diff options
Diffstat (limited to 'drivers/block')
37 files changed, 6780 insertions, 1530 deletions
diff --git a/drivers/block/DAC960.c b/drivers/block/DAC960.c index e086fbbbe853..8db9089127c5 100644 --- a/drivers/block/DAC960.c +++ b/drivers/block/DAC960.c @@ -1177,7 +1177,8 @@ static bool DAC960_V1_EnableMemoryMailboxInterface(DAC960_Controller_T int TimeoutCounter; int i; - + memset(&CommandMailbox, 0, sizeof(DAC960_V1_CommandMailbox_T)); + if (pci_set_dma_mask(Controller->PCIDevice, DMA_BIT_MASK(32))) return DAC960_Failure(Controller, "DMA mask out of range"); Controller->BounceBufferLimit = DMA_BIT_MASK(32); @@ -4627,7 +4628,8 @@ static void DAC960_V2_ProcessCompletedCommand(DAC960_Command_T *Command) DAC960_Controller_T *Controller = Command->Controller; DAC960_CommandType_T CommandType = Command->CommandType; DAC960_V2_CommandMailbox_T *CommandMailbox = &Command->V2.CommandMailbox; - DAC960_V2_IOCTL_Opcode_T CommandOpcode = CommandMailbox->Common.IOCTL_Opcode; + DAC960_V2_IOCTL_Opcode_T IOCTLOpcode = CommandMailbox->Common.IOCTL_Opcode; + DAC960_V2_CommandOpcode_T CommandOpcode = CommandMailbox->SCSI_10.CommandOpcode; DAC960_V2_CommandStatus_T CommandStatus = Command->V2.CommandStatus; if (CommandType == DAC960_ReadCommand || @@ -4699,7 +4701,7 @@ static void DAC960_V2_ProcessCompletedCommand(DAC960_Command_T *Command) { if (Controller->ShutdownMonitoringTimer) return; - if (CommandOpcode == DAC960_V2_GetControllerInfo) + if (IOCTLOpcode == DAC960_V2_GetControllerInfo) { DAC960_V2_ControllerInfo_T *NewControllerInfo = Controller->V2.NewControllerInformation; @@ -4719,14 +4721,14 @@ static void DAC960_V2_ProcessCompletedCommand(DAC960_Command_T *Command) memcpy(ControllerInfo, NewControllerInfo, sizeof(DAC960_V2_ControllerInfo_T)); } - else if (CommandOpcode == DAC960_V2_GetEvent) + else if (IOCTLOpcode == DAC960_V2_GetEvent) { if (CommandStatus == DAC960_V2_NormalCompletion) { DAC960_V2_ReportEvent(Controller, Controller->V2.Event); } Controller->V2.NextEventSequenceNumber++; } - else if (CommandOpcode == DAC960_V2_GetPhysicalDeviceInfoValid && + else if (IOCTLOpcode == DAC960_V2_GetPhysicalDeviceInfoValid && CommandStatus == DAC960_V2_NormalCompletion) { DAC960_V2_PhysicalDeviceInfo_T *NewPhysicalDeviceInfo = @@ -4915,7 +4917,7 @@ static void DAC960_V2_ProcessCompletedCommand(DAC960_Command_T *Command) NewPhysicalDeviceInfo->LogicalUnit++; Controller->V2.PhysicalDeviceIndex++; } - else if (CommandOpcode == DAC960_V2_GetPhysicalDeviceInfoValid) + else if (IOCTLOpcode == DAC960_V2_GetPhysicalDeviceInfoValid) { unsigned int DeviceIndex; for (DeviceIndex = Controller->V2.PhysicalDeviceIndex; @@ -4938,7 +4940,7 @@ static void DAC960_V2_ProcessCompletedCommand(DAC960_Command_T *Command) } Controller->V2.NeedPhysicalDeviceInformation = false; } - else if (CommandOpcode == DAC960_V2_GetLogicalDeviceInfoValid && + else if (IOCTLOpcode == DAC960_V2_GetLogicalDeviceInfoValid && CommandStatus == DAC960_V2_NormalCompletion) { DAC960_V2_LogicalDeviceInfo_T *NewLogicalDeviceInfo = @@ -5065,7 +5067,7 @@ static void DAC960_V2_ProcessCompletedCommand(DAC960_Command_T *Command) [LogicalDeviceNumber] = true; NewLogicalDeviceInfo->LogicalDeviceNumber++; } - else if (CommandOpcode == DAC960_V2_GetLogicalDeviceInfoValid) + else if (IOCTLOpcode == DAC960_V2_GetLogicalDeviceInfoValid) { int LogicalDriveNumber; for (LogicalDriveNumber = 0; diff --git a/drivers/block/Kconfig b/drivers/block/Kconfig index 6f07ec1c2f58..a796407123c7 100644 --- a/drivers/block/Kconfig +++ b/drivers/block/Kconfig @@ -116,6 +116,8 @@ config PARIDE source "drivers/block/paride/Kconfig" +source "drivers/block/mtip32xx/Kconfig" + config BLK_CPQ_DA tristate "Compaq SMART2 support" depends on PCI && VIRT_TO_BUS @@ -315,6 +317,17 @@ config BLK_DEV_NBD If unsure, say N. +config BLK_DEV_NVME + tristate "NVM Express block device" + depends on PCI + ---help--- + The NVM Express driver is for solid state drives directly + connected to the PCI or PCI Express bus. If you know you + don't have one of these, it is safe to answer N. + + To compile this driver as a module, choose M here: the + module will be called nvme. + config BLK_DEV_OSD tristate "OSD object-as-blkdev support" depends on SCSI_OSD_ULD @@ -341,7 +354,7 @@ config BLK_DEV_SX8 Use devices /dev/sx8/$N and /dev/sx8/$Np$M. config BLK_DEV_UB - tristate "Low Performance USB Block driver" + tristate "Low Performance USB Block driver (deprecated)" depends on USB help This driver supports certain USB attached storage devices diff --git a/drivers/block/Makefile b/drivers/block/Makefile index 76646e9a1c91..5b795059f8fb 100644 --- a/drivers/block/Makefile +++ b/drivers/block/Makefile @@ -23,6 +23,7 @@ obj-$(CONFIG_XILINX_SYSACE) += xsysace.o obj-$(CONFIG_CDROM_PKTCDVD) += pktcdvd.o obj-$(CONFIG_MG_DISK) += mg_disk.o obj-$(CONFIG_SUNVDC) += sunvdc.o +obj-$(CONFIG_BLK_DEV_NVME) += nvme.o obj-$(CONFIG_BLK_DEV_OSD) += osdblk.o obj-$(CONFIG_BLK_DEV_UMEM) += umem.o @@ -39,5 +40,6 @@ obj-$(CONFIG_XEN_BLKDEV_FRONTEND) += xen-blkfront.o obj-$(CONFIG_XEN_BLKDEV_BACKEND) += xen-blkback/ obj-$(CONFIG_BLK_DEV_DRBD) += drbd/ obj-$(CONFIG_BLK_DEV_RBD) += rbd.o +obj-$(CONFIG_BLK_DEV_PCIESSD_MTIP32XX) += mtip32xx/ swim_mod-y := swim.o swim_asm.o diff --git a/drivers/block/brd.c b/drivers/block/brd.c index ec246437f5a4..531ceb31d0ff 100644 --- a/drivers/block/brd.c +++ b/drivers/block/brd.c @@ -242,9 +242,9 @@ static void copy_to_brd(struct brd_device *brd, const void *src, page = brd_lookup_page(brd, sector); BUG_ON(!page); - dst = kmap_atomic(page, KM_USER1); + dst = kmap_atomic(page); memcpy(dst + offset, src, copy); - kunmap_atomic(dst, KM_USER1); + kunmap_atomic(dst); if (copy < n) { src += copy; @@ -253,9 +253,9 @@ static void copy_to_brd(struct brd_device *brd, const void *src, page = brd_lookup_page(brd, sector); BUG_ON(!page); - dst = kmap_atomic(page, KM_USER1); + dst = kmap_atomic(page); memcpy(dst, src, copy); - kunmap_atomic(dst, KM_USER1); + kunmap_atomic(dst); } } @@ -273,9 +273,9 @@ static void copy_from_brd(void *dst, struct brd_device *brd, copy = min_t(size_t, n, PAGE_SIZE - offset); page = brd_lookup_page(brd, sector); if (page) { - src = kmap_atomic(page, KM_USER1); + src = kmap_atomic(page); memcpy(dst, src + offset, copy); - kunmap_atomic(src, KM_USER1); + kunmap_atomic(src); } else memset(dst, 0, copy); @@ -285,9 +285,9 @@ static void copy_from_brd(void *dst, struct brd_device *brd, copy = n - copy; page = brd_lookup_page(brd, sector); if (page) { - src = kmap_atomic(page, KM_USER1); + src = kmap_atomic(page); memcpy(dst, src, copy); - kunmap_atomic(src, KM_USER1); + kunmap_atomic(src); } else memset(dst, 0, copy); } @@ -309,7 +309,7 @@ static int brd_do_bvec(struct brd_device *brd, struct page *page, goto out; } - mem = kmap_atomic(page, KM_USER0); + mem = kmap_atomic(page); if (rw == READ) { copy_from_brd(mem + off, brd, sector, len); flush_dcache_page(page); @@ -317,7 +317,7 @@ static int brd_do_bvec(struct brd_device *brd, struct page *page, flush_dcache_page(page); copy_to_brd(brd, mem + off, sector, len); } - kunmap_atomic(mem, KM_USER0); + kunmap_atomic(mem); out: return err; diff --git a/drivers/block/cciss.c b/drivers/block/cciss.c index 587cce57adae..b0f553b26d0f 100644 --- a/drivers/block/cciss.c +++ b/drivers/block/cciss.c @@ -1735,7 +1735,7 @@ static int cciss_ioctl(struct block_device *bdev, fmode_t mode, case CCISS_BIG_PASSTHRU: return cciss_bigpassthru(h, argp); - /* scsi_cmd_ioctl handles these, below, though some are not */ + /* scsi_cmd_blk_ioctl handles these, below, though some are not */ /* very meaningful for cciss. SG_IO is the main one people want. */ case SG_GET_VERSION_NUM: @@ -1746,9 +1746,9 @@ static int cciss_ioctl(struct block_device *bdev, fmode_t mode, case SG_EMULATED_HOST: case SG_IO: case SCSI_IOCTL_SEND_COMMAND: - return scsi_cmd_ioctl(disk->queue, disk, mode, cmd, argp); + return scsi_cmd_blk_ioctl(bdev, mode, cmd, argp); - /* scsi_cmd_ioctl would normally handle these, below, but */ + /* scsi_cmd_blk_ioctl would normally handle these, below, but */ /* they aren't a good fit for cciss, as CD-ROMs are */ /* not supported, and we don't have any bus/target/lun */ /* which we present to the kernel. */ diff --git a/drivers/block/drbd/drbd_bitmap.c b/drivers/block/drbd/drbd_bitmap.c index 912f585a760f..3030201c69d8 100644 --- a/drivers/block/drbd/drbd_bitmap.c +++ b/drivers/block/drbd/drbd_bitmap.c @@ -289,25 +289,25 @@ static unsigned int bm_bit_to_page_idx(struct drbd_bitmap *b, u64 bitnr) return page_nr; } -static unsigned long *__bm_map_pidx(struct drbd_bitmap *b, unsigned int idx, const enum km_type km) +static unsigned long *__bm_map_pidx(struct drbd_bitmap *b, unsigned int idx) { struct page *page = b->bm_pages[idx]; - return (unsigned long *) kmap_atomic(page, km); + return (unsigned long *) kmap_atomic(page); } static unsigned long *bm_map_pidx(struct drbd_bitmap *b, unsigned int idx) { - return __bm_map_pidx(b, idx, KM_IRQ1); + return __bm_map_pidx(b, idx); } -static void __bm_unmap(unsigned long *p_addr, const enum km_type km) +static void __bm_unmap(unsigned long *p_addr) { - kunmap_atomic(p_addr, km); + kunmap_atomic(p_addr); }; static void bm_unmap(unsigned long *p_addr) { - return __bm_unmap(p_addr, KM_IRQ1); + return __bm_unmap(p_addr); } /* long word offset of _bitmap_ sector */ @@ -543,15 +543,15 @@ static unsigned long bm_count_bits(struct drbd_bitmap *b) /* all but last page */ for (idx = 0; idx < b->bm_number_of_pages - 1; idx++) { - p_addr = __bm_map_pidx(b, idx, KM_USER0); + p_addr = __bm_map_pidx(b, idx); for (i = 0; i < LWPP; i++) bits += hweight_long(p_addr[i]); - __bm_unmap(p_addr, KM_USER0); + __bm_unmap(p_addr); cond_resched(); } /* last (or only) page */ last_word = ((b->bm_bits - 1) & BITS_PER_PAGE_MASK) >> LN2_BPL; - p_addr = __bm_map_pidx(b, idx, KM_USER0); + p_addr = __bm_map_pidx(b, idx); for (i = 0; i < last_word; i++) bits += hweight_long(p_addr[i]); p_addr[last_word] &= cpu_to_lel(mask); @@ -559,7 +559,7 @@ static unsigned long bm_count_bits(struct drbd_bitmap *b) /* 32bit arch, may have an unused padding long */ if (BITS_PER_LONG == 32 && (last_word & 1) == 0) p_addr[last_word+1] = 0; - __bm_unmap(p_addr, KM_USER0); + __bm_unmap(p_addr); return bits; } @@ -970,11 +970,11 @@ static void bm_page_io_async(struct bm_aio_ctx *ctx, int page_nr, int rw) __must * to use pre-allocated page pool */ void *src, *dest; page = alloc_page(__GFP_HIGHMEM|__GFP_WAIT); - dest = kmap_atomic(page, KM_USER0); - src = kmap_atomic(b->bm_pages[page_nr], KM_USER1); + dest = kmap_atomic(page); + src = kmap_atomic(b->bm_pages[page_nr]); memcpy(dest, src, PAGE_SIZE); - kunmap_atomic(src, KM_USER1); - kunmap_atomic(dest, KM_USER0); + kunmap_atomic(src); + kunmap_atomic(dest); bm_store_page_idx(page, page_nr); } else page = b->bm_pages[page_nr]; @@ -1163,7 +1163,7 @@ int drbd_bm_write_page(struct drbd_conf *mdev, unsigned int idx) __must_hold(loc * this returns a bit number, NOT a sector! */ static unsigned long __bm_find_next(struct drbd_conf *mdev, unsigned long bm_fo, - const int find_zero_bit, const enum km_type km) + const int find_zero_bit) { struct drbd_bitmap *b = mdev->bitmap; unsigned long *p_addr; @@ -1178,7 +1178,7 @@ static unsigned long __bm_find_next(struct drbd_conf *mdev, unsigned long bm_fo, while (bm_fo < b->bm_bits) { /* bit offset of the first bit in the page */ bit_offset = bm_fo & ~BITS_PER_PAGE_MASK; - p_addr = __bm_map_pidx(b, bm_bit_to_page_idx(b, bm_fo), km); + p_addr = __bm_map_pidx(b, bm_bit_to_page_idx(b, bm_fo)); if (find_zero_bit) i = find_next_zero_bit_le(p_addr, @@ -1187,7 +1187,7 @@ static unsigned long __bm_find_next(struct drbd_conf *mdev, unsigned long bm_fo, i = find_next_bit_le(p_addr, PAGE_SIZE*8, bm_fo & BITS_PER_PAGE_MASK); - __bm_unmap(p_addr, km); + __bm_unmap(p_addr); if (i < PAGE_SIZE*8) { bm_fo = bit_offset + i; if (bm_fo >= b->bm_bits) @@ -1215,7 +1215,7 @@ static unsigned long bm_find_next(struct drbd_conf *mdev, if (BM_DONT_TEST & b->bm_flags) bm_print_lock_info(mdev); - i = __bm_find_next(mdev, bm_fo, find_zero_bit, KM_IRQ1); + i = __bm_find_next(mdev, bm_fo, find_zero_bit); spin_unlock_irq(&b->bm_lock); return i; @@ -1239,13 +1239,13 @@ unsigned long drbd_bm_find_next_zero(struct drbd_conf *mdev, unsigned long bm_fo unsigned long _drbd_bm_find_next(struct drbd_conf *mdev, unsigned long bm_fo) { /* WARN_ON(!(BM_DONT_SET & mdev->b->bm_flags)); */ - return __bm_find_next(mdev, bm_fo, 0, KM_USER1); + return __bm_find_next(mdev, bm_fo, 0); } unsigned long _drbd_bm_find_next_zero(struct drbd_conf *mdev, unsigned long bm_fo) { /* WARN_ON(!(BM_DONT_SET & mdev->b->bm_flags)); */ - return __bm_find_next(mdev, bm_fo, 1, KM_USER1); + return __bm_find_next(mdev, bm_fo, 1); } /* returns number of bits actually changed. @@ -1273,14 +1273,14 @@ static int __bm_change_bits_to(struct drbd_conf *mdev, const unsigned long s, unsigned int page_nr = bm_bit_to_page_idx(b, bitnr); if (page_nr != last_page_nr) { if (p_addr) - __bm_unmap(p_addr, KM_IRQ1); + __bm_unmap(p_addr); if (c < 0) bm_set_page_lazy_writeout(b->bm_pages[last_page_nr]); else if (c > 0) bm_set_page_need_writeout(b->bm_pages[last_page_nr]); changed_total += c; c = 0; - p_addr = __bm_map_pidx(b, page_nr, KM_IRQ1); + p_addr = __bm_map_pidx(b, page_nr); last_page_nr = page_nr; } if (val) @@ -1289,7 +1289,7 @@ static int __bm_change_bits_to(struct drbd_conf *mdev, const unsigned long s, c -= (0 != __test_and_clear_bit_le(bitnr & BITS_PER_PAGE_MASK, p_addr)); } if (p_addr) - __bm_unmap(p_addr, KM_IRQ1); + __bm_unmap(p_addr); if (c < 0) bm_set_page_lazy_writeout(b->bm_pages[last_page_nr]); else if (c > 0) @@ -1342,13 +1342,13 @@ static inline void bm_set_full_words_within_one_page(struct drbd_bitmap *b, { int i; int bits; - unsigned long *paddr = kmap_atomic(b->bm_pages[page_nr], KM_IRQ1); + unsigned long *paddr = kmap_atomic(b->bm_pages[page_nr]); for (i = first_word; i < last_word; i++) { bits = hweight_long(paddr[i]); paddr[i] = ~0UL; b->bm_set += BITS_PER_LONG - bits; } - kunmap_atomic(paddr, KM_IRQ1); + kunmap_atomic(paddr); } /* Same thing as drbd_bm_set_bits, diff --git a/drivers/block/drbd/drbd_int.h b/drivers/block/drbd/drbd_int.h index 9cf20355ceec..8d680562ba73 100644 --- a/drivers/block/drbd/drbd_int.h +++ b/drivers/block/drbd/drbd_int.h @@ -59,8 +59,8 @@ /* module parameter, defined in drbd_main.c */ extern unsigned int minor_count; -extern int disable_sendpage; -extern int allow_oos; +extern bool disable_sendpage; +extern bool allow_oos; extern unsigned int cn_idx; #ifdef CONFIG_DRBD_FAULT_INJECTION diff --git a/drivers/block/drbd/drbd_main.c b/drivers/block/drbd/drbd_main.c index 0358e55356c8..211fc44f84be 100644 --- a/drivers/block/drbd/drbd_main.c +++ b/drivers/block/drbd/drbd_main.c @@ -117,8 +117,8 @@ module_param(fault_devs, int, 0644); /* module parameter, defined */ unsigned int minor_count = DRBD_MINOR_COUNT_DEF; -int disable_sendpage; -int allow_oos; +bool disable_sendpage; +bool allow_oos; unsigned int cn_idx = CN_IDX_DRBD; int proc_details; /* Detail level in proc drbd*/ diff --git a/drivers/block/drbd/drbd_nl.c b/drivers/block/drbd/drbd_nl.c index af2a25049bce..abfaacaaf346 100644 --- a/drivers/block/drbd/drbd_nl.c +++ b/drivers/block/drbd/drbd_nl.c @@ -179,7 +179,7 @@ int drbd_khelper(struct drbd_conf *mdev, char *cmd) dev_info(DEV, "helper command: %s %s %s\n", usermode_helper, cmd, mb); drbd_bcast_ev_helper(mdev, cmd); - ret = call_usermodehelper(usermode_helper, argv, envp, 1); + ret = call_usermodehelper(usermode_helper, argv, envp, UMH_WAIT_PROC); if (ret) dev_warn(DEV, "helper command: %s %s %s exit code %u (0x%x)\n", usermode_helper, cmd, mb, @@ -2526,10 +2526,10 @@ void drbd_bcast_ee(struct drbd_conf *mdev, page = e->pages; page_chain_for_each(page) { - void *d = kmap_atomic(page, KM_USER0); + void *d = kmap_atomic(page); unsigned l = min_t(unsigned, len, PAGE_SIZE); memcpy(tl, d, l); - kunmap_atomic(d, KM_USER0); + kunmap_atomic(d); tl = (unsigned short*)((char*)tl + l); len -= l; if (len == 0) diff --git a/drivers/block/floppy.c b/drivers/block/floppy.c index 510fb10ec45a..b0b00d70c166 100644 --- a/drivers/block/floppy.c +++ b/drivers/block/floppy.c @@ -202,7 +202,6 @@ static int slow_floppy; #include <asm/dma.h> #include <asm/irq.h> -#include <asm/system.h> static int FLOPPY_IRQ = 6; static int FLOPPY_DMA = 2; @@ -1031,37 +1030,6 @@ static int fd_wait_for_completion(unsigned long delay, timeout_fn function) return 0; } -static DEFINE_SPINLOCK(floppy_hlt_lock); -static int hlt_disabled; -static void floppy_disable_hlt(void) -{ - unsigned long flags; - - WARN_ONCE(1, "floppy_disable_hlt() scheduled for removal in 2012"); - spin_lock_irqsave(&floppy_hlt_lock, flags); - if (!hlt_disabled) { - hlt_disabled = 1; -#ifdef HAVE_DISABLE_HLT - disable_hlt(); -#endif - } - spin_unlock_irqrestore(&floppy_hlt_lock, flags); -} - -static void floppy_enable_hlt(void) -{ - unsigned long flags; - - spin_lock_irqsave(&floppy_hlt_lock, flags); - if (hlt_disabled) { - hlt_disabled = 0; -#ifdef HAVE_DISABLE_HLT - enable_hlt(); -#endif - } - spin_unlock_irqrestore(&floppy_hlt_lock, flags); -} - static void setup_DMA(void) { unsigned long f; @@ -1106,7 +1074,6 @@ static void setup_DMA(void) fd_enable_dma(); release_dma_lock(f); #endif - floppy_disable_hlt(); } static void show_floppy(void); @@ -1708,7 +1675,6 @@ irqreturn_t floppy_interrupt(int irq, void *dev_id) fd_disable_dma(); release_dma_lock(f); - floppy_enable_hlt(); do_floppy = NULL; if (fdc >= N_FDC || FDCS->address == -1) { /* we don't even know which FDC is the culprit */ @@ -1857,8 +1823,6 @@ static void floppy_shutdown(unsigned long data) show_floppy(); cancel_activity(); - floppy_enable_hlt(); - flags = claim_dma_lock(); fd_disable_dma(); release_dma_lock(flags); @@ -3832,7 +3796,7 @@ static int __floppy_read_block_0(struct block_device *bdev) bio.bi_size = size; bio.bi_bdev = bdev; bio.bi_sector = 0; - bio.bi_flags = BIO_QUIET; + bio.bi_flags = (1 << BIO_QUIET); init_completion(&complete); bio.bi_private = &complete; bio.bi_end_io = floppy_rb0_complete; @@ -4368,8 +4332,14 @@ out_unreg_blkdev: out_put_disk: while (dr--) { del_timer_sync(&motor_off_timer[dr]); - if (disks[dr]->queue) + if (disks[dr]->queue) { blk_cleanup_queue(disks[dr]->queue); + /* + * put_disk() is not paired with add_disk() and + * will put queue reference one extra time. fix it. + */ + disks[dr]->queue = NULL; + } put_disk(disks[dr]); } return err; @@ -4503,7 +4473,6 @@ static void floppy_release_irq_and_dma(void) #if N_FDC > 1 set_dor(1, ~8, 0); #endif - floppy_enable_hlt(); if (floppy_track_buffer && max_buffer_sectors) { tmpsize = max_buffer_sectors * 1024; @@ -4579,6 +4548,15 @@ static void __exit floppy_module_exit(void) platform_device_unregister(&floppy_device[drive]); } blk_cleanup_queue(disks[drive]->queue); + + /* + * These disks have not called add_disk(). Don't put down + * queue reference in put_disk(). + */ + if (!(allowed_drive_mask & (1 << drive)) || + fdc_state[FDC(drive)].version == FDC_NONE) + disks[drive]->queue = NULL; + put_disk(disks[drive]); } diff --git a/drivers/block/hd.c b/drivers/block/hd.c index b52c9ca146fc..bf397bf108b7 100644 --- a/drivers/block/hd.c +++ b/drivers/block/hd.c @@ -44,7 +44,6 @@ #define HD_IRQ 14 #define REALLY_SLOW_IO -#include <asm/system.h> #include <asm/io.h> #include <asm/uaccess.h> diff --git a/drivers/block/loop.c b/drivers/block/loop.c index f00257782fcc..bbca966f8f66 100644 --- a/drivers/block/loop.c +++ b/drivers/block/loop.c @@ -93,16 +93,16 @@ static int transfer_none(struct loop_device *lo, int cmd, struct page *loop_page, unsigned loop_off, int size, sector_t real_block) { - char *raw_buf = kmap_atomic(raw_page, KM_USER0) + raw_off; - char *loop_buf = kmap_atomic(loop_page, KM_USER1) + loop_off; + char *raw_buf = kmap_atomic(raw_page) + raw_off; + char *loop_buf = kmap_atomic(loop_page) + loop_off; if (cmd == READ) memcpy(loop_buf, raw_buf, size); else memcpy(raw_buf, loop_buf, size); - kunmap_atomic(loop_buf, KM_USER1); - kunmap_atomic(raw_buf, KM_USER0); + kunmap_atomic(loop_buf); + kunmap_atomic(raw_buf); cond_resched(); return 0; } @@ -112,8 +112,8 @@ static int transfer_xor(struct loop_device *lo, int cmd, struct page *loop_page, unsigned loop_off, int size, sector_t real_block) { - char *raw_buf = kmap_atomic(raw_page, KM_USER0) + raw_off; - char *loop_buf = kmap_atomic(loop_page, KM_USER1) + loop_off; + char *raw_buf = kmap_atomic(raw_page) + raw_off; + char *loop_buf = kmap_atomic(loop_page) + loop_off; char *in, *out, *key; int i, keysize; @@ -130,8 +130,8 @@ static int transfer_xor(struct loop_device *lo, int cmd, for (i = 0; i < size; i++) *out++ = *in++ ^ key[(i & 511) % keysize]; - kunmap_atomic(loop_buf, KM_USER1); - kunmap_atomic(raw_buf, KM_USER0); + kunmap_atomic(loop_buf); + kunmap_atomic(raw_buf); cond_resched(); return 0; } @@ -356,14 +356,14 @@ lo_direct_splice_actor(struct pipe_inode_info *pipe, struct splice_desc *sd) return __splice_from_pipe(pipe, sd, lo_splice_actor); } -static int +static ssize_t do_lo_receive(struct loop_device *lo, struct bio_vec *bvec, int bsize, loff_t pos) { struct lo_read_data cookie; struct splice_desc sd; struct file *file; - long retval; + ssize_t retval; cookie.lo = lo; cookie.page = bvec->bv_page; @@ -379,26 +379,28 @@ do_lo_receive(struct loop_device *lo, file = lo->lo_backing_file; retval = splice_direct_to_actor(file, &sd, lo_direct_splice_actor); - if (retval < 0) - return retval; - if (retval != bvec->bv_len) - return -EIO; - return 0; + return retval; } static int lo_receive(struct loop_device *lo, struct bio *bio, int bsize, loff_t pos) { struct bio_vec *bvec; - int i, ret = 0; + ssize_t s; + int i; bio_for_each_segment(bvec, bio, i) { - ret = do_lo_receive(lo, bvec, bsize, pos); - if (ret < 0) + s = do_lo_receive(lo, bvec, bsize, pos); + if (s < 0) + return s; + + if (s != bvec->bv_len) { + zero_fill_bio(bio); break; + } pos += bvec->bv_len; } - return ret; + return 0; } static int do_bio_filebacked(struct loop_device *lo, struct bio *bio) diff --git a/drivers/block/mtip32xx/Kconfig b/drivers/block/mtip32xx/Kconfig new file mode 100644 index 000000000000..b5dd14e072f2 --- /dev/null +++ b/drivers/block/mtip32xx/Kconfig @@ -0,0 +1,9 @@ +# +# mtip32xx device driver configuration +# + +config BLK_DEV_PCIESSD_MTIP32XX + tristate "Block Device Driver for Micron PCIe SSDs" + depends on HOTPLUG_PCI_PCIE + help + This enables the block driver for Micron PCIe SSDs. diff --git a/drivers/block/mtip32xx/Makefile b/drivers/block/mtip32xx/Makefile new file mode 100644 index 000000000000..4fbef8c8329b --- /dev/null +++ b/drivers/block/mtip32xx/Makefile @@ -0,0 +1,5 @@ +# +# Makefile for Block device driver for Micron PCIe SSD +# + +obj-$(CONFIG_BLK_DEV_PCIESSD_MTIP32XX) += mtip32xx.o diff --git a/drivers/block/mtip32xx/mtip32xx.c b/drivers/block/mtip32xx/mtip32xx.c new file mode 100644 index 000000000000..8eb81c96608f --- /dev/null +++ b/drivers/block/mtip32xx/mtip32xx.c @@ -0,0 +1,3650 @@ +/* + * Driver for the Micron P320 SSD + * Copyright (C) 2011 Micron Technology, Inc. + * + * Portions of this code were derived from works subjected to the + * following copyright: + * Copyright (C) 2009 Integrated Device Technology, Inc. + * + * This program is free software; you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation; either version 2 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + */ + +#include <linux/pci.h> +#include <linux/interrupt.h> +#include <linux/ata.h> +#include <linux/delay.h> +#include <linux/hdreg.h> +#include <linux/uaccess.h> +#include <linux/random.h> +#include <linux/smp.h> +#include <linux/compat.h> +#include <linux/fs.h> +#include <linux/module.h> +#include <linux/genhd.h> +#include <linux/blkdev.h> +#include <linux/bio.h> +#include <linux/dma-mapping.h> +#include <linux/idr.h> +#include <linux/kthread.h> +#include <../drivers/ata/ahci.h> +#include "mtip32xx.h" + +#define HW_CMD_SLOT_SZ (MTIP_MAX_COMMAND_SLOTS * 32) +#define HW_CMD_TBL_SZ (AHCI_CMD_TBL_HDR_SZ + (MTIP_MAX_SG * 16)) +#define HW_CMD_TBL_AR_SZ (HW_CMD_TBL_SZ * MTIP_MAX_COMMAND_SLOTS) +#define HW_PORT_PRIV_DMA_SZ \ + (HW_CMD_SLOT_SZ + HW_CMD_TBL_AR_SZ + AHCI_RX_FIS_SZ) + +#define HOST_HSORG 0xFC +#define HSORG_DISABLE_SLOTGRP_INTR (1<<24) +#define HSORG_DISABLE_SLOTGRP_PXIS (1<<16) +#define HSORG_HWREV 0xFF00 +#define HSORG_STYLE 0x8 +#define HSORG_SLOTGROUPS 0x7 + +#define PORT_COMMAND_ISSUE 0x38 +#define PORT_SDBV 0x7C + +#define PORT_OFFSET 0x100 +#define PORT_MEM_SIZE 0x80 + +#define PORT_IRQ_ERR \ + (PORT_IRQ_HBUS_ERR | PORT_IRQ_IF_ERR | PORT_IRQ_CONNECT | \ + PORT_IRQ_PHYRDY | PORT_IRQ_UNK_FIS | PORT_IRQ_BAD_PMP | \ + PORT_IRQ_TF_ERR | PORT_IRQ_HBUS_DATA_ERR | PORT_IRQ_IF_NONFATAL | \ + PORT_IRQ_OVERFLOW) +#define PORT_IRQ_LEGACY \ + (PORT_IRQ_PIOS_FIS | PORT_IRQ_D2H_REG_FIS) +#define PORT_IRQ_HANDLED \ + (PORT_IRQ_SDB_FIS | PORT_IRQ_LEGACY | \ + PORT_IRQ_TF_ERR | PORT_IRQ_IF_ERR | \ + PORT_IRQ_CONNECT | PORT_IRQ_PHYRDY) +#define DEF_PORT_IRQ \ + (PORT_IRQ_ERR | PORT_IRQ_LEGACY | PORT_IRQ_SDB_FIS) + +/* product numbers */ +#define MTIP_PRODUCT_UNKNOWN 0x00 +#define MTIP_PRODUCT_ASICFPGA 0x11 + +/* Device instance number, incremented each time a device is probed. */ +static int instance; + +/* + * Global variable used to hold the major block device number + * allocated in mtip_init(). + */ +static int mtip_major; + +static DEFINE_SPINLOCK(rssd_index_lock); +static DEFINE_IDA(rssd_index_ida); + +static int mtip_block_initialize(struct driver_data *dd); + +#ifdef CONFIG_COMPAT +struct mtip_compat_ide_task_request_s { + __u8 io_ports[8]; + __u8 hob_ports[8]; + ide_reg_valid_t out_flags; + ide_reg_valid_t in_flags; + int data_phase; + int req_cmd; + compat_ulong_t out_size; + compat_ulong_t in_size; +}; +#endif + +/* + * This function check_for_surprise_removal is called + * while card is removed from the system and it will + * read the vendor id from the configration space + * + * @pdev Pointer to the pci_dev structure. + * + * return value + * true if device removed, else false + */ +static bool mtip_check_surprise_removal(struct pci_dev *pdev) +{ + u16 vendor_id = 0; + + /* Read the vendorID from the configuration space */ + pci_read_config_word(pdev, 0x00, &vendor_id); + if (vendor_id == 0xFFFF) + return true; /* device removed */ + + return false; /* device present */ +} + +/* + * This function is called for clean the pending command in the + * command slot during the surprise removal of device and return + * error to the upper layer. + * + * @dd Pointer to the DRIVER_DATA structure. + * + * return value + * None + */ +static void mtip_command_cleanup(struct driver_data *dd) +{ + int group = 0, commandslot = 0, commandindex = 0; + struct mtip_cmd *command; + struct mtip_port *port = dd->port; + + for (group = 0; group < 4; group++) { + for (commandslot = 0; commandslot < 32; commandslot++) { + if (!(port->allocated[group] & (1 << commandslot))) + continue; + + commandindex = group << 5 | commandslot; + command = &port->commands[commandindex]; + + if (atomic_read(&command->active) + && (command->async_callback)) { + command->async_callback(command->async_data, + -ENODEV); + command->async_callback = NULL; + command->async_data = NULL; + } + + dma_unmap_sg(&port->dd->pdev->dev, + command->sg, + command->scatter_ents, + command->direction); + } + } + + up(&port->cmd_slot); + + atomic_set(&dd->drv_cleanup_done, true); +} + +/* + * Obtain an empty command slot. + * + * This function needs to be reentrant since it could be called + * at the same time on multiple CPUs. The allocation of the + * command slot must be atomic. + * + * @port Pointer to the port data structure. + * + * return value + * >= 0 Index of command slot obtained. + * -1 No command slots available. + */ +static int get_slot(struct mtip_port *port) +{ + int slot, i; + unsigned int num_command_slots = port->dd->slot_groups * 32; + + /* + * Try 10 times, because there is a small race here. + * that's ok, because it's still cheaper than a lock. + * + * Race: Since this section is not protected by lock, same bit + * could be chosen by different process contexts running in + * different processor. So instead of costly lock, we are going + * with loop. + */ + for (i = 0; i < 10; i++) { + slot = find_next_zero_bit(port->allocated, + num_command_slots, 1); + if ((slot < num_command_slots) && + (!test_and_set_bit(slot, port->allocated))) + return slot; + } + dev_warn(&port->dd->pdev->dev, "Failed to get a tag.\n"); + + if (mtip_check_surprise_removal(port->dd->pdev)) { + /* Device not present, clean outstanding commands */ + mtip_command_cleanup(port->dd); + } + return -1; +} + +/* + * Release a command slot. + * + * @port Pointer to the port data structure. + * @tag Tag of command to release + * + * return value + * None + */ +static inline void release_slot(struct mtip_port *port, int tag) +{ + smp_mb__before_clear_bit(); + clear_bit(tag, port->allocated); + smp_mb__after_clear_bit(); +} + +/* + * Reset the HBA (without sleeping) + * + * Just like hba_reset, except does not call sleep, so can be + * run from interrupt/tasklet context. + * + * @dd Pointer to the driver data structure. + * + * return value + * 0 The reset was successful. + * -1 The HBA Reset bit did not clear. + */ +static int hba_reset_nosleep(struct driver_data *dd) +{ + unsigned long timeout; + + /* Chip quirk: quiesce any chip function */ + mdelay(10); + + /* Set the reset bit */ + writel(HOST_RESET, dd->mmio + HOST_CTL); + + /* Flush */ + readl(dd->mmio + HOST_CTL); + + /* + * Wait 10ms then spin for up to 1 second + * waiting for reset acknowledgement + */ + timeout = jiffies + msecs_to_jiffies(1000); + mdelay(10); + while ((readl(dd->mmio + HOST_CTL) & HOST_RESET) + && time_before(jiffies, timeout)) + mdelay(1); + + if (readl(dd->mmio + HOST_CTL) & HOST_RESET) + return -1; + + return 0; +} + +/* + * Issue a command to the hardware. + * + * Set the appropriate bit in the s_active and Command Issue hardware + * registers, causing hardware command processing to begin. + * + * @port Pointer to the port structure. + * @tag The tag of the command to be issued. + * + * return value + * None + */ +static inline void mtip_issue_ncq_command(struct mtip_port *port, int tag) +{ + unsigned long flags = 0; + + atomic_set(&port->commands[tag].active, 1); + + spin_lock_irqsave(&port->cmd_issue_lock, flags); + + writel((1 << MTIP_TAG_BIT(tag)), + port->s_active[MTIP_TAG_INDEX(tag)]); + writel((1 << MTIP_TAG_BIT(tag)), + port->cmd_issue[MTIP_TAG_INDEX(tag)]); + + spin_unlock_irqrestore(&port->cmd_issue_lock, flags); +} + +/* + * Enable/disable the reception of FIS + * + * @port Pointer to the port data structure + * @enable 1 to enable, 0 to disable + * + * return value + * Previous state: 1 enabled, 0 disabled + */ +static int mtip_enable_fis(struct mtip_port *port, int enable) +{ + u32 tmp; + + /* enable FIS reception */ + tmp = readl(port->mmio + PORT_CMD); + if (enable) + writel(tmp | PORT_CMD_FIS_RX, port->mmio + PORT_CMD); + else + writel(tmp & ~PORT_CMD_FIS_RX, port->mmio + PORT_CMD); + + /* Flush */ + readl(port->mmio + PORT_CMD); + + return (((tmp & PORT_CMD_FIS_RX) == PORT_CMD_FIS_RX)); +} + +/* + * Enable/disable the DMA engine + * + * @port Pointer to the port data structure + * @enable 1 to enable, 0 to disable + * + * return value + * Previous state: 1 enabled, 0 disabled. + */ +static int mtip_enable_engine(struct mtip_port *port, int enable) +{ + u32 tmp; + + /* enable FIS reception */ + tmp = readl(port->mmio + PORT_CMD); + if (enable) + writel(tmp | PORT_CMD_START, port->mmio + PORT_CMD); + else + writel(tmp & ~PORT_CMD_START, port->mmio + PORT_CMD); + + readl(port->mmio + PORT_CMD); + return (((tmp & PORT_CMD_START) == PORT_CMD_START)); +} + +/* + * Enables the port DMA engine and FIS reception. + * + * return value + * None + */ +static inline void mtip_start_port(struct mtip_port *port) +{ + /* Enable FIS reception */ + mtip_enable_fis(port, 1); + + /* Enable the DMA engine */ + mtip_enable_engine(port, 1); +} + +/* + * Deinitialize a port by disabling port interrupts, the DMA engine, + * and FIS reception. + * + * @port Pointer to the port structure + * + * return value + * None + */ +static inline void mtip_deinit_port(struct mtip_port *port) +{ + /* Disable interrupts on this port */ + writel(0, port->mmio + PORT_IRQ_MASK); + + /* Disable the DMA engine */ + mtip_enable_engine(port, 0); + + /* Disable FIS reception */ + mtip_enable_fis(port, 0); +} + +/* + * Initialize a port. + * + * This function deinitializes the port by calling mtip_deinit_port() and + * then initializes it by setting the command header and RX FIS addresses, + * clearing the SError register and any pending port interrupts before + * re-enabling the default set of port interrupts. + * + * @port Pointer to the port structure. + * + * return value + * None + */ +static void mtip_init_port(struct mtip_port *port) +{ + int i; + mtip_deinit_port(port); + + /* Program the command list base and FIS base addresses */ + if (readl(port->dd->mmio + HOST_CAP) & HOST_CAP_64) { + writel((port->command_list_dma >> 16) >> 16, + port->mmio + PORT_LST_ADDR_HI); + writel((port->rxfis_dma >> 16) >> 16, + port->mmio + PORT_FIS_ADDR_HI); + } + + writel(port->command_list_dma & 0xFFFFFFFF, + port->mmio + PORT_LST_ADDR); + writel(port->rxfis_dma & 0xFFFFFFFF, port->mmio + PORT_FIS_ADDR); + + /* Clear SError */ + writel(readl(port->mmio + PORT_SCR_ERR), port->mmio + PORT_SCR_ERR); + + /* reset the completed registers.*/ + for (i = 0; i < port->dd->slot_groups; i++) + writel(0xFFFFFFFF, port->completed[i]); + + /* Clear any pending interrupts for this port */ + writel(readl(port->mmio + PORT_IRQ_STAT), port->mmio + PORT_IRQ_STAT); + + /* Enable port interrupts */ + writel(DEF_PORT_IRQ, port->mmio + PORT_IRQ_MASK); +} + +/* + * Restart a port + * + * @port Pointer to the port data structure. + * + * return value + * None + */ +static void mtip_restart_port(struct mtip_port *port) +{ + unsigned long timeout; + + /* Disable the DMA engine */ + mtip_enable_engine(port, 0); + + /* Chip quirk: wait up to 500ms for PxCMD.CR == 0 */ + timeout = jiffies + msecs_to_jiffies(500); + while ((readl(port->mmio + PORT_CMD) & PORT_CMD_LIST_ON) + && time_before(jiffies, timeout)) + ; + + /* + * Chip quirk: escalate to hba reset if + * PxCMD.CR not clear after 500 ms + */ + if (readl(port->mmio + PORT_CMD) & PORT_CMD_LIST_ON) { + dev_warn(&port->dd->pdev->dev, + "PxCMD.CR not clear, escalating reset\n"); + + if (hba_reset_nosleep(port->dd)) + dev_err(&port->dd->pdev->dev, + "HBA reset escalation failed.\n"); + + /* 30 ms delay before com reset to quiesce chip */ + mdelay(30); + } + + dev_warn(&port->dd->pdev->dev, "Issuing COM reset\n"); + + /* Set PxSCTL.DET */ + writel(readl(port->mmio + PORT_SCR_CTL) | + 1, port->mmio + PORT_SCR_CTL); + readl(port->mmio + PORT_SCR_CTL); + + /* Wait 1 ms to quiesce chip function */ + timeout = jiffies + msecs_to_jiffies(1); + while (time_before(jiffies, timeout)) + ; + + /* Clear PxSCTL.DET */ + writel(readl(port->mmio + PORT_SCR_CTL) & ~1, + port->mmio + PORT_SCR_CTL); + readl(port->mmio + PORT_SCR_CTL); + + /* Wait 500 ms for bit 0 of PORT_SCR_STS to be set */ + timeout = jiffies + msecs_to_jiffies(500); + while (((readl(port->mmio + PORT_SCR_STAT) & 0x01) == 0) + && time_before(jiffies, timeout)) + ; + + if ((readl(port->mmio + PORT_SCR_STAT) & 0x01) == 0) + dev_warn(&port->dd->pdev->dev, + "COM reset failed\n"); + + /* Clear SError, the PxSERR.DIAG.x should be set so clear it */ + writel(readl(port->mmio + PORT_SCR_ERR), port->mmio + PORT_SCR_ERR); + + /* Enable the DMA engine */ + mtip_enable_engine(port, 1); +} + +/* + * Called periodically to see if any read/write commands are + * taking too long to complete. + * + * @data Pointer to the PORT data structure. + * + * return value + * None + */ +static void mtip_timeout_function(unsigned long int data) +{ + struct mtip_port *port = (struct mtip_port *) data; + struct host_to_dev_fis *fis; + struct mtip_cmd *command; + int tag, cmdto_cnt = 0; + unsigned int bit, group; + unsigned int num_command_slots = port->dd->slot_groups * 32; + + if (unlikely(!port)) + return; + + if (atomic_read(&port->dd->resumeflag) == true) { + mod_timer(&port->cmd_timer, + jiffies + msecs_to_jiffies(30000)); + return; + } + + for (tag = 0; tag < num_command_slots; tag++) { + /* + * Skip internal command slot as it has + * its own timeout mechanism + */ + if (tag == MTIP_TAG_INTERNAL) + continue; + + if (atomic_read(&port->commands[tag].active) && + (time_after(jiffies, port->commands[tag].comp_time))) { + group = tag >> 5; + bit = tag & 0x1F; + + command = &port->commands[tag]; + fis = (struct host_to_dev_fis *) command->command; + + dev_warn(&port->dd->pdev->dev, + "Timeout for command tag %d\n", tag); + + cmdto_cnt++; + if (cmdto_cnt == 1) + set_bit(MTIP_FLAG_EH_ACTIVE_BIT, &port->flags); + + /* + * Clear the completed bit. This should prevent + * any interrupt handlers from trying to retire + * the command. + */ + writel(1 << bit, port->completed[group]); + + /* Call the async completion callback. */ + if (likely(command->async_callback)) + command->async_callback(command->async_data, + -EIO); + command->async_callback = NULL; + command->comp_func = NULL; + + /* Unmap the DMA scatter list entries */ + dma_unmap_sg(&port->dd->pdev->dev, + command->sg, + command->scatter_ents, + command->direction); + + /* + * Clear the allocated bit and active tag for the + * command. + */ + atomic_set(&port->commands[tag].active, 0); + release_slot(port, tag); + + up(&port->cmd_slot); + } + } + + if (cmdto_cnt) { + dev_warn(&port->dd->pdev->dev, + "%d commands timed out: restarting port", + cmdto_cnt); + mtip_restart_port(port); + clear_bit(MTIP_FLAG_EH_ACTIVE_BIT, &port->flags); + wake_up_interruptible(&port->svc_wait); + } + + /* Restart the timer */ + mod_timer(&port->cmd_timer, + jiffies + msecs_to_jiffies(MTIP_TIMEOUT_CHECK_PERIOD)); +} + +/* + * IO completion function. + * + * This completion function is called by the driver ISR when a + * command that was issued by the kernel completes. It first calls the + * asynchronous completion function which normally calls back into the block + * layer passing the asynchronous callback data, then unmaps the + * scatter list associated with the completed command, and finally + * clears the allocated bit associated with the completed command. + * + * @port Pointer to the port data structure. + * @tag Tag of the command. + * @data Pointer to driver_data. + * @status Completion status. + * + * return value + * None + */ +static void mtip_async_complete(struct mtip_port *port, + int tag, + void *data, + int status) +{ + struct mtip_cmd *command; + struct driver_data *dd = data; + int cb_status = status ? -EIO : 0; + + if (unlikely(!dd) || unlikely(!port)) + return; + + command = &port->commands[tag]; + + if (unlikely(status == PORT_IRQ_TF_ERR)) { + dev_warn(&port->dd->pdev->dev, + "Command tag %d failed due to TFE\n", tag); + } + + /* Upper layer callback */ + if (likely(command->async_callback)) + command->async_callback(command->async_data, cb_status); + + command->async_callback = NULL; + command->comp_func = NULL; + + /* Unmap the DMA scatter list entries */ + dma_unmap_sg(&dd->pdev->dev, + command->sg, + command->scatter_ents, + command->direction); + + /* Clear the allocated and active bits for the command */ + atomic_set(&port->commands[tag].active, 0); + release_slot(port, tag); + + up(&port->cmd_slot); +} + +/* + * Internal command completion callback function. + * + * This function is normally called by the driver ISR when an internal + * command completed. This function signals the command completion by + * calling complete(). + * + * @port Pointer to the port data structure. + * @tag Tag of the command that has completed. + * @data Pointer to a completion structure. + * @status Completion status. + * + * return value + * None + */ +static void mtip_completion(struct mtip_port *port, + int tag, + void *data, + int status) +{ + struct mtip_cmd *command = &port->commands[tag]; + struct completion *waiting = data; + if (unlikely(status == PORT_IRQ_TF_ERR)) + dev_warn(&port->dd->pdev->dev, + "Internal command %d completed with TFE\n", tag); + + command->async_callback = NULL; + command->comp_func = NULL; + + complete(waiting); +} + +/* + * Helper function for tag logging + */ +static void print_tags(struct driver_data *dd, + char *msg, + unsigned long *tagbits) +{ + unsigned int tag, count = 0; + + for (tag = 0; tag < (dd->slot_groups) * 32; tag++) { + if (test_bit(tag, tagbits)) + count++; + } + if (count) + dev_info(&dd->pdev->dev, "%s [%i tags]\n", msg, count); +} + +/* + * Handle an error. + * + * @dd Pointer to the DRIVER_DATA structure. + * + * return value + * None + */ +static void mtip_handle_tfe(struct driver_data *dd) +{ + int group, tag, bit, reissue; + struct mtip_port *port; + struct mtip_cmd *command; + u32 completed; + struct host_to_dev_fis *fis; + unsigned long tagaccum[SLOTBITS_IN_LONGS]; + + dev_warn(&dd->pdev->dev, "Taskfile error\n"); + + port = dd->port; + + /* Stop the timer to prevent command timeouts. */ + del_timer(&port->cmd_timer); + + /* Set eh_active */ + set_bit(MTIP_FLAG_EH_ACTIVE_BIT, &port->flags); + + /* Loop through all the groups */ + for (group = 0; group < dd->slot_groups; group++) { + completed = readl(port->completed[group]); + + /* clear completed status register in the hardware.*/ + writel(completed, port->completed[group]); + + /* clear the tag accumulator */ + memset(tagaccum, 0, SLOTBITS_IN_LONGS * sizeof(long)); + + /* Process successfully completed commands */ + for (bit = 0; bit < 32 && completed; bit++) { + if (!(completed & (1<<bit))) + continue; + tag = (group << 5) + bit; + + /* Skip the internal command slot */ + if (tag == MTIP_TAG_INTERNAL) + continue; + + command = &port->commands[tag]; + if (likely(command->comp_func)) { + set_bit(tag, tagaccum); + atomic_set(&port->commands[tag].active, 0); + command->comp_func(port, + tag, + command->comp_data, + 0); + } else { + dev_err(&port->dd->pdev->dev, + "Missing completion func for tag %d", + tag); + if (mtip_check_surprise_removal(dd->pdev)) { + mtip_command_cleanup(dd); + /* don't proceed further */ + return; + } + } + } + } + print_tags(dd, "TFE tags completed:", tagaccum); + + /* Restart the port */ + mdelay(20); + mtip_restart_port(port); + + /* clear the tag accumulator */ + memset(tagaccum, 0, SLOTBITS_IN_LONGS * sizeof(long)); + + /* Loop through all the groups */ + for (group = 0; group < dd->slot_groups; group++) { + for (bit = 0; bit < 32; bit++) { + reissue = 1; + tag = (group << 5) + bit; + + /* If the active bit is set re-issue the command */ + if (atomic_read(&port->commands[tag].active) == 0) + continue; + + fis = (struct host_to_dev_fis *) + port->commands[tag].command; + + /* Should re-issue? */ + if (tag == MTIP_TAG_INTERNAL || + fis->command == ATA_CMD_SET_FEATURES) + reissue = 0; + + /* + * First check if this command has + * exceeded its retries. + */ + if (reissue && + (port->commands[tag].retries-- > 0)) { + + set_bit(tag, tagaccum); + + /* Update the timeout value. */ + port->commands[tag].comp_time = + jiffies + msecs_to_jiffies( + MTIP_NCQ_COMMAND_TIMEOUT_MS); + /* Re-issue the command. */ + mtip_issue_ncq_command(port, tag); + + continue; + } + + /* Retire a command that will not be reissued */ + dev_warn(&port->dd->pdev->dev, + "retiring tag %d\n", tag); + atomic_set(&port->commands[tag].active, 0); + + if (port->commands[tag].comp_func) + port->commands[tag].comp_func( + port, + tag, + port->commands[tag].comp_data, + PORT_IRQ_TF_ERR); + else + dev_warn(&port->dd->pdev->dev, + "Bad completion for tag %d\n", + tag); + } + } + print_tags(dd, "TFE tags reissued:", tagaccum); + + /* clear eh_active */ + clear_bit(MTIP_FLAG_EH_ACTIVE_BIT, &port->flags); + wake_up_interruptible(&port->svc_wait); + + mod_timer(&port->cmd_timer, + jiffies + msecs_to_jiffies(MTIP_TIMEOUT_CHECK_PERIOD)); +} + +/* + * Handle a set device bits interrupt + */ +static inline void mtip_process_sdbf(struct driver_data *dd) +{ + struct mtip_port *port = dd->port; + int group, tag, bit; + u32 completed; + struct mtip_cmd *command; + + /* walk all bits in all slot groups */ + for (group = 0; group < dd->slot_groups; group++) { + completed = readl(port->completed[group]); + + /* clear completed status register in the hardware.*/ + writel(completed, port->completed[group]); + + /* Process completed commands. */ + for (bit = 0; + (bit < 32) && completed; + bit++, completed >>= 1) { + if (completed & 0x01) { + tag = (group << 5) | bit; + + /* skip internal command slot. */ + if (unlikely(tag == MTIP_TAG_INTERNAL)) + continue; + + command = &port->commands[tag]; + /* make internal callback */ + if (likely(command->comp_func)) { + command->comp_func( + port, + tag, + command->comp_data, + 0); + } else { + dev_warn(&dd->pdev->dev, + "Null completion " + "for tag %d", + tag); + + if (mtip_check_surprise_removal( + dd->pdev)) { + mtip_command_cleanup(dd); + return; + } + } + } + } + } +} + +/* + * Process legacy pio and d2h interrupts + */ +static inline void mtip_process_legacy(struct driver_data *dd, u32 port_stat) +{ + struct mtip_port *port = dd->port; + struct mtip_cmd *cmd = &port->commands[MTIP_TAG_INTERNAL]; + + if (test_bit(MTIP_FLAG_IC_ACTIVE_BIT, &port->flags) && + (cmd != NULL) && !(readl(port->cmd_issue[MTIP_TAG_INTERNAL]) + & (1 << MTIP_TAG_INTERNAL))) { + if (cmd->comp_func) { + cmd->comp_func(port, + MTIP_TAG_INTERNAL, + cmd->comp_data, + 0); + return; + } + } + + dev_warn(&dd->pdev->dev, "IRQ status 0x%x ignored.\n", port_stat); + + return; +} + +/* + * Demux and handle errors + */ +static inline void mtip_process_errors(struct driver_data *dd, u32 port_stat) +{ + if (likely(port_stat & (PORT_IRQ_TF_ERR | PORT_IRQ_IF_ERR))) + mtip_handle_tfe(dd); + + if (unlikely(port_stat & PORT_IRQ_CONNECT)) { + dev_warn(&dd->pdev->dev, + "Clearing PxSERR.DIAG.x\n"); + writel((1 << 26), dd->port->mmio + PORT_SCR_ERR); + } + + if (unlikely(port_stat & PORT_IRQ_PHYRDY)) { + dev_warn(&dd->pdev->dev, + "Clearing PxSERR.DIAG.n\n"); + writel((1 << 16), dd->port->mmio + PORT_SCR_ERR); + } + + if (unlikely(port_stat & ~PORT_IRQ_HANDLED)) { + dev_warn(&dd->pdev->dev, + "Port stat errors %x unhandled\n", + (port_stat & ~PORT_IRQ_HANDLED)); + } +} + +static inline irqreturn_t mtip_handle_irq(struct driver_data *data) +{ + struct driver_data *dd = (struct driver_data *) data; + struct mtip_port *port = dd->port; + u32 hba_stat, port_stat; + int rv = IRQ_NONE; + + hba_stat = readl(dd->mmio + HOST_IRQ_STAT); + if (hba_stat) { + rv = IRQ_HANDLED; + + /* Acknowledge the interrupt status on the port.*/ + port_stat = readl(port->mmio + PORT_IRQ_STAT); + writel(port_stat, port->mmio + PORT_IRQ_STAT); + + /* Demux port status */ + if (likely(port_stat & PORT_IRQ_SDB_FIS)) + mtip_process_sdbf(dd); + + if (unlikely(port_stat & PORT_IRQ_ERR)) { + if (unlikely(mtip_check_surprise_removal(dd->pdev))) { + mtip_command_cleanup(dd); + /* don't proceed further */ + return IRQ_HANDLED; + } + + mtip_process_errors(dd, port_stat & PORT_IRQ_ERR); + } + + if (unlikely(port_stat & PORT_IRQ_LEGACY)) + mtip_process_legacy(dd, port_stat & PORT_IRQ_LEGACY); + } + + /* acknowledge interrupt */ + writel(hba_stat, dd->mmio + HOST_IRQ_STAT); + + return rv; +} + +/* + * Wrapper for mtip_handle_irq + * (ignores return code) + */ +static void mtip_tasklet(unsigned long data) +{ + mtip_handle_irq((struct driver_data *) data); +} + +/* + * HBA interrupt subroutine. + * + * @irq IRQ number. + * @instance Pointer to the driver data structure. + * + * return value + * IRQ_HANDLED A HBA interrupt was pending and handled. + * IRQ_NONE This interrupt was not for the HBA. + */ +static irqreturn_t mtip_irq_handler(int irq, void *instance) +{ + struct driver_data *dd = instance; + tasklet_schedule(&dd->tasklet); + return IRQ_HANDLED; +} + +static void mtip_issue_non_ncq_command(struct mtip_port *port, int tag) +{ + atomic_set(&port->commands[tag].active, 1); + writel(1 << MTIP_TAG_BIT(tag), + port->cmd_issue[MTIP_TAG_INDEX(tag)]); +} + +/* + * Wait for port to quiesce + * + * @port Pointer to port data structure + * @timeout Max duration to wait (ms) + * + * return value + * 0 Success + * -EBUSY Commands still active + */ +static int mtip_quiesce_io(struct mtip_port *port, unsigned long timeout) +{ + unsigned long to; + unsigned int n; + unsigned int active = 1; + + to = jiffies + msecs_to_jiffies(timeout); + do { + if (test_bit(MTIP_FLAG_SVC_THD_ACTIVE_BIT, &port->flags) && + test_bit(MTIP_FLAG_ISSUE_CMDS_BIT, &port->flags)) { + msleep(20); + continue; /* svc thd is actively issuing commands */ + } + /* + * Ignore s_active bit 0 of array element 0. + * This bit will always be set + */ + active = readl(port->s_active[0]) & 0xFFFFFFFE; + for (n = 1; n < port->dd->slot_groups; n++) + active |= readl(port->s_active[n]); + + if (!active) + break; + + msleep(20); + } while (time_before(jiffies, to)); + + return active ? -EBUSY : 0; +} + +/* + * Execute an internal command and wait for the completion. + * + * @port Pointer to the port data structure. + * @fis Pointer to the FIS that describes the command. + * @fis_len Length in WORDS of the FIS. + * @buffer DMA accessible for command data. + * @buf_len Length, in bytes, of the data buffer. + * @opts Command header options, excluding the FIS length + * and the number of PRD entries. + * @timeout Time in ms to wait for the command to complete. + * + * return value + * 0 Command completed successfully. + * -EFAULT The buffer address is not correctly aligned. + * -EBUSY Internal command or other IO in progress. + * -EAGAIN Time out waiting for command to complete. + */ +static int mtip_exec_internal_command(struct mtip_port *port, + void *fis, + int fis_len, + dma_addr_t buffer, + int buf_len, + u32 opts, + gfp_t atomic, + unsigned long timeout) +{ + struct mtip_cmd_sg *command_sg; + DECLARE_COMPLETION_ONSTACK(wait); + int rv = 0; + struct mtip_cmd *int_cmd = &port->commands[MTIP_TAG_INTERNAL]; + + /* Make sure the buffer is 8 byte aligned. This is asic specific. */ + if (buffer & 0x00000007) { + dev_err(&port->dd->pdev->dev, + "SG buffer is not 8 byte aligned\n"); + return -EFAULT; + } + + /* Only one internal command should be running at a time */ + if (test_and_set_bit(MTIP_TAG_INTERNAL, port->allocated)) { + dev_warn(&port->dd->pdev->dev, + "Internal command already active\n"); + return -EBUSY; + } + set_bit(MTIP_FLAG_IC_ACTIVE_BIT, &port->flags); + + if (atomic == GFP_KERNEL) { + /* wait for io to complete if non atomic */ + if (mtip_quiesce_io(port, 5000) < 0) { + dev_warn(&port->dd->pdev->dev, + "Failed to quiesce IO\n"); + release_slot(port, MTIP_TAG_INTERNAL); + clear_bit(MTIP_FLAG_IC_ACTIVE_BIT, &port->flags); + wake_up_interruptible(&port->svc_wait); + return -EBUSY; + } + + /* Set the completion function and data for the command. */ + int_cmd->comp_data = &wait; + int_cmd->comp_func = mtip_completion; + + } else { + /* Clear completion - we're going to poll */ + int_cmd->comp_data = NULL; + int_cmd->comp_func = NULL; + } + + /* Copy the command to the command table */ + memcpy(int_cmd->command, fis, fis_len*4); + + /* Populate the SG list */ + int_cmd->command_header->opts = + __force_bit2int cpu_to_le32(opts | fis_len); + if (buf_len) { + command_sg = int_cmd->command + AHCI_CMD_TBL_HDR_SZ; + + command_sg->info = + __force_bit2int cpu_to_le32((buf_len-1) & 0x3FFFFF); + command_sg->dba = + __force_bit2int cpu_to_le32(buffer & 0xFFFFFFFF); + command_sg->dba_upper = + __force_bit2int cpu_to_le32((buffer >> 16) >> 16); + + int_cmd->command_header->opts |= + __force_bit2int cpu_to_le32((1 << 16)); + } + + /* Populate the command header */ + int_cmd->command_header->byte_count = 0; + + /* Issue the command to the hardware */ + mtip_issue_non_ncq_command(port, MTIP_TAG_INTERNAL); + + /* Poll if atomic, wait_for_completion otherwise */ + if (atomic == GFP_KERNEL) { + /* Wait for the command to complete or timeout. */ + if (wait_for_completion_timeout( + &wait, + msecs_to_jiffies(timeout)) == 0) { + dev_err(&port->dd->pdev->dev, + "Internal command did not complete [%d] " + "within timeout of %lu ms\n", + atomic, timeout); + rv = -EAGAIN; + } + + if (readl(port->cmd_issue[MTIP_TAG_INTERNAL]) + & (1 << MTIP_TAG_INTERNAL)) { + dev_warn(&port->dd->pdev->dev, + "Retiring internal command but CI is 1.\n"); + } + + } else { + /* Spin for <timeout> checking if command still outstanding */ + timeout = jiffies + msecs_to_jiffies(timeout); + + while ((readl( + port->cmd_issue[MTIP_TAG_INTERNAL]) + & (1 << MTIP_TAG_INTERNAL)) + && time_before(jiffies, timeout)) + ; + + if (readl(port->cmd_issue[MTIP_TAG_INTERNAL]) + & (1 << MTIP_TAG_INTERNAL)) { + dev_err(&port->dd->pdev->dev, + "Internal command did not complete [%d]\n", + atomic); + rv = -EAGAIN; + } + } + + /* Clear the allocated and active bits for the internal command. */ + atomic_set(&int_cmd->active, 0); + release_slot(port, MTIP_TAG_INTERNAL); + clear_bit(MTIP_FLAG_IC_ACTIVE_BIT, &port->flags); + wake_up_interruptible(&port->svc_wait); + + return rv; +} + +/* + * Byte-swap ATA ID strings. + * + * ATA identify data contains strings in byte-swapped 16-bit words. + * They must be swapped (on all architectures) to be usable as C strings. + * This function swaps bytes in-place. + * + * @buf The buffer location of the string + * @len The number of bytes to swap + * + * return value + * None + */ +static inline void ata_swap_string(u16 *buf, unsigned int len) +{ + int i; + for (i = 0; i < (len/2); i++) + be16_to_cpus(&buf[i]); +} + +/* + * Request the device identity information. + * + * If a user space buffer is not specified, i.e. is NULL, the + * identify information is still read from the drive and placed + * into the identify data buffer (@e port->identify) in the + * port data structure. + * When the identify buffer contains valid identify information @e + * port->identify_valid is non-zero. + * + * @port Pointer to the port structure. + * @user_buffer A user space buffer where the identify data should be + * copied. + * + * return value + * 0 Command completed successfully. + * -EFAULT An error occurred while coping data to the user buffer. + * -1 Command failed. + */ +static int mtip_get_identify(struct mtip_port *port, void __user *user_buffer) +{ + int rv = 0; + struct host_to_dev_fis fis; + + /* Build the FIS. */ + memset(&fis, 0, sizeof(struct host_to_dev_fis)); + fis.type = 0x27; + fis.opts = 1 << 7; + fis.command = ATA_CMD_ID_ATA; + + /* Set the identify information as invalid. */ + port->identify_valid = 0; + + /* Clear the identify information. */ + memset(port->identify, 0, sizeof(u16) * ATA_ID_WORDS); + + /* Execute the command. */ + if (mtip_exec_internal_command(port, + &fis, + 5, + port->identify_dma, + sizeof(u16) * ATA_ID_WORDS, + 0, + GFP_KERNEL, + MTIP_INTERNAL_COMMAND_TIMEOUT_MS) + < 0) { + rv = -1; + goto out; + } + + /* + * Perform any necessary byte-swapping. Yes, the kernel does in fact + * perform field-sensitive swapping on the string fields. + * See the kernel use of ata_id_string() for proof of this. + */ +#ifdef __LITTLE_ENDIAN + ata_swap_string(port->identify + 27, 40); /* model string*/ + ata_swap_string(port->identify + 23, 8); /* firmware string*/ + ata_swap_string(port->identify + 10, 20); /* serial# string*/ +#else + { + int i; + for (i = 0; i < ATA_ID_WORDS; i++) + port->identify[i] = le16_to_cpu(port->identify[i]); + } +#endif + + /* Set the identify buffer as valid. */ + port->identify_valid = 1; + + if (user_buffer) { + if (copy_to_user( + user_buffer, + port->identify, + ATA_ID_WORDS * sizeof(u16))) { + rv = -EFAULT; + goto out; + } + } + +out: + return rv; +} + +/* + * Issue a standby immediate command to the device. + * + * @port Pointer to the port structure. + * + * return value + * 0 Command was executed successfully. + * -1 An error occurred while executing the command. + */ +static int mtip_standby_immediate(struct mtip_port *port) +{ + int rv; + struct host_to_dev_fis fis; + + /* Build the FIS. */ + memset(&fis, 0, sizeof(struct host_to_dev_fis)); + fis.type = 0x27; + fis.opts = 1 << 7; + fis.command = ATA_CMD_STANDBYNOW1; + + /* Execute the command. Use a 15-second timeout for large drives. */ + rv = mtip_exec_internal_command(port, + &fis, + 5, + 0, + 0, + 0, + GFP_KERNEL, + 15000); + + return rv; +} + +/* + * Get the drive capacity. + * + * @dd Pointer to the device data structure. + * @sectors Pointer to the variable that will receive the sector count. + * + * return value + * 1 Capacity was returned successfully. + * 0 The identify information is invalid. + */ +static bool mtip_hw_get_capacity(struct driver_data *dd, sector_t *sectors) +{ + struct mtip_port *port = dd->port; + u64 total, raw0, raw1, raw2, raw3; + raw0 = port->identify[100]; + raw1 = port->identify[101]; + raw2 = port->identify[102]; + raw3 = port->identify[103]; + total = raw0 | raw1<<16 | raw2<<32 | raw3<<48; + *sectors = total; + return (bool) !!port->identify_valid; +} + +/* + * Reset the HBA. + * + * Resets the HBA by setting the HBA Reset bit in the Global + * HBA Control register. After setting the HBA Reset bit the + * function waits for 1 second before reading the HBA Reset + * bit to make sure it has cleared. If HBA Reset is not clear + * an error is returned. Cannot be used in non-blockable + * context. + * + * @dd Pointer to the driver data structure. + * + * return value + * 0 The reset was successful. + * -1 The HBA Reset bit did not clear. + */ +static int mtip_hba_reset(struct driver_data *dd) +{ + mtip_deinit_port(dd->port); + + /* Set the reset bit */ + writel(HOST_RESET, dd->mmio + HOST_CTL); + + /* Flush */ + readl(dd->mmio + HOST_CTL); + + /* Wait for reset to clear */ + ssleep(1); + + /* Check the bit has cleared */ + if (readl(dd->mmio + HOST_CTL) & HOST_RESET) { + dev_err(&dd->pdev->dev, + "Reset bit did not clear.\n"); + return -1; + } + + return 0; +} + +/* + * Display the identify command data. + * + * @port Pointer to the port data structure. + * + * return value + * None + */ +static void mtip_dump_identify(struct mtip_port *port) +{ + sector_t sectors; + unsigned short revid; + char cbuf[42]; + + if (!port->identify_valid) + return; + + strlcpy(cbuf, (char *)(port->identify+10), 21); + dev_info(&port->dd->pdev->dev, + "Serial No.: %s\n", cbuf); + + strlcpy(cbuf, (char *)(port->identify+23), 9); + dev_info(&port->dd->pdev->dev, + "Firmware Ver.: %s\n", cbuf); + + strlcpy(cbuf, (char *)(port->identify+27), 41); + dev_info(&port->dd->pdev->dev, "Model: %s\n", cbuf); + + if (mtip_hw_get_capacity(port->dd, §ors)) + dev_info(&port->dd->pdev->dev, + "Capacity: %llu sectors (%llu MB)\n", + (u64)sectors, + ((u64)sectors) * ATA_SECT_SIZE >> 20); + + pci_read_config_word(port->dd->pdev, PCI_REVISION_ID, &revid); + switch (revid & 0xFF) { + case 0x1: + strlcpy(cbuf, "A0", 3); + break; + case 0x3: + strlcpy(cbuf, "A2", 3); + break; + default: + strlcpy(cbuf, "?", 2); + break; + } + dev_info(&port->dd->pdev->dev, + "Card Type: %s\n", cbuf); +} + +/* + * Map the commands scatter list into the command table. + * + * @command Pointer to the command. + * @nents Number of scatter list entries. + * + * return value + * None + */ +static inline void fill_command_sg(struct driver_data *dd, + struct mtip_cmd *command, + int nents) +{ + int n; + unsigned int dma_len; + struct mtip_cmd_sg *command_sg; + struct scatterlist *sg = command->sg; + + command_sg = command->command + AHCI_CMD_TBL_HDR_SZ; + + for (n = 0; n < nents; n++) { + dma_len = sg_dma_len(sg); + if (dma_len > 0x400000) + dev_err(&dd->pdev->dev, + "DMA segment length truncated\n"); + command_sg->info = __force_bit2int + cpu_to_le32((dma_len-1) & 0x3FFFFF); + command_sg->dba = __force_bit2int + cpu_to_le32(sg_dma_address(sg)); + command_sg->dba_upper = __force_bit2int + cpu_to_le32((sg_dma_address(sg) >> 16) >> 16); + command_sg++; + sg++; + } +} + +/* + * @brief Execute a drive command. + * + * return value 0 The command completed successfully. + * return value -1 An error occurred while executing the command. + */ +static int exec_drive_task(struct mtip_port *port, u8 *command) +{ + struct host_to_dev_fis fis; + struct host_to_dev_fis *reply = (port->rxfis + RX_FIS_D2H_REG); + + /* Build the FIS. */ + memset(&fis, 0, sizeof(struct host_to_dev_fis)); + fis.type = 0x27; + fis.opts = 1 << 7; + fis.command = command[0]; + fis.features = command[1]; + fis.sect_count = command[2]; + fis.sector = command[3]; + fis.cyl_low = command[4]; + fis.cyl_hi = command[5]; + fis.device = command[6] & ~0x10; /* Clear the dev bit*/ + + + dbg_printk(MTIP_DRV_NAME "%s: User Command: cmd %x, feat %x, " + "nsect %x, sect %x, lcyl %x, " + "hcyl %x, sel %x\n", + __func__, + command[0], + command[1], + command[2], + command[3], + command[4], + command[5], + command[6]); + + /* Execute the command. */ + if (mtip_exec_internal_command(port, + &fis, + 5, + 0, + 0, + 0, + GFP_KERNEL, + MTIP_IOCTL_COMMAND_TIMEOUT_MS) < 0) { + return -1; + } + + command[0] = reply->command; /* Status*/ + command[1] = reply->features; /* Error*/ + command[4] = reply->cyl_low; + command[5] = reply->cyl_hi; + + dbg_printk(MTIP_DRV_NAME "%s: Completion Status: stat %x, " + "err %x , cyl_lo %x cyl_hi %x\n", + __func__, + command[0], + command[1], + command[4], + command[5]); + + return 0; +} + +/* + * @brief Execute a drive command. + * + * @param port Pointer to the port data structure. + * @param command Pointer to the user specified command parameters. + * @param user_buffer Pointer to the user space buffer where read sector + * data should be copied. + * + * return value 0 The command completed successfully. + * return value -EFAULT An error occurred while copying the completion + * data to the user space buffer. + * return value -1 An error occurred while executing the command. + */ +static int exec_drive_command(struct mtip_port *port, u8 *command, + void __user *user_buffer) +{ + struct host_to_dev_fis fis; + struct host_to_dev_fis *reply = (port->rxfis + RX_FIS_D2H_REG); + + /* Build the FIS. */ + memset(&fis, 0, sizeof(struct host_to_dev_fis)); + fis.type = 0x27; + fis.opts = 1 << 7; + fis.command = command[0]; + fis.features = command[2]; + fis.sect_count = command[3]; + if (fis.command == ATA_CMD_SMART) { + fis.sector = command[1]; + fis.cyl_low = 0x4F; + fis.cyl_hi = 0xC2; + } + + dbg_printk(MTIP_DRV_NAME + "%s: User Command: cmd %x, sect %x, " + "feat %x, sectcnt %x\n", + __func__, + command[0], + command[1], + command[2], + command[3]); + + memset(port->sector_buffer, 0x00, ATA_SECT_SIZE); + + /* Execute the command. */ + if (mtip_exec_internal_command(port, + &fis, + 5, + port->sector_buffer_dma, + (command[3] != 0) ? ATA_SECT_SIZE : 0, + 0, + GFP_KERNEL, + MTIP_IOCTL_COMMAND_TIMEOUT_MS) + < 0) { + return -1; + } + + /* Collect the completion status. */ + command[0] = reply->command; /* Status*/ + command[1] = reply->features; /* Error*/ + command[2] = command[3]; + + dbg_printk(MTIP_DRV_NAME + "%s: Completion Status: stat %x, " + "err %x, cmd %x\n", + __func__, + command[0], + command[1], + command[2]); + + if (user_buffer && command[3]) { + if (copy_to_user(user_buffer, + port->sector_buffer, + ATA_SECT_SIZE * command[3])) { + return -EFAULT; + } + } + + return 0; +} + +/* + * Indicates whether a command has a single sector payload. + * + * @command passed to the device to perform the certain event. + * @features passed to the device to perform the certain event. + * + * return value + * 1 command is one that always has a single sector payload, + * regardless of the value in the Sector Count field. + * 0 otherwise + * + */ +static unsigned int implicit_sector(unsigned char command, + unsigned char features) +{ + unsigned int rv = 0; + + /* list of commands that have an implicit sector count of 1 */ + switch (command) { + case ATA_CMD_SEC_SET_PASS: + case ATA_CMD_SEC_UNLOCK: + case ATA_CMD_SEC_ERASE_PREP: + case ATA_CMD_SEC_ERASE_UNIT: + case ATA_CMD_SEC_FREEZE_LOCK: + case ATA_CMD_SEC_DISABLE_PASS: + case ATA_CMD_PMP_READ: + case ATA_CMD_PMP_WRITE: + rv = 1; + break; + case ATA_CMD_SET_MAX: + if (features == ATA_SET_MAX_UNLOCK) + rv = 1; + break; + case ATA_CMD_SMART: + if ((features == ATA_SMART_READ_VALUES) || + (features == ATA_SMART_READ_THRESHOLDS)) + rv = 1; + break; + case ATA_CMD_CONF_OVERLAY: + if ((features == ATA_DCO_IDENTIFY) || + (features == ATA_DCO_SET)) + rv = 1; + break; + } + return rv; +} + +/* + * Executes a taskfile + * See ide_taskfile_ioctl() for derivation + */ +static int exec_drive_taskfile(struct driver_data *dd, + void __user *buf, + ide_task_request_t *req_task, + int outtotal) +{ + struct host_to_dev_fis fis; + struct host_to_dev_fis *reply; + u8 *outbuf = NULL; + u8 *inbuf = NULL; + dma_addr_t outbuf_dma = 0; + dma_addr_t inbuf_dma = 0; + dma_addr_t dma_buffer = 0; + int err = 0; + unsigned int taskin = 0; + unsigned int taskout = 0; + u8 nsect = 0; + unsigned int timeout = MTIP_IOCTL_COMMAND_TIMEOUT_MS; + unsigned int force_single_sector; + unsigned int transfer_size; + unsigned long task_file_data; + int intotal = outtotal + req_task->out_size; + + taskout = req_task->out_size; + taskin = req_task->in_size; + /* 130560 = 512 * 0xFF*/ + if (taskin > 130560 || taskout > 130560) { + err = -EINVAL; + goto abort; + } + + if (taskout) { + outbuf = kzalloc(taskout, GFP_KERNEL); + if (outbuf == NULL) { + err = -ENOMEM; + goto abort; + } + if (copy_from_user(outbuf, buf + outtotal, taskout)) { + err = -EFAULT; + goto abort; + } + outbuf_dma = pci_map_single(dd->pdev, + outbuf, + taskout, + DMA_TO_DEVICE); + if (outbuf_dma == 0) { + err = -ENOMEM; + goto abort; + } + dma_buffer = outbuf_dma; + } + + if (taskin) { + inbuf = kzalloc(taskin, GFP_KERNEL); + if (inbuf == NULL) { + err = -ENOMEM; + goto abort; + } + + if (copy_from_user(inbuf, buf + intotal, taskin)) { + err = -EFAULT; + goto abort; + } + inbuf_dma = pci_map_single(dd->pdev, + inbuf, + taskin, DMA_FROM_DEVICE); + if (inbuf_dma == 0) { + err = -ENOMEM; + goto abort; + } + dma_buffer = inbuf_dma; + } + + /* only supports PIO and non-data commands from this ioctl. */ + switch (req_task->data_phase) { + case TASKFILE_OUT: + nsect = taskout / ATA_SECT_SIZE; + reply = (dd->port->rxfis + RX_FIS_PIO_SETUP); + break; + case TASKFILE_IN: + reply = (dd->port->rxfis + RX_FIS_PIO_SETUP); + break; + case TASKFILE_NO_DATA: + reply = (dd->port->rxfis + RX_FIS_D2H_REG); + break; + default: + err = -EINVAL; + goto abort; + } + + /* Build the FIS. */ + memset(&fis, 0, sizeof(struct host_to_dev_fis)); + + fis.type = 0x27; + fis.opts = 1 << 7; + fis.command = req_task->io_ports[7]; + fis.features = req_task->io_ports[1]; + fis.sect_count = req_task->io_ports[2]; + fis.lba_low = req_task->io_ports[3]; + fis.lba_mid = req_task->io_ports[4]; + fis.lba_hi = req_task->io_ports[5]; + /* Clear the dev bit*/ + fis.device = req_task->io_ports[6] & ~0x10; + + if ((req_task->in_flags.all == 0) && (req_task->out_flags.all & 1)) { + req_task->in_flags.all = + IDE_TASKFILE_STD_IN_FLAGS | + (IDE_HOB_STD_IN_FLAGS << 8); + fis.lba_low_ex = req_task->hob_ports[3]; + fis.lba_mid_ex = req_task->hob_ports[4]; + fis.lba_hi_ex = req_task->hob_ports[5]; + fis.features_ex = req_task->hob_ports[1]; + fis.sect_cnt_ex = req_task->hob_ports[2]; + + } else { + req_task->in_flags.all = IDE_TASKFILE_STD_IN_FLAGS; + } + + force_single_sector = implicit_sector(fis.command, fis.features); + + if ((taskin || taskout) && (!fis.sect_count)) { + if (nsect) + fis.sect_count = nsect; + else { + if (!force_single_sector) { + dev_warn(&dd->pdev->dev, + "data movement but " + "sect_count is 0\n"); + err = -EINVAL; + goto abort; + } + } + } + + dbg_printk(MTIP_DRV_NAME + "taskfile: cmd %x, feat %x, nsect %x," + " sect/lbal %x, lcyl/lbam %x, hcyl/lbah %x," + " head/dev %x\n", + fis.command, + fis.features, + fis.sect_count, + fis.lba_low, + fis.lba_mid, + fis.lba_hi, + fis.device); + + switch (fis.command) { + case ATA_CMD_DOWNLOAD_MICRO: + /* Change timeout for Download Microcode to 60 seconds.*/ + timeout = 60000; + break; + case ATA_CMD_SEC_ERASE_UNIT: + /* Change timeout for Security Erase Unit to 4 minutes.*/ + timeout = 240000; + break; + case ATA_CMD_STANDBYNOW1: + /* Change timeout for standby immediate to 10 seconds.*/ + timeout = 10000; + break; + case 0xF7: + case 0xFA: + /* Change timeout for vendor unique command to 10 secs */ + timeout = 10000; + break; + case ATA_CMD_SMART: + /* Change timeout for vendor unique command to 10 secs */ + timeout = 10000; + break; + default: + timeout = MTIP_IOCTL_COMMAND_TIMEOUT_MS; + break; + } + + /* Determine the correct transfer size.*/ + if (force_single_sector) + transfer_size = ATA_SECT_SIZE; + else + transfer_size = ATA_SECT_SIZE * fis.sect_count; + + /* Execute the command.*/ + if (mtip_exec_internal_command(dd->port, + &fis, + 5, + dma_buffer, + transfer_size, + 0, + GFP_KERNEL, + timeout) < 0) { + err = -EIO; + goto abort; + } + + task_file_data = readl(dd->port->mmio+PORT_TFDATA); + + if ((req_task->data_phase == TASKFILE_IN) && !(task_file_data & 1)) { + reply = dd->port->rxfis + RX_FIS_PIO_SETUP; + req_task->io_ports[7] = reply->control; + } else { + reply = dd->port->rxfis + RX_FIS_D2H_REG; + req_task->io_ports[7] = reply->command; + } + + /* reclaim the DMA buffers.*/ + if (inbuf_dma) + pci_unmap_single(dd->pdev, inbuf_dma, + taskin, DMA_FROM_DEVICE); + if (outbuf_dma) + pci_unmap_single(dd->pdev, outbuf_dma, + taskout, DMA_TO_DEVICE); + inbuf_dma = 0; + outbuf_dma = 0; + + /* return the ATA registers to the caller.*/ + req_task->io_ports[1] = reply->features; + req_task->io_ports[2] = reply->sect_count; + req_task->io_ports[3] = reply->lba_low; + req_task->io_ports[4] = reply->lba_mid; + req_task->io_ports[5] = reply->lba_hi; + req_task->io_ports[6] = reply->device; + + if (req_task->out_flags.all & 1) { + + req_task->hob_ports[3] = reply->lba_low_ex; + req_task->hob_ports[4] = reply->lba_mid_ex; + req_task->hob_ports[5] = reply->lba_hi_ex; + req_task->hob_ports[1] = reply->features_ex; + req_task->hob_ports[2] = reply->sect_cnt_ex; + } + + /* Com rest after secure erase or lowlevel format */ + if (((fis.command == ATA_CMD_SEC_ERASE_UNIT) || + ((fis.command == 0xFC) && + (fis.features == 0x27 || fis.features == 0x72 || + fis.features == 0x62 || fis.features == 0x26))) && + !(reply->command & 1)) { + mtip_restart_port(dd->port); + } + + dbg_printk(MTIP_DRV_NAME + "%s: Completion: stat %x," + "err %x, sect_cnt %x, lbalo %x," + "lbamid %x, lbahi %x, dev %x\n", + __func__, + req_task->io_ports[7], + req_task->io_ports[1], + req_task->io_ports[2], + req_task->io_ports[3], + req_task->io_ports[4], + req_task->io_ports[5], + req_task->io_ports[6]); + + if (taskout) { + if (copy_to_user(buf + outtotal, outbuf, taskout)) { + err = -EFAULT; + goto abort; + } + } + if (taskin) { + if (copy_to_user(buf + intotal, inbuf, taskin)) { + err = -EFAULT; + goto abort; + } + } +abort: + if (inbuf_dma) + pci_unmap_single(dd->pdev, inbuf_dma, + taskin, DMA_FROM_DEVICE); + if (outbuf_dma) + pci_unmap_single(dd->pdev, outbuf_dma, + taskout, DMA_TO_DEVICE); + kfree(outbuf); + kfree(inbuf); + + return err; +} + +/* + * Handle IOCTL calls from the Block Layer. + * + * This function is called by the Block Layer when it receives an IOCTL + * command that it does not understand. If the IOCTL command is not supported + * this function returns -ENOTTY. + * + * @dd Pointer to the driver data structure. + * @cmd IOCTL command passed from the Block Layer. + * @arg IOCTL argument passed from the Block Layer. + * + * return value + * 0 The IOCTL completed successfully. + * -ENOTTY The specified command is not supported. + * -EFAULT An error occurred copying data to a user space buffer. + * -EIO An error occurred while executing the command. + */ +static int mtip_hw_ioctl(struct driver_data *dd, unsigned int cmd, + unsigned long arg) +{ + switch (cmd) { + case HDIO_GET_IDENTITY: + if (mtip_get_identify(dd->port, (void __user *) arg) < 0) { + dev_warn(&dd->pdev->dev, + "Unable to read identity\n"); + return -EIO; + } + + break; + case HDIO_DRIVE_CMD: + { + u8 drive_command[4]; + + /* Copy the user command info to our buffer. */ + if (copy_from_user(drive_command, + (void __user *) arg, + sizeof(drive_command))) + return -EFAULT; + + /* Execute the drive command. */ + if (exec_drive_command(dd->port, + drive_command, + (void __user *) (arg+4))) + return -EIO; + + /* Copy the status back to the users buffer. */ + if (copy_to_user((void __user *) arg, + drive_command, + sizeof(drive_command))) + return -EFAULT; + + break; + } + case HDIO_DRIVE_TASK: + { + u8 drive_command[7]; + + /* Copy the user command info to our buffer. */ + if (copy_from_user(drive_command, + (void __user *) arg, + sizeof(drive_command))) + return -EFAULT; + + /* Execute the drive command. */ + if (exec_drive_task(dd->port, drive_command)) + return -EIO; + + /* Copy the status back to the users buffer. */ + if (copy_to_user((void __user *) arg, + drive_command, + sizeof(drive_command))) + return -EFAULT; + + break; + } + case HDIO_DRIVE_TASKFILE: { + ide_task_request_t req_task; + int ret, outtotal; + + if (copy_from_user(&req_task, (void __user *) arg, + sizeof(req_task))) + return -EFAULT; + + outtotal = sizeof(req_task); + + ret = exec_drive_taskfile(dd, (void __user *) arg, + &req_task, outtotal); + + if (copy_to_user((void __user *) arg, &req_task, + sizeof(req_task))) + return -EFAULT; + + return ret; + } + + default: + return -EINVAL; + } + return 0; +} + +/* + * Submit an IO to the hw + * + * This function is called by the block layer to issue an io + * to the device. Upon completion, the callback function will + * be called with the data parameter passed as the callback data. + * + * @dd Pointer to the driver data structure. + * @start First sector to read. + * @nsect Number of sectors to read. + * @nents Number of entries in scatter list for the read command. + * @tag The tag of this read command. + * @callback Pointer to the function that should be called + * when the read completes. + * @data Callback data passed to the callback function + * when the read completes. + * @dir Direction (read or write) + * + * return value + * None + */ +static void mtip_hw_submit_io(struct driver_data *dd, sector_t start, + int nsect, int nents, int tag, void *callback, + void *data, int dir) +{ + struct host_to_dev_fis *fis; + struct mtip_port *port = dd->port; + struct mtip_cmd *command = &port->commands[tag]; + + /* Map the scatter list for DMA access */ + if (dir == READ) + nents = dma_map_sg(&dd->pdev->dev, command->sg, + nents, DMA_FROM_DEVICE); + else + nents = dma_map_sg(&dd->pdev->dev, command->sg, + nents, DMA_TO_DEVICE); + + command->scatter_ents = nents; + + /* + * The number of retries for this command before it is + * reported as a failure to the upper layers. + */ + command->retries = MTIP_MAX_RETRIES; + + /* Fill out fis */ + fis = command->command; + fis->type = 0x27; + fis->opts = 1 << 7; + fis->command = + (dir == READ ? ATA_CMD_FPDMA_READ : ATA_CMD_FPDMA_WRITE); + *((unsigned int *) &fis->lba_low) = (start & 0xFFFFFF); + *((unsigned int *) &fis->lba_low_ex) = ((start >> 24) & 0xFFFFFF); + fis->device = 1 << 6; + fis->features = nsect & 0xFF; + fis->features_ex = (nsect >> 8) & 0xFF; + fis->sect_count = ((tag << 3) | (tag >> 5)); + fis->sect_cnt_ex = 0; + fis->control = 0; + fis->res2 = 0; + fis->res3 = 0; + fill_command_sg(dd, command, nents); + + /* Populate the command header */ + command->command_header->opts = + __force_bit2int cpu_to_le32( + (nents << 16) | 5 | AHCI_CMD_PREFETCH); + command->command_header->byte_count = 0; + + /* + * Set the completion function and data for the command + * within this layer. + */ + command->comp_data = dd; + command->comp_func = mtip_async_complete; + command->direction = (dir == READ ? DMA_FROM_DEVICE : DMA_TO_DEVICE); + + /* + * Set the completion function and data for the command passed + * from the upper layer. + */ + command->async_data = data; + command->async_callback = callback; + + /* + * To prevent this command from being issued + * if an internal command is in progress or error handling is active. + */ + if (unlikely(test_bit(MTIP_FLAG_IC_ACTIVE_BIT, &port->flags) || + test_bit(MTIP_FLAG_EH_ACTIVE_BIT, &port->flags))) { + set_bit(tag, port->cmds_to_issue); + set_bit(MTIP_FLAG_ISSUE_CMDS_BIT, &port->flags); + return; + } + + /* Issue the command to the hardware */ + mtip_issue_ncq_command(port, tag); + + /* Set the command's timeout value.*/ + port->commands[tag].comp_time = jiffies + msecs_to_jiffies( + MTIP_NCQ_COMMAND_TIMEOUT_MS); +} + +/* + * Release a command slot. + * + * @dd Pointer to the driver data structure. + * @tag Slot tag + * + * return value + * None + */ +static void mtip_hw_release_scatterlist(struct driver_data *dd, int tag) +{ + release_slot(dd->port, tag); +} + +/* + * Obtain a command slot and return its associated scatter list. + * + * @dd Pointer to the driver data structure. + * @tag Pointer to an int that will receive the allocated command + * slot tag. + * + * return value + * Pointer to the scatter list for the allocated command slot + * or NULL if no command slots are available. + */ +static struct scatterlist *mtip_hw_get_scatterlist(struct driver_data *dd, + int *tag) +{ + /* + * It is possible that, even with this semaphore, a thread + * may think that no command slots are available. Therefore, we + * need to make an attempt to get_slot(). + */ + down(&dd->port->cmd_slot); + *tag = get_slot(dd->port); + + if (unlikely(*tag < 0)) + return NULL; + + return dd->port->commands[*tag].sg; +} + +/* + * Sysfs register/status dump. + * + * @dev Pointer to the device structure, passed by the kernrel. + * @attr Pointer to the device_attribute structure passed by the kernel. + * @buf Pointer to the char buffer that will receive the stats info. + * + * return value + * The size, in bytes, of the data copied into buf. + */ +static ssize_t hw_show_registers(struct device *dev, + struct device_attribute *attr, + char *buf) +{ + u32 group_allocated; + struct driver_data *dd = dev_to_disk(dev)->private_data; + int size = 0; + int n; + + size += sprintf(&buf[size], "%s:\ns_active:\n", __func__); + + for (n = 0; n < dd->slot_groups; n++) + size += sprintf(&buf[size], "0x%08x\n", + readl(dd->port->s_active[n])); + + size += sprintf(&buf[size], "Command Issue:\n"); + + for (n = 0; n < dd->slot_groups; n++) + size += sprintf(&buf[size], "0x%08x\n", + readl(dd->port->cmd_issue[n])); + + size += sprintf(&buf[size], "Allocated:\n"); + + for (n = 0; n < dd->slot_groups; n++) { + if (sizeof(long) > sizeof(u32)) + group_allocated = + dd->port->allocated[n/2] >> (32*(n&1)); + else + group_allocated = dd->port->allocated[n]; + size += sprintf(&buf[size], "0x%08x\n", + group_allocated); + } + + size += sprintf(&buf[size], "completed:\n"); + + for (n = 0; n < dd->slot_groups; n++) + size += sprintf(&buf[size], "0x%08x\n", + readl(dd->port->completed[n])); + + size += sprintf(&buf[size], "PORT_IRQ_STAT 0x%08x\n", + readl(dd->port->mmio + PORT_IRQ_STAT)); + size += sprintf(&buf[size], "HOST_IRQ_STAT 0x%08x\n", + readl(dd->mmio + HOST_IRQ_STAT)); + + return size; +} +static DEVICE_ATTR(registers, S_IRUGO, hw_show_registers, NULL); + +/* + * Create the sysfs related attributes. + * + * @dd Pointer to the driver data structure. + * @kobj Pointer to the kobj for the block device. + * + * return value + * 0 Operation completed successfully. + * -EINVAL Invalid parameter. + */ +static int mtip_hw_sysfs_init(struct driver_data *dd, struct kobject *kobj) +{ + if (!kobj || !dd) + return -EINVAL; + + if (sysfs_create_file(kobj, &dev_attr_registers.attr)) + dev_warn(&dd->pdev->dev, + "Error creating registers sysfs entry\n"); + return 0; +} + +/* + * Remove the sysfs related attributes. + * + * @dd Pointer to the driver data structure. + * @kobj Pointer to the kobj for the block device. + * + * return value + * 0 Operation completed successfully. + * -EINVAL Invalid parameter. + */ +static int mtip_hw_sysfs_exit(struct driver_data *dd, struct kobject *kobj) +{ + if (!kobj || !dd) + return -EINVAL; + + sysfs_remove_file(kobj, &dev_attr_registers.attr); + + return 0; +} + +/* + * Perform any init/resume time hardware setup + * + * @dd Pointer to the driver data structure. + * + * return value + * None + */ +static inline void hba_setup(struct driver_data *dd) +{ + u32 hwdata; + hwdata = readl(dd->mmio + HOST_HSORG); + + /* interrupt bug workaround: use only 1 IS bit.*/ + writel(hwdata | + HSORG_DISABLE_SLOTGRP_INTR | + HSORG_DISABLE_SLOTGRP_PXIS, + dd->mmio + HOST_HSORG); +} + +/* + * Detect the details of the product, and store anything needed + * into the driver data structure. This includes product type and + * version and number of slot groups. + * + * @dd Pointer to the driver data structure. + * + * return value + * None + */ +static void mtip_detect_product(struct driver_data *dd) +{ + u32 hwdata; + unsigned int rev, slotgroups; + + /* + * HBA base + 0xFC [15:0] - vendor-specific hardware interface + * info register: + * [15:8] hardware/software interface rev# + * [ 3] asic-style interface + * [ 2:0] number of slot groups, minus 1 (only valid for asic-style). + */ + hwdata = readl(dd->mmio + HOST_HSORG); + + dd->product_type = MTIP_PRODUCT_UNKNOWN; + dd->slot_groups = 1; + + if (hwdata & 0x8) { + dd->product_type = MTIP_PRODUCT_ASICFPGA; + rev = (hwdata & HSORG_HWREV) >> 8; + slotgroups = (hwdata & HSORG_SLOTGROUPS) + 1; + dev_info(&dd->pdev->dev, + "ASIC-FPGA design, HS rev 0x%x, " + "%i slot groups [%i slots]\n", + rev, + slotgroups, + slotgroups * 32); + + if (slotgroups > MTIP_MAX_SLOT_GROUPS) { + dev_warn(&dd->pdev->dev, + "Warning: driver only supports " + "%i slot groups.\n", MTIP_MAX_SLOT_GROUPS); + slotgroups = MTIP_MAX_SLOT_GROUPS; + } + dd->slot_groups = slotgroups; + return; + } + + dev_warn(&dd->pdev->dev, "Unrecognized product id\n"); +} + +/* + * Blocking wait for FTL rebuild to complete + * + * @dd Pointer to the DRIVER_DATA structure. + * + * return value + * 0 FTL rebuild completed successfully + * -EFAULT FTL rebuild error/timeout/interruption + */ +static int mtip_ftl_rebuild_poll(struct driver_data *dd) +{ + unsigned long timeout, cnt = 0, start; + + dev_warn(&dd->pdev->dev, + "FTL rebuild in progress. Polling for completion.\n"); + + start = jiffies; + dd->ftlrebuildflag = 1; + timeout = jiffies + msecs_to_jiffies(MTIP_FTL_REBUILD_TIMEOUT_MS); + + do { + if (mtip_check_surprise_removal(dd->pdev)) + return -EFAULT; + + if (mtip_get_identify(dd->port, NULL) < 0) + return -EFAULT; + + if (*(dd->port->identify + MTIP_FTL_REBUILD_OFFSET) == + MTIP_FTL_REBUILD_MAGIC) { + ssleep(1); + /* Print message every 3 minutes */ + if (cnt++ >= 180) { + dev_warn(&dd->pdev->dev, + "FTL rebuild in progress (%d secs).\n", + jiffies_to_msecs(jiffies - start) / 1000); + cnt = 0; + } + } else { + dev_warn(&dd->pdev->dev, + "FTL rebuild complete (%d secs).\n", + jiffies_to_msecs(jiffies - start) / 1000); + dd->ftlrebuildflag = 0; + mtip_block_initialize(dd); + break; + } + ssleep(10); + } while (time_before(jiffies, timeout)); + + /* Check for timeout */ + if (dd->ftlrebuildflag) { + dev_err(&dd->pdev->dev, + "Timed out waiting for FTL rebuild to complete (%d secs).\n", + jiffies_to_msecs(jiffies - start) / 1000); + return -EFAULT; + } + + return 0; +} + +/* + * service thread to issue queued commands + * + * @data Pointer to the driver data structure. + * + * return value + * 0 + */ + +static int mtip_service_thread(void *data) +{ + struct driver_data *dd = (struct driver_data *)data; + unsigned long slot, slot_start, slot_wrap; + unsigned int num_cmd_slots = dd->slot_groups * 32; + struct mtip_port *port = dd->port; + + while (1) { + /* + * the condition is to check neither an internal command is + * is in progress nor error handling is active + */ + wait_event_interruptible(port->svc_wait, (port->flags) && + !test_bit(MTIP_FLAG_IC_ACTIVE_BIT, &port->flags) && + !test_bit(MTIP_FLAG_EH_ACTIVE_BIT, &port->flags)); + + if (kthread_should_stop()) + break; + + set_bit(MTIP_FLAG_SVC_THD_ACTIVE_BIT, &port->flags); + if (test_bit(MTIP_FLAG_ISSUE_CMDS_BIT, &port->flags)) { + slot = 1; + /* used to restrict the loop to one iteration */ + slot_start = num_cmd_slots; + slot_wrap = 0; + while (1) { + slot = find_next_bit(port->cmds_to_issue, + num_cmd_slots, slot); + if (slot_wrap == 1) { + if ((slot_start >= slot) || + (slot >= num_cmd_slots)) + break; + } + if (unlikely(slot_start == num_cmd_slots)) + slot_start = slot; + + if (unlikely(slot == num_cmd_slots)) { + slot = 1; + slot_wrap = 1; + continue; + } + + /* Issue the command to the hardware */ + mtip_issue_ncq_command(port, slot); + + /* Set the command's timeout value.*/ + port->commands[slot].comp_time = jiffies + + msecs_to_jiffies(MTIP_NCQ_COMMAND_TIMEOUT_MS); + + clear_bit(slot, port->cmds_to_issue); + } + + clear_bit(MTIP_FLAG_ISSUE_CMDS_BIT, &port->flags); + } else if (test_bit(MTIP_FLAG_REBUILD_BIT, &port->flags)) { + mtip_ftl_rebuild_poll(dd); + clear_bit(MTIP_FLAG_REBUILD_BIT, &port->flags); + } + clear_bit(MTIP_FLAG_SVC_THD_ACTIVE_BIT, &port->flags); + + if (test_bit(MTIP_FLAG_SVC_THD_SHOULD_STOP_BIT, &port->flags)) + break; + } + return 0; +} + +/* + * Called once for each card. + * + * @dd Pointer to the driver data structure. + * + * return value + * 0 on success, else an error code. + */ +static int mtip_hw_init(struct driver_data *dd) +{ + int i; + int rv; + unsigned int num_command_slots; + + dd->mmio = pcim_iomap_table(dd->pdev)[MTIP_ABAR]; + + mtip_detect_product(dd); + if (dd->product_type == MTIP_PRODUCT_UNKNOWN) { + rv = -EIO; + goto out1; + } + num_command_slots = dd->slot_groups * 32; + + hba_setup(dd); + + tasklet_init(&dd->tasklet, mtip_tasklet, (unsigned long)dd); + + dd->port = kzalloc(sizeof(struct mtip_port), GFP_KERNEL); + if (!dd->port) { + dev_err(&dd->pdev->dev, + "Memory allocation: port structure\n"); + return -ENOMEM; + } + + /* Counting semaphore to track command slot usage */ + sema_init(&dd->port->cmd_slot, num_command_slots - 1); + + /* Spinlock to prevent concurrent issue */ + spin_lock_init(&dd->port->cmd_issue_lock); + + /* Set the port mmio base address. */ + dd->port->mmio = dd->mmio + PORT_OFFSET; + dd->port->dd = dd; + + /* Allocate memory for the command list. */ + dd->port->command_list = + dmam_alloc_coherent(&dd->pdev->dev, + HW_PORT_PRIV_DMA_SZ + (ATA_SECT_SIZE * 2), + &dd->port->command_list_dma, + GFP_KERNEL); + if (!dd->port->command_list) { + dev_err(&dd->pdev->dev, + "Memory allocation: command list\n"); + rv = -ENOMEM; + goto out1; + } + + /* Clear the memory we have allocated. */ + memset(dd->port->command_list, + 0, + HW_PORT_PRIV_DMA_SZ + (ATA_SECT_SIZE * 2)); + + /* Setup the addresse of the RX FIS. */ + dd->port->rxfis = dd->port->command_list + HW_CMD_SLOT_SZ; + dd->port->rxfis_dma = dd->port->command_list_dma + HW_CMD_SLOT_SZ; + + /* Setup the address of the command tables. */ + dd->port->command_table = dd->port->rxfis + AHCI_RX_FIS_SZ; + dd->port->command_tbl_dma = dd->port->rxfis_dma + AHCI_RX_FIS_SZ; + + /* Setup the address of the identify data. */ + dd->port->identify = dd->port->command_table + + HW_CMD_TBL_AR_SZ; + dd->port->identify_dma = dd->port->command_tbl_dma + + HW_CMD_TBL_AR_SZ; + + /* Setup the address of the sector buffer. */ + dd->port->sector_buffer = (void *) dd->port->identify + ATA_SECT_SIZE; + dd->port->sector_buffer_dma = dd->port->identify_dma + ATA_SECT_SIZE; + + /* Point the command headers at the command tables. */ + for (i = 0; i < num_command_slots; i++) { + dd->port->commands[i].command_header = + dd->port->command_list + + (sizeof(struct mtip_cmd_hdr) * i); + dd->port->commands[i].command_header_dma = + dd->port->command_list_dma + + (sizeof(struct mtip_cmd_hdr) * i); + + dd->port->commands[i].command = + dd->port->command_table + (HW_CMD_TBL_SZ * i); + dd->port->commands[i].command_dma = + dd->port->command_tbl_dma + (HW_CMD_TBL_SZ * i); + + if (readl(dd->mmio + HOST_CAP) & HOST_CAP_64) + dd->port->commands[i].command_header->ctbau = + __force_bit2int cpu_to_le32( + (dd->port->commands[i].command_dma >> 16) >> 16); + dd->port->commands[i].command_header->ctba = + __force_bit2int cpu_to_le32( + dd->port->commands[i].command_dma & 0xFFFFFFFF); + + /* + * If this is not done, a bug is reported by the stock + * FC11 i386. Due to the fact that it has lots of kernel + * debugging enabled. + */ + sg_init_table(dd->port->commands[i].sg, MTIP_MAX_SG); + + /* Mark all commands as currently inactive.*/ + atomic_set(&dd->port->commands[i].active, 0); + } + + /* Setup the pointers to the extended s_active and CI registers. */ + for (i = 0; i < dd->slot_groups; i++) { + dd->port->s_active[i] = + dd->port->mmio + i*0x80 + PORT_SCR_ACT; + dd->port->cmd_issue[i] = + dd->port->mmio + i*0x80 + PORT_COMMAND_ISSUE; + dd->port->completed[i] = + dd->port->mmio + i*0x80 + PORT_SDBV; + } + + /* Reset the HBA. */ + if (mtip_hba_reset(dd) < 0) { + dev_err(&dd->pdev->dev, + "Card did not reset within timeout\n"); + rv = -EIO; + goto out2; + } + + mtip_init_port(dd->port); + mtip_start_port(dd->port); + + /* Setup the ISR and enable interrupts. */ + rv = devm_request_irq(&dd->pdev->dev, + dd->pdev->irq, + mtip_irq_handler, + IRQF_SHARED, + dev_driver_string(&dd->pdev->dev), + dd); + + if (rv) { + dev_err(&dd->pdev->dev, + "Unable to allocate IRQ %d\n", dd->pdev->irq); + goto out2; + } + + /* Enable interrupts on the HBA. */ + writel(readl(dd->mmio + HOST_CTL) | HOST_IRQ_EN, + dd->mmio + HOST_CTL); + + init_timer(&dd->port->cmd_timer); + init_waitqueue_head(&dd->port->svc_wait); + + dd->port->cmd_timer.data = (unsigned long int) dd->port; + dd->port->cmd_timer.function = mtip_timeout_function; + mod_timer(&dd->port->cmd_timer, + jiffies + msecs_to_jiffies(MTIP_TIMEOUT_CHECK_PERIOD)); + + if (mtip_get_identify(dd->port, NULL) < 0) { + rv = -EFAULT; + goto out3; + } + + if (*(dd->port->identify + MTIP_FTL_REBUILD_OFFSET) == + MTIP_FTL_REBUILD_MAGIC) { + set_bit(MTIP_FLAG_REBUILD_BIT, &dd->port->flags); + return MTIP_FTL_REBUILD_MAGIC; + } + mtip_dump_identify(dd->port); + return rv; + +out3: + del_timer_sync(&dd->port->cmd_timer); + + /* Disable interrupts on the HBA. */ + writel(readl(dd->mmio + HOST_CTL) & ~HOST_IRQ_EN, + dd->mmio + HOST_CTL); + + /*Release the IRQ. */ + devm_free_irq(&dd->pdev->dev, dd->pdev->irq, dd); + +out2: + mtip_deinit_port(dd->port); + + /* Free the command/command header memory. */ + dmam_free_coherent(&dd->pdev->dev, + HW_PORT_PRIV_DMA_SZ + (ATA_SECT_SIZE * 2), + dd->port->command_list, + dd->port->command_list_dma); +out1: + /* Free the memory allocated for the for structure. */ + kfree(dd->port); + + return rv; +} + +/* + * Called to deinitialize an interface. + * + * @dd Pointer to the driver data structure. + * + * return value + * 0 + */ +static int mtip_hw_exit(struct driver_data *dd) +{ + /* + * Send standby immediate (E0h) to the drive so that it + * saves its state. + */ + if (atomic_read(&dd->drv_cleanup_done) != true) { + + mtip_standby_immediate(dd->port); + + /* de-initialize the port. */ + mtip_deinit_port(dd->port); + + /* Disable interrupts on the HBA. */ + writel(readl(dd->mmio + HOST_CTL) & ~HOST_IRQ_EN, + dd->mmio + HOST_CTL); + } + + del_timer_sync(&dd->port->cmd_timer); + + /* Release the IRQ. */ + devm_free_irq(&dd->pdev->dev, dd->pdev->irq, dd); + + /* Stop the bottom half tasklet. */ + tasklet_kill(&dd->tasklet); + + /* Free the command/command header memory. */ + dmam_free_coherent(&dd->pdev->dev, + HW_PORT_PRIV_DMA_SZ + (ATA_SECT_SIZE * 2), + dd->port->command_list, + dd->port->command_list_dma); + /* Free the memory allocated for the for structure. */ + kfree(dd->port); + + return 0; +} + +/* + * Issue a Standby Immediate command to the device. + * + * This function is called by the Block Layer just before the + * system powers off during a shutdown. + * + * @dd Pointer to the driver data structure. + * + * return value + * 0 + */ +static int mtip_hw_shutdown(struct driver_data *dd) +{ + /* + * Send standby immediate (E0h) to the drive so that it + * saves its state. + */ + mtip_standby_immediate(dd->port); + + return 0; +} + +/* + * Suspend function + * + * This function is called by the Block Layer just before the + * system hibernates. + * + * @dd Pointer to the driver data structure. + * + * return value + * 0 Suspend was successful + * -EFAULT Suspend was not successful + */ +static int mtip_hw_suspend(struct driver_data *dd) +{ + /* + * Send standby immediate (E0h) to the drive + * so that it saves its state. + */ + if (mtip_standby_immediate(dd->port) != 0) { + dev_err(&dd->pdev->dev, + "Failed standby-immediate command\n"); + return -EFAULT; + } + + /* Disable interrupts on the HBA.*/ + writel(readl(dd->mmio + HOST_CTL) & ~HOST_IRQ_EN, + dd->mmio + HOST_CTL); + mtip_deinit_port(dd->port); + + return 0; +} + +/* + * Resume function + * + * This function is called by the Block Layer as the + * system resumes. + * + * @dd Pointer to the driver data structure. + * + * return value + * 0 Resume was successful + * -EFAULT Resume was not successful + */ +static int mtip_hw_resume(struct driver_data *dd) +{ + /* Perform any needed hardware setup steps */ + hba_setup(dd); + + /* Reset the HBA */ + if (mtip_hba_reset(dd) != 0) { + dev_err(&dd->pdev->dev, + "Unable to reset the HBA\n"); + return -EFAULT; + } + + /* + * Enable the port, DMA engine, and FIS reception specific + * h/w in controller. + */ + mtip_init_port(dd->port); + mtip_start_port(dd->port); + + /* Enable interrupts on the HBA.*/ + writel(readl(dd->mmio + HOST_CTL) | HOST_IRQ_EN, + dd->mmio + HOST_CTL); + + return 0; +} + +/* + * Helper function for reusing disk name + * upon hot insertion. + */ +static int rssd_disk_name_format(char *prefix, + int index, + char *buf, + int buflen) +{ + const int base = 'z' - 'a' + 1; + char *begin = buf + strlen(prefix); + char *end = buf + buflen; + char *p; + int unit; + + p = end - 1; + *p = '\0'; + unit = base; + do { + if (p == begin) + return -EINVAL; + *--p = 'a' + (index % unit); + index = (index / unit) - 1; + } while (index >= 0); + + memmove(begin, p, end - p); + memcpy(buf, prefix, strlen(prefix)); + + return 0; +} + +/* + * Block layer IOCTL handler. + * + * @dev Pointer to the block_device structure. + * @mode ignored + * @cmd IOCTL command passed from the user application. + * @arg Argument passed from the user application. + * + * return value + * 0 IOCTL completed successfully. + * -ENOTTY IOCTL not supported or invalid driver data + * structure pointer. + */ +static int mtip_block_ioctl(struct block_device *dev, + fmode_t mode, + unsigned cmd, + unsigned long arg) +{ + struct driver_data *dd = dev->bd_disk->private_data; + + if (!capable(CAP_SYS_ADMIN)) + return -EACCES; + + if (!dd) + return -ENOTTY; + + switch (cmd) { + case BLKFLSBUF: + return -ENOTTY; + default: + return mtip_hw_ioctl(dd, cmd, arg); + } +} + +#ifdef CONFIG_COMPAT +/* + * Block layer compat IOCTL handler. + * + * @dev Pointer to the block_device structure. + * @mode ignored + * @cmd IOCTL command passed from the user application. + * @arg Argument passed from the user application. + * + * return value + * 0 IOCTL completed successfully. + * -ENOTTY IOCTL not supported or invalid driver data + * structure pointer. + */ +static int mtip_block_compat_ioctl(struct block_device *dev, + fmode_t mode, + unsigned cmd, + unsigned long arg) +{ + struct driver_data *dd = dev->bd_disk->private_data; + + if (!capable(CAP_SYS_ADMIN)) + return -EACCES; + + if (!dd) + return -ENOTTY; + + switch (cmd) { + case BLKFLSBUF: + return -ENOTTY; + case HDIO_DRIVE_TASKFILE: { + struct mtip_compat_ide_task_request_s __user *compat_req_task; + ide_task_request_t req_task; + int compat_tasksize, outtotal, ret; + + compat_tasksize = + sizeof(struct mtip_compat_ide_task_request_s); + + compat_req_task = + (struct mtip_compat_ide_task_request_s __user *) arg; + + if (copy_from_user(&req_task, (void __user *) arg, + compat_tasksize - (2 * sizeof(compat_long_t)))) + return -EFAULT; + + if (get_user(req_task.out_size, &compat_req_task->out_size)) + return -EFAULT; + + if (get_user(req_task.in_size, &compat_req_task->in_size)) + return -EFAULT; + + outtotal = sizeof(struct mtip_compat_ide_task_request_s); + + ret = exec_drive_taskfile(dd, (void __user *) arg, + &req_task, outtotal); + + if (copy_to_user((void __user *) arg, &req_task, + compat_tasksize - + (2 * sizeof(compat_long_t)))) + return -EFAULT; + + if (put_user(req_task.out_size, &compat_req_task->out_size)) + return -EFAULT; + + if (put_user(req_task.in_size, &compat_req_task->in_size)) + return -EFAULT; + + return ret; + } + default: + return mtip_hw_ioctl(dd, cmd, arg); + } +} +#endif + +/* + * Obtain the geometry of the device. + * + * You may think that this function is obsolete, but some applications, + * fdisk for example still used CHS values. This function describes the + * device as having 224 heads and 56 sectors per cylinder. These values are + * chosen so that each cylinder is aligned on a 4KB boundary. Since a + * partition is described in terms of a start and end cylinder this means + * that each partition is also 4KB aligned. Non-aligned partitions adversely + * affects performance. + * + * @dev Pointer to the block_device strucutre. + * @geo Pointer to a hd_geometry structure. + * + * return value + * 0 Operation completed successfully. + * -ENOTTY An error occurred while reading the drive capacity. + */ +static int mtip_block_getgeo(struct block_device *dev, + struct hd_geometry *geo) +{ + struct driver_data *dd = dev->bd_disk->private_data; + sector_t capacity; + + if (!dd) + return -ENOTTY; + + if (!(mtip_hw_get_capacity(dd, &capacity))) { + dev_warn(&dd->pdev->dev, + "Could not get drive capacity.\n"); + return -ENOTTY; + } + + geo->heads = 224; + geo->sectors = 56; + sector_div(capacity, (geo->heads * geo->sectors)); + geo->cylinders = capacity; + return 0; +} + +/* + * Block device operation function. + * + * This structure contains pointers to the functions required by the block + * layer. + */ +static const struct block_device_operations mtip_block_ops = { + .ioctl = mtip_block_ioctl, +#ifdef CONFIG_COMPAT + .compat_ioctl = mtip_block_compat_ioctl, +#endif + .getgeo = mtip_block_getgeo, + .owner = THIS_MODULE +}; + +/* + * Block layer make request function. + * + * This function is called by the kernel to process a BIO for + * the P320 device. + * + * @queue Pointer to the request queue. Unused other than to obtain + * the driver data structure. + * @bio Pointer to the BIO. + * + */ +static void mtip_make_request(struct request_queue *queue, struct bio *bio) +{ + struct driver_data *dd = queue->queuedata; + struct scatterlist *sg; + struct bio_vec *bvec; + int nents = 0; + int tag = 0; + + if (unlikely(!bio_has_data(bio))) { + blk_queue_flush(queue, 0); + bio_endio(bio, 0); + return; + } + + sg = mtip_hw_get_scatterlist(dd, &tag); + if (likely(sg != NULL)) { + blk_queue_bounce(queue, &bio); + + if (unlikely((bio)->bi_vcnt > MTIP_MAX_SG)) { + dev_warn(&dd->pdev->dev, + "Maximum number of SGL entries exceeded"); + bio_io_error(bio); + mtip_hw_release_scatterlist(dd, tag); + return; + } + + /* Create the scatter list for this bio. */ + bio_for_each_segment(bvec, bio, nents) { + sg_set_page(&sg[nents], + bvec->bv_page, + bvec->bv_len, + bvec->bv_offset); + } + + /* Issue the read/write. */ + mtip_hw_submit_io(dd, + bio->bi_sector, + bio_sectors(bio), + nents, + tag, + bio_endio, + bio, + bio_data_dir(bio)); + } else + bio_io_error(bio); +} + +/* + * Block layer initialization function. + * + * This function is called once by the PCI layer for each P320 + * device that is connected to the system. + * + * @dd Pointer to the driver data structure. + * + * return value + * 0 on success else an error code. + */ +static int mtip_block_initialize(struct driver_data *dd) +{ + int rv = 0, wait_for_rebuild = 0; + sector_t capacity; + unsigned int index = 0; + struct kobject *kobj; + unsigned char thd_name[16]; + + if (dd->disk) + goto skip_create_disk; /* hw init done, before rebuild */ + + /* Initialize the protocol layer. */ + wait_for_rebuild = mtip_hw_init(dd); + if (wait_for_rebuild < 0) { + dev_err(&dd->pdev->dev, + "Protocol layer initialization failed\n"); + rv = -EINVAL; + goto protocol_init_error; + } + + dd->disk = alloc_disk(MTIP_MAX_MINORS); + if (dd->disk == NULL) { + dev_err(&dd->pdev->dev, + "Unable to allocate gendisk structure\n"); + rv = -EINVAL; + goto alloc_disk_error; + } + + /* Generate the disk name, implemented same as in sd.c */ + do { + if (!ida_pre_get(&rssd_index_ida, GFP_KERNEL)) + goto ida_get_error; + + spin_lock(&rssd_index_lock); + rv = ida_get_new(&rssd_index_ida, &index); + spin_unlock(&rssd_index_lock); + } while (rv == -EAGAIN); + + if (rv) + goto ida_get_error; + + rv = rssd_disk_name_format("rssd", + index, + dd->disk->disk_name, + DISK_NAME_LEN); + if (rv) + goto disk_index_error; + + dd->disk->driverfs_dev = &dd->pdev->dev; + dd->disk->major = dd->major; + dd->disk->first_minor = dd->instance * MTIP_MAX_MINORS; + dd->disk->fops = &mtip_block_ops; + dd->disk->private_data = dd; + dd->index = index; + + /* + * if rebuild pending, start the service thread, and delay the block + * queue creation and add_disk() + */ + if (wait_for_rebuild == MTIP_FTL_REBUILD_MAGIC) + goto start_service_thread; + +skip_create_disk: + /* Allocate the request queue. */ + dd->queue = blk_alloc_queue(GFP_KERNEL); + if (dd->queue == NULL) { + dev_err(&dd->pdev->dev, + "Unable to allocate request queue\n"); + rv = -ENOMEM; + goto block_queue_alloc_init_error; + } + + /* Attach our request function to the request queue. */ + blk_queue_make_request(dd->queue, mtip_make_request); + + dd->disk->queue = dd->queue; + dd->queue->queuedata = dd; + + /* Set device limits. */ + set_bit(QUEUE_FLAG_NONROT, &dd->queue->queue_flags); + blk_queue_max_segments(dd->queue, MTIP_MAX_SG); + blk_queue_physical_block_size(dd->queue, 4096); + blk_queue_io_min(dd->queue, 4096); + /* + * write back cache is not supported in the device. FUA depends on + * write back cache support, hence setting flush support to zero. + */ + blk_queue_flush(dd->queue, 0); + + /* Set the capacity of the device in 512 byte sectors. */ + if (!(mtip_hw_get_capacity(dd, &capacity))) { + dev_warn(&dd->pdev->dev, + "Could not read drive capacity\n"); + rv = -EIO; + goto read_capacity_error; + } + set_capacity(dd->disk, capacity); + + /* Enable the block device and add it to /dev */ + add_disk(dd->disk); + + /* + * Now that the disk is active, initialize any sysfs attributes + * managed by the protocol layer. + */ + kobj = kobject_get(&disk_to_dev(dd->disk)->kobj); + if (kobj) { + mtip_hw_sysfs_init(dd, kobj); + kobject_put(kobj); + } + + if (dd->mtip_svc_handler) + return rv; /* service thread created for handling rebuild */ + +start_service_thread: + sprintf(thd_name, "mtip_svc_thd_%02d", index); + + dd->mtip_svc_handler = kthread_run(mtip_service_thread, + dd, thd_name); + + if (IS_ERR(dd->mtip_svc_handler)) { + printk(KERN_ERR "mtip32xx: service thread failed to start\n"); + dd->mtip_svc_handler = NULL; + rv = -EFAULT; + goto kthread_run_error; + } + + return rv; + +kthread_run_error: + /* Delete our gendisk. This also removes the device from /dev */ + del_gendisk(dd->disk); + +read_capacity_error: + blk_cleanup_queue(dd->queue); + +block_queue_alloc_init_error: +disk_index_error: + spin_lock(&rssd_index_lock); + ida_remove(&rssd_index_ida, index); + spin_unlock(&rssd_index_lock); + +ida_get_error: + put_disk(dd->disk); + +alloc_disk_error: + mtip_hw_exit(dd); /* De-initialize the protocol layer. */ + +protocol_init_error: + return rv; +} + +/* + * Block layer deinitialization function. + * + * Called by the PCI layer as each P320 device is removed. + * + * @dd Pointer to the driver data structure. + * + * return value + * 0 + */ +static int mtip_block_remove(struct driver_data *dd) +{ + struct kobject *kobj; + + if (dd->mtip_svc_handler) { + set_bit(MTIP_FLAG_SVC_THD_SHOULD_STOP_BIT, &dd->port->flags); + wake_up_interruptible(&dd->port->svc_wait); + kthread_stop(dd->mtip_svc_handler); + } + + /* Clean up the sysfs attributes managed by the protocol layer. */ + kobj = kobject_get(&disk_to_dev(dd->disk)->kobj); + if (kobj) { + mtip_hw_sysfs_exit(dd, kobj); + kobject_put(kobj); + } + + /* + * Delete our gendisk structure. This also removes the device + * from /dev + */ + del_gendisk(dd->disk); + blk_cleanup_queue(dd->queue); + dd->disk = NULL; + dd->queue = NULL; + + /* De-initialize the protocol layer. */ + mtip_hw_exit(dd); + + return 0; +} + +/* + * Function called by the PCI layer when just before the + * machine shuts down. + * + * If a protocol layer shutdown function is present it will be called + * by this function. + * + * @dd Pointer to the driver data structure. + * + * return value + * 0 + */ +static int mtip_block_shutdown(struct driver_data *dd) +{ + dev_info(&dd->pdev->dev, + "Shutting down %s ...\n", dd->disk->disk_name); + + /* Delete our gendisk structure, and cleanup the blk queue. */ + del_gendisk(dd->disk); + blk_cleanup_queue(dd->queue); + dd->disk = NULL; + dd->queue = NULL; + + mtip_hw_shutdown(dd); + return 0; +} + +static int mtip_block_suspend(struct driver_data *dd) +{ + dev_info(&dd->pdev->dev, + "Suspending %s ...\n", dd->disk->disk_name); + mtip_hw_suspend(dd); + return 0; +} + +static int mtip_block_resume(struct driver_data *dd) +{ + dev_info(&dd->pdev->dev, "Resuming %s ...\n", + dd->disk->disk_name); + mtip_hw_resume(dd); + return 0; +} + +/* + * Called for each supported PCI device detected. + * + * This function allocates the private data structure, enables the + * PCI device and then calls the block layer initialization function. + * + * return value + * 0 on success else an error code. + */ +static int mtip_pci_probe(struct pci_dev *pdev, + const struct pci_device_id *ent) +{ + int rv = 0; + struct driver_data *dd = NULL; + + /* Allocate memory for this devices private data. */ + dd = kzalloc(sizeof(struct driver_data), GFP_KERNEL); + if (dd == NULL) { + dev_err(&pdev->dev, + "Unable to allocate memory for driver data\n"); + return -ENOMEM; + } + + /* Set the atomic variable as 1 in case of SRSI */ + atomic_set(&dd->drv_cleanup_done, true); + + atomic_set(&dd->resumeflag, false); + + /* Attach the private data to this PCI device. */ + pci_set_drvdata(pdev, dd); + + rv = pcim_enable_device(pdev); + if (rv < 0) { + dev_err(&pdev->dev, "Unable to enable device\n"); + goto iomap_err; + } + + /* Map BAR5 to memory. */ + rv = pcim_iomap_regions(pdev, 1 << MTIP_ABAR, MTIP_DRV_NAME); + if (rv < 0) { + dev_err(&pdev->dev, "Unable to map regions\n"); + goto iomap_err; + } + + if (!pci_set_dma_mask(pdev, DMA_BIT_MASK(64))) { + rv = pci_set_consistent_dma_mask(pdev, DMA_BIT_MASK(64)); + + if (rv) { + rv = pci_set_consistent_dma_mask(pdev, + DMA_BIT_MASK(32)); + if (rv) { + dev_warn(&pdev->dev, + "64-bit DMA enable failed\n"); + goto setmask_err; + } + } + } + + pci_set_master(pdev); + + if (pci_enable_msi(pdev)) { + dev_warn(&pdev->dev, + "Unable to enable MSI interrupt.\n"); + goto block_initialize_err; + } + + /* Copy the info we may need later into the private data structure. */ + dd->major = mtip_major; + dd->instance = instance; + dd->pdev = pdev; + + /* Initialize the block layer. */ + rv = mtip_block_initialize(dd); + if (rv < 0) { + dev_err(&pdev->dev, + "Unable to initialize block layer\n"); + goto block_initialize_err; + } + + /* + * Increment the instance count so that each device has a unique + * instance number. + */ + instance++; + + goto done; + +block_initialize_err: + pci_disable_msi(pdev); + +setmask_err: + pcim_iounmap_regions(pdev, 1 << MTIP_ABAR); + +iomap_err: + kfree(dd); + pci_set_drvdata(pdev, NULL); + return rv; +done: + /* Set the atomic variable as 0 in case of SRSI */ + atomic_set(&dd->drv_cleanup_done, true); + + return rv; +} + +/* + * Called for each probed device when the device is removed or the + * driver is unloaded. + * + * return value + * None + */ +static void mtip_pci_remove(struct pci_dev *pdev) +{ + struct driver_data *dd = pci_get_drvdata(pdev); + int counter = 0; + + if (mtip_check_surprise_removal(pdev)) { + while (atomic_read(&dd->drv_cleanup_done) == false) { + counter++; + msleep(20); + if (counter == 10) { + /* Cleanup the outstanding commands */ + mtip_command_cleanup(dd); + break; + } + } + } + /* Set the atomic variable as 1 in case of SRSI */ + atomic_set(&dd->drv_cleanup_done, true); + + /* Clean up the block layer. */ + mtip_block_remove(dd); + + pci_disable_msi(pdev); + + kfree(dd); + pcim_iounmap_regions(pdev, 1 << MTIP_ABAR); +} + +/* + * Called for each probed device when the device is suspended. + * + * return value + * 0 Success + * <0 Error + */ +static int mtip_pci_suspend(struct pci_dev *pdev, pm_message_t mesg) +{ + int rv = 0; + struct driver_data *dd = pci_get_drvdata(pdev); + + if (!dd) { + dev_err(&pdev->dev, + "Driver private datastructure is NULL\n"); + return -EFAULT; + } + + atomic_set(&dd->resumeflag, true); + + /* Disable ports & interrupts then send standby immediate */ + rv = mtip_block_suspend(dd); + if (rv < 0) { + dev_err(&pdev->dev, + "Failed to suspend controller\n"); + return rv; + } + + /* + * Save the pci config space to pdev structure & + * disable the device + */ + pci_save_state(pdev); + pci_disable_device(pdev); + + /* Move to Low power state*/ + pci_set_power_state(pdev, PCI_D3hot); + + return rv; +} + +/* + * Called for each probed device when the device is resumed. + * + * return value + * 0 Success + * <0 Error + */ +static int mtip_pci_resume(struct pci_dev *pdev) +{ + int rv = 0; + struct driver_data *dd; + + dd = pci_get_drvdata(pdev); + if (!dd) { + dev_err(&pdev->dev, + "Driver private datastructure is NULL\n"); + return -EFAULT; + } + + /* Move the device to active State */ + pci_set_power_state(pdev, PCI_D0); + + /* Restore PCI configuration space */ + pci_restore_state(pdev); + + /* Enable the PCI device*/ + rv = pcim_enable_device(pdev); + if (rv < 0) { + dev_err(&pdev->dev, + "Failed to enable card during resume\n"); + goto err; + } + pci_set_master(pdev); + + /* + * Calls hbaReset, initPort, & startPort function + * then enables interrupts + */ + rv = mtip_block_resume(dd); + if (rv < 0) + dev_err(&pdev->dev, "Unable to resume\n"); + +err: + atomic_set(&dd->resumeflag, false); + + return rv; +} + +/* + * Shutdown routine + * + * return value + * None + */ +static void mtip_pci_shutdown(struct pci_dev *pdev) +{ + struct driver_data *dd = pci_get_drvdata(pdev); + if (dd) + mtip_block_shutdown(dd); +} + +/* Table of device ids supported by this driver. */ +static DEFINE_PCI_DEVICE_TABLE(mtip_pci_tbl) = { + { PCI_DEVICE(PCI_VENDOR_ID_MICRON, P320_DEVICE_ID) }, + { 0 } +}; + +/* Structure that describes the PCI driver functions. */ +static struct pci_driver mtip_pci_driver = { + .name = MTIP_DRV_NAME, + .id_table = mtip_pci_tbl, + .probe = mtip_pci_probe, + .remove = mtip_pci_remove, + .suspend = mtip_pci_suspend, + .resume = mtip_pci_resume, + .shutdown = mtip_pci_shutdown, +}; + +MODULE_DEVICE_TABLE(pci, mtip_pci_tbl); + +/* + * Module initialization function. + * + * Called once when the module is loaded. This function allocates a major + * block device number to the Cyclone devices and registers the PCI layer + * of the driver. + * + * Return value + * 0 on success else error code. + */ +static int __init mtip_init(void) +{ + printk(KERN_INFO MTIP_DRV_NAME " Version " MTIP_DRV_VERSION "\n"); + + /* Allocate a major block device number to use with this driver. */ + mtip_major = register_blkdev(0, MTIP_DRV_NAME); + if (mtip_major < 0) { + printk(KERN_ERR "Unable to register block device (%d)\n", + mtip_major); + return -EBUSY; + } + + /* Register our PCI operations. */ + return pci_register_driver(&mtip_pci_driver); +} + +/* + * Module de-initialization function. + * + * Called once when the module is unloaded. This function deallocates + * the major block device number allocated by mtip_init() and + * unregisters the PCI layer of the driver. + * + * Return value + * none + */ +static void __exit mtip_exit(void) +{ + /* Release the allocated major block device number. */ + unregister_blkdev(mtip_major, MTIP_DRV_NAME); + + /* Unregister the PCI driver. */ + pci_unregister_driver(&mtip_pci_driver); +} + +MODULE_AUTHOR("Micron Technology, Inc"); +MODULE_DESCRIPTION("Micron RealSSD PCIe Block Driver"); +MODULE_LICENSE("GPL"); +MODULE_VERSION(MTIP_DRV_VERSION); + +module_init(mtip_init); +module_exit(mtip_exit); diff --git a/drivers/block/mtip32xx/mtip32xx.h b/drivers/block/mtip32xx/mtip32xx.h new file mode 100644 index 000000000000..e0554a8f2233 --- /dev/null +++ b/drivers/block/mtip32xx/mtip32xx.h @@ -0,0 +1,418 @@ +/* + * mtip32xx.h - Header file for the P320 SSD Block Driver + * Copyright (C) 2011 Micron Technology, Inc. + * + * Portions of this code were derived from works subjected to the + * following copyright: + * Copyright (C) 2009 Integrated Device Technology, Inc. + * + * This program is free software; you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation; either version 2 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + */ + +#ifndef __MTIP32XX_H__ +#define __MTIP32XX_H__ + +#include <linux/spinlock.h> +#include <linux/rwsem.h> +#include <linux/ata.h> +#include <linux/interrupt.h> +#include <linux/genhd.h> +#include <linux/version.h> + +/* Offset of Subsystem Device ID in pci confoguration space */ +#define PCI_SUBSYSTEM_DEVICEID 0x2E + +/* offset of Device Control register in PCIe extended capabilites space */ +#define PCIE_CONFIG_EXT_DEVICE_CONTROL_OFFSET 0x48 + +/* # of times to retry timed out IOs */ +#define MTIP_MAX_RETRIES 5 + +/* Various timeout values in ms */ +#define MTIP_NCQ_COMMAND_TIMEOUT_MS 5000 +#define MTIP_IOCTL_COMMAND_TIMEOUT_MS 5000 +#define MTIP_INTERNAL_COMMAND_TIMEOUT_MS 5000 + +/* check for timeouts every 500ms */ +#define MTIP_TIMEOUT_CHECK_PERIOD 500 + +/* ftl rebuild */ +#define MTIP_FTL_REBUILD_OFFSET 142 +#define MTIP_FTL_REBUILD_MAGIC 0xED51 +#define MTIP_FTL_REBUILD_TIMEOUT_MS 2400000 + +/* Macro to extract the tag bit number from a tag value. */ +#define MTIP_TAG_BIT(tag) (tag & 0x1F) + +/* + * Macro to extract the tag index from a tag value. The index + * is used to access the correct s_active/Command Issue register based + * on the tag value. + */ +#define MTIP_TAG_INDEX(tag) (tag >> 5) + +/* + * Maximum number of scatter gather entries + * a single command may have. + */ +#define MTIP_MAX_SG 128 + +/* + * Maximum number of slot groups (Command Issue & s_active registers) + * NOTE: This is the driver maximum; check dd->slot_groups for actual value. + */ +#define MTIP_MAX_SLOT_GROUPS 8 + +/* Internal command tag. */ +#define MTIP_TAG_INTERNAL 0 + +/* Micron Vendor ID & P320x SSD Device ID */ +#define PCI_VENDOR_ID_MICRON 0x1344 +#define P320_DEVICE_ID 0x5150 + +/* Driver name and version strings */ +#define MTIP_DRV_NAME "mtip32xx" +#define MTIP_DRV_VERSION "1.2.6os3" + +/* Maximum number of minor device numbers per device. */ +#define MTIP_MAX_MINORS 16 + +/* Maximum number of supported command slots. */ +#define MTIP_MAX_COMMAND_SLOTS (MTIP_MAX_SLOT_GROUPS * 32) + +/* + * Per-tag bitfield size in longs. + * Linux bit manipulation functions + * (i.e. test_and_set_bit, find_next_zero_bit) + * manipulate memory in longs, so we try to make the math work. + * take the slot groups and find the number of longs, rounding up. + * Careful! i386 and x86_64 use different size longs! + */ +#define U32_PER_LONG (sizeof(long) / sizeof(u32)) +#define SLOTBITS_IN_LONGS ((MTIP_MAX_SLOT_GROUPS + \ + (U32_PER_LONG-1))/U32_PER_LONG) + +/* BAR number used to access the HBA registers. */ +#define MTIP_ABAR 5 + +#ifdef DEBUG + #define dbg_printk(format, arg...) \ + printk(pr_fmt(format), ##arg); +#else + #define dbg_printk(format, arg...) +#endif + +#define __force_bit2int (unsigned int __force) + +/* below are bit numbers in 'flags' defined in mtip_port */ +#define MTIP_FLAG_IC_ACTIVE_BIT 0 +#define MTIP_FLAG_EH_ACTIVE_BIT 1 +#define MTIP_FLAG_SVC_THD_ACTIVE_BIT 2 +#define MTIP_FLAG_ISSUE_CMDS_BIT 4 +#define MTIP_FLAG_REBUILD_BIT 5 +#define MTIP_FLAG_SVC_THD_SHOULD_STOP_BIT 8 + +/* Register Frame Information Structure (FIS), host to device. */ +struct host_to_dev_fis { + /* + * FIS type. + * - 27h Register FIS, host to device. + * - 34h Register FIS, device to host. + * - 39h DMA Activate FIS, device to host. + * - 41h DMA Setup FIS, bi-directional. + * - 46h Data FIS, bi-directional. + * - 58h BIST Activate FIS, bi-directional. + * - 5Fh PIO Setup FIS, device to host. + * - A1h Set Device Bits FIS, device to host. + */ + unsigned char type; + unsigned char opts; + unsigned char command; + unsigned char features; + + union { + unsigned char lba_low; + unsigned char sector; + }; + union { + unsigned char lba_mid; + unsigned char cyl_low; + }; + union { + unsigned char lba_hi; + unsigned char cyl_hi; + }; + union { + unsigned char device; + unsigned char head; + }; + + union { + unsigned char lba_low_ex; + unsigned char sector_ex; + }; + union { + unsigned char lba_mid_ex; + unsigned char cyl_low_ex; + }; + union { + unsigned char lba_hi_ex; + unsigned char cyl_hi_ex; + }; + unsigned char features_ex; + + unsigned char sect_count; + unsigned char sect_cnt_ex; + unsigned char res2; + unsigned char control; + + unsigned int res3; +}; + +/* Command header structure. */ +struct mtip_cmd_hdr { + /* + * Command options. + * - Bits 31:16 Number of PRD entries. + * - Bits 15:8 Unused in this implementation. + * - Bit 7 Prefetch bit, informs the drive to prefetch PRD entries. + * - Bit 6 Write bit, should be set when writing data to the device. + * - Bit 5 Unused in this implementation. + * - Bits 4:0 Length of the command FIS in DWords (DWord = 4 bytes). + */ + unsigned int opts; + /* This field is unsed when using NCQ. */ + union { + unsigned int byte_count; + unsigned int status; + }; + /* + * Lower 32 bits of the command table address associated with this + * header. The command table addresses must be 128 byte aligned. + */ + unsigned int ctba; + /* + * If 64 bit addressing is used this field is the upper 32 bits + * of the command table address associated with this command. + */ + unsigned int ctbau; + /* Reserved and unused. */ + unsigned int res[4]; +}; + +/* Command scatter gather structure (PRD). */ +struct mtip_cmd_sg { + /* + * Low 32 bits of the data buffer address. For P320 this + * address must be 8 byte aligned signified by bits 2:0 being + * set to 0. + */ + unsigned int dba; + /* + * When 64 bit addressing is used this field is the upper + * 32 bits of the data buffer address. + */ + unsigned int dba_upper; + /* Unused. */ + unsigned int reserved; + /* + * Bit 31: interrupt when this data block has been transferred. + * Bits 30..22: reserved + * Bits 21..0: byte count (minus 1). For P320 the byte count must be + * 8 byte aligned signified by bits 2:0 being set to 1. + */ + unsigned int info; +}; +struct mtip_port; + +/* Structure used to describe a command. */ +struct mtip_cmd { + + struct mtip_cmd_hdr *command_header; /* ptr to command header entry */ + + dma_addr_t command_header_dma; /* corresponding physical address */ + + void *command; /* ptr to command table entry */ + + dma_addr_t command_dma; /* corresponding physical address */ + + void *comp_data; /* data passed to completion function comp_func() */ + /* + * Completion function called by the ISR upon completion of + * a command. + */ + void (*comp_func)(struct mtip_port *port, + int tag, + void *data, + int status); + /* Additional callback function that may be called by comp_func() */ + void (*async_callback)(void *data, int status); + + void *async_data; /* Addl. data passed to async_callback() */ + + int scatter_ents; /* Number of scatter list entries used */ + + struct scatterlist sg[MTIP_MAX_SG]; /* Scatter list entries */ + + int retries; /* The number of retries left for this command. */ + + int direction; /* Data transfer direction */ + + unsigned long comp_time; /* command completion time, in jiffies */ + + atomic_t active; /* declares if this command sent to the drive. */ +}; + +/* Structure used to describe a port. */ +struct mtip_port { + /* Pointer back to the driver data for this port. */ + struct driver_data *dd; + /* + * Used to determine if the data pointed to by the + * identify field is valid. + */ + unsigned long identify_valid; + /* Base address of the memory mapped IO for the port. */ + void __iomem *mmio; + /* Array of pointers to the memory mapped s_active registers. */ + void __iomem *s_active[MTIP_MAX_SLOT_GROUPS]; + /* Array of pointers to the memory mapped completed registers. */ + void __iomem *completed[MTIP_MAX_SLOT_GROUPS]; + /* Array of pointers to the memory mapped Command Issue registers. */ + void __iomem *cmd_issue[MTIP_MAX_SLOT_GROUPS]; + /* + * Pointer to the beginning of the command header memory as used + * by the driver. + */ + void *command_list; + /* + * Pointer to the beginning of the command header memory as used + * by the DMA. + */ + dma_addr_t command_list_dma; + /* + * Pointer to the beginning of the RX FIS memory as used + * by the driver. + */ + void *rxfis; + /* + * Pointer to the beginning of the RX FIS memory as used + * by the DMA. + */ + dma_addr_t rxfis_dma; + /* + * Pointer to the beginning of the command table memory as used + * by the driver. + */ + void *command_table; + /* + * Pointer to the beginning of the command table memory as used + * by the DMA. + */ + dma_addr_t command_tbl_dma; + /* + * Pointer to the beginning of the identify data memory as used + * by the driver. + */ + u16 *identify; + /* + * Pointer to the beginning of the identify data memory as used + * by the DMA. + */ + dma_addr_t identify_dma; + /* + * Pointer to the beginning of a sector buffer that is used + * by the driver when issuing internal commands. + */ + u16 *sector_buffer; + /* + * Pointer to the beginning of a sector buffer that is used + * by the DMA when the driver issues internal commands. + */ + dma_addr_t sector_buffer_dma; + /* + * Bit significant, used to determine if a command slot has + * been allocated. i.e. the slot is in use. Bits are cleared + * when the command slot and all associated data structures + * are no longer needed. + */ + unsigned long allocated[SLOTBITS_IN_LONGS]; + /* + * used to queue commands when an internal command is in progress + * or error handling is active + */ + unsigned long cmds_to_issue[SLOTBITS_IN_LONGS]; + /* + * Array of command slots. Structure includes pointers to the + * command header and command table, and completion function and data + * pointers. + */ + struct mtip_cmd commands[MTIP_MAX_COMMAND_SLOTS]; + /* Used by mtip_service_thread to wait for an event */ + wait_queue_head_t svc_wait; + /* + * indicates the state of the port. Also, helps the service thread + * to determine its action on wake up. + */ + unsigned long flags; + /* + * Timer used to complete commands that have been active for too long. + */ + struct timer_list cmd_timer; + /* + * Semaphore used to block threads if there are no + * command slots available. + */ + struct semaphore cmd_slot; + /* Spinlock for working around command-issue bug. */ + spinlock_t cmd_issue_lock; +}; + +/* + * Driver private data structure. + * + * One structure is allocated per probed device. + */ +struct driver_data { + void __iomem *mmio; /* Base address of the HBA registers. */ + + int major; /* Major device number. */ + + int instance; /* Instance number. First device probed is 0, ... */ + + struct gendisk *disk; /* Pointer to our gendisk structure. */ + + struct pci_dev *pdev; /* Pointer to the PCI device structure. */ + + struct request_queue *queue; /* Our request queue. */ + + struct mtip_port *port; /* Pointer to the port data structure. */ + + /* Tasklet used to process the bottom half of the ISR. */ + struct tasklet_struct tasklet; + + unsigned product_type; /* magic value declaring the product type */ + + unsigned slot_groups; /* number of slot groups the product supports */ + + atomic_t drv_cleanup_done; /* Atomic variable for SRSI */ + + unsigned long index; /* Index to determine the disk name */ + + unsigned int ftlrebuildflag; /* FTL rebuild flag */ + + atomic_t resumeflag; /* Atomic variable to track suspend/resume */ + + struct task_struct *mtip_svc_handler; /* task_struct of svc thd */ +}; + +#endif diff --git a/drivers/block/nbd.c b/drivers/block/nbd.c index c3f0ee16594d..061427a75d37 100644 --- a/drivers/block/nbd.c +++ b/drivers/block/nbd.c @@ -34,12 +34,11 @@ #include <linux/kthread.h> #include <asm/uaccess.h> -#include <asm/system.h> #include <asm/types.h> #include <linux/nbd.h> -#define LO_MAGIC 0x68797548 +#define NBD_MAGIC 0x68797548 #ifdef NDEBUG #define dprintk(flags, fmt...) @@ -116,7 +115,7 @@ static void nbd_end_request(struct request *req) spin_unlock_irqrestore(q->queue_lock, flags); } -static void sock_shutdown(struct nbd_device *lo, int lock) +static void sock_shutdown(struct nbd_device *nbd, int lock) { /* Forcibly shutdown the socket causing all listeners * to error @@ -125,14 +124,14 @@ static void sock_shutdown(struct nbd_device *lo, int lock) * there should be a more generic interface rather than * calling socket ops directly here */ if (lock) - mutex_lock(&lo->tx_lock); - if (lo->sock) { - dev_warn(disk_to_dev(lo->disk), "shutting down socket\n"); - kernel_sock_shutdown(lo->sock, SHUT_RDWR); - lo->sock = NULL; + mutex_lock(&nbd->tx_lock); + if (nbd->sock) { + dev_warn(disk_to_dev(nbd->disk), "shutting down socket\n"); + kernel_sock_shutdown(nbd->sock, SHUT_RDWR); + nbd->sock = NULL; } if (lock) - mutex_unlock(&lo->tx_lock); + mutex_unlock(&nbd->tx_lock); } static void nbd_xmit_timeout(unsigned long arg) @@ -147,17 +146,17 @@ static void nbd_xmit_timeout(unsigned long arg) /* * Send or receive packet. */ -static int sock_xmit(struct nbd_device *lo, int send, void *buf, int size, +static int sock_xmit(struct nbd_device *nbd, int send, void *buf, int size, int msg_flags) { - struct socket *sock = lo->sock; + struct socket *sock = nbd->sock; int result; struct msghdr msg; struct kvec iov; sigset_t blocked, oldset; if (unlikely(!sock)) { - dev_err(disk_to_dev(lo->disk), + dev_err(disk_to_dev(nbd->disk), "Attempted %s on closed socket in sock_xmit\n", (send ? "send" : "recv")); return -EINVAL; @@ -181,15 +180,15 @@ static int sock_xmit(struct nbd_device *lo, int send, void *buf, int size, if (send) { struct timer_list ti; - if (lo->xmit_timeout) { + if (nbd->xmit_timeout) { init_timer(&ti); ti.function = nbd_xmit_timeout; ti.data = (unsigned long)current; - ti.expires = jiffies + lo->xmit_timeout; + ti.expires = jiffies + nbd->xmit_timeout; add_timer(&ti); } result = kernel_sendmsg(sock, &msg, &iov, 1, size); - if (lo->xmit_timeout) + if (nbd->xmit_timeout) del_timer_sync(&ti); } else result = kernel_recvmsg(sock, &msg, &iov, 1, size, @@ -201,7 +200,7 @@ static int sock_xmit(struct nbd_device *lo, int send, void *buf, int size, task_pid_nr(current), current->comm, dequeue_signal_lock(current, ¤t->blocked, &info)); result = -EINTR; - sock_shutdown(lo, !send); + sock_shutdown(nbd, !send); break; } @@ -219,18 +218,19 @@ static int sock_xmit(struct nbd_device *lo, int send, void *buf, int size, return result; } -static inline int sock_send_bvec(struct nbd_device *lo, struct bio_vec *bvec, +static inline int sock_send_bvec(struct nbd_device *nbd, struct bio_vec *bvec, int flags) { int result; void *kaddr = kmap(bvec->bv_page); - result = sock_xmit(lo, 1, kaddr + bvec->bv_offset, bvec->bv_len, flags); + result = sock_xmit(nbd, 1, kaddr + bvec->bv_offset, + bvec->bv_len, flags); kunmap(bvec->bv_page); return result; } /* always call with the tx_lock held */ -static int nbd_send_req(struct nbd_device *lo, struct request *req) +static int nbd_send_req(struct nbd_device *nbd, struct request *req) { int result, flags; struct nbd_request request; @@ -243,14 +243,14 @@ static int nbd_send_req(struct nbd_device *lo, struct request *req) memcpy(request.handle, &req, sizeof(req)); dprintk(DBG_TX, "%s: request %p: sending control (%s@%llu,%uB)\n", - lo->disk->disk_name, req, + nbd->disk->disk_name, req, nbdcmd_to_ascii(nbd_cmd(req)), (unsigned long long)blk_rq_pos(req) << 9, blk_rq_bytes(req)); - result = sock_xmit(lo, 1, &request, sizeof(request), + result = sock_xmit(nbd, 1, &request, sizeof(request), (nbd_cmd(req) == NBD_CMD_WRITE) ? MSG_MORE : 0); if (result <= 0) { - dev_err(disk_to_dev(lo->disk), + dev_err(disk_to_dev(nbd->disk), "Send control failed (result %d)\n", result); goto error_out; } @@ -267,10 +267,10 @@ static int nbd_send_req(struct nbd_device *lo, struct request *req) if (!rq_iter_last(req, iter)) flags = MSG_MORE; dprintk(DBG_TX, "%s: request %p: sending %d bytes data\n", - lo->disk->disk_name, req, bvec->bv_len); - result = sock_send_bvec(lo, bvec, flags); + nbd->disk->disk_name, req, bvec->bv_len); + result = sock_send_bvec(nbd, bvec, flags); if (result <= 0) { - dev_err(disk_to_dev(lo->disk), + dev_err(disk_to_dev(nbd->disk), "Send data failed (result %d)\n", result); goto error_out; @@ -283,25 +283,25 @@ error_out: return -EIO; } -static struct request *nbd_find_request(struct nbd_device *lo, +static struct request *nbd_find_request(struct nbd_device *nbd, struct request *xreq) { struct request *req, *tmp; int err; - err = wait_event_interruptible(lo->active_wq, lo->active_req != xreq); + err = wait_event_interruptible(nbd->active_wq, nbd->active_req != xreq); if (unlikely(err)) goto out; - spin_lock(&lo->queue_lock); - list_for_each_entry_safe(req, tmp, &lo->queue_head, queuelist) { + spin_lock(&nbd->queue_lock); + list_for_each_entry_safe(req, tmp, &nbd->queue_head, queuelist) { if (req != xreq) continue; list_del_init(&req->queuelist); - spin_unlock(&lo->queue_lock); + spin_unlock(&nbd->queue_lock); return req; } - spin_unlock(&lo->queue_lock); + spin_unlock(&nbd->queue_lock); err = -ENOENT; @@ -309,78 +309,78 @@ out: return ERR_PTR(err); } -static inline int sock_recv_bvec(struct nbd_device *lo, struct bio_vec *bvec) +static inline int sock_recv_bvec(struct nbd_device *nbd, struct bio_vec *bvec) { int result; void *kaddr = kmap(bvec->bv_page); - result = sock_xmit(lo, 0, kaddr + bvec->bv_offset, bvec->bv_len, + result = sock_xmit(nbd, 0, kaddr + bvec->bv_offset, bvec->bv_len, MSG_WAITALL); kunmap(bvec->bv_page); return result; } /* NULL returned = something went wrong, inform userspace */ -static struct request *nbd_read_stat(struct nbd_device *lo) +static struct request *nbd_read_stat(struct nbd_device *nbd) { int result; struct nbd_reply reply; struct request *req; reply.magic = 0; - result = sock_xmit(lo, 0, &reply, sizeof(reply), MSG_WAITALL); + result = sock_xmit(nbd, 0, &reply, sizeof(reply), MSG_WAITALL); if (result <= 0) { - dev_err(disk_to_dev(lo->disk), + dev_err(disk_to_dev(nbd->disk), "Receive control failed (result %d)\n", result); goto harderror; } if (ntohl(reply.magic) != NBD_REPLY_MAGIC) { - dev_err(disk_to_dev(lo->disk), "Wrong magic (0x%lx)\n", + dev_err(disk_to_dev(nbd->disk), "Wrong magic (0x%lx)\n", (unsigned long)ntohl(reply.magic)); result = -EPROTO; goto harderror; } - req = nbd_find_request(lo, *(struct request **)reply.handle); + req = nbd_find_request(nbd, *(struct request **)reply.handle); if (IS_ERR(req)) { result = PTR_ERR(req); if (result != -ENOENT) goto harderror; - dev_err(disk_to_dev(lo->disk), "Unexpected reply (%p)\n", + dev_err(disk_to_dev(nbd->disk), "Unexpected reply (%p)\n", reply.handle); result = -EBADR; goto harderror; } if (ntohl(reply.error)) { - dev_err(disk_to_dev(lo->disk), "Other side returned error (%d)\n", + dev_err(disk_to_dev(nbd->disk), "Other side returned error (%d)\n", ntohl(reply.error)); req->errors++; return req; } dprintk(DBG_RX, "%s: request %p: got reply\n", - lo->disk->disk_name, req); + nbd->disk->disk_name, req); if (nbd_cmd(req) == NBD_CMD_READ) { struct req_iterator iter; struct bio_vec *bvec; rq_for_each_segment(bvec, req, iter) { - result = sock_recv_bvec(lo, bvec); + result = sock_recv_bvec(nbd, bvec); if (result <= 0) { - dev_err(disk_to_dev(lo->disk), "Receive data failed (result %d)\n", + dev_err(disk_to_dev(nbd->disk), "Receive data failed (result %d)\n", result); req->errors++; return req; } dprintk(DBG_RX, "%s: request %p: got %d bytes data\n", - lo->disk->disk_name, req, bvec->bv_len); + nbd->disk->disk_name, req, bvec->bv_len); } } return req; harderror: - lo->harderror = result; + nbd->harderror = result; return NULL; } @@ -398,48 +398,48 @@ static struct device_attribute pid_attr = { .show = pid_show, }; -static int nbd_do_it(struct nbd_device *lo) +static int nbd_do_it(struct nbd_device *nbd) { struct request *req; int ret; - BUG_ON(lo->magic != LO_MAGIC); + BUG_ON(nbd->magic != NBD_MAGIC); - lo->pid = task_pid_nr(current); - ret = device_create_file(disk_to_dev(lo->disk), &pid_attr); + nbd->pid = task_pid_nr(current); + ret = device_create_file(disk_to_dev(nbd->disk), &pid_attr); if (ret) { - dev_err(disk_to_dev(lo->disk), "device_create_file failed!\n"); - lo->pid = 0; + dev_err(disk_to_dev(nbd->disk), "device_create_file failed!\n"); + nbd->pid = 0; return ret; } - while ((req = nbd_read_stat(lo)) != NULL) + while ((req = nbd_read_stat(nbd)) != NULL) nbd_end_request(req); - device_remove_file(disk_to_dev(lo->disk), &pid_attr); - lo->pid = 0; + device_remove_file(disk_to_dev(nbd->disk), &pid_attr); + nbd->pid = 0; return 0; } -static void nbd_clear_que(struct nbd_device *lo) +static void nbd_clear_que(struct nbd_device *nbd) { struct request *req; - BUG_ON(lo->magic != LO_MAGIC); + BUG_ON(nbd->magic != NBD_MAGIC); /* - * Because we have set lo->sock to NULL under the tx_lock, all + * Because we have set nbd->sock to NULL under the tx_lock, all * modifications to the list must have completed by now. For * the same reason, the active_req must be NULL. * * As a consequence, we don't need to take the spin lock while * purging the list here. */ - BUG_ON(lo->sock); - BUG_ON(lo->active_req); + BUG_ON(nbd->sock); + BUG_ON(nbd->active_req); - while (!list_empty(&lo->queue_head)) { - req = list_entry(lo->queue_head.next, struct request, + while (!list_empty(&nbd->queue_head)) { + req = list_entry(nbd->queue_head.next, struct request, queuelist); list_del_init(&req->queuelist); req->errors++; @@ -448,7 +448,7 @@ static void nbd_clear_que(struct nbd_device *lo) } -static void nbd_handle_req(struct nbd_device *lo, struct request *req) +static void nbd_handle_req(struct nbd_device *nbd, struct request *req) { if (req->cmd_type != REQ_TYPE_FS) goto error_out; @@ -456,8 +456,8 @@ static void nbd_handle_req(struct nbd_device *lo, struct request *req) nbd_cmd(req) = NBD_CMD_READ; if (rq_data_dir(req) == WRITE) { nbd_cmd(req) = NBD_CMD_WRITE; - if (lo->flags & NBD_READ_ONLY) { - dev_err(disk_to_dev(lo->disk), + if (nbd->flags & NBD_READ_ONLY) { + dev_err(disk_to_dev(nbd->disk), "Write on read-only\n"); goto error_out; } @@ -465,29 +465,29 @@ static void nbd_handle_req(struct nbd_device *lo, struct request *req) req->errors = 0; - mutex_lock(&lo->tx_lock); - if (unlikely(!lo->sock)) { - mutex_unlock(&lo->tx_lock); - dev_err(disk_to_dev(lo->disk), + mutex_lock(&nbd->tx_lock); + if (unlikely(!nbd->sock)) { + mutex_unlock(&nbd->tx_lock); + dev_err(disk_to_dev(nbd->disk), "Attempted send on closed socket\n"); goto error_out; } - lo->active_req = req; + nbd->active_req = req; - if (nbd_send_req(lo, req) != 0) { - dev_err(disk_to_dev(lo->disk), "Request send failed\n"); + if (nbd_send_req(nbd, req) != 0) { + dev_err(disk_to_dev(nbd->disk), "Request send failed\n"); req->errors++; nbd_end_request(req); } else { - spin_lock(&lo->queue_lock); - list_add(&req->queuelist, &lo->queue_head); - spin_unlock(&lo->queue_lock); + spin_lock(&nbd->queue_lock); + list_add(&req->queuelist, &nbd->queue_head); + spin_unlock(&nbd->queue_lock); } - lo->active_req = NULL; - mutex_unlock(&lo->tx_lock); - wake_up_all(&lo->active_wq); + nbd->active_req = NULL; + mutex_unlock(&nbd->tx_lock); + wake_up_all(&nbd->active_wq); return; @@ -498,28 +498,28 @@ error_out: static int nbd_thread(void *data) { - struct nbd_device *lo = data; + struct nbd_device *nbd = data; struct request *req; set_user_nice(current, -20); - while (!kthread_should_stop() || !list_empty(&lo->waiting_queue)) { + while (!kthread_should_stop() || !list_empty(&nbd->waiting_queue)) { /* wait for something to do */ - wait_event_interruptible(lo->waiting_wq, + wait_event_interruptible(nbd->waiting_wq, kthread_should_stop() || - !list_empty(&lo->waiting_queue)); + !list_empty(&nbd->waiting_queue)); /* extract request */ - if (list_empty(&lo->waiting_queue)) + if (list_empty(&nbd->waiting_queue)) continue; - spin_lock_irq(&lo->queue_lock); - req = list_entry(lo->waiting_queue.next, struct request, + spin_lock_irq(&nbd->queue_lock); + req = list_entry(nbd->waiting_queue.next, struct request, queuelist); list_del_init(&req->queuelist); - spin_unlock_irq(&lo->queue_lock); + spin_unlock_irq(&nbd->queue_lock); /* handle request */ - nbd_handle_req(lo, req); + nbd_handle_req(nbd, req); } return 0; } @@ -527,7 +527,7 @@ static int nbd_thread(void *data) /* * We always wait for result of write, for now. It would be nice to make it optional * in future - * if ((rq_data_dir(req) == WRITE) && (lo->flags & NBD_WRITE_NOCHK)) + * if ((rq_data_dir(req) == WRITE) && (nbd->flags & NBD_WRITE_NOCHK)) * { printk( "Warning: Ignoring result!\n"); nbd_end_request( req ); } */ @@ -536,19 +536,19 @@ static void do_nbd_request(struct request_queue *q) struct request *req; while ((req = blk_fetch_request(q)) != NULL) { - struct nbd_device *lo; + struct nbd_device *nbd; spin_unlock_irq(q->queue_lock); dprintk(DBG_BLKDEV, "%s: request %p: dequeued (flags=%x)\n", req->rq_disk->disk_name, req, req->cmd_type); - lo = req->rq_disk->private_data; + nbd = req->rq_disk->private_data; - BUG_ON(lo->magic != LO_MAGIC); + BUG_ON(nbd->magic != NBD_MAGIC); - if (unlikely(!lo->sock)) { - dev_err(disk_to_dev(lo->disk), + if (unlikely(!nbd->sock)) { + dev_err(disk_to_dev(nbd->disk), "Attempted send on closed socket\n"); req->errors++; nbd_end_request(req); @@ -556,11 +556,11 @@ static void do_nbd_request(struct request_queue *q) continue; } - spin_lock_irq(&lo->queue_lock); - list_add_tail(&req->queuelist, &lo->waiting_queue); - spin_unlock_irq(&lo->queue_lock); + spin_lock_irq(&nbd->queue_lock); + list_add_tail(&req->queuelist, &nbd->waiting_queue); + spin_unlock_irq(&nbd->queue_lock); - wake_up(&lo->waiting_wq); + wake_up(&nbd->waiting_wq); spin_lock_irq(q->queue_lock); } @@ -568,32 +568,32 @@ static void do_nbd_request(struct request_queue *q) /* Must be called with tx_lock held */ -static int __nbd_ioctl(struct block_device *bdev, struct nbd_device *lo, +static int __nbd_ioctl(struct block_device *bdev, struct nbd_device *nbd, unsigned int cmd, unsigned long arg) { switch (cmd) { case NBD_DISCONNECT: { struct request sreq; - dev_info(disk_to_dev(lo->disk), "NBD_DISCONNECT\n"); + dev_info(disk_to_dev(nbd->disk), "NBD_DISCONNECT\n"); blk_rq_init(NULL, &sreq); sreq.cmd_type = REQ_TYPE_SPECIAL; nbd_cmd(&sreq) = NBD_CMD_DISC; - if (!lo->sock) + if (!nbd->sock) return -EINVAL; - nbd_send_req(lo, &sreq); + nbd_send_req(nbd, &sreq); return 0; } case NBD_CLEAR_SOCK: { struct file *file; - lo->sock = NULL; - file = lo->file; - lo->file = NULL; - nbd_clear_que(lo); - BUG_ON(!list_empty(&lo->queue_head)); + nbd->sock = NULL; + file = nbd->file; + nbd->file = NULL; + nbd_clear_que(nbd); + BUG_ON(!list_empty(&nbd->queue_head)); if (file) fput(file); return 0; @@ -601,14 +601,14 @@ static int __nbd_ioctl(struct block_device *bdev, struct nbd_device *lo, case NBD_SET_SOCK: { struct file *file; - if (lo->file) + if (nbd->file) return -EBUSY; file = fget(arg); if (file) { struct inode *inode = file->f_path.dentry->d_inode; if (S_ISSOCK(inode->i_mode)) { - lo->file = file; - lo->sock = SOCKET_I(inode); + nbd->file = file; + nbd->sock = SOCKET_I(inode); if (max_part > 0) bdev->bd_invalidated = 1; return 0; @@ -620,29 +620,29 @@ static int __nbd_ioctl(struct block_device *bdev, struct nbd_device *lo, } case NBD_SET_BLKSIZE: - lo->blksize = arg; - lo->bytesize &= ~(lo->blksize-1); - bdev->bd_inode->i_size = lo->bytesize; - set_blocksize(bdev, lo->blksize); - set_capacity(lo->disk, lo->bytesize >> 9); + nbd->blksize = arg; + nbd->bytesize &= ~(nbd->blksize-1); + bdev->bd_inode->i_size = nbd->bytesize; + set_blocksize(bdev, nbd->blksize); + set_capacity(nbd->disk, nbd->bytesize >> 9); return 0; case NBD_SET_SIZE: - lo->bytesize = arg & ~(lo->blksize-1); - bdev->bd_inode->i_size = lo->bytesize; - set_blocksize(bdev, lo->blksize); - set_capacity(lo->disk, lo->bytesize >> 9); + nbd->bytesize = arg & ~(nbd->blksize-1); + bdev->bd_inode->i_size = nbd->bytesize; + set_blocksize(bdev, nbd->blksize); + set_capacity(nbd->disk, nbd->bytesize >> 9); return 0; case NBD_SET_TIMEOUT: - lo->xmit_timeout = arg * HZ; + nbd->xmit_timeout = arg * HZ; return 0; case NBD_SET_SIZE_BLOCKS: - lo->bytesize = ((u64) arg) * lo->blksize; - bdev->bd_inode->i_size = lo->bytesize; - set_blocksize(bdev, lo->blksize); - set_capacity(lo->disk, lo->bytesize >> 9); + nbd->bytesize = ((u64) arg) * nbd->blksize; + bdev->bd_inode->i_size = nbd->bytesize; + set_blocksize(bdev, nbd->blksize); + set_capacity(nbd->disk, nbd->bytesize >> 9); return 0; case NBD_DO_IT: { @@ -650,38 +650,38 @@ static int __nbd_ioctl(struct block_device *bdev, struct nbd_device *lo, struct file *file; int error; - if (lo->pid) + if (nbd->pid) return -EBUSY; - if (!lo->file) + if (!nbd->file) return -EINVAL; - mutex_unlock(&lo->tx_lock); + mutex_unlock(&nbd->tx_lock); - thread = kthread_create(nbd_thread, lo, lo->disk->disk_name); + thread = kthread_create(nbd_thread, nbd, nbd->disk->disk_name); if (IS_ERR(thread)) { - mutex_lock(&lo->tx_lock); + mutex_lock(&nbd->tx_lock); return PTR_ERR(thread); } wake_up_process(thread); - error = nbd_do_it(lo); + error = nbd_do_it(nbd); kthread_stop(thread); - mutex_lock(&lo->tx_lock); + mutex_lock(&nbd->tx_lock); if (error) return error; - sock_shutdown(lo, 0); - file = lo->file; - lo->file = NULL; - nbd_clear_que(lo); - dev_warn(disk_to_dev(lo->disk), "queue cleared\n"); + sock_shutdown(nbd, 0); + file = nbd->file; + nbd->file = NULL; + nbd_clear_que(nbd); + dev_warn(disk_to_dev(nbd->disk), "queue cleared\n"); if (file) fput(file); - lo->bytesize = 0; + nbd->bytesize = 0; bdev->bd_inode->i_size = 0; - set_capacity(lo->disk, 0); + set_capacity(nbd->disk, 0); if (max_part > 0) ioctl_by_bdev(bdev, BLKRRPART, 0); - return lo->harderror; + return nbd->harderror; } case NBD_CLEAR_QUE: @@ -689,14 +689,14 @@ static int __nbd_ioctl(struct block_device *bdev, struct nbd_device *lo, * This is for compatibility only. The queue is always cleared * by NBD_DO_IT or NBD_CLEAR_SOCK. */ - BUG_ON(!lo->sock && !list_empty(&lo->queue_head)); + BUG_ON(!nbd->sock && !list_empty(&nbd->queue_head)); return 0; case NBD_PRINT_DEBUG: - dev_info(disk_to_dev(lo->disk), + dev_info(disk_to_dev(nbd->disk), "next = %p, prev = %p, head = %p\n", - lo->queue_head.next, lo->queue_head.prev, - &lo->queue_head); + nbd->queue_head.next, nbd->queue_head.prev, + &nbd->queue_head); return 0; } return -ENOTTY; @@ -705,21 +705,21 @@ static int __nbd_ioctl(struct block_device *bdev, struct nbd_device *lo, static int nbd_ioctl(struct block_device *bdev, fmode_t mode, unsigned int cmd, unsigned long arg) { - struct nbd_device *lo = bdev->bd_disk->private_data; + struct nbd_device *nbd = bdev->bd_disk->private_data; int error; if (!capable(CAP_SYS_ADMIN)) return -EPERM; - BUG_ON(lo->magic != LO_MAGIC); + BUG_ON(nbd->magic != NBD_MAGIC); /* Anyone capable of this syscall can do *real bad* things */ dprintk(DBG_IOCTL, "%s: nbd_ioctl cmd=%s(0x%x) arg=%lu\n", - lo->disk->disk_name, ioctl_cmd_to_ascii(cmd), cmd, arg); + nbd->disk->disk_name, ioctl_cmd_to_ascii(cmd), cmd, arg); - mutex_lock(&lo->tx_lock); - error = __nbd_ioctl(bdev, lo, cmd, arg); - mutex_unlock(&lo->tx_lock); + mutex_lock(&nbd->tx_lock); + error = __nbd_ioctl(bdev, nbd, cmd, arg); + mutex_unlock(&nbd->tx_lock); return error; } @@ -805,7 +805,7 @@ static int __init nbd_init(void) for (i = 0; i < nbds_max; i++) { struct gendisk *disk = nbd_dev[i].disk; nbd_dev[i].file = NULL; - nbd_dev[i].magic = LO_MAGIC; + nbd_dev[i].magic = NBD_MAGIC; nbd_dev[i].flags = 0; INIT_LIST_HEAD(&nbd_dev[i].waiting_queue); spin_lock_init(&nbd_dev[i].queue_lock); diff --git a/drivers/block/nvme.c b/drivers/block/nvme.c new file mode 100644 index 000000000000..38a2d0631882 --- /dev/null +++ b/drivers/block/nvme.c @@ -0,0 +1,1740 @@ +/* + * NVM Express device driver + * Copyright (c) 2011, Intel Corporation. + * + * This program is free software; you can redistribute it and/or modify it + * under the terms and conditions of the GNU General Public License, + * version 2, as published by the Free Software Foundation. + * + * This program is distributed in the hope it will be useful, but WITHOUT + * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or + * FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for + * more details. + * + * You should have received a copy of the GNU General Public License along with + * this program; if not, write to the Free Software Foundation, Inc., + * 51 Franklin St - Fifth Floor, Boston, MA 02110-1301 USA. + */ + +#include <linux/nvme.h> +#include <linux/bio.h> +#include <linux/bitops.h> +#include <linux/blkdev.h> +#include <linux/delay.h> +#include <linux/errno.h> +#include <linux/fs.h> +#include <linux/genhd.h> +#include <linux/idr.h> +#include <linux/init.h> +#include <linux/interrupt.h> +#include <linux/io.h> +#include <linux/kdev_t.h> +#include <linux/kthread.h> +#include <linux/kernel.h> +#include <linux/mm.h> +#include <linux/module.h> +#include <linux/moduleparam.h> +#include <linux/pci.h> +#include <linux/poison.h> +#include <linux/sched.h> +#include <linux/slab.h> +#include <linux/types.h> + +#include <asm-generic/io-64-nonatomic-lo-hi.h> + +#define NVME_Q_DEPTH 1024 +#define SQ_SIZE(depth) (depth * sizeof(struct nvme_command)) +#define CQ_SIZE(depth) (depth * sizeof(struct nvme_completion)) +#define NVME_MINORS 64 +#define NVME_IO_TIMEOUT (5 * HZ) +#define ADMIN_TIMEOUT (60 * HZ) + +static int nvme_major; +module_param(nvme_major, int, 0); + +static int use_threaded_interrupts; +module_param(use_threaded_interrupts, int, 0); + +static DEFINE_SPINLOCK(dev_list_lock); +static LIST_HEAD(dev_list); +static struct task_struct *nvme_thread; + +/* + * Represents an NVM Express device. Each nvme_dev is a PCI function. + */ +struct nvme_dev { + struct list_head node; + struct nvme_queue **queues; + u32 __iomem *dbs; + struct pci_dev *pci_dev; + struct dma_pool *prp_page_pool; + struct dma_pool *prp_small_pool; + int instance; + int queue_count; + int db_stride; + u32 ctrl_config; + struct msix_entry *entry; + struct nvme_bar __iomem *bar; + struct list_head namespaces; + char serial[20]; + char model[40]; + char firmware_rev[8]; +}; + +/* + * An NVM Express namespace is equivalent to a SCSI LUN + */ +struct nvme_ns { + struct list_head list; + + struct nvme_dev *dev; + struct request_queue *queue; + struct gendisk *disk; + + int ns_id; + int lba_shift; +}; + +/* + * An NVM Express queue. Each device has at least two (one for admin + * commands and one for I/O commands). + */ +struct nvme_queue { + struct device *q_dmadev; + struct nvme_dev *dev; + spinlock_t q_lock; + struct nvme_command *sq_cmds; + volatile struct nvme_completion *cqes; + dma_addr_t sq_dma_addr; + dma_addr_t cq_dma_addr; + wait_queue_head_t sq_full; + wait_queue_t sq_cong_wait; + struct bio_list sq_cong; + u32 __iomem *q_db; + u16 q_depth; + u16 cq_vector; + u16 sq_head; + u16 sq_tail; + u16 cq_head; + u16 cq_phase; + unsigned long cmdid_data[]; +}; + +/* + * Check we didin't inadvertently grow the command struct + */ +static inline void _nvme_check_size(void) +{ + BUILD_BUG_ON(sizeof(struct nvme_rw_command) != 64); + BUILD_BUG_ON(sizeof(struct nvme_create_cq) != 64); + BUILD_BUG_ON(sizeof(struct nvme_create_sq) != 64); + BUILD_BUG_ON(sizeof(struct nvme_delete_queue) != 64); + BUILD_BUG_ON(sizeof(struct nvme_features) != 64); + BUILD_BUG_ON(sizeof(struct nvme_command) != 64); + BUILD_BUG_ON(sizeof(struct nvme_id_ctrl) != 4096); + BUILD_BUG_ON(sizeof(struct nvme_id_ns) != 4096); + BUILD_BUG_ON(sizeof(struct nvme_lba_range_type) != 64); +} + +typedef void (*nvme_completion_fn)(struct nvme_dev *, void *, + struct nvme_completion *); + +struct nvme_cmd_info { + nvme_completion_fn fn; + void *ctx; + unsigned long timeout; +}; + +static struct nvme_cmd_info *nvme_cmd_info(struct nvme_queue *nvmeq) +{ + return (void *)&nvmeq->cmdid_data[BITS_TO_LONGS(nvmeq->q_depth)]; +} + +/** + * alloc_cmdid() - Allocate a Command ID + * @nvmeq: The queue that will be used for this command + * @ctx: A pointer that will be passed to the handler + * @handler: The function to call on completion + * + * Allocate a Command ID for a queue. The data passed in will + * be passed to the completion handler. This is implemented by using + * the bottom two bits of the ctx pointer to store the handler ID. + * Passing in a pointer that's not 4-byte aligned will cause a BUG. + * We can change this if it becomes a problem. + * + * May be called with local interrupts disabled and the q_lock held, + * or with interrupts enabled and no locks held. + */ +static int alloc_cmdid(struct nvme_queue *nvmeq, void *ctx, + nvme_completion_fn handler, unsigned timeout) +{ + int depth = nvmeq->q_depth - 1; + struct nvme_cmd_info *info = nvme_cmd_info(nvmeq); + int cmdid; + + do { + cmdid = find_first_zero_bit(nvmeq->cmdid_data, depth); + if (cmdid >= depth) + return -EBUSY; + } while (test_and_set_bit(cmdid, nvmeq->cmdid_data)); + + info[cmdid].fn = handler; + info[cmdid].ctx = ctx; + info[cmdid].timeout = jiffies + timeout; + return cmdid; +} + +static int alloc_cmdid_killable(struct nvme_queue *nvmeq, void *ctx, + nvme_completion_fn handler, unsigned timeout) +{ + int cmdid; + wait_event_killable(nvmeq->sq_full, + (cmdid = alloc_cmdid(nvmeq, ctx, handler, timeout)) >= 0); + return (cmdid < 0) ? -EINTR : cmdid; +} + +/* Special values must be less than 0x1000 */ +#define CMD_CTX_BASE ((void *)POISON_POINTER_DELTA) +#define CMD_CTX_CANCELLED (0x30C + CMD_CTX_BASE) +#define CMD_CTX_COMPLETED (0x310 + CMD_CTX_BASE) +#define CMD_CTX_INVALID (0x314 + CMD_CTX_BASE) +#define CMD_CTX_FLUSH (0x318 + CMD_CTX_BASE) + +static void special_completion(struct nvme_dev *dev, void *ctx, + struct nvme_completion *cqe) +{ + if (ctx == CMD_CTX_CANCELLED) + return; + if (ctx == CMD_CTX_FLUSH) + return; + if (ctx == CMD_CTX_COMPLETED) { + dev_warn(&dev->pci_dev->dev, + "completed id %d twice on queue %d\n", + cqe->command_id, le16_to_cpup(&cqe->sq_id)); + return; + } + if (ctx == CMD_CTX_INVALID) { + dev_warn(&dev->pci_dev->dev, + "invalid id %d completed on queue %d\n", + cqe->command_id, le16_to_cpup(&cqe->sq_id)); + return; + } + + dev_warn(&dev->pci_dev->dev, "Unknown special completion %p\n", ctx); +} + +/* + * Called with local interrupts disabled and the q_lock held. May not sleep. + */ +static void *free_cmdid(struct nvme_queue *nvmeq, int cmdid, + nvme_completion_fn *fn) +{ + void *ctx; + struct nvme_cmd_info *info = nvme_cmd_info(nvmeq); + + if (cmdid >= nvmeq->q_depth) { + *fn = special_completion; + return CMD_CTX_INVALID; + } + *fn = info[cmdid].fn; + ctx = info[cmdid].ctx; + info[cmdid].fn = special_completion; + info[cmdid].ctx = CMD_CTX_COMPLETED; + clear_bit(cmdid, nvmeq->cmdid_data); + wake_up(&nvmeq->sq_full); + return ctx; +} + +static void *cancel_cmdid(struct nvme_queue *nvmeq, int cmdid, + nvme_completion_fn *fn) +{ + void *ctx; + struct nvme_cmd_info *info = nvme_cmd_info(nvmeq); + if (fn) + *fn = info[cmdid].fn; + ctx = info[cmdid].ctx; + info[cmdid].fn = special_completion; + info[cmdid].ctx = CMD_CTX_CANCELLED; + return ctx; +} + +static struct nvme_queue *get_nvmeq(struct nvme_dev *dev) +{ + return dev->queues[get_cpu() + 1]; +} + +static void put_nvmeq(struct nvme_queue *nvmeq) +{ + put_cpu(); +} + +/** + * nvme_submit_cmd() - Copy a command into a queue and ring the doorbell + * @nvmeq: The queue to use + * @cmd: The command to send + * + * Safe to use from interrupt context + */ +static int nvme_submit_cmd(struct nvme_queue *nvmeq, struct nvme_command *cmd) +{ + unsigned long flags; + u16 tail; + spin_lock_irqsave(&nvmeq->q_lock, flags); + tail = nvmeq->sq_tail; + memcpy(&nvmeq->sq_cmds[tail], cmd, sizeof(*cmd)); + if (++tail == nvmeq->q_depth) + tail = 0; + writel(tail, nvmeq->q_db); + nvmeq->sq_tail = tail; + spin_unlock_irqrestore(&nvmeq->q_lock, flags); + + return 0; +} + +/* + * The nvme_iod describes the data in an I/O, including the list of PRP + * entries. You can't see it in this data structure because C doesn't let + * me express that. Use nvme_alloc_iod to ensure there's enough space + * allocated to store the PRP list. + */ +struct nvme_iod { + void *private; /* For the use of the submitter of the I/O */ + int npages; /* In the PRP list. 0 means small pool in use */ + int offset; /* Of PRP list */ + int nents; /* Used in scatterlist */ + int length; /* Of data, in bytes */ + dma_addr_t first_dma; + struct scatterlist sg[0]; +}; + +static __le64 **iod_list(struct nvme_iod *iod) +{ + return ((void *)iod) + iod->offset; +} + +/* + * Will slightly overestimate the number of pages needed. This is OK + * as it only leads to a small amount of wasted memory for the lifetime of + * the I/O. + */ +static int nvme_npages(unsigned size) +{ + unsigned nprps = DIV_ROUND_UP(size + PAGE_SIZE, PAGE_SIZE); + return DIV_ROUND_UP(8 * nprps, PAGE_SIZE - 8); +} + +static struct nvme_iod * +nvme_alloc_iod(unsigned nseg, unsigned nbytes, gfp_t gfp) +{ + struct nvme_iod *iod = kmalloc(sizeof(struct nvme_iod) + + sizeof(__le64 *) * nvme_npages(nbytes) + + sizeof(struct scatterlist) * nseg, gfp); + + if (iod) { + iod->offset = offsetof(struct nvme_iod, sg[nseg]); + iod->npages = -1; + iod->length = nbytes; + } + + return iod; +} + +static void nvme_free_iod(struct nvme_dev *dev, struct nvme_iod *iod) +{ + const int last_prp = PAGE_SIZE / 8 - 1; + int i; + __le64 **list = iod_list(iod); + dma_addr_t prp_dma = iod->first_dma; + + if (iod->npages == 0) + dma_pool_free(dev->prp_small_pool, list[0], prp_dma); + for (i = 0; i < iod->npages; i++) { + __le64 *prp_list = list[i]; + dma_addr_t next_prp_dma = le64_to_cpu(prp_list[last_prp]); + dma_pool_free(dev->prp_page_pool, prp_list, prp_dma); + prp_dma = next_prp_dma; + } + kfree(iod); +} + +static void requeue_bio(struct nvme_dev *dev, struct bio *bio) +{ + struct nvme_queue *nvmeq = get_nvmeq(dev); + if (bio_list_empty(&nvmeq->sq_cong)) + add_wait_queue(&nvmeq->sq_full, &nvmeq->sq_cong_wait); + bio_list_add(&nvmeq->sq_cong, bio); + put_nvmeq(nvmeq); + wake_up_process(nvme_thread); +} + +static void bio_completion(struct nvme_dev *dev, void *ctx, + struct nvme_completion *cqe) +{ + struct nvme_iod *iod = ctx; + struct bio *bio = iod->private; + u16 status = le16_to_cpup(&cqe->status) >> 1; + + dma_unmap_sg(&dev->pci_dev->dev, iod->sg, iod->nents, + bio_data_dir(bio) ? DMA_TO_DEVICE : DMA_FROM_DEVICE); + nvme_free_iod(dev, iod); + if (status) { + bio_endio(bio, -EIO); + } else if (bio->bi_vcnt > bio->bi_idx) { + requeue_bio(dev, bio); + } else { + bio_endio(bio, 0); + } +} + +/* length is in bytes. gfp flags indicates whether we may sleep. */ +static int nvme_setup_prps(struct nvme_dev *dev, + struct nvme_common_command *cmd, struct nvme_iod *iod, + int total_len, gfp_t gfp) +{ + struct dma_pool *pool; + int length = total_len; + struct scatterlist *sg = iod->sg; + int dma_len = sg_dma_len(sg); + u64 dma_addr = sg_dma_address(sg); + int offset = offset_in_page(dma_addr); + __le64 *prp_list; + __le64 **list = iod_list(iod); + dma_addr_t prp_dma; + int nprps, i; + + cmd->prp1 = cpu_to_le64(dma_addr); + length -= (PAGE_SIZE - offset); + if (length <= 0) + return total_len; + + dma_len -= (PAGE_SIZE - offset); + if (dma_len) { + dma_addr += (PAGE_SIZE - offset); + } else { + sg = sg_next(sg); + dma_addr = sg_dma_address(sg); + dma_len = sg_dma_len(sg); + } + + if (length <= PAGE_SIZE) { + cmd->prp2 = cpu_to_le64(dma_addr); + return total_len; + } + + nprps = DIV_ROUND_UP(length, PAGE_SIZE); + if (nprps <= (256 / 8)) { + pool = dev->prp_small_pool; + iod->npages = 0; + } else { + pool = dev->prp_page_pool; + iod->npages = 1; + } + + prp_list = dma_pool_alloc(pool, gfp, &prp_dma); + if (!prp_list) { + cmd->prp2 = cpu_to_le64(dma_addr); + iod->npages = -1; + return (total_len - length) + PAGE_SIZE; + } + list[0] = prp_list; + iod->first_dma = prp_dma; + cmd->prp2 = cpu_to_le64(prp_dma); + i = 0; + for (;;) { + if (i == PAGE_SIZE / 8) { + __le64 *old_prp_list = prp_list; + prp_list = dma_pool_alloc(pool, gfp, &prp_dma); + if (!prp_list) + return total_len - length; + list[iod->npages++] = prp_list; + prp_list[0] = old_prp_list[i - 1]; + old_prp_list[i - 1] = cpu_to_le64(prp_dma); + i = 1; + } + prp_list[i++] = cpu_to_le64(dma_addr); + dma_len -= PAGE_SIZE; + dma_addr += PAGE_SIZE; + length -= PAGE_SIZE; + if (length <= 0) + break; + if (dma_len > 0) + continue; + BUG_ON(dma_len < 0); + sg = sg_next(sg); + dma_addr = sg_dma_address(sg); + dma_len = sg_dma_len(sg); + } + + return total_len; +} + +/* NVMe scatterlists require no holes in the virtual address */ +#define BIOVEC_NOT_VIRT_MERGEABLE(vec1, vec2) ((vec2)->bv_offset || \ + (((vec1)->bv_offset + (vec1)->bv_len) % PAGE_SIZE)) + +static int nvme_map_bio(struct device *dev, struct nvme_iod *iod, + struct bio *bio, enum dma_data_direction dma_dir, int psegs) +{ + struct bio_vec *bvec, *bvprv = NULL; + struct scatterlist *sg = NULL; + int i, old_idx, length = 0, nsegs = 0; + + sg_init_table(iod->sg, psegs); + old_idx = bio->bi_idx; + bio_for_each_segment(bvec, bio, i) { + if (bvprv && BIOVEC_PHYS_MERGEABLE(bvprv, bvec)) { + sg->length += bvec->bv_len; + } else { + if (bvprv && BIOVEC_NOT_VIRT_MERGEABLE(bvprv, bvec)) + break; + sg = sg ? sg + 1 : iod->sg; + sg_set_page(sg, bvec->bv_page, bvec->bv_len, + bvec->bv_offset); + nsegs++; + } + length += bvec->bv_len; + bvprv = bvec; + } + bio->bi_idx = i; + iod->nents = nsegs; + sg_mark_end(sg); + if (dma_map_sg(dev, iod->sg, iod->nents, dma_dir) == 0) { + bio->bi_idx = old_idx; + return -ENOMEM; + } + return length; +} + +static int nvme_submit_flush(struct nvme_queue *nvmeq, struct nvme_ns *ns, + int cmdid) +{ + struct nvme_command *cmnd = &nvmeq->sq_cmds[nvmeq->sq_tail]; + + memset(cmnd, 0, sizeof(*cmnd)); + cmnd->common.opcode = nvme_cmd_flush; + cmnd->common.command_id = cmdid; + cmnd->common.nsid = cpu_to_le32(ns->ns_id); + + if (++nvmeq->sq_tail == nvmeq->q_depth) + nvmeq->sq_tail = 0; + writel(nvmeq->sq_tail, nvmeq->q_db); + + return 0; +} + +static int nvme_submit_flush_data(struct nvme_queue *nvmeq, struct nvme_ns *ns) +{ + int cmdid = alloc_cmdid(nvmeq, (void *)CMD_CTX_FLUSH, + special_completion, NVME_IO_TIMEOUT); + if (unlikely(cmdid < 0)) + return cmdid; + + return nvme_submit_flush(nvmeq, ns, cmdid); +} + +/* + * Called with local interrupts disabled and the q_lock held. May not sleep. + */ +static int nvme_submit_bio_queue(struct nvme_queue *nvmeq, struct nvme_ns *ns, + struct bio *bio) +{ + struct nvme_command *cmnd; + struct nvme_iod *iod; + enum dma_data_direction dma_dir; + int cmdid, length, result = -ENOMEM; + u16 control; + u32 dsmgmt; + int psegs = bio_phys_segments(ns->queue, bio); + + if ((bio->bi_rw & REQ_FLUSH) && psegs) { + result = nvme_submit_flush_data(nvmeq, ns); + if (result) + return result; + } + + iod = nvme_alloc_iod(psegs, bio->bi_size, GFP_ATOMIC); + if (!iod) + goto nomem; + iod->private = bio; + + result = -EBUSY; + cmdid = alloc_cmdid(nvmeq, iod, bio_completion, NVME_IO_TIMEOUT); + if (unlikely(cmdid < 0)) + goto free_iod; + + if ((bio->bi_rw & REQ_FLUSH) && !psegs) + return nvme_submit_flush(nvmeq, ns, cmdid); + + control = 0; + if (bio->bi_rw & REQ_FUA) + control |= NVME_RW_FUA; + if (bio->bi_rw & (REQ_FAILFAST_DEV | REQ_RAHEAD)) + control |= NVME_RW_LR; + + dsmgmt = 0; + if (bio->bi_rw & REQ_RAHEAD) + dsmgmt |= NVME_RW_DSM_FREQ_PREFETCH; + + cmnd = &nvmeq->sq_cmds[nvmeq->sq_tail]; + + memset(cmnd, 0, sizeof(*cmnd)); + if (bio_data_dir(bio)) { + cmnd->rw.opcode = nvme_cmd_write; + dma_dir = DMA_TO_DEVICE; + } else { + cmnd->rw.opcode = nvme_cmd_read; + dma_dir = DMA_FROM_DEVICE; + } + + result = nvme_map_bio(nvmeq->q_dmadev, iod, bio, dma_dir, psegs); + if (result < 0) + goto free_iod; + length = result; + + cmnd->rw.command_id = cmdid; + cmnd->rw.nsid = cpu_to_le32(ns->ns_id); + length = nvme_setup_prps(nvmeq->dev, &cmnd->common, iod, length, + GFP_ATOMIC); + cmnd->rw.slba = cpu_to_le64(bio->bi_sector >> (ns->lba_shift - 9)); + cmnd->rw.length = cpu_to_le16((length >> ns->lba_shift) - 1); + cmnd->rw.control = cpu_to_le16(control); + cmnd->rw.dsmgmt = cpu_to_le32(dsmgmt); + + bio->bi_sector += length >> 9; + + if (++nvmeq->sq_tail == nvmeq->q_depth) + nvmeq->sq_tail = 0; + writel(nvmeq->sq_tail, nvmeq->q_db); + + return 0; + + free_iod: + nvme_free_iod(nvmeq->dev, iod); + nomem: + return result; +} + +static void nvme_make_request(struct request_queue *q, struct bio *bio) +{ + struct nvme_ns *ns = q->queuedata; + struct nvme_queue *nvmeq = get_nvmeq(ns->dev); + int result = -EBUSY; + + spin_lock_irq(&nvmeq->q_lock); + if (bio_list_empty(&nvmeq->sq_cong)) + result = nvme_submit_bio_queue(nvmeq, ns, bio); + if (unlikely(result)) { + if (bio_list_empty(&nvmeq->sq_cong)) + add_wait_queue(&nvmeq->sq_full, &nvmeq->sq_cong_wait); + bio_list_add(&nvmeq->sq_cong, bio); + } + + spin_unlock_irq(&nvmeq->q_lock); + put_nvmeq(nvmeq); +} + +static irqreturn_t nvme_process_cq(struct nvme_queue *nvmeq) +{ + u16 head, phase; + + head = nvmeq->cq_head; + phase = nvmeq->cq_phase; + + for (;;) { + void *ctx; + nvme_completion_fn fn; + struct nvme_completion cqe = nvmeq->cqes[head]; + if ((le16_to_cpu(cqe.status) & 1) != phase) + break; + nvmeq->sq_head = le16_to_cpu(cqe.sq_head); + if (++head == nvmeq->q_depth) { + head = 0; + phase = !phase; + } + + ctx = free_cmdid(nvmeq, cqe.command_id, &fn); + fn(nvmeq->dev, ctx, &cqe); + } + + /* If the controller ignores the cq head doorbell and continuously + * writes to the queue, it is theoretically possible to wrap around + * the queue twice and mistakenly return IRQ_NONE. Linux only + * requires that 0.1% of your interrupts are handled, so this isn't + * a big problem. + */ + if (head == nvmeq->cq_head && phase == nvmeq->cq_phase) + return IRQ_NONE; + + writel(head, nvmeq->q_db + (1 << nvmeq->dev->db_stride)); + nvmeq->cq_head = head; + nvmeq->cq_phase = phase; + + return IRQ_HANDLED; +} + +static irqreturn_t nvme_irq(int irq, void *data) +{ + irqreturn_t result; + struct nvme_queue *nvmeq = data; + spin_lock(&nvmeq->q_lock); + result = nvme_process_cq(nvmeq); + spin_unlock(&nvmeq->q_lock); + return result; +} + +static irqreturn_t nvme_irq_check(int irq, void *data) +{ + struct nvme_queue *nvmeq = data; + struct nvme_completion cqe = nvmeq->cqes[nvmeq->cq_head]; + if ((le16_to_cpu(cqe.status) & 1) != nvmeq->cq_phase) + return IRQ_NONE; + return IRQ_WAKE_THREAD; +} + +static void nvme_abort_command(struct nvme_queue *nvmeq, int cmdid) +{ + spin_lock_irq(&nvmeq->q_lock); + cancel_cmdid(nvmeq, cmdid, NULL); + spin_unlock_irq(&nvmeq->q_lock); +} + +struct sync_cmd_info { + struct task_struct *task; + u32 result; + int status; +}; + +static void sync_completion(struct nvme_dev *dev, void *ctx, + struct nvme_completion *cqe) +{ + struct sync_cmd_info *cmdinfo = ctx; + cmdinfo->result = le32_to_cpup(&cqe->result); + cmdinfo->status = le16_to_cpup(&cqe->status) >> 1; + wake_up_process(cmdinfo->task); +} + +/* + * Returns 0 on success. If the result is negative, it's a Linux error code; + * if the result is positive, it's an NVM Express status code + */ +static int nvme_submit_sync_cmd(struct nvme_queue *nvmeq, + struct nvme_command *cmd, u32 *result, unsigned timeout) +{ + int cmdid; + struct sync_cmd_info cmdinfo; + + cmdinfo.task = current; + cmdinfo.status = -EINTR; + + cmdid = alloc_cmdid_killable(nvmeq, &cmdinfo, sync_completion, + timeout); + if (cmdid < 0) + return cmdid; + cmd->common.command_id = cmdid; + + set_current_state(TASK_KILLABLE); + nvme_submit_cmd(nvmeq, cmd); + schedule(); + + if (cmdinfo.status == -EINTR) { + nvme_abort_command(nvmeq, cmdid); + return -EINTR; + } + + if (result) + *result = cmdinfo.result; + + return cmdinfo.status; +} + +static int nvme_submit_admin_cmd(struct nvme_dev *dev, struct nvme_command *cmd, + u32 *result) +{ + return nvme_submit_sync_cmd(dev->queues[0], cmd, result, ADMIN_TIMEOUT); +} + +static int adapter_delete_queue(struct nvme_dev *dev, u8 opcode, u16 id) +{ + int status; + struct nvme_command c; + + memset(&c, 0, sizeof(c)); + c.delete_queue.opcode = opcode; + c.delete_queue.qid = cpu_to_le16(id); + + status = nvme_submit_admin_cmd(dev, &c, NULL); + if (status) + return -EIO; + return 0; +} + +static int adapter_alloc_cq(struct nvme_dev *dev, u16 qid, + struct nvme_queue *nvmeq) +{ + int status; + struct nvme_command c; + int flags = NVME_QUEUE_PHYS_CONTIG | NVME_CQ_IRQ_ENABLED; + + memset(&c, 0, sizeof(c)); + c.create_cq.opcode = nvme_admin_create_cq; + c.create_cq.prp1 = cpu_to_le64(nvmeq->cq_dma_addr); + c.create_cq.cqid = cpu_to_le16(qid); + c.create_cq.qsize = cpu_to_le16(nvmeq->q_depth - 1); + c.create_cq.cq_flags = cpu_to_le16(flags); + c.create_cq.irq_vector = cpu_to_le16(nvmeq->cq_vector); + + status = nvme_submit_admin_cmd(dev, &c, NULL); + if (status) + return -EIO; + return 0; +} + +static int adapter_alloc_sq(struct nvme_dev *dev, u16 qid, + struct nvme_queue *nvmeq) +{ + int status; + struct nvme_command c; + int flags = NVME_QUEUE_PHYS_CONTIG | NVME_SQ_PRIO_MEDIUM; + + memset(&c, 0, sizeof(c)); + c.create_sq.opcode = nvme_admin_create_sq; + c.create_sq.prp1 = cpu_to_le64(nvmeq->sq_dma_addr); + c.create_sq.sqid = cpu_to_le16(qid); + c.create_sq.qsize = cpu_to_le16(nvmeq->q_depth - 1); + c.create_sq.sq_flags = cpu_to_le16(flags); + c.create_sq.cqid = cpu_to_le16(qid); + + status = nvme_submit_admin_cmd(dev, &c, NULL); + if (status) + return -EIO; + return 0; +} + +static int adapter_delete_cq(struct nvme_dev *dev, u16 cqid) +{ + return adapter_delete_queue(dev, nvme_admin_delete_cq, cqid); +} + +static int adapter_delete_sq(struct nvme_dev *dev, u16 sqid) +{ + return adapter_delete_queue(dev, nvme_admin_delete_sq, sqid); +} + +static int nvme_identify(struct nvme_dev *dev, unsigned nsid, unsigned cns, + dma_addr_t dma_addr) +{ + struct nvme_command c; + + memset(&c, 0, sizeof(c)); + c.identify.opcode = nvme_admin_identify; + c.identify.nsid = cpu_to_le32(nsid); + c.identify.prp1 = cpu_to_le64(dma_addr); + c.identify.cns = cpu_to_le32(cns); + + return nvme_submit_admin_cmd(dev, &c, NULL); +} + +static int nvme_get_features(struct nvme_dev *dev, unsigned fid, + unsigned dword11, dma_addr_t dma_addr) +{ + struct nvme_command c; + + memset(&c, 0, sizeof(c)); + c.features.opcode = nvme_admin_get_features; + c.features.prp1 = cpu_to_le64(dma_addr); + c.features.fid = cpu_to_le32(fid); + c.features.dword11 = cpu_to_le32(dword11); + + return nvme_submit_admin_cmd(dev, &c, NULL); +} + +static int nvme_set_features(struct nvme_dev *dev, unsigned fid, + unsigned dword11, dma_addr_t dma_addr, u32 *result) +{ + struct nvme_command c; + + memset(&c, 0, sizeof(c)); + c.features.opcode = nvme_admin_set_features; + c.features.prp1 = cpu_to_le64(dma_addr); + c.features.fid = cpu_to_le32(fid); + c.features.dword11 = cpu_to_le32(dword11); + + return nvme_submit_admin_cmd(dev, &c, result); +} + +static void nvme_free_queue(struct nvme_dev *dev, int qid) +{ + struct nvme_queue *nvmeq = dev->queues[qid]; + int vector = dev->entry[nvmeq->cq_vector].vector; + + irq_set_affinity_hint(vector, NULL); + free_irq(vector, nvmeq); + + /* Don't tell the adapter to delete the admin queue */ + if (qid) { + adapter_delete_sq(dev, qid); + adapter_delete_cq(dev, qid); + } + + dma_free_coherent(nvmeq->q_dmadev, CQ_SIZE(nvmeq->q_depth), + (void *)nvmeq->cqes, nvmeq->cq_dma_addr); + dma_free_coherent(nvmeq->q_dmadev, SQ_SIZE(nvmeq->q_depth), + nvmeq->sq_cmds, nvmeq->sq_dma_addr); + kfree(nvmeq); +} + +static struct nvme_queue *nvme_alloc_queue(struct nvme_dev *dev, int qid, + int depth, int vector) +{ + struct device *dmadev = &dev->pci_dev->dev; + unsigned extra = (depth / 8) + (depth * sizeof(struct nvme_cmd_info)); + struct nvme_queue *nvmeq = kzalloc(sizeof(*nvmeq) + extra, GFP_KERNEL); + if (!nvmeq) + return NULL; + + nvmeq->cqes = dma_alloc_coherent(dmadev, CQ_SIZE(depth), + &nvmeq->cq_dma_addr, GFP_KERNEL); + if (!nvmeq->cqes) + goto free_nvmeq; + memset((void *)nvmeq->cqes, 0, CQ_SIZE(depth)); + + nvmeq->sq_cmds = dma_alloc_coherent(dmadev, SQ_SIZE(depth), + &nvmeq->sq_dma_addr, GFP_KERNEL); + if (!nvmeq->sq_cmds) + goto free_cqdma; + + nvmeq->q_dmadev = dmadev; + nvmeq->dev = dev; + spin_lock_init(&nvmeq->q_lock); + nvmeq->cq_head = 0; + nvmeq->cq_phase = 1; + init_waitqueue_head(&nvmeq->sq_full); + init_waitqueue_entry(&nvmeq->sq_cong_wait, nvme_thread); + bio_list_init(&nvmeq->sq_cong); + nvmeq->q_db = &dev->dbs[qid << (dev->db_stride + 1)]; + nvmeq->q_depth = depth; + nvmeq->cq_vector = vector; + + return nvmeq; + + free_cqdma: + dma_free_coherent(dmadev, CQ_SIZE(nvmeq->q_depth), (void *)nvmeq->cqes, + nvmeq->cq_dma_addr); + free_nvmeq: + kfree(nvmeq); + return NULL; +} + +static int queue_request_irq(struct nvme_dev *dev, struct nvme_queue *nvmeq, + const char *name) +{ + if (use_threaded_interrupts) + return request_threaded_irq(dev->entry[nvmeq->cq_vector].vector, + nvme_irq_check, nvme_irq, + IRQF_DISABLED | IRQF_SHARED, + name, nvmeq); + return request_irq(dev->entry[nvmeq->cq_vector].vector, nvme_irq, + IRQF_DISABLED | IRQF_SHARED, name, nvmeq); +} + +static __devinit struct nvme_queue *nvme_create_queue(struct nvme_dev *dev, + int qid, int cq_size, int vector) +{ + int result; + struct nvme_queue *nvmeq = nvme_alloc_queue(dev, qid, cq_size, vector); + + if (!nvmeq) + return ERR_PTR(-ENOMEM); + + result = adapter_alloc_cq(dev, qid, nvmeq); + if (result < 0) + goto free_nvmeq; + + result = adapter_alloc_sq(dev, qid, nvmeq); + if (result < 0) + goto release_cq; + + result = queue_request_irq(dev, nvmeq, "nvme"); + if (result < 0) + goto release_sq; + + return nvmeq; + + release_sq: + adapter_delete_sq(dev, qid); + release_cq: + adapter_delete_cq(dev, qid); + free_nvmeq: + dma_free_coherent(nvmeq->q_dmadev, CQ_SIZE(nvmeq->q_depth), + (void *)nvmeq->cqes, nvmeq->cq_dma_addr); + dma_free_coherent(nvmeq->q_dmadev, SQ_SIZE(nvmeq->q_depth), + nvmeq->sq_cmds, nvmeq->sq_dma_addr); + kfree(nvmeq); + return ERR_PTR(result); +} + +static int __devinit nvme_configure_admin_queue(struct nvme_dev *dev) +{ + int result; + u32 aqa; + u64 cap; + unsigned long timeout; + struct nvme_queue *nvmeq; + + dev->dbs = ((void __iomem *)dev->bar) + 4096; + + nvmeq = nvme_alloc_queue(dev, 0, 64, 0); + if (!nvmeq) + return -ENOMEM; + + aqa = nvmeq->q_depth - 1; + aqa |= aqa << 16; + + dev->ctrl_config = NVME_CC_ENABLE | NVME_CC_CSS_NVM; + dev->ctrl_config |= (PAGE_SHIFT - 12) << NVME_CC_MPS_SHIFT; + dev->ctrl_config |= NVME_CC_ARB_RR | NVME_CC_SHN_NONE; + dev->ctrl_config |= NVME_CC_IOSQES | NVME_CC_IOCQES; + + writel(0, &dev->bar->cc); + writel(aqa, &dev->bar->aqa); + writeq(nvmeq->sq_dma_addr, &dev->bar->asq); + writeq(nvmeq->cq_dma_addr, &dev->bar->acq); + writel(dev->ctrl_config, &dev->bar->cc); + + cap = readq(&dev->bar->cap); + timeout = ((NVME_CAP_TIMEOUT(cap) + 1) * HZ / 2) + jiffies; + dev->db_stride = NVME_CAP_STRIDE(cap); + + while (!(readl(&dev->bar->csts) & NVME_CSTS_RDY)) { + msleep(100); + if (fatal_signal_pending(current)) + return -EINTR; + if (time_after(jiffies, timeout)) { + dev_err(&dev->pci_dev->dev, + "Device not ready; aborting initialisation\n"); + return -ENODEV; + } + } + + result = queue_request_irq(dev, nvmeq, "nvme admin"); + dev->queues[0] = nvmeq; + return result; +} + +static struct nvme_iod *nvme_map_user_pages(struct nvme_dev *dev, int write, + unsigned long addr, unsigned length) +{ + int i, err, count, nents, offset; + struct scatterlist *sg; + struct page **pages; + struct nvme_iod *iod; + + if (addr & 3) + return ERR_PTR(-EINVAL); + if (!length) + return ERR_PTR(-EINVAL); + + offset = offset_in_page(addr); + count = DIV_ROUND_UP(offset + length, PAGE_SIZE); + pages = kcalloc(count, sizeof(*pages), GFP_KERNEL); + + err = get_user_pages_fast(addr, count, 1, pages); + if (err < count) { + count = err; + err = -EFAULT; + goto put_pages; + } + + iod = nvme_alloc_iod(count, length, GFP_KERNEL); + sg = iod->sg; + sg_init_table(sg, count); + for (i = 0; i < count; i++) { + sg_set_page(&sg[i], pages[i], + min_t(int, length, PAGE_SIZE - offset), offset); + length -= (PAGE_SIZE - offset); + offset = 0; + } + sg_mark_end(&sg[i - 1]); + iod->nents = count; + + err = -ENOMEM; + nents = dma_map_sg(&dev->pci_dev->dev, sg, count, + write ? DMA_TO_DEVICE : DMA_FROM_DEVICE); + if (!nents) + goto free_iod; + + kfree(pages); + return iod; + + free_iod: + kfree(iod); + put_pages: + for (i = 0; i < count; i++) + put_page(pages[i]); + kfree(pages); + return ERR_PTR(err); +} + +static void nvme_unmap_user_pages(struct nvme_dev *dev, int write, + struct nvme_iod *iod) +{ + int i; + + dma_unmap_sg(&dev->pci_dev->dev, iod->sg, iod->nents, + write ? DMA_TO_DEVICE : DMA_FROM_DEVICE); + + for (i = 0; i < iod->nents; i++) + put_page(sg_page(&iod->sg[i])); +} + +static int nvme_submit_io(struct nvme_ns *ns, struct nvme_user_io __user *uio) +{ + struct nvme_dev *dev = ns->dev; + struct nvme_queue *nvmeq; + struct nvme_user_io io; + struct nvme_command c; + unsigned length; + int status; + struct nvme_iod *iod; + + if (copy_from_user(&io, uio, sizeof(io))) + return -EFAULT; + length = (io.nblocks + 1) << ns->lba_shift; + + switch (io.opcode) { + case nvme_cmd_write: + case nvme_cmd_read: + case nvme_cmd_compare: + iod = nvme_map_user_pages(dev, io.opcode & 1, io.addr, length); + break; + default: + return -EINVAL; + } + + if (IS_ERR(iod)) + return PTR_ERR(iod); + + memset(&c, 0, sizeof(c)); + c.rw.opcode = io.opcode; + c.rw.flags = io.flags; + c.rw.nsid = cpu_to_le32(ns->ns_id); + c.rw.slba = cpu_to_le64(io.slba); + c.rw.length = cpu_to_le16(io.nblocks); + c.rw.control = cpu_to_le16(io.control); + c.rw.dsmgmt = cpu_to_le16(io.dsmgmt); + c.rw.reftag = io.reftag; + c.rw.apptag = io.apptag; + c.rw.appmask = io.appmask; + /* XXX: metadata */ + length = nvme_setup_prps(dev, &c.common, iod, length, GFP_KERNEL); + + nvmeq = get_nvmeq(dev); + /* + * Since nvme_submit_sync_cmd sleeps, we can't keep preemption + * disabled. We may be preempted at any point, and be rescheduled + * to a different CPU. That will cause cacheline bouncing, but no + * additional races since q_lock already protects against other CPUs. + */ + put_nvmeq(nvmeq); + if (length != (io.nblocks + 1) << ns->lba_shift) + status = -ENOMEM; + else + status = nvme_submit_sync_cmd(nvmeq, &c, NULL, NVME_IO_TIMEOUT); + + nvme_unmap_user_pages(dev, io.opcode & 1, iod); + nvme_free_iod(dev, iod); + return status; +} + +static int nvme_user_admin_cmd(struct nvme_ns *ns, + struct nvme_admin_cmd __user *ucmd) +{ + struct nvme_dev *dev = ns->dev; + struct nvme_admin_cmd cmd; + struct nvme_command c; + int status, length; + struct nvme_iod *iod; + + if (!capable(CAP_SYS_ADMIN)) + return -EACCES; + if (copy_from_user(&cmd, ucmd, sizeof(cmd))) + return -EFAULT; + + memset(&c, 0, sizeof(c)); + c.common.opcode = cmd.opcode; + c.common.flags = cmd.flags; + c.common.nsid = cpu_to_le32(cmd.nsid); + c.common.cdw2[0] = cpu_to_le32(cmd.cdw2); + c.common.cdw2[1] = cpu_to_le32(cmd.cdw3); + c.common.cdw10[0] = cpu_to_le32(cmd.cdw10); + c.common.cdw10[1] = cpu_to_le32(cmd.cdw11); + c.common.cdw10[2] = cpu_to_le32(cmd.cdw12); + c.common.cdw10[3] = cpu_to_le32(cmd.cdw13); + c.common.cdw10[4] = cpu_to_le32(cmd.cdw14); + c.common.cdw10[5] = cpu_to_le32(cmd.cdw15); + + length = cmd.data_len; + if (cmd.data_len) { + iod = nvme_map_user_pages(dev, cmd.opcode & 1, cmd.addr, + length); + if (IS_ERR(iod)) + return PTR_ERR(iod); + length = nvme_setup_prps(dev, &c.common, iod, length, + GFP_KERNEL); + } + + if (length != cmd.data_len) + status = -ENOMEM; + else + status = nvme_submit_admin_cmd(dev, &c, NULL); + + if (cmd.data_len) { + nvme_unmap_user_pages(dev, cmd.opcode & 1, iod); + nvme_free_iod(dev, iod); + } + return status; +} + +static int nvme_ioctl(struct block_device *bdev, fmode_t mode, unsigned int cmd, + unsigned long arg) +{ + struct nvme_ns *ns = bdev->bd_disk->private_data; + + switch (cmd) { + case NVME_IOCTL_ID: + return ns->ns_id; + case NVME_IOCTL_ADMIN_CMD: + return nvme_user_admin_cmd(ns, (void __user *)arg); + case NVME_IOCTL_SUBMIT_IO: + return nvme_submit_io(ns, (void __user *)arg); + default: + return -ENOTTY; + } +} + +static const struct block_device_operations nvme_fops = { + .owner = THIS_MODULE, + .ioctl = nvme_ioctl, + .compat_ioctl = nvme_ioctl, +}; + +static void nvme_timeout_ios(struct nvme_queue *nvmeq) +{ + int depth = nvmeq->q_depth - 1; + struct nvme_cmd_info *info = nvme_cmd_info(nvmeq); + unsigned long now = jiffies; + int cmdid; + + for_each_set_bit(cmdid, nvmeq->cmdid_data, depth) { + void *ctx; + nvme_completion_fn fn; + static struct nvme_completion cqe = { .status = cpu_to_le16(NVME_SC_ABORT_REQ) << 1, }; + + if (!time_after(now, info[cmdid].timeout)) + continue; + dev_warn(nvmeq->q_dmadev, "Timing out I/O %d\n", cmdid); + ctx = cancel_cmdid(nvmeq, cmdid, &fn); + fn(nvmeq->dev, ctx, &cqe); + } +} + +static void nvme_resubmit_bios(struct nvme_queue *nvmeq) +{ + while (bio_list_peek(&nvmeq->sq_cong)) { + struct bio *bio = bio_list_pop(&nvmeq->sq_cong); + struct nvme_ns *ns = bio->bi_bdev->bd_disk->private_data; + if (nvme_submit_bio_queue(nvmeq, ns, bio)) { + bio_list_add_head(&nvmeq->sq_cong, bio); + break; + } + if (bio_list_empty(&nvmeq->sq_cong)) + remove_wait_queue(&nvmeq->sq_full, + &nvmeq->sq_cong_wait); + } +} + +static int nvme_kthread(void *data) +{ + struct nvme_dev *dev; + + while (!kthread_should_stop()) { + __set_current_state(TASK_RUNNING); + spin_lock(&dev_list_lock); + list_for_each_entry(dev, &dev_list, node) { + int i; + for (i = 0; i < dev->queue_count; i++) { + struct nvme_queue *nvmeq = dev->queues[i]; + if (!nvmeq) + continue; + spin_lock_irq(&nvmeq->q_lock); + if (nvme_process_cq(nvmeq)) + printk("process_cq did something\n"); + nvme_timeout_ios(nvmeq); + nvme_resubmit_bios(nvmeq); + spin_unlock_irq(&nvmeq->q_lock); + } + } + spin_unlock(&dev_list_lock); + set_current_state(TASK_INTERRUPTIBLE); + schedule_timeout(HZ); + } + return 0; +} + +static DEFINE_IDA(nvme_index_ida); + +static int nvme_get_ns_idx(void) +{ + int index, error; + + do { + if (!ida_pre_get(&nvme_index_ida, GFP_KERNEL)) + return -1; + + spin_lock(&dev_list_lock); + error = ida_get_new(&nvme_index_ida, &index); + spin_unlock(&dev_list_lock); + } while (error == -EAGAIN); + + if (error) + index = -1; + return index; +} + +static void nvme_put_ns_idx(int index) +{ + spin_lock(&dev_list_lock); + ida_remove(&nvme_index_ida, index); + spin_unlock(&dev_list_lock); +} + +static struct nvme_ns *nvme_alloc_ns(struct nvme_dev *dev, int nsid, + struct nvme_id_ns *id, struct nvme_lba_range_type *rt) +{ + struct nvme_ns *ns; + struct gendisk *disk; + int lbaf; + + if (rt->attributes & NVME_LBART_ATTRIB_HIDE) + return NULL; + + ns = kzalloc(sizeof(*ns), GFP_KERNEL); + if (!ns) + return NULL; + ns->queue = blk_alloc_queue(GFP_KERNEL); + if (!ns->queue) + goto out_free_ns; + ns->queue->queue_flags = QUEUE_FLAG_DEFAULT; + queue_flag_set_unlocked(QUEUE_FLAG_NOMERGES, ns->queue); + queue_flag_set_unlocked(QUEUE_FLAG_NONROT, ns->queue); +/* queue_flag_set_unlocked(QUEUE_FLAG_DISCARD, ns->queue); */ + blk_queue_make_request(ns->queue, nvme_make_request); + ns->dev = dev; + ns->queue->queuedata = ns; + + disk = alloc_disk(NVME_MINORS); + if (!disk) + goto out_free_queue; + ns->ns_id = nsid; + ns->disk = disk; + lbaf = id->flbas & 0xf; + ns->lba_shift = id->lbaf[lbaf].ds; + + disk->major = nvme_major; + disk->minors = NVME_MINORS; + disk->first_minor = NVME_MINORS * nvme_get_ns_idx(); + disk->fops = &nvme_fops; + disk->private_data = ns; + disk->queue = ns->queue; + disk->driverfs_dev = &dev->pci_dev->dev; + sprintf(disk->disk_name, "nvme%dn%d", dev->instance, nsid); + set_capacity(disk, le64_to_cpup(&id->nsze) << (ns->lba_shift - 9)); + + return ns; + + out_free_queue: + blk_cleanup_queue(ns->queue); + out_free_ns: + kfree(ns); + return NULL; +} + +static void nvme_ns_free(struct nvme_ns *ns) +{ + int index = ns->disk->first_minor / NVME_MINORS; + put_disk(ns->disk); + nvme_put_ns_idx(index); + blk_cleanup_queue(ns->queue); + kfree(ns); +} + +static int set_queue_count(struct nvme_dev *dev, int count) +{ + int status; + u32 result; + u32 q_count = (count - 1) | ((count - 1) << 16); + + status = nvme_set_features(dev, NVME_FEAT_NUM_QUEUES, q_count, 0, + &result); + if (status) + return -EIO; + return min(result & 0xffff, result >> 16) + 1; +} + +static int __devinit nvme_setup_io_queues(struct nvme_dev *dev) +{ + int result, cpu, i, nr_io_queues, db_bar_size; + + nr_io_queues = num_online_cpus(); + result = set_queue_count(dev, nr_io_queues); + if (result < 0) + return result; + if (result < nr_io_queues) + nr_io_queues = result; + + /* Deregister the admin queue's interrupt */ + free_irq(dev->entry[0].vector, dev->queues[0]); + + db_bar_size = 4096 + ((nr_io_queues + 1) << (dev->db_stride + 3)); + if (db_bar_size > 8192) { + iounmap(dev->bar); + dev->bar = ioremap(pci_resource_start(dev->pci_dev, 0), + db_bar_size); + dev->dbs = ((void __iomem *)dev->bar) + 4096; + dev->queues[0]->q_db = dev->dbs; + } + + for (i = 0; i < nr_io_queues; i++) + dev->entry[i].entry = i; + for (;;) { + result = pci_enable_msix(dev->pci_dev, dev->entry, + nr_io_queues); + if (result == 0) { + break; + } else if (result > 0) { + nr_io_queues = result; + continue; + } else { + nr_io_queues = 1; + break; + } + } + + result = queue_request_irq(dev, dev->queues[0], "nvme admin"); + /* XXX: handle failure here */ + + cpu = cpumask_first(cpu_online_mask); + for (i = 0; i < nr_io_queues; i++) { + irq_set_affinity_hint(dev->entry[i].vector, get_cpu_mask(cpu)); + cpu = cpumask_next(cpu, cpu_online_mask); + } + + for (i = 0; i < nr_io_queues; i++) { + dev->queues[i + 1] = nvme_create_queue(dev, i + 1, + NVME_Q_DEPTH, i); + if (IS_ERR(dev->queues[i + 1])) + return PTR_ERR(dev->queues[i + 1]); + dev->queue_count++; + } + + for (; i < num_possible_cpus(); i++) { + int target = i % rounddown_pow_of_two(dev->queue_count - 1); + dev->queues[i + 1] = dev->queues[target + 1]; + } + + return 0; +} + +static void nvme_free_queues(struct nvme_dev *dev) +{ + int i; + + for (i = dev->queue_count - 1; i >= 0; i--) + nvme_free_queue(dev, i); +} + +static int __devinit nvme_dev_add(struct nvme_dev *dev) +{ + int res, nn, i; + struct nvme_ns *ns, *next; + struct nvme_id_ctrl *ctrl; + struct nvme_id_ns *id_ns; + void *mem; + dma_addr_t dma_addr; + + res = nvme_setup_io_queues(dev); + if (res) + return res; + + mem = dma_alloc_coherent(&dev->pci_dev->dev, 8192, &dma_addr, + GFP_KERNEL); + + res = nvme_identify(dev, 0, 1, dma_addr); + if (res) { + res = -EIO; + goto out_free; + } + + ctrl = mem; + nn = le32_to_cpup(&ctrl->nn); + memcpy(dev->serial, ctrl->sn, sizeof(ctrl->sn)); + memcpy(dev->model, ctrl->mn, sizeof(ctrl->mn)); + memcpy(dev->firmware_rev, ctrl->fr, sizeof(ctrl->fr)); + + id_ns = mem; + for (i = 1; i <= nn; i++) { + res = nvme_identify(dev, i, 0, dma_addr); + if (res) + continue; + + if (id_ns->ncap == 0) + continue; + + res = nvme_get_features(dev, NVME_FEAT_LBA_RANGE, i, + dma_addr + 4096); + if (res) + continue; + + ns = nvme_alloc_ns(dev, i, mem, mem + 4096); + if (ns) + list_add_tail(&ns->list, &dev->namespaces); + } + list_for_each_entry(ns, &dev->namespaces, list) + add_disk(ns->disk); + + goto out; + + out_free: + list_for_each_entry_safe(ns, next, &dev->namespaces, list) { + list_del(&ns->list); + nvme_ns_free(ns); + } + + out: + dma_free_coherent(&dev->pci_dev->dev, 8192, mem, dma_addr); + return res; +} + +static int nvme_dev_remove(struct nvme_dev *dev) +{ + struct nvme_ns *ns, *next; + + spin_lock(&dev_list_lock); + list_del(&dev->node); + spin_unlock(&dev_list_lock); + + /* TODO: wait all I/O finished or cancel them */ + + list_for_each_entry_safe(ns, next, &dev->namespaces, list) { + list_del(&ns->list); + del_gendisk(ns->disk); + nvme_ns_free(ns); + } + + nvme_free_queues(dev); + + return 0; +} + +static int nvme_setup_prp_pools(struct nvme_dev *dev) +{ + struct device *dmadev = &dev->pci_dev->dev; + dev->prp_page_pool = dma_pool_create("prp list page", dmadev, + PAGE_SIZE, PAGE_SIZE, 0); + if (!dev->prp_page_pool) + return -ENOMEM; + + /* Optimisation for I/Os between 4k and 128k */ + dev->prp_small_pool = dma_pool_create("prp list 256", dmadev, + 256, 256, 0); + if (!dev->prp_small_pool) { + dma_pool_destroy(dev->prp_page_pool); + return -ENOMEM; + } + return 0; +} + +static void nvme_release_prp_pools(struct nvme_dev *dev) +{ + dma_pool_destroy(dev->prp_page_pool); + dma_pool_destroy(dev->prp_small_pool); +} + +/* XXX: Use an ida or something to let remove / add work correctly */ +static void nvme_set_instance(struct nvme_dev *dev) +{ + static int instance; + dev->instance = instance++; +} + +static void nvme_release_instance(struct nvme_dev *dev) +{ +} + +static int __devinit nvme_probe(struct pci_dev *pdev, + const struct pci_device_id *id) +{ + int bars, result = -ENOMEM; + struct nvme_dev *dev; + + dev = kzalloc(sizeof(*dev), GFP_KERNEL); + if (!dev) + return -ENOMEM; + dev->entry = kcalloc(num_possible_cpus(), sizeof(*dev->entry), + GFP_KERNEL); + if (!dev->entry) + goto free; + dev->queues = kcalloc(num_possible_cpus() + 1, sizeof(void *), + GFP_KERNEL); + if (!dev->queues) + goto free; + + if (pci_enable_device_mem(pdev)) + goto free; + pci_set_master(pdev); + bars = pci_select_bars(pdev, IORESOURCE_MEM); + if (pci_request_selected_regions(pdev, bars, "nvme")) + goto disable; + + INIT_LIST_HEAD(&dev->namespaces); + dev->pci_dev = pdev; + pci_set_drvdata(pdev, dev); + dma_set_mask(&pdev->dev, DMA_BIT_MASK(64)); + dma_set_coherent_mask(&pdev->dev, DMA_BIT_MASK(64)); + nvme_set_instance(dev); + dev->entry[0].vector = pdev->irq; + + result = nvme_setup_prp_pools(dev); + if (result) + goto disable_msix; + + dev->bar = ioremap(pci_resource_start(pdev, 0), 8192); + if (!dev->bar) { + result = -ENOMEM; + goto disable_msix; + } + + result = nvme_configure_admin_queue(dev); + if (result) + goto unmap; + dev->queue_count++; + + spin_lock(&dev_list_lock); + list_add(&dev->node, &dev_list); + spin_unlock(&dev_list_lock); + + result = nvme_dev_add(dev); + if (result) + goto delete; + + return 0; + + delete: + spin_lock(&dev_list_lock); + list_del(&dev->node); + spin_unlock(&dev_list_lock); + + nvme_free_queues(dev); + unmap: + iounmap(dev->bar); + disable_msix: + pci_disable_msix(pdev); + nvme_release_instance(dev); + nvme_release_prp_pools(dev); + disable: + pci_disable_device(pdev); + pci_release_regions(pdev); + free: + kfree(dev->queues); + kfree(dev->entry); + kfree(dev); + return result; +} + +static void __devexit nvme_remove(struct pci_dev *pdev) +{ + struct nvme_dev *dev = pci_get_drvdata(pdev); + nvme_dev_remove(dev); + pci_disable_msix(pdev); + iounmap(dev->bar); + nvme_release_instance(dev); + nvme_release_prp_pools(dev); + pci_disable_device(pdev); + pci_release_regions(pdev); + kfree(dev->queues); + kfree(dev->entry); + kfree(dev); +} + +/* These functions are yet to be implemented */ +#define nvme_error_detected NULL +#define nvme_dump_registers NULL +#define nvme_link_reset NULL +#define nvme_slot_reset NULL +#define nvme_error_resume NULL +#define nvme_suspend NULL +#define nvme_resume NULL + +static struct pci_error_handlers nvme_err_handler = { + .error_detected = nvme_error_detected, + .mmio_enabled = nvme_dump_registers, + .link_reset = nvme_link_reset, + .slot_reset = nvme_slot_reset, + .resume = nvme_error_resume, +}; + +/* Move to pci_ids.h later */ +#define PCI_CLASS_STORAGE_EXPRESS 0x010802 + +static DEFINE_PCI_DEVICE_TABLE(nvme_id_table) = { + { PCI_DEVICE_CLASS(PCI_CLASS_STORAGE_EXPRESS, 0xffffff) }, + { 0, } +}; +MODULE_DEVICE_TABLE(pci, nvme_id_table); + +static struct pci_driver nvme_driver = { + .name = "nvme", + .id_table = nvme_id_table, + .probe = nvme_probe, + .remove = __devexit_p(nvme_remove), + .suspend = nvme_suspend, + .resume = nvme_resume, + .err_handler = &nvme_err_handler, +}; + +static int __init nvme_init(void) +{ + int result = -EBUSY; + + nvme_thread = kthread_run(nvme_kthread, NULL, "nvme"); + if (IS_ERR(nvme_thread)) + return PTR_ERR(nvme_thread); + + nvme_major = register_blkdev(nvme_major, "nvme"); + if (nvme_major <= 0) + goto kill_kthread; + + result = pci_register_driver(&nvme_driver); + if (result) + goto unregister_blkdev; + return 0; + + unregister_blkdev: + unregister_blkdev(nvme_major, "nvme"); + kill_kthread: + kthread_stop(nvme_thread); + return result; +} + +static void __exit nvme_exit(void) +{ + pci_unregister_driver(&nvme_driver); + unregister_blkdev(nvme_major, "nvme"); + kthread_stop(nvme_thread); +} + +MODULE_AUTHOR("Matthew Wilcox <willy@linux.intel.com>"); +MODULE_LICENSE("GPL"); +MODULE_VERSION("0.8"); +module_init(nvme_init); +module_exit(nvme_exit); diff --git a/drivers/block/paride/bpck6.c b/drivers/block/paride/bpck6.c index ad124525ac23..ec64e7f5d1ce 100644 --- a/drivers/block/paride/bpck6.c +++ b/drivers/block/paride/bpck6.c @@ -20,9 +20,6 @@ */ -/* PARAMETERS */ -static int verbose; /* set this to 1 to see debugging messages and whatnot */ - #define BACKPACK_VERSION "2.0.2" #include <linux/module.h> @@ -36,6 +33,8 @@ static int verbose; /* set this to 1 to see debugging messages and whatnot */ #include "ppc6lnx.c" #include "paride.h" +/* PARAMETERS */ +static bool verbose; /* set this to 1 to see debugging messages and whatnot */ #define PPCSTRUCT(pi) ((Interface *)(pi->private)) diff --git a/drivers/block/paride/pcd.c b/drivers/block/paride/pcd.c index 46b8136c31bb..ba2b6b5e5910 100644 --- a/drivers/block/paride/pcd.c +++ b/drivers/block/paride/pcd.c @@ -144,7 +144,7 @@ enum {D_PRT, D_PRO, D_UNI, D_MOD, D_SLV, D_DLY}; static DEFINE_MUTEX(pcd_mutex); static DEFINE_SPINLOCK(pcd_lock); -module_param(verbose, bool, 0644); +module_param(verbose, int, 0644); module_param(major, int, 0); module_param(name, charp, 0); module_param(nice, int, 0); diff --git a/drivers/block/paride/pd.c b/drivers/block/paride/pd.c index 869e7676d46f..831e3ac156e6 100644 --- a/drivers/block/paride/pd.c +++ b/drivers/block/paride/pd.c @@ -124,8 +124,9 @@ by default. */ +#include <linux/types.h> -static int verbose = 0; +static bool verbose = 0; static int major = PD_MAJOR; static char *name = PD_NAME; static int cluster = 64; diff --git a/drivers/block/paride/pf.c b/drivers/block/paride/pf.c index f21b520ef419..ec8f9ed6326e 100644 --- a/drivers/block/paride/pf.c +++ b/drivers/block/paride/pf.c @@ -118,13 +118,15 @@ #define PF_NAME "pf" #define PF_UNITS 4 +#include <linux/types.h> + /* Here are things one can override from the insmod command. Most are autoprobed by paride unless set here. Verbose is off by default. */ -static int verbose = 0; +static bool verbose = 0; static int major = PF_MAJOR; static char *name = PF_NAME; static int cluster = 64; diff --git a/drivers/block/paride/pg.c b/drivers/block/paride/pg.c index a79fb4f7ff62..4a27b1de5fcb 100644 --- a/drivers/block/paride/pg.c +++ b/drivers/block/paride/pg.c @@ -130,13 +130,14 @@ #define PI_PG 4 #endif +#include <linux/types.h> /* Here are things one can override from the insmod command. Most are autoprobed by paride unless set here. Verbose is 0 by default. */ -static int verbose = 0; +static bool verbose = 0; static int major = PG_MAJOR; static char *name = PG_NAME; static int disable = 0; diff --git a/drivers/block/paride/pt.c b/drivers/block/paride/pt.c index 7179f79d7468..2596042eb987 100644 --- a/drivers/block/paride/pt.c +++ b/drivers/block/paride/pt.c @@ -109,13 +109,15 @@ #define PT_NAME "pt" #define PT_UNITS 4 +#include <linux/types.h> + /* Here are things one can override from the insmod command. Most are autoprobed by paride unless set here. Verbose is on by default. */ -static int verbose = 0; +static bool verbose = 0; static int major = PT_MAJOR; static char *name = PT_NAME; static int disable = 0; diff --git a/drivers/block/pktcdvd.c b/drivers/block/pktcdvd.c index d59edeabd93f..ba66e4445f41 100644 --- a/drivers/block/pktcdvd.c +++ b/drivers/block/pktcdvd.c @@ -987,14 +987,14 @@ static void pkt_copy_bio_data(struct bio *src_bio, int seg, int offs, struct pag while (copy_size > 0) { struct bio_vec *src_bvl = bio_iovec_idx(src_bio, seg); - void *vfrom = kmap_atomic(src_bvl->bv_page, KM_USER0) + + void *vfrom = kmap_atomic(src_bvl->bv_page) + src_bvl->bv_offset + offs; void *vto = page_address(dst_page) + dst_offs; int len = min_t(int, copy_size, src_bvl->bv_len - offs); BUG_ON(len < 0); memcpy(vto, vfrom, len); - kunmap_atomic(vfrom, KM_USER0); + kunmap_atomic(vfrom); seg++; offs = 0; @@ -1019,10 +1019,10 @@ static void pkt_make_local_copy(struct packet_data *pkt, struct bio_vec *bvec) offs = 0; for (f = 0; f < pkt->frames; f++) { if (bvec[f].bv_page != pkt->pages[p]) { - void *vfrom = kmap_atomic(bvec[f].bv_page, KM_USER0) + bvec[f].bv_offset; + void *vfrom = kmap_atomic(bvec[f].bv_page) + bvec[f].bv_offset; void *vto = page_address(pkt->pages[p]) + offs; memcpy(vto, vfrom, CD_FRAMESIZE); - kunmap_atomic(vfrom, KM_USER0); + kunmap_atomic(vfrom); bvec[f].bv_page = pkt->pages[p]; bvec[f].bv_offset = offs; } else { diff --git a/drivers/block/rbd.c b/drivers/block/rbd.c index 148ab944378d..013c7a549fb6 100644 --- a/drivers/block/rbd.c +++ b/drivers/block/rbd.c @@ -41,19 +41,35 @@ #include "rbd_types.h" -#define DRV_NAME "rbd" -#define DRV_NAME_LONG "rbd (rados block device)" +/* + * The basic unit of block I/O is a sector. It is interpreted in a + * number of contexts in Linux (blk, bio, genhd), but the default is + * universally 512 bytes. These symbols are just slightly more + * meaningful than the bare numbers they represent. + */ +#define SECTOR_SHIFT 9 +#define SECTOR_SIZE (1ULL << SECTOR_SHIFT) + +#define RBD_DRV_NAME "rbd" +#define RBD_DRV_NAME_LONG "rbd (rados block device)" #define RBD_MINORS_PER_MAJOR 256 /* max minors per blkdev */ -#define RBD_MAX_MD_NAME_LEN (96 + sizeof(RBD_SUFFIX)) +#define RBD_MAX_MD_NAME_LEN (RBD_MAX_OBJ_NAME_LEN + sizeof(RBD_SUFFIX)) #define RBD_MAX_POOL_NAME_LEN 64 #define RBD_MAX_SNAP_NAME_LEN 32 #define RBD_MAX_OPT_LEN 1024 #define RBD_SNAP_HEAD_NAME "-" +/* + * An RBD device name will be "rbd#", where the "rbd" comes from + * RBD_DRV_NAME above, and # is a unique integer identifier. + * MAX_INT_FORMAT_WIDTH is used in ensuring DEV_NAME_LEN is big + * enough to hold all possible device names. + */ #define DEV_NAME_LEN 32 +#define MAX_INT_FORMAT_WIDTH ((5 * sizeof (int)) / 2 + 1) #define RBD_NOTIFY_TIMEOUT_DEFAULT 10 @@ -66,7 +82,6 @@ struct rbd_image_header { __u8 obj_order; __u8 crypt_type; __u8 comp_type; - struct rw_semaphore snap_rwsem; struct ceph_snap_context *snapc; size_t snap_names_len; u64 snap_seq; @@ -83,7 +98,7 @@ struct rbd_options { }; /* - * an instance of the client. multiple devices may share a client. + * an instance of the client. multiple devices may share an rbd client. */ struct rbd_client { struct ceph_client *client; @@ -92,20 +107,9 @@ struct rbd_client { struct list_head node; }; -struct rbd_req_coll; - /* - * a single io request + * a request completion status */ -struct rbd_request { - struct request *rq; /* blk layer request */ - struct bio *bio; /* cloned bio */ - struct page **pages; /* list of used pages */ - u64 len; - int coll_index; - struct rbd_req_coll *coll; -}; - struct rbd_req_status { int done; int rc; @@ -122,6 +126,18 @@ struct rbd_req_coll { struct rbd_req_status status[0]; }; +/* + * a single io request + */ +struct rbd_request { + struct request *rq; /* blk layer request */ + struct bio *bio; /* cloned bio */ + struct page **pages; /* list of used pages */ + u64 len; + int coll_index; + struct rbd_req_coll *coll; +}; + struct rbd_snap { struct device dev; const char *name; @@ -140,7 +156,6 @@ struct rbd_device { struct gendisk *disk; /* blkdev's gendisk and rq */ struct request_queue *q; - struct ceph_client *client; struct rbd_client *rbd_client; char name[DEV_NAME_LEN]; /* blkdev name, e.g. rbd3 */ @@ -157,6 +172,8 @@ struct rbd_device { struct ceph_osd_event *watch_event; struct ceph_osd_request *watch_request; + /* protects updating the header */ + struct rw_semaphore header_rwsem; char snap_name[RBD_MAX_SNAP_NAME_LEN]; u32 cur_snap; /* index+1 of current snapshot within snap context 0 - for the head */ @@ -171,15 +188,13 @@ struct rbd_device { struct device dev; }; -static struct bus_type rbd_bus_type = { - .name = "rbd", -}; - -static spinlock_t node_lock; /* protects client get/put */ - static DEFINE_MUTEX(ctl_mutex); /* Serialize open/close/setup/teardown */ + static LIST_HEAD(rbd_dev_list); /* devices */ -static LIST_HEAD(rbd_client_list); /* clients */ +static DEFINE_SPINLOCK(rbd_dev_list_lock); + +static LIST_HEAD(rbd_client_list); /* clients */ +static DEFINE_SPINLOCK(rbd_client_list_lock); static int __rbd_init_snaps_header(struct rbd_device *rbd_dev); static void rbd_dev_release(struct device *dev); @@ -190,12 +205,32 @@ static ssize_t rbd_snap_add(struct device *dev, static void __rbd_remove_snap_dev(struct rbd_device *rbd_dev, struct rbd_snap *snap); +static ssize_t rbd_add(struct bus_type *bus, const char *buf, + size_t count); +static ssize_t rbd_remove(struct bus_type *bus, const char *buf, + size_t count); + +static struct bus_attribute rbd_bus_attrs[] = { + __ATTR(add, S_IWUSR, NULL, rbd_add), + __ATTR(remove, S_IWUSR, NULL, rbd_remove), + __ATTR_NULL +}; -static struct rbd_device *dev_to_rbd(struct device *dev) +static struct bus_type rbd_bus_type = { + .name = "rbd", + .bus_attrs = rbd_bus_attrs, +}; + +static void rbd_root_dev_release(struct device *dev) { - return container_of(dev, struct rbd_device, dev); } +static struct device rbd_root_dev = { + .init_name = "rbd", + .release = rbd_root_dev_release, +}; + + static struct device *rbd_get_dev(struct rbd_device *rbd_dev) { return get_device(&rbd_dev->dev); @@ -210,8 +245,7 @@ static int __rbd_update_snaps(struct rbd_device *rbd_dev); static int rbd_open(struct block_device *bdev, fmode_t mode) { - struct gendisk *disk = bdev->bd_disk; - struct rbd_device *rbd_dev = disk->private_data; + struct rbd_device *rbd_dev = bdev->bd_disk->private_data; rbd_get_dev(rbd_dev); @@ -256,9 +290,11 @@ static struct rbd_client *rbd_client_create(struct ceph_options *opt, kref_init(&rbdc->kref); INIT_LIST_HEAD(&rbdc->node); + mutex_lock_nested(&ctl_mutex, SINGLE_DEPTH_NESTING); + rbdc->client = ceph_create_client(opt, rbdc, 0, 0); if (IS_ERR(rbdc->client)) - goto out_rbdc; + goto out_mutex; opt = NULL; /* Now rbdc->client is responsible for opt */ ret = ceph_open_session(rbdc->client); @@ -267,16 +303,19 @@ static struct rbd_client *rbd_client_create(struct ceph_options *opt, rbdc->rbd_opts = rbd_opts; - spin_lock(&node_lock); + spin_lock(&rbd_client_list_lock); list_add_tail(&rbdc->node, &rbd_client_list); - spin_unlock(&node_lock); + spin_unlock(&rbd_client_list_lock); + + mutex_unlock(&ctl_mutex); dout("rbd_client_create created %p\n", rbdc); return rbdc; out_err: ceph_destroy_client(rbdc->client); -out_rbdc: +out_mutex: + mutex_unlock(&ctl_mutex); kfree(rbdc); out_opt: if (opt) @@ -324,7 +363,7 @@ static int parse_rbd_opts_token(char *c, void *private) substring_t argstr[MAX_OPT_ARGS]; int token, intval, ret; - token = match_token((char *)c, rbdopt_tokens, argstr); + token = match_token(c, rbdopt_tokens, argstr); if (token < 0) return -EINVAL; @@ -357,64 +396,61 @@ static int parse_rbd_opts_token(char *c, void *private) * Get a ceph client with specific addr and configuration, if one does * not exist create it. */ -static int rbd_get_client(struct rbd_device *rbd_dev, const char *mon_addr, - char *options) +static struct rbd_client *rbd_get_client(const char *mon_addr, + size_t mon_addr_len, + char *options) { struct rbd_client *rbdc; struct ceph_options *opt; - int ret; struct rbd_options *rbd_opts; rbd_opts = kzalloc(sizeof(*rbd_opts), GFP_KERNEL); if (!rbd_opts) - return -ENOMEM; + return ERR_PTR(-ENOMEM); rbd_opts->notify_timeout = RBD_NOTIFY_TIMEOUT_DEFAULT; - ret = ceph_parse_options(&opt, options, mon_addr, - mon_addr + strlen(mon_addr), parse_rbd_opts_token, rbd_opts); - if (ret < 0) - goto done_err; + opt = ceph_parse_options(options, mon_addr, + mon_addr + mon_addr_len, + parse_rbd_opts_token, rbd_opts); + if (IS_ERR(opt)) { + kfree(rbd_opts); + return ERR_CAST(opt); + } - spin_lock(&node_lock); + spin_lock(&rbd_client_list_lock); rbdc = __rbd_client_find(opt); if (rbdc) { - ceph_destroy_options(opt); - /* using an existing client */ kref_get(&rbdc->kref); - rbd_dev->rbd_client = rbdc; - rbd_dev->client = rbdc->client; - spin_unlock(&node_lock); - return 0; + spin_unlock(&rbd_client_list_lock); + + ceph_destroy_options(opt); + kfree(rbd_opts); + + return rbdc; } - spin_unlock(&node_lock); + spin_unlock(&rbd_client_list_lock); rbdc = rbd_client_create(opt, rbd_opts); - if (IS_ERR(rbdc)) { - ret = PTR_ERR(rbdc); - goto done_err; - } - rbd_dev->rbd_client = rbdc; - rbd_dev->client = rbdc->client; - return 0; -done_err: - kfree(rbd_opts); - return ret; + if (IS_ERR(rbdc)) + kfree(rbd_opts); + + return rbdc; } /* * Destroy ceph client + * + * Caller must hold rbd_client_list_lock. */ static void rbd_client_release(struct kref *kref) { struct rbd_client *rbdc = container_of(kref, struct rbd_client, kref); dout("rbd_release_client %p\n", rbdc); - spin_lock(&node_lock); list_del(&rbdc->node); - spin_unlock(&node_lock); ceph_destroy_client(rbdc->client); kfree(rbdc->rbd_opts); @@ -427,9 +463,10 @@ static void rbd_client_release(struct kref *kref) */ static void rbd_put_client(struct rbd_device *rbd_dev) { + spin_lock(&rbd_client_list_lock); kref_put(&rbd_dev->rbd_client->kref, rbd_client_release); + spin_unlock(&rbd_client_list_lock); rbd_dev->rbd_client = NULL; - rbd_dev->client = NULL; } /* @@ -454,21 +491,19 @@ static int rbd_header_from_disk(struct rbd_image_header *header, gfp_t gfp_flags) { int i; - u32 snap_count = le32_to_cpu(ondisk->snap_count); - int ret = -ENOMEM; + u32 snap_count; - if (memcmp(ondisk, RBD_HEADER_TEXT, sizeof(RBD_HEADER_TEXT))) { + if (memcmp(ondisk, RBD_HEADER_TEXT, sizeof(RBD_HEADER_TEXT))) return -ENXIO; - } - init_rwsem(&header->snap_rwsem); - header->snap_names_len = le64_to_cpu(ondisk->snap_names_len); + snap_count = le32_to_cpu(ondisk->snap_count); header->snapc = kmalloc(sizeof(struct ceph_snap_context) + - snap_count * - sizeof(struct rbd_image_snap_ondisk), + snap_count * sizeof (*ondisk), gfp_flags); if (!header->snapc) return -ENOMEM; + + header->snap_names_len = le64_to_cpu(ondisk->snap_names_len); if (snap_count) { header->snap_names = kmalloc(header->snap_names_len, GFP_KERNEL); @@ -495,8 +530,7 @@ static int rbd_header_from_disk(struct rbd_image_header *header, header->snapc->num_snaps = snap_count; header->total_snaps = snap_count; - if (snap_count && - allocated_snaps == snap_count) { + if (snap_count && allocated_snaps == snap_count) { for (i = 0; i < snap_count; i++) { header->snapc->snaps[i] = le64_to_cpu(ondisk->snaps[i].id); @@ -515,7 +549,7 @@ err_names: kfree(header->snap_names); err_snapc: kfree(header->snapc); - return ret; + return -ENOMEM; } static int snap_index(struct rbd_image_header *header, int snap_num) @@ -539,35 +573,34 @@ static int snap_by_name(struct rbd_image_header *header, const char *snap_name, int i; char *p = header->snap_names; - for (i = 0; i < header->total_snaps; i++, p += strlen(p) + 1) { - if (strcmp(snap_name, p) == 0) - break; - } - if (i == header->total_snaps) - return -ENOENT; - if (seq) - *seq = header->snapc->snaps[i]; + for (i = 0; i < header->total_snaps; i++) { + if (!strcmp(snap_name, p)) { - if (size) - *size = header->snap_sizes[i]; + /* Found it. Pass back its id and/or size */ - return i; + if (seq) + *seq = header->snapc->snaps[i]; + if (size) + *size = header->snap_sizes[i]; + return i; + } + p += strlen(p) + 1; /* Skip ahead to the next name */ + } + return -ENOENT; } -static int rbd_header_set_snap(struct rbd_device *dev, - const char *snap_name, - u64 *size) +static int rbd_header_set_snap(struct rbd_device *dev, u64 *size) { struct rbd_image_header *header = &dev->header; struct ceph_snap_context *snapc = header->snapc; int ret = -ENOENT; - down_write(&header->snap_rwsem); + BUILD_BUG_ON(sizeof (dev->snap_name) < sizeof (RBD_SNAP_HEAD_NAME)); - if (!snap_name || - !*snap_name || - strcmp(snap_name, "-") == 0 || - strcmp(snap_name, RBD_SNAP_HEAD_NAME) == 0) { + down_write(&dev->header_rwsem); + + if (!memcmp(dev->snap_name, RBD_SNAP_HEAD_NAME, + sizeof (RBD_SNAP_HEAD_NAME))) { if (header->total_snaps) snapc->seq = header->snap_seq; else @@ -577,7 +610,7 @@ static int rbd_header_set_snap(struct rbd_device *dev, if (size) *size = header->image_size; } else { - ret = snap_by_name(header, snap_name, &snapc->seq, size); + ret = snap_by_name(header, dev->snap_name, &snapc->seq, size); if (ret < 0) goto done; @@ -587,7 +620,7 @@ static int rbd_header_set_snap(struct rbd_device *dev, ret = 0; done: - up_write(&header->snap_rwsem); + up_write(&dev->header_rwsem); return ret; } @@ -714,7 +747,7 @@ static struct bio *bio_chain_clone(struct bio **old, struct bio **next, /* split the bio. We'll release it either in the next call, or it will have to be released outside */ - bp = bio_split(old_chain, (len - total) / 512ULL); + bp = bio_split(old_chain, (len - total) / SECTOR_SIZE); if (!bp) goto err_out; @@ -854,7 +887,7 @@ static int rbd_do_request(struct request *rq, struct timespec mtime = CURRENT_TIME; struct rbd_request *req_data; struct ceph_osd_request_head *reqhead; - struct rbd_image_header *header = &dev->header; + struct ceph_osd_client *osdc; req_data = kzalloc(sizeof(*req_data), GFP_NOIO); if (!req_data) { @@ -871,15 +904,13 @@ static int rbd_do_request(struct request *rq, dout("rbd_do_request obj=%s ofs=%lld len=%lld\n", obj, len, ofs); - down_read(&header->snap_rwsem); + down_read(&dev->header_rwsem); - req = ceph_osdc_alloc_request(&dev->client->osdc, flags, - snapc, - ops, - false, - GFP_NOIO, pages, bio); + osdc = &dev->rbd_client->client->osdc; + req = ceph_osdc_alloc_request(osdc, flags, snapc, ops, + false, GFP_NOIO, pages, bio); if (!req) { - up_read(&header->snap_rwsem); + up_read(&dev->header_rwsem); ret = -ENOMEM; goto done_pages; } @@ -906,27 +937,27 @@ static int rbd_do_request(struct request *rq, layout->fl_object_size = cpu_to_le32(1 << RBD_MAX_OBJ_ORDER); layout->fl_pg_preferred = cpu_to_le32(-1); layout->fl_pg_pool = cpu_to_le32(dev->poolid); - ceph_calc_raw_layout(&dev->client->osdc, layout, snapid, - ofs, &len, &bno, req, ops); + ceph_calc_raw_layout(osdc, layout, snapid, ofs, &len, &bno, + req, ops); ceph_osdc_build_request(req, ofs, &len, ops, snapc, &mtime, req->r_oid, req->r_oid_len); - up_read(&header->snap_rwsem); + up_read(&dev->header_rwsem); if (linger_req) { - ceph_osdc_set_request_linger(&dev->client->osdc, req); + ceph_osdc_set_request_linger(osdc, req); *linger_req = req; } - ret = ceph_osdc_start_request(&dev->client->osdc, req, false); + ret = ceph_osdc_start_request(osdc, req, false); if (ret < 0) goto done_err; if (!rbd_cb) { - ret = ceph_osdc_wait_request(&dev->client->osdc, req); + ret = ceph_osdc_wait_request(osdc, req); if (ver) *ver = le64_to_cpu(req->r_reassert_version.version); dout("reassert_ver=%lld\n", @@ -1210,8 +1241,8 @@ static void rbd_watch_cb(u64 ver, u64 notify_id, u8 opcode, void *data) rc = __rbd_update_snaps(dev); mutex_unlock(&ctl_mutex); if (rc) - pr_warning(DRV_NAME "%d got notification but failed to update" - " snaps: %d\n", dev->major, rc); + pr_warning(RBD_DRV_NAME "%d got notification but failed to " + " update snaps: %d\n", dev->major, rc); rbd_req_sync_notify_ack(dev, ver, notify_id, dev->obj_md_name); } @@ -1224,7 +1255,7 @@ static int rbd_req_sync_watch(struct rbd_device *dev, u64 ver) { struct ceph_osd_req_op *ops; - struct ceph_osd_client *osdc = &dev->client->osdc; + struct ceph_osd_client *osdc = &dev->rbd_client->client->osdc; int ret = rbd_create_rw_ops(&ops, 1, CEPH_OSD_OP_WATCH, 0); if (ret < 0) @@ -1311,7 +1342,7 @@ static int rbd_req_sync_notify(struct rbd_device *dev, const char *obj) { struct ceph_osd_req_op *ops; - struct ceph_osd_client *osdc = &dev->client->osdc; + struct ceph_osd_client *osdc = &dev->rbd_client->client->osdc; struct ceph_osd_event *event; struct rbd_notify_info info; int payload_len = sizeof(u32) + sizeof(u32); @@ -1418,9 +1449,7 @@ static void rbd_rq_fn(struct request_queue *q) struct request *rq; struct bio_pair *bp = NULL; - rq = blk_fetch_request(q); - - while (1) { + while ((rq = blk_fetch_request(q))) { struct bio *bio; struct bio *rq_bio, *next_bio = NULL; bool do_write; @@ -1438,32 +1467,32 @@ static void rbd_rq_fn(struct request_queue *q) /* filter out block requests we don't understand */ if ((rq->cmd_type != REQ_TYPE_FS)) { __blk_end_request_all(rq, 0); - goto next; + continue; } /* deduce our operation (read, write) */ do_write = (rq_data_dir(rq) == WRITE); size = blk_rq_bytes(rq); - ofs = blk_rq_pos(rq) * 512ULL; + ofs = blk_rq_pos(rq) * SECTOR_SIZE; rq_bio = rq->bio; if (do_write && rbd_dev->read_only) { __blk_end_request_all(rq, -EROFS); - goto next; + continue; } spin_unlock_irq(q->queue_lock); dout("%s 0x%x bytes at 0x%llx\n", do_write ? "write" : "read", - size, blk_rq_pos(rq) * 512ULL); + size, blk_rq_pos(rq) * SECTOR_SIZE); num_segs = rbd_get_num_segments(&rbd_dev->header, ofs, size); coll = rbd_alloc_coll(num_segs); if (!coll) { spin_lock_irq(q->queue_lock); __blk_end_request_all(rq, -ENOMEM); - goto next; + continue; } do { @@ -1509,8 +1538,6 @@ next_seg: if (bp) bio_pair_release(bp); spin_lock_irq(q->queue_lock); -next: - rq = blk_fetch_request(q); } } @@ -1523,13 +1550,17 @@ static int rbd_merge_bvec(struct request_queue *q, struct bvec_merge_data *bmd, struct bio_vec *bvec) { struct rbd_device *rbd_dev = q->queuedata; - unsigned int chunk_sectors = 1 << (rbd_dev->header.obj_order - 9); - sector_t sector = bmd->bi_sector + get_start_sect(bmd->bi_bdev); - unsigned int bio_sectors = bmd->bi_size >> 9; + unsigned int chunk_sectors; + sector_t sector; + unsigned int bio_sectors; int max; + chunk_sectors = 1 << (rbd_dev->header.obj_order - SECTOR_SHIFT); + sector = bmd->bi_sector + get_start_sect(bmd->bi_bdev); + bio_sectors = bmd->bi_size >> SECTOR_SHIFT; + max = (chunk_sectors - ((sector & (chunk_sectors - 1)) - + bio_sectors)) << 9; + + bio_sectors)) << SECTOR_SHIFT; if (max < 0) max = 0; /* bio_add cannot handle a negative return */ if (max <= bvec->bv_len && bio_sectors == 0) @@ -1562,15 +1593,16 @@ static int rbd_read_header(struct rbd_device *rbd_dev, ssize_t rc; struct rbd_image_header_ondisk *dh; int snap_count = 0; - u64 snap_names_len = 0; u64 ver; + size_t len; + /* + * First reads the fixed-size header to determine the number + * of snapshots, then re-reads it, along with all snapshot + * records as well as their stored names. + */ + len = sizeof (*dh); while (1) { - int len = sizeof(*dh) + - snap_count * sizeof(struct rbd_image_snap_ondisk) + - snap_names_len; - - rc = -ENOMEM; dh = kmalloc(len, GFP_KERNEL); if (!dh) return -ENOMEM; @@ -1585,21 +1617,22 @@ static int rbd_read_header(struct rbd_device *rbd_dev, rc = rbd_header_from_disk(header, dh, snap_count, GFP_KERNEL); if (rc < 0) { - if (rc == -ENXIO) { + if (rc == -ENXIO) pr_warning("unrecognized header format" " for image %s", rbd_dev->obj); - } goto out_dh; } - if (snap_count != header->total_snaps) { - snap_count = header->total_snaps; - snap_names_len = header->snap_names_len; - rbd_header_free(header); - kfree(dh); - continue; - } - break; + if (snap_count == header->total_snaps) + break; + + snap_count = header->total_snaps; + len = sizeof (*dh) + + snap_count * sizeof(struct rbd_image_snap_ondisk) + + header->snap_names_len; + + rbd_header_free(header); + kfree(dh); } header->obj_version = ver; @@ -1620,13 +1653,14 @@ static int rbd_header_add_snap(struct rbd_device *dev, int ret; void *data, *p, *e; u64 ver; + struct ceph_mon_client *monc; /* we should create a snapshot only if we're pointing at the head */ if (dev->cur_snap) return -EINVAL; - ret = ceph_monc_create_snapid(&dev->client->monc, dev->poolid, - &new_snapid); + monc = &dev->rbd_client->client->monc; + ret = ceph_monc_create_snapid(monc, dev->poolid, &new_snapid); dout("created snapid=%lld\n", new_snapid); if (ret < 0) return ret; @@ -1681,9 +1715,9 @@ static int __rbd_update_snaps(struct rbd_device *rbd_dev) return ret; /* resized? */ - set_capacity(rbd_dev->disk, h.image_size / 512ULL); + set_capacity(rbd_dev->disk, h.image_size / SECTOR_SIZE); - down_write(&rbd_dev->header.snap_rwsem); + down_write(&rbd_dev->header_rwsem); snap_seq = rbd_dev->header.snapc->seq; if (rbd_dev->header.total_snaps && @@ -1708,7 +1742,7 @@ static int __rbd_update_snaps(struct rbd_device *rbd_dev) ret = __rbd_init_snaps_header(rbd_dev); - up_write(&rbd_dev->header.snap_rwsem); + up_write(&rbd_dev->header_rwsem); return ret; } @@ -1718,6 +1752,7 @@ static int rbd_init_disk(struct rbd_device *rbd_dev) struct gendisk *disk; struct request_queue *q; int rc; + u64 segment_size; u64 total_size = 0; /* contact OSD, request size info about the object being mapped */ @@ -1730,7 +1765,7 @@ static int rbd_init_disk(struct rbd_device *rbd_dev) if (rc) return rc; - rc = rbd_header_set_snap(rbd_dev, rbd_dev->snap_name, &total_size); + rc = rbd_header_set_snap(rbd_dev, &total_size); if (rc) return rc; @@ -1740,7 +1775,7 @@ static int rbd_init_disk(struct rbd_device *rbd_dev) if (!disk) goto out; - snprintf(disk->disk_name, sizeof(disk->disk_name), DRV_NAME "%d", + snprintf(disk->disk_name, sizeof(disk->disk_name), RBD_DRV_NAME "%d", rbd_dev->id); disk->major = rbd_dev->major; disk->first_minor = 0; @@ -1753,11 +1788,15 @@ static int rbd_init_disk(struct rbd_device *rbd_dev) if (!q) goto out_disk; + /* We use the default size, but let's be explicit about it. */ + blk_queue_physical_block_size(q, SECTOR_SIZE); + /* set io sizes to object size */ - blk_queue_max_hw_sectors(q, rbd_obj_bytes(&rbd_dev->header) / 512ULL); - blk_queue_max_segment_size(q, rbd_obj_bytes(&rbd_dev->header)); - blk_queue_io_min(q, rbd_obj_bytes(&rbd_dev->header)); - blk_queue_io_opt(q, rbd_obj_bytes(&rbd_dev->header)); + segment_size = rbd_obj_bytes(&rbd_dev->header); + blk_queue_max_hw_sectors(q, segment_size / SECTOR_SIZE); + blk_queue_max_segment_size(q, segment_size); + blk_queue_io_min(q, segment_size); + blk_queue_io_opt(q, segment_size); blk_queue_merge_bvec(q, rbd_merge_bvec); disk->queue = q; @@ -1768,7 +1807,7 @@ static int rbd_init_disk(struct rbd_device *rbd_dev) rbd_dev->q = q; /* finally, announce the disk to the world */ - set_capacity(disk, total_size / 512ULL); + set_capacity(disk, total_size / SECTOR_SIZE); add_disk(disk); pr_info("%s: added with size 0x%llx\n", @@ -1785,10 +1824,15 @@ out: sysfs */ +static struct rbd_device *dev_to_rbd_dev(struct device *dev) +{ + return container_of(dev, struct rbd_device, dev); +} + static ssize_t rbd_size_show(struct device *dev, struct device_attribute *attr, char *buf) { - struct rbd_device *rbd_dev = dev_to_rbd(dev); + struct rbd_device *rbd_dev = dev_to_rbd_dev(dev); return sprintf(buf, "%llu\n", (unsigned long long)rbd_dev->header.image_size); } @@ -1796,7 +1840,7 @@ static ssize_t rbd_size_show(struct device *dev, static ssize_t rbd_major_show(struct device *dev, struct device_attribute *attr, char *buf) { - struct rbd_device *rbd_dev = dev_to_rbd(dev); + struct rbd_device *rbd_dev = dev_to_rbd_dev(dev); return sprintf(buf, "%d\n", rbd_dev->major); } @@ -1804,15 +1848,16 @@ static ssize_t rbd_major_show(struct device *dev, static ssize_t rbd_client_id_show(struct device *dev, struct device_attribute *attr, char *buf) { - struct rbd_device *rbd_dev = dev_to_rbd(dev); + struct rbd_device *rbd_dev = dev_to_rbd_dev(dev); - return sprintf(buf, "client%lld\n", ceph_client_id(rbd_dev->client)); + return sprintf(buf, "client%lld\n", + ceph_client_id(rbd_dev->rbd_client->client)); } static ssize_t rbd_pool_show(struct device *dev, struct device_attribute *attr, char *buf) { - struct rbd_device *rbd_dev = dev_to_rbd(dev); + struct rbd_device *rbd_dev = dev_to_rbd_dev(dev); return sprintf(buf, "%s\n", rbd_dev->pool_name); } @@ -1820,7 +1865,7 @@ static ssize_t rbd_pool_show(struct device *dev, static ssize_t rbd_name_show(struct device *dev, struct device_attribute *attr, char *buf) { - struct rbd_device *rbd_dev = dev_to_rbd(dev); + struct rbd_device *rbd_dev = dev_to_rbd_dev(dev); return sprintf(buf, "%s\n", rbd_dev->obj); } @@ -1829,7 +1874,7 @@ static ssize_t rbd_snap_show(struct device *dev, struct device_attribute *attr, char *buf) { - struct rbd_device *rbd_dev = dev_to_rbd(dev); + struct rbd_device *rbd_dev = dev_to_rbd_dev(dev); return sprintf(buf, "%s\n", rbd_dev->snap_name); } @@ -1839,7 +1884,7 @@ static ssize_t rbd_image_refresh(struct device *dev, const char *buf, size_t size) { - struct rbd_device *rbd_dev = dev_to_rbd(dev); + struct rbd_device *rbd_dev = dev_to_rbd_dev(dev); int rc; int ret = size; @@ -1904,7 +1949,7 @@ static ssize_t rbd_snap_size_show(struct device *dev, { struct rbd_snap *snap = container_of(dev, struct rbd_snap, dev); - return sprintf(buf, "%lld\n", (long long)snap->size); + return sprintf(buf, "%zd\n", snap->size); } static ssize_t rbd_snap_id_show(struct device *dev, @@ -1913,7 +1958,7 @@ static ssize_t rbd_snap_id_show(struct device *dev, { struct rbd_snap *snap = container_of(dev, struct rbd_snap, dev); - return sprintf(buf, "%lld\n", (long long)snap->id); + return sprintf(buf, "%llu\n", (unsigned long long) snap->id); } static DEVICE_ATTR(snap_size, S_IRUGO, rbd_snap_size_show, NULL); @@ -2085,19 +2130,9 @@ static int __rbd_init_snaps_header(struct rbd_device *rbd_dev) return 0; } - -static void rbd_root_dev_release(struct device *dev) -{ -} - -static struct device rbd_root_dev = { - .init_name = "rbd", - .release = rbd_root_dev_release, -}; - static int rbd_bus_add_dev(struct rbd_device *rbd_dev) { - int ret = -ENOMEM; + int ret; struct device *dev; struct rbd_snap *snap; @@ -2111,7 +2146,7 @@ static int rbd_bus_add_dev(struct rbd_device *rbd_dev) dev_set_name(dev, "%d", rbd_dev->id); ret = device_register(dev); if (ret < 0) - goto done_free; + goto out; list_for_each_entry(snap, &rbd_dev->snaps, node) { ret = rbd_register_snap_dev(rbd_dev, snap, @@ -2119,10 +2154,7 @@ static int rbd_bus_add_dev(struct rbd_device *rbd_dev) if (ret < 0) break; } - - mutex_unlock(&ctl_mutex); - return 0; -done_free: +out: mutex_unlock(&ctl_mutex); return ret; } @@ -2151,102 +2183,250 @@ static int rbd_init_watch_dev(struct rbd_device *rbd_dev) return ret; } +static atomic64_t rbd_id_max = ATOMIC64_INIT(0); + +/* + * Get a unique rbd identifier for the given new rbd_dev, and add + * the rbd_dev to the global list. The minimum rbd id is 1. + */ +static void rbd_id_get(struct rbd_device *rbd_dev) +{ + rbd_dev->id = atomic64_inc_return(&rbd_id_max); + + spin_lock(&rbd_dev_list_lock); + list_add_tail(&rbd_dev->node, &rbd_dev_list); + spin_unlock(&rbd_dev_list_lock); +} + +/* + * Remove an rbd_dev from the global list, and record that its + * identifier is no longer in use. + */ +static void rbd_id_put(struct rbd_device *rbd_dev) +{ + struct list_head *tmp; + int rbd_id = rbd_dev->id; + int max_id; + + BUG_ON(rbd_id < 1); + + spin_lock(&rbd_dev_list_lock); + list_del_init(&rbd_dev->node); + + /* + * If the id being "put" is not the current maximum, there + * is nothing special we need to do. + */ + if (rbd_id != atomic64_read(&rbd_id_max)) { + spin_unlock(&rbd_dev_list_lock); + return; + } + + /* + * We need to update the current maximum id. Search the + * list to find out what it is. We're more likely to find + * the maximum at the end, so search the list backward. + */ + max_id = 0; + list_for_each_prev(tmp, &rbd_dev_list) { + struct rbd_device *rbd_dev; + + rbd_dev = list_entry(tmp, struct rbd_device, node); + if (rbd_id > max_id) + max_id = rbd_id; + } + spin_unlock(&rbd_dev_list_lock); + + /* + * The max id could have been updated by rbd_id_get(), in + * which case it now accurately reflects the new maximum. + * Be careful not to overwrite the maximum value in that + * case. + */ + atomic64_cmpxchg(&rbd_id_max, rbd_id, max_id); +} + +/* + * Skips over white space at *buf, and updates *buf to point to the + * first found non-space character (if any). Returns the length of + * the token (string of non-white space characters) found. Note + * that *buf must be terminated with '\0'. + */ +static inline size_t next_token(const char **buf) +{ + /* + * These are the characters that produce nonzero for + * isspace() in the "C" and "POSIX" locales. + */ + const char *spaces = " \f\n\r\t\v"; + + *buf += strspn(*buf, spaces); /* Find start of token */ + + return strcspn(*buf, spaces); /* Return token length */ +} + +/* + * Finds the next token in *buf, and if the provided token buffer is + * big enough, copies the found token into it. The result, if + * copied, is guaranteed to be terminated with '\0'. Note that *buf + * must be terminated with '\0' on entry. + * + * Returns the length of the token found (not including the '\0'). + * Return value will be 0 if no token is found, and it will be >= + * token_size if the token would not fit. + * + * The *buf pointer will be updated to point beyond the end of the + * found token. Note that this occurs even if the token buffer is + * too small to hold it. + */ +static inline size_t copy_token(const char **buf, + char *token, + size_t token_size) +{ + size_t len; + + len = next_token(buf); + if (len < token_size) { + memcpy(token, *buf, len); + *(token + len) = '\0'; + } + *buf += len; + + return len; +} + +/* + * This fills in the pool_name, obj, obj_len, snap_name, obj_len, + * rbd_dev, rbd_md_name, and name fields of the given rbd_dev, based + * on the list of monitor addresses and other options provided via + * /sys/bus/rbd/add. + */ +static int rbd_add_parse_args(struct rbd_device *rbd_dev, + const char *buf, + const char **mon_addrs, + size_t *mon_addrs_size, + char *options, + size_t options_size) +{ + size_t len; + + /* The first four tokens are required */ + + len = next_token(&buf); + if (!len) + return -EINVAL; + *mon_addrs_size = len + 1; + *mon_addrs = buf; + + buf += len; + + len = copy_token(&buf, options, options_size); + if (!len || len >= options_size) + return -EINVAL; + + len = copy_token(&buf, rbd_dev->pool_name, sizeof (rbd_dev->pool_name)); + if (!len || len >= sizeof (rbd_dev->pool_name)) + return -EINVAL; + + len = copy_token(&buf, rbd_dev->obj, sizeof (rbd_dev->obj)); + if (!len || len >= sizeof (rbd_dev->obj)) + return -EINVAL; + + /* We have the object length in hand, save it. */ + + rbd_dev->obj_len = len; + + BUILD_BUG_ON(RBD_MAX_MD_NAME_LEN + < RBD_MAX_OBJ_NAME_LEN + sizeof (RBD_SUFFIX)); + sprintf(rbd_dev->obj_md_name, "%s%s", rbd_dev->obj, RBD_SUFFIX); + + /* + * The snapshot name is optional, but it's an error if it's + * too long. If no snapshot is supplied, fill in the default. + */ + len = copy_token(&buf, rbd_dev->snap_name, sizeof (rbd_dev->snap_name)); + if (!len) + memcpy(rbd_dev->snap_name, RBD_SNAP_HEAD_NAME, + sizeof (RBD_SNAP_HEAD_NAME)); + else if (len >= sizeof (rbd_dev->snap_name)) + return -EINVAL; + + return 0; +} + static ssize_t rbd_add(struct bus_type *bus, const char *buf, size_t count) { - struct ceph_osd_client *osdc; struct rbd_device *rbd_dev; - ssize_t rc = -ENOMEM; - int irc, new_id = 0; - struct list_head *tmp; - char *mon_dev_name; - char *options; + const char *mon_addrs = NULL; + size_t mon_addrs_size = 0; + char *options = NULL; + struct ceph_osd_client *osdc; + int rc = -ENOMEM; if (!try_module_get(THIS_MODULE)) return -ENODEV; - mon_dev_name = kmalloc(RBD_MAX_OPT_LEN, GFP_KERNEL); - if (!mon_dev_name) - goto err_out_mod; - - options = kmalloc(RBD_MAX_OPT_LEN, GFP_KERNEL); - if (!options) - goto err_mon_dev; - - /* new rbd_device object */ rbd_dev = kzalloc(sizeof(*rbd_dev), GFP_KERNEL); if (!rbd_dev) - goto err_out_opt; + goto err_nomem; + options = kmalloc(count, GFP_KERNEL); + if (!options) + goto err_nomem; /* static rbd_device initialization */ spin_lock_init(&rbd_dev->lock); INIT_LIST_HEAD(&rbd_dev->node); INIT_LIST_HEAD(&rbd_dev->snaps); + init_rwsem(&rbd_dev->header_rwsem); - /* generate unique id: find highest unique id, add one */ - mutex_lock_nested(&ctl_mutex, SINGLE_DEPTH_NESTING); - - list_for_each(tmp, &rbd_dev_list) { - struct rbd_device *rbd_dev; + init_rwsem(&rbd_dev->header_rwsem); - rbd_dev = list_entry(tmp, struct rbd_device, node); - if (rbd_dev->id >= new_id) - new_id = rbd_dev->id + 1; - } - - rbd_dev->id = new_id; + /* generate unique id: find highest unique id, add one */ + rbd_id_get(rbd_dev); - /* add to global list */ - list_add_tail(&rbd_dev->node, &rbd_dev_list); + /* Fill in the device name, now that we have its id. */ + BUILD_BUG_ON(DEV_NAME_LEN + < sizeof (RBD_DRV_NAME) + MAX_INT_FORMAT_WIDTH); + sprintf(rbd_dev->name, "%s%d", RBD_DRV_NAME, rbd_dev->id); /* parse add command */ - if (sscanf(buf, "%" __stringify(RBD_MAX_OPT_LEN) "s " - "%" __stringify(RBD_MAX_OPT_LEN) "s " - "%" __stringify(RBD_MAX_POOL_NAME_LEN) "s " - "%" __stringify(RBD_MAX_OBJ_NAME_LEN) "s" - "%" __stringify(RBD_MAX_SNAP_NAME_LEN) "s", - mon_dev_name, options, rbd_dev->pool_name, - rbd_dev->obj, rbd_dev->snap_name) < 4) { - rc = -EINVAL; - goto err_out_slot; - } - - if (rbd_dev->snap_name[0] == 0) - rbd_dev->snap_name[0] = '-'; - - rbd_dev->obj_len = strlen(rbd_dev->obj); - snprintf(rbd_dev->obj_md_name, sizeof(rbd_dev->obj_md_name), "%s%s", - rbd_dev->obj, RBD_SUFFIX); - - /* initialize rest of new object */ - snprintf(rbd_dev->name, DEV_NAME_LEN, DRV_NAME "%d", rbd_dev->id); - rc = rbd_get_client(rbd_dev, mon_dev_name, options); - if (rc < 0) - goto err_out_slot; + rc = rbd_add_parse_args(rbd_dev, buf, &mon_addrs, &mon_addrs_size, + options, count); + if (rc) + goto err_put_id; - mutex_unlock(&ctl_mutex); + rbd_dev->rbd_client = rbd_get_client(mon_addrs, mon_addrs_size - 1, + options); + if (IS_ERR(rbd_dev->rbd_client)) { + rc = PTR_ERR(rbd_dev->rbd_client); + goto err_put_id; + } /* pick the pool */ - osdc = &rbd_dev->client->osdc; + osdc = &rbd_dev->rbd_client->client->osdc; rc = ceph_pg_poolid_by_name(osdc->osdmap, rbd_dev->pool_name); if (rc < 0) goto err_out_client; rbd_dev->poolid = rc; /* register our block device */ - irc = register_blkdev(0, rbd_dev->name); - if (irc < 0) { - rc = irc; + rc = register_blkdev(0, rbd_dev->name); + if (rc < 0) goto err_out_client; - } - rbd_dev->major = irc; + rbd_dev->major = rc; rc = rbd_bus_add_dev(rbd_dev); if (rc) goto err_out_blkdev; - /* set up and announce blkdev mapping */ + /* + * At this point cleanup in the event of an error is the job + * of the sysfs code (initiated by rbd_bus_del_dev()). + * + * Set up and announce blkdev mapping. + */ rc = rbd_init_disk(rbd_dev); if (rc) goto err_out_bus; @@ -2258,35 +2438,26 @@ static ssize_t rbd_add(struct bus_type *bus, return count; err_out_bus: - mutex_lock_nested(&ctl_mutex, SINGLE_DEPTH_NESTING); - list_del_init(&rbd_dev->node); - mutex_unlock(&ctl_mutex); - /* this will also clean up rest of rbd_dev stuff */ rbd_bus_del_dev(rbd_dev); kfree(options); - kfree(mon_dev_name); return rc; err_out_blkdev: unregister_blkdev(rbd_dev->major, rbd_dev->name); err_out_client: rbd_put_client(rbd_dev); - mutex_lock_nested(&ctl_mutex, SINGLE_DEPTH_NESTING); -err_out_slot: - list_del_init(&rbd_dev->node); - mutex_unlock(&ctl_mutex); - - kfree(rbd_dev); -err_out_opt: +err_put_id: + rbd_id_put(rbd_dev); +err_nomem: kfree(options); -err_mon_dev: - kfree(mon_dev_name); -err_out_mod: + kfree(rbd_dev); + dout("Error adding device %s\n", buf); module_put(THIS_MODULE); - return rc; + + return (ssize_t) rc; } static struct rbd_device *__rbd_get_dev(unsigned long id) @@ -2294,22 +2465,28 @@ static struct rbd_device *__rbd_get_dev(unsigned long id) struct list_head *tmp; struct rbd_device *rbd_dev; + spin_lock(&rbd_dev_list_lock); list_for_each(tmp, &rbd_dev_list) { rbd_dev = list_entry(tmp, struct rbd_device, node); - if (rbd_dev->id == id) + if (rbd_dev->id == id) { + spin_unlock(&rbd_dev_list_lock); return rbd_dev; + } } + spin_unlock(&rbd_dev_list_lock); return NULL; } static void rbd_dev_release(struct device *dev) { - struct rbd_device *rbd_dev = - container_of(dev, struct rbd_device, dev); + struct rbd_device *rbd_dev = dev_to_rbd_dev(dev); - if (rbd_dev->watch_request) - ceph_osdc_unregister_linger_request(&rbd_dev->client->osdc, + if (rbd_dev->watch_request) { + struct ceph_client *client = rbd_dev->rbd_client->client; + + ceph_osdc_unregister_linger_request(&client->osdc, rbd_dev->watch_request); + } if (rbd_dev->watch_event) rbd_req_sync_unwatch(rbd_dev, rbd_dev->obj_md_name); @@ -2318,6 +2495,9 @@ static void rbd_dev_release(struct device *dev) /* clean up and free blkdev */ rbd_free_disk(rbd_dev); unregister_blkdev(rbd_dev->major, rbd_dev->name); + + /* done with the id, and with the rbd_dev */ + rbd_id_put(rbd_dev); kfree(rbd_dev); /* release module ref */ @@ -2350,8 +2530,6 @@ static ssize_t rbd_remove(struct bus_type *bus, goto done; } - list_del_init(&rbd_dev->node); - __rbd_remove_all_snaps(rbd_dev); rbd_bus_del_dev(rbd_dev); @@ -2365,7 +2543,7 @@ static ssize_t rbd_snap_add(struct device *dev, const char *buf, size_t count) { - struct rbd_device *rbd_dev = dev_to_rbd(dev); + struct rbd_device *rbd_dev = dev_to_rbd_dev(dev); int ret; char *name = kmalloc(count + 1, GFP_KERNEL); if (!name) @@ -2401,12 +2579,6 @@ err_unlock: return ret; } -static struct bus_attribute rbd_bus_attrs[] = { - __ATTR(add, S_IWUSR, NULL, rbd_add), - __ATTR(remove, S_IWUSR, NULL, rbd_remove), - __ATTR_NULL -}; - /* * create control files in sysfs * /sys/bus/rbd/... @@ -2415,21 +2587,21 @@ static int rbd_sysfs_init(void) { int ret; - rbd_bus_type.bus_attrs = rbd_bus_attrs; - - ret = bus_register(&rbd_bus_type); - if (ret < 0) + ret = device_register(&rbd_root_dev); + if (ret < 0) return ret; - ret = device_register(&rbd_root_dev); + ret = bus_register(&rbd_bus_type); + if (ret < 0) + device_unregister(&rbd_root_dev); return ret; } static void rbd_sysfs_cleanup(void) { - device_unregister(&rbd_root_dev); bus_unregister(&rbd_bus_type); + device_unregister(&rbd_root_dev); } int __init rbd_init(void) @@ -2439,8 +2611,7 @@ int __init rbd_init(void) rc = rbd_sysfs_init(); if (rc) return rc; - spin_lock_init(&node_lock); - pr_info("loaded " DRV_NAME_LONG "\n"); + pr_info("loaded " RBD_DRV_NAME_LONG "\n"); return 0; } diff --git a/drivers/block/rbd_types.h b/drivers/block/rbd_types.h index fc6c678aa2cb..950708688f17 100644 --- a/drivers/block/rbd_types.h +++ b/drivers/block/rbd_types.h @@ -41,10 +41,6 @@ #define RBD_HEADER_SIGNATURE "RBD" #define RBD_HEADER_VERSION "001.005" -struct rbd_info { - __le64 max_id; -} __attribute__ ((packed)); - struct rbd_image_snap_ondisk { __le64 id; __le64 image_size; diff --git a/drivers/block/sunvdc.c b/drivers/block/sunvdc.c index 48e8fee9f2d4..9dcf76a10bb6 100644 --- a/drivers/block/sunvdc.c +++ b/drivers/block/sunvdc.c @@ -839,10 +839,7 @@ static struct vio_driver vdc_port_driver = { .id_table = vdc_port_match, .probe = vdc_port_probe, .remove = vdc_port_remove, - .driver = { - .name = "vdc_port", - .owner = THIS_MODULE, - } + .name = "vdc_port", }; static int __init vdc_init(void) diff --git a/drivers/block/sx8.c b/drivers/block/sx8.c index b70f0fca9a42..3fb6ab4c8b4e 100644 --- a/drivers/block/sx8.c +++ b/drivers/block/sx8.c @@ -619,8 +619,10 @@ static int carm_array_info (struct carm_host *host, unsigned int array_idx) host->state == HST_DEV_SCAN); spin_unlock_irq(&host->lock); - DPRINTK("blk_insert_request, tag == %u\n", idx); - blk_insert_request(host->oob_q, crq->rq, 1, crq); + DPRINTK("blk_execute_rq_nowait, tag == %u\n", idx); + crq->rq->cmd_type = REQ_TYPE_SPECIAL; + crq->rq->special = crq; + blk_execute_rq_nowait(host->oob_q, NULL, crq->rq, true, NULL); return 0; @@ -658,8 +660,10 @@ static int carm_send_special (struct carm_host *host, carm_sspc_t func) BUG_ON(rc < 0); crq->msg_bucket = (u32) rc; - DPRINTK("blk_insert_request, tag == %u\n", idx); - blk_insert_request(host->oob_q, crq->rq, 1, crq); + DPRINTK("blk_execute_rq_nowait, tag == %u\n", idx); + crq->rq->cmd_type = REQ_TYPE_SPECIAL; + crq->rq->special = crq; + blk_execute_rq_nowait(host->oob_q, NULL, crq->rq, true, NULL); return 0; } @@ -1116,7 +1120,7 @@ static inline void carm_handle_resp(struct carm_host *host, break; case MISC_GET_FW_VER: { struct carm_fw_ver *ver = (struct carm_fw_ver *) - mem + sizeof(struct carm_msg_get_fw_ver); + (mem + sizeof(struct carm_msg_get_fw_ver)); if (!error) { host->fw_ver = le32_to_cpu(ver->version); host->flags |= (ver->features & FL_FW_VER_MASK); diff --git a/drivers/block/ub.c b/drivers/block/ub.c index 0e376d46bdd1..fcec0225ac76 100644 --- a/drivers/block/ub.c +++ b/drivers/block/ub.c @@ -119,43 +119,6 @@ /* */ - -/* command block wrapper */ -struct bulk_cb_wrap { - __le32 Signature; /* contains 'USBC' */ - u32 Tag; /* unique per command id */ - __le32 DataTransferLength; /* size of data */ - u8 Flags; /* direction in bit 0 */ - u8 Lun; /* LUN */ - u8 Length; /* of of the CDB */ - u8 CDB[UB_MAX_CDB_SIZE]; /* max command */ -}; - -#define US_BULK_CB_WRAP_LEN 31 -#define US_BULK_CB_SIGN 0x43425355 /*spells out USBC */ -#define US_BULK_FLAG_IN 1 -#define US_BULK_FLAG_OUT 0 - -/* command status wrapper */ -struct bulk_cs_wrap { - __le32 Signature; /* should = 'USBS' */ - u32 Tag; /* same as original command */ - __le32 Residue; /* amount not transferred */ - u8 Status; /* see below */ -}; - -#define US_BULK_CS_WRAP_LEN 13 -#define US_BULK_CS_SIGN 0x53425355 /* spells out 'USBS' */ -#define US_BULK_STAT_OK 0 -#define US_BULK_STAT_FAIL 1 -#define US_BULK_STAT_PHASE 2 - -/* bulk-only class specific requests */ -#define US_BULK_RESET_REQUEST 0xff -#define US_BULK_GET_MAX_LUN 0xfe - -/* - */ struct ub_dev; #define UB_MAX_REQ_SG 9 /* cdrecord requires 32KB and maybe a header */ @@ -1744,12 +1707,11 @@ static int ub_bd_release(struct gendisk *disk, fmode_t mode) static int ub_bd_ioctl(struct block_device *bdev, fmode_t mode, unsigned int cmd, unsigned long arg) { - struct gendisk *disk = bdev->bd_disk; void __user *usermem = (void __user *) arg; int ret; mutex_lock(&ub_mutex); - ret = scsi_cmd_ioctl(disk->queue, disk, mode, cmd, usermem); + ret = scsi_cmd_blk_ioctl(bdev, mode, cmd, usermem); mutex_unlock(&ub_mutex); return ret; @@ -2478,6 +2440,8 @@ static int __init ub_init(void) int rc; int i; + pr_info("'Low Performance USB Block' driver is deprecated. " + "Please switch to usb-storage\n"); for (i = 0; i < UB_QLOCK_NUM; i++) spin_lock_init(&ub_qlockv[i]); diff --git a/drivers/block/viodasd.c b/drivers/block/viodasd.c deleted file mode 100644 index 9a5b2a2d616d..000000000000 --- a/drivers/block/viodasd.c +++ /dev/null @@ -1,809 +0,0 @@ -/* -*- linux-c -*- - * viodasd.c - * Authors: Dave Boutcher <boutcher@us.ibm.com> - * Ryan Arnold <ryanarn@us.ibm.com> - * Colin Devilbiss <devilbis@us.ibm.com> - * Stephen Rothwell - * - * (C) Copyright 2000-2004 IBM Corporation - * - * This program is free software; you can redistribute it and/or - * modify it under the terms of the GNU General Public License as - * published by the Free Software Foundation; either version 2 of the - * License, or (at your option) any later version. - * - * This program is distributed in the hope that it will be useful, - * but WITHOUT ANY WARRANTY; without even the implied warranty of - * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the - * GNU General Public License for more details. - * - * You should have received a copy of the GNU General Public License - * along with this program; if not, write to the Free Software - * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA - * - * This routine provides access to disk space (termed "DASD" in historical - * IBM terms) owned and managed by an OS/400 partition running on the - * same box as this Linux partition. - * - * All disk operations are performed by sending messages back and forth to - * the OS/400 partition. - */ - -#define pr_fmt(fmt) "viod: " fmt - -#include <linux/major.h> -#include <linux/fs.h> -#include <linux/module.h> -#include <linux/kernel.h> -#include <linux/blkdev.h> -#include <linux/genhd.h> -#include <linux/hdreg.h> -#include <linux/errno.h> -#include <linux/init.h> -#include <linux/string.h> -#include <linux/mutex.h> -#include <linux/dma-mapping.h> -#include <linux/completion.h> -#include <linux/device.h> -#include <linux/scatterlist.h> - -#include <asm/uaccess.h> -#include <asm/vio.h> -#include <asm/iseries/hv_types.h> -#include <asm/iseries/hv_lp_event.h> -#include <asm/iseries/hv_lp_config.h> -#include <asm/iseries/vio.h> -#include <asm/firmware.h> - -MODULE_DESCRIPTION("iSeries Virtual DASD"); -MODULE_AUTHOR("Dave Boutcher"); -MODULE_LICENSE("GPL"); - -/* - * We only support 7 partitions per physical disk....so with minor - * numbers 0-255 we get a maximum of 32 disks. - */ -#define VIOD_GENHD_NAME "iseries/vd" - -#define VIOD_VERS "1.64" - -enum { - PARTITION_SHIFT = 3, - MAX_DISKNO = HVMAXARCHITECTEDVIRTUALDISKS, - MAX_DISK_NAME = FIELD_SIZEOF(struct gendisk, disk_name) -}; - -static DEFINE_MUTEX(viodasd_mutex); -static DEFINE_SPINLOCK(viodasd_spinlock); - -#define VIOMAXREQ 16 - -#define DEVICE_NO(cell) ((struct viodasd_device *)(cell) - &viodasd_devices[0]) - -struct viodasd_waitevent { - struct completion com; - int rc; - u16 sub_result; - int max_disk; /* open */ -}; - -static const struct vio_error_entry viodasd_err_table[] = { - { 0x0201, EINVAL, "Invalid Range" }, - { 0x0202, EINVAL, "Invalid Token" }, - { 0x0203, EIO, "DMA Error" }, - { 0x0204, EIO, "Use Error" }, - { 0x0205, EIO, "Release Error" }, - { 0x0206, EINVAL, "Invalid Disk" }, - { 0x0207, EBUSY, "Can't Lock" }, - { 0x0208, EIO, "Already Locked" }, - { 0x0209, EIO, "Already Unlocked" }, - { 0x020A, EIO, "Invalid Arg" }, - { 0x020B, EIO, "Bad IFS File" }, - { 0x020C, EROFS, "Read Only Device" }, - { 0x02FF, EIO, "Internal Error" }, - { 0x0000, 0, NULL }, -}; - -/* - * Figure out the biggest I/O request (in sectors) we can accept - */ -#define VIODASD_MAXSECTORS (4096 / 512 * VIOMAXBLOCKDMA) - -/* - * Number of disk I/O requests we've sent to OS/400 - */ -static int num_req_outstanding; - -/* - * This is our internal structure for keeping track of disk devices - */ -struct viodasd_device { - u16 cylinders; - u16 tracks; - u16 sectors; - u16 bytes_per_sector; - u64 size; - int read_only; - spinlock_t q_lock; - struct gendisk *disk; - struct device *dev; -} viodasd_devices[MAX_DISKNO]; - -/* - * External open entry point. - */ -static int viodasd_open(struct block_device *bdev, fmode_t mode) -{ - struct viodasd_device *d = bdev->bd_disk->private_data; - HvLpEvent_Rc hvrc; - struct viodasd_waitevent we; - u16 flags = 0; - - if (d->read_only) { - if (mode & FMODE_WRITE) - return -EROFS; - flags = vioblockflags_ro; - } - - init_completion(&we.com); - - /* Send the open event to OS/400 */ - hvrc = HvCallEvent_signalLpEventFast(viopath_hostLp, - HvLpEvent_Type_VirtualIo, - viomajorsubtype_blockio | vioblockopen, - HvLpEvent_AckInd_DoAck, HvLpEvent_AckType_ImmediateAck, - viopath_sourceinst(viopath_hostLp), - viopath_targetinst(viopath_hostLp), - (u64)(unsigned long)&we, VIOVERSION << 16, - ((u64)DEVICE_NO(d) << 48) | ((u64)flags << 32), - 0, 0, 0); - if (hvrc != 0) { - pr_warning("HV open failed %d\n", (int)hvrc); - return -EIO; - } - - wait_for_completion(&we.com); - - /* Check the return code */ - if (we.rc != 0) { - const struct vio_error_entry *err = - vio_lookup_rc(viodasd_err_table, we.sub_result); - - pr_warning("bad rc opening disk: %d:0x%04x (%s)\n", - (int)we.rc, we.sub_result, err->msg); - return -EIO; - } - - return 0; -} - -static int viodasd_unlocked_open(struct block_device *bdev, fmode_t mode) -{ - int ret; - - mutex_lock(&viodasd_mutex); - ret = viodasd_open(bdev, mode); - mutex_unlock(&viodasd_mutex); - - return ret; -} - - -/* - * External release entry point. - */ -static int viodasd_release(struct gendisk *disk, fmode_t mode) -{ - struct viodasd_device *d = disk->private_data; - HvLpEvent_Rc hvrc; - - mutex_lock(&viodasd_mutex); - /* Send the event to OS/400. We DON'T expect a response */ - hvrc = HvCallEvent_signalLpEventFast(viopath_hostLp, - HvLpEvent_Type_VirtualIo, - viomajorsubtype_blockio | vioblockclose, - HvLpEvent_AckInd_NoAck, HvLpEvent_AckType_ImmediateAck, - viopath_sourceinst(viopath_hostLp), - viopath_targetinst(viopath_hostLp), - 0, VIOVERSION << 16, - ((u64)DEVICE_NO(d) << 48) /* | ((u64)flags << 32) */, - 0, 0, 0); - if (hvrc != 0) - pr_warning("HV close call failed %d\n", (int)hvrc); - - mutex_unlock(&viodasd_mutex); - - return 0; -} - - -/* External ioctl entry point. - */ -static int viodasd_getgeo(struct block_device *bdev, struct hd_geometry *geo) -{ - struct gendisk *disk = bdev->bd_disk; - struct viodasd_device *d = disk->private_data; - - geo->sectors = d->sectors ? d->sectors : 32; - geo->heads = d->tracks ? d->tracks : 64; - geo->cylinders = d->cylinders ? d->cylinders : - get_capacity(disk) / (geo->sectors * geo->heads); - - return 0; -} - -/* - * Our file operations table - */ -static const struct block_device_operations viodasd_fops = { - .owner = THIS_MODULE, - .open = viodasd_unlocked_open, - .release = viodasd_release, - .getgeo = viodasd_getgeo, -}; - -/* - * End a request - */ -static void viodasd_end_request(struct request *req, int error, - int num_sectors) -{ - __blk_end_request(req, error, num_sectors << 9); -} - -/* - * Send an actual I/O request to OS/400 - */ -static int send_request(struct request *req) -{ - u64 start; - int direction; - int nsg; - u16 viocmd; - HvLpEvent_Rc hvrc; - struct vioblocklpevent *bevent; - struct HvLpEvent *hev; - struct scatterlist sg[VIOMAXBLOCKDMA]; - int sgindex; - struct viodasd_device *d; - unsigned long flags; - - start = (u64)blk_rq_pos(req) << 9; - - if (rq_data_dir(req) == READ) { - direction = DMA_FROM_DEVICE; - viocmd = viomajorsubtype_blockio | vioblockread; - } else { - direction = DMA_TO_DEVICE; - viocmd = viomajorsubtype_blockio | vioblockwrite; - } - - d = req->rq_disk->private_data; - - /* Now build the scatter-gather list */ - sg_init_table(sg, VIOMAXBLOCKDMA); - nsg = blk_rq_map_sg(req->q, req, sg); - nsg = dma_map_sg(d->dev, sg, nsg, direction); - - spin_lock_irqsave(&viodasd_spinlock, flags); - num_req_outstanding++; - - /* This optimization handles a single DMA block */ - if (nsg == 1) - hvrc = HvCallEvent_signalLpEventFast(viopath_hostLp, - HvLpEvent_Type_VirtualIo, viocmd, - HvLpEvent_AckInd_DoAck, - HvLpEvent_AckType_ImmediateAck, - viopath_sourceinst(viopath_hostLp), - viopath_targetinst(viopath_hostLp), - (u64)(unsigned long)req, VIOVERSION << 16, - ((u64)DEVICE_NO(d) << 48), start, - ((u64)sg_dma_address(&sg[0])) << 32, - sg_dma_len(&sg[0])); - else { - bevent = (struct vioblocklpevent *) - vio_get_event_buffer(viomajorsubtype_blockio); - if (bevent == NULL) { - pr_warning("error allocating disk event buffer\n"); - goto error_ret; - } - - /* - * Now build up the actual request. Note that we store - * the pointer to the request in the correlation - * token so we can match the response up later - */ - memset(bevent, 0, sizeof(struct vioblocklpevent)); - hev = &bevent->event; - hev->flags = HV_LP_EVENT_VALID | HV_LP_EVENT_DO_ACK | - HV_LP_EVENT_INT; - hev->xType = HvLpEvent_Type_VirtualIo; - hev->xSubtype = viocmd; - hev->xSourceLp = HvLpConfig_getLpIndex(); - hev->xTargetLp = viopath_hostLp; - hev->xSizeMinus1 = - offsetof(struct vioblocklpevent, u.rw_data.dma_info) + - (sizeof(bevent->u.rw_data.dma_info[0]) * nsg) - 1; - hev->xSourceInstanceId = viopath_sourceinst(viopath_hostLp); - hev->xTargetInstanceId = viopath_targetinst(viopath_hostLp); - hev->xCorrelationToken = (u64)req; - bevent->version = VIOVERSION; - bevent->disk = DEVICE_NO(d); - bevent->u.rw_data.offset = start; - - /* - * Copy just the dma information from the sg list - * into the request - */ - for (sgindex = 0; sgindex < nsg; sgindex++) { - bevent->u.rw_data.dma_info[sgindex].token = - sg_dma_address(&sg[sgindex]); - bevent->u.rw_data.dma_info[sgindex].len = - sg_dma_len(&sg[sgindex]); - } - - /* Send the request */ - hvrc = HvCallEvent_signalLpEvent(&bevent->event); - vio_free_event_buffer(viomajorsubtype_blockio, bevent); - } - - if (hvrc != HvLpEvent_Rc_Good) { - pr_warning("error sending disk event to OS/400 (rc %d)\n", - (int)hvrc); - goto error_ret; - } - spin_unlock_irqrestore(&viodasd_spinlock, flags); - return 0; - -error_ret: - num_req_outstanding--; - spin_unlock_irqrestore(&viodasd_spinlock, flags); - dma_unmap_sg(d->dev, sg, nsg, direction); - return -1; -} - -/* - * This is the external request processing routine - */ -static void do_viodasd_request(struct request_queue *q) -{ - struct request *req; - - /* - * If we already have the maximum number of requests - * outstanding to OS/400 just bail out. We'll come - * back later. - */ - while (num_req_outstanding < VIOMAXREQ) { - req = blk_fetch_request(q); - if (req == NULL) - return; - /* check that request contains a valid command */ - if (req->cmd_type != REQ_TYPE_FS) { - viodasd_end_request(req, -EIO, blk_rq_sectors(req)); - continue; - } - /* Try sending the request */ - if (send_request(req) != 0) - viodasd_end_request(req, -EIO, blk_rq_sectors(req)); - } -} - -/* - * Probe a single disk and fill in the viodasd_device structure - * for it. - */ -static int probe_disk(struct viodasd_device *d) -{ - HvLpEvent_Rc hvrc; - struct viodasd_waitevent we; - int dev_no = DEVICE_NO(d); - struct gendisk *g; - struct request_queue *q; - u16 flags = 0; - -retry: - init_completion(&we.com); - - /* Send the open event to OS/400 */ - hvrc = HvCallEvent_signalLpEventFast(viopath_hostLp, - HvLpEvent_Type_VirtualIo, - viomajorsubtype_blockio | vioblockopen, - HvLpEvent_AckInd_DoAck, HvLpEvent_AckType_ImmediateAck, - viopath_sourceinst(viopath_hostLp), - viopath_targetinst(viopath_hostLp), - (u64)(unsigned long)&we, VIOVERSION << 16, - ((u64)dev_no << 48) | ((u64)flags<< 32), - 0, 0, 0); - if (hvrc != 0) { - pr_warning("bad rc on HV open %d\n", (int)hvrc); - return 0; - } - - wait_for_completion(&we.com); - - if (we.rc != 0) { - if (flags != 0) - return 0; - /* try again with read only flag set */ - flags = vioblockflags_ro; - goto retry; - } - if (we.max_disk > (MAX_DISKNO - 1)) { - printk_once(KERN_INFO pr_fmt("Only examining the first %d of %d disks connected\n"), - MAX_DISKNO, we.max_disk + 1); - } - - /* Send the close event to OS/400. We DON'T expect a response */ - hvrc = HvCallEvent_signalLpEventFast(viopath_hostLp, - HvLpEvent_Type_VirtualIo, - viomajorsubtype_blockio | vioblockclose, - HvLpEvent_AckInd_NoAck, HvLpEvent_AckType_ImmediateAck, - viopath_sourceinst(viopath_hostLp), - viopath_targetinst(viopath_hostLp), - 0, VIOVERSION << 16, - ((u64)dev_no << 48) | ((u64)flags << 32), - 0, 0, 0); - if (hvrc != 0) { - pr_warning("bad rc sending event to OS/400 %d\n", (int)hvrc); - return 0; - } - - if (d->dev == NULL) { - /* this is when we reprobe for new disks */ - if (vio_create_viodasd(dev_no) == NULL) { - pr_warning("cannot allocate virtual device for disk %d\n", - dev_no); - return 0; - } - /* - * The vio_create_viodasd will have recursed into this - * routine with d->dev set to the new vio device and - * will finish the setup of the disk below. - */ - return 1; - } - - /* create the request queue for the disk */ - spin_lock_init(&d->q_lock); - q = blk_init_queue(do_viodasd_request, &d->q_lock); - if (q == NULL) { - pr_warning("cannot allocate queue for disk %d\n", dev_no); - return 0; - } - g = alloc_disk(1 << PARTITION_SHIFT); - if (g == NULL) { - pr_warning("cannot allocate disk structure for disk %d\n", - dev_no); - blk_cleanup_queue(q); - return 0; - } - - d->disk = g; - blk_queue_max_segments(q, VIOMAXBLOCKDMA); - blk_queue_max_hw_sectors(q, VIODASD_MAXSECTORS); - g->major = VIODASD_MAJOR; - g->first_minor = dev_no << PARTITION_SHIFT; - if (dev_no >= 26) - snprintf(g->disk_name, sizeof(g->disk_name), - VIOD_GENHD_NAME "%c%c", - 'a' + (dev_no / 26) - 1, 'a' + (dev_no % 26)); - else - snprintf(g->disk_name, sizeof(g->disk_name), - VIOD_GENHD_NAME "%c", 'a' + (dev_no % 26)); - g->fops = &viodasd_fops; - g->queue = q; - g->private_data = d; - g->driverfs_dev = d->dev; - set_capacity(g, d->size >> 9); - - pr_info("disk %d: %lu sectors (%lu MB) CHS=%d/%d/%d sector size %d%s\n", - dev_no, (unsigned long)(d->size >> 9), - (unsigned long)(d->size >> 20), - (int)d->cylinders, (int)d->tracks, - (int)d->sectors, (int)d->bytes_per_sector, - d->read_only ? " (RO)" : ""); - - /* register us in the global list */ - add_disk(g); - return 1; -} - -/* returns the total number of scatterlist elements converted */ -static int block_event_to_scatterlist(const struct vioblocklpevent *bevent, - struct scatterlist *sg, int *total_len) -{ - int i, numsg; - const struct rw_data *rw_data = &bevent->u.rw_data; - static const int offset = - offsetof(struct vioblocklpevent, u.rw_data.dma_info); - static const int element_size = sizeof(rw_data->dma_info[0]); - - numsg = ((bevent->event.xSizeMinus1 + 1) - offset) / element_size; - if (numsg > VIOMAXBLOCKDMA) - numsg = VIOMAXBLOCKDMA; - - *total_len = 0; - sg_init_table(sg, VIOMAXBLOCKDMA); - for (i = 0; (i < numsg) && (rw_data->dma_info[i].len > 0); ++i) { - sg_dma_address(&sg[i]) = rw_data->dma_info[i].token; - sg_dma_len(&sg[i]) = rw_data->dma_info[i].len; - *total_len += rw_data->dma_info[i].len; - } - return i; -} - -/* - * Restart all queues, starting with the one _after_ the disk given, - * thus reducing the chance of starvation of higher numbered disks. - */ -static void viodasd_restart_all_queues_starting_from(int first_index) -{ - int i; - - for (i = first_index + 1; i < MAX_DISKNO; ++i) - if (viodasd_devices[i].disk) - blk_run_queue(viodasd_devices[i].disk->queue); - for (i = 0; i <= first_index; ++i) - if (viodasd_devices[i].disk) - blk_run_queue(viodasd_devices[i].disk->queue); -} - -/* - * For read and write requests, decrement the number of outstanding requests, - * Free the DMA buffers we allocated. - */ -static int viodasd_handle_read_write(struct vioblocklpevent *bevent) -{ - int num_sg, num_sect, pci_direction, total_len; - struct request *req; - struct scatterlist sg[VIOMAXBLOCKDMA]; - struct HvLpEvent *event = &bevent->event; - unsigned long irq_flags; - struct viodasd_device *d; - int error; - spinlock_t *qlock; - - num_sg = block_event_to_scatterlist(bevent, sg, &total_len); - num_sect = total_len >> 9; - if (event->xSubtype == (viomajorsubtype_blockio | vioblockread)) - pci_direction = DMA_FROM_DEVICE; - else - pci_direction = DMA_TO_DEVICE; - req = (struct request *)bevent->event.xCorrelationToken; - d = req->rq_disk->private_data; - - dma_unmap_sg(d->dev, sg, num_sg, pci_direction); - - /* - * Since this is running in interrupt mode, we need to make sure - * we're not stepping on any global I/O operations - */ - spin_lock_irqsave(&viodasd_spinlock, irq_flags); - num_req_outstanding--; - spin_unlock_irqrestore(&viodasd_spinlock, irq_flags); - - error = (event->xRc == HvLpEvent_Rc_Good) ? 0 : -EIO; - if (error) { - const struct vio_error_entry *err; - err = vio_lookup_rc(viodasd_err_table, bevent->sub_result); - pr_warning("read/write error %d:0x%04x (%s)\n", - event->xRc, bevent->sub_result, err->msg); - num_sect = blk_rq_sectors(req); - } - qlock = req->q->queue_lock; - spin_lock_irqsave(qlock, irq_flags); - viodasd_end_request(req, error, num_sect); - spin_unlock_irqrestore(qlock, irq_flags); - - /* Finally, try to get more requests off of this device's queue */ - viodasd_restart_all_queues_starting_from(DEVICE_NO(d)); - - return 0; -} - -/* This routine handles incoming block LP events */ -static void handle_block_event(struct HvLpEvent *event) -{ - struct vioblocklpevent *bevent = (struct vioblocklpevent *)event; - struct viodasd_waitevent *pwe; - - if (event == NULL) - /* Notification that a partition went away! */ - return; - /* First, we should NEVER get an int here...only acks */ - if (hvlpevent_is_int(event)) { - pr_warning("Yikes! got an int in viodasd event handler!\n"); - if (hvlpevent_need_ack(event)) { - event->xRc = HvLpEvent_Rc_InvalidSubtype; - HvCallEvent_ackLpEvent(event); - } - } - - switch (event->xSubtype & VIOMINOR_SUBTYPE_MASK) { - case vioblockopen: - /* - * Handle a response to an open request. We get all the - * disk information in the response, so update it. The - * correlation token contains a pointer to a waitevent - * structure that has a completion in it. update the - * return code in the waitevent structure and post the - * completion to wake up the guy who sent the request - */ - pwe = (struct viodasd_waitevent *)event->xCorrelationToken; - pwe->rc = event->xRc; - pwe->sub_result = bevent->sub_result; - if (event->xRc == HvLpEvent_Rc_Good) { - const struct open_data *data = &bevent->u.open_data; - struct viodasd_device *device = - &viodasd_devices[bevent->disk]; - device->read_only = - bevent->flags & vioblockflags_ro; - device->size = data->disk_size; - device->cylinders = data->cylinders; - device->tracks = data->tracks; - device->sectors = data->sectors; - device->bytes_per_sector = data->bytes_per_sector; - pwe->max_disk = data->max_disk; - } - complete(&pwe->com); - break; - case vioblockclose: - break; - case vioblockread: - case vioblockwrite: - viodasd_handle_read_write(bevent); - break; - - default: - pr_warning("invalid subtype!"); - if (hvlpevent_need_ack(event)) { - event->xRc = HvLpEvent_Rc_InvalidSubtype; - HvCallEvent_ackLpEvent(event); - } - } -} - -/* - * Get the driver to reprobe for more disks. - */ -static ssize_t probe_disks(struct device_driver *drv, const char *buf, - size_t count) -{ - struct viodasd_device *d; - - for (d = viodasd_devices; d < &viodasd_devices[MAX_DISKNO]; d++) { - if (d->disk == NULL) - probe_disk(d); - } - return count; -} -static DRIVER_ATTR(probe, S_IWUSR, NULL, probe_disks); - -static int viodasd_probe(struct vio_dev *vdev, const struct vio_device_id *id) -{ - struct viodasd_device *d = &viodasd_devices[vdev->unit_address]; - - d->dev = &vdev->dev; - if (!probe_disk(d)) - return -ENODEV; - return 0; -} - -static int viodasd_remove(struct vio_dev *vdev) -{ - struct viodasd_device *d; - - d = &viodasd_devices[vdev->unit_address]; - if (d->disk) { - del_gendisk(d->disk); - blk_cleanup_queue(d->disk->queue); - put_disk(d->disk); - d->disk = NULL; - } - d->dev = NULL; - return 0; -} - -/** - * viodasd_device_table: Used by vio.c to match devices that we - * support. - */ -static struct vio_device_id viodasd_device_table[] __devinitdata = { - { "block", "IBM,iSeries-viodasd" }, - { "", "" } -}; -MODULE_DEVICE_TABLE(vio, viodasd_device_table); - -static struct vio_driver viodasd_driver = { - .id_table = viodasd_device_table, - .probe = viodasd_probe, - .remove = viodasd_remove, - .driver = { - .name = "viodasd", - .owner = THIS_MODULE, - } -}; - -static int need_delete_probe; - -/* - * Initialize the whole device driver. Handle module and non-module - * versions - */ -static int __init viodasd_init(void) -{ - int rc; - - if (!firmware_has_feature(FW_FEATURE_ISERIES)) { - rc = -ENODEV; - goto early_fail; - } - - /* Try to open to our host lp */ - if (viopath_hostLp == HvLpIndexInvalid) - vio_set_hostlp(); - - if (viopath_hostLp == HvLpIndexInvalid) { - pr_warning("invalid hosting partition\n"); - rc = -EIO; - goto early_fail; - } - - pr_info("vers " VIOD_VERS ", hosting partition %d\n", viopath_hostLp); - - /* register the block device */ - rc = register_blkdev(VIODASD_MAJOR, VIOD_GENHD_NAME); - if (rc) { - pr_warning("Unable to get major number %d for %s\n", - VIODASD_MAJOR, VIOD_GENHD_NAME); - goto early_fail; - } - /* Actually open the path to the hosting partition */ - rc = viopath_open(viopath_hostLp, viomajorsubtype_blockio, - VIOMAXREQ + 2); - if (rc) { - pr_warning("error opening path to host partition %d\n", - viopath_hostLp); - goto unregister_blk; - } - - /* Initialize our request handler */ - vio_setHandler(viomajorsubtype_blockio, handle_block_event); - - rc = vio_register_driver(&viodasd_driver); - if (rc) { - pr_warning("vio_register_driver failed\n"); - goto unset_handler; - } - - /* - * If this call fails, it just means that we cannot dynamically - * add virtual disks, but the driver will still work fine for - * all existing disk, so ignore the failure. - */ - if (!driver_create_file(&viodasd_driver.driver, &driver_attr_probe)) - need_delete_probe = 1; - - return 0; - -unset_handler: - vio_clearHandler(viomajorsubtype_blockio); - viopath_close(viopath_hostLp, viomajorsubtype_blockio, VIOMAXREQ + 2); -unregister_blk: - unregister_blkdev(VIODASD_MAJOR, VIOD_GENHD_NAME); -early_fail: - return rc; -} -module_init(viodasd_init); - -void __exit viodasd_exit(void) -{ - if (need_delete_probe) - driver_remove_file(&viodasd_driver.driver, &driver_attr_probe); - vio_unregister_driver(&viodasd_driver); - vio_clearHandler(viomajorsubtype_blockio); - viopath_close(viopath_hostLp, viomajorsubtype_blockio, VIOMAXREQ + 2); - unregister_blkdev(VIODASD_MAJOR, VIOD_GENHD_NAME); -} -module_exit(viodasd_exit); diff --git a/drivers/block/virtio_blk.c b/drivers/block/virtio_blk.c index 4d0b70adf5f7..c4a60badf252 100644 --- a/drivers/block/virtio_blk.c +++ b/drivers/block/virtio_blk.c @@ -4,6 +4,7 @@ #include <linux/blkdev.h> #include <linux/hdreg.h> #include <linux/module.h> +#include <linux/mutex.h> #include <linux/virtio.h> #include <linux/virtio_blk.h> #include <linux/scatterlist.h> @@ -36,6 +37,12 @@ struct virtio_blk /* Process context for config space updates */ struct work_struct config_work; + /* Lock for config space updates */ + struct mutex config_lock; + + /* enable config space updates */ + bool config_enable; + /* What host tells us, plus 2 for header & tailer. */ unsigned int sg_elems; @@ -172,7 +179,7 @@ static bool do_req(struct request_queue *q, struct virtio_blk *vblk, } } - if (virtqueue_add_buf(vblk->vq, vblk->sg, out, in, vbr) < 0) { + if (virtqueue_add_buf(vblk->vq, vblk->sg, out, in, vbr, GFP_ATOMIC)<0) { mempool_free(vbr, vblk->pool); return false; } @@ -243,8 +250,8 @@ static int virtblk_ioctl(struct block_device *bdev, fmode_t mode, if (!virtio_has_feature(vblk->vdev, VIRTIO_BLK_F_SCSI)) return -ENOTTY; - return scsi_cmd_ioctl(disk->queue, disk, mode, cmd, - (void __user *)data); + return scsi_cmd_blk_ioctl(bdev, mode, cmd, + (void __user *)data); } /* We provide getgeo only to please some old bootloader/partitioning tools */ @@ -318,6 +325,10 @@ static void virtblk_config_changed_work(struct work_struct *work) char cap_str_2[10], cap_str_10[10]; u64 capacity, size; + mutex_lock(&vblk->config_lock); + if (!vblk->config_enable) + goto done; + /* Host must always specify the capacity. */ vdev->config->get(vdev, offsetof(struct virtio_blk_config, capacity), &capacity, sizeof(capacity)); @@ -340,6 +351,8 @@ static void virtblk_config_changed_work(struct work_struct *work) cap_str_10, cap_str_2); set_capacity(vblk->disk, capacity); +done: + mutex_unlock(&vblk->config_lock); } static void virtblk_config_changed(struct virtio_device *vdev) @@ -349,6 +362,18 @@ static void virtblk_config_changed(struct virtio_device *vdev) queue_work(virtblk_wq, &vblk->config_work); } +static int init_vq(struct virtio_blk *vblk) +{ + int err = 0; + + /* We expect one virtqueue, for output. */ + vblk->vq = virtio_find_single_vq(vblk->vdev, blk_done, "requests"); + if (IS_ERR(vblk->vq)) + err = PTR_ERR(vblk->vq); + + return err; +} + static int __devinit virtblk_probe(struct virtio_device *vdev) { struct virtio_blk *vblk; @@ -388,14 +413,13 @@ static int __devinit virtblk_probe(struct virtio_device *vdev) vblk->vdev = vdev; vblk->sg_elems = sg_elems; sg_init_table(vblk->sg, vblk->sg_elems); + mutex_init(&vblk->config_lock); INIT_WORK(&vblk->config_work, virtblk_config_changed_work); + vblk->config_enable = true; - /* We expect one virtqueue, for output. */ - vblk->vq = virtio_find_single_vq(vdev, blk_done, "requests"); - if (IS_ERR(vblk->vq)) { - err = PTR_ERR(vblk->vq); + err = init_vq(vblk); + if (err) goto out_free_vblk; - } vblk->pool = mempool_create_kmalloc_pool(1,sizeof(struct virtblk_req)); if (!vblk->pool) { @@ -542,7 +566,10 @@ static void __devexit virtblk_remove(struct virtio_device *vdev) struct virtio_blk *vblk = vdev->priv; int index = vblk->index; - flush_work(&vblk->config_work); + /* Prevent config work handler from accessing the device. */ + mutex_lock(&vblk->config_lock); + vblk->config_enable = false; + mutex_unlock(&vblk->config_lock); /* Nothing should be pending. */ BUG_ON(!list_empty(&vblk->reqs)); @@ -550,6 +577,8 @@ static void __devexit virtblk_remove(struct virtio_device *vdev) /* Stop all the virtqueues. */ vdev->config->reset(vdev); + flush_work(&vblk->config_work); + del_gendisk(vblk->disk); blk_cleanup_queue(vblk->disk->queue); put_disk(vblk->disk); @@ -559,6 +588,46 @@ static void __devexit virtblk_remove(struct virtio_device *vdev) ida_simple_remove(&vd_index_ida, index); } +#ifdef CONFIG_PM +static int virtblk_freeze(struct virtio_device *vdev) +{ + struct virtio_blk *vblk = vdev->priv; + + /* Ensure we don't receive any more interrupts */ + vdev->config->reset(vdev); + + /* Prevent config work handler from accessing the device. */ + mutex_lock(&vblk->config_lock); + vblk->config_enable = false; + mutex_unlock(&vblk->config_lock); + + flush_work(&vblk->config_work); + + spin_lock_irq(vblk->disk->queue->queue_lock); + blk_stop_queue(vblk->disk->queue); + spin_unlock_irq(vblk->disk->queue->queue_lock); + blk_sync_queue(vblk->disk->queue); + + vdev->config->del_vqs(vdev); + return 0; +} + +static int virtblk_restore(struct virtio_device *vdev) +{ + struct virtio_blk *vblk = vdev->priv; + int ret; + + vblk->config_enable = true; + ret = init_vq(vdev->priv); + if (!ret) { + spin_lock_irq(vblk->disk->queue->queue_lock); + blk_start_queue(vblk->disk->queue); + spin_unlock_irq(vblk->disk->queue->queue_lock); + } + return ret; +} +#endif + static const struct virtio_device_id id_table[] = { { VIRTIO_ID_BLOCK, VIRTIO_DEV_ANY_ID }, { 0 }, @@ -584,6 +653,10 @@ static struct virtio_driver __refdata virtio_blk = { .probe = virtblk_probe, .remove = __devexit_p(virtblk_remove), .config_changed = virtblk_config_changed, +#ifdef CONFIG_PM + .freeze = virtblk_freeze, + .restore = virtblk_restore, +#endif }; static int __init init(void) diff --git a/drivers/block/xd.c b/drivers/block/xd.c index 4abd2bcd20fb..ff540520bada 100644 --- a/drivers/block/xd.c +++ b/drivers/block/xd.c @@ -52,7 +52,6 @@ #include <linux/io.h> #include <linux/gfp.h> -#include <asm/system.h> #include <asm/uaccess.h> #include <asm/dma.h> @@ -148,7 +147,7 @@ static volatile int xdc_busy; static struct timer_list xd_watchdog_int; static volatile u_char xd_error; -static int nodma = XD_DONT_USE_DMA; +static bool nodma = XD_DONT_USE_DMA; static struct request_queue *xd_queue; diff --git a/drivers/block/xen-blkback/blkback.c b/drivers/block/xen-blkback/blkback.c index 15ec4db194d1..0088bf60f368 100644 --- a/drivers/block/xen-blkback/blkback.c +++ b/drivers/block/xen-blkback/blkback.c @@ -39,9 +39,6 @@ #include <linux/list.h> #include <linux/delay.h> #include <linux/freezer.h> -#include <linux/loop.h> -#include <linux/falloc.h> -#include <linux/fs.h> #include <xen/events.h> #include <xen/page.h> @@ -362,7 +359,7 @@ static int xen_blkbk_map(struct blkif_request *req, { struct gnttab_map_grant_ref map[BLKIF_MAX_SEGMENTS_PER_REQUEST]; int i; - int nseg = req->nr_segments; + int nseg = req->u.rw.nr_segments; int ret = 0; /* @@ -416,30 +413,25 @@ static int xen_blkbk_map(struct blkif_request *req, return ret; } -static void xen_blk_discard(struct xen_blkif *blkif, struct blkif_request *req) +static int dispatch_discard_io(struct xen_blkif *blkif, + struct blkif_request *req) { int err = 0; int status = BLKIF_RSP_OKAY; struct block_device *bdev = blkif->vbd.bdev; - if (blkif->blk_backend_type == BLKIF_BACKEND_PHY) - /* just forward the discard request */ + blkif->st_ds_req++; + + xen_blkif_get(blkif); + if (blkif->blk_backend_type == BLKIF_BACKEND_PHY || + blkif->blk_backend_type == BLKIF_BACKEND_FILE) { + unsigned long secure = (blkif->vbd.discard_secure && + (req->u.discard.flag & BLKIF_DISCARD_SECURE)) ? + BLKDEV_DISCARD_SECURE : 0; err = blkdev_issue_discard(bdev, req->u.discard.sector_number, req->u.discard.nr_sectors, - GFP_KERNEL, 0); - else if (blkif->blk_backend_type == BLKIF_BACKEND_FILE) { - /* punch a hole in the backing file */ - struct loop_device *lo = bdev->bd_disk->private_data; - struct file *file = lo->lo_backing_file; - - if (file->f_op->fallocate) - err = file->f_op->fallocate(file, - FALLOC_FL_KEEP_SIZE | FALLOC_FL_PUNCH_HOLE, - req->u.discard.sector_number << 9, - req->u.discard.nr_sectors << 9); - else - err = -EOPNOTSUPP; + GFP_KERNEL, secure); } else err = -EOPNOTSUPP; @@ -449,7 +441,9 @@ static void xen_blk_discard(struct xen_blkif *blkif, struct blkif_request *req) } else if (err) status = BLKIF_RSP_ERROR; - make_response(blkif, req->id, req->operation, status); + make_response(blkif, req->u.discard.id, req->operation, status); + xen_blkif_put(blkif); + return err; } static void xen_blk_drain_io(struct xen_blkif *blkif) @@ -573,8 +567,11 @@ __do_block_io_op(struct xen_blkif *blkif) /* Apply all sanity checks to /private copy/ of request. */ barrier(); - - if (dispatch_rw_block_io(blkif, &req, pending_req)) + if (unlikely(req.operation == BLKIF_OP_DISCARD)) { + free_req(pending_req); + if (dispatch_discard_io(blkif, &req)) + break; + } else if (dispatch_rw_block_io(blkif, &req, pending_req)) break; /* Yield point for this unbounded loop. */ @@ -633,10 +630,6 @@ static int dispatch_rw_block_io(struct xen_blkif *blkif, blkif->st_f_req++; operation = WRITE_FLUSH; break; - case BLKIF_OP_DISCARD: - blkif->st_ds_req++; - operation = REQ_DISCARD; - break; default: operation = 0; /* make gcc happy */ goto fail_response; @@ -644,9 +637,9 @@ static int dispatch_rw_block_io(struct xen_blkif *blkif, } /* Check that the number of segments is sane. */ - nseg = req->nr_segments; - if (unlikely(nseg == 0 && operation != WRITE_FLUSH && - operation != REQ_DISCARD) || + nseg = req->u.rw.nr_segments; + + if (unlikely(nseg == 0 && operation != WRITE_FLUSH) || unlikely(nseg > BLKIF_MAX_SEGMENTS_PER_REQUEST)) { pr_debug(DRV_PFX "Bad number of segments in request (%d)\n", nseg); @@ -654,12 +647,12 @@ static int dispatch_rw_block_io(struct xen_blkif *blkif, goto fail_response; } - preq.dev = req->handle; + preq.dev = req->u.rw.handle; preq.sector_number = req->u.rw.sector_number; preq.nr_sects = 0; pending_req->blkif = blkif; - pending_req->id = req->id; + pending_req->id = req->u.rw.id; pending_req->operation = req->operation; pending_req->status = BLKIF_RSP_OKAY; pending_req->nr_pages = nseg; @@ -707,7 +700,7 @@ static int dispatch_rw_block_io(struct xen_blkif *blkif, * the hypercall to unmap the grants - that is all done in * xen_blkbk_unmap. */ - if (operation != REQ_DISCARD && xen_blkbk_map(req, pending_req, seg)) + if (xen_blkbk_map(req, pending_req, seg)) goto fail_flush; /* @@ -739,23 +732,16 @@ static int dispatch_rw_block_io(struct xen_blkif *blkif, /* This will be hit if the operation was a flush or discard. */ if (!bio) { - BUG_ON(operation != WRITE_FLUSH && operation != REQ_DISCARD); + BUG_ON(operation != WRITE_FLUSH); - if (operation == WRITE_FLUSH) { - bio = bio_alloc(GFP_KERNEL, 0); - if (unlikely(bio == NULL)) - goto fail_put_bio; + bio = bio_alloc(GFP_KERNEL, 0); + if (unlikely(bio == NULL)) + goto fail_put_bio; - biolist[nbio++] = bio; - bio->bi_bdev = preq.bdev; - bio->bi_private = pending_req; - bio->bi_end_io = end_block_io_op; - } else if (operation == REQ_DISCARD) { - xen_blk_discard(blkif, req); - xen_blkif_put(blkif); - free_req(pending_req); - return 0; - } + biolist[nbio++] = bio; + bio->bi_bdev = preq.bdev; + bio->bi_private = pending_req; + bio->bi_end_io = end_block_io_op; } /* @@ -784,7 +770,7 @@ static int dispatch_rw_block_io(struct xen_blkif *blkif, xen_blkbk_unmap(pending_req); fail_response: /* Haven't submitted any bio's yet. */ - make_response(blkif, req->id, req->operation, BLKIF_RSP_ERROR); + make_response(blkif, req->u.rw.id, req->operation, BLKIF_RSP_ERROR); free_req(pending_req); msleep(1); /* back off a bit */ return -EIO; diff --git a/drivers/block/xen-blkback/common.h b/drivers/block/xen-blkback/common.h index dfb1b3a43a5d..d0ee7edc9be8 100644 --- a/drivers/block/xen-blkback/common.h +++ b/drivers/block/xen-blkback/common.h @@ -60,58 +60,66 @@ struct blkif_common_response { char dummy; }; -/* i386 protocol version */ -#pragma pack(push, 4) - struct blkif_x86_32_request_rw { + uint8_t nr_segments; /* number of segments */ + blkif_vdev_t handle; /* only for read/write requests */ + uint64_t id; /* private guest value, echoed in resp */ blkif_sector_t sector_number;/* start sector idx on disk (r/w only) */ struct blkif_request_segment seg[BLKIF_MAX_SEGMENTS_PER_REQUEST]; -}; +} __attribute__((__packed__)); struct blkif_x86_32_request_discard { + uint8_t flag; /* BLKIF_DISCARD_SECURE or zero */ + blkif_vdev_t _pad1; /* was "handle" for read/write requests */ + uint64_t id; /* private guest value, echoed in resp */ blkif_sector_t sector_number;/* start sector idx on disk (r/w only) */ - uint64_t nr_sectors; -}; + uint64_t nr_sectors; +} __attribute__((__packed__)); struct blkif_x86_32_request { uint8_t operation; /* BLKIF_OP_??? */ - uint8_t nr_segments; /* number of segments */ - blkif_vdev_t handle; /* only for read/write requests */ - uint64_t id; /* private guest value, echoed in resp */ union { struct blkif_x86_32_request_rw rw; struct blkif_x86_32_request_discard discard; } u; -}; +} __attribute__((__packed__)); + +/* i386 protocol version */ +#pragma pack(push, 4) struct blkif_x86_32_response { uint64_t id; /* copied from request */ uint8_t operation; /* copied from request */ int16_t status; /* BLKIF_RSP_??? */ }; #pragma pack(pop) - /* x86_64 protocol version */ struct blkif_x86_64_request_rw { + uint8_t nr_segments; /* number of segments */ + blkif_vdev_t handle; /* only for read/write requests */ + uint32_t _pad1; /* offsetof(blkif_reqest..,u.rw.id)==8 */ + uint64_t id; blkif_sector_t sector_number;/* start sector idx on disk (r/w only) */ struct blkif_request_segment seg[BLKIF_MAX_SEGMENTS_PER_REQUEST]; -}; +} __attribute__((__packed__)); struct blkif_x86_64_request_discard { + uint8_t flag; /* BLKIF_DISCARD_SECURE or zero */ + blkif_vdev_t _pad1; /* was "handle" for read/write requests */ + uint32_t _pad2; /* offsetof(blkif_..,u.discard.id)==8 */ + uint64_t id; blkif_sector_t sector_number;/* start sector idx on disk (r/w only) */ - uint64_t nr_sectors; -}; + uint64_t nr_sectors; +} __attribute__((__packed__)); struct blkif_x86_64_request { uint8_t operation; /* BLKIF_OP_??? */ - uint8_t nr_segments; /* number of segments */ - blkif_vdev_t handle; /* only for read/write requests */ - uint64_t __attribute__((__aligned__(8))) id; union { struct blkif_x86_64_request_rw rw; struct blkif_x86_64_request_discard discard; } u; -}; +} __attribute__((__packed__)); + struct blkif_x86_64_response { uint64_t __attribute__((__aligned__(8))) id; uint8_t operation; /* copied from request */ @@ -156,6 +164,7 @@ struct xen_vbd { /* Cached size parameter. */ sector_t size; bool flush_support; + bool discard_secure; }; struct backend_info; @@ -237,22 +246,23 @@ static inline void blkif_get_x86_32_req(struct blkif_request *dst, { int i, n = BLKIF_MAX_SEGMENTS_PER_REQUEST; dst->operation = src->operation; - dst->nr_segments = src->nr_segments; - dst->handle = src->handle; - dst->id = src->id; switch (src->operation) { case BLKIF_OP_READ: case BLKIF_OP_WRITE: case BLKIF_OP_WRITE_BARRIER: case BLKIF_OP_FLUSH_DISKCACHE: + dst->u.rw.nr_segments = src->u.rw.nr_segments; + dst->u.rw.handle = src->u.rw.handle; + dst->u.rw.id = src->u.rw.id; dst->u.rw.sector_number = src->u.rw.sector_number; barrier(); - if (n > dst->nr_segments) - n = dst->nr_segments; + if (n > dst->u.rw.nr_segments) + n = dst->u.rw.nr_segments; for (i = 0; i < n; i++) dst->u.rw.seg[i] = src->u.rw.seg[i]; break; case BLKIF_OP_DISCARD: + dst->u.discard.flag = src->u.discard.flag; dst->u.discard.sector_number = src->u.discard.sector_number; dst->u.discard.nr_sectors = src->u.discard.nr_sectors; break; @@ -266,22 +276,23 @@ static inline void blkif_get_x86_64_req(struct blkif_request *dst, { int i, n = BLKIF_MAX_SEGMENTS_PER_REQUEST; dst->operation = src->operation; - dst->nr_segments = src->nr_segments; - dst->handle = src->handle; - dst->id = src->id; switch (src->operation) { case BLKIF_OP_READ: case BLKIF_OP_WRITE: case BLKIF_OP_WRITE_BARRIER: case BLKIF_OP_FLUSH_DISKCACHE: + dst->u.rw.nr_segments = src->u.rw.nr_segments; + dst->u.rw.handle = src->u.rw.handle; + dst->u.rw.id = src->u.rw.id; dst->u.rw.sector_number = src->u.rw.sector_number; barrier(); - if (n > dst->nr_segments) - n = dst->nr_segments; + if (n > dst->u.rw.nr_segments) + n = dst->u.rw.nr_segments; for (i = 0; i < n; i++) dst->u.rw.seg[i] = src->u.rw.seg[i]; break; case BLKIF_OP_DISCARD: + dst->u.discard.flag = src->u.discard.flag; dst->u.discard.sector_number = src->u.discard.sector_number; dst->u.discard.nr_sectors = src->u.discard.nr_sectors; break; diff --git a/drivers/block/xen-blkback/xenbus.c b/drivers/block/xen-blkback/xenbus.c index 37c794d31264..24a2fb57e5d0 100644 --- a/drivers/block/xen-blkback/xenbus.c +++ b/drivers/block/xen-blkback/xenbus.c @@ -338,6 +338,9 @@ static int xen_vbd_create(struct xen_blkif *blkif, blkif_vdev_t handle, if (q && q->flush_flags) vbd->flush_support = true; + if (q && blk_queue_secdiscard(q)) + vbd->discard_secure = true; + DPRINTK("Successful creation of handle=%04x (dom=%u)\n", handle, blkif->domid); return 0; @@ -420,6 +423,15 @@ int xen_blkbk_discard(struct xenbus_transaction xbt, struct backend_info *be) state = 1; blkif->blk_backend_type = BLKIF_BACKEND_PHY; } + /* Optional. */ + err = xenbus_printf(xbt, dev->nodename, + "discard-secure", "%d", + blkif->vbd.discard_secure); + if (err) { + xenbus_dev_fatal(dev, err, + "writting discard-secure"); + goto kfree; + } } } else { err = PTR_ERR(type); diff --git a/drivers/block/xen-blkfront.c b/drivers/block/xen-blkfront.c index 19b6005a323e..98cbeba8cd53 100644 --- a/drivers/block/xen-blkfront.c +++ b/drivers/block/xen-blkfront.c @@ -98,7 +98,8 @@ struct blkfront_info unsigned long shadow_free; unsigned int feature_flush; unsigned int flush_op; - unsigned int feature_discard; + unsigned int feature_discard:1; + unsigned int feature_secdiscard:1; unsigned int discard_granularity; unsigned int discard_alignment; int is_ready; @@ -135,15 +136,15 @@ static int get_id_from_freelist(struct blkfront_info *info) { unsigned long free = info->shadow_free; BUG_ON(free >= BLK_RING_SIZE); - info->shadow_free = info->shadow[free].req.id; - info->shadow[free].req.id = 0x0fffffee; /* debug */ + info->shadow_free = info->shadow[free].req.u.rw.id; + info->shadow[free].req.u.rw.id = 0x0fffffee; /* debug */ return free; } static void add_id_to_freelist(struct blkfront_info *info, unsigned long id) { - info->shadow[id].req.id = info->shadow_free; + info->shadow[id].req.u.rw.id = info->shadow_free; info->shadow[id].request = NULL; info->shadow_free = id; } @@ -156,7 +157,7 @@ static int xlbd_reserve_minors(unsigned int minor, unsigned int nr) if (end > nr_minors) { unsigned long *bitmap, *old; - bitmap = kzalloc(BITS_TO_LONGS(end) * sizeof(*bitmap), + bitmap = kcalloc(BITS_TO_LONGS(end), sizeof(*bitmap), GFP_KERNEL); if (bitmap == NULL) return -ENOMEM; @@ -287,9 +288,9 @@ static int blkif_queue_request(struct request *req) id = get_id_from_freelist(info); info->shadow[id].request = req; - ring_req->id = id; + ring_req->u.rw.id = id; ring_req->u.rw.sector_number = (blkif_sector_t)blk_rq_pos(req); - ring_req->handle = info->handle; + ring_req->u.rw.handle = info->handle; ring_req->operation = rq_data_dir(req) ? BLKIF_OP_WRITE : BLKIF_OP_READ; @@ -305,16 +306,21 @@ static int blkif_queue_request(struct request *req) ring_req->operation = info->flush_op; } - if (unlikely(req->cmd_flags & REQ_DISCARD)) { + if (unlikely(req->cmd_flags & (REQ_DISCARD | REQ_SECURE))) { /* id, sector_number and handle are set above. */ ring_req->operation = BLKIF_OP_DISCARD; - ring_req->nr_segments = 0; ring_req->u.discard.nr_sectors = blk_rq_sectors(req); + if ((req->cmd_flags & REQ_SECURE) && info->feature_secdiscard) + ring_req->u.discard.flag = BLKIF_DISCARD_SECURE; + else + ring_req->u.discard.flag = 0; } else { - ring_req->nr_segments = blk_rq_map_sg(req->q, req, info->sg); - BUG_ON(ring_req->nr_segments > BLKIF_MAX_SEGMENTS_PER_REQUEST); + ring_req->u.rw.nr_segments = blk_rq_map_sg(req->q, req, + info->sg); + BUG_ON(ring_req->u.rw.nr_segments > + BLKIF_MAX_SEGMENTS_PER_REQUEST); - for_each_sg(info->sg, sg, ring_req->nr_segments, i) { + for_each_sg(info->sg, sg, ring_req->u.rw.nr_segments, i) { buffer_mfn = pfn_to_mfn(page_to_pfn(sg_page(sg))); fsect = sg->offset >> 9; lsect = fsect + (sg->length >> 9) - 1; @@ -424,6 +430,8 @@ static int xlvbd_init_blk_queue(struct gendisk *gd, u16 sector_size) blk_queue_max_discard_sectors(rq, get_capacity(gd)); rq->limits.discard_granularity = info->discard_granularity; rq->limits.discard_alignment = info->discard_alignment; + if (info->feature_secdiscard) + queue_flag_set_unlocked(QUEUE_FLAG_SECDISCARD, rq); } /* Hard sector size and max sectors impersonate the equiv. hardware. */ @@ -705,7 +713,9 @@ static void blkif_free(struct blkfront_info *info, int suspend) static void blkif_completion(struct blk_shadow *s) { int i; - for (i = 0; i < s->req.nr_segments; i++) + /* Do not let BLKIF_OP_DISCARD as nr_segment is in the same place + * flag. */ + for (i = 0; i < s->req.u.rw.nr_segments; i++) gnttab_end_foreign_access(s->req.u.rw.seg[i].gref, 0, 0UL); } @@ -736,7 +746,8 @@ static irqreturn_t blkif_interrupt(int irq, void *dev_id) id = bret->id; req = info->shadow[id].request; - blkif_completion(&info->shadow[id]); + if (bret->operation != BLKIF_OP_DISCARD) + blkif_completion(&info->shadow[id]); add_id_to_freelist(info, id); @@ -749,7 +760,9 @@ static irqreturn_t blkif_interrupt(int irq, void *dev_id) info->gd->disk_name); error = -EOPNOTSUPP; info->feature_discard = 0; + info->feature_secdiscard = 0; queue_flag_clear(QUEUE_FLAG_DISCARD, rq); + queue_flag_clear(QUEUE_FLAG_SECDISCARD, rq); } __blk_end_request_all(req, error); break; @@ -763,7 +776,7 @@ static irqreturn_t blkif_interrupt(int irq, void *dev_id) error = -EOPNOTSUPP; } if (unlikely(bret->status == BLKIF_RSP_ERROR && - info->shadow[id].req.nr_segments == 0)) { + info->shadow[id].req.u.rw.nr_segments == 0)) { printk(KERN_WARNING "blkfront: %s: empty write %s op failed\n", info->flush_op == BLKIF_OP_WRITE_BARRIER ? "barrier" : "flush disk cache", @@ -984,8 +997,8 @@ static int blkfront_probe(struct xenbus_device *dev, INIT_WORK(&info->work, blkif_restart_queue); for (i = 0; i < BLK_RING_SIZE; i++) - info->shadow[i].req.id = i+1; - info->shadow[BLK_RING_SIZE-1].req.id = 0x0fffffff; + info->shadow[i].req.u.rw.id = i+1; + info->shadow[BLK_RING_SIZE-1].req.u.rw.id = 0x0fffffff; /* Front end dir is a number, which is used as the id. */ info->handle = simple_strtoul(strrchr(dev->nodename, '/')+1, NULL, 0); @@ -1019,9 +1032,9 @@ static int blkif_recover(struct blkfront_info *info) /* Stage 2: Set up free list. */ memset(&info->shadow, 0, sizeof(info->shadow)); for (i = 0; i < BLK_RING_SIZE; i++) - info->shadow[i].req.id = i+1; + info->shadow[i].req.u.rw.id = i+1; info->shadow_free = info->ring.req_prod_pvt; - info->shadow[BLK_RING_SIZE-1].req.id = 0x0fffffff; + info->shadow[BLK_RING_SIZE-1].req.u.rw.id = 0x0fffffff; /* Stage 3: Find pending requests and requeue them. */ for (i = 0; i < BLK_RING_SIZE; i++) { @@ -1034,17 +1047,19 @@ static int blkif_recover(struct blkfront_info *info) *req = copy[i].req; /* We get a new request id, and must reset the shadow state. */ - req->id = get_id_from_freelist(info); - memcpy(&info->shadow[req->id], ©[i], sizeof(copy[i])); + req->u.rw.id = get_id_from_freelist(info); + memcpy(&info->shadow[req->u.rw.id], ©[i], sizeof(copy[i])); + if (req->operation != BLKIF_OP_DISCARD) { /* Rewrite any grant references invalidated by susp/resume. */ - for (j = 0; j < req->nr_segments; j++) - gnttab_grant_foreign_access_ref( - req->u.rw.seg[j].gref, - info->xbdev->otherend_id, - pfn_to_mfn(info->shadow[req->id].frame[j]), - rq_data_dir(info->shadow[req->id].request)); - info->shadow[req->id].req = *req; + for (j = 0; j < req->u.rw.nr_segments; j++) + gnttab_grant_foreign_access_ref( + req->u.rw.seg[j].gref, + info->xbdev->otherend_id, + pfn_to_mfn(info->shadow[req->u.rw.id].frame[j]), + rq_data_dir(info->shadow[req->u.rw.id].request)); + } + info->shadow[req->u.rw.id].req = *req; info->ring.req_prod_pvt++; } @@ -1135,11 +1150,13 @@ static void blkfront_setup_discard(struct blkfront_info *info) char *type; unsigned int discard_granularity; unsigned int discard_alignment; + unsigned int discard_secure; type = xenbus_read(XBT_NIL, info->xbdev->otherend, "type", NULL); if (IS_ERR(type)) return; + info->feature_secdiscard = 0; if (strncmp(type, "phy", 3) == 0) { err = xenbus_gather(XBT_NIL, info->xbdev->otherend, "discard-granularity", "%u", &discard_granularity, @@ -1150,6 +1167,12 @@ static void blkfront_setup_discard(struct blkfront_info *info) info->discard_granularity = discard_granularity; info->discard_alignment = discard_alignment; } + err = xenbus_gather(XBT_NIL, info->xbdev->otherend, + "discard-secure", "%d", &discard_secure, + NULL); + if (!err) + info->feature_secdiscard = discard_secure; + } else if (strncmp(type, "file", 4) == 0) info->feature_discard = 1; |