summaryrefslogtreecommitdiffstats
path: root/drivers/gpu/drm/amd/amdgpu
diff options
context:
space:
mode:
Diffstat (limited to 'drivers/gpu/drm/amd/amdgpu')
-rw-r--r--drivers/gpu/drm/amd/amdgpu/amdgpu.h142
-rw-r--r--drivers/gpu/drm/amd/amdgpu/amdgpu_acpi.c114
-rw-r--r--drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.c29
-rw-r--r--drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gpuvm.c171
-rw-r--r--drivers/gpu/drm/amd/amdgpu/amdgpu_csa.c2
-rw-r--r--drivers/gpu/drm/amd/amdgpu/amdgpu_ctx.c1
-rw-r--r--drivers/gpu/drm/amd/amdgpu/amdgpu_device.c393
-rw-r--r--drivers/gpu/drm/amd/amdgpu/amdgpu_display.c12
-rw-r--r--drivers/gpu/drm/amd/amdgpu/amdgpu_doorbell.h243
-rw-r--r--drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c16
-rw-r--r--drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.c2
-rw-r--r--drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c6
-rw-r--r--drivers/gpu/drm/amd/amdgpu/amdgpu_mode.h10
-rw-r--r--drivers/gpu/drm/amd/amdgpu/amdgpu_pm.c26
-rw-r--r--drivers/gpu/drm/amd/amdgpu/amdgpu_uvd.c2
-rw-r--r--drivers/gpu/drm/amd/amdgpu/amdgpu_uvd.h2
-rw-r--r--drivers/gpu/drm/amd/amdgpu/amdgpu_virt.c2
-rw-r--r--drivers/gpu/drm/amd/amdgpu/amdgpu_virt.h24
-rw-r--r--drivers/gpu/drm/amd/amdgpu/amdgpu_xgmi.c68
-rw-r--r--drivers/gpu/drm/amd/amdgpu/amdgpu_xgmi.h39
-rw-r--r--drivers/gpu/drm/amd/amdgpu/cik.c1
-rw-r--r--drivers/gpu/drm/amd/amdgpu/cik.h1
-rw-r--r--drivers/gpu/drm/amd/amdgpu/gfx_v7_0.c2
-rw-r--r--drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c11
-rw-r--r--drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c8
-rw-r--r--drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c9
-rw-r--r--drivers/gpu/drm/amd/amdgpu/mxgpu_vi.c1
-rw-r--r--drivers/gpu/drm/amd/amdgpu/psp_v10_0.c3
-rw-r--r--drivers/gpu/drm/amd/amdgpu/psp_v3_1.c5
-rw-r--r--drivers/gpu/drm/amd/amdgpu/sdma_v3_0.c2
-rw-r--r--drivers/gpu/drm/amd/amdgpu/sdma_v4_0.c77
-rw-r--r--drivers/gpu/drm/amd/amdgpu/soc15.c23
-rw-r--r--drivers/gpu/drm/amd/amdgpu/soc15.h2
-rw-r--r--drivers/gpu/drm/amd/amdgpu/tonga_ih.c2
-rw-r--r--drivers/gpu/drm/amd/amdgpu/uvd_v4_2.c8
-rw-r--r--drivers/gpu/drm/amd/amdgpu/uvd_v5_0.c8
-rw-r--r--drivers/gpu/drm/amd/amdgpu/uvd_v6_0.c8
-rw-r--r--drivers/gpu/drm/amd/amdgpu/uvd_v7_0.c12
-rw-r--r--drivers/gpu/drm/amd/amdgpu/vce_v3_0.c1
-rw-r--r--drivers/gpu/drm/amd/amdgpu/vce_v4_0.c4
-rw-r--r--drivers/gpu/drm/amd/amdgpu/vega10_ih.c2
-rw-r--r--drivers/gpu/drm/amd/amdgpu/vega10_reg_init.c28
-rw-r--r--drivers/gpu/drm/amd/amdgpu/vega20_reg_init.c33
-rw-r--r--drivers/gpu/drm/amd/amdgpu/vi.c19
-rw-r--r--drivers/gpu/drm/amd/amdgpu/vi.h1
45 files changed, 1065 insertions, 510 deletions
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu.h b/drivers/gpu/drm/amd/amdgpu/amdgpu.h
index 42f882c633ee..c8ad6bf6618a 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu.h
@@ -81,6 +81,7 @@
#include "amdgpu_job.h"
#include "amdgpu_bo_list.h"
#include "amdgpu_gem.h"
+#include "amdgpu_doorbell.h"
#define MAX_GPU_INSTANCE 16
@@ -361,123 +362,6 @@ int amdgpu_fence_slab_init(void);
void amdgpu_fence_slab_fini(void);
/*
- * GPU doorbell structures, functions & helpers
- */
-typedef enum _AMDGPU_DOORBELL_ASSIGNMENT
-{
- AMDGPU_DOORBELL_KIQ = 0x000,
- AMDGPU_DOORBELL_HIQ = 0x001,
- AMDGPU_DOORBELL_DIQ = 0x002,
- AMDGPU_DOORBELL_MEC_RING0 = 0x010,
- AMDGPU_DOORBELL_MEC_RING1 = 0x011,
- AMDGPU_DOORBELL_MEC_RING2 = 0x012,
- AMDGPU_DOORBELL_MEC_RING3 = 0x013,
- AMDGPU_DOORBELL_MEC_RING4 = 0x014,
- AMDGPU_DOORBELL_MEC_RING5 = 0x015,
- AMDGPU_DOORBELL_MEC_RING6 = 0x016,
- AMDGPU_DOORBELL_MEC_RING7 = 0x017,
- AMDGPU_DOORBELL_GFX_RING0 = 0x020,
- AMDGPU_DOORBELL_sDMA_ENGINE0 = 0x1E0,
- AMDGPU_DOORBELL_sDMA_ENGINE1 = 0x1E1,
- AMDGPU_DOORBELL_IH = 0x1E8,
- AMDGPU_DOORBELL_MAX_ASSIGNMENT = 0x3FF,
- AMDGPU_DOORBELL_INVALID = 0xFFFF
-} AMDGPU_DOORBELL_ASSIGNMENT;
-
-struct amdgpu_doorbell {
- /* doorbell mmio */
- resource_size_t base;
- resource_size_t size;
- u32 __iomem *ptr;
- u32 num_doorbells; /* Number of doorbells actually reserved for amdgpu. */
-};
-
-/*
- * 64bit doorbell, offset are in QWORD, occupy 2KB doorbell space
- */
-typedef enum _AMDGPU_DOORBELL64_ASSIGNMENT
-{
- /*
- * All compute related doorbells: kiq, hiq, diq, traditional compute queue, user queue, should locate in
- * a continues range so that programming CP_MEC_DOORBELL_RANGE_LOWER/UPPER can cover this range.
- * Compute related doorbells are allocated from 0x00 to 0x8a
- */
-
-
- /* kernel scheduling */
- AMDGPU_DOORBELL64_KIQ = 0x00,
-
- /* HSA interface queue and debug queue */
- AMDGPU_DOORBELL64_HIQ = 0x01,
- AMDGPU_DOORBELL64_DIQ = 0x02,
-
- /* Compute engines */
- AMDGPU_DOORBELL64_MEC_RING0 = 0x03,
- AMDGPU_DOORBELL64_MEC_RING1 = 0x04,
- AMDGPU_DOORBELL64_MEC_RING2 = 0x05,
- AMDGPU_DOORBELL64_MEC_RING3 = 0x06,
- AMDGPU_DOORBELL64_MEC_RING4 = 0x07,
- AMDGPU_DOORBELL64_MEC_RING5 = 0x08,
- AMDGPU_DOORBELL64_MEC_RING6 = 0x09,
- AMDGPU_DOORBELL64_MEC_RING7 = 0x0a,
-
- /* User queue doorbell range (128 doorbells) */
- AMDGPU_DOORBELL64_USERQUEUE_START = 0x0b,
- AMDGPU_DOORBELL64_USERQUEUE_END = 0x8a,
-
- /* Graphics engine */
- AMDGPU_DOORBELL64_GFX_RING0 = 0x8b,
-
- /*
- * Other graphics doorbells can be allocated here: from 0x8c to 0xdf
- * Graphics voltage island aperture 1
- * default non-graphics QWORD index is 0xe0 - 0xFF inclusive
- */
-
- /* sDMA engines reserved from 0xe0 -0xef */
- AMDGPU_DOORBELL64_sDMA_ENGINE0 = 0xE0,
- AMDGPU_DOORBELL64_sDMA_HI_PRI_ENGINE0 = 0xE1,
- AMDGPU_DOORBELL64_sDMA_ENGINE1 = 0xE8,
- AMDGPU_DOORBELL64_sDMA_HI_PRI_ENGINE1 = 0xE9,
-
- /* For vega10 sriov, the sdma doorbell must be fixed as follow
- * to keep the same setting with host driver, or it will
- * happen conflicts
- */
- AMDGPU_VEGA10_DOORBELL64_sDMA_ENGINE0 = 0xF0,
- AMDGPU_VEGA10_DOORBELL64_sDMA_HI_PRI_ENGINE0 = 0xF1,
- AMDGPU_VEGA10_DOORBELL64_sDMA_ENGINE1 = 0xF2,
- AMDGPU_VEGA10_DOORBELL64_sDMA_HI_PRI_ENGINE1 = 0xF3,
-
- /* Interrupt handler */
- AMDGPU_DOORBELL64_IH = 0xF4, /* For legacy interrupt ring buffer */
- AMDGPU_DOORBELL64_IH_RING1 = 0xF5, /* For page migration request log */
- AMDGPU_DOORBELL64_IH_RING2 = 0xF6, /* For page migration translation/invalidation log */
-
- /* VCN engine use 32 bits doorbell */
- AMDGPU_DOORBELL64_VCN0_1 = 0xF8, /* lower 32 bits for VNC0 and upper 32 bits for VNC1 */
- AMDGPU_DOORBELL64_VCN2_3 = 0xF9,
- AMDGPU_DOORBELL64_VCN4_5 = 0xFA,
- AMDGPU_DOORBELL64_VCN6_7 = 0xFB,
-
- /* overlap the doorbell assignment with VCN as they are mutually exclusive
- * VCE engine's doorbell is 32 bit and two VCE ring share one QWORD
- */
- AMDGPU_DOORBELL64_UVD_RING0_1 = 0xF8,
- AMDGPU_DOORBELL64_UVD_RING2_3 = 0xF9,
- AMDGPU_DOORBELL64_UVD_RING4_5 = 0xFA,
- AMDGPU_DOORBELL64_UVD_RING6_7 = 0xFB,
-
- AMDGPU_DOORBELL64_VCE_RING0_1 = 0xFC,
- AMDGPU_DOORBELL64_VCE_RING2_3 = 0xFD,
- AMDGPU_DOORBELL64_VCE_RING4_5 = 0xFE,
- AMDGPU_DOORBELL64_VCE_RING6_7 = 0xFF,
-
- AMDGPU_DOORBELL64_MAX_ASSIGNMENT = 0xFF,
- AMDGPU_DOORBELL64_INVALID = 0xFFFF
-} AMDGPU_DOORBELL64_ASSIGNMENT;
-
-/*
* IRQS.
*/
@@ -654,6 +538,8 @@ struct amdgpu_asic_funcs {
struct amdgpu_ring *ring);
/* check if the asic needs a full reset of if soft reset will work */
bool (*need_full_reset)(struct amdgpu_device *adev);
+ /* initialize doorbell layout for specific asic*/
+ void (*init_doorbell_index)(struct amdgpu_device *adev);
};
/*
@@ -1023,6 +909,8 @@ struct amdgpu_device {
unsigned long last_mm_index;
bool in_gpu_reset;
struct mutex lock_reset;
+ struct amdgpu_doorbell_index doorbell_index;
+ int asic_reset_res;
};
static inline struct amdgpu_device *amdgpu_ttm_adev(struct ttm_bo_device *bdev)
@@ -1047,11 +935,6 @@ 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_mm_rdoorbell(struct amdgpu_device *adev, u32 index);
-void amdgpu_mm_wdoorbell(struct amdgpu_device *adev, u32 index, u32 v);
-u64 amdgpu_mm_rdoorbell64(struct amdgpu_device *adev, u32 index);
-void amdgpu_mm_wdoorbell64(struct amdgpu_device *adev, u32 index, u64 v);
-
bool amdgpu_device_asic_has_dc_support(enum amd_asic_type asic_type);
bool amdgpu_device_has_dc_support(struct amdgpu_device *adev);
@@ -1113,11 +996,6 @@ int emu_soc_asic_init(struct amdgpu_device *adev);
#define RREG32_IO(reg) amdgpu_io_rreg(adev, (reg))
#define WREG32_IO(reg, v) amdgpu_io_wreg(adev, (reg), (v))
-#define RDOORBELL32(index) amdgpu_mm_rdoorbell(adev, (index))
-#define WDOORBELL32(index, v) amdgpu_mm_wdoorbell(adev, (index), (v))
-#define RDOORBELL64(index) amdgpu_mm_rdoorbell64(adev, (index))
-#define WDOORBELL64(index, v) amdgpu_mm_wdoorbell64(adev, (index), (v))
-
#define REG_FIELD_SHIFT(reg, field) reg##__##field##__SHIFT
#define REG_FIELD_MASK(reg, field) reg##__##field##_MASK
@@ -1159,6 +1037,7 @@ int emu_soc_asic_init(struct amdgpu_device *adev);
#define amdgpu_asic_flush_hdp(adev, r) (adev)->asic_funcs->flush_hdp((adev), (r))
#define amdgpu_asic_invalidate_hdp(adev, r) (adev)->asic_funcs->invalidate_hdp((adev), (r))
#define amdgpu_asic_need_full_reset(adev) (adev)->asic_funcs->need_full_reset((adev))
+#define amdgpu_asic_init_doorbell_index(adev) (adev)->asic_funcs->init_doorbell_index((adev))
/* Common functions */
bool amdgpu_device_should_recover_gpu(struct amdgpu_device *adev);
@@ -1219,12 +1098,6 @@ void amdgpu_disable_vblank_kms(struct drm_device *dev, unsigned int pipe);
long amdgpu_kms_compat_ioctl(struct file *filp, unsigned int cmd,
unsigned long arg);
-
-/*
- * functions used by amdgpu_xgmi.c
- */
-int amdgpu_xgmi_add_device(struct amdgpu_device *adev);
-
/*
* functions used by amdgpu_encoder.c
*/
@@ -1252,6 +1125,9 @@ bool amdgpu_acpi_is_pcie_performance_request_supported(struct amdgpu_device *ade
int amdgpu_acpi_pcie_performance_request(struct amdgpu_device *adev,
u8 perf_req, bool advertise);
int amdgpu_acpi_pcie_notify_device_ready(struct amdgpu_device *adev);
+
+void amdgpu_acpi_get_backlight_caps(struct amdgpu_device *adev,
+ struct amdgpu_dm_backlight_caps *caps);
#else
static inline int amdgpu_acpi_init(struct amdgpu_device *adev) { return 0; }
static inline void amdgpu_acpi_fini(struct amdgpu_device *adev) { }
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_acpi.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_acpi.c
index 7f0afc526419..47db65926d71 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_acpi.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_acpi.c
@@ -41,28 +41,21 @@ struct amdgpu_atif_notification_cfg {
};
struct amdgpu_atif_notifications {
- bool display_switch;
- bool expansion_mode_change;
bool thermal_state;
bool forced_power_state;
bool system_power_state;
- bool display_conf_change;
- bool px_gfx_switch;
bool brightness_change;
bool dgpu_display_event;
+ bool gpu_package_power_limit;
};
struct amdgpu_atif_functions {
bool system_params;
bool sbios_requests;
- bool select_active_disp;
- bool lid_state;
- bool get_tv_standard;
- bool set_tv_standard;
- bool get_panel_expansion_mode;
- bool set_panel_expansion_mode;
bool temperature_change;
- bool graphics_device_types;
+ bool query_backlight_transfer_characteristics;
+ bool ready_to_undock;
+ bool external_gpu_information;
};
struct amdgpu_atif {
@@ -72,6 +65,7 @@ struct amdgpu_atif {
struct amdgpu_atif_functions functions;
struct amdgpu_atif_notification_cfg notification_cfg;
struct amdgpu_encoder *encoder_for_bl;
+ struct amdgpu_dm_backlight_caps backlight_caps;
};
/* Call the ATIF method
@@ -137,15 +131,12 @@ static union acpi_object *amdgpu_atif_call(struct amdgpu_atif *atif,
*/
static void amdgpu_atif_parse_notification(struct amdgpu_atif_notifications *n, u32 mask)
{
- n->display_switch = mask & ATIF_DISPLAY_SWITCH_REQUEST_SUPPORTED;
- n->expansion_mode_change = mask & ATIF_EXPANSION_MODE_CHANGE_REQUEST_SUPPORTED;
n->thermal_state = mask & ATIF_THERMAL_STATE_CHANGE_REQUEST_SUPPORTED;
n->forced_power_state = mask & ATIF_FORCED_POWER_STATE_CHANGE_REQUEST_SUPPORTED;
n->system_power_state = mask & ATIF_SYSTEM_POWER_SOURCE_CHANGE_REQUEST_SUPPORTED;
- n->display_conf_change = mask & ATIF_DISPLAY_CONF_CHANGE_REQUEST_SUPPORTED;
- n->px_gfx_switch = mask & ATIF_PX_GFX_SWITCH_REQUEST_SUPPORTED;
n->brightness_change = mask & ATIF_PANEL_BRIGHTNESS_CHANGE_REQUEST_SUPPORTED;
n->dgpu_display_event = mask & ATIF_DGPU_DISPLAY_EVENT_SUPPORTED;
+ n->gpu_package_power_limit = mask & ATIF_GPU_PACKAGE_POWER_LIMIT_REQUEST_SUPPORTED;
}
/**
@@ -162,14 +153,11 @@ static void amdgpu_atif_parse_functions(struct amdgpu_atif_functions *f, u32 mas
{
f->system_params = mask & ATIF_GET_SYSTEM_PARAMETERS_SUPPORTED;
f->sbios_requests = mask & ATIF_GET_SYSTEM_BIOS_REQUESTS_SUPPORTED;
- f->select_active_disp = mask & ATIF_SELECT_ACTIVE_DISPLAYS_SUPPORTED;
- f->lid_state = mask & ATIF_GET_LID_STATE_SUPPORTED;
- f->get_tv_standard = mask & ATIF_GET_TV_STANDARD_FROM_CMOS_SUPPORTED;
- f->set_tv_standard = mask & ATIF_SET_TV_STANDARD_IN_CMOS_SUPPORTED;
- f->get_panel_expansion_mode = mask & ATIF_GET_PANEL_EXPANSION_MODE_FROM_CMOS_SUPPORTED;
- f->set_panel_expansion_mode = mask & ATIF_SET_PANEL_EXPANSION_MODE_IN_CMOS_SUPPORTED;
f->temperature_change = mask & ATIF_TEMPERATURE_CHANGE_NOTIFICATION_SUPPORTED;
- f->graphics_device_types = mask & ATIF_GET_GRAPHICS_DEVICE_TYPES_SUPPORTED;
+ f->query_backlight_transfer_characteristics =
+ mask & ATIF_QUERY_BACKLIGHT_TRANSFER_CHARACTERISTICS_SUPPORTED;
+ f->ready_to_undock = mask & ATIF_READY_TO_UNDOCK_NOTIFICATION_SUPPORTED;
+ f->external_gpu_information = mask & ATIF_GET_EXTERNAL_GPU_INFORMATION_SUPPORTED;
}
/**
@@ -311,6 +299,65 @@ out:
}
/**
+ * amdgpu_atif_query_backlight_caps - get min and max backlight input signal
+ *
+ * @handle: acpi handle
+ *
+ * Execute the QUERY_BRIGHTNESS_TRANSFER_CHARACTERISTICS ATIF function
+ * to determine the acceptable range of backlight values
+ *
+ * Backlight_caps.caps_valid will be set to true if the query is successful
+ *
+ * The input signals are in range 0-255
+ *
+ * This function assumes the display with backlight is the first LCD
+ *
+ * Returns 0 on success, error on failure.
+ */
+static int amdgpu_atif_query_backlight_caps(struct amdgpu_atif *atif)
+{
+ union acpi_object *info;
+ struct atif_qbtc_output characteristics;
+ struct atif_qbtc_arguments arguments;
+ struct acpi_buffer params;
+ size_t size;
+ int err = 0;
+
+ arguments.size = sizeof(arguments);
+ arguments.requested_display = ATIF_QBTC_REQUEST_LCD1;
+
+ params.length = sizeof(arguments);
+ params.pointer = (void *)&arguments;
+
+ info = amdgpu_atif_call(atif,
+ ATIF_FUNCTION_QUERY_BRIGHTNESS_TRANSFER_CHARACTERISTICS,
+ &params);
+ if (!info) {
+ err = -EIO;
+ goto out;
+ }
+
+ size = *(u16 *) info->buffer.pointer;
+ if (size < 10) {
+ err = -EINVAL;
+ goto out;
+ }
+
+ memset(&characteristics, 0, sizeof(characteristics));
+ size = min(sizeof(characteristics), size);
+ memcpy(&characteristics, info->buffer.pointer, size);
+
+ atif->backlight_caps.caps_valid = true;
+ atif->backlight_caps.min_input_signal =
+ characteristics.min_input_signal;
+ atif->backlight_caps.max_input_signal =
+ characteristics.max_input_signal;
+out:
+ kfree(info);
+ return err;
+}
+
+/**
* amdgpu_atif_get_sbios_requests - get requested sbios event
*
* @handle: acpi handle
@@ -799,6 +846,17 @@ int amdgpu_acpi_init(struct amdgpu_device *adev)
}
}
+ if (atif->functions.query_backlight_transfer_characteristics) {
+ ret = amdgpu_atif_query_backlight_caps(atif);
+ if (ret) {
+ DRM_DEBUG_DRIVER("Call to QUERY_BACKLIGHT_TRANSFER_CHARACTERISTICS failed: %d\n",
+ ret);
+ atif->backlight_caps.caps_valid = false;
+ }
+ } else {
+ atif->backlight_caps.caps_valid = false;
+ }
+
out:
adev->acpi_nb.notifier_call = amdgpu_acpi_event;
register_acpi_notifier(&adev->acpi_nb);
@@ -806,6 +864,18 @@ out:
return ret;
}
+void amdgpu_acpi_get_backlight_caps(struct amdgpu_device *adev,
+ struct amdgpu_dm_backlight_caps *caps)
+{
+ if (!adev->atif) {
+ caps->caps_valid = false;
+ return;
+ }
+ caps->caps_valid = adev->atif->backlight_caps.caps_valid;
+ caps->min_input_signal = adev->atif->backlight_caps.min_input_signal;
+ caps->max_input_signal = adev->atif->backlight_caps.max_input_signal;
+}
+
/**
* amdgpu_acpi_fini - tear down driver acpi support
*
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.c
index bcf1666fb31d..d693b8047653 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.c
@@ -73,9 +73,11 @@ void amdgpu_amdkfd_device_probe(struct amdgpu_device *adev)
case CHIP_FIJI:
case CHIP_POLARIS10:
case CHIP_POLARIS11:
+ case CHIP_POLARIS12:
kfd2kgd = amdgpu_amdkfd_gfx_8_0_get_functions();
break;
case CHIP_VEGA10:
+ case CHIP_VEGA12:
case CHIP_VEGA20:
case CHIP_RAVEN:
kfd2kgd = amdgpu_amdkfd_gfx_9_0_get_functions();
@@ -179,25 +181,14 @@ void amdgpu_amdkfd_device_init(struct amdgpu_device *adev)
* process in case of 64-bit doorbells so we
* can use each doorbell assignment twice.
*/
- if (adev->asic_type == CHIP_VEGA10) {
- gpu_resources.sdma_doorbell[0][i] =
- AMDGPU_VEGA10_DOORBELL64_sDMA_ENGINE0 + (i >> 1);
- gpu_resources.sdma_doorbell[0][i+1] =
- AMDGPU_VEGA10_DOORBELL64_sDMA_ENGINE0 + 0x200 + (i >> 1);
- gpu_resources.sdma_doorbell[1][i] =
- AMDGPU_VEGA10_DOORBELL64_sDMA_ENGINE1 + (i >> 1);
- gpu_resources.sdma_doorbell[1][i+1] =
- AMDGPU_VEGA10_DOORBELL64_sDMA_ENGINE1 + 0x200 + (i >> 1);
- } else {
- gpu_resources.sdma_doorbell[0][i] =
- AMDGPU_DOORBELL64_sDMA_ENGINE0 + (i >> 1);
- gpu_resources.sdma_doorbell[0][i+1] =
- AMDGPU_DOORBELL64_sDMA_ENGINE0 + 0x200 + (i >> 1);
- gpu_resources.sdma_doorbell[1][i] =
- AMDGPU_DOORBELL64_sDMA_ENGINE1 + (i >> 1);
- gpu_resources.sdma_doorbell[1][i+1] =
- AMDGPU_DOORBELL64_sDMA_ENGINE1 + 0x200 + (i >> 1);
- }
+ gpu_resources.sdma_doorbell[0][i] =
+ adev->doorbell_index.sdma_engine0 + (i >> 1);
+ gpu_resources.sdma_doorbell[0][i+1] =
+ adev->doorbell_index.sdma_engine0 + 0x200 + (i >> 1);
+ gpu_resources.sdma_doorbell[1][i] =
+ adev->doorbell_index.sdma_engine1 + (i >> 1);
+ gpu_resources.sdma_doorbell[1][i+1] =
+ adev->doorbell_index.sdma_engine1 + 0x200 + (i >> 1);
}
/* Doorbells 0x0e0-0ff and 0x2e0-2ff are reserved for
* SDMA, IH and VCN. So don't use them for the CP.
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gpuvm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gpuvm.c
index df0a059565f9..f3129b912714 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gpuvm.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gpuvm.c
@@ -46,9 +46,9 @@
/* Impose limit on how much memory KFD can use */
static struct {
uint64_t max_system_mem_limit;
- uint64_t max_userptr_mem_limit;
+ uint64_t max_ttm_mem_limit;
int64_t system_mem_used;
- int64_t userptr_mem_used;
+ int64_t ttm_mem_used;
spinlock_t mem_limit_lock;
} kfd_mem_limit;
@@ -90,8 +90,8 @@ static bool check_if_add_bo_to_vm(struct amdgpu_vm *avm,
}
/* Set memory usage limits. Current, limits are
- * System (kernel) memory - 3/8th System RAM
- * Userptr memory - 3/4th System RAM
+ * System (TTM + userptr) memory - 3/4th System RAM
+ * TTM memory - 3/8th System RAM
*/
void amdgpu_amdkfd_gpuvm_init_mem_limits(void)
{
@@ -103,48 +103,54 @@ void amdgpu_amdkfd_gpuvm_init_mem_limits(void)
mem *= si.mem_unit;
spin_lock_init(&kfd_mem_limit.mem_limit_lock);
- kfd_mem_limit.max_system_mem_limit = (mem >> 1) - (mem >> 3);
- kfd_mem_limit.max_userptr_mem_limit = mem - (mem >> 2);
- pr_debug("Kernel memory limit %lluM, userptr limit %lluM\n",
+ kfd_mem_limit.max_system_mem_limit = (mem >> 1) + (mem >> 2);
+ kfd_mem_limit.max_ttm_mem_limit = (mem >> 1) - (mem >> 3);
+ pr_debug("Kernel memory limit %lluM, TTM limit %lluM\n",
(kfd_mem_limit.max_system_mem_limit >> 20),
- (kfd_mem_limit.max_userptr_mem_limit >> 20));
+ (kfd_mem_limit.max_ttm_mem_limit >> 20));
}
static int amdgpu_amdkfd_reserve_system_mem_limit(struct amdgpu_device *adev,
- uint64_t size, u32 domain)
+ uint64_t size, u32 domain, bool sg)
{
- size_t acc_size;
+ size_t acc_size, system_mem_needed, ttm_mem_needed;
int ret = 0;
acc_size = ttm_bo_dma_acc_size(&adev->mman.bdev, size,
sizeof(struct amdgpu_bo));
spin_lock(&kfd_mem_limit.mem_limit_lock);
+
if (domain == AMDGPU_GEM_DOMAIN_GTT) {
- if (kfd_mem_limit.system_mem_used + (acc_size + size) >
- kfd_mem_limit.max_system_mem_limit) {
- ret = -ENOMEM;
- goto err_no_mem;
- }
- kfd_mem_limit.system_mem_used += (acc_size + size);
- } else if (domain == AMDGPU_GEM_DOMAIN_CPU) {
- if ((kfd_mem_limit.system_mem_used + acc_size >
- kfd_mem_limit.max_system_mem_limit) ||
- (kfd_mem_limit.userptr_mem_used + (size + acc_size) >
- kfd_mem_limit.max_userptr_mem_limit)) {
- ret = -ENOMEM;
- goto err_no_mem;
- }
- kfd_mem_limit.system_mem_used += acc_size;
- kfd_mem_limit.userptr_mem_used += size;
+ /* TTM GTT memory */
+ system_mem_needed = acc_size + size;
+ ttm_mem_needed = acc_size + size;
+ } else if (domain == AMDGPU_GEM_DOMAIN_CPU && !sg) {
+ /* Userptr */
+ system_mem_needed = acc_size + size;
+ ttm_mem_needed = acc_size;
+ } else {
+ /* VRAM and SG */
+ system_mem_needed = acc_size;
+ ttm_mem_needed = acc_size;
}
-err_no_mem:
+
+ if ((kfd_mem_limit.system_mem_used + system_mem_needed >
+ kfd_mem_limit.max_system_mem_limit) ||
+ (kfd_mem_limit.ttm_mem_used + ttm_mem_needed >
+ kfd_mem_limit.max_ttm_mem_limit))
+ ret = -ENOMEM;
+ else {
+ kfd_mem_limit.system_mem_used += system_mem_needed;
+ kfd_mem_limit.ttm_mem_used += ttm_mem_needed;
+ }
+
spin_unlock(&kfd_mem_limit.mem_limit_lock);
return ret;
}
static void unreserve_system_mem_limit(struct amdgpu_device *adev,
- uint64_t size, u32 domain)
+ uint64_t size, u32 domain, bool sg)
{
size_t acc_size;
@@ -154,14 +160,18 @@ static void unreserve_system_mem_limit(struct amdgpu_device *adev,
spin_lock(&kfd_mem_limit.mem_limit_lock);
if (domain == AMDGPU_GEM_DOMAIN_GTT) {
kfd_mem_limit.system_mem_used -= (acc_size + size);
- } else if (domain == AMDGPU_GEM_DOMAIN_CPU) {
+ kfd_mem_limit.ttm_mem_used -= (acc_size + size);
+ } else if (domain == AMDGPU_GEM_DOMAIN_CPU && !sg) {
+ kfd_mem_limit.system_mem_used -= (acc_size + size);
+ kfd_mem_limit.ttm_mem_used -= acc_size;
+ } else {
kfd_mem_limit.system_mem_used -= acc_size;
- kfd_mem_limit.userptr_mem_used -= size;
+ kfd_mem_limit.ttm_mem_used -= acc_size;
}
WARN_ONCE(kfd_mem_limit.system_mem_used < 0,
"kfd system memory accounting unbalanced");
- WARN_ONCE(kfd_mem_limit.userptr_mem_used < 0,
- "kfd userptr memory accounting unbalanced");
+ WARN_ONCE(kfd_mem_limit.ttm_mem_used < 0,
+ "kfd TTM memory accounting unbalanced");
spin_unlock(&kfd_mem_limit.mem_limit_lock);
}
@@ -171,16 +181,22 @@ void amdgpu_amdkfd_unreserve_system_memory_limit(struct amdgpu_bo *bo)
spin_lock(&kfd_mem_limit.mem_limit_lock);
if (bo->flags & AMDGPU_AMDKFD_USERPTR_BO) {
- kfd_mem_limit.system_mem_used -= bo->tbo.acc_size;
- kfd_mem_limit.userptr_mem_used -= amdgpu_bo_size(bo);
+ kfd_mem_limit.system_mem_used -=
+ (bo->tbo.acc_size + amdgpu_bo_size(bo));
+ kfd_mem_limit.ttm_mem_used -= bo->tbo.acc_size;
} else if (bo->preferred_domains == AMDGPU_GEM_DOMAIN_GTT) {
kfd_mem_limit.system_mem_used -=
(bo->tbo.acc_size + amdgpu_bo_size(bo));
+ kfd_mem_limit.ttm_mem_used -=
+ (bo->tbo.acc_size + amdgpu_bo_size(bo));
+ } else {
+ kfd_mem_limit.system_mem_used -= bo->tbo.acc_size;
+ kfd_mem_limit.ttm_mem_used -= bo->tbo.acc_size;
}
WARN_ONCE(kfd_mem_limit.system_mem_used < 0,
"kfd system memory accounting unbalanced");
- WARN_ONCE(kfd_mem_limit.userptr_mem_used < 0,
- "kfd userptr memory accounting unbalanced");
+ WARN_ONCE(kfd_mem_limit.ttm_mem_used < 0,
+ "kfd TTM memory accounting unbalanced");
spin_unlock(&kfd_mem_limit.mem_limit_lock);
}
@@ -395,23 +411,6 @@ static int vm_validate_pt_pd_bos(struct amdgpu_vm *vm)
return 0;
}
-static int sync_vm_fence(struct amdgpu_device *adev, struct amdgpu_sync *sync,
- struct dma_fence *f)
-{
- int ret = amdgpu_sync_fence(adev, sync, f, false);
-
- /* Sync objects can't handle multiple GPUs (contexts) updating
- * sync->last_vm_update. Fortunately we don't need it for
- * KFD's purposes, so we can just drop that fence.
- */
- if (sync->last_vm_update) {
- dma_fence_put(sync->last_vm_update);
- sync->last_vm_update = NULL;
- }
-
- return ret;
-}
-
static int vm_update_pds(struct amdgpu_vm *vm, struct amdgpu_sync *sync)
{
struct amdgpu_bo *pd = vm->root.base.bo;
@@ -422,7 +421,7 @@ static int vm_update_pds(struct amdgpu_vm *vm, struct amdgpu_sync *sync)
if (ret)
return ret;
- return sync_vm_fence(adev, sync, vm->last_update);
+ return amdgpu_sync_fence(NULL, sync, vm->last_update, false);
}
/* add_bo_to_vm - Add a BO to a VM
@@ -826,7 +825,7 @@ static int unmap_bo_from_gpuvm(struct amdgpu_device *adev,
/* Add the eviction fence back */
amdgpu_bo_fence(pd, &vm->process_info->eviction_fence->base, true);
- sync_vm_fence(adev, sync, bo_va->last_pt_update);
+ amdgpu_sync_fence(NULL, sync, bo_va->last_pt_update, false);
return 0;
}
@@ -851,7 +850,7 @@ static int update_gpuvm_pte(struct amdgpu_device *adev,
return ret;
}
- return sync_vm_fence(adev, sync, bo_va->last_pt_update);
+ return amdgpu_sync_fence(NULL, sync, bo_va->last_pt_update, false);
}
static int map_bo_to_gpuvm(struct amdgpu_device *adev,
@@ -901,6 +900,26 @@ static int process_validate_vms(struct amdkfd_process_info *process_info)
return 0;
}
+static int process_sync_pds_resv(struct amdkfd_process_info *process_info,
+ struct amdgpu_sync *sync)
+{
+ struct amdgpu_vm *peer_vm;
+ int ret;
+
+ list_for_each_entry(peer_vm, &process_info->vm_list_head,
+ vm_list_node) {
+ struct amdgpu_bo *pd = peer_vm->root.base.bo;
+
+ ret = amdgpu_sync_resv(NULL,
+ sync, pd->tbo.resv,
+ AMDGPU_FENCE_OWNER_UNDEFINED, false);
+ if (ret)
+ return ret;
+ }
+
+ return 0;
+}
+
static int process_update_pds(struct amdkfd_process_info *process_info,
struct amdgpu_sync *sync)
{
@@ -1199,7 +1218,8 @@ int amdgpu_amdkfd_gpuvm_alloc_memory_of_gpu(
byte_align = (adev->family == AMDGPU_FAMILY_VI &&
adev->asic_type != CHIP_FIJI &&
adev->asic_type != CHIP_POLARIS10 &&
- adev->asic_type != CHIP_POLARIS11) ?
+ adev->asic_type != CHIP_POLARIS11 &&
+ adev->asic_type != CHIP_POLARIS12) ?
VI_BO_SIZE_ALIGN : 1;
mapping_flags = AMDGPU_VM_PAGE_READABLE;
@@ -1215,10 +1235,11 @@ int amdgpu_amdkfd_gpuvm_alloc_memory_of_gpu(
amdgpu_sync_create(&(*mem)->sync);
- ret = amdgpu_amdkfd_reserve_system_mem_limit(adev, size, alloc_domain);
+ ret = amdgpu_amdkfd_reserve_system_mem_limit(adev, size,
+ alloc_domain, false);
if (ret) {
pr_debug("Insufficient system memory\n");
- goto err_reserve_system_mem;
+ goto err_reserve_limit;
}
pr_debug("\tcreate BO VA 0x%llx size 0x%llx domain %s\n",
@@ -1266,10 +1287,10 @@ int amdgpu_amdkfd_gpuvm_alloc_memory_of_gpu(
allocate_init_user_pages_failed:
amdgpu_bo_unref(&bo);
/* Don't unreserve system mem limit twice */
- goto err_reserve_system_mem;
+ goto err_reserve_limit;
err_bo_create:
- unreserve_system_mem_limit(adev, size, alloc_domain);
-err_reserve_system_mem:
+ unreserve_system_mem_limit(adev, size, alloc_domain, false);
+err_reserve_limit:
mutex_destroy(&(*mem)->lock);
kfree(*mem);
return ret;
@@ -1405,7 +1426,8 @@ int amdgpu_amdkfd_gpuvm_map_memory_to_gpu(
* the queues are still stopped and we can leave mapping for
* the next restore worker
*/
- if (bo->tbo.mem.mem_type == TTM_PL_SYSTEM)
+ if (amdgpu_ttm_tt_get_usermm(bo->tbo.ttm) &&
+ bo->tbo.mem.mem_type == TTM_PL_SYSTEM)
is_invalid_userptr = true;
if (check_if_add_bo_to_vm(avm, mem)) {
@@ -2044,13 +2066,10 @@ int amdgpu_amdkfd_gpuvm_restore_process_bos(void *info, struct dma_fence **ef)
if (ret)
goto validate_map_fail;
- /* Wait for PD/PTs validate to finish */
- /* FIXME: I think this isn't needed */
- list_for_each_entry(peer_vm, &process_info->vm_list_head,
- vm_list_node) {
- struct amdgpu_bo *bo = peer_vm->root.base.bo;
-
- ttm_bo_wait(&bo->tbo, false, false);
+ ret = process_sync_pds_resv(process_info, &sync_obj);
+ if (ret) {
+ pr_debug("Memory eviction: Failed to sync to PD BO moving fence. Try again\n");
+ goto validate_map_fail;
}
/* Validate BOs and map them to GPUVM (update VM page tables). */
@@ -2066,7 +2085,11 @@ int amdgpu_amdkfd_gpuvm_restore_process_bos(void *info, struct dma_fence **ef)
pr_debug("Memory eviction: Validate BOs failed. Try again\n");
goto validate_map_fail;
}
-
+ ret = amdgpu_sync_fence(NULL, &sync_obj, bo->tbo.moving, false);
+ if (ret) {
+ pr_debug("Memory eviction: Sync BO fence failed. Try again\n");
+ goto validate_map_fail;
+ }
list_for_each_entry(bo_va_entry, &mem->bo_va_list,
bo_list) {
ret = update_gpuvm_pte((struct amdgpu_device *)
@@ -2087,6 +2110,7 @@ int amdgpu_amdkfd_gpuvm_restore_process_bos(void *info, struct dma_fence **ef)
goto validate_map_fail;
}
+ /* Wait for validate and PT updates to finish */
amdgpu_sync_wait(&sync_obj, false);
/* Release old eviction fence and create new one, because fence only
@@ -2105,10 +2129,7 @@ int amdgpu_amdkfd_gpuvm_restore_process_bos(void *info, struct dma_fence **ef)
process_info->eviction_fence = new_fence;
*ef = dma_fence_get(&new_fence->base);
- /* Wait for validate to finish and attach new eviction fence */
- list_for_each_entry(mem, &process_info->kfd_bo_list,
- validate_list.head)
- ttm_bo_wait(&mem->bo->tbo, false, false);
+ /* Attach new eviction fence to all BOs */
list_for_each_entry(mem, &process_info->kfd_bo_list,
validate_list.head)
amdgpu_bo_fence(mem->bo,
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_csa.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_csa.c
index 0c590ddf250a..5b550706ee76 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_csa.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_csa.c
@@ -43,7 +43,7 @@ int amdgpu_allocate_static_csa(struct amdgpu_device *adev, struct amdgpu_bo **bo
r = amdgpu_bo_create_kernel(adev, size, PAGE_SIZE,
domain, bo,
NULL, &ptr);
- if (!bo)
+ if (!*bo)
return -ENOMEM;
memset(ptr, 0, size);
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ctx.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ctx.c
index f9b54236102d..95f4c4139fc6 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ctx.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ctx.c
@@ -39,6 +39,7 @@ const unsigned int amdgpu_ctx_num_entities[AMDGPU_HW_IP_NUM] = {
[AMDGPU_HW_IP_UVD_ENC] = 1,
[AMDGPU_HW_IP_VCN_DEC] = 1,
[AMDGPU_HW_IP_VCN_ENC] = 1,
+ [AMDGPU_HW_IP_VCN_JPEG] = 1,
};
static int amdgput_ctx_total_num_entities(void)
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
index 590588a82471..c75badfa5c4c 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
@@ -59,6 +59,8 @@
#include "amdgpu_amdkfd.h"
#include "amdgpu_pm.h"
+#include "amdgpu_xgmi.h"
+
MODULE_FIRMWARE("amdgpu/vega10_gpu_info.bin");
MODULE_FIRMWARE("amdgpu/vega12_gpu_info.bin");
MODULE_FIRMWARE("amdgpu/raven_gpu_info.bin");
@@ -513,6 +515,8 @@ void amdgpu_device_pci_config_reset(struct amdgpu_device *adev)
*/
static int amdgpu_device_doorbell_init(struct amdgpu_device *adev)
{
+ amdgpu_asic_init_doorbell_index(adev);
+
/* No doorbell on SI hardware generation */
if (adev->asic_type < CHIP_BONAIRE) {
adev->doorbell.base = 0;
@@ -530,10 +534,19 @@ static int amdgpu_device_doorbell_init(struct amdgpu_device *adev)
adev->doorbell.size = pci_resource_len(adev->pdev, 2);
adev->doorbell.num_doorbells = min_t(u32, adev->doorbell.size / sizeof(u32),
- AMDGPU_DOORBELL_MAX_ASSIGNMENT+1);
+ adev->doorbell_index.max_assignment+1);
if (adev->doorbell.num_doorbells == 0)
return -EINVAL;
+ /* For Vega, reserve and map two pages on doorbell BAR since SDMA
+ * paging queue doorbell use the second page. The
+ * AMDGPU_DOORBELL64_MAX_ASSIGNMENT definition assumes all the
+ * doorbells are in the first page. So with paging queue enabled,
+ * the max num_doorbells should + 1 page (0x400 in dword)
+ */
+ if (adev->asic_type >= CHIP_VEGA10)
+ adev->doorbell.num_doorbells += 0x400;
+
adev->doorbell.ptr = ioremap(adev->doorbell.base,
adev->doorbell.num_doorbells *
sizeof(u32));
@@ -2458,9 +2471,6 @@ 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);
- /* doorbell bar mapping */
- amdgpu_device_doorbell_init(adev);
-
/* io port mapping */
for (i = 0; i < DEVICE_COUNT_RESOURCE; i++) {
if (pci_resource_flags(adev->pdev, i) & IORESOURCE_IO) {
@@ -2479,6 +2489,9 @@ int amdgpu_device_init(struct amdgpu_device *adev,
if (r)
return r;
+ /* 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 */
@@ -3151,86 +3164,6 @@ static int amdgpu_device_recover_vram(struct amdgpu_device *adev)
return 0;
}
-/**
- * amdgpu_device_reset - reset ASIC/GPU for bare-metal or passthrough
- *
- * @adev: amdgpu device pointer
- *
- * attempt to do soft-reset or full-reset and reinitialize Asic
- * return 0 means succeeded otherwise failed
- */
-static int amdgpu_device_reset(struct amdgpu_device *adev)
-{
- bool need_full_reset, vram_lost = 0;
- int r;
-
- need_full_reset = amdgpu_device_ip_need_full_reset(adev);
-
- if (!need_full_reset) {
- amdgpu_device_ip_pre_soft_reset(adev);
- r = amdgpu_device_ip_soft_reset(adev);
- amdgpu_device_ip_post_soft_reset(adev);
- if (r || amdgpu_device_ip_check_soft_reset(adev)) {
- DRM_INFO("soft reset failed, will fallback to full reset!\n");
- need_full_reset = true;
- }
- }
-
- if (need_full_reset) {
- r = amdgpu_device_ip_suspend(adev);
-
-retry:
- r = amdgpu_asic_reset(adev);
- /* post card */
- amdgpu_atom_asic_init(adev->mode_info.atom_context);
-
- if (!r) {
- dev_info(adev->dev, "GPU reset succeeded, trying to resume\n");
- r = amdgpu_device_ip_resume_phase1(adev);
- if (r)
- goto out;
-
- vram_lost = amdgpu_device_check_vram_lost(adev);
- if (vram_lost) {
- DRM_ERROR("VRAM is lost!\n");
- atomic_inc(&adev->vram_lost_counter);
- }
-
- r = amdgpu_gtt_mgr_recover(
- &adev->mman.bdev.man[TTM_PL_TT]);
- if (r)
- goto out;
-
- r = amdgpu_device_fw_loading(adev);
- if (r)
- return r;
-
- r = amdgpu_device_ip_resume_phase2(adev);
- if (r)
- goto out;
-
- if (vram_lost)
- amdgpu_device_fill_reset_magic(adev);
- }
- }
-
-out:
- if (!r) {
- amdgpu_irq_gpu_reset_resume_helper(adev);
- r = amdgpu_ib_ring_tests(adev);
- if (r) {
- dev_err(adev->dev, "ib ring test failed (%d).\n", r);
- r = amdgpu_device_ip_suspend(adev);
- need_full_reset = true;
- goto retry;
- }
- }
-
- if (!r)
- r = amdgpu_device_recover_vram(adev);
-
- return r;
-}
/**
* amdgpu_device_reset_sriov - reset ASIC for SR-IOV vf
@@ -3329,31 +3262,13 @@ disabled:
return false;
}
-/**
- * amdgpu_device_gpu_recover - reset the asic and recover scheduler
- *
- * @adev: amdgpu device pointer
- * @job: which job trigger hang
- *
- * Attempt to reset the GPU if it has hung (all asics).
- * Returns 0 for success or an error on failure.
- */
-int amdgpu_device_gpu_recover(struct amdgpu_device *adev,
- struct amdgpu_job *job)
-{
- int i, r, resched;
- dev_info(adev->dev, "GPU reset begin!\n");
-
- mutex_lock(&adev->lock_reset);
- atomic_inc(&adev->gpu_reset_counter);
- adev->in_gpu_reset = 1;
-
- /* Block kfd */
- amdgpu_amdkfd_pre_reset(adev);
-
- /* block TTM */
- resched = ttm_bo_lock_delayed_workqueue(&adev->mman.bdev);
+static int amdgpu_device_pre_asic_reset(struct amdgpu_device *adev,
+ struct amdgpu_job *job,
+ bool *need_full_reset_arg)
+{
+ int i, r = 0;
+ bool need_full_reset = *need_full_reset_arg;
/* block all schedulers and reset given job's ring */
for (i = 0; i < AMDGPU_MAX_RINGS; ++i) {
@@ -3373,10 +3288,123 @@ int amdgpu_device_gpu_recover(struct amdgpu_device *adev,
amdgpu_fence_driver_force_completion(ring);
}
- if (amdgpu_sriov_vf(adev))
- r = amdgpu_device_reset_sriov(adev, job ? false : true);
- else
- r = amdgpu_device_reset(adev);
+
+
+ if (!amdgpu_sriov_vf(adev)) {
+
+ if (!need_full_reset)
+ need_full_reset = amdgpu_device_ip_need_full_reset(adev);
+
+ if (!need_full_reset) {
+ amdgpu_device_ip_pre_soft_reset(adev);
+ r = amdgpu_device_ip_soft_reset(adev);
+ amdgpu_device_ip_post_soft_reset(adev);
+ if (r || amdgpu_device_ip_check_soft_reset(adev)) {
+ DRM_INFO("soft reset failed, will fallback to full reset!\n");
+ need_full_reset = true;
+ }
+ }
+
+ if (need_full_reset)
+ r = amdgpu_device_ip_suspend(adev);
+
+ *need_full_reset_arg = need_full_reset;
+ }
+
+ return r;
+}
+
+static int amdgpu_do_asic_reset(struct amdgpu_hive_info *hive,
+ struct list_head *device_list_handle,
+ bool *need_full_reset_arg)
+{
+ struct amdgpu_device *tmp_adev = NULL;
+ bool need_full_reset = *need_full_reset_arg, vram_lost = false;
+ int r = 0;
+
+ /*
+ * ASIC reset has to be done on all HGMI hive nodes ASAP
+ * to allow proper links negotiation in FW (within 1 sec)
+ */
+ if (need_full_reset) {
+ list_for_each_entry(tmp_adev, device_list_handle, gmc.xgmi.head) {
+ r = amdgpu_asic_reset(tmp_adev);
+ if (r)
+ DRM_WARN("ASIC reset failed with err r, %d for drm dev, %s",
+ r, tmp_adev->ddev->unique);
+ }
+ }
+
+
+ list_for_each_entry(tmp_adev, device_list_handle, gmc.xgmi.head) {
+ if (need_full_reset) {
+ /* post card */
+ if (amdgpu_atom_asic_init(tmp_adev->mode_info.atom_context))
+ DRM_WARN("asic atom init failed!");
+
+ if (!r) {
+ dev_info(tmp_adev->dev, "GPU reset succeeded, trying to resume\n");
+ r = amdgpu_device_ip_resume_phase1(tmp_adev);
+ if (r)
+ goto out;
+
+ vram_lost = amdgpu_device_check_vram_lost(tmp_adev);
+ if (vram_lost) {
+ DRM_ERROR("VRAM is lost!\n");
+ atomic_inc(&tmp_adev->vram_lost_counter);
+ }
+
+ r = amdgpu_gtt_mgr_recover(
+ &tmp_adev->mman.bdev.man[TTM_PL_TT]);
+ if (r)
+ goto out;
+
+ r = amdgpu_device_fw_loading(tmp_adev);
+ if (r)
+ return r;
+
+ r = amdgpu_device_ip_resume_phase2(tmp_adev);
+ if (r)
+ goto out;
+
+ if (vram_lost)
+ amdgpu_device_fill_reset_magic(tmp_adev);
+
+ /* Update PSP FW topology after reset */
+ if (hive && tmp_adev->gmc.xgmi.num_physical_nodes > 1)
+ r = amdgpu_xgmi_update_topology(hive, tmp_adev);
+ }
+ }
+
+
+out:
+ 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 = amdgpu_device_ip_suspend(tmp_adev);
+ need_full_reset = true;
+ r = -EAGAIN;
+ goto end;
+ }
+ }
+
+ if (!r)
+ r = amdgpu_device_recover_vram(tmp_adev);
+ else
+ tmp_adev->asic_reset_res = r;
+ }
+
+end:
+ *need_full_reset_arg = need_full_reset;
+ return r;
+}
+
+static void amdgpu_device_post_asic_reset(struct amdgpu_device *adev,
+ struct amdgpu_job *job)
+{
+ int i;
for (i = 0; i < AMDGPU_MAX_RINGS; ++i) {
struct amdgpu_ring *ring = adev->rings[i];
@@ -3388,7 +3416,7 @@ int amdgpu_device_gpu_recover(struct amdgpu_device *adev,
* or all rings (in the case @job is NULL)
* after above amdgpu_reset accomplished
*/
- if ((!job || job->base.sched == &ring->sched) && !r)
+ if ((!job || job->base.sched == &ring->sched) && !adev->asic_reset_res)
drm_sched_job_recovery(&ring->sched);
kthread_unpark(ring->sched.thread);
@@ -3398,21 +3426,144 @@ int amdgpu_device_gpu_recover(struct amdgpu_device *adev,
drm_helper_resume_force_mode(adev->ddev);
}
- ttm_bo_unlock_delayed_workqueue(&adev->mman.bdev, resched);
+ adev->asic_reset_res = 0;
+}
- if (r) {
- /* bad news, how to tell it to userspace ? */
- dev_info(adev->dev, "GPU reset(%d) failed\n", atomic_read(&adev->gpu_reset_counter));
- amdgpu_vf_error_put(adev, AMDGIM_ERROR_VF_GPU_RESET_FAIL, 0, r);
- } else {
- dev_info(adev->dev, "GPU reset(%d) succeeded!\n",atomic_read(&adev->gpu_reset_counter));
- }
+static void amdgpu_device_lock_adev(struct amdgpu_device *adev)
+{
+ mutex_lock(&adev->lock_reset);
+ atomic_inc(&adev->gpu_reset_counter);
+ adev->in_gpu_reset = 1;
+ /* Block kfd */
+ amdgpu_amdkfd_pre_reset(adev);
+}
+static void amdgpu_device_unlock_adev(struct amdgpu_device *adev)
+{
/*unlock kfd */
amdgpu_amdkfd_post_reset(adev);
amdgpu_vf_error_trans_all(adev);
adev->in_gpu_reset = 0;
mutex_unlock(&adev->lock_reset);
+}
+
+
+/**
+ * amdgpu_device_gpu_recover - reset the asic and recover scheduler
+ *
+ * @adev: amdgpu device pointer
+ * @job: which job trigger hang
+ *
+ * Attempt to reset the GPU if it has hung (all asics).
+ * Attempt to do soft-reset or full-reset and reinitialize Asic
+ * Returns 0 for success or an error on failure.
+ */
+
+int amdgpu_device_gpu_recover(struct amdgpu_device *adev,
+ struct amdgpu_job *job)
+{
+ int r;
+ struct amdgpu_hive_info *hive = NULL;
+ bool need_full_reset = false;
+ struct amdgpu_device *tmp_adev = NULL;
+ struct list_head device_list, *device_list_handle = NULL;
+
+ INIT_LIST_HEAD(&device_list);
+
+ dev_info(adev->dev, "GPU reset begin!\n");
+
+ /*
+ * In case of XGMI hive disallow concurrent resets to be triggered
+ * by different nodes. No point also since the one node already executing
+ * reset will also reset all the other nodes in the hive.
+ */
+ hive = amdgpu_get_xgmi_hive(adev);
+ if (hive && adev->gmc.xgmi.num_physical_nodes > 1 &&
+ !mutex_trylock(&hive->hive_lock))
+ return 0;
+
+ /* Start with adev pre asic reset first for soft reset check.*/
+ amdgpu_device_lock_adev(adev);
+ r = amdgpu_device_pre_asic_reset(adev,
+ job,
+ &need_full_reset);
+ if (r) {
+ /*TODO Should we stop ?*/
+ DRM_ERROR("GPU pre asic reset failed with err, %d for drm dev, %s ",
+ r, adev->ddev->unique);
+ adev->asic_reset_res = r;
+ }
+
+ /* Build list of devices to reset */
+ if (need_full_reset && adev->gmc.xgmi.num_physical_nodes > 1) {
+ if (!hive) {
+ amdgpu_device_unlock_adev(adev);
+ return -ENODEV;
+ }
+
+ /*
+ * In case we are in XGMI hive mode device reset is done for all the
+ * nodes in the hive to retrain all XGMI links and hence the reset
+ * sequence is executed in loop on all nodes.
+ */
+ device_list_handle = &hive->device_list;
+ } else {
+ list_add_tail(&adev->gmc.xgmi.head, &device_list);
+ device_list_handle = &device_list;
+ }
+
+retry: /* Rest of adevs pre asic reset from XGMI hive. */
+ list_for_each_entry(tmp_adev, device_list_handle, gmc.xgmi.head) {
+
+ if (tmp_adev == adev)
+ continue;
+
+ dev_info(tmp_adev->dev, "GPU reset begin for drm dev %s!\n", adev->ddev->unique);
+
+ amdgpu_device_lock_adev(tmp_adev);
+ r = amdgpu_device_pre_asic_reset(tmp_adev,
+ NULL,
+ &need_full_reset);
+ /*TODO Should we stop ?*/
+ if (r) {
+ DRM_ERROR("GPU pre asic reset failed with err, %d for drm dev, %s ",
+ r, tmp_adev->ddev->unique);
+ tmp_adev->asic_reset_res = r;
+ }
+ }
+
+ /* Actual ASIC resets if needed.*/
+ /* TODO Implement XGMI hive reset logic for SRIOV */
+ if (amdgpu_sriov_vf(adev)) {
+ r = amdgpu_device_reset_sriov(adev, job ? false : true);
+ if (r)
+ adev->asic_reset_res = r;
+ } else {
+ r = amdgpu_do_asic_reset(hive, device_list_handle, &need_full_reset);
+ if (r && r == -EAGAIN)
+ goto retry;
+ }
+
+ /* Post ASIC reset for all devs .*/
+ list_for_each_entry(tmp_adev, device_list_handle, gmc.xgmi.head) {
+ amdgpu_device_post_asic_reset(tmp_adev, tmp_adev == adev ? job : NULL);
+
+ if (r) {
+ /* bad news, how to tell it to userspace ? */
+ dev_info(tmp_adev->dev, "GPU reset(%d) failed\n", atomic_read(&adev->gpu_reset_counter));
+ amdgpu_vf_error_put(tmp_adev, AMDGIM_ERROR_VF_GPU_RESET_FAIL, 0, r);
+ } else {
+ dev_info(tmp_adev->dev, "GPU reset(%d) succeeded!\n", atomic_read(&adev->gpu_reset_counter));
+ }
+
+ amdgpu_device_unlock_adev(tmp_adev);
+ }
+
+ if (hive && adev->gmc.xgmi.num_physical_nodes > 1)
+ mutex_unlock(&hive->hive_lock);
+
+ if (r)
+ dev_info(adev->dev, "GPU reset end with ret = %d\n", r);
return r;
}
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_display.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_display.c
index 686a26de50f9..15ce7e681d67 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_display.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_display.c
@@ -631,6 +631,11 @@ int amdgpu_display_modeset_create_props(struct amdgpu_device *adev)
drm_property_create_range(adev->ddev, 0, "max bpc", 8, 16);
if (!adev->mode_info.max_bpc_property)
return -ENOMEM;
+ adev->mode_info.abm_level_property =
+ drm_property_create_range(adev->ddev, 0,
+ "abm level", 0, 4);
+ if (!adev->mode_info.abm_level_property)
+ return -ENOMEM;
}
return 0;
@@ -857,7 +862,12 @@ int amdgpu_display_get_crtc_scanoutpos(struct drm_device *dev,
/* Inside "upper part" of vblank area? Apply corrective offset if so: */
if (in_vbl && (*vpos >= vbl_start)) {
vtotal = mode->crtc_vtotal;
- *vpos = *vpos - vtotal;
+
+ /* With variable refresh rate displays the vpos can exceed
+ * the vtotal value. Clamp to 0 to return -vbl_end instead
+ * of guessing the remaining number of lines until scanout.
+ */
+ *vpos = (*vpos < vtotal) ? (*vpos - vtotal) : 0;
}
/* Correct for shifted end of vbl at vbl_end. */
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_doorbell.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_doorbell.h
new file mode 100644
index 000000000000..be620b29f4aa
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_doorbell.h
@@ -0,0 +1,243 @@
+/*
+ * Copyright 2018 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.
+ *
+ */
+
+/*
+ * GPU doorbell structures, functions & helpers
+ */
+struct amdgpu_doorbell {
+ /* doorbell mmio */
+ resource_size_t base;
+ resource_size_t size;
+ u32 __iomem *ptr;
+ u32 num_doorbells; /* Number of doorbells actually reserved for amdgpu. */
+};
+
+/* Reserved doorbells for amdgpu (including multimedia).
+ * KFD can use all the rest in the 2M doorbell bar.
+ * For asic before vega10, doorbell is 32-bit, so the
+ * index/offset is in dword. For vega10 and after, doorbell
+ * can be 64-bit, so the index defined is in qword.
+ */
+struct amdgpu_doorbell_index {
+ uint32_t kiq;
+ uint32_t mec_ring0;
+ uint32_t mec_ring1;
+ uint32_t mec_ring2;
+ uint32_t mec_ring3;
+ uint32_t mec_ring4;
+ uint32_t mec_ring5;
+ uint32_t mec_ring6;
+ uint32_t mec_ring7;
+ uint32_t userqueue_start;
+ uint32_t userqueue_end;
+ uint32_t gfx_ring0;
+ uint32_t sdma_engine0;
+ uint32_t sdma_engine1;
+ uint32_t sdma_engine2;
+ uint32_t sdma_engine3;
+ uint32_t sdma_engine4;
+ uint32_t sdma_engine5;
+ uint32_t sdma_engine6;
+ uint32_t sdma_engine7;
+ uint32_t ih;
+ union {
+ struct {
+ uint32_t vcn_ring0_1;
+ uint32_t vcn_ring2_3;
+ uint32_t vcn_ring4_5;
+ uint32_t vcn_ring6_7;
+ } vcn;
+ struct {
+ uint32_t uvd_ring0_1;
+ uint32_t uvd_ring2_3;
+ uint32_t uvd_ring4_5;
+ uint32_t uvd_ring6_7;
+ uint32_t vce_ring0_1;
+ uint32_t vce_ring2_3;
+ uint32_t vce_ring4_5;
+ uint32_t vce_ring6_7;
+ } uvd_vce;
+ };
+ uint32_t max_assignment;
+};
+
+typedef enum _AMDGPU_DOORBELL_ASSIGNMENT
+{
+ AMDGPU_DOORBELL_KIQ = 0x000,
+ AMDGPU_DOORBELL_HIQ = 0x001,
+ AMDGPU_DOORBELL_DIQ = 0x002,
+ AMDGPU_DOORBELL_MEC_RING0 = 0x010,
+ AMDGPU_DOORBELL_MEC_RING1 = 0x011,
+ AMDGPU_DOORBELL_MEC_RING2 = 0x012,
+ AMDGPU_DOORBELL_MEC_RING3 = 0x013,
+ AMDGPU_DOORBELL_MEC_RING4 = 0x014,
+ AMDGPU_DOORBELL_MEC_RING5 = 0x015,
+ AMDGPU_DOORBELL_MEC_RING6 = 0x016,
+ AMDGPU_DOORBELL_MEC_RING7 = 0x017,
+ AMDGPU_DOORBELL_GFX_RING0 = 0x020,
+ AMDGPU_DOORBELL_sDMA_ENGINE0 = 0x1E0,
+ AMDGPU_DOORBELL_sDMA_ENGINE1 = 0x1E1,
+ AMDGPU_DOORBELL_IH = 0x1E8,
+ AMDGPU_DOORBELL_MAX_ASSIGNMENT = 0x3FF,
+ AMDGPU_DOORBELL_INVALID = 0xFFFF
+} AMDGPU_DOORBELL_ASSIGNMENT;
+
+typedef enum _AMDGPU_VEGA20_DOORBELL_ASSIGNMENT
+{
+ /* Compute + GFX: 0~255 */
+ AMDGPU_VEGA20_DOORBELL_KIQ = 0x000,
+ AMDGPU_VEGA20_DOORBELL_HIQ = 0x001,
+ AMDGPU_VEGA20_DOORBELL_DIQ = 0x002,
+ AMDGPU_VEGA20_DOORBELL_MEC_RING0 = 0x003,
+ AMDGPU_VEGA20_DOORBELL_MEC_RING1 = 0x004,
+ AMDGPU_VEGA20_DOORBELL_MEC_RING2 = 0x005,
+ AMDGPU_VEGA20_DOORBELL_MEC_RING3 = 0x006,
+ AMDGPU_VEGA20_DOORBELL_MEC_RING4 = 0x007,
+ AMDGPU_VEGA20_DOORBELL_MEC_RING5 = 0x008,
+ AMDGPU_VEGA20_DOORBELL_MEC_RING6 = 0x009,
+ AMDGPU_VEGA20_DOORBELL_MEC_RING7 = 0x00A,
+ AMDGPU_VEGA20_DOORBELL_USERQUEUE_START = 0x00B,
+ AMDGPU_VEGA20_DOORBELL_USERQUEUE_END = 0x08A,
+ AMDGPU_VEGA20_DOORBELL_GFX_RING0 = 0x08B,
+ /* SDMA:256~335*/
+ AMDGPU_VEGA20_DOORBELL_sDMA_ENGINE0 = 0x100,
+ AMDGPU_VEGA20_DOORBELL_sDMA_ENGINE1 = 0x10A,
+ AMDGPU_VEGA20_DOORBELL_sDMA_ENGINE2 = 0x114,
+ AMDGPU_VEGA20_DOORBELL_sDMA_ENGINE3 = 0x11E,
+ AMDGPU_VEGA20_DOORBELL_sDMA_ENGINE4 = 0x128,
+ AMDGPU_VEGA20_DOORBELL_sDMA_ENGINE5 = 0x132,
+ AMDGPU_VEGA20_DOORBELL_sDMA_ENGINE6 = 0x13C,
+ AMDGPU_VEGA20_DOORBELL_sDMA_ENGINE7 = 0x146,
+ /* IH: 376~391 */
+ AMDGPU_VEGA20_DOORBELL_IH = 0x178,
+ /* MMSCH: 392~407
+ * overlap the doorbell assignment with VCN as they are mutually exclusive
+ * VCE engine's doorbell is 32 bit and two VCE ring share one QWORD
+ */
+ AMDGPU_VEGA20_DOORBELL64_VCN0_1 = 0x188, /* lower 32 bits for VNC0 and upper 32 bits for VNC1 */
+ AMDGPU_VEGA20_DOORBELL64_VCN2_3 = 0x189,
+ AMDGPU_VEGA20_DOORBELL64_VCN4_5 = 0x18A,
+ AMDGPU_VEGA20_DOORBELL64_VCN6_7 = 0x18B,
+
+ AMDGPU_VEGA20_DOORBELL64_UVD_RING0_1 = 0x188,
+ AMDGPU_VEGA20_DOORBELL64_UVD_RING2_3 = 0x189,
+ AMDGPU_VEGA20_DOORBELL64_UVD_RING4_5 = 0x18A,
+ AMDGPU_VEGA20_DOORBELL64_UVD_RING6_7 = 0x18B,
+
+ AMDGPU_VEGA20_DOORBELL64_VCE_RING0_1 = 0x18C,
+ AMDGPU_VEGA20_DOORBELL64_VCE_RING2_3 = 0x18D,
+ AMDGPU_VEGA20_DOORBELL64_VCE_RING4_5 = 0x18E,
+ AMDGPU_VEGA20_DOORBELL64_VCE_RING6_7 = 0x18F,
+ AMDGPU_VEGA20_DOORBELL_MAX_ASSIGNMENT = 0x18F,
+ AMDGPU_VEGA20_DOORBELL_INVALID = 0xFFFF
+} AMDGPU_VEGA20_DOORBELL_ASSIGNMENT;
+
+/*
+ * 64bit doorbell, offset are in QWORD, occupy 2KB doorbell space
+ */
+typedef enum _AMDGPU_DOORBELL64_ASSIGNMENT
+{
+ /*
+ * All compute related doorbells: kiq, hiq, diq, traditional compute queue, user queue, should locate in
+ * a continues range so that programming CP_MEC_DOORBELL_RANGE_LOWER/UPPER can cover this range.
+ * Compute related doorbells are allocated from 0x00 to 0x8a
+ */
+
+
+ /* kernel scheduling */
+ AMDGPU_DOORBELL64_KIQ = 0x00,
+
+ /* HSA interface queue and debug queue */
+ AMDGPU_DOORBELL64_HIQ = 0x01,
+ AMDGPU_DOORBELL64_DIQ = 0x02,
+
+ /* Compute engines */
+ AMDGPU_DOORBELL64_MEC_RING0 = 0x03,
+ AMDGPU_DOORBELL64_MEC_RING1 = 0x04,
+ AMDGPU_DOORBELL64_MEC_RING2 = 0x05,
+ AMDGPU_DOORBELL64_MEC_RING3 = 0x06,
+ AMDGPU_DOORBELL64_MEC_RING4 = 0x07,
+ AMDGPU_DOORBELL64_MEC_RING5 = 0x08,
+ AMDGPU_DOORBELL64_MEC_RING6 = 0x09,
+ AMDGPU_DOORBELL64_MEC_RING7 = 0x0a,
+
+ /* User queue doorbell range (128 doorbells) */
+ AMDGPU_DOORBELL64_USERQUEUE_START = 0x0b,
+ AMDGPU_DOORBELL64_USERQUEUE_END = 0x8a,
+
+ /* Graphics engine */
+ AMDGPU_DOORBELL64_GFX_RING0 = 0x8b,
+
+ /*
+ * Other graphics doorbells can be allocated here: from 0x8c to 0xdf
+ * Graphics voltage island aperture 1
+ * default non-graphics QWORD index is 0xe0 - 0xFF inclusive
+ */
+
+ /* For vega10 sriov, the sdma doorbell must be fixed as follow
+ * to keep the same setting with host driver, or it will
+ * happen conflicts
+ */
+ AMDGPU_DOORBELL64_sDMA_ENGINE0 = 0xF0,
+ AMDGPU_DOORBELL64_sDMA_HI_PRI_ENGINE0 = 0xF1,
+ AMDGPU_DOORBELL64_sDMA_ENGINE1 = 0xF2,
+ AMDGPU_DOORBELL64_sDMA_HI_PRI_ENGINE1 = 0xF3,
+
+ /* Interrupt handler */
+ AMDGPU_DOORBELL64_IH = 0xF4, /* For legacy interrupt ring buffer */
+ AMDGPU_DOORBELL64_IH_RING1 = 0xF5, /* For page migration request log */
+ AMDGPU_DOORBELL64_IH_RING2 = 0xF6, /* For page migration translation/invalidation log */
+
+ /* VCN engine use 32 bits doorbell */
+ AMDGPU_DOORBELL64_VCN0_1 = 0xF8, /* lower 32 bits for VNC0 and upper 32 bits for VNC1 */
+ AMDGPU_DOORBELL64_VCN2_3 = 0xF9,
+ AMDGPU_DOORBELL64_VCN4_5 = 0xFA,
+ AMDGPU_DOORBELL64_VCN6_7 = 0xFB,
+
+ /* overlap the doorbell assignment with VCN as they are mutually exclusive
+ * VCE engine's doorbell is 32 bit and two VCE ring share one QWORD
+ */
+ AMDGPU_DOORBELL64_UVD_RING0_1 = 0xF8,
+ AMDGPU_DOORBELL64_UVD_RING2_3 = 0xF9,
+ AMDGPU_DOORBELL64_UVD_RING4_5 = 0xFA,
+ AMDGPU_DOORBELL64_UVD_RING6_7 = 0xFB,
+
+ AMDGPU_DOORBELL64_VCE_RING0_1 = 0xFC,
+ AMDGPU_DOORBELL64_VCE_RING2_3 = 0xFD,
+ AMDGPU_DOORBELL64_VCE_RING4_5 = 0xFE,
+ AMDGPU_DOORBELL64_VCE_RING6_7 = 0xFF,
+
+ AMDGPU_DOORBELL64_MAX_ASSIGNMENT = 0xFF,
+ AMDGPU_DOORBELL64_INVALID = 0xFFFF
+} AMDGPU_DOORBELL64_ASSIGNMENT;
+
+u32 amdgpu_mm_rdoorbell(struct amdgpu_device *adev, u32 index);
+void amdgpu_mm_wdoorbell(struct amdgpu_device *adev, u32 index, u32 v);
+u64 amdgpu_mm_rdoorbell64(struct amdgpu_device *adev, u32 index);
+void amdgpu_mm_wdoorbell64(struct amdgpu_device *adev, u32 index, u64 v);
+
+#define RDOORBELL32(index) amdgpu_mm_rdoorbell(adev, (index))
+#define WDOORBELL32(index, v) amdgpu_mm_wdoorbell(adev, (index), (v))
+#define RDOORBELL64(index) amdgpu_mm_rdoorbell64(adev, (index))
+#define WDOORBELL64(index, v) amdgpu_mm_wdoorbell64(adev, (index), (v))
+
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c
index 8de55f7f1a3a..90f474f98b6e 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c
@@ -454,9 +454,10 @@ module_param_named(cntl_sb_buf_per_se, amdgpu_cntl_sb_buf_per_se, int, 0444);
/**
* DOC: param_buf_per_se (int)
- * Override the size of Off-Chip Pramater Cache per Shader Engine in Byte. The default is 0 (depending on gfx).
+ * Override the size of Off-Chip Parameter Cache per Shader Engine in Byte.
+ * The default is 0 (depending on gfx).
*/
-MODULE_PARM_DESC(param_buf_per_se, "the size of Off-Chip Pramater Cache per Shader Engine (default depending on gfx)");
+MODULE_PARM_DESC(param_buf_per_se, "the size of Off-Chip Parameter Cache per Shader Engine (default depending on gfx)");
module_param_named(param_buf_per_se, amdgpu_param_buf_per_se, int, 0444);
/**
@@ -1220,9 +1221,6 @@ static struct drm_driver kms_driver = {
.patchlevel = KMS_DRIVER_PATCHLEVEL,
};
-static struct drm_driver *driver;
-static struct pci_driver *pdriver;
-
static struct pci_driver amdgpu_kms_pci_driver = {
.name = DRIVER_NAME,
.id_table = pciidlist,
@@ -1252,16 +1250,14 @@ static int __init amdgpu_init(void)
goto error_fence;
DRM_INFO("amdgpu kernel modesetting enabled.\n");
- driver = &kms_driver;
- pdriver = &amdgpu_kms_pci_driver;
- driver->num_ioctls = amdgpu_max_kms_ioctl;
+ kms_driver.num_ioctls = amdgpu_max_kms_ioctl;
amdgpu_register_atpx_handler();
/* Ignore KFD init failures. Normal when CONFIG_HSA_AMD is not set. */
amdgpu_amdkfd_init();
/* let modprobe override vga console setting */
- return pci_register_driver(pdriver);
+ return pci_register_driver(&amdgpu_kms_pci_driver);
error_fence:
amdgpu_sync_fini();
@@ -1273,7 +1269,7 @@ error_sync:
static void __exit amdgpu_exit(void)
{
amdgpu_amdkfd_fini();
- pci_unregister_driver(pdriver);
+ pci_unregister_driver(&amdgpu_kms_pci_driver);
amdgpu_unregister_atpx_handler();
amdgpu_sync_fini();
amdgpu_fence_slab_fini();
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.c
index 6a70c0b7105f..97a60da62004 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.c
@@ -250,7 +250,7 @@ int amdgpu_gfx_kiq_init_ring(struct amdgpu_device *adev,
ring->adev = NULL;
ring->ring_obj = NULL;
ring->use_doorbell = true;
- ring->doorbell_index = AMDGPU_DOORBELL_KIQ;
+ ring->doorbell_index = adev->doorbell_index.kiq;
r = amdgpu_gfx_kiq_acquire(adev, ring);
if (r)
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c
index 9b3164c0f861..bc62bf41b7e9 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c
@@ -467,9 +467,6 @@ static int amdgpu_info_ioctl(struct drm_device *dev, void *data, struct drm_file
if (!info->return_size || !info->return_pointer)
return -EINVAL;
- /* Ensure IB tests are run on ring */
- flush_delayed_work(&adev->late_init_work);
-
switch (info->query) {
case AMDGPU_INFO_ACCEL_WORKING:
ui32 = adev->accel_working;
@@ -950,6 +947,9 @@ int amdgpu_driver_open_kms(struct drm_device *dev, struct drm_file *file_priv)
struct amdgpu_fpriv *fpriv;
int r, pasid;
+ /* Ensure IB tests are run on ring */
+ flush_delayed_work(&adev->late_init_work);
+
file_priv->driver_priv = NULL;
r = pm_runtime_get_sync(dev->dev);
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_mode.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_mode.h
index 0dc2c5c57015..aadd0fa42e43 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_mode.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_mode.h
@@ -38,7 +38,6 @@
#include <drm/drm_crtc_helper.h>
#include <drm/drm_fb_helper.h>
#include <drm/drm_plane_helper.h>
-#include <drm/drm_fb_helper.h>
#include <linux/i2c.h>
#include <linux/i2c-algo-bit.h>
#include <linux/hrtimer.h>
@@ -294,13 +293,6 @@ struct amdgpu_display_funcs {
uint16_t connector_object_id,
struct amdgpu_hpd *hpd,
struct amdgpu_router *router);
- /* it is used to enter or exit into free sync mode */
- int (*notify_freesync)(struct drm_device *dev, void *data,
- struct drm_file *filp);
- /* it is used to allow enablement of freesync mode */
- int (*set_freesync_property)(struct drm_connector *connector,
- struct drm_property *property,
- uint64_t val);
};
@@ -340,6 +332,8 @@ struct amdgpu_mode_info {
struct drm_property *dither_property;
/* maximum number of bits per channel for monitor color */
struct drm_property *max_bpc_property;
+ /* Adaptive Backlight Modulation (power feature) */
+ struct drm_property *abm_level_property;
/* hardcoded DFP edid from BIOS */
struct edid *bios_hardcoded_edid;
int bios_hardcoded_edid_size;
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_pm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_pm.c
index 7235cd0b0fa9..1f61ed95727c 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_pm.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_pm.c
@@ -33,6 +33,8 @@
#include <linux/hwmon.h>
#include <linux/hwmon-sysfs.h>
#include <linux/nospec.h>
+#include "hwmgr.h"
+#define WIDTH_4K 3840
static int amdgpu_debugfs_pm_init(struct amdgpu_device *adev);
@@ -1642,6 +1644,19 @@ static umode_t hwmon_attributes_visible(struct kobject *kobj,
attr == &sensor_dev_attr_fan1_enable.dev_attr.attr))
return 0;
+ /* Skip fan attributes on APU */
+ if ((adev->flags & AMD_IS_APU) &&
+ (attr == &sensor_dev_attr_pwm1.dev_attr.attr ||
+ attr == &sensor_dev_attr_pwm1_enable.dev_attr.attr ||
+ attr == &sensor_dev_attr_pwm1_max.dev_attr.attr ||
+ attr == &sensor_dev_attr_pwm1_min.dev_attr.attr ||
+ attr == &sensor_dev_attr_fan1_input.dev_attr.attr ||
+ attr == &sensor_dev_attr_fan1_min.dev_attr.attr ||
+ attr == &sensor_dev_attr_fan1_max.dev_attr.attr ||
+ attr == &sensor_dev_attr_fan1_target.dev_attr.attr ||
+ attr == &sensor_dev_attr_fan1_enable.dev_attr.attr))
+ return 0;
+
/* Skip limit attributes if DPM is not enabled */
if (!adev->pm.dpm_enabled &&
(attr == &sensor_dev_attr_temp1_crit.dev_attr.attr ||
@@ -1956,6 +1971,17 @@ void amdgpu_dpm_enable_uvd(struct amdgpu_device *adev, bool enable)
amdgpu_dpm_set_powergating_by_smu(adev, AMD_IP_BLOCK_TYPE_UVD, !enable);
mutex_unlock(&adev->pm.mutex);
}
+ /* enable/disable Low Memory PState for UVD (4k videos) */
+ if (adev->asic_type == CHIP_STONEY &&
+ adev->uvd.decode_image_width >= WIDTH_4K) {
+ struct pp_hwmgr *hwmgr = adev->powerplay.pp_handle;
+
+ if (hwmgr && hwmgr->hwmgr_func &&
+ hwmgr->hwmgr_func->update_nbdpm_pstate)
+ hwmgr->hwmgr_func->update_nbdpm_pstate(hwmgr,
+ !enable,
+ true);
+ }
}
void amdgpu_dpm_enable_vce(struct amdgpu_device *adev, bool enable)
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_uvd.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_uvd.c
index 69896f451e8a..4e5d13e41f6a 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_uvd.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_uvd.c
@@ -692,6 +692,8 @@ static int amdgpu_uvd_cs_msg_decode(struct amdgpu_device *adev, uint32_t *msg,
buf_sizes[0x1] = dpb_size;
buf_sizes[0x2] = image_size;
buf_sizes[0x4] = min_ctx_size;
+ /* store image width to adjust nb memory pstate */
+ adev->uvd.decode_image_width = width;
return 0;
}
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_uvd.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_uvd.h
index a3ab1a41060f..5eb63288d157 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_uvd.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_uvd.h
@@ -65,6 +65,8 @@ struct amdgpu_uvd {
struct drm_sched_entity entity;
struct delayed_work idle_work;
unsigned harvest_config;
+ /* store image width to adjust nb memory state */
+ unsigned decode_image_width;
};
int amdgpu_uvd_sw_init(struct amdgpu_device *adev);
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_virt.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_virt.c
index cfee74732edb..462a04e0f5e6 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_virt.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_virt.c
@@ -334,7 +334,7 @@ void amdgpu_virt_init_data_exchange(struct amdgpu_device *adev)
if (adev->fw_vram_usage.va != NULL) {
adev->virt.fw_reserve.p_pf2vf =
- (struct amdgim_pf2vf_info_header *)(
+ (struct amd_sriov_msg_pf2vf_info_header *)(
adev->fw_vram_usage.va + AMDGIM_DATAEXCHANGE_OFFSET);
AMDGPU_FW_VRAM_PF2VF_READ(adev, header.size, &pf2vf_size);
AMDGPU_FW_VRAM_PF2VF_READ(adev, checksum, &checksum);
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_virt.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_virt.h
index 0728fbc9a692..722deefc0a7e 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_virt.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_virt.h
@@ -63,8 +63,8 @@ struct amdgpu_virt_ops {
* Firmware Reserve Frame buffer
*/
struct amdgpu_virt_fw_reserve {
- struct amdgim_pf2vf_info_header *p_pf2vf;
- struct amdgim_vf2pf_info_header *p_vf2pf;
+ struct amd_sriov_msg_pf2vf_info_header *p_pf2vf;
+ struct amd_sriov_msg_vf2pf_info_header *p_vf2pf;
unsigned int checksum_key;
};
/*
@@ -85,15 +85,17 @@ enum AMDGIM_FEATURE_FLAG {
AMDGIM_FEATURE_GIM_FLR_VRAMLOST = 0x4,
};
-struct amdgim_pf2vf_info_header {
+struct amd_sriov_msg_pf2vf_info_header {
/* the total structure size in byte. */
uint32_t size;
/* version of this structure, written by the GIM */
uint32_t version;
+ /* reserved */
+ uint32_t reserved[2];
} __aligned(4);
struct amdgim_pf2vf_info_v1 {
/* header contains size and version */
- struct amdgim_pf2vf_info_header header;
+ struct amd_sriov_msg_pf2vf_info_header header;
/* max_width * max_height */
unsigned int uvd_enc_max_pixels_count;
/* 16x16 pixels/sec, codec independent */
@@ -112,7 +114,7 @@ struct amdgim_pf2vf_info_v1 {
struct amdgim_pf2vf_info_v2 {
/* header contains size and version */
- struct amdgim_pf2vf_info_header header;
+ struct amd_sriov_msg_pf2vf_info_header header;
/* use private key from mailbox 2 to create chueksum */
uint32_t checksum;
/* The features flags of the GIM driver supports. */
@@ -137,20 +139,22 @@ struct amdgim_pf2vf_info_v2 {
uint64_t vcefw_kboffset;
/* VCE FW size in KB */
uint32_t vcefw_ksize;
- uint32_t reserved[AMDGIM_GET_STRUCTURE_RESERVED_SIZE(256, 0, 0, (9 + sizeof(struct amdgim_pf2vf_info_header)/sizeof(uint32_t)), 3)];
+ uint32_t reserved[AMDGIM_GET_STRUCTURE_RESERVED_SIZE(256, 0, 0, (9 + sizeof(struct amd_sriov_msg_pf2vf_info_header)/sizeof(uint32_t)), 3)];
} __aligned(4);
-struct amdgim_vf2pf_info_header {
+struct amd_sriov_msg_vf2pf_info_header {
/* the total structure size in byte. */
uint32_t size;
/*version of this structure, written by the guest */
uint32_t version;
+ /* reserved */
+ uint32_t reserved[2];
} __aligned(4);
struct amdgim_vf2pf_info_v1 {
/* header contains size and version */
- struct amdgim_vf2pf_info_header header;
+ struct amd_sriov_msg_vf2pf_info_header header;
/* driver version */
char driver_version[64];
/* driver certification, 1=WHQL, 0=None */
@@ -180,7 +184,7 @@ struct amdgim_vf2pf_info_v1 {
struct amdgim_vf2pf_info_v2 {
/* header contains size and version */
- struct amdgim_vf2pf_info_header header;
+ struct amd_sriov_msg_vf2pf_info_header header;
uint32_t checksum;
/* driver version */
uint8_t driver_version[64];
@@ -206,7 +210,7 @@ struct amdgim_vf2pf_info_v2 {
uint32_t uvd_enc_usage;
/* guest uvd engine usage percentage. 0xffff means N/A. */
uint32_t uvd_enc_health;
- uint32_t reserved[AMDGIM_GET_STRUCTURE_RESERVED_SIZE(256, 64, 0, (12 + sizeof(struct amdgim_vf2pf_info_header)/sizeof(uint32_t)), 0)];
+ uint32_t reserved[AMDGIM_GET_STRUCTURE_RESERVED_SIZE(256, 64, 0, (12 + sizeof(struct amd_sriov_msg_vf2pf_info_header)/sizeof(uint32_t)), 0)];
} __aligned(4);
#define AMDGPU_FW_VRAM_VF2PF_VER 2
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_xgmi.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_xgmi.c
index 909216a9b447..fb37e69f1bba 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_xgmi.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_xgmi.c
@@ -23,7 +23,7 @@
*/
#include <linux/list.h>
#include "amdgpu.h"
-#include "amdgpu_psp.h"
+#include "amdgpu_xgmi.h"
static DEFINE_MUTEX(xgmi_mutex);
@@ -31,15 +31,16 @@ static DEFINE_MUTEX(xgmi_mutex);
#define AMDGPU_MAX_XGMI_HIVE 8
#define AMDGPU_MAX_XGMI_DEVICE_PER_HIVE 4
-struct amdgpu_hive_info {
- uint64_t hive_id;
- struct list_head device_list;
-};
-
static struct amdgpu_hive_info xgmi_hives[AMDGPU_MAX_XGMI_HIVE];
static unsigned hive_count = 0;
-static struct amdgpu_hive_info *amdgpu_get_xgmi_hive(struct amdgpu_device *adev)
+
+void *amdgpu_xgmi_hive_try_lock(struct amdgpu_hive_info *hive)
+{
+ return &hive->device_list;
+}
+
+struct amdgpu_hive_info *amdgpu_get_xgmi_hive(struct amdgpu_device *adev)
{
int i;
struct amdgpu_hive_info *tmp;
@@ -58,15 +59,38 @@ static struct amdgpu_hive_info *amdgpu_get_xgmi_hive(struct amdgpu_device *adev)
tmp = &xgmi_hives[hive_count++];
tmp->hive_id = adev->gmc.xgmi.hive_id;
INIT_LIST_HEAD(&tmp->device_list);
+ mutex_init(&tmp->hive_lock);
+
return tmp;
}
+int amdgpu_xgmi_update_topology(struct amdgpu_hive_info *hive, struct amdgpu_device *adev)
+{
+ int ret = -EINVAL;
+
+ /* Each psp need to set the latest topology */
+ ret = psp_xgmi_set_topology_info(&adev->psp,
+ hive->number_devices,
+ &hive->topology_info);
+ if (ret)
+ dev_err(adev->dev,
+ "XGMI: Set topology failure on device %llx, hive %llx, ret %d",
+ adev->gmc.xgmi.node_id,
+ adev->gmc.xgmi.hive_id, ret);
+ else
+ dev_info(adev->dev, "XGMI: Add node %d to hive 0x%llx.\n",
+ adev->gmc.xgmi.physical_node_id,
+ adev->gmc.xgmi.hive_id);
+
+ return ret;
+}
+
int amdgpu_xgmi_add_device(struct amdgpu_device *adev)
{
- struct psp_xgmi_topology_info *tmp_topology;
+ struct psp_xgmi_topology_info *hive_topology;
struct amdgpu_hive_info *hive;
struct amdgpu_xgmi *entry;
- struct amdgpu_device *tmp_adev;
+ struct amdgpu_device *tmp_adev = NULL;
int count = 0, ret = -EINVAL;
@@ -76,21 +100,21 @@ int amdgpu_xgmi_add_device(struct amdgpu_device *adev)
adev->gmc.xgmi.node_id = psp_xgmi_get_node_id(&adev->psp);
adev->gmc.xgmi.hive_id = psp_xgmi_get_hive_id(&adev->psp);
- tmp_topology = kzalloc(sizeof(struct psp_xgmi_topology_info), GFP_KERNEL);
- if (!tmp_topology)
- return -ENOMEM;
mutex_lock(&xgmi_mutex);
hive = amdgpu_get_xgmi_hive(adev);
if (!hive)
goto exit;
+ hive_topology = &hive->topology_info;
+
list_add_tail(&adev->gmc.xgmi.head, &hive->device_list);
list_for_each_entry(entry, &hive->device_list, head)
- tmp_topology->nodes[count++].node_id = entry->node_id;
+ hive_topology->nodes[count++].node_id = entry->node_id;
+ hive->number_devices = count;
/* Each psp need to get the latest topology */
list_for_each_entry(tmp_adev, &hive->device_list, gmc.xgmi.head) {
- ret = psp_xgmi_get_topology_info(&tmp_adev->psp, count, tmp_topology);
+ ret = psp_xgmi_get_topology_info(&tmp_adev->psp, count, hive_topology);
if (ret) {
dev_err(tmp_adev->dev,
"XGMI: Get topology failure on device %llx, hive %llx, ret %d",
@@ -101,25 +125,13 @@ int amdgpu_xgmi_add_device(struct amdgpu_device *adev)
}
}
- /* Each psp need to set the latest topology */
list_for_each_entry(tmp_adev, &hive->device_list, gmc.xgmi.head) {
- ret = psp_xgmi_set_topology_info(&tmp_adev->psp, count, tmp_topology);
- if (ret) {
- dev_err(tmp_adev->dev,
- "XGMI: Set topology failure on device %llx, hive %llx, ret %d",
- tmp_adev->gmc.xgmi.node_id,
- tmp_adev->gmc.xgmi.hive_id, ret);
- /* To do : continue with some node failed or disable the whole hive */
+ ret = amdgpu_xgmi_update_topology(hive, tmp_adev);
+ if (ret)
break;
- }
}
- if (!ret)
- dev_info(adev->dev, "XGMI: Add node %d to hive 0x%llx.\n",
- adev->gmc.xgmi.physical_node_id,
- adev->gmc.xgmi.hive_id);
exit:
mutex_unlock(&xgmi_mutex);
- kfree(tmp_topology);
return ret;
}
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_xgmi.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_xgmi.h
new file mode 100644
index 000000000000..6335bfdcc51d
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_xgmi.h
@@ -0,0 +1,39 @@
+/*
+ * 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 __AMDGPU_XGMI_H__
+#define __AMDGPU_XGMI_H__
+
+#include "amdgpu_psp.h"
+
+struct amdgpu_hive_info {
+ uint64_t hive_id;
+ struct list_head device_list;
+ struct psp_xgmi_topology_info topology_info;
+ int number_devices;
+ struct mutex hive_lock;
+};
+
+struct amdgpu_hive_info *amdgpu_get_xgmi_hive(struct amdgpu_device *adev);
+int amdgpu_xgmi_update_topology(struct amdgpu_hive_info *hive, struct amdgpu_device *adev);
+int amdgpu_xgmi_add_device(struct amdgpu_device *adev);
+
+#endif
diff --git a/drivers/gpu/drm/amd/amdgpu/cik.c b/drivers/gpu/drm/amd/amdgpu/cik.c
index f41f5f57e9f3..71c50d8900e3 100644
--- a/drivers/gpu/drm/amd/amdgpu/cik.c
+++ b/drivers/gpu/drm/amd/amdgpu/cik.c
@@ -1755,6 +1755,7 @@ static const struct amdgpu_asic_funcs cik_asic_funcs =
.flush_hdp = &cik_flush_hdp,
.invalidate_hdp = &cik_invalidate_hdp,
.need_full_reset = &cik_need_full_reset,
+ .init_doorbell_index = &legacy_doorbell_index_init,
};
static int cik_common_early_init(void *handle)
diff --git a/drivers/gpu/drm/amd/amdgpu/cik.h b/drivers/gpu/drm/amd/amdgpu/cik.h
index e49c6f15a0a0..54c625a2e570 100644
--- a/drivers/gpu/drm/amd/amdgpu/cik.h
+++ b/drivers/gpu/drm/amd/amdgpu/cik.h
@@ -30,4 +30,5 @@ void cik_srbm_select(struct amdgpu_device *adev,
u32 me, u32 pipe, u32 queue, u32 vmid);
int cik_set_ip_blocks(struct amdgpu_device *adev);
+void legacy_doorbell_index_init(struct amdgpu_device *adev);
#endif
diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v7_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v7_0.c
index f467b9bd090d..3a9fb6018c16 100644
--- a/drivers/gpu/drm/amd/amdgpu/gfx_v7_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/gfx_v7_0.c
@@ -4363,7 +4363,7 @@ static int gfx_v7_0_compute_ring_init(struct amdgpu_device *adev, int ring_id,
ring->ring_obj = NULL;
ring->use_doorbell = true;
- ring->doorbell_index = AMDGPU_DOORBELL_MEC_RING0 + ring_id;
+ ring->doorbell_index = adev->doorbell_index.mec_ring0 + ring_id;
sprintf(ring->name, "comp_%d.%d.%d", ring->me, ring->pipe, ring->queue);
irq_type = AMDGPU_CP_IRQ_COMPUTE_MEC1_PIPE0_EOP
diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c
index cb066a8dccd7..1454fc306783 100644
--- a/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c
@@ -44,7 +44,6 @@
#include "gca/gfx_8_0_d.h"
#include "gca/gfx_8_0_enum.h"
#include "gca/gfx_8_0_sh_mask.h"
-#include "gca/gfx_8_0_enum.h"
#include "dce/dce_10_0_d.h"
#include "dce/dce_10_0_sh_mask.h"
@@ -1891,7 +1890,7 @@ static int gfx_v8_0_compute_ring_init(struct amdgpu_device *adev, int ring_id,
ring->ring_obj = NULL;
ring->use_doorbell = true;
- ring->doorbell_index = AMDGPU_DOORBELL_MEC_RING0 + ring_id;
+ ring->doorbell_index = adev->doorbell_index.mec_ring0 + ring_id;
ring->eop_gpu_addr = adev->gfx.mec.hpd_eop_gpu_addr
+ (ring_id * GFX8_MEC_HPD_SIZE);
sprintf(ring->name, "comp_%d.%d.%d", ring->me, ring->pipe, ring->queue);
@@ -2002,7 +2001,7 @@ static int gfx_v8_0_sw_init(void *handle)
/* no gfx doorbells on iceland */
if (adev->asic_type != CHIP_TOPAZ) {
ring->use_doorbell = true;
- ring->doorbell_index = AMDGPU_DOORBELL_GFX_RING0;
+ ring->doorbell_index = adev->doorbell_index.gfx_ring0;
}
r = amdgpu_ring_init(adev, ring, 1024, &adev->gfx.eop_irq,
@@ -4216,7 +4215,7 @@ static void gfx_v8_0_set_cpg_door_bell(struct amdgpu_device *adev, struct amdgpu
tmp = REG_SET_FIELD(0, CP_RB_DOORBELL_RANGE_LOWER,
DOORBELL_RANGE_LOWER,
- AMDGPU_DOORBELL_GFX_RING0);
+ adev->doorbell_index.gfx_ring0);
WREG32(mmCP_RB_DOORBELL_RANGE_LOWER, tmp);
WREG32(mmCP_RB_DOORBELL_RANGE_UPPER,
@@ -4645,8 +4644,8 @@ static int gfx_v8_0_kcq_init_queue(struct amdgpu_ring *ring)
static void gfx_v8_0_set_mec_doorbell_range(struct amdgpu_device *adev)
{
if (adev->asic_type > CHIP_TONGA) {
- WREG32(mmCP_MEC_DOORBELL_RANGE_LOWER, AMDGPU_DOORBELL_KIQ << 2);
- WREG32(mmCP_MEC_DOORBELL_RANGE_UPPER, AMDGPU_DOORBELL_MEC_RING7 << 2);
+ WREG32(mmCP_MEC_DOORBELL_RANGE_LOWER, adev->doorbell_index.kiq << 2);
+ WREG32(mmCP_MEC_DOORBELL_RANGE_UPPER, adev->doorbell_index.mec_ring7 << 2);
}
/* enable doorbells */
WREG32_FIELD(CP_PQ_STATUS, DOORBELL_ENABLE, 1);
diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c
index c27caa144c57..af8ccb014be3 100644
--- a/drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c
@@ -1566,7 +1566,7 @@ static int gfx_v9_0_compute_ring_init(struct amdgpu_device *adev, int ring_id,
ring->ring_obj = NULL;
ring->use_doorbell = true;
- ring->doorbell_index = (AMDGPU_DOORBELL_MEC_RING0 + ring_id) << 1;
+ ring->doorbell_index = (adev->doorbell_index.mec_ring0 + ring_id) << 1;
ring->eop_gpu_addr = adev->gfx.mec.hpd_eop_gpu_addr
+ (ring_id * GFX9_MEC_HPD_SIZE);
sprintf(ring->name, "comp_%d.%d.%d", ring->me, ring->pipe, ring->queue);
@@ -1655,7 +1655,7 @@ static int gfx_v9_0_sw_init(void *handle)
else
sprintf(ring->name, "gfx_%d", i);
ring->use_doorbell = true;
- ring->doorbell_index = AMDGPU_DOORBELL64_GFX_RING0 << 1;
+ ring->doorbell_index = adev->doorbell_index.gfx_ring0 << 1;
r = amdgpu_ring_init(adev, ring, 1024,
&adev->gfx.eop_irq, AMDGPU_CP_IRQ_GFX_EOP);
if (r)
@@ -2981,9 +2981,9 @@ static int gfx_v9_0_kiq_init_register(struct amdgpu_ring *ring)
/* enable the doorbell if requested */
if (ring->use_doorbell) {
WREG32_SOC15(GC, 0, mmCP_MEC_DOORBELL_RANGE_LOWER,
- (AMDGPU_DOORBELL64_KIQ *2) << 2);
+ (adev->doorbell_index.kiq * 2) << 2);
WREG32_SOC15(GC, 0, mmCP_MEC_DOORBELL_RANGE_UPPER,
- (AMDGPU_DOORBELL64_USERQUEUE_END * 2) << 2);
+ (adev->doorbell_index.userqueue_end * 2) << 2);
}
WREG32_SOC15(GC, 0, mmCP_HQD_PQ_DOORBELL_CONTROL,
diff --git a/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c b/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c
index 811231e4ec53..3a4e5d8d5162 100644
--- a/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c
@@ -338,9 +338,12 @@ static void gmc_v9_0_flush_gpu_tlb(struct amdgpu_device *adev,
struct amdgpu_vmhub *hub = &adev->vmhub[i];
u32 tmp = gmc_v9_0_get_invalidate_req(vmid, flush_type);
- if (i == AMDGPU_GFXHUB && !adev->in_gpu_reset &&
- adev->gfx.kiq.ring.sched.ready &&
- (amdgpu_sriov_runtime(adev) || !amdgpu_sriov_vf(adev))) {
+ /* This is necessary for a HW workaround under SRIOV as well
+ * as GFXOFF under bare metal
+ */
+ if (adev->gfx.kiq.ring.sched.ready &&
+ (amdgpu_sriov_runtime(adev) || !amdgpu_sriov_vf(adev)) &&
+ !adev->in_gpu_reset) {
uint32_t req = hub->vm_inv_eng0_req + eng;
uint32_t ack = hub->vm_inv_eng0_ack + eng;
diff --git a/drivers/gpu/drm/amd/amdgpu/mxgpu_vi.c b/drivers/gpu/drm/amd/amdgpu/mxgpu_vi.c
index 64e875d528dd..6a0fcd67662a 100644
--- a/drivers/gpu/drm/amd/amdgpu/mxgpu_vi.c
+++ b/drivers/gpu/drm/amd/amdgpu/mxgpu_vi.c
@@ -37,7 +37,6 @@
#include "gmc/gmc_8_2_sh_mask.h"
#include "oss/oss_3_0_d.h"
#include "oss/oss_3_0_sh_mask.h"
-#include "gca/gfx_8_0_sh_mask.h"
#include "dce/dce_10_0_d.h"
#include "dce/dce_10_0_sh_mask.h"
#include "smu/smu_7_1_3_d.h"
diff --git a/drivers/gpu/drm/amd/amdgpu/psp_v10_0.c b/drivers/gpu/drm/amd/amdgpu/psp_v10_0.c
index 295c2205485a..d78b4306a36f 100644
--- a/drivers/gpu/drm/amd/amdgpu/psp_v10_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/psp_v10_0.c
@@ -240,12 +240,9 @@ static int psp_v10_0_ring_stop(struct psp_context *psp,
enum psp_ring_type ring_type)
{
int ret = 0;
- struct psp_ring *ring;
unsigned int psp_ring_reg = 0;
struct amdgpu_device *adev = psp->adev;
- ring = &psp->km_ring;
-
/* Write the ring destroy command to C2PMSG_64 */
psp_ring_reg = 3 << 16;
WREG32_SOC15(MP0, 0, mmMP0_SMN_C2PMSG_64, psp_ring_reg);
diff --git a/drivers/gpu/drm/amd/amdgpu/psp_v3_1.c b/drivers/gpu/drm/amd/amdgpu/psp_v3_1.c
index 9cea0bbe4525..7efb823dd3b1 100644
--- a/drivers/gpu/drm/amd/amdgpu/psp_v3_1.c
+++ b/drivers/gpu/drm/amd/amdgpu/psp_v3_1.c
@@ -356,12 +356,9 @@ static int psp_v3_1_ring_stop(struct psp_context *psp,
enum psp_ring_type ring_type)
{
int ret = 0;
- struct psp_ring *ring;
unsigned int psp_ring_reg = 0;
struct amdgpu_device *adev = psp->adev;
- ring = &psp->km_ring;
-
/* Write the ring destroy command to C2PMSG_64 */
psp_ring_reg = 3 << 16;
WREG32_SOC15(MP0, 0, mmMP0_SMN_C2PMSG_64, psp_ring_reg);
@@ -593,7 +590,7 @@ static int psp_v3_1_mode1_reset(struct psp_context *psp)
}
/*send the mode 1 reset command*/
- WREG32(offset, 0x70000);
+ WREG32(offset, GFX_CTRL_CMD_ID_MODE1_RST);
mdelay(1000);
diff --git a/drivers/gpu/drm/amd/amdgpu/sdma_v3_0.c b/drivers/gpu/drm/amd/amdgpu/sdma_v3_0.c
index b6a25f92d566..1bccc5fe2d9d 100644
--- a/drivers/gpu/drm/amd/amdgpu/sdma_v3_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/sdma_v3_0.c
@@ -1146,7 +1146,7 @@ static int sdma_v3_0_sw_init(void *handle)
if (!amdgpu_sriov_vf(adev)) {
ring->use_doorbell = true;
ring->doorbell_index = (i == 0) ?
- AMDGPU_DOORBELL_sDMA_ENGINE0 : AMDGPU_DOORBELL_sDMA_ENGINE1;
+ adev->doorbell_index.sdma_engine0 : adev->doorbell_index.sdma_engine1;
} else {
ring->use_pollmem = true;
}
diff --git a/drivers/gpu/drm/amd/amdgpu/sdma_v4_0.c b/drivers/gpu/drm/amd/amdgpu/sdma_v4_0.c
index f4490cdd9804..4b6d3e5c821f 100644
--- a/drivers/gpu/drm/amd/amdgpu/sdma_v4_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/sdma_v4_0.c
@@ -925,11 +925,9 @@ static void sdma_v4_0_page_resume(struct amdgpu_device *adev, unsigned int i)
OFFSET, ring->doorbell_index);
WREG32_SDMA(i, mmSDMA0_PAGE_DOORBELL, doorbell);
WREG32_SDMA(i, mmSDMA0_PAGE_DOORBELL_OFFSET, doorbell_offset);
- /* TODO: enable doorbell support */
- /*adev->nbio_funcs->sdma_doorbell_range(adev, i, ring->use_doorbell,
- ring->doorbell_index);*/
- sdma_v4_0_ring_set_wptr(ring);
+ /* paging queue doorbell range is setup at sdma_v4_0_gfx_resume */
+ sdma_v4_0_page_ring_set_wptr(ring);
/* set minor_ptr_update to 0 after wptr programed */
WREG32_SDMA(i, mmSDMA0_PAGE_MINOR_PTR_UPDATE, 0);
@@ -1449,23 +1447,46 @@ static void sdma_v4_0_ring_emit_reg_wait(struct amdgpu_ring *ring, uint32_t reg,
sdma_v4_0_wait_reg_mem(ring, 0, 0, reg, 0, val, mask, 10);
}
+static bool sdma_v4_0_fw_support_paging_queue(struct amdgpu_device *adev)
+{
+ uint fw_version = adev->sdma.instance[0].fw_version;
+
+ switch (adev->asic_type) {
+ case CHIP_VEGA10:
+ return fw_version >= 430;
+ case CHIP_VEGA12:
+ /*return fw_version >= 31;*/
+ return false;
+ case CHIP_VEGA20:
+ /*return fw_version >= 115;*/
+ return false;
+ default:
+ return false;
+ }
+}
+
static int sdma_v4_0_early_init(void *handle)
{
struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+ int r;
- if (adev->asic_type == CHIP_RAVEN) {
+ if (adev->asic_type == CHIP_RAVEN)
adev->sdma.num_instances = 1;
- adev->sdma.has_page_queue = false;
- } else {
+ else
adev->sdma.num_instances = 2;
- /* TODO: Page queue breaks driver reload under SRIOV */
- if ((adev->asic_type == CHIP_VEGA10) && amdgpu_sriov_vf((adev)))
- adev->sdma.has_page_queue = false;
- else if (adev->asic_type != CHIP_VEGA20 &&
- adev->asic_type != CHIP_VEGA12)
- adev->sdma.has_page_queue = true;
+
+ r = sdma_v4_0_init_microcode(adev);
+ if (r) {
+ DRM_ERROR("Failed to load sdma firmware!\n");
+ return r;
}
+ /* TODO: Page queue breaks driver reload under SRIOV */
+ if ((adev->asic_type == CHIP_VEGA10) && amdgpu_sriov_vf((adev)))
+ adev->sdma.has_page_queue = false;
+ else if (sdma_v4_0_fw_support_paging_queue(adev))
+ adev->sdma.has_page_queue = true;
+
sdma_v4_0_set_ring_funcs(adev);
sdma_v4_0_set_buffer_funcs(adev);
sdma_v4_0_set_vm_pte_funcs(adev);
@@ -1474,7 +1495,6 @@ static int sdma_v4_0_early_init(void *handle)
return 0;
}
-
static int sdma_v4_0_sw_init(void *handle)
{
struct amdgpu_ring *ring;
@@ -1493,12 +1513,6 @@ static int sdma_v4_0_sw_init(void *handle)
if (r)
return r;
- r = sdma_v4_0_init_microcode(adev);
- if (r) {
- DRM_ERROR("Failed to load sdma firmware!\n");
- return r;
- }
-
for (i = 0; i < adev->sdma.num_instances; i++) {
ring = &adev->sdma.instance[i].ring;
ring->ring_obj = NULL;
@@ -1507,15 +1521,10 @@ static int sdma_v4_0_sw_init(void *handle)
DRM_INFO("use_doorbell being set to: [%s]\n",
ring->use_doorbell?"true":"false");
- if (adev->asic_type == CHIP_VEGA10)
- ring->doorbell_index = (i == 0) ?
- (AMDGPU_VEGA10_DOORBELL64_sDMA_ENGINE0 << 1) //get DWORD offset
- : (AMDGPU_VEGA10_DOORBELL64_sDMA_ENGINE1 << 1); // get DWORD offset
- else
- ring->doorbell_index = (i == 0) ?
- (AMDGPU_DOORBELL64_sDMA_ENGINE0 << 1) //get DWORD offset
- : (AMDGPU_DOORBELL64_sDMA_ENGINE1 << 1); // get DWORD offset
-
+ /* doorbell size is 2 dwords, get DWORD offset */
+ ring->doorbell_index = (i == 0) ?
+ (adev->doorbell_index.sdma_engine0 << 1)
+ : (adev->doorbell_index.sdma_engine1 << 1);
sprintf(ring->name, "sdma%d", i);
r = amdgpu_ring_init(adev, ring, 1024,
@@ -1529,7 +1538,15 @@ static int sdma_v4_0_sw_init(void *handle)
if (adev->sdma.has_page_queue) {
ring = &adev->sdma.instance[i].page;
ring->ring_obj = NULL;
- ring->use_doorbell = false;
+ ring->use_doorbell = true;
+
+ /* paging queue use same doorbell index/routing as gfx queue
+ * with 0x400 (4096 dwords) offset on second doorbell page
+ */
+ ring->doorbell_index = (i == 0) ?
+ (adev->doorbell_index.sdma_engine0 << 1)
+ : (adev->doorbell_index.sdma_engine1 << 1);
+ ring->doorbell_index += 0x400;
sprintf(ring->name, "page%d", i);
r = amdgpu_ring_init(adev, ring, 1024,
diff --git a/drivers/gpu/drm/amd/amdgpu/soc15.c b/drivers/gpu/drm/amd/amdgpu/soc15.c
index 4cc0dcb1a187..83624e150ca7 100644
--- a/drivers/gpu/drm/amd/amdgpu/soc15.c
+++ b/drivers/gpu/drm/amd/amdgpu/soc15.c
@@ -613,6 +613,24 @@ static const struct amdgpu_asic_funcs soc15_asic_funcs =
.flush_hdp = &soc15_flush_hdp,
.invalidate_hdp = &soc15_invalidate_hdp,
.need_full_reset = &soc15_need_full_reset,
+ .init_doorbell_index = &vega10_doorbell_index_init,
+};
+
+static const struct amdgpu_asic_funcs vega20_asic_funcs =
+{
+ .read_disabled_bios = &soc15_read_disabled_bios,
+ .read_bios_from_rom = &soc15_read_bios_from_rom,
+ .read_register = &soc15_read_register,
+ .reset = &soc15_asic_reset,
+ .set_vga_state = &soc15_vga_set_state,
+ .get_xclk = &soc15_get_xclk,
+ .set_uvd_clocks = &soc15_set_uvd_clocks,
+ .set_vce_clocks = &soc15_set_vce_clocks,
+ .get_config_memsize = &soc15_get_config_memsize,
+ .flush_hdp = &soc15_flush_hdp,
+ .invalidate_hdp = &soc15_invalidate_hdp,
+ .need_full_reset = &soc15_need_full_reset,
+ .init_doorbell_index = &vega20_doorbell_index_init,
};
static int soc15_common_early_init(void *handle)
@@ -632,11 +650,11 @@ static int soc15_common_early_init(void *handle)
adev->se_cac_rreg = &soc15_se_cac_rreg;
adev->se_cac_wreg = &soc15_se_cac_wreg;
- adev->asic_funcs = &soc15_asic_funcs;
adev->external_rev_id = 0xFF;
switch (adev->asic_type) {
case CHIP_VEGA10:
+ adev->asic_funcs = &soc15_asic_funcs;
adev->cg_flags = AMD_CG_SUPPORT_GFX_MGCG |
AMD_CG_SUPPORT_GFX_MGLS |
AMD_CG_SUPPORT_GFX_RLC_LS |
@@ -660,6 +678,7 @@ static int soc15_common_early_init(void *handle)
adev->external_rev_id = 0x1;
break;
case CHIP_VEGA12:
+ adev->asic_funcs = &soc15_asic_funcs;
adev->cg_flags = AMD_CG_SUPPORT_GFX_MGCG |
AMD_CG_SUPPORT_GFX_MGLS |
AMD_CG_SUPPORT_GFX_CGCG |
@@ -682,6 +701,7 @@ static int soc15_common_early_init(void *handle)
adev->external_rev_id = adev->rev_id + 0x14;
break;
case CHIP_VEGA20:
+ adev->asic_funcs = &vega20_asic_funcs;
adev->cg_flags = AMD_CG_SUPPORT_GFX_MGCG |
AMD_CG_SUPPORT_GFX_MGLS |
AMD_CG_SUPPORT_GFX_CGCG |
@@ -704,6 +724,7 @@ static int soc15_common_early_init(void *handle)
adev->external_rev_id = adev->rev_id + 0x28;
break;
case CHIP_RAVEN:
+ adev->asic_funcs = &soc15_asic_funcs;
if (adev->rev_id >= 0x8)
adev->external_rev_id = adev->rev_id + 0x81;
else if (adev->pdev->device == 0x15d8)
diff --git a/drivers/gpu/drm/amd/amdgpu/soc15.h b/drivers/gpu/drm/amd/amdgpu/soc15.h
index f8ad7804dc40..a66c8bfbbaa6 100644
--- a/drivers/gpu/drm/amd/amdgpu/soc15.h
+++ b/drivers/gpu/drm/amd/amdgpu/soc15.h
@@ -58,4 +58,6 @@ 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);
+void vega10_doorbell_index_init(struct amdgpu_device *adev);
+void vega20_doorbell_index_init(struct amdgpu_device *adev);
#endif
diff --git a/drivers/gpu/drm/amd/amdgpu/tonga_ih.c b/drivers/gpu/drm/amd/amdgpu/tonga_ih.c
index 3abffd06b5c7..dcdbb4d72472 100644
--- a/drivers/gpu/drm/amd/amdgpu/tonga_ih.c
+++ b/drivers/gpu/drm/amd/amdgpu/tonga_ih.c
@@ -322,7 +322,7 @@ static int tonga_ih_sw_init(void *handle)
return r;
adev->irq.ih.use_doorbell = true;
- adev->irq.ih.doorbell_index = AMDGPU_DOORBELL_IH;
+ adev->irq.ih.doorbell_index = adev->doorbell_index.ih;
r = amdgpu_irq_init(adev);
diff --git a/drivers/gpu/drm/amd/amdgpu/uvd_v4_2.c b/drivers/gpu/drm/amd/amdgpu/uvd_v4_2.c
index 90bbcee00f28..d69c8f6daaf8 100644
--- a/drivers/gpu/drm/amd/amdgpu/uvd_v4_2.c
+++ b/drivers/gpu/drm/amd/amdgpu/uvd_v4_2.c
@@ -116,16 +116,16 @@ static int uvd_v4_2_sw_init(void *handle)
if (r)
return r;
- r = amdgpu_uvd_resume(adev);
- if (r)
- return r;
-
ring = &adev->uvd.inst->ring;
sprintf(ring->name, "uvd");
r = amdgpu_ring_init(adev, ring, 512, &adev->uvd.inst->irq, 0);
if (r)
return r;
+ r = amdgpu_uvd_resume(adev);
+ if (r)
+ return r;
+
r = amdgpu_uvd_entity_init(adev);
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 1c5e12703103..ee8cd06ddc38 100644
--- a/drivers/gpu/drm/amd/amdgpu/uvd_v5_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/uvd_v5_0.c
@@ -113,16 +113,16 @@ static int uvd_v5_0_sw_init(void *handle)
if (r)
return r;
- r = amdgpu_uvd_resume(adev);
- if (r)
- return r;
-
ring = &adev->uvd.inst->ring;
sprintf(ring->name, "uvd");
r = amdgpu_ring_init(adev, ring, 512, &adev->uvd.inst->irq, 0);
if (r)
return r;
+ r = amdgpu_uvd_resume(adev);
+ if (r)
+ return r;
+
r = amdgpu_uvd_entity_init(adev);
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 f184842ef2a2..d4f4a66f8324 100644
--- a/drivers/gpu/drm/amd/amdgpu/uvd_v6_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/uvd_v6_0.c
@@ -400,16 +400,16 @@ static int uvd_v6_0_sw_init(void *handle)
DRM_INFO("UVD ENC is disabled\n");
}
- r = amdgpu_uvd_resume(adev);
- if (r)
- return r;
-
ring = &adev->uvd.inst->ring;
sprintf(ring->name, "uvd");
r = amdgpu_ring_init(adev, ring, 512, &adev->uvd.inst->irq, 0);
if (r)
return r;
+ r = amdgpu_uvd_resume(adev);
+ if (r)
+ return r;
+
if (uvd_v6_0_enc_support(adev)) {
for (i = 0; i < adev->uvd.num_enc_rings; ++i) {
ring = &adev->uvd.inst->ring_enc[i];
diff --git a/drivers/gpu/drm/amd/amdgpu/uvd_v7_0.c b/drivers/gpu/drm/amd/amdgpu/uvd_v7_0.c
index 8a4595968d98..089645e78f98 100644
--- a/drivers/gpu/drm/amd/amdgpu/uvd_v7_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/uvd_v7_0.c
@@ -430,10 +430,6 @@ static int uvd_v7_0_sw_init(void *handle)
DRM_INFO("PSP loading UVD firmware\n");
}
- r = amdgpu_uvd_resume(adev);
- if (r)
- return r;
-
for (j = 0; j < adev->uvd.num_uvd_inst; j++) {
if (adev->uvd.harvest_config & (1 << j))
continue;
@@ -455,9 +451,9 @@ static int uvd_v7_0_sw_init(void *handle)
* sriov, so set unused location for other unused rings.
*/
if (i == 0)
- ring->doorbell_index = AMDGPU_DOORBELL64_UVD_RING0_1 * 2;
+ ring->doorbell_index = adev->doorbell_index.uvd_vce.uvd_ring0_1 * 2;
else
- ring->doorbell_index = AMDGPU_DOORBELL64_UVD_RING2_3 * 2 + 1;
+ ring->doorbell_index = adev->doorbell_index.uvd_vce.uvd_ring2_3 * 2 + 1;
}
r = amdgpu_ring_init(adev, ring, 512, &adev->uvd.inst[j].irq, 0);
if (r)
@@ -465,6 +461,10 @@ static int uvd_v7_0_sw_init(void *handle)
}
}
+ r = amdgpu_uvd_resume(adev);
+ if (r)
+ return r;
+
r = amdgpu_uvd_entity_init(adev);
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 3e84840859a7..2668effadd27 100644
--- a/drivers/gpu/drm/amd/amdgpu/vce_v3_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/vce_v3_0.c
@@ -37,7 +37,6 @@
#include "gca/gfx_8_0_d.h"
#include "smu/smu_7_1_2_d.h"
#include "smu/smu_7_1_2_sh_mask.h"
-#include "gca/gfx_8_0_d.h"
#include "gca/gfx_8_0_sh_mask.h"
#include "ivsrcid/ivsrcid_vislands30.h"
diff --git a/drivers/gpu/drm/amd/amdgpu/vce_v4_0.c b/drivers/gpu/drm/amd/amdgpu/vce_v4_0.c
index 0054ba1b9a68..9fb34b7d8e03 100644
--- a/drivers/gpu/drm/amd/amdgpu/vce_v4_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/vce_v4_0.c
@@ -466,9 +466,9 @@ static int vce_v4_0_sw_init(void *handle)
* so set unused location for other unused rings.
*/
if (i == 0)
- ring->doorbell_index = AMDGPU_DOORBELL64_VCE_RING0_1 * 2;
+ ring->doorbell_index = adev->doorbell_index.uvd_vce.vce_ring0_1 * 2;
else
- ring->doorbell_index = AMDGPU_DOORBELL64_VCE_RING2_3 * 2 + 1;
+ ring->doorbell_index = adev->doorbell_index.uvd_vce.vce_ring2_3 * 2 + 1;
}
r = amdgpu_ring_init(adev, ring, 512, &adev->vce.irq, 0);
if (r)
diff --git a/drivers/gpu/drm/amd/amdgpu/vega10_ih.c b/drivers/gpu/drm/amd/amdgpu/vega10_ih.c
index a0fda6f9252a..d84b687240d1 100644
--- a/drivers/gpu/drm/amd/amdgpu/vega10_ih.c
+++ b/drivers/gpu/drm/amd/amdgpu/vega10_ih.c
@@ -385,7 +385,7 @@ static int vega10_ih_sw_init(void *handle)
return r;
adev->irq.ih.use_doorbell = true;
- adev->irq.ih.doorbell_index = AMDGPU_DOORBELL64_IH << 1;
+ adev->irq.ih.doorbell_index = adev->doorbell_index.ih << 1;
r = amdgpu_irq_init(adev);
diff --git a/drivers/gpu/drm/amd/amdgpu/vega10_reg_init.c b/drivers/gpu/drm/amd/amdgpu/vega10_reg_init.c
index c5c9b2bc190d..422674bb3cdf 100644
--- a/drivers/gpu/drm/amd/amdgpu/vega10_reg_init.c
+++ b/drivers/gpu/drm/amd/amdgpu/vega10_reg_init.c
@@ -56,4 +56,32 @@ int vega10_reg_base_init(struct amdgpu_device *adev)
return 0;
}
+void vega10_doorbell_index_init(struct amdgpu_device *adev)
+{
+ adev->doorbell_index.kiq = AMDGPU_DOORBELL64_KIQ;
+ adev->doorbell_index.mec_ring0 = AMDGPU_DOORBELL64_MEC_RING0;
+ adev->doorbell_index.mec_ring1 = AMDGPU_DOORBELL64_MEC_RING1;
+ adev->doorbell_index.mec_ring2 = AMDGPU_DOORBELL64_MEC_RING2;
+ adev->doorbell_index.mec_ring3 = AMDGPU_DOORBELL64_MEC_RING3;
+ adev->doorbell_index.mec_ring4 = AMDGPU_DOORBELL64_MEC_RING4;
+ adev->doorbell_index.mec_ring5 = AMDGPU_DOORBELL64_MEC_RING5;
+ adev->doorbell_index.mec_ring6 = AMDGPU_DOORBELL64_MEC_RING6;
+ adev->doorbell_index.mec_ring7 = AMDGPU_DOORBELL64_MEC_RING7;
+ adev->doorbell_index.userqueue_start = AMDGPU_DOORBELL64_USERQUEUE_START;
+ adev->doorbell_index.userqueue_end = AMDGPU_DOORBELL64_USERQUEUE_END;
+ adev->doorbell_index.gfx_ring0 = AMDGPU_DOORBELL64_GFX_RING0;
+ adev->doorbell_index.sdma_engine0 = AMDGPU_DOORBELL64_sDMA_ENGINE0;
+ adev->doorbell_index.sdma_engine1 = AMDGPU_DOORBELL64_sDMA_ENGINE1;
+ adev->doorbell_index.ih = AMDGPU_DOORBELL64_IH;
+ adev->doorbell_index.uvd_vce.uvd_ring0_1 = AMDGPU_DOORBELL64_UVD_RING0_1;
+ adev->doorbell_index.uvd_vce.uvd_ring2_3 = AMDGPU_DOORBELL64_UVD_RING2_3;
+ adev->doorbell_index.uvd_vce.uvd_ring4_5 = AMDGPU_DOORBELL64_UVD_RING4_5;
+ adev->doorbell_index.uvd_vce.uvd_ring6_7 = AMDGPU_DOORBELL64_UVD_RING6_7;
+ adev->doorbell_index.uvd_vce.vce_ring0_1 = AMDGPU_DOORBELL64_VCE_RING0_1;
+ adev->doorbell_index.uvd_vce.vce_ring2_3 = AMDGPU_DOORBELL64_VCE_RING2_3;
+ adev->doorbell_index.uvd_vce.vce_ring4_5 = AMDGPU_DOORBELL64_VCE_RING4_5;
+ adev->doorbell_index.uvd_vce.vce_ring6_7 = AMDGPU_DOORBELL64_VCE_RING6_7;
+ /* In unit of dword doorbell */
+ adev->doorbell_index.max_assignment = AMDGPU_DOORBELL64_MAX_ASSIGNMENT << 1;
+}
diff --git a/drivers/gpu/drm/amd/amdgpu/vega20_reg_init.c b/drivers/gpu/drm/amd/amdgpu/vega20_reg_init.c
index d13fc4fcb517..edce413fda9a 100644
--- a/drivers/gpu/drm/amd/amdgpu/vega20_reg_init.c
+++ b/drivers/gpu/drm/amd/amdgpu/vega20_reg_init.c
@@ -54,4 +54,37 @@ int vega20_reg_base_init(struct amdgpu_device *adev)
return 0;
}
+void vega20_doorbell_index_init(struct amdgpu_device *adev)
+{
+ adev->doorbell_index.kiq = AMDGPU_VEGA20_DOORBELL_KIQ;
+ adev->doorbell_index.mec_ring0 = AMDGPU_VEGA20_DOORBELL_MEC_RING0;
+ adev->doorbell_index.mec_ring1 = AMDGPU_VEGA20_DOORBELL_MEC_RING1;
+ adev->doorbell_index.mec_ring2 = AMDGPU_VEGA20_DOORBELL_MEC_RING2;
+ adev->doorbell_index.mec_ring3 = AMDGPU_VEGA20_DOORBELL_MEC_RING3;
+ adev->doorbell_index.mec_ring4 = AMDGPU_VEGA20_DOORBELL_MEC_RING4;
+ adev->doorbell_index.mec_ring5 = AMDGPU_VEGA20_DOORBELL_MEC_RING5;
+ adev->doorbell_index.mec_ring6 = AMDGPU_VEGA20_DOORBELL_MEC_RING6;
+ adev->doorbell_index.mec_ring7 = AMDGPU_VEGA20_DOORBELL_MEC_RING7;
+ adev->doorbell_index.userqueue_start = AMDGPU_VEGA20_DOORBELL_USERQUEUE_START;
+ adev->doorbell_index.userqueue_end = AMDGPU_VEGA20_DOORBELL_USERQUEUE_END;
+ adev->doorbell_index.gfx_ring0 = AMDGPU_VEGA20_DOORBELL_GFX_RING0;
+ adev->doorbell_index.sdma_engine0 = AMDGPU_VEGA20_DOORBELL_sDMA_ENGINE0;
+ adev->doorbell_index.sdma_engine1 = AMDGPU_VEGA20_DOORBELL_sDMA_ENGINE1;
+ adev->doorbell_index.sdma_engine2 = AMDGPU_VEGA20_DOORBELL_sDMA_ENGINE2;
+ adev->doorbell_index.sdma_engine3 = AMDGPU_VEGA20_DOORBELL_sDMA_ENGINE3;
+ adev->doorbell_index.sdma_engine4 = AMDGPU_VEGA20_DOORBELL_sDMA_ENGINE4;
+ adev->doorbell_index.sdma_engine5 = AMDGPU_VEGA20_DOORBELL_sDMA_ENGINE5;
+ adev->doorbell_index.sdma_engine6 = AMDGPU_VEGA20_DOORBELL_sDMA_ENGINE6;
+ adev->doorbell_index.sdma_engine7 = AMDGPU_VEGA20_DOORBELL_sDMA_ENGINE7;
+ adev->doorbell_index.ih = AMDGPU_VEGA20_DOORBELL_IH;
+ adev->doorbell_index.uvd_vce.uvd_ring0_1 = AMDGPU_VEGA20_DOORBELL64_UVD_RING0_1;
+ adev->doorbell_index.uvd_vce.uvd_ring2_3 = AMDGPU_VEGA20_DOORBELL64_UVD_RING2_3;
+ adev->doorbell_index.uvd_vce.uvd_ring4_5 = AMDGPU_VEGA20_DOORBELL64_UVD_RING4_5;
+ adev->doorbell_index.uvd_vce.uvd_ring6_7 = AMDGPU_VEGA20_DOORBELL64_UVD_RING6_7;
+ adev->doorbell_index.uvd_vce.vce_ring0_1 = AMDGPU_VEGA20_DOORBELL64_VCE_RING0_1;
+ adev->doorbell_index.uvd_vce.vce_ring2_3 = AMDGPU_VEGA20_DOORBELL64_VCE_RING2_3;
+ adev->doorbell_index.uvd_vce.vce_ring4_5 = AMDGPU_VEGA20_DOORBELL64_VCE_RING4_5;
+ adev->doorbell_index.uvd_vce.vce_ring6_7 = AMDGPU_VEGA20_DOORBELL64_VCE_RING6_7;
+ adev->doorbell_index.max_assignment = AMDGPU_VEGA20_DOORBELL_MAX_ASSIGNMENT << 1;
+}
diff --git a/drivers/gpu/drm/amd/amdgpu/vi.c b/drivers/gpu/drm/amd/amdgpu/vi.c
index 07880d35e9de..ff2906c215fa 100644
--- a/drivers/gpu/drm/amd/amdgpu/vi.c
+++ b/drivers/gpu/drm/amd/amdgpu/vi.c
@@ -955,6 +955,7 @@ static const struct amdgpu_asic_funcs vi_asic_funcs =
.flush_hdp = &vi_flush_hdp,
.invalidate_hdp = &vi_invalidate_hdp,
.need_full_reset = &vi_need_full_reset,
+ .init_doorbell_index = &legacy_doorbell_index_init,
};
#define CZ_REV_BRISTOL(rev) \
@@ -1712,3 +1713,21 @@ int vi_set_ip_blocks(struct amdgpu_device *adev)
return 0;
}
+
+void legacy_doorbell_index_init(struct amdgpu_device *adev)
+{
+ adev->doorbell_index.kiq = AMDGPU_DOORBELL_KIQ;
+ adev->doorbell_index.mec_ring0 = AMDGPU_DOORBELL_MEC_RING0;
+ adev->doorbell_index.mec_ring1 = AMDGPU_DOORBELL_MEC_RING1;
+ adev->doorbell_index.mec_ring2 = AMDGPU_DOORBELL_MEC_RING2;
+ adev->doorbell_index.mec_ring3 = AMDGPU_DOORBELL_MEC_RING3;
+ adev->doorbell_index.mec_ring4 = AMDGPU_DOORBELL_MEC_RING4;
+ adev->doorbell_index.mec_ring5 = AMDGPU_DOORBELL_MEC_RING5;
+ adev->doorbell_index.mec_ring6 = AMDGPU_DOORBELL_MEC_RING6;
+ adev->doorbell_index.mec_ring7 = AMDGPU_DOORBELL_MEC_RING7;
+ adev->doorbell_index.gfx_ring0 = AMDGPU_DOORBELL_GFX_RING0;
+ adev->doorbell_index.sdma_engine0 = AMDGPU_DOORBELL_sDMA_ENGINE0;
+ adev->doorbell_index.sdma_engine1 = AMDGPU_DOORBELL_sDMA_ENGINE1;
+ adev->doorbell_index.ih = AMDGPU_DOORBELL_IH;
+ adev->doorbell_index.max_assignment = AMDGPU_DOORBELL_MAX_ASSIGNMENT;
+}
diff --git a/drivers/gpu/drm/amd/amdgpu/vi.h b/drivers/gpu/drm/amd/amdgpu/vi.h
index 0429fe332269..8de0772f986c 100644
--- a/drivers/gpu/drm/amd/amdgpu/vi.h
+++ b/drivers/gpu/drm/amd/amdgpu/vi.h
@@ -30,4 +30,5 @@ void vi_srbm_select(struct amdgpu_device *adev,
u32 me, u32 pipe, u32 queue, u32 vmid);
int vi_set_ip_blocks(struct amdgpu_device *adev);
+void legacy_doorbell_index_init(struct amdgpu_device *adev);
#endif