.funcs = &smuio_v13_0_funcs; else adev->smuio.funcs = &smuio_v9_0_funcs; @@ -826,6 +1044,27 @@ int soc15_set_ip_blocks(struct amdgpu_device *adev) amdgpu_device_ip_block_add(adev, &vcn_v2_0_ip_block); amdgpu_device_ip_block_add(adev, &jpeg_v2_0_ip_block); break; + case CHIP_ALDEBARAN: + amdgpu_device_ip_block_add(adev, &vega10_common_ip_block); + amdgpu_device_ip_block_add(adev, &gmc_v9_0_ip_block); + + if (amdgpu_sriov_vf(adev)) { + if (likely(adev->firmware.load_type == AMDGPU_FW_LOAD_PSP)) + amdgpu_device_ip_block_add(adev, &psp_v13_0_ip_block); + amdgpu_device_ip_block_add(adev, &vega20_ih_ip_block); + } else { + amdgpu_device_ip_block_add(adev, &vega20_ih_ip_block); + if (likely(adev->firmware.load_type == AMDGPU_FW_LOAD_PSP)) + amdgpu_device_ip_block_add(adev, &psp_v13_0_ip_block); + } + + amdgpu_device_ip_block_add(adev, &gfx_v9_0_ip_block); + amdgpu_device_ip_block_add(adev, &sdma_v4_0_ip_block); + + amdgpu_device_ip_block_add(adev, &smu_v13_0_ip_block); + amdgpu_device_ip_block_add(adev, &vcn_v2_6_ip_block); + amdgpu_device_ip_block_add(adev, &jpeg_v2_6_ip_block); + break; default: return -EINVAL; } @@ -994,6 +1233,7 @@ static const struct amdgpu_asic_funcs soc15_asic_funcs = .get_pcie_replay_count = &soc15_get_pcie_replay_count, .supports_baco = &soc15_supports_baco, .pre_asic_init = &soc15_pre_asic_init, + .query_video_codecs = &soc15_query_video_codecs, }; static const struct amdgpu_asic_funcs vega20_asic_funcs = @@ -1015,6 +1255,7 @@ static const struct amdgpu_asic_funcs vega20_asic_funcs = .get_pcie_replay_count = &soc15_get_pcie_replay_count, .supports_baco = &soc15_supports_baco, .pre_asic_init = &soc15_pre_asic_init, + .query_video_codecs = &soc15_query_video_codecs, }; static int soc15_common_early_init(void *handle) @@ -1244,6 +1485,21 @@ static int soc15_common_early_init(void *handle) AMD_PG_SUPPORT_JPEG | AMD_PG_SUPPORT_VCN_DPG; break; + case CHIP_ALDEBARAN: + adev->asic_funcs = &vega20_asic_funcs; + adev->cg_flags = AMD_CG_SUPPORT_GFX_MGCG | + AMD_CG_SUPPORT_GFX_MGLS | + AMD_CG_SUPPORT_GFX_CGCG | + AMD_CG_SUPPORT_GFX_CGLS | + AMD_CG_SUPPORT_GFX_CP_LS | + AMD_CG_SUPPORT_HDP_LS | + AMD_CG_SUPPORT_SDMA_MGCG | + AMD_CG_SUPPORT_SDMA_LS | + AMD_CG_SUPPORT_IH_CG | + AMD_CG_SUPPORT_VCN_MGCG | AMD_CG_SUPPORT_JPEG_MGCG; + adev->pg_flags = AMD_PG_SUPPORT_VCN_DPG; + adev->external_rev_id = adev->rev_id + 0x3c; + break; default: /* FIXME: not supported yet */ return -EINVAL; @@ -1268,8 +1524,9 @@ static int soc15_common_late_init(void *handle) if (adev->hdp.funcs->reset_ras_error_count) adev->hdp.funcs->reset_ras_error_count(adev); - if (adev->nbio.funcs->ras_late_init) - r = adev->nbio.funcs->ras_late_init(adev); + if (adev->nbio.ras_funcs && + adev->nbio.ras_funcs->ras_late_init) + r = adev->nbio.ras_funcs->ras_late_init(adev); return r; } @@ -1290,7 +1547,9 @@ static int soc15_common_sw_fini(void *handle) { struct amdgpu_device *adev = (struct amdgpu_device *)handle; - amdgpu_nbio_ras_fini(adev); + if (adev->nbio.ras_funcs && + adev->nbio.ras_funcs->ras_fini) + adev->nbio.ras_funcs->ras_fini(adev); adev->df.funcs->sw_fini(adev); return 0; } @@ -1354,9 +1613,11 @@ static int soc15_common_hw_fini(void *handle) if (adev->nbio.ras_if && amdgpu_ras_is_supported(adev, adev->nbio.ras_if->block)) { - if (adev->nbio.funcs->init_ras_controller_interrupt) + if (adev->nbio.ras_funcs && + adev->nbio.ras_funcs->init_ras_controller_interrupt) amdgpu_irq_put(adev, &adev->nbio.ras_controller_irq, 0); - if (adev->nbio.funcs->init_ras_err_event_athub_interrupt) + if (adev->nbio.ras_funcs && + adev->nbio.ras_funcs->init_ras_err_event_athub_interrupt) amdgpu_irq_put(adev, &adev->nbio.ras_err_event_athub_irq, 0); } @@ -1477,6 +1738,7 @@ static int soc15_common_set_clockgating_state(void *handle, state == AMD_CG_STATE_GATE); break; case CHIP_ARCTURUS: + case CHIP_ALDEBARAN: adev->hdp.funcs->update_clock_gating(adev, state == AMD_CG_STATE_GATE); break; @@ -1498,15 +1760,18 @@ static void soc15_common_get_clockgating_state(void *handle, u32 *flags) adev->hdp.funcs->get_clock_gating_state(adev, flags); - /* AMD_CG_SUPPORT_DRM_MGCG */ - data = RREG32(SOC15_REG_OFFSET(MP0, 0, mmMP0_MISC_CGTT_CTRL0)); - if (!(data & 0x01000000)) - *flags |= AMD_CG_SUPPORT_DRM_MGCG; + if (adev->asic_type != CHIP_ALDEBARAN) { + + /* AMD_CG_SUPPORT_DRM_MGCG */ + data = RREG32(SOC15_REG_OFFSET(MP0, 0, mmMP0_MISC_CGTT_CTRL0)); + if (!(data & 0x01000000)) + *flags |= AMD_CG_SUPPORT_DRM_MGCG; - /* AMD_CG_SUPPORT_DRM_LS */ - data = RREG32(SOC15_REG_OFFSET(MP0, 0, mmMP0_MISC_LIGHT_SLEEP_CTRL)); - if (data & 0x1) - *flags |= AMD_CG_SUPPORT_DRM_LS; + /* AMD_CG_SUPPORT_DRM_LS */ + data = RREG32(SOC15_REG_OFFSET(MP0, 0, mmMP0_MISC_LIGHT_SLEEP_CTRL)); + if (data & 0x1) + *flags |= AMD_CG_SUPPORT_DRM_LS; + } /* AMD_CG_SUPPORT_ROM_MGCG */ adev->smuio.funcs->get_clock_gating_state(adev, flags); diff --git a/drivers/gpu/drm/amd/amdgpu/soc15.h b/drivers/gpu/drm/amd/amdgpu/soc15.h index 8f38f047265b..034cfdfc4dbe 100644 --- a/drivers/gpu/drm/amd/amdgpu/soc15.h +++ b/drivers/gpu/drm/amd/amdgpu/soc15.h @@ -49,6 +49,13 @@ struct soc15_reg_rlcg { u32 reg; }; +struct soc15_reg { + uint32_t hwip; + uint32_t inst; + uint32_t seg; + uint32_t reg_offset; +}; + struct soc15_reg_entry { uint32_t hwip; uint32_t inst; @@ -88,6 +95,10 @@ struct soc15_ras_field_entry { #define SOC15_REG_FIELD(reg, field) reg##__##field##_MASK, reg##__##field##__SHIFT +#define SOC15_REG_FIELD_VAL(val, mask, shift) (((val) & mask) >> shift) + +#define SOC15_RAS_REG_FIELD_VAL(val, entry, field) SOC15_REG_FIELD_VAL((val), (entry).field##_count_mask, (entry).field##_count_shift) + void soc15_grbm_select(struct amdgpu_device *adev, u32 me, u32 pipe, u32 queue, u32 vmid); void soc15_set_virt_ops(struct amdgpu_device *adev); @@ -100,6 +111,7 @@ void soc15_program_register_sequence(struct amdgpu_device *adev, int vega10_reg_base_init(struct amdgpu_device *adev); int vega20_reg_base_init(struct amdgpu_device *adev); int arct_reg_base_init(struct amdgpu_device *adev); +int aldebaran_reg_base_init(struct amdgpu_device *adev); void vega10_doorbell_index_init(struct amdgpu_device *adev); void vega20_doorbell_index_init(struct amdgpu_device *adev); diff --git a/drivers/gpu/drm/amd/amdgpu/soc15_common.h b/drivers/gpu/drm/amd/amdgpu/soc15_common.h index a5c00ab8b021..14bd794bbea6 100644 --- a/drivers/gpu/drm/amd/amdgpu/soc15_common.h +++ b/drivers/gpu/drm/amd/amdgpu/soc15_common.h @@ -77,13 +77,21 @@ }) #define WREG32_RLC(reg, value) \ + do { \ + if (adev->gfx.rlc.funcs->rlcg_wreg) \ + adev->gfx.rlc.funcs->rlcg_wreg(adev, reg, value, 0); \ + else \ + WREG32(reg, value); \ + } while (0) + +#define WREG32_RLC_EX(prefix, reg, value) \ do { \ if (amdgpu_sriov_fullaccess(adev)) { \ uint32_t i = 0; \ uint32_t retries = 50000; \ - uint32_t r0 = adev->reg_offset[GC_HWIP][0][mmSCRATCH_REG0_BASE_IDX] + mmSCRATCH_REG0; \ - uint32_t r1 = adev->reg_offset[GC_HWIP][0][mmSCRATCH_REG1_BASE_IDX] + mmSCRATCH_REG1; \ - uint32_t spare_int = adev->reg_offset[GC_HWIP][0][mmRLC_SPARE_INT_BASE_IDX] + mmRLC_SPARE_INT; \ + uint32_t r0 = adev->reg_offset[GC_HWIP][0][prefix##SCRATCH_REG0_BASE_IDX] + prefix##SCRATCH_REG0; \ + uint32_t r1 = adev->reg_offset[GC_HWIP][0][prefix##SCRATCH_REG1_BASE_IDX] + prefix##SCRATCH_REG1; \ + uint32_t spare_int = adev->reg_offset[GC_HWIP][0][prefix##RLC_SPARE_INT_BASE_IDX] + prefix##RLC_SPARE_INT; \ WREG32(r0, value); \ WREG32(r1, (reg | 0x80000000)); \ WREG32(spare_int, 0x1); \ @@ -101,13 +109,32 @@ } while (0) #define WREG32_SOC15_RLC_SHADOW(ip, inst, reg, value) \ + WREG32_RLC((adev->reg_offset[ip##_HWIP][inst][reg##_BASE_IDX] + reg), value) + +#define RREG32_RLC(reg) \ + (adev->gfx.rlc.funcs->rlcg_rreg ? \ + adev->gfx.rlc.funcs->rlcg_rreg(adev, reg, 0) : RREG32(reg)) + +#define WREG32_RLC_NO_KIQ(reg, value) \ + do { \ + if (adev->gfx.rlc.funcs->rlcg_wreg) \ + adev->gfx.rlc.funcs->rlcg_wreg(adev, reg, value, AMDGPU_REGS_NO_KIQ); \ + else \ + WREG32_NO_KIQ(reg, value); \ + } while (0) + +#define RREG32_RLC_NO_KIQ(reg) \ + (adev->gfx.rlc.funcs->rlcg_rreg ? \ + adev->gfx.rlc.funcs->rlcg_rreg(adev, reg, AMDGPU_REGS_NO_KIQ) : RREG32_NO_KIQ(reg)) + +#define WREG32_SOC15_RLC_SHADOW_EX(prefix, ip, inst, reg, value) \ do { \ uint32_t target_reg = adev->reg_offset[ip##_HWIP][inst][reg##_BASE_IDX] + reg;\ if (amdgpu_sriov_fullaccess(adev)) { \ - uint32_t r2 = adev->reg_offset[GC_HWIP][0][mmSCRATCH_REG1_BASE_IDX] + mmSCRATCH_REG2; \ - uint32_t r3 = adev->reg_offset[GC_HWIP][0][mmSCRATCH_REG1_BASE_IDX] + mmSCRATCH_REG3; \ - uint32_t grbm_cntl = adev->reg_offset[GC_HWIP][0][mmGRBM_GFX_CNTL_BASE_IDX] + mmGRBM_GFX_CNTL; \ - uint32_t grbm_idx = adev->reg_offset[GC_HWIP][0][mmGRBM_GFX_INDEX_BASE_IDX] + mmGRBM_GFX_INDEX; \ + uint32_t r2 = adev->reg_offset[GC_HWIP][0][prefix##SCRATCH_REG1_BASE_IDX] + prefix##SCRATCH_REG2; \ + uint32_t r3 = adev->reg_offset[GC_HWIP][0][prefix##SCRATCH_REG1_BASE_IDX] + prefix##SCRATCH_REG3; \ + uint32_t grbm_cntl = adev->reg_offset[GC_HWIP][0][prefix##GRBM_GFX_CNTL_BASE_IDX] + prefix##GRBM_GFX_CNTL; \ + uint32_t grbm_idx = adev->reg_offset[GC_HWIP][0][prefix##GRBM_GFX_INDEX_BASE_IDX] + prefix##GRBM_GFX_INDEX; \ if (target_reg == grbm_cntl) \ WREG32(r2, value); \ else if (target_reg == grbm_idx) \ @@ -118,18 +145,30 @@ } \ } while (0) +#define RREG32_SOC15_RLC(ip, inst, reg) \ + RREG32_RLC(adev->reg_offset[ip##_HWIP][inst][reg##_BASE_IDX] + reg) + #define WREG32_SOC15_RLC(ip, inst, reg, value) \ do { \ + uint32_t target_reg = adev->reg_offset[ip##_HWIP][0][reg##_BASE_IDX] + reg;\ + WREG32_RLC(target_reg, value); \ + } while (0) + +#define WREG32_SOC15_RLC_EX(prefix, ip, inst, reg, value) \ + do { \ uint32_t target_reg = adev->reg_offset[GC_HWIP][0][reg##_BASE_IDX] + reg;\ - WREG32_RLC(target_reg, value); \ + WREG32_RLC_EX(prefix, target_reg, value); \ } while (0) #define WREG32_FIELD15_RLC(ip, idx, reg, field, val) \ - WREG32_RLC((adev->reg_offset[ip##_HWIP][idx][mm##reg##_BASE_IDX] + mm##reg), \ - (RREG32(adev->reg_offset[ip##_HWIP][idx][mm##reg##_BASE_IDX] + mm##reg) \ - & ~REG_FIELD_MASK(reg, field)) | (val) << REG_FIELD_SHIFT(reg, field)) + WREG32_RLC((adev->reg_offset[ip##_HWIP][idx][mm##reg##_BASE_IDX] + mm##reg), \ + (RREG32_RLC(adev->reg_offset[ip##_HWIP][idx][mm##reg##_BASE_IDX] + mm##reg) \ + & ~REG_FIELD_MASK(reg, field)) | (val) << REG_FIELD_SHIFT(reg, field)) #define WREG32_SOC15_OFFSET_RLC(ip, inst, reg, offset, value) \ - WREG32_RLC(((adev->reg_offset[ip##_HWIP][inst][reg##_BASE_IDX] + reg) + offset), value) + WREG32_RLC(((adev->reg_offset[ip##_HWIP][inst][reg##_BASE_IDX] + reg) + offset), value) + +#define RREG32_SOC15_OFFSET_RLC(ip, inst, reg, offset) \ + RREG32_RLC(((adev->reg_offset[ip##_HWIP][inst][reg##_BASE_IDX] + reg) + offset)) #endif diff --git a/drivers/gpu/drm/amd/amdgpu/ta_secureDisplay_if.h b/drivers/gpu/drm/amd/amdgpu/ta_secureDisplay_if.h index 5039375bb1d4..cf8ff064dc72 100644 --- a/drivers/gpu/drm/amd/amdgpu/ta_secureDisplay_if.h +++ b/drivers/gpu/drm/amd/amdgpu/ta_secureDisplay_if.h @@ -50,6 +50,7 @@ enum ta_securedisplay_status { TA_SECUREDISPLAY_STATUS__I2C_WRITE_ERROR = 0x04, /* Fail to Write to I2C */ TA_SECUREDISPLAY_STATUS__READ_DIO_SCRATCH_ERROR = 0x05, /*Fail Read DIO Scratch Register*/ TA_SECUREDISPLAY_STATUS__READ_CRC_ERROR = 0x06, /* Fail to Read CRC*/ + TA_SECUREDISPLAY_STATUS__I2C_INIT_ERROR = 0x07, /* Failed to initialize I2C */ TA_SECUREDISPLAY_STATUS__MAX = 0x7FFFFFFF,/* Maximum Value for status*/ }; diff --git a/drivers/gpu/drm/amd/amdgpu/umc_v6_1.c b/drivers/gpu/drm/amd/amdgpu/umc_v6_1.c index 96d7769609f4..20b44983ac94 100644 --- a/drivers/gpu/drm/amd/amdgpu/umc_v6_1.c +++ b/drivers/gpu/drm/amd/amdgpu/umc_v6_1.c @@ -22,6 +22,7 @@ */ #include "umc_v6_1.h" #include "amdgpu_ras.h" +#include "amdgpu_umc.h" #include "amdgpu.h" #include "rsmu/rsmu_0_0_2_offset.h" @@ -464,9 +465,10 @@ static void umc_v6_1_err_cnt_init(struct amdgpu_device *adev) umc_v6_1_enable_umc_index_mode(adev); } -const struct amdgpu_umc_funcs umc_v6_1_funcs = { +const struct amdgpu_umc_ras_funcs umc_v6_1_ras_funcs = { .err_cnt_init = umc_v6_1_err_cnt_init, .ras_late_init = amdgpu_umc_ras_late_init, + .ras_fini = amdgpu_umc_ras_fini, .query_ras_error_count = umc_v6_1_query_ras_error_count, .query_ras_error_address = umc_v6_1_query_ras_error_address, }; diff --git a/drivers/gpu/drm/amd/amdgpu/umc_v6_1.h b/drivers/gpu/drm/amd/amdgpu/umc_v6_1.h index 0ce1d323cfdd..5dc36c730bb2 100644 --- a/drivers/gpu/drm/amd/amdgpu/umc_v6_1.h +++ b/drivers/gpu/drm/amd/amdgpu/umc_v6_1.h @@ -45,7 +45,7 @@ /* umc ce count initial value */ #define UMC_V6_1_CE_CNT_INIT (UMC_V6_1_CE_CNT_MAX - UMC_V6_1_CE_INT_THRESHOLD) -extern const struct amdgpu_umc_funcs umc_v6_1_funcs; +extern const struct amdgpu_umc_ras_funcs umc_v6_1_ras_funcs; extern const uint32_t umc_v6_1_channel_idx_tbl[UMC_V6_1_UMC_INSTANCE_NUM][UMC_V6_1_CHANNEL_INSTANCE_NUM]; diff --git a/drivers/gpu/drm/amd/amdgpu/umc_v6_7.c b/drivers/gpu/drm/amd/amdgpu/umc_v6_7.c new file mode 100644 index 000000000000..3a8f787374c0 --- /dev/null +++ b/drivers/gpu/drm/amd/amdgpu/umc_v6_7.c @@ -0,0 +1,281 @@ +/* + * Copyright 2021 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + */ +#include "umc_v6_7.h" +#include "amdgpu_ras.h" +#include "amdgpu_umc.h" +#include "amdgpu.h" + +#include "umc/umc_6_7_0_offset.h" +#include "umc/umc_6_7_0_sh_mask.h" + +static inline uint32_t get_umc_v6_7_reg_offset(struct amdgpu_device *adev, + uint32_t umc_inst, + uint32_t ch_inst) +{ + return adev->umc.channel_offs * ch_inst + UMC_V6_7_INST_DIST * umc_inst; +} + +static void umc_v6_7_query_correctable_error_count(struct amdgpu_device *adev, + uint32_t umc_reg_offset, + unsigned long *error_count) +{ + uint32_t ecc_err_cnt_sel, ecc_err_cnt_sel_addr; + uint32_t ecc_err_cnt, ecc_err_cnt_addr; + uint64_t mc_umc_status; + uint32_t mc_umc_status_addr; + + /* UMC 6_1_1 registers */ + ecc_err_cnt_sel_addr = + SOC15_REG_OFFSET(UMC, 0, regUMCCH0_0_EccErrCntSel); + ecc_err_cnt_addr = + SOC15_REG_OFFSET(UMC, 0, regUMCCH0_0_EccErrCnt); + mc_umc_status_addr = + SOC15_REG_OFFSET(UMC, 0, regMCA_UMC_UMC0_MCUMC_STATUST0); + + /* select the lower chip and check the error count */ + ecc_err_cnt_sel = RREG32_PCIE((ecc_err_cnt_sel_addr + umc_reg_offset) * 4); + ecc_err_cnt_sel = REG_SET_FIELD(ecc_err_cnt_sel, UMCCH0_0_EccErrCntSel, + EccErrCntCsSel, 0); + WREG32_PCIE((ecc_err_cnt_sel_addr + umc_reg_offset) * 4, ecc_err_cnt_sel); + + ecc_err_cnt = RREG32_PCIE((ecc_err_cnt_addr + umc_reg_offset) * 4); + *error_count += + (REG_GET_FIELD(ecc_err_cnt, UMCCH0_0_EccErrCnt, EccErrCnt) - + UMC_V6_7_CE_CNT_INIT); + + /* select the higher chip and check the err counter */ + ecc_err_cnt_sel = REG_SET_FIELD(ecc_err_cnt_sel, UMCCH0_0_EccErrCntSel, + EccErrCntCsSel, 1); + WREG32_PCIE((ecc_err_cnt_sel_addr + umc_reg_offset) * 4, ecc_err_cnt_sel); + + ecc_err_cnt = RREG32_PCIE((ecc_err_cnt_addr + umc_reg_offset) * 4); + *error_count += + (REG_GET_FIELD(ecc_err_cnt, UMCCH0_0_EccErrCnt, EccErrCnt) - + UMC_V6_7_CE_CNT_INIT); + + /* check for SRAM correctable error + MCUMC_STATUS is a 64 bit register */ + mc_umc_status = RREG64_PCIE((mc_umc_status_addr + umc_reg_offset) * 4); + if (REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, Val) == 1 && + REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, CECC) == 1) + *error_count += 1; +} + +static void umc_v6_7_querry_uncorrectable_error_count(struct amdgpu_device *adev, + uint32_t umc_reg_offset, + unsigned long *error_count) +{ + uint64_t mc_umc_status; + uint32_t mc_umc_status_addr; + + mc_umc_status_addr = + SOC15_REG_OFFSET(UMC, 0, regMCA_UMC_UMC0_MCUMC_STATUST0); + + /* check the MCUMC_STATUS */ + mc_umc_status = RREG64_PCIE((mc_umc_status_addr + umc_reg_offset) * 4); + if ((REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, Val) == 1) && + (REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, Deferred) == 1 || + REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, UECC) == 1 || + REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, PCC) == 1 || + REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, UC) == 1 || + REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, TCC) == 1)) + *error_count += 1; +} + +static void umc_v6_7_reset_error_count_per_channel(struct amdgpu_device *adev, + uint32_t umc_reg_offset) +{ + uint32_t ecc_err_cnt_addr; + uint32_t ecc_err_cnt_sel, ecc_err_cnt_sel_addr; + + ecc_err_cnt_sel_addr = + SOC15_REG_OFFSET(UMC, 0, + regUMCCH0_0_EccErrCntSel); + ecc_err_cnt_addr = + SOC15_REG_OFFSET(UMC, 0, + regUMCCH0_0_EccErrCnt); + + /* select the lower chip */ + ecc_err_cnt_sel = RREG32_PCIE((ecc_err_cnt_sel_addr + + umc_reg_offset) * 4); + ecc_err_cnt_sel = REG_SET_FIELD(ecc_err_cnt_sel, + UMCCH0_0_EccErrCntSel, + EccErrCntCsSel, 0); + WREG32_PCIE((ecc_err_cnt_sel_addr + umc_reg_offset) * 4, + ecc_err_cnt_sel); + + /* clear lower chip error count */ + WREG32_PCIE((ecc_err_cnt_addr + umc_reg_offset) * 4, + UMC_V6_7_CE_CNT_INIT); + + /* select the higher chip */ + ecc_err_cnt_sel = RREG32_PCIE((ecc_err_cnt_sel_addr + + umc_reg_offset) * 4); + ecc_err_cnt_sel = REG_SET_FIELD(ecc_err_cnt_sel, + UMCCH0_0_EccErrCntSel, + EccErrCntCsSel, 1); + WREG32_PCIE((ecc_err_cnt_sel_addr + umc_reg_offset) * 4, + ecc_err_cnt_sel); + + /* clear higher chip error count */ + WREG32_PCIE((ecc_err_cnt_addr + umc_reg_offset) * 4, + UMC_V6_7_CE_CNT_INIT); +} + +static void umc_v6_7_reset_error_count(struct amdgpu_device *adev) +{ + uint32_t umc_inst = 0; + uint32_t ch_inst = 0; + uint32_t umc_reg_offset = 0; + + LOOP_UMC_INST_AND_CH(umc_inst, ch_inst) { + umc_reg_offset = get_umc_v6_7_reg_offset(adev, + umc_inst, + ch_inst); + + umc_v6_7_reset_error_count_per_channel(adev, + umc_reg_offset); + } +} + +static void umc_v6_7_query_ras_error_count(struct amdgpu_device *adev, + void *ras_error_status) +{ + struct ras_err_data *err_data = (struct ras_err_data *)ras_error_status; + + uint32_t umc_inst = 0; + uint32_t ch_inst = 0; + uint32_t umc_reg_offset = 0; + + /*TODO: driver needs to toggle DF Cstate to ensure + * safe access of UMC registers. Will add the protection */ + LOOP_UMC_INST_AND_CH(umc_inst, ch_inst) { + umc_reg_offset = get_umc_v6_7_reg_offset(adev, + umc_inst, + ch_inst); + umc_v6_7_query_correctable_error_count(adev, + umc_reg_offset, + &(err_data->ce_count)); + umc_v6_7_querry_uncorrectable_error_count(adev, + umc_reg_offset, + &(err_data->ue_count)); + } + + umc_v6_7_reset_error_count(adev); +} + +static void umc_v6_7_query_error_address(struct amdgpu_device *adev, + struct ras_err_data *err_data, + uint32_t umc_reg_offset, + uint32_t ch_inst, + uint32_t umc_inst) +{ + uint32_t mc_umc_status_addr; + uint64_t mc_umc_status, err_addr, retired_page, mc_umc_addrt0; + struct eeprom_table_record *err_rec; + uint32_t channel_index; + + mc_umc_status_addr = + SOC15_REG_OFFSET(UMC, 0, regMCA_UMC_UMC0_MCUMC_STATUST0); + mc_umc_addrt0 = + SOC15_REG_OFFSET(UMC, 0, regMCA_UMC_UMC0_MCUMC_ADDRT0); + + mc_umc_status = RREG64_PCIE((mc_umc_status_addr + umc_reg_offset) * 4); + + if (mc_umc_status == 0) + return; + + if (!err_data->err_addr) { + /* clear umc status */ + WREG64_PCIE((mc_umc_status_addr + umc_reg_offset) * 4, 0x0ULL); + return; + } + + err_rec = &err_data->err_addr[err_data->err_addr_cnt]; + + channel_index = + adev->umc.channel_idx_tbl[umc_inst * adev->umc.channel_inst_num + ch_inst]; + + /* calculate error address if ue/ce error is detected */ + if (REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, Val) == 1 && + (REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, UECC) == 1 || + REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, CECC) == 1)) { + + err_addr = RREG64_PCIE((mc_umc_addrt0 + umc_reg_offset) * 4); + err_addr = REG_GET_FIELD(err_addr, MCA_UMC_UMC0_MCUMC_ADDRT0, ErrorAddr); + + /* translate umc channel address to soc pa, 3 parts are included */ + retired_page = ADDR_OF_8KB_BLOCK(err_addr) | + ADDR_OF_256B_BLOCK(channel_index) | + OFFSET_IN_256B_BLOCK(err_addr); + + /* we only save ue error information currently, ce is skipped */ + if (REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, UECC) + == 1) { + err_rec->address = err_addr; + /* page frame address is saved */ + err_rec->retired_page = retired_page >> AMDGPU_GPU_PAGE_SHIFT; + err_rec->ts = (uint64_t)ktime_get_real_seconds(); + err_rec->err_type = AMDGPU_RAS_EEPROM_ERR_NON_RECOVERABLE; + err_rec->cu = 0; + err_rec->mem_channel = channel_index; + err_rec->mcumc_id = umc_inst; + + err_data->err_addr_cnt++; + } + } + + /* clear umc status */ + WREG64_PCIE((mc_umc_status_addr + umc_reg_offset) * 4, 0x0ULL); +} + +static void umc_v6_7_query_ras_error_address(struct amdgpu_device *adev, + void *ras_error_status) +{ + struct ras_err_data *err_data = (struct ras_err_data *)ras_error_status; + + uint32_t umc_inst = 0; + uint32_t ch_inst = 0; + uint32_t umc_reg_offset = 0; + + /*TODO: driver needs to toggle DF Cstate to ensure + * safe access of UMC resgisters. Will add the protection + * when firmware interface is ready */ + LOOP_UMC_INST_AND_CH(umc_inst, ch_inst) { + umc_reg_offset = get_umc_v6_7_reg_offset(adev, + umc_inst, + ch_inst); + umc_v6_7_query_error_address(adev, + err_data, + umc_reg_offset, + ch_inst, + umc_inst); + } +} + +const struct amdgpu_umc_ras_funcs umc_v6_7_ras_funcs = { + .ras_late_init = amdgpu_umc_ras_late_init, + .ras_fini = amdgpu_umc_ras_fini, + .query_ras_error_count = umc_v6_7_query_ras_error_count, + .query_ras_error_address = umc_v6_7_query_ras_error_address, +}; diff --git a/drivers/gpu/drm/amd/amdgpu/umc_v6_7.h b/drivers/gpu/drm/amd/amdgpu/umc_v6_7.h new file mode 100644 index 000000000000..4eb85f247e96 --- /dev/null +++ b/drivers/gpu/drm/amd/amdgpu/umc_v6_7.h @@ -0,0 +1,37 @@ +/* + * Copyright 2021 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + */ +#ifndef __UMC_V6_7_H__ +#define __UMC_V6_7_H__ + +/* EccErrCnt max value */ +#define UMC_V6_7_CE_CNT_MAX 0xffff +/* umc ce interrupt threshold */ +#define UMC_V6_7_CE_INT_THRESHOLD 0xffff +/* umc ce count initial value */ +#define UMC_V6_7_CE_CNT_INIT (UMC_V6_7_CE_CNT_MAX - UMC_V6_7_CE_INT_THRESHOLD) + +#define UMC_V6_7_INST_DIST 0x40000 + +extern const struct amdgpu_umc_ras_funcs umc_v6_7_ras_funcs; + +#endif diff --git a/drivers/gpu/drm/amd/amdgpu/umc_v8_7.c b/drivers/gpu/drm/amd/amdgpu/umc_v8_7.c index a064c097690c..89d20adfa001 100644 --- a/drivers/gpu/drm/amd/amdgpu/umc_v8_7.c +++ b/drivers/gpu/drm/amd/amdgpu/umc_v8_7.c @@ -22,6 +22,7 @@ */ #include "umc_v8_7.h" #include "amdgpu_ras.h" +#include "amdgpu_umc.h" #include "amdgpu.h" #include "rsmu/rsmu_0_0_2_offset.h" @@ -323,9 +324,10 @@ static void umc_v8_7_err_cnt_init(struct amdgpu_device *adev) } } -const struct amdgpu_umc_funcs umc_v8_7_funcs = { +const struct amdgpu_umc_ras_funcs umc_v8_7_ras_funcs = { .err_cnt_init = umc_v8_7_err_cnt_init, .ras_late_init = amdgpu_umc_ras_late_init, + .ras_fini = amdgpu_umc_ras_fini, .query_ras_error_count = umc_v8_7_query_ras_error_count, .query_ras_error_address = umc_v8_7_query_ras_error_address, }; diff --git a/drivers/gpu/drm/amd/amdgpu/umc_v8_7.h b/drivers/gpu/drm/amd/amdgpu/umc_v8_7.h index d4d0468e3df5..37e6dc7c28e0 100644 --- a/drivers/gpu/drm/amd/amdgpu/umc_v8_7.h +++ b/drivers/gpu/drm/amd/amdgpu/umc_v8_7.h @@ -44,7 +44,7 @@ /* umc ce count initial value */ #define UMC_V8_7_CE_CNT_INIT (UMC_V8_7_CE_CNT_MAX - UMC_V8_7_CE_INT_THRESHOLD) -extern const struct amdgpu_umc_funcs umc_v8_7_funcs; +extern const struct amdgpu_umc_ras_funcs umc_v8_7_ras_funcs; extern const uint32_t umc_v8_7_channel_idx_tbl[UMC_V8_7_UMC_INSTANCE_NUM][UMC_V8_7_CHANNEL_INSTANCE_NUM]; diff --git a/drivers/gpu/drm/amd/amdgpu/uvd_v3_1.c b/drivers/gpu/drm/amd/amdgpu/uvd_v3_1.c index 10ecae257b18..284447d7a579 100644 --- a/drivers/gpu/drm/amd/amdgpu/uvd_v3_1.c +++ b/drivers/gpu/drm/amd/amdgpu/uvd_v3_1.c @@ -562,7 +562,7 @@ static int uvd_v3_1_sw_init(void *handle) ring = &adev->uvd.inst->ring; sprintf(ring->name, "uvd"); r = amdgpu_ring_init(adev, ring, 512, &adev->uvd.inst->irq, 0, - AMDGPU_RING_PRIO_DEFAULT); + AMDGPU_RING_PRIO_DEFAULT, NULL); if (r) return r; diff --git a/drivers/gpu/drm/amd/amdgpu/uvd_v4_2.c b/drivers/gpu/drm/amd/amdgpu/uvd_v4_2.c index a70d2a0de316..a301518e4957 100644 --- a/drivers/gpu/drm/amd/amdgpu/uvd_v4_2.c +++ b/drivers/gpu/drm/amd/amdgpu/uvd_v4_2.c @@ -119,7 +119,7 @@ static int uvd_v4_2_sw_init(void *handle) ring = &adev->uvd.inst->ring; sprintf(ring->name, "uvd"); r = amdgpu_ring_init(adev, ring, 512, &adev->uvd.inst->irq, 0, - AMDGPU_RING_PRIO_DEFAULT); + AMDGPU_RING_PRIO_DEFAULT, NULL); if (r) return r; diff --git a/drivers/gpu/drm/amd/amdgpu/uvd_v5_0.c b/drivers/gpu/drm/amd/amdgpu/uvd_v5_0.c index f3b0a927101b..a4d5bd21c83c 100644 --- a/drivers/gpu/drm/amd/amdgpu/uvd_v5_0.c +++ b/drivers/gpu/drm/amd/amdgpu/uvd_v5_0.c @@ -117,7 +117,7 @@ static int uvd_v5_0_sw_init(void *handle) ring = &adev->uvd.inst->ring; sprintf(ring->name, "uvd"); r = amdgpu_ring_init(adev, ring, 512, &adev->uvd.inst->irq, 0, - AMDGPU_RING_PRIO_DEFAULT); + AMDGPU_RING_PRIO_DEFAULT, NULL); if (r) return r; diff --git a/drivers/gpu/drm/amd/amdgpu/uvd_v6_0.c b/drivers/gpu/drm/amd/amdgpu/uvd_v6_0.c index 760859880c1e..2bab9c77952f 100644 --- a/drivers/gpu/drm/amd/amdgpu/uvd_v6_0.c +++ b/drivers/gpu/drm/amd/amdgpu/uvd_v6_0.c @@ -420,7 +420,7 @@ static int uvd_v6_0_sw_init(void *handle) ring = &adev->uvd.inst->ring; sprintf(ring->name, "uvd"); r = amdgpu_ring_init(adev, ring, 512, &adev->uvd.inst->irq, 0, - AMDGPU_RING_PRIO_DEFAULT); + AMDGPU_RING_PRIO_DEFAULT, NULL); if (r) return r; @@ -434,7 +434,7 @@ static int uvd_v6_0_sw_init(void *handle) sprintf(ring->name, "uvd_enc%d", i); r = amdgpu_ring_init(adev, ring, 512, &adev->uvd.inst->irq, 0, - AMDGPU_RING_PRIO_DEFAULT); + AMDGPU_RING_PRIO_DEFAULT, NULL); if (r) return r; } diff --git a/drivers/gpu/drm/amd/amdgpu/uvd_v7_0.c b/drivers/gpu/drm/amd/amdgpu/uvd_v7_0.c index 7cd67cb2ac5f..0cd98fcb1f9f 100644 --- a/drivers/gpu/drm/amd/amdgpu/uvd_v7_0.c +++ b/drivers/gpu/drm/amd/amdgpu/uvd_v7_0.c @@ -454,7 +454,7 @@ static int uvd_v7_0_sw_init(void *handle) sprintf(ring->name, "uvd_%d", ring->me); r = amdgpu_ring_init(adev, ring, 512, &adev->uvd.inst[j].irq, 0, - AMDGPU_RING_PRIO_DEFAULT); + AMDGPU_RING_PRIO_DEFAULT, NULL); if (r) return r; } @@ -475,7 +475,7 @@ static int uvd_v7_0_sw_init(void *handle) } r = amdgpu_ring_init(adev, ring, 512, &adev->uvd.inst[j].irq, 0, - AMDGPU_RING_PRIO_DEFAULT); + AMDGPU_RING_PRIO_DEFAULT, NULL); if (r) return r; } diff --git a/drivers/gpu/drm/amd/amdgpu/vce_v2_0.c b/drivers/gpu/drm/amd/amdgpu/vce_v2_0.c index 0e2945baf0f1..c7d28c169be5 100644 --- a/drivers/gpu/drm/amd/amdgpu/vce_v2_0.c +++ b/drivers/gpu/drm/amd/amdgpu/vce_v2_0.c @@ -433,9 +433,8 @@ static int vce_v2_0_sw_init(void *handle) for (i = 0; i < adev->vce.num_rings; i++) { ring = &adev->vce.ring[i]; sprintf(ring->name, "vce%d", i); - r = amdgpu_ring_init(adev, ring, 512, - &adev->vce.irq, 0, - AMDGPU_RING_PRIO_DEFAULT); + r = amdgpu_ring_init(adev, ring, 512, &adev->vce.irq, 0, + AMDGPU_RING_PRIO_DEFAULT, NULL); if (r) return r; } diff --git a/drivers/gpu/drm/amd/amdgpu/vce_v3_0.c b/drivers/gpu/drm/amd/amdgpu/vce_v3_0.c index 6d9108fa22e0..3b82fb289ef6 100644 --- a/drivers/gpu/drm/amd/amdgpu/vce_v3_0.c +++ b/drivers/gpu/drm/amd/amdgpu/vce_v3_0.c @@ -443,7 +443,7 @@ static int vce_v3_0_sw_init(void *handle) ring = &adev->vce.ring[i]; sprintf(ring->name, "vce%d", i); r = amdgpu_ring_init(adev, ring, 512, &adev->vce.irq, 0, - AMDGPU_RING_PRIO_DEFAULT); + AMDGPU_RING_PRIO_DEFAULT, NULL); if (r) return r; } diff --git a/drivers/gpu/drm/amd/amdgpu/vce_v4_0.c b/drivers/gpu/drm/amd/amdgpu/vce_v4_0.c index 37fa163393fd..8e238dea7bef 100644 --- a/drivers/gpu/drm/amd/amdgpu/vce_v4_0.c +++ b/drivers/gpu/drm/amd/amdgpu/vce_v4_0.c @@ -477,7 +477,7 @@ static int vce_v4_0_sw_init(void *handle) ring->doorbell_index = adev->doorbell_index.uvd_vce.vce_ring2_3 * 2 + 1; } r = amdgpu_ring_init(adev, ring, 512, &adev->vce.irq, 0, - AMDGPU_RING_PRIO_DEFAULT); + AMDGPU_RING_PRIO_DEFAULT, NULL); if (r) return r; } diff --git a/drivers/gpu/drm/amd/amdgpu/vcn_v1_0.c b/drivers/gpu/drm/amd/amdgpu/vcn_v1_0.c index 6117931fa8d7..51a773a37a35 100644 --- a/drivers/gpu/drm/amd/amdgpu/vcn_v1_0.c +++ b/drivers/gpu/drm/amd/amdgpu/vcn_v1_0.c @@ -129,7 +129,7 @@ static int vcn_v1_0_sw_init(void *handle) ring = &adev->vcn.inst->ring_dec; sprintf(ring->name, "vcn_dec"); r = amdgpu_ring_init(adev, ring, 512, &adev->vcn.inst->irq, 0, - AMDGPU_RING_PRIO_DEFAULT); + AMDGPU_RING_PRIO_DEFAULT, NULL); if (r) return r; @@ -148,7 +148,7 @@ static int vcn_v1_0_sw_init(void *handle) ring = &adev->vcn.inst->ring_enc[i]; sprintf(ring->name, "vcn_enc%d", i); r = amdgpu_ring_init(adev, ring, 512, &adev->vcn.inst->irq, 0, - AMDGPU_RING_PRIO_DEFAULT); + AMDGPU_RING_PRIO_DEFAULT, NULL); if (r) return r; } diff --git a/drivers/gpu/drm/amd/amdgpu/vcn_v2_0.c b/drivers/gpu/drm/amd/amdgpu/vcn_v2_0.c index d63198c945bf..116b9643d5ba 100644 --- a/drivers/gpu/drm/amd/amdgpu/vcn_v2_0.c +++ b/drivers/gpu/drm/amd/amdgpu/vcn_v2_0.c @@ -136,7 +136,7 @@ static int vcn_v2_0_sw_init(void *handle) sprintf(ring->name, "vcn_dec"); r = amdgpu_ring_init(adev, ring, 512, &adev->vcn.inst->irq, 0, - AMDGPU_RING_PRIO_DEFAULT); + AMDGPU_RING_PRIO_DEFAULT, NULL); if (r) return r; @@ -167,7 +167,7 @@ static int vcn_v2_0_sw_init(void *handle) ring->doorbell_index = (adev->doorbell_index.vcn.vcn_ring0_1 << 1) + 1 + i; sprintf(ring->name, "vcn_enc%d", i); r = amdgpu_ring_init(adev, ring, 512, &adev->vcn.inst->irq, 0, - AMDGPU_RING_PRIO_DEFAULT); + AMDGPU_RING_PRIO_DEFAULT, NULL); if (r) return r; } diff --git a/drivers/gpu/drm/amd/amdgpu/vcn_v2_5.c b/drivers/gpu/drm/amd/amdgpu/vcn_v2_5.c index b6e0f4ba6272..948813d7caa0 100644 --- a/drivers/gpu/drm/amd/amdgpu/vcn_v2_5.c +++ b/drivers/gpu/drm/amd/amdgpu/vcn_v2_5.c @@ -189,7 +189,7 @@ static int vcn_v2_5_sw_init(void *handle) (amdgpu_sriov_vf(adev) ? 2*j : 8*j); sprintf(ring->name, "vcn_dec_%d", j); r = amdgpu_ring_init(adev, ring, 512, &adev->vcn.inst[j].irq, - 0, AMDGPU_RING_PRIO_DEFAULT); + 0, AMDGPU_RING_PRIO_DEFAULT, NULL); if (r) return r; @@ -203,7 +203,7 @@ static int vcn_v2_5_sw_init(void *handle) sprintf(ring->name, "vcn_enc_%d.%d", j, i); r = amdgpu_ring_init(adev, ring, 512, &adev->vcn.inst[j].irq, 0, - AMDGPU_RING_PRIO_DEFAULT); + AMDGPU_RING_PRIO_DEFAULT, NULL); if (r) return r; } @@ -1545,6 +1545,36 @@ static const struct amdgpu_ring_funcs vcn_v2_5_dec_ring_vm_funcs = { .emit_reg_write_reg_wait = amdgpu_ring_emit_reg_write_reg_wait_helper, }; +static const struct amdgpu_ring_funcs vcn_v2_6_dec_ring_vm_funcs = { + .type = AMDGPU_RING_TYPE_VCN_DEC, + .align_mask = 0xf, + .vmhub = AMDGPU_MMHUB_0, + .get_rptr = vcn_v2_5_dec_ring_get_rptr, + .get_wptr = vcn_v2_5_dec_ring_get_wptr, + .set_wptr = vcn_v2_5_dec_ring_set_wptr, + .emit_frame_size = + SOC15_FLUSH_GPU_TLB_NUM_WREG * 6 + + SOC15_FLUSH_GPU_TLB_NUM_REG_WAIT * 8 + + 8 + /* vcn_v2_0_dec_ring_emit_vm_flush */ + 14 + 14 + /* vcn_v2_0_dec_ring_emit_fence x2 vm fence */ + 6, + .emit_ib_size = 8, /* vcn_v2_0_dec_ring_emit_ib */ + .emit_ib = vcn_v2_0_dec_ring_emit_ib, + .emit_fence = vcn_v2_0_dec_ring_emit_fence, + .emit_vm_flush = vcn_v2_0_dec_ring_emit_vm_flush, + .test_ring = vcn_v2_0_dec_ring_test_ring, + .test_ib = amdgpu_vcn_dec_ring_test_ib, + .insert_nop = vcn_v2_0_dec_ring_insert_nop, + .insert_start = vcn_v2_0_dec_ring_insert_start, + .insert_end = vcn_v2_0_dec_ring_insert_end, + .pad_ib = amdgpu_ring_generic_pad_ib, + .begin_use = amdgpu_vcn_ring_begin_use, + .end_use = amdgpu_vcn_ring_end_use, + .emit_wreg = vcn_v2_0_dec_ring_emit_wreg, + .emit_reg_wait = vcn_v2_0_dec_ring_emit_reg_wait, + .emit_reg_write_reg_wait = amdgpu_ring_emit_reg_write_reg_wait_helper, +}; + /** * vcn_v2_5_enc_ring_get_rptr - get enc read pointer * @@ -1644,6 +1674,36 @@ static const struct amdgpu_ring_funcs vcn_v2_5_enc_ring_vm_funcs = { .emit_reg_write_reg_wait = amdgpu_ring_emit_reg_write_reg_wait_helper, }; +static const struct amdgpu_ring_funcs vcn_v2_6_enc_ring_vm_funcs = { + .type = AMDGPU_RING_TYPE_VCN_ENC, + .align_mask = 0x3f, + .nop = VCN_ENC_CMD_NO_OP, + .vmhub = AMDGPU_MMHUB_0, + .get_rptr = vcn_v2_5_enc_ring_get_rptr, + .get_wptr = vcn_v2_5_enc_ring_get_wptr, + .set_wptr = vcn_v2_5_enc_ring_set_wptr, + .emit_frame_size = + SOC15_FLUSH_GPU_TLB_NUM_WREG * 3 + + SOC15_FLUSH_GPU_TLB_NUM_REG_WAIT * 4 + + 4 + /* vcn_v2_0_enc_ring_emit_vm_flush */ + 5 + 5 + /* vcn_v2_0_enc_ring_emit_fence x2 vm fence */ + 1, /* vcn_v2_0_enc_ring_insert_end */ + .emit_ib_size = 5, /* vcn_v2_0_enc_ring_emit_ib */ + .emit_ib = vcn_v2_0_enc_ring_emit_ib, + .emit_fence = vcn_v2_0_enc_ring_emit_fence, + .emit_vm_flush = vcn_v2_0_enc_ring_emit_vm_flush, + .test_ring = amdgpu_vcn_enc_ring_test_ring, + .test_ib = amdgpu_vcn_enc_ring_test_ib, + .insert_nop = amdgpu_ring_insert_nop, + .insert_end = vcn_v2_0_enc_ring_insert_end, + .pad_ib = amdgpu_ring_generic_pad_ib, + .begin_use = amdgpu_vcn_ring_begin_use, + .end_use = amdgpu_vcn_ring_end_use, + .emit_wreg = vcn_v2_0_enc_ring_emit_wreg, + .emit_reg_wait = vcn_v2_0_enc_ring_emit_reg_wait, + .emit_reg_write_reg_wait = amdgpu_ring_emit_reg_write_reg_wait_helper, +}; + static void vcn_v2_5_set_dec_ring_funcs(struct amdgpu_device *adev) { int i; @@ -1651,7 +1711,10 @@ static void vcn_v2_5_set_dec_ring_funcs(struct amdgpu_device *adev) for (i = 0; i < adev->vcn.num_vcn_inst; ++i) { if (adev->vcn.harvest_config & (1 << i)) continue; - adev->vcn.inst[i].ring_dec.funcs = &vcn_v2_5_dec_ring_vm_funcs; + if (adev->asic_type == CHIP_ARCTURUS) + adev->vcn.inst[i].ring_dec.funcs = &vcn_v2_5_dec_ring_vm_funcs; + else /* CHIP_ALDEBARAN */ + adev->vcn.inst[i].ring_dec.funcs = &vcn_v2_6_dec_ring_vm_funcs; adev->vcn.inst[i].ring_dec.me = i; DRM_INFO("VCN(%d) decode is enabled in VM mode\n", i); } @@ -1665,7 +1728,10 @@ static void vcn_v2_5_set_enc_ring_funcs(struct amdgpu_device *adev) if (adev->vcn.harvest_config & (1 << j)) continue; for (i = 0; i < adev->vcn.num_enc_rings; ++i) { - adev->vcn.inst[j].ring_enc[i].funcs = &vcn_v2_5_enc_ring_vm_funcs; + if (adev->asic_type == CHIP_ARCTURUS) + adev->vcn.inst[j].ring_enc[i].funcs = &vcn_v2_5_enc_ring_vm_funcs; + else /* CHIP_ALDEBARAN */ + adev->vcn.inst[j].ring_enc[i].funcs = &vcn_v2_6_enc_ring_vm_funcs; adev->vcn.inst[j].ring_enc[i].me = j; } DRM_INFO("VCN(%d) encode is enabled in VM mode\n", j); @@ -1830,6 +1896,26 @@ static const struct amd_ip_funcs vcn_v2_5_ip_funcs = { .set_powergating_state = vcn_v2_5_set_powergating_state, }; +static const struct amd_ip_funcs vcn_v2_6_ip_funcs = { + .name = "vcn_v2_6", + .early_init = vcn_v2_5_early_init, + .late_init = NULL, + .sw_init = vcn_v2_5_sw_init, + .sw_fini = vcn_v2_5_sw_fini, + .hw_init = vcn_v2_5_hw_init, + .hw_fini = vcn_v2_5_hw_fini, + .suspend = vcn_v2_5_suspend, + .resume = vcn_v2_5_resume, + .is_idle = vcn_v2_5_is_idle, + .wait_for_idle = vcn_v2_5_wait_for_idle, + .check_soft_reset = NULL, + .pre_soft_reset = NULL, + .soft_reset = NULL, + .post_soft_reset = NULL, + .set_clockgating_state = vcn_v2_5_set_clockgating_state, + .set_powergating_state = vcn_v2_5_set_powergating_state, +}; + const struct amdgpu_ip_block_version vcn_v2_5_ip_block = { .type = AMD_IP_BLOCK_TYPE_VCN, @@ -1838,3 +1924,12 @@ const struct amdgpu_ip_block_version vcn_v2_5_ip_block = .rev = 0, .funcs = &vcn_v2_5_ip_funcs, }; + +const struct amdgpu_ip_block_version vcn_v2_6_ip_block = +{ + .type = AMD_IP_BLOCK_TYPE_VCN, + .major = 2, + .minor = 6, + .rev = 0, + .funcs = &vcn_v2_6_ip_funcs, +}; diff --git a/drivers/gpu/drm/amd/amdgpu/vcn_v2_5.h b/drivers/gpu/drm/amd/amdgpu/vcn_v2_5.h index 8d9c0800b8e0..e72f799ed0fd 100644 --- a/drivers/gpu/drm/amd/amdgpu/vcn_v2_5.h +++ b/drivers/gpu/drm/amd/amdgpu/vcn_v2_5.h @@ -25,5 +25,6 @@ #define __VCN_V2_5_H__ extern const struct amdgpu_ip_block_version vcn_v2_5_ip_block; +extern const struct amdgpu_ip_block_version vcn_v2_6_ip_block; #endif /* __VCN_V2_5_H__ */ diff --git a/drivers/gpu/drm/amd/amdgpu/vcn_v3_0.c b/drivers/gpu/drm/amd/amdgpu/vcn_v3_0.c index def583916294..cf165ab5dd26 100644 --- a/drivers/gpu/drm/amd/amdgpu/vcn_v3_0.c +++ b/drivers/gpu/drm/amd/amdgpu/vcn_v3_0.c @@ -50,6 +50,9 @@ #define VCN_INSTANCES_SIENNA_CICHLID 2 #define DEC_SW_RING_ENABLED FALSE +#define RDECODE_MSG_CREATE 0x00000000 +#define RDECODE_MESSAGE_CREATE 0x00000001 + static int amdgpu_ih_clientid_vcns[] = { SOC15_IH_CLIENTID_VCN, SOC15_IH_CLIENTID_VCN1 @@ -171,6 +174,7 @@ static int vcn_v3_0_sw_init(void *handle) for (i = 0; i < adev->vcn.num_vcn_inst; i++) { volatile struct amdgpu_fw_shared *fw_shared; + if (adev->vcn.harvest_config & (1 << i)) continue; @@ -198,6 +202,8 @@ static int vcn_v3_0_sw_init(void *handle) if (r) return r; + atomic_set(&adev->vcn.inst[i].sched_score, 0); + ring = &adev->vcn.inst[i].ring_dec; ring->use_doorbell = true; if (amdgpu_sriov_vf(adev)) { @@ -205,11 +211,10 @@ static int vcn_v3_0_sw_init(void *handle) } else { ring->doorbell_index = (adev->doorbell_index.vcn.vcn_ring0_1 << 1) + 8 * i; } - if (adev->asic_type == CHIP_SIENNA_CICHLID && i != 0) - ring->no_scheduler = true; sprintf(ring->name, "vcn_dec_%d", i); r = amdgpu_ring_init(adev, ring, 512, &adev->vcn.inst[i].irq, 0, - AMDGPU_RING_PRIO_DEFAULT); + AMDGPU_RING_PRIO_DEFAULT, + &adev->vcn.inst[i].sched_score); if (r) return r; @@ -227,18 +232,18 @@ static int vcn_v3_0_sw_init(void *handle) } else { ring->doorbell_index = (adev->doorbell_index.vcn.vcn_ring0_1 << 1) + 2 + j + 8 * i; } - if (adev->asic_type == CHIP_SIENNA_CICHLID && i != 1) - ring->no_scheduler = true; sprintf(ring->name, "vcn_enc_%d.%d", i, j); r = amdgpu_ring_init(adev, ring, 512, &adev->vcn.inst[i].irq, 0, - AMDGPU_RING_PRIO_DEFAULT); + AMDGPU_RING_PRIO_DEFAULT, + &adev->vcn.inst[i].sched_score); if (r) return r; } fw_shared = adev->vcn.inst[i].fw_shared_cpu_addr; fw_shared->present_flag_0 |= cpu_to_le32(AMDGPU_VCN_SW_RING_FLAG) | - cpu_to_le32(AMDGPU_VCN_MULTI_QUEUE_FLAG); + cpu_to_le32(AMDGPU_VCN_MULTI_QUEUE_FLAG) | + cpu_to_le32(AMDGPU_VCN_FW_SHARED_FLAG_0_RB); fw_shared->sw_ring.is_enabled = cpu_to_le32(DEC_SW_RING_ENABLED); } @@ -584,6 +589,10 @@ static void vcn_v3_0_mc_resume_dpg_mode(struct amdgpu_device *adev, int inst_idx WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET( VCN, inst_idx, mmUVD_VCPU_NONCACHE_SIZE0), AMDGPU_GPU_PAGE_ALIGN(sizeof(struct amdgpu_fw_shared)), 0, indirect); + + /* VCN global tiling registers */ + WREG32_SOC15_DPG_MODE(0, SOC15_DPG_MODE_OFFSET( + UVD, 0, mmUVD_GFX10_ADDR_CONFIG), adev->gfx.config.gb_addr_config, 0, indirect); } static void vcn_v3_0_disable_static_power_gating(struct amdgpu_device *adev, int inst) @@ -1074,7 +1083,13 @@ static int vcn_v3_0_start_dpg_mode(struct amdgpu_device *adev, int inst_idx, boo WREG32_SOC15(VCN, inst_idx, mmUVD_RBC_RB_WPTR, lower_32_bits(ring->wptr)); + /* Reset FW shared memory RBC WPTR/RPTR */ + fw_shared->rb.rptr = 0; + fw_shared->rb.wptr = lower_32_bits(ring->wptr); + + /*resetting done, fw can check RB ring */ fw_shared->multi_queue.decode_queue_mode &= cpu_to_le32(~FW_QUEUE_RING_RESET); + /* Unstall DPG */ WREG32_P(SOC15_REG_OFFSET(VCN, inst_idx, mmUVD_POWER_STATUS), 0, ~UVD_POWER_STATUS__STALL_DPG_POWER_UP_MASK); @@ -1239,9 +1254,11 @@ static int vcn_v3_0_start(struct amdgpu_device *adev) /* Initialize the ring buffer's read and write pointers */ WREG32_SOC15(VCN, i, mmUVD_RBC_RB_RPTR, 0); + WREG32_SOC15(VCN, i, mmUVD_SCRATCH2, 0); ring->wptr = RREG32_SOC15(VCN, i, mmUVD_RBC_RB_RPTR); WREG32_SOC15(VCN, i, mmUVD_RBC_RB_WPTR, lower_32_bits(ring->wptr)); + fw_shared->rb.wptr = lower_32_bits(ring->wptr); fw_shared->multi_queue.decode_queue_mode &= cpu_to_le32(~FW_QUEUE_RING_RESET); fw_shared->multi_queue.encode_generalpurpose_queue_mode |= cpu_to_le32(FW_QUEUE_RING_RESET); @@ -1662,6 +1679,10 @@ static int vcn_v3_0_pause_dpg_mode(struct amdgpu_device *adev, WREG32_SOC15(VCN, inst_idx, mmUVD_RB_WPTR2, lower_32_bits(ring->wptr)); fw_shared->multi_queue.encode_lowlatency_queue_mode &= cpu_to_le32(~FW_QUEUE_RING_RESET); + /* restore wptr/rptr with pointers saved in FW shared memory*/ + WREG32_SOC15(VCN, inst_idx, mmUVD_RBC_RB_RPTR, fw_shared->rb.rptr); + WREG32_SOC15(VCN, inst_idx, mmUVD_RBC_RB_WPTR, fw_shared->rb.wptr); + /* Unstall DPG */ WREG32_P(SOC15_REG_OFFSET(VCN, inst_idx, mmUVD_POWER_STATUS), 0, ~UVD_POWER_STATUS__STALL_DPG_POWER_UP_MASK); @@ -1721,6 +1742,15 @@ static uint64_t vcn_v3_0_dec_ring_get_wptr(struct amdgpu_ring *ring) static void vcn_v3_0_dec_ring_set_wptr(struct amdgpu_ring *ring) { struct amdgpu_device *adev = ring->adev; + volatile struct amdgpu_fw_shared *fw_shared; + + if (adev->pg_flags & AMD_PG_SUPPORT_VCN_DPG) { + /*whenever update RBC_RB_WPTR, we save the wptr in shared rb.wptr and scratch2 */ + fw_shared = adev->vcn.inst[ring->me].fw_shared_cpu_addr; + fw_shared->rb.wptr = lower_32_bits(ring->wptr); + WREG32_SOC15(VCN, ring->me, mmUVD_SCRATCH2, + lower_32_bits(ring->wptr)); + } if (ring->use_doorbell) { adev->wb.wb[ring->wptr_offs] = lower_32_bits(ring->wptr); @@ -1822,6 +1852,132 @@ static const struct amdgpu_ring_funcs vcn_v3_0_dec_sw_ring_vm_funcs = { .emit_reg_write_reg_wait = amdgpu_ring_emit_reg_write_reg_wait_helper, }; +static int vcn_v3_0_limit_sched(struct amdgpu_cs_parser *p) +{ + struct drm_gpu_scheduler **scheds; + + /* The create msg must be in the first IB submitted */ + if (atomic_read(&p->entity->fence_seq)) + return -EINVAL; + + scheds = p->adev->gpu_sched[AMDGPU_HW_IP_VCN_DEC] + [AMDGPU_RING_PRIO_DEFAULT].sched; + drm_sched_entity_modify_sched(p->entity, scheds, 1); + return 0; +} + +static int vcn_v3_0_dec_msg(struct amdgpu_cs_parser *p, uint64_t addr) +{ + struct ttm_operation_ctx ctx = { false, false }; + struct amdgpu_bo_va_mapping *map; + uint32_t *msg, num_buffers; + struct amdgpu_bo *bo; + uint64_t start, end; + unsigned int i; + void * ptr; + int r; + + addr &= AMDGPU_GMC_HOLE_MASK; + r = amdgpu_cs_find_mapping(p, addr, &bo, &map); + if (r) { + DRM_ERROR("Can't find BO for addr 0x%08Lx\n", addr); + return r; + } + + start = map->start * AMDGPU_GPU_PAGE_SIZE; + end = (map->last + 1) * AMDGPU_GPU_PAGE_SIZE; + if (addr & 0x7) { + DRM_ERROR("VCN messages must be 8 byte aligned!\n"); + return -EINVAL; + } + + bo->flags |= AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED; + amdgpu_bo_placement_from_domain(bo, bo->allowed_domains); + r = ttm_bo_validate(&bo->tbo, &bo->placement, &ctx); + if (r) { + DRM_ERROR("Failed validating the VCN message BO (%d)!\n", r); + return r; + } + + r = amdgpu_bo_kmap(bo, &ptr); + if (r) { + DRM_ERROR("Failed mapping the VCN message (%d)!\n", r); + return r; + } + + msg = ptr + addr - start; + + /* Check length */ + if (msg[1] > end - addr) { + r = -EINVAL; + goto out; + } + + if (msg[3] != RDECODE_MSG_CREATE) + goto out; + + num_buffers = msg[2]; + for (i = 0, msg = &msg[6]; i < num_buffers; ++i, msg += 4) { + uint32_t offset, size, *create; + + if (msg[0] != RDECODE_MESSAGE_CREATE) + continue; + + offset = msg[1]; + size = msg[2]; + + if (offset + size > end) { + r = -EINVAL; + goto out; + } + + create = ptr + addr + offset - start; + + /* H246, HEVC and VP9 can run on any instance */ + if (create[0] == 0x7 || create[0] == 0x10 || create[0] == 0x11) + continue; + + r = vcn_v3_0_limit_sched(p); + if (r) + goto out; + } + +out: + amdgpu_bo_kunmap(bo); + return r; +} + +static int vcn_v3_0_ring_patch_cs_in_place(struct amdgpu_cs_parser *p, + uint32_t ib_idx) +{ + struct amdgpu_ring *ring = to_amdgpu_ring(p->entity->rq->sched); + struct amdgpu_ib *ib = &p->job->ibs[ib_idx]; + uint32_t msg_lo = 0, msg_hi = 0; + unsigned i; + int r; + + /* The first instance can decode anything */ + if (!ring->me) + return 0; + + for (i = 0; i < ib->length_dw; i += 2) { + uint32_t reg = amdgpu_get_ib_value(p, ib_idx, i); + uint32_t val = amdgpu_get_ib_value(p, ib_idx, i + 1); + + if (reg == PACKET0(p->adev->vcn.internal.data0, 0)) { + msg_lo = val; + } else if (reg == PACKET0(p->adev->vcn.internal.data1, 0)) { + msg_hi = val; + } else if (reg == PACKET0(p->adev->vcn.internal.cmd, 0) && + val == 0) { + r = vcn_v3_0_dec_msg(p, ((u64)msg_hi) << 32 | msg_lo); + if (r) + return r; + } + } + return 0; +} + static const struct amdgpu_ring_funcs vcn_v3_0_dec_ring_vm_funcs = { .type = AMDGPU_RING_TYPE_VCN_DEC, .align_mask = 0xf, @@ -1829,6 +1985,7 @@ static const struct amdgpu_ring_funcs vcn_v3_0_dec_ring_vm_funcs = { .get_rptr = vcn_v3_0_dec_ring_get_rptr, .get_wptr = vcn_v3_0_dec_ring_get_wptr, .set_wptr = vcn_v3_0_dec_ring_set_wptr, + .patch_cs_in_place = vcn_v3_0_ring_patch_cs_in_place, .emit_frame_size = SOC15_FLUSH_GPU_TLB_NUM_WREG * 6 + SOC15_FLUSH_GPU_TLB_NUM_REG_WAIT * 8 + diff --git a/drivers/gpu/drm/amd/amdgpu/vega10_ih.c b/drivers/gpu/drm/amd/amdgpu/vega10_ih.c index 88626d83e07b..ca8efa5c6978 100644 --- a/drivers/gpu/drm/amd/amdgpu/vega10_ih.c +++ b/drivers/gpu/drm/amd/amdgpu/vega10_ih.c @@ -220,10 +220,8 @@ static int vega10_ih_enable_ring(struct amdgpu_device *adev, tmp = vega10_ih_rb_cntl(ih, tmp); if (ih == &adev->irq.ih) tmp = REG_SET_FIELD(tmp, IH_RB_CNTL, RPTR_REARM, !!adev->irq.msi_enabled); - if (ih == &adev->irq.ih1) { - tmp = REG_SET_FIELD(tmp, IH_RB_CNTL, WPTR_OVERFLOW_ENABLE, 0); + if (ih == &adev->irq.ih1) tmp = REG_SET_FIELD(tmp, IH_RB_CNTL, RB_FULL_DRAIN_ENABLE, 1); - } if (amdgpu_sriov_vf(adev)) { if (psp_reg_program(&adev->psp, ih_regs->psp_reg_id, tmp)) { dev_err(adev->dev, "PSP program IH_RB_CNTL failed!\n"); @@ -265,7 +263,6 @@ static int vega10_ih_irq_init(struct amdgpu_device *adev) u32 ih_chicken; int ret; int i; - u32 tmp; /* disable irqs */ ret = vega10_ih_toggle_interrupts(adev, false); @@ -291,15 +288,6 @@ static int vega10_ih_irq_init(struct amdgpu_device *adev) } } - tmp = RREG32_SOC15(OSSSYS, 0, mmIH_STORM_CLIENT_LIST_CNTL); - tmp = REG_SET_FIELD(tmp, IH_STORM_CLIENT_LIST_CNTL, - CLIENT18_IS_STORM_CLIENT, 1); - WREG32_SOC15(OSSSYS, 0, mmIH_STORM_CLIENT_LIST_CNTL, tmp); - - tmp = RREG32_SOC15(OSSSYS, 0, mmIH_INT_FLOOD_CNTL); - tmp = REG_SET_FIELD(tmp, IH_INT_FLOOD_CNTL, FLOOD_CNTL_ENABLE, 1); - WREG32_SOC15(OSSSYS, 0, mmIH_INT_FLOOD_CNTL, tmp); - pci_set_master(adev->pdev); /* enable interrupts */ @@ -345,11 +333,17 @@ static u32 vega10_ih_get_wptr(struct amdgpu_device *adev, u32 wptr, tmp; struct amdgpu_ih_regs *ih_regs; - wptr = le32_to_cpu(*ih->wptr_cpu); - ih_regs = &ih->ih_regs; + if (ih == &adev->irq.ih) { + /* Only ring0 supports writeback. On other rings fall back + * to register-based code with overflow checking below. + */ + wptr = le32_to_cpu(*ih->wptr_cpu); - if (!REG_GET_FIELD(wptr, IH_RB_WPTR, RB_OVERFLOW)) - goto out; + if (!REG_GET_FIELD(wptr, IH_RB_WPTR, RB_OVERFLOW)) + goto out; + } + + ih_regs = &ih->ih_regs; /* Double check that the overflow wasn't already cleared. */ wptr = RREG32_NO_KIQ(ih_regs->ih_rb_wptr); @@ -440,15 +434,11 @@ static int vega10_ih_self_irq(struct amdgpu_device *adev, struct amdgpu_irq_src *source, struct amdgpu_iv_entry *entry) { - uint32_t wptr = cpu_to_le32(entry->src_data[0]); - switch (entry->ring_id) { case 1: - *adev->irq.ih1.wptr_cpu = wptr; schedule_work(&adev->irq.ih1_work); break; case 2: - *adev->irq.ih2.wptr_cpu = wptr; schedule_work(&adev->irq.ih2_work); break; default: break; diff --git a/drivers/gpu/drm/amd/amdgpu/vega20_ih.c b/drivers/gpu/drm/amd/amdgpu/vega20_ih.c index 5a3c867d5881..8a122b413bf5 100644 --- a/drivers/gpu/drm/amd/amdgpu/vega20_ih.c +++ b/drivers/gpu/drm/amd/amdgpu/vega20_ih.c @@ -35,6 +35,9 @@ #define MAX_REARM_RETRY 10 +#define mmIH_CHICKEN_ALDEBARAN 0x18d +#define mmIH_CHICKEN_ALDEBARAN_BASE_IDX 0 + static void vega20_ih_set_interrupt_funcs(struct amdgpu_device *adev); /** @@ -104,6 +107,8 @@ static int vega20_ih_toggle_ring_interrupts(struct amdgpu_device *adev, tmp = RREG32(ih_regs->ih_rb_cntl); tmp = REG_SET_FIELD(tmp, IH_RB_CNTL, RB_ENABLE, (enable ? 1 : 0)); + tmp = REG_SET_FIELD(tmp, IH_RB_CNTL, RB_GPU_TS_ENABLE, 1); + /* enable_intr field is only valid in ring0 */ if (ih == &adev->irq.ih) tmp = REG_SET_FIELD(tmp, IH_RB_CNTL, ENABLE_INTR, (enable ? 1 : 0)); @@ -220,10 +225,8 @@ static int vega20_ih_enable_ring(struct amdgpu_device *adev, tmp = vega20_ih_rb_cntl(ih, tmp); if (ih == &adev->irq.ih) tmp = REG_SET_FIELD(tmp, IH_RB_CNTL, RPTR_REARM, !!adev->irq.msi_enabled); - if (ih == &adev->irq.ih1) { - tmp = REG_SET_FIELD(tmp, IH_RB_CNTL, WPTR_OVERFLOW_ENABLE, 0); + if (ih == &adev->irq.ih1) tmp = REG_SET_FIELD(tmp, IH_RB_CNTL, RB_FULL_DRAIN_ENABLE, 1); - } if (amdgpu_sriov_vf(adev)) { if (psp_reg_program(&adev->psp, ih_regs->psp_reg_id, tmp)) { dev_err(adev->dev, "PSP program IH_RB_CNTL failed!\n"); @@ -261,10 +264,10 @@ static void vega20_ih_reroute_ih(struct amdgpu_device *adev) { uint32_t tmp; - /* vega20 ih reroute will go through psp - * this function is only used for arcturus + /* vega20 ih reroute will go through psp this + * function is used for newer asics starting arcturus */ - if (adev->asic_type == CHIP_ARCTURUS) { + if (adev->asic_type >= CHIP_ARCTURUS) { /* Reroute to IH ring 1 for VMC */ WREG32_SOC15(OSSSYS, 0, mmIH_CLIENT_CFG_INDEX, 0x12); tmp = RREG32_SOC15(OSSSYS, 0, mmIH_CLIENT_CFG_DATA); @@ -297,7 +300,6 @@ static int vega20_ih_irq_init(struct amdgpu_device *adev) u32 ih_chicken; int ret; int i; - u32 tmp; /* disable irqs */ ret = vega20_ih_toggle_interrupts(adev, false); @@ -316,6 +318,18 @@ static int vega20_ih_irq_init(struct amdgpu_device *adev) WREG32_SOC15(OSSSYS, 0, mmIH_CHICKEN, ih_chicken); } + /* psp firmware won't program IH_CHICKEN for aldebaran + * driver needs to program it properly according to + * MC_SPACE type in IH_RB_CNTL */ + if (adev->asic_type == CHIP_ALDEBARAN) { + ih_chicken = RREG32_SOC15(OSSSYS, 0, mmIH_CHICKEN_ALDEBARAN); + if (adev->irq.ih.use_bus_addr) { + ih_chicken = REG_SET_FIELD(ih_chicken, IH_CHICKEN, + MC_SPACE_GPA_ENABLE, 1); + } + WREG32_SOC15(OSSSYS, 0, mmIH_CHICKEN_ALDEBARAN, ih_chicken); + } + for (i = 0; i < ARRAY_SIZE(ih); i++) { if (ih[i]->ring_size) { if (i == 1) @@ -326,15 +340,6 @@ static int vega20_ih_irq_init(struct amdgpu_device *adev) } } - tmp = RREG32_SOC15(OSSSYS, 0, mmIH_STORM_CLIENT_LIST_CNTL); - tmp = REG_SET_FIELD(tmp, IH_STORM_CLIENT_LIST_CNTL, - CLIENT18_IS_STORM_CLIENT, 1); - WREG32_SOC15(OSSSYS, 0, mmIH_STORM_CLIENT_LIST_CNTL, tmp); - - tmp = RREG32_SOC15(OSSSYS, 0, mmIH_INT_FLOOD_CNTL); - tmp = REG_SET_FIELD(tmp, IH_INT_FLOOD_CNTL, FLOOD_CNTL_ENABLE, 1); - WREG32_SOC15(OSSSYS, 0, mmIH_INT_FLOOD_CNTL, tmp); - pci_set_master(adev->pdev); /* enable interrupts */ @@ -380,11 +385,17 @@ static u32 vega20_ih_get_wptr(struct amdgpu_device *adev, u32 wptr, tmp; struct amdgpu_ih_regs *ih_regs; - wptr = le32_to_cpu(*ih->wptr_cpu); - ih_regs = &ih->ih_regs; + if (ih == &adev->irq.ih) { + /* Only ring0 supports writeback. On other rings fall back + * to register-based code with overflow checking below. + */ + wptr = le32_to_cpu(*ih->wptr_cpu); - if (!REG_GET_FIELD(wptr, IH_RB_WPTR, RB_OVERFLOW)) - goto out; + if (!REG_GET_FIELD(wptr, IH_RB_WPTR, RB_OVERFLOW)) + goto out; + } + + ih_regs = &ih->ih_regs; /* Double check that the overflow wasn't already cleared. */ wptr = RREG32_NO_KIQ(ih_regs->ih_rb_wptr); @@ -476,15 +487,11 @@ static int vega20_ih_self_irq(struct amdgpu_device *adev, struct amdgpu_irq_src *source, struct amdgpu_iv_entry *entry) { - uint32_t wptr = cpu_to_le32(entry->src_data[0]); - switch (entry->ring_id) { case 1: - *adev->irq.ih1.wptr_cpu = wptr; schedule_work(&adev->irq.ih1_work); break; case 2: - *adev->irq.ih2.wptr_cpu = wptr; schedule_work(&adev->irq.ih2_work); break; default: break; diff --git a/drivers/gpu/drm/amd/amdgpu/vi.c b/drivers/gpu/drm/amd/amdgpu/vi.c index eafb76aebd00..735ebbd1148f 100644 --- a/drivers/gpu/drm/amd/amdgpu/vi.c +++ b/drivers/gpu/drm/amd/amdgpu/vi.c @@ -24,6 +24,8 @@ #include <linux/pci.h> #include <linux/slab.h> +#include <drm/amdgpu_drm.h> + #include "amdgpu.h" #include "amdgpu_atombios.h" #include "amdgpu_ih.h" @@ -79,6 +81,217 @@ #include "mxgpu_vi.h" #include "amdgpu_dm.h" +#define ixPCIE_LC_L1_PM_SUBSTATE 0x100100C6 +#define PCIE_LC_L1_PM_SUBSTATE__LC_L1_SUBSTATES_OVERRIDE_EN_MASK 0x00000001L +#define PCIE_LC_L1_PM_SUBSTATE__LC_PCI_PM_L1_2_OVERRIDE_MASK 0x00000002L +#define PCIE_LC_L1_PM_SUBSTATE__LC_PCI_PM_L1_1_OVERRIDE_MASK 0x00000004L +#define PCIE_LC_L1_PM_SUBSTATE__LC_ASPM_L1_2_OVERRIDE_MASK 0x00000008L +#define PCIE_LC_L1_PM_SUBSTATE__LC_ASPM_L1_1_OVERRIDE_MASK 0x00000010L +#define ixPCIE_L1_PM_SUB_CNTL 0x378 +#define PCIE_L1_PM_SUB_CNTL__ASPM_L1_2_EN_MASK 0x00000004L +#define PCIE_L1_PM_SUB_CNTL__ASPM_L1_1_EN_MASK 0x00000008L +#define PCIE_L1_PM_SUB_CNTL__PCI_PM_L1_2_EN_MASK 0x00000001L +#define PCIE_L1_PM_SUB_CNTL__PCI_PM_L1_1_EN_MASK 0x00000002L +#define PCIE_LC_CNTL6__LC_L1_POWERDOWN_MASK 0x00200000L +#define LINK_CAP 0x64 +#define PCIE_LINK_CAP__CLOCK_POWER_MANAGEMENT_MASK 0x00040000L +#define ixCPM_CONTROL 0x1400118 +#define ixPCIE_LC_CNTL7 0x100100BC +#define PCIE_LC_CNTL7__LC_L1_SIDEBAND_CLKREQ_PDWN_EN_MASK 0x00000400L +#define PCIE_LC_CNTL__LC_L0S_INACTIVITY_DEFAULT 0x00000007 +#define PCIE_LC_CNTL__LC_L1_INACTIVITY_DEFAULT 0x00000009 +#define CPM_CONTROL__CLKREQb_UNGATE_TXCLK_ENABLE_MASK 0x01000000L +#define PCIE_L1_PM_SUB_CNTL 0x378 +#define ASIC_IS_P22(asic_type, rid) ((asic_type >= CHIP_POLARIS10) && \ + (asic_type <= CHIP_POLARIS12) && \ + (rid >= 0x6E)) +/* Topaz */ +static const struct amdgpu_video_codecs topaz_video_codecs_encode = +{ + .codec_count = 0, + .codec_array = NULL, +}; + +/* Tonga, CZ, ST, Fiji */ +static const struct amdgpu_video_codec_info tonga_video_codecs_encode_array[] = +{ + { + .codec_type = AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_MPEG4_AVC, + .max_width = 4096, + .max_height = 2304, + .max_pixels_per_frame = 4096 * 2304, + .max_level = 0, + }, +}; + +static const struct amdgpu_video_codecs tonga_video_codecs_encode = +{ + .codec_count = ARRAY_SIZE(tonga_video_codecs_encode_array), + .codec_array = tonga_video_codecs_encode_array, +}; + +/* Polaris */ +static const struct amdgpu_video_codec_info polaris_video_codecs_encode_array[] = +{ + { + .codec_type = AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_MPEG4_AVC, + .max_width = 4096, + .max_height = 2304, + .max_pixels_per_frame = 4096 * 2304, + .max_level = 0, + }, + { + .codec_type = AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_HEVC, + .max_width = 4096, + .max_height = 2304, + .max_pixels_per_frame = 4096 * 2304, + .max_level = 0, + }, +}; + +static const struct amdgpu_video_codecs polaris_video_codecs_encode = +{ + .codec_count = ARRAY_SIZE(polaris_video_codecs_encode_array), + .codec_array = polaris_video_codecs_encode_array, +}; + +/* Topaz */ +static const struct amdgpu_video_codecs topaz_video_codecs_decode = +{ + .codec_count = 0, + .codec_array = NULL, +}; + +/* Tonga */ +static const struct amdgpu_video_codec_info tonga_video_codecs_decode_array[] = +{ + { + .codec_type = AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_MPEG2, + .max_width = 4096, + .max_height = 4096, + .max_pixels_per_frame = 4096 * 4096, + .max_level = 3, + }, + { + .codec_type = AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_MPEG4, + .max_width = 4096, + .max_height = 4096, + .max_pixels_per_frame = 4096 * 4096, + .max_level = 5, + }, + { + .codec_type = AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_MPEG4_AVC, + .max_width = 4096, + .max_height = 4096, + .max_pixels_per_frame = 4096 * 4096, + .max_level = 52, + }, + { + .codec_type = AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_VC1, + .max_width = 4096, + .max_height = 4096, + .max_pixels_per_frame = 4096 * 4096, + .max_level = 4, + }, +}; + +static const struct amdgpu_video_codecs tonga_video_codecs_decode = +{ + .codec_count = ARRAY_SIZE(tonga_video_codecs_decode_array), + .codec_array = tonga_video_codecs_decode_array, +}; + +/* CZ, ST, Fiji, Polaris */ +static const struct amdgpu_video_codec_info cz_video_codecs_decode_array[] = +{ + { + .codec_type = AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_MPEG2, + .max_width = 4096, + .max_height = 4096, + .max_pixels_per_frame = 4096 * 4096, + .max_level = 3, + }, + { + .codec_type = AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_MPEG4, + .max_width = 4096, + .max_height = 4096, + .max_pixels_per_frame = 4096 * 4096, + .max_level = 5, + }, + { + .codec_type = AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_MPEG4_AVC, + .max_width = 4096, + .max_height = 4096, + .max_pixels_per_frame = 4096 * 4096, + .max_level = 52, + }, + { + .codec_type = AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_VC1, + .max_width = 4096, + .max_height = 4096, + .max_pixels_per_frame = 4096 * 4096, + .max_level = 4, + }, + { + .codec_type = AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_HEVC, + .max_width = 4096, + .max_height = 4096, + .max_pixels_per_frame = 4096 * 4096, + .max_level = 186, + }, + { + .codec_type = AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_JPEG, + .max_width = 4096, + .max_height = 4096, + .max_pixels_per_frame = 4096 * 4096, + .max_level = 0, + }, +}; + +static const struct amdgpu_video_codecs cz_video_codecs_decode = +{ + .codec_count = ARRAY_SIZE(cz_video_codecs_decode_array), + .codec_array = cz_video_codecs_decode_array, +}; + +static int vi_query_video_codecs(struct amdgpu_device *adev, bool encode, + const struct amdgpu_video_codecs **codecs) +{ + switch (adev->asic_type) { + case CHIP_TOPAZ: + if (encode) + *codecs = &topaz_video_codecs_encode; + else + *codecs = &topaz_video_codecs_decode; + return 0; + case CHIP_TONGA: + if (encode) + *codecs = &tonga_video_codecs_encode; + else + *codecs = &tonga_video_codecs_decode; + return 0; + case CHIP_POLARIS10: + case CHIP_POLARIS11: + case CHIP_POLARIS12: + case CHIP_VEGAM: + if (encode) + *codecs = &polaris_video_codecs_encode; + else + *codecs = &cz_video_codecs_decode; + return 0; + case CHIP_FIJI: + case CHIP_CARRIZO: + case CHIP_STONEY: + if (encode) + *codecs = &tonga_video_codecs_encode; + else + *codecs = &cz_video_codecs_decode; + return 0; + default: + return -EINVAL; + } +} + /* * Indirect registers accessor */ @@ -902,13 +1115,178 @@ static void vi_pcie_gen3_enable(struct amdgpu_device *adev) /* todo */ } +static void vi_enable_aspm(struct amdgpu_device *adev) +{ + u32 data, orig; + + orig = data = RREG32_PCIE(ixPCIE_LC_CNTL); + data |= PCIE_LC_CNTL__LC_L0S_INACTIVITY_DEFAULT << + PCIE_LC_CNTL__LC_L0S_INACTIVITY__SHIFT; + data |= PCIE_LC_CNTL__LC_L1_INACTIVITY_DEFAULT << + PCIE_LC_CNTL__LC_L1_INACTIVITY__SHIFT; + data &= ~PCIE_LC_CNTL__LC_PMI_TO_L1_DIS_MASK; + data |= PCIE_LC_CNTL__LC_DELAY_L1_EXIT_MASK; + if (orig != data) + WREG32_PCIE(ixPCIE_LC_CNTL, data); +} + static void vi_program_aspm(struct amdgpu_device *adev) { + u32 data, data1, orig; + bool bL1SS = false; + bool bClkReqSupport = true; - if (amdgpu_aspm == 0) + if (amdgpu_aspm != 1) return; - /* todo */ + if (adev->flags & AMD_IS_APU || + adev->asic_type < CHIP_POLARIS10) + return; + + orig = data = RREG32_PCIE(ixPCIE_LC_CNTL); + data &= ~PCIE_LC_CNTL__LC_L1_INACTIVITY_MASK; + data &= ~PCIE_LC_CNTL__LC_L0S_INACTIVITY_MASK; + data |= PCIE_LC_CNTL__LC_PMI_TO_L1_DIS_MASK; + if (orig != data) + WREG32_PCIE(ixPCIE_LC_CNTL, data); + + orig = data = RREG32_PCIE(ixPCIE_LC_N_FTS_CNTL); + data &= ~PCIE_LC_N_FTS_CNTL__LC_XMIT_N_FTS_MASK; + data |= 0x0024 << PCIE_LC_N_FTS_CNTL__LC_XMIT_N_FTS__SHIFT; + data |= PCIE_LC_N_FTS_CNTL__LC_XMIT_N_FTS_OVERRIDE_EN_MASK; + if (orig != data) + WREG32_PCIE(ixPCIE_LC_N_FTS_CNTL, data); + + orig = data = RREG32_PCIE(ixPCIE_LC_CNTL3); + data |= PCIE_LC_CNTL3__LC_GO_TO_RECOVERY_MASK; + if (orig != data) + WREG32_PCIE(ixPCIE_LC_CNTL3, data); + + orig = data = RREG32_PCIE(ixPCIE_P_CNTL); + data |= PCIE_P_CNTL__P_IGNORE_EDB_ERR_MASK; + if (orig != data) + WREG32_PCIE(ixPCIE_P_CNTL, data); + + data = RREG32_PCIE(ixPCIE_LC_L1_PM_SUBSTATE); + pci_read_config_dword(adev->pdev, PCIE_L1_PM_SUB_CNTL, &data1); + if (data & PCIE_LC_L1_PM_SUBSTATE__LC_L1_SUBSTATES_OVERRIDE_EN_MASK && + (data & (PCIE_LC_L1_PM_SUBSTATE__LC_PCI_PM_L1_2_OVERRIDE_MASK | + PCIE_LC_L1_PM_SUBSTATE__LC_PCI_PM_L1_1_OVERRIDE_MASK | + PCIE_LC_L1_PM_SUBSTATE__LC_ASPM_L1_2_OVERRIDE_MASK | + PCIE_LC_L1_PM_SUBSTATE__LC_ASPM_L1_1_OVERRIDE_MASK))) { + bL1SS = true; + } else if (data1 & (PCIE_L1_PM_SUB_CNTL__ASPM_L1_2_EN_MASK | + PCIE_L1_PM_SUB_CNTL__ASPM_L1_1_EN_MASK | + PCIE_L1_PM_SUB_CNTL__PCI_PM_L1_2_EN_MASK | + PCIE_L1_PM_SUB_CNTL__PCI_PM_L1_1_EN_MASK)) { + bL1SS = true; + } + + orig = data = RREG32_PCIE(ixPCIE_LC_CNTL6); + data |= PCIE_LC_CNTL6__LC_L1_POWERDOWN_MASK; + if (orig != data) + WREG32_PCIE(ixPCIE_LC_CNTL6, data); + + orig = data = RREG32_PCIE(ixPCIE_LC_LINK_WIDTH_CNTL); + data |= PCIE_LC_LINK_WIDTH_CNTL__LC_DYN_LANES_PWR_STATE_MASK; + if (orig != data) + WREG32_PCIE(ixPCIE_LC_LINK_WIDTH_CNTL, data); + + pci_read_config_dword(adev->pdev, LINK_CAP, &data); + if (!(data & PCIE_LINK_CAP__CLOCK_POWER_MANAGEMENT_MASK)) + bClkReqSupport = false; + + if (bClkReqSupport) { + orig = data = RREG32_SMC(ixTHM_CLK_CNTL); + data &= ~(THM_CLK_CNTL__CMON_CLK_SEL_MASK | THM_CLK_CNTL__TMON_CLK_SEL_MASK); + data |= (1 << THM_CLK_CNTL__CMON_CLK_SEL__SHIFT) | + (1 << THM_CLK_CNTL__TMON_CLK_SEL__SHIFT); + if (orig != data) + WREG32_SMC(ixTHM_CLK_CNTL, data); + + orig = data = RREG32_SMC(ixMISC_CLK_CTRL); + data &= ~(MISC_CLK_CTRL__DEEP_SLEEP_CLK_SEL_MASK | + MISC_CLK_CTRL__ZCLK_SEL_MASK | MISC_CLK_CTRL__DFT_SMS_PG_CLK_SEL_MASK); + data |= (1 << MISC_CLK_CTRL__DEEP_SLEEP_CLK_SEL__SHIFT) | + (1 << MISC_CLK_CTRL__ZCLK_SEL__SHIFT); + data |= (0x20 << MISC_CLK_CTRL__DFT_SMS_PG_CLK_SEL__SHIFT); + if (orig != data) + WREG32_SMC(ixMISC_CLK_CTRL, data); + + orig = data = RREG32_SMC(ixCG_CLKPIN_CNTL); + data |= CG_CLKPIN_CNTL__XTALIN_DIVIDE_MASK; + if (orig != data) + WREG32_SMC(ixCG_CLKPIN_CNTL, data); + + orig = data = RREG32_SMC(ixCG_CLKPIN_CNTL_2); + data |= CG_CLKPIN_CNTL_2__ENABLE_XCLK_MASK; + if (orig != data) + WREG32_SMC(ixCG_CLKPIN_CNTL, data); + + orig = data = RREG32_SMC(ixMPLL_BYPASSCLK_SEL); + data &= ~MPLL_BYPASSCLK_SEL__MPLL_CLKOUT_SEL_MASK; + data |= (4 << MPLL_BYPASSCLK_SEL__MPLL_CLKOUT_SEL__SHIFT); + if (orig != data) + WREG32_SMC(ixMPLL_BYPASSCLK_SEL, data); + + orig = data = RREG32_PCIE(ixCPM_CONTROL); + data |= (CPM_CONTROL__REFCLK_XSTCLK_ENABLE_MASK | + CPM_CONTROL__CLKREQb_UNGATE_TXCLK_ENABLE_MASK); + if (orig != data) + WREG32_PCIE(ixCPM_CONTROL, data); + + orig = data = RREG32_PCIE(ixPCIE_CONFIG_CNTL); + data &= ~PCIE_CONFIG_CNTL__DYN_CLK_LATENCY_MASK; + data |= (0xE << PCIE_CONFIG_CNTL__DYN_CLK_LATENCY__SHIFT); + if (orig != data) + WREG32_PCIE(ixPCIE_CONFIG_CNTL, data); + + orig = data = RREG32(mmBIF_CLK_CTRL); + data |= BIF_CLK_CTRL__BIF_XSTCLK_READY_MASK; + if (orig != data) + WREG32(mmBIF_CLK_CTRL, data); + + orig = data = RREG32_PCIE(ixPCIE_LC_CNTL7); + data |= PCIE_LC_CNTL7__LC_L1_SIDEBAND_CLKREQ_PDWN_EN_MASK; + if (orig != data) + WREG32_PCIE(ixPCIE_LC_CNTL7, data); + + orig = data = RREG32_PCIE(ixPCIE_HW_DEBUG); + data |= PCIE_HW_DEBUG__HW_01_DEBUG_MASK; + if (orig != data) + WREG32_PCIE(ixPCIE_HW_DEBUG, data); + + orig = data = RREG32_PCIE(ixPCIE_LC_CNTL2); + data |= PCIE_LC_CNTL2__LC_ALLOW_PDWN_IN_L23_MASK; + data |= PCIE_LC_CNTL2__LC_ALLOW_PDWN_IN_L1_MASK; + if (bL1SS) + data &= ~PCIE_LC_CNTL2__LC_ALLOW_PDWN_IN_L1_MASK; + if (orig != data) + WREG32_PCIE(ixPCIE_LC_CNTL2, data); + + } + + vi_enable_aspm(adev); + + data = RREG32_PCIE(ixPCIE_LC_N_FTS_CNTL); + data1 = RREG32_PCIE(ixPCIE_LC_STATUS1); + if (((data & PCIE_LC_N_FTS_CNTL__LC_N_FTS_MASK) == PCIE_LC_N_FTS_CNTL__LC_N_FTS_MASK) && + data1 & PCIE_LC_STATUS1__LC_REVERSE_XMIT_MASK && + data1 & PCIE_LC_STATUS1__LC_REVERSE_RCVR_MASK) { + orig = data = RREG32_PCIE(ixPCIE_LC_CNTL); + data &= ~PCIE_LC_CNTL__LC_L0S_INACTIVITY_MASK; + if (orig != data) + WREG32_PCIE(ixPCIE_LC_CNTL, data); + } + + if ((adev->asic_type == CHIP_POLARIS12 && + !(ASICID_IS_P23(adev->pdev->device, adev->pdev->revision))) || + ASIC_IS_P22(adev->asic_type, adev->external_rev_id)) { + orig = data = RREG32_PCIE(ixPCIE_LC_TRAINING_CNTL); + data &= ~PCIE_LC_TRAINING_CNTL__LC_DISABLE_TRAINING_BIT_ARCH_MASK; + if (orig != data) + WREG32_PCIE(ixPCIE_LC_TRAINING_CNTL, data); + } } static void vi_enable_doorbell_aperture(struct amdgpu_device *adev, @@ -1085,6 +1463,7 @@ static const struct amdgpu_asic_funcs vi_asic_funcs = .get_pcie_replay_count = &vi_get_pcie_replay_count, .supports_baco = &vi_asic_supports_baco, .pre_asic_init = &vi_pre_asic_init, + .query_video_codecs = &vi_query_video_codecs, }; #define CZ_REV_BRISTOL(rev) \ |