diff options
| author | Linus Torvalds <torvalds@linux-foundation.org> | 2013-05-09 16:35:00 -0700 |
|---|---|---|
| committer | Linus Torvalds <torvalds@linux-foundation.org> | 2013-05-09 16:35:00 -0700 |
| commit | 2d4fe27850420606155fb1f7d18ab2b40153e67b (patch) | |
| tree | 56b0d465e1189babf4ac668a5b048a747bfb9682 | |
| parent | 2e99f3a12b20ab3afad0e042cc0bdd0ee855dca0 (diff) | |
| parent | 94f370cab6e5ac514b658c6b2b3aa308cefc5c7a (diff) | |
| download | linux-2d4fe27850420606155fb1f7d18ab2b40153e67b.tar.gz | |
Merge git://git.infradead.org/users/willy/linux-nvme
Pull NVMe driver update from Matthew Wilcox: "Lots of exciting new features in the NVM Express driver this time, including support for emulating SCSI commands, discard support and the ability to submit per-sector metadata with I/Os. It's still mostly bugfixes though!" * git://git.infradead.org/users/willy/linux-nvme: (27 commits) NVMe: Use user defined admin ioctl timeout NVMe: Simplify Firmware Activate code slightly NVMe: Only clear the enable bit when disabling controller NVMe: Wait for device to acknowledge shutdown NVMe: Schedule timeout for sync commands NVMe: Meta-data support in NVME_IOCTL_SUBMIT_IO NVMe: Device specific stripe size handling NVMe: Split non-mergeable bio requests NVMe: Remove dead code in nvme_dev_add NVMe: Check for NULL memory in nvme_dev_add NVMe: Fix error clean-up on nvme_alloc_queue NVMe: Free admin queue on request_irq error NVMe: Add scsi unmap to SG_IO NVMe: queue usage fixes in nvme-scsi NVMe: Set TASK_INTERRUPTIBLE before processing queues NVMe: Add a character device for each nvme device NVMe: Fix endian-related problems in user I/O submission path NVMe: Fix I/O cancellation status on big-endian machines NVMe: Fix sparse warnings in scsi emulation NVMe: Don't fail initialisation unnecessarily ...
| -rw-r--r-- | drivers/block/Makefile | 1 | ||||
| -rw-r--r-- | drivers/block/nvme-core.c (renamed from drivers/block/nvme.c) | 594 | ||||
| -rw-r--r-- | drivers/block/nvme-scsi.c | 3053 | ||||
| -rw-r--r-- | include/linux/nvme.h | 158 |
4 files changed, 3640 insertions, 166 deletions
diff --git a/drivers/block/Makefile b/drivers/block/Makefile index a3b40232c6ab..ca07399a8d99 100644 --- a/drivers/block/Makefile +++ b/drivers/block/Makefile @@ -42,4 +42,5 @@ obj-$(CONFIG_BLK_DEV_PCIESSD_MTIP32XX) += mtip32xx/ obj-$(CONFIG_BLK_DEV_RSXX) += rsxx/ +nvme-y := nvme-core.o nvme-scsi.o swim_mod-y := swim.o swim_asm.o diff --git a/drivers/block/nvme.c b/drivers/block/nvme-core.c index 9dcefe40380b..8efdfaa44a59 100644 --- a/drivers/block/nvme.c +++ b/drivers/block/nvme-core.c @@ -39,14 +39,13 @@ #include <linux/sched.h> #include <linux/slab.h> #include <linux/types.h> - +#include <scsi/sg.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; @@ -60,43 +59,6 @@ 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]; - u32 max_hw_sectors; -}; - -/* - * 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). */ @@ -131,6 +93,7 @@ static inline void _nvme_check_size(void) 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_format_cmd) != 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); @@ -261,12 +224,12 @@ static void *cancel_cmdid(struct nvme_queue *nvmeq, int cmdid, return ctx; } -static struct nvme_queue *get_nvmeq(struct nvme_dev *dev) +struct nvme_queue *get_nvmeq(struct nvme_dev *dev) { return dev->queues[get_cpu() + 1]; } -static void put_nvmeq(struct nvme_queue *nvmeq) +void put_nvmeq(struct nvme_queue *nvmeq) { put_cpu(); } @@ -294,22 +257,6 @@ static int nvme_submit_cmd(struct nvme_queue *nvmeq, struct nvme_command *cmd) 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; @@ -343,7 +290,7 @@ nvme_alloc_iod(unsigned nseg, unsigned nbytes, gfp_t gfp) return iod; } -static void nvme_free_iod(struct nvme_dev *dev, struct nvme_iod *iod) +void nvme_free_iod(struct nvme_dev *dev, struct nvme_iod *iod) { const int last_prp = PAGE_SIZE / 8 - 1; int i; @@ -361,16 +308,6 @@ static void nvme_free_iod(struct nvme_dev *dev, struct nvme_iod *iod) 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) { @@ -382,19 +319,15 @@ static void bio_completion(struct nvme_dev *dev, void *ctx, 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) { + if (status) bio_endio(bio, -EIO); - } else if (bio->bi_vcnt > bio->bi_idx) { - requeue_bio(dev, bio); - } else { + 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) +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; @@ -473,43 +406,193 @@ static int nvme_setup_prps(struct nvme_dev *dev, return total_len; } +struct nvme_bio_pair { + struct bio b1, b2, *parent; + struct bio_vec *bv1, *bv2; + int err; + atomic_t cnt; +}; + +static void nvme_bio_pair_endio(struct bio *bio, int err) +{ + struct nvme_bio_pair *bp = bio->bi_private; + + if (err) + bp->err = err; + + if (atomic_dec_and_test(&bp->cnt)) { + bio_endio(bp->parent, bp->err); + if (bp->bv1) + kfree(bp->bv1); + if (bp->bv2) + kfree(bp->bv2); + kfree(bp); + } +} + +static struct nvme_bio_pair *nvme_bio_split(struct bio *bio, int idx, + int len, int offset) +{ + struct nvme_bio_pair *bp; + + BUG_ON(len > bio->bi_size); + BUG_ON(idx > bio->bi_vcnt); + + bp = kmalloc(sizeof(*bp), GFP_ATOMIC); + if (!bp) + return NULL; + bp->err = 0; + + bp->b1 = *bio; + bp->b2 = *bio; + + bp->b1.bi_size = len; + bp->b2.bi_size -= len; + bp->b1.bi_vcnt = idx; + bp->b2.bi_idx = idx; + bp->b2.bi_sector += len >> 9; + + if (offset) { + bp->bv1 = kmalloc(bio->bi_max_vecs * sizeof(struct bio_vec), + GFP_ATOMIC); + if (!bp->bv1) + goto split_fail_1; + + bp->bv2 = kmalloc(bio->bi_max_vecs * sizeof(struct bio_vec), + GFP_ATOMIC); + if (!bp->bv2) + goto split_fail_2; + + memcpy(bp->bv1, bio->bi_io_vec, + bio->bi_max_vecs * sizeof(struct bio_vec)); + memcpy(bp->bv2, bio->bi_io_vec, + bio->bi_max_vecs * sizeof(struct bio_vec)); + + bp->b1.bi_io_vec = bp->bv1; + bp->b2.bi_io_vec = bp->bv2; + bp->b2.bi_io_vec[idx].bv_offset += offset; + bp->b2.bi_io_vec[idx].bv_len -= offset; + bp->b1.bi_io_vec[idx].bv_len = offset; + bp->b1.bi_vcnt++; + } else + bp->bv1 = bp->bv2 = NULL; + + bp->b1.bi_private = bp; + bp->b2.bi_private = bp; + + bp->b1.bi_end_io = nvme_bio_pair_endio; + bp->b2.bi_end_io = nvme_bio_pair_endio; + + bp->parent = bio; + atomic_set(&bp->cnt, 2); + + return bp; + + split_fail_2: + kfree(bp->bv1); + split_fail_1: + kfree(bp); + return NULL; +} + +static int nvme_split_and_submit(struct bio *bio, struct nvme_queue *nvmeq, + int idx, int len, int offset) +{ + struct nvme_bio_pair *bp = nvme_bio_split(bio, idx, len, offset); + if (!bp) + return -ENOMEM; + + if (bio_list_empty(&nvmeq->sq_cong)) + add_wait_queue(&nvmeq->sq_full, &nvmeq->sq_cong_wait); + bio_list_add(&nvmeq->sq_cong, &bp->b1); + bio_list_add(&nvmeq->sq_cong, &bp->b2); + + return 0; +} + /* 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, +static int nvme_map_bio(struct nvme_queue *nvmeq, 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; + int i, length = 0, nsegs = 0, split_len = bio->bi_size; + + if (nvmeq->dev->stripe_size) + split_len = nvmeq->dev->stripe_size - + ((bio->bi_sector << 9) & (nvmeq->dev->stripe_size - 1)); 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; + return nvme_split_and_submit(bio, nvmeq, i, + length, 0); + sg = sg ? sg + 1 : iod->sg; sg_set_page(sg, bvec->bv_page, bvec->bv_len, bvec->bv_offset); nsegs++; } + + if (split_len - length < bvec->bv_len) + return nvme_split_and_submit(bio, nvmeq, i, split_len, + split_len - length); 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; + if (dma_map_sg(nvmeq->q_dmadev, iod->sg, iod->nents, dma_dir) == 0) return -ENOMEM; - } + + BUG_ON(length != bio->bi_size); return length; } +/* + * We reuse the small pool to allocate the 16-byte range here as it is not + * worth having a special pool for these or additional cases to handle freeing + * the iod. + */ +static int nvme_submit_discard(struct nvme_queue *nvmeq, struct nvme_ns *ns, + struct bio *bio, struct nvme_iod *iod, int cmdid) +{ + struct nvme_dsm_range *range; + struct nvme_command *cmnd = &nvmeq->sq_cmds[nvmeq->sq_tail]; + + range = dma_pool_alloc(nvmeq->dev->prp_small_pool, GFP_ATOMIC, + &iod->first_dma); + if (!range) + return -ENOMEM; + + iod_list(iod)[0] = (__le64 *)range; + iod->npages = 0; + + range->cattr = cpu_to_le32(0); + range->nlb = cpu_to_le32(bio->bi_size >> ns->lba_shift); + range->slba = cpu_to_le64(nvme_block_nr(ns, bio->bi_sector)); + + memset(cmnd, 0, sizeof(*cmnd)); + cmnd->dsm.opcode = nvme_cmd_dsm; + cmnd->dsm.command_id = cmdid; + cmnd->dsm.nsid = cpu_to_le32(ns->ns_id); + cmnd->dsm.prp1 = cpu_to_le64(iod->first_dma); + cmnd->dsm.nr = 0; + cmnd->dsm.attributes = cpu_to_le32(NVME_DSMGMT_AD); + + 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(struct nvme_queue *nvmeq, struct nvme_ns *ns, int cmdid) { @@ -527,7 +610,7 @@ static int nvme_submit_flush(struct nvme_queue *nvmeq, struct nvme_ns *ns, return 0; } -static int nvme_submit_flush_data(struct nvme_queue *nvmeq, struct nvme_ns *ns) +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); @@ -567,6 +650,12 @@ static int nvme_submit_bio_queue(struct nvme_queue *nvmeq, struct nvme_ns *ns, if (unlikely(cmdid < 0)) goto free_iod; + if (bio->bi_rw & REQ_DISCARD) { + result = nvme_submit_discard(nvmeq, ns, bio, iod, cmdid); + if (result) + goto free_cmdid; + return result; + } if ((bio->bi_rw & REQ_FLUSH) && !psegs) return nvme_submit_flush(nvmeq, ns, cmdid); @@ -591,8 +680,8 @@ static int nvme_submit_bio_queue(struct nvme_queue *nvmeq, struct nvme_ns *ns, dma_dir = DMA_FROM_DEVICE; } - result = nvme_map_bio(nvmeq->q_dmadev, iod, bio, dma_dir, psegs); - if (result < 0) + result = nvme_map_bio(nvmeq, iod, bio, dma_dir, psegs); + if (result <= 0) goto free_cmdid; length = result; @@ -600,13 +689,11 @@ static int nvme_submit_bio_queue(struct nvme_queue *nvmeq, struct nvme_ns *ns, 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.slba = cpu_to_le64(nvme_block_nr(ns, bio->bi_sector)); 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); @@ -724,8 +811,8 @@ static void sync_completion(struct nvme_dev *dev, void *ctx, * 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 nvme_submit_sync_cmd(struct nvme_queue *nvmeq, struct nvme_command *cmd, + u32 *result, unsigned timeout) { int cmdid; struct sync_cmd_info cmdinfo; @@ -741,7 +828,7 @@ static int nvme_submit_sync_cmd(struct nvme_queue *nvmeq, set_current_state(TASK_KILLABLE); nvme_submit_cmd(nvmeq, cmd); - schedule(); + schedule_timeout(timeout); if (cmdinfo.status == -EINTR) { nvme_abort_command(nvmeq, cmdid); @@ -754,7 +841,7 @@ static int nvme_submit_sync_cmd(struct nvme_queue *nvmeq, return cmdinfo.status; } -static int nvme_submit_admin_cmd(struct nvme_dev *dev, struct nvme_command *cmd, +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); @@ -827,7 +914,7 @@ 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, +int nvme_identify(struct nvme_dev *dev, unsigned nsid, unsigned cns, dma_addr_t dma_addr) { struct nvme_command c; @@ -841,7 +928,7 @@ static int nvme_identify(struct nvme_dev *dev, unsigned nsid, unsigned cns, return nvme_submit_admin_cmd(dev, &c, NULL); } -static int nvme_get_features(struct nvme_dev *dev, unsigned fid, unsigned nsid, +int nvme_get_features(struct nvme_dev *dev, unsigned fid, unsigned nsid, dma_addr_t dma_addr, u32 *result) { struct nvme_command c; @@ -855,8 +942,8 @@ static int nvme_get_features(struct nvme_dev *dev, unsigned fid, unsigned nsid, return nvme_submit_admin_cmd(dev, &c, result); } -static int nvme_set_features(struct nvme_dev *dev, unsigned fid, - unsigned dword11, dma_addr_t dma_addr, u32 *result) +int nvme_set_features(struct nvme_dev *dev, unsigned fid, unsigned dword11, + dma_addr_t dma_addr, u32 *result) { struct nvme_command c; @@ -885,7 +972,7 @@ static void nvme_cancel_ios(struct nvme_queue *nvmeq, bool timeout) void *ctx; nvme_completion_fn fn; static struct nvme_completion cqe = { - .status = cpu_to_le16(NVME_SC_ABORT_REQ) << 1, + .status = cpu_to_le16(NVME_SC_ABORT_REQ << 1), }; if (timeout && !time_after(now, info[cmdid].timeout)) @@ -966,7 +1053,7 @@ static struct nvme_queue *nvme_alloc_queue(struct nvme_dev *dev, int qid, return nvmeq; free_cqdma: - dma_free_coherent(dmadev, CQ_SIZE(nvmeq->q_depth), (void *)nvmeq->cqes, + dma_free_coherent(dmadev, CQ_SIZE(depth), (void *)nvmeq->cqes, nvmeq->cq_dma_addr); free_nvmeq: kfree(nvmeq); @@ -1021,15 +1108,60 @@ static struct nvme_queue *nvme_create_queue(struct nvme_dev *dev, int qid, return ERR_PTR(result); } +static int nvme_wait_ready(struct nvme_dev *dev, u64 cap, bool enabled) +{ + unsigned long timeout; + u32 bit = enabled ? NVME_CSTS_RDY : 0; + + timeout = ((NVME_CAP_TIMEOUT(cap) + 1) * HZ / 2) + jiffies; + + while ((readl(&dev->bar->csts) & NVME_CSTS_RDY) != bit) { + 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; + } + } + + return 0; +} + +/* + * If the device has been passed off to us in an enabled state, just clear + * the enabled bit. The spec says we should set the 'shutdown notification + * bits', but doing so may cause the device to complete commands to the + * admin queue ... and we don't know what memory that might be pointing at! + */ +static int nvme_disable_ctrl(struct nvme_dev *dev, u64 cap) +{ + u32 cc = readl(&dev->bar->cc); + + if (cc & NVME_CC_ENABLE) + writel(cc & ~NVME_CC_ENABLE, &dev->bar->cc); + return nvme_wait_ready(dev, cap, false); +} + +static int nvme_enable_ctrl(struct nvme_dev *dev, u64 cap) +{ + return nvme_wait_ready(dev, cap, true); +} + static int nvme_configure_admin_queue(struct nvme_dev *dev) { - int result = 0; + int result; u32 aqa; - u64 cap; - unsigned long timeout; + u64 cap = readq(&dev->bar->cap); struct nvme_queue *nvmeq; dev->dbs = ((void __iomem *)dev->bar) + 4096; + dev->db_stride = NVME_CAP_STRIDE(cap); + + result = nvme_disable_ctrl(dev, cap); + if (result < 0) + return result; nvmeq = nvme_alloc_queue(dev, 0, 64, 0); if (!nvmeq) @@ -1043,38 +1175,28 @@ static int nvme_configure_admin_queue(struct nvme_dev *dev) 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 (!result && !(readl(&dev->bar->csts) & NVME_CSTS_RDY)) { - msleep(100); - if (fatal_signal_pending(current)) - result = -EINTR; - if (time_after(jiffies, timeout)) { - dev_err(&dev->pci_dev->dev, - "Device not ready; aborting initialisation\n"); - result = -ENODEV; - } - } - - if (result) { - nvme_free_queue_mem(nvmeq); - return result; - } + result = nvme_enable_ctrl(dev, cap); + if (result) + goto free_q; result = queue_request_irq(dev, nvmeq, "nvme admin"); + if (result) + goto free_q; + dev->queues[0] = nvmeq; return result; + + free_q: + nvme_free_queue_mem(nvmeq); + return result; } -static struct nvme_iod *nvme_map_user_pages(struct nvme_dev *dev, int write, +struct nvme_iod *nvme_map_user_pages(struct nvme_dev *dev, int write, unsigned long addr, unsigned length) { int i, err, count, nents, offset; @@ -1130,7 +1252,7 @@ static struct nvme_iod *nvme_map_user_pages(struct nvme_dev *dev, int write, return ERR_PTR(err); } -static void nvme_unmap_user_pages(struct nvme_dev *dev, int write, +void nvme_unmap_user_pages(struct nvme_dev *dev, int write, struct nvme_iod *iod) { int i; @@ -1148,13 +1270,19 @@ static int nvme_submit_io(struct nvme_ns *ns, struct nvme_user_io __user *uio) struct nvme_queue *nvmeq; struct nvme_user_io io; struct nvme_command c; - unsigned length; - int status; - struct nvme_iod *iod; + unsigned length, meta_len; + int status, i; + struct nvme_iod *iod, *meta_iod = NULL; + dma_addr_t meta_dma_addr; + void *meta, *uninitialized_var(meta_mem); if (copy_from_user(&io, uio, sizeof(io))) return -EFAULT; length = (io.nblocks + 1) << ns->lba_shift; + meta_len = (io.nblocks + 1) * ns->ms; + + if (meta_len && ((io.metadata & 3) || !io.metadata)) + return -EINVAL; switch (io.opcode) { case nvme_cmd_write: @@ -1176,11 +1304,42 @@ static int nvme_submit_io(struct nvme_ns *ns, struct nvme_user_io __user *uio) 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 */ + c.rw.dsmgmt = cpu_to_le32(io.dsmgmt); + c.rw.reftag = cpu_to_le32(io.reftag); + c.rw.apptag = cpu_to_le16(io.apptag); + c.rw.appmask = cpu_to_le16(io.appmask); + + if (meta_len) { + meta_iod = nvme_map_user_pages(dev, io.opcode & 1, io.metadata, meta_len); + if (IS_ERR(meta_iod)) { + status = PTR_ERR(meta_iod); + meta_iod = NULL; + goto unmap; + } + + meta_mem = dma_alloc_coherent(&dev->pci_dev->dev, meta_len, + &meta_dma_addr, GFP_KERNEL); + if (!meta_mem) { + status = -ENOMEM; + goto unmap; + } + + if (io.opcode & 1) { + int meta_offset = 0; + + for (i = 0; i < meta_iod->nents; i++) { + meta = kmap_atomic(sg_page(&meta_iod->sg[i])) + + meta_iod->sg[i].offset; + memcpy(meta_mem + meta_offset, meta, + meta_iod->sg[i].length); + kunmap_atomic(meta); + meta_offset += meta_iod->sg[i].length; + } + } + + c.rw.metadata = cpu_to_le64(meta_dma_addr); + } + length = nvme_setup_prps(dev, &c.common, iod, length, GFP_KERNEL); nvmeq = get_nvmeq(dev); @@ -1196,8 +1355,33 @@ static int nvme_submit_io(struct nvme_ns *ns, struct nvme_user_io __user *uio) else status = nvme_submit_sync_cmd(nvmeq, &c, NULL, NVME_IO_TIMEOUT); + if (meta_len) { + if (status == NVME_SC_SUCCESS && !(io.opcode & 1)) { + int meta_offset = 0; + + for (i = 0; i < meta_iod->nents; i++) { + meta = kmap_atomic(sg_page(&meta_iod->sg[i])) + + meta_iod->sg[i].offset; + memcpy(meta, meta_mem + meta_offset, + meta_iod->sg[i].length); + kunmap_atomic(meta); + meta_offset += meta_iod->sg[i].length; + } + } + + dma_free_coherent(&dev->pci_dev->dev, meta_len, meta_mem, + meta_dma_addr); + } + + unmap: nvme_unmap_user_pages(dev, io.opcode & 1, iod); nvme_free_iod(dev, iod); + + if (meta_iod) { + nvme_unmap_user_pages(dev, io.opcode & 1, meta_iod); + nvme_free_iod(dev, meta_iod); + } + return status; } @@ -1208,6 +1392,7 @@ static int nvme_user_admin_cmd(struct nvme_dev *dev, struct nvme_command c; int status, length; struct nvme_iod *uninitialized_var(iod); + unsigned timeout; if (!capable(CAP_SYS_ADMIN)) return -EACCES; @@ -1237,10 +1422,13 @@ static int nvme_user_admin_cmd(struct nvme_dev *dev, GFP_KERNEL); } + timeout = cmd.timeout_ms ? msecs_to_jiffies(cmd.timeout_ms) : + ADMIN_TIMEOUT; if (length != cmd.data_len) status = -ENOMEM; else - status = nvme_submit_admin_cmd(dev, &c, &cmd.result); + status = nvme_submit_sync_cmd(dev->queues[0], &c, &cmd.result, + timeout); if (cmd.data_len) { nvme_unmap_user_pages(dev, cmd.opcode & 1, iod); @@ -1266,6 +1454,10 @@ static int nvme_ioctl(struct block_device *bdev, fmode_t mode, unsigned int cmd, return nvme_user_admin_cmd(ns->dev, (void __user *)arg); case NVME_IOCTL_SUBMIT_IO: return nvme_submit_io(ns, (void __user *)arg); + case SG_GET_VERSION_NUM: + return nvme_sg_get_version_num((void __user *)arg); + case SG_IO: + return nvme_sg_io(ns, (void __user *)arg); default: return -ENOTTY; } @@ -1282,13 +1474,17 @@ 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 (bio_list_empty(&nvmeq->sq_cong)) + remove_wait_queue(&nvmeq->sq_full, + &nvmeq->sq_cong_wait); if (nvme_submit_bio_queue(nvmeq, ns, bio)) { + if (bio_list_empty(&nvmeq->sq_cong)) + add_wait_queue(&nvmeq->sq_full, + &nvmeq->sq_cong_wait); 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); } } @@ -1297,7 +1493,7 @@ static int nvme_kthread(void *data) struct nvme_dev *dev; while (!kthread_should_stop()) { - __set_current_state(TASK_RUNNING); + set_current_state(TASK_INTERRUPTIBLE); spin_lock(&dev_list_lock); list_for_each_entry(dev, &dev_list, node) { int i; @@ -1314,8 +1510,7 @@ static int nvme_kthread(void *data) } } spin_unlock(&dev_list_lock); - set_current_state(TASK_INTERRUPTIBLE); - schedule_timeout(HZ); + schedule_timeout(round_jiffies_relative(HZ)); } return 0; } @@ -1347,6 +1542,16 @@ static void nvme_put_ns_idx(int index) spin_unlock(&dev_list_lock); } +static void nvme_config_discard(struct nvme_ns *ns) +{ + u32 logical_block_size = queue_logical_block_size(ns->queue); + ns->queue->limits.discard_zeroes_data = 0; + ns->queue->limits.discard_alignment = logical_block_size; + ns->queue->limits.discard_granularity = logical_block_size; + ns->queue->limits.max_discard_sectors = 0xffffffff; + queue_flag_set_unlocked(QUEUE_FLAG_DISCARD, ns->queue); +} + static struct nvme_ns *nvme_alloc_ns(struct nvme_dev *dev, int nsid, struct nvme_id_ns *id, struct nvme_lba_range_type *rt) { @@ -1366,7 +1571,6 @@ static struct nvme_ns *nvme_alloc_ns(struct nvme_dev *dev, int nsid, 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; @@ -1378,6 +1582,7 @@ static struct nvme_ns *nvme_alloc_ns(struct nvme_dev *dev, int nsid, ns->disk = disk; lbaf = id->flbas & 0xf; ns->lba_shift = id->lbaf[lbaf].ds; + ns->ms = le16_to_cpu(id->lbaf[lbaf].ms); blk_queue_logical_block_size(ns->queue, 1 << ns->lba_shift); if (dev->max_hw_sectors) blk_queue_max_hw_sectors(ns->queue, dev->max_hw_sectors); @@ -1392,6 +1597,9 @@ static struct nvme_ns *nvme_alloc_ns(struct nvme_dev *dev, int nsid, sprintf(disk->disk_name, "nvme%dn%d", dev->instance, nsid); set_capacity(disk, le64_to_cpup(&id->nsze) << (ns->lba_shift - 9)); + if (dev->oncs & NVME_CTRL_ONCS_DSM) + nvme_config_discard(ns); + return ns; out_free_queue: @@ -1496,14 +1704,21 @@ static void nvme_free_queues(struct nvme_dev *dev) nvme_free_queue(dev, i); } +/* + * Return: error value if an error occurred setting up the queues or calling + * Identify Device. 0 if these succeeded, even if adding some of the + * namespaces failed. At the moment, these failures are silent. TBD which + * failures should be reported. + */ static int nvme_dev_add(struct nvme_dev *dev) { int res, nn, i; - struct nvme_ns *ns, *next; + struct nvme_ns *ns; struct nvme_id_ctrl *ctrl; struct nvme_id_ns *id_ns; void *mem; dma_addr_t dma_addr; + int shift = NVME_CAP_MPSMIN(readq(&dev->bar->cap)) + 12; res = nvme_setup_io_queues(dev); if (res) @@ -1511,22 +1726,26 @@ static int nvme_dev_add(struct nvme_dev *dev) mem = dma_alloc_coherent(&dev->pci_dev->dev, 8192, &dma_addr, GFP_KERNEL); + if (!mem) + return -ENOMEM; res = nvme_identify(dev, 0, 1, dma_addr); if (res) { res = -EIO; - goto out_free; + goto out; } ctrl = mem; nn = le32_to_cpup(&ctrl->nn); + dev->oncs = le16_to_cpup(&ctrl->oncs); 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)); - if (ctrl->mdts) { - int shift = NVME_CAP_MPSMIN(readq(&dev->bar->cap)) + 12; + if (ctrl->mdts) dev->max_hw_sectors = 1 << (ctrl->mdts + shift - 9); - } + if ((dev->pci_dev->vendor == PCI_VENDOR_ID_INTEL) && + (dev->pci_dev->device == 0x0953) && ctrl->vs[3]) + dev->stripe_size = 1 << (ctrl->vs[3] + shift); id_ns = mem; for (i = 1; i <= nn; i++) { @@ -1548,14 +1767,7 @@ static int nvme_dev_add(struct nvme_dev *dev) } 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); - } + res = 0; out: dma_free_coherent(&dev->pci_dev->dev, 8192, mem, dma_addr); @@ -1634,6 +1846,56 @@ static void nvme_release_instance(struct nvme_dev *dev) spin_unlock(&dev_list_lock); } +static void nvme_free_dev(struct kref *kref) +{ + struct nvme_dev *dev = container_of(kref, struct nvme_dev, kref); + nvme_dev_remove(dev); + pci_disable_msix(dev->pci_dev); + iounmap(dev->bar); + nvme_release_instance(dev); + nvme_release_prp_pools(dev); + pci_disable_device(dev->pci_dev); + pci_release_regions(dev->pci_dev); + kfree(dev->queues); + kfree(dev->entry); + kfree(dev); +} + +static int nvme_dev_open(struct inode *inode, struct file *f) +{ + struct nvme_dev *dev = container_of(f->private_data, struct nvme_dev, + miscdev); + kref_get(&dev->kref); + f->private_data = dev; + return 0; +} + +static int nvme_dev_release(struct inode *inode, struct file *f) +{ + struct nvme_dev *dev = f->private_data; + kref_put(&dev->kref, nvme_free_dev); + return 0; +} + +static long nvme_dev_ioctl(struct file *f, unsigned int cmd, unsigned long arg) +{ + struct nvme_dev *dev = f->private_data; + switch (cmd) { + case NVME_IOCTL_ADMIN_CMD: + return nvme_user_admin_cmd(dev, (void __user *)arg); + default: + return -ENOTTY; + } +} + +static const struct file_operations nvme_dev_fops = { + .owner = THIS_MODULE, + .open = nvme_dev_open, + .release = nvme_dev_release, + .unlocked_ioctl = nvme_dev_ioctl, + .compat_ioctl = nvme_dev_ioctl, +}; + static int nvme_probe(struct pci_dev *pdev, const struct pci_device_id *id) { int bars, result = -ENOMEM; @@ -1692,8 +1954,20 @@ static int nvme_probe(struct pci_dev *pdev, const struct pci_device_id *id) if (result) goto delete; + scnprintf(dev->name, sizeof(dev->name), "nvme%d", dev->instance); + dev->miscdev.minor = MISC_DYNAMIC_MINOR; + dev->miscdev.parent = &pdev->dev; + dev->miscdev.name = dev->name; + dev->miscdev.fops = &nvme_dev_fops; + result = misc_register(&dev->miscdev); + if (result) + goto remove; + + kref_init(&dev->kref); return 0; + remove: + nvme_dev_remove(dev); delete: spin_lock(&dev_list_lock); list_del(&dev->node); @@ -1719,16 +1993,8 @@ static int nvme_probe(struct pci_dev *pdev, const struct pci_device_id *id) static void 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); + misc_deregister(&dev->miscdev); + kref_put(&dev->kref, nvme_free_dev); } /* These functions are yet to be implemented */ diff --git a/drivers/block/nvme-scsi.c b/drivers/block/nvme-scsi.c new file mode 100644 index 000000000000..fed54b039893 --- /dev/null +++ b/drivers/block/nvme-scsi.c @@ -0,0 +1,3053 @@ +/* + * 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. + */ + +/* + * Refer to the SCSI-NVMe Translation spec for details on how + * each command is translated. + */ + +#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 <linux/version.h> +#include <scsi/sg.h> +#include <scsi/scsi.h> + + +static int sg_version_num = 30534; /* 2 digits for each component */ + +#define SNTI_TRANSLATION_SUCCESS 0 +#define SNTI_INTERNAL_ERROR 1 + +/* VPD Page Codes */ +#define VPD_SUPPORTED_PAGES 0x00 +#define VPD_SERIAL_NUMBER 0x80 +#define VPD_DEVICE_IDENTIFIERS 0x83 +#define VPD_EXTENDED_INQUIRY 0x86 +#define VPD_BLOCK_DEV_CHARACTERISTICS 0xB1 + +/* CDB offsets */ +#define REPORT_LUNS_CDB_ALLOC_LENGTH_OFFSET 6 +#define REPORT_LUNS_SR_OFFSET 2 +#define READ_CAP_16_CDB_ALLOC_LENGTH_OFFSET 10 +#define REQUEST_SENSE_CDB_ALLOC_LENGTH_OFFSET 4 +#define REQUEST_SENSE_DESC_OFFSET 1 +#define REQUEST_SENSE_DESC_MASK 0x01 +#define DESCRIPTOR_FORMAT_SENSE_DATA_TYPE 1 +#define INQUIRY_EVPD_BYTE_OFFSET 1 +#define INQUIRY_PAGE_CODE_BYTE_OFFSET 2 +#define INQUIRY_EVPD_BIT_MASK 1 +#define INQUIRY_CDB_ALLOCATION_LENGTH_OFFSET 3 +#define START_STOP_UNIT_CDB_IMMED_OFFSET 1 +#define START_STOP_UNIT_CDB_IMMED_MASK 0x1 +#define START_STOP_UNIT_CDB_POWER_COND_MOD_OFFSET 3 +#define START_STOP_UNIT_CDB_POWER_COND_MOD_MASK 0xF +#define START_STOP_UNIT_CDB_POWER_COND_OFFSET 4 +#define START_STOP_UNIT_CDB_POWER_COND_MASK 0xF0 +#define START_STOP_UNIT_CDB_NO_FLUSH_OFFSET 4 +#define START_STOP_UNIT_CDB_NO_FLUSH_MASK 0x4 +#define START_STOP_UNIT_CDB_START_OFFSET 4 +#define START_STOP_UNIT_CDB_START_MASK 0x1 +#define WRITE_BUFFER_CDB_MODE_OFFSET 1 +#define WRITE_BUFFER_CDB_MODE_MASK 0x1F +#define WRITE_BUFFER_CDB_BUFFER_ID_OFFSET 2 +#define WRITE_BUFFER_CDB_BUFFER_OFFSET_OFFSET 3 +#define WRITE_BUFFER_CDB_PARM_LIST_LENGTH_OFFSET 6 +#define FORMAT_UNIT_CDB_FORMAT_PROT_INFO_OFFSET 1 +#define FORMAT_UNIT_CDB_FORMAT_PROT_INFO_MASK 0xC0 +#define FORMAT_UNIT_CDB_FORMAT_PROT_INFO_SHIFT 6 +#define FORMAT_UNIT_CDB_LONG_LIST_OFFSET 1 +#define FORMAT_UNIT_CDB_LONG_LIST_MASK 0x20 +#define FORMAT_UNIT_CDB_FORMAT_DATA_OFFSET 1 +#define FORMAT_UNIT_CDB_FORMAT_DATA_MASK 0x10 +#define FORMAT_UNIT_SHORT_PARM_LIST_LEN 4 +#define FORMAT_UNIT_LONG_PARM_LIST_LEN 8 +#define FORMAT_UNIT_PROT_INT_OFFSET 3 +#define FORMAT_UNIT_PROT_FIELD_USAGE_OFFSET 0 +#define FORMAT_UNIT_PROT_FIELD_USAGE_MASK 0x07 +#define UNMAP_CDB_PARAM_LIST_LENGTH_OFFSET 7 + +/* Misc. defines */ +#define NIBBLE_SHIFT 4 +#define FIXED_SENSE_DATA 0x70 +#define DESC_FORMAT_SENSE_DATA 0x72 +#define FIXED_SENSE_DATA_ADD_LENGTH 10 +#define LUN_ENTRY_SIZE 8 +#define LUN_DATA_HEADER_SIZE 8 +#define ALL_LUNS_RETURNED 0x02 +#define ALL_WELL_KNOWN_LUNS_RETURNED 0x01 +#define RESTRICTED_LUNS_RETURNED 0x00 +#define NVME_POWER_STATE_START_VALID 0x00 +#define NVME_POWER_STATE_ACTIVE 0x01 +#define NVME_POWER_STATE_IDLE 0x02 +#define NVME_POWER_STATE_STANDBY 0x03 +#define NVME_POWER_STATE_LU_CONTROL 0x07 +#define POWER_STATE_0 0 +#define POWER_STATE_1 1 +#define POWER_STATE_2 2 +#define POWER_STATE_3 3 +#define DOWNLOAD_SAVE_ACTIVATE 0x05 +#define DOWNLOAD_SAVE_DEFER_ACTIVATE 0x0E +#define ACTIVATE_DEFERRED_MICROCODE 0x0F +#define FORMAT_UNIT_IMMED_MASK 0x2 +#define FORMAT_UNIT_IMMED_OFFSET 1 +#define KELVIN_TEMP_FACTOR 273 +#define FIXED_FMT_SENSE_DATA_SIZE 18 +#define DESC_FMT_SENSE_DATA_SIZE 8 + +/* SCSI/NVMe defines and bit masks */ +#define INQ_STANDARD_INQUIRY_PAGE 0x00 +#define INQ_SUPPORTED_VPD_PAGES_PAGE 0x00 +#define INQ_UNIT_SERIAL_NUMBER_PAGE 0x80 +#define INQ_DEVICE_IDENTIFICATION_PAGE 0x83 +#define INQ_EXTENDED_INQUIRY_DATA_PAGE 0x86 +#define INQ_BDEV_CHARACTERISTICS_PAGE 0xB1 +#define INQ_SERIAL_NUMBER_LENGTH 0x14 +#define INQ_NUM_SUPPORTED_VPD_PAGES 5 +#define VERSION_SPC_4 0x06 +#define ACA_UNSUPPORTED 0 +#define STANDARD_INQUIRY_LENGTH 36 +#define ADDITIONAL_STD_INQ_LENGTH 31 +#define EXTENDED_INQUIRY_DATA_PAGE_LENGTH 0x3C +#define RESERVED_FIELD 0 + +/* SCSI READ/WRITE Defines */ +#define IO_CDB_WP_MASK 0xE0 +#define IO_CDB_WP_SHIFT 5 +#define IO_CDB_FUA_MASK 0x8 +#define IO_6_CDB_LBA_OFFSET 0 +#define IO_6_CDB_LBA_MASK 0x001FFFFF +#define IO_6_CDB_TX_LEN_OFFSET 4 +#define IO_6_DEFAULT_TX_LEN 256 +#define IO_10_CDB_LBA_OFFSET 2 +#define IO_10_CDB_TX_LEN_OFFSET 7 +#define IO_10_CDB_WP_OFFSET 1 +#define IO_10_CDB_FUA_OFFSET 1 +#define IO_12_CDB_LBA_OFFSET 2 +#define IO_12_CDB_TX_LEN_OFFSET 6 +#define IO_12_CDB_WP_OFFSET 1 +#define IO_12_CDB_FUA_OFFSET 1 +#define IO_16_CDB_FUA_OFFSET 1 +#define IO_16_CDB_WP_OFFSET 1 +#define IO_16_CDB_LBA_OFFSET 2 +#define IO_16_CDB_TX_LEN_OFFSET 10 + +/* Mode Sense/Select defines */ +#define MODE_PAGE_INFO_EXCEP 0x1C +#define MODE_PAGE_CACHING 0x08 +#define MODE_PAGE_CONTROL 0x0A +#define MODE_PAGE_POWER_CONDITION 0x1A +#define MODE_PAGE_RETURN_ALL 0x3F +#define MODE_PAGE_BLK_DES_LEN 0x08 +#define MODE_PAGE_LLBAA_BLK_DES_LEN 0x10 +#define MODE_PAGE_CACHING_LEN 0x14 +#define MODE_PAGE_CONTROL_LEN 0x0C +#define MODE_PAGE_POW_CND_LEN 0x28 +#define MODE_PAGE_INF_EXC_LEN 0x0C +#define MODE_PAGE_ALL_LEN 0x54 +#define MODE_SENSE6_MPH_SIZE 4 +#define MODE_SENSE6_ALLOC_LEN_OFFSET 4 +#define MODE_SENSE_PAGE_CONTROL_OFFSET 2 +#define MODE_SENSE_PAGE_CONTROL_MASK 0xC0 +#define MODE_SENSE_PAGE_CODE_OFFSET 2 +#define MODE_SENSE_PAGE_CODE_MASK 0x3F +#define MODE_SENSE_LLBAA_OFFSET 1 +#define MODE_SENSE_LLBAA_MASK 0x10 +#define MODE_SENSE_LLBAA_SHIFT 4 +#define MODE_SENSE_DBD_OFFSET 1 +#define MODE_SENSE_DBD_MASK 8 +#define MODE_SENSE_DBD_SHIFT 3 +#define MODE_SENSE10_MPH_SIZE 8 +#define MODE_SENSE10_ALLOC_LEN_OFFSET 7 +#define MODE_SELECT_CDB_PAGE_FORMAT_OFFSET 1 +#define MODE_SELECT_CDB_SAVE_PAGES_OFFSET 1 +#define MODE_SELECT_6_CDB_PARAM_LIST_LENGTH_OFFSET 4 +#define MODE_SELECT_10_CDB_PARAM_LIST_LENGTH_OFFSET 7 +#define MODE_SELECT_CDB_PAGE_FORMAT_MASK 0x10 +#define MODE_SELECT_CDB_SAVE_PAGES_MASK 0x1 +#define MODE_SELECT_6_BD_OFFSET 3 +#define MODE_SELECT_10_BD_OFFSET 6 +#define MODE_SELECT_10_LLBAA_OFFSET 4 +#define MODE_SELECT_10_LLBAA_MASK 1 +#define MODE_SELECT_6_MPH_SIZE 4 +#define MODE_SELECT_10_MPH_SIZE 8 +#define CACHING_MODE_PAGE_WCE_MASK 0x04 +#define MODE_SENSE_BLK_DESC_ENABLED 0 +#define MODE_SENSE_BLK_DESC_COUNT 1 +#define MODE_SELECT_PAGE_CODE_MASK 0x3F +#define SHORT_DESC_BLOCK 8 +#define LONG_DESC_BLOCK 16 +#define MODE_PAGE_POW_CND_LEN_FIELD 0x26 +#define MODE_PAGE_INF_EXC_LEN_FIELD 0x0A +#define MODE_PAGE_CACHING_LEN_FIELD 0x12 +#define MODE_PAGE_CONTROL_LEN_FIELD 0x0A +#define MODE_SENSE_PC_CURRENT_VALUES 0 + +/* Log Sense defines */ +#define LOG_PAGE_SUPPORTED_LOG_PAGES_PAGE 0x00 +#define LOG_PAGE_SUPPORTED_LOG_PAGES_LENGTH 0x07 +#define LOG_PAGE_INFORMATIONAL_EXCEPTIONS_PAGE 0x2F +#define LOG_PAGE_TEMPERATURE_PAGE 0x0D +#define LOG_SENSE_CDB_SP_OFFSET 1 +#define LOG_SENSE_CDB_SP_NOT_ENABLED 0 +#define LOG_SENSE_CDB_PC_OFFSET 2 +#define LOG_SENSE_CDB_PC_MASK 0xC0 +#define LOG_SENSE_CDB_PC_SHIFT 6 +#define LOG_SENSE_CDB_PC_CUMULATIVE_VALUES 1 +#define LOG_SENSE_CDB_PAGE_CODE_MASK 0x3F +#define LOG_SENSE_CDB_ALLOC_LENGTH_OFFSET 7 +#define REMAINING_INFO_EXCP_PAGE_LENGTH 0x8 +#define LOG_INFO_EXCP_PAGE_LENGTH 0xC +#define REMAINING_TEMP_PAGE_LENGTH 0xC +#define LOG_TEMP_PAGE_LENGTH 0x10 +#define LOG_TEMP_UNKNOWN 0xFF +#define SUPPORTED_LOG_PAGES_PAGE_LENGTH 0x3 + +/* Read Capacity defines */ +#define READ_CAP_10_RESP_SIZE 8 +#define READ_CAP_16_RESP_SIZE 32 + +/* NVMe Namespace and Command Defines */ +#define NVME_GET_SMART_LOG_PAGE 0x02 +#define NVME_GET_FEAT_TEMP_THRESH 0x04 +#define BYTES_TO_DWORDS 4 +#define NVME_MAX_FIRMWARE_SLOT 7 + +/* Report LUNs defines */ +#define REPORT_LUNS_FIRST_LUN_OFFSET 8 + +/* SCSI ADDITIONAL SENSE Codes */ + +#define SCSI_ASC_NO_SENSE 0x00 +#define SCSI_ASC_PERIPHERAL_DEV_WRITE_FAULT 0x03 +#define SCSI_ASC_LUN_NOT_READY 0x04 +#define SCSI_ASC_WARNING 0x0B +#define SCSI_ASC_LOG_BLOCK_GUARD_CHECK_FAILED 0x10 +#define SCSI_ASC_LOG_BLOCK_APPTAG_CHECK_FAILED 0x10 +#define SCSI_ASC_LOG_BLOCK_REFTAG_CHECK_FAILED 0x10 +#define SCSI_ASC_UNRECOVERED_READ_ERROR 0x11 +#define SCSI_ASC_MISCOMPARE_DURING_VERIFY 0x1D +#define SCSI_ASC_ACCESS_DENIED_INVALID_LUN_ID 0x20 +#define SCSI_ASC_ILLEGAL_COMMAND 0x20 +#define SCSI_ASC_ILLEGAL_BLOCK 0x21 +#define SCSI_ASC_INVALID_CDB 0x24 +#define SCSI_ASC_INVALID_LUN 0x25 +#define SCSI_ASC_INVALID_PARAMETER 0x26 +#define SCSI_ASC_FORMAT_COMMAND_FAILED 0x31 +#define SCSI_ASC_INTERNAL_TARGET_FAILURE 0x44 + +/* SCSI ADDITIONAL SENSE Code Qualifiers */ + +#define SCSI_ASCQ_CAUSE_NOT_REPORTABLE 0x00 +#define SCSI_ASCQ_FORMAT_COMMAND_FAILED 0x01 +#define SCSI_ASCQ_LOG_BLOCK_GUARD_CHECK_FAILED 0x01 +#define SCSI_ASCQ_LOG_BLOCK_APPTAG_CHECK_FAILED 0x02 +#define SCSI_ASCQ_LOG_BLOCK_REFTAG_CHECK_FAILED 0x03 +#define SCSI_ASCQ_FORMAT_IN_PROGRESS 0x04 +#define SCSI_ASCQ_POWER_LOSS_EXPECTED 0x08 +#define SCSI_ASCQ_INVALID_LUN_ID 0x09 + +/** + * DEVICE_SPECIFIC_PARAMETER in mode parameter header (see sbc2r16) to + * enable DPOFUA support type 0x10 value. + */ +#define DEVICE_SPECIFIC_PARAMETER 0 +#define VPD_ID_DESCRIPTOR_LENGTH sizeof(VPD_IDENTIFICATION_DESCRIPTOR) + +/* MACROs to extract information from CDBs */ + +#define GET_OPCODE(cdb) cdb[0] + +#define GET_U8_FROM_CDB(cdb, index) (cdb[index] << 0) + +#define GET_U16_FROM_CDB(cdb, index) ((cdb[index] << 8) | (cdb[index + 1] << 0)) + +#define GET_U24_FROM_CDB(cdb, index) ((cdb[index] << 16) | \ +(cdb[index + 1] << 8) | \ +(cdb[index + 2] << 0)) + +#define GET_U32_FROM_CDB(cdb, index) ((cdb[index] << 24) | \ +(cdb[index + 1] << 16) | \ +(cdb[index + 2] << 8) | \ +(cdb[index + 3] << 0)) + +#define GET_U64_FROM_CDB(cdb, index) ((((u64)cdb[index]) << 56) | \ +(((u64)cdb[index + 1]) << 48) | \ +(((u64)cdb[index + 2]) << 40) | \ +(((u64)cdb[index + 3]) << 32) | \ +(((u64)cdb[index + 4]) << 24) | \ +(((u64)cdb[index + 5]) << 16) | \ +(((u64)cdb[index + 6]) << 8) | \ +(((u64)cdb[index + 7]) << 0)) + +/* Inquiry Helper Macros */ +#define GET_INQ_EVPD_BIT(cdb) \ +((GET_U8_FROM_CDB(cdb, INQUIRY_EVPD_BYTE_OFFSET) & \ +INQUIRY_EVPD_BIT_MASK) ? 1 : 0) + +#define GET_INQ_PAGE_CODE(cdb) \ +(GET_U8_FROM_CDB(cdb, INQUIRY_PAGE_CODE_BYTE_OFFSET)) + +#define GET_INQ_ALLOC_LENGTH(cdb) \ +(GET_U16_FROM_CDB(cdb, INQUIRY_CDB_ALLOCATION_LENGTH_OFFSET)) + +/* Report LUNs Helper Macros */ +#define GET_REPORT_LUNS_ALLOC_LENGTH(cdb) \ +(GET_U32_FROM_CDB(cdb, REPORT_LUNS_CDB_ALLOC_LENGTH_OFFSET)) + +/* Read Capacity Helper Macros */ +#define GET_READ_CAP_16_ALLOC_LENGTH(cdb) \ +(GET_U32_FROM_CDB(cdb, READ_CAP_16_CDB_ALLOC_LENGTH_OFFSET)) + +#define IS_READ_CAP_16(cdb) \ +((cdb[0] == SERVICE_ACTION_IN && cdb[1] == SAI_READ_CAPACITY_16) ? 1 : 0) + +/* Request Sense Helper Macros */ +#define GET_REQUEST_SENSE_ALLOC_LENGTH(cdb) \ +(GET_U8_FROM_CDB(cdb, REQUEST_SENSE_CDB_ALLOC_LENGTH_OFFSET)) + +/* Mode Sense Helper Macros */ +#define GET_MODE_SENSE_DBD(cdb) \ +((GET_U8_FROM_CDB(cdb, MODE_SENSE_DBD_OFFSET) & MODE_SENSE_DBD_MASK) >> \ +MODE_SENSE_DBD_SHIFT) + +#define GET_MODE_SENSE_LLBAA(cdb) \ +((GET_U8_FROM_CDB(cdb, MODE_SENSE_LLBAA_OFFSET) & \ +MODE_SENSE_LLBAA_MASK) >> MODE_SENSE_LLBAA_SHIFT) + +#define GET_MODE_SENSE_MPH_SIZE(cdb10) \ +(cdb10 ? MODE_SENSE10_MPH_SIZE : MODE_SENSE6_MPH_SIZE) + + +/* Struct to gather data that needs to be extracted from a SCSI CDB. + Not conforming to any particular CDB variant, but compatible with all. */ + +struct nvme_trans_io_cdb { + u8 fua; + u8 prot_info; + u64 lba; + u32 xfer_len; +}; + + +/* Internal Helper Functions */ + + +/* Copy data to userspace memory */ + +static int nvme_trans_copy_to_user(struct sg_io_hdr *hdr, void *from, + unsigned long n) +{ + int res = SNTI_TRANSLATION_SUCCESS; + unsigned long not_copied; + int i; + void *index = from; + size_t remaining = n; + size_t xfer_len; + + if (hdr->iovec_count > 0) { + struct sg_iovec sgl; + + for (i = 0; i < hdr->iovec_count; i++) { + not_copied = copy_from_user(&sgl, hdr->dxferp + + i * sizeof(struct sg_iovec), + sizeof(struct sg_iovec)); + if (not_copied) + return -EFAULT; + xfer_len = min(remaining, sgl.iov_len); + not_copied = copy_to_user(sgl.iov_base, index, + xfer_len); + if (not_copied) { + res = -EFAULT; + break; + } + index += xfer_len; + remaining -= xfer_len; + if (remaining == 0) + break; + } + return res; + } + not_copied = copy_to_user(hdr->dxferp, from, n); + if (not_copied) + res = -EFAULT; + return res; +} + +/* Copy data from userspace memory */ + +static int nvme_trans_copy_from_user(struct sg_io_hdr *hdr, void *to, + unsigned long n) +{ + int res = SNTI_TRANSLATION_SUCCESS; + unsigned long not_copied; + int i; + void *index = to; + size_t remaining = n; + size_t xfer_len; + + if (hdr->iovec_count > 0) { + struct sg_iovec sgl; + + for (i = 0; i < hdr->iovec_count; i++) { + not_copied = copy_from_user(&sgl, hdr->dxferp + + i * sizeof(struct sg_iovec), + sizeof(struct sg_iovec)); + if (not_copied) + return -EFAULT; + xfer_len = min(remaining, sgl.iov_len); + not_copied = copy_from_user(index, sgl.iov_base, + xfer_len); + if (not_copied) { + res = -EFAULT; + break; + } + index += xfer_len; + remaining -= xfer_len; + if (remaining == 0) + break; + } + return res; + } + + not_copied = copy_from_user(to, hdr->dxferp, n); + if (not_copied) + res = -EFAULT; + return res; +} + +/* Status/Sense Buffer Writeback */ + +static int nvme_trans_completion(struct sg_io_hdr *hdr, u8 status, u8 sense_key, + u8 asc, u8 ascq) +{ + int res = SNTI_TRANSLATION_SUCCESS; + u8 xfer_len; + u8 resp[DESC_FMT_SENSE_DATA_SIZE]; + + if (scsi_status_is_good(status)) { + hdr->status = SAM_STAT_GOOD; + hdr->masked_status = GOOD; + hdr->host_status = DID_OK; + hdr->driver_status = DRIVER_OK; + hdr->sb_len_wr = 0; + } else { + hdr->status = status; + hdr->masked_status = status >> 1; + hdr->host_status = DID_OK; + hdr->driver_status = DRIVER_OK; + + memset(resp, 0, DESC_FMT_SENSE_DATA_SIZE); + resp[0] = DESC_FORMAT_SENSE_DATA; + resp[1] = sense_key; + resp[2] = asc; + resp[3] = ascq; + + xfer_len = min_t(u8, hdr->mx_sb_len, DESC_FMT_SENSE_DATA_SIZE); + hdr->sb_len_wr = xfer_len; + if (copy_to_user(hdr->sbp, resp, xfer_len) > 0) + res = -EFAULT; + } + + return res; +} + +static int nvme_trans_status_code(struct sg_io_hdr *hdr, int nvme_sc) +{ + u8 status, sense_key, asc, ascq; + int res = SNTI_TRANSLATION_SUCCESS; + + /* For non-nvme (Linux) errors, simply return the error code */ + if (nvme_sc < 0) + return nvme_sc; + + /* Mask DNR, More, and reserved fields */ + nvme_sc &= 0x7FF; + + switch (nvme_sc) { + /* Generic Command Status */ + case NVME_SC_SUCCESS: + status = SAM_STAT_GOOD; + sense_key = NO_SENSE; + asc = SCSI_ASC_NO_SENSE; + ascq = SCSI_ASCQ_CAUSE_NOT_REPORTABLE; + break; + case NVME_SC_INVALID_OPCODE: + status = SAM_STAT_CHECK_CONDITION; + sense_key = ILLEGAL_REQUEST; + asc = SCSI_ASC_ILLEGAL_COMMAND; + ascq = SCSI_ASCQ_CAUSE_NOT_REPORTABLE; + break; + case NVME_SC_INVALID_FIELD: + status = SAM_STAT_CHECK_CONDITION; + sense_key = ILLEGAL_REQUEST; + asc = SCSI_ASC_INVALID_CDB; + ascq = SCSI_ASCQ_CAUSE_NOT_REPORTABLE; + break; + case NVME_SC_DATA_XFER_ERROR: + status = SAM_STAT_CHECK_CONDITION; + sense_key = MEDIUM_ERROR; + asc = SCSI_ASC_NO_SENSE; + ascq = SCSI_ASCQ_CAUSE_NOT_REPORTABLE; + break; + case NVME_SC_POWER_LOSS: + status = SAM_STAT_TASK_ABORTED; + sense_key = ABORTED_COMMAND; + asc = SCSI_ASC_WARNING; + ascq = SCSI_ASCQ_POWER_LOSS_EXPECTED; + break; + case NVME_SC_INTERNAL: + status = SAM_STAT_CHECK_CONDITION; + sense_key = HARDWARE_ERROR; + asc = SCSI_ASC_INTERNAL_TARGET_FAILURE; + ascq = SCSI_ASCQ_CAUSE_NOT_REPORTABLE; + break; + case NVME_SC_ABORT_REQ: + status = SAM_STAT_TASK_ABORTED; + sense_key = ABORTED_COMMAND; + asc = SCSI_ASC_NO_SENSE; + ascq = SCSI_ASCQ_CAUSE_NOT_REPORTABLE; + break; + case NVME_SC_ABORT_QUEUE: + status = SAM_STAT_TASK_ABORTED; + sense_key = ABORTED_COMMAND; + asc = SCSI_ASC_NO_SENSE; + ascq = SCSI_ASCQ_CAUSE_NOT_REPORTABLE; + break; + case NVME_SC_FUSED_FAIL: + status = SAM_STAT_TASK_ABORTED; + sense_key = ABORTED_COMMAND; + asc = SCSI_ASC_NO_SENSE; + ascq = SCSI_ASCQ_CAUSE_NOT_REPORTABLE; + break; + case NVME_SC_FUSED_MISSING: + status = SAM_STAT_TASK_ABORTED; + sense_key = ABORTED_COMMAND; + asc = SCSI_ASC_NO_SENSE; + ascq = SCSI_ASCQ_CAUSE_NOT_REPORTABLE; + break; + case NVME_SC_INVALID_NS: + status = SAM_STAT_CHECK_CONDITION; + sense_key = ILLEGAL_REQUEST; + asc = SCSI_ASC_ACCESS_DENIED_INVALID_LUN_ID; + ascq = SCSI_ASCQ_INVALID_LUN_ID; + break; + case NVME_SC_LBA_RANGE: + status = SAM_STAT_CHECK_CONDITION; + sense_key = ILLEGAL_REQUEST; + asc = SCSI_ASC_ILLEGAL_BLOCK; + ascq = SCSI_ASCQ_CAUSE_NOT_REPORTABLE; + break; + case NVME_SC_CAP_EXCEEDED: + status = SAM_STAT_CHECK_CONDITION; + sense_key = MEDIUM_ERROR; + asc = SCSI_ASC_NO_SENSE; + ascq = SCSI_ASCQ_CAUSE_NOT_REPORTABLE; + break; + case NVME_SC_NS_NOT_READY: + status = SAM_STAT_CHECK_CONDITION; + sense_key = NOT_READY; + asc = SCSI_ASC_LUN_NOT_READY; + ascq = SCSI_ASCQ_CAUSE_NOT_REPORTABLE; + break; + + /* Command Specific Status */ + case NVME_SC_INVALID_FORMAT: + status = SAM_STAT_CHECK_CONDITION; + sense_key = ILLEGAL_REQUEST; + asc = SCSI_ASC_FORMAT_COMMAND_FAILED; + ascq = SCSI_ASCQ_FORMAT_COMMAND_FAILED; + break; + case NVME_SC_BAD_ATTRIBUTES: + status = SAM_STAT_CHECK_CONDITION; + sense_key = ILLEGAL_REQUEST; + asc = SCSI_ASC_INVALID_CDB; + ascq = SCSI_ASCQ_CAUSE_NOT_REPORTABLE; + break; + + /* Media Errors */ + case NVME_SC_WRITE_FAULT: + status = SAM_STAT_CHECK_CONDITION; + sense_key = MEDIUM_ERROR; + asc = SCSI_ASC_PERIPHERAL_DEV_WRITE_FAULT; + ascq = SCSI_ASCQ_CAUSE_NOT_REPORTABLE; + break; + case NVME_SC_READ_ERROR: + status = SAM_STAT_CHECK_CONDITION; + sense_key = MEDIUM_ERROR; + asc = SCSI_ASC_UNRECOVERED_READ_ERROR; + ascq = SCSI_ASCQ_CAUSE_NOT_REPORTABLE; + break; + case NVME_SC_GUARD_CHECK: + status = SAM_STAT_CHECK_CONDITION; + sense_key = MEDIUM_ERROR; + asc = SCSI_ASC_LOG_BLOCK_GUARD_CHECK_FAILED; + ascq = SCSI_ASCQ_LOG_BLOCK_GUARD_CHECK_FAILED; + break; + case NVME_SC_APPTAG_CHECK: + status = SAM_STAT_CHECK_CONDITION; + sense_key = MEDIUM_ERROR; + asc = SCSI_ASC_LOG_BLOCK_APPTAG_CHECK_FAILED; + ascq = SCSI_ASCQ_LOG_BLOCK_APPTAG_CHECK_FAILED; + break; + case NVME_SC_REFTAG_CHECK: + status = SAM_STAT_CHECK_CONDITION; + sense_key = MEDIUM_ERROR; + asc = SCSI_ASC_LOG_BLOCK_REFTAG_CHECK_FAILED; + ascq = SCSI_ASCQ_LOG_BLOCK_REFTAG_CHECK_FAILED; + break; + case NVME_SC_COMPARE_FAILED: + status = SAM_STAT_CHECK_CONDITION; + sense_key = MISCOMPARE; + asc = SCSI_ASC_MISCOMPARE_DURING_VERIFY; + ascq = SCSI_ASCQ_CAUSE_NOT_REPORTABLE; + break; + case NVME_SC_ACCESS_DENIED: + status = SAM_STAT_CHECK_CONDITION; + sense_key = ILLEGAL_REQUEST; + asc = SCSI_ASC_ACCESS_DENIED_INVALID_LUN_ID; + ascq = SCSI_ASCQ_INVALID_LUN_ID; + break; + + /* Unspecified/Default */ + case NVME_SC_CMDID_CONFLICT: + case NVME_SC_CMD_SEQ_ERROR: + case NVME_SC_CQ_INVALID: + case NVME_SC_QID_INVALID: + case NVME_SC_QUEUE_SIZE: + case NVME_SC_ABORT_LIMIT: + case NVME_SC_ABORT_MISSING: + case NVME_SC_ASYNC_LIMIT: + case NVME_SC_FIRMWARE_SLOT: + case NVME_SC_FIRMWARE_IMAGE: + case NVME_SC_INVALID_VECTOR: + case NVME_SC_INVALID_LOG_PAGE: + default: + status = SAM_STAT_CHECK_CONDITION; + sense_key = ILLEGAL_REQUEST; + asc = SCSI_ASC_NO_SENSE; + ascq = SCSI_ASCQ_CAUSE_NOT_REPORTABLE; + break; + } + + res = nvme_trans_completion(hdr, status, sense_key, asc, ascq); + + return res; +} + +/* INQUIRY Helper Functions */ + +static int nvme_trans_standard_inquiry_page(struct nvme_ns *ns, + struct sg_io_hdr *hdr, u8 *inq_response, + int alloc_len) +{ + struct nvme_dev *dev = ns->dev; + dma_addr_t dma_addr; + void *mem; + struct nvme_id_ns *id_ns; + int res = SNTI_TRANSLATION_SUCCESS; + int nvme_sc; + int xfer_len; + u8 resp_data_format = 0x02; + u8 protect; + u8 cmdque = 0x01 << 1; + + mem = dma_alloc_coherent(&dev->pci_dev->dev, sizeof(struct nvme_id_ns), + &dma_addr, GFP_KERNEL); + if (mem == NULL) { + res = -ENOMEM; + goto out_dma; + } + + /* nvme ns identify - use DPS value for PROTECT field */ + nvme_sc = nvme_identify(dev, ns->ns_id, 0, dma_addr); + res = nvme_trans_status_code(hdr, nvme_sc); + /* + * If nvme_sc was -ve, res will be -ve here. + * If nvme_sc was +ve, the status would bace been translated, and res + * can only be 0 or -ve. + * - If 0 && nvme_sc > 0, then go into next if where res gets nvme_sc + * - If -ve, return because its a Linux error. + */ + if (res) + goto out_free; + if (nvme_sc) { + res = nvme_sc; + goto out_free; + } + id_ns = mem; + (id_ns->dps) ? (protect = 0x01) : (protect = 0); + + memset(inq_response, 0, STANDARD_INQUIRY_LENGTH); + inq_response[2] = VERSION_SPC_4; + inq_response[3] = resp_data_format; /*normaca=0 | hisup=0 */ + inq_response[4] = ADDITIONAL_STD_INQ_LENGTH; + inq_response[5] = protect; /* sccs=0 | acc=0 | tpgs=0 | pc3=0 */ + inq_response[7] = cmdque; /* wbus16=0 | sync=0 | vs=0 */ + strncpy(&inq_response[8], "NVMe ", 8); + strncpy(&inq_response[16], dev->model, 16); + strncpy(&inq_response[32], dev->firmware_rev, 4); + + xfer_len = min(alloc_len, STANDARD_INQUIRY_LENGTH); + res = nvme_trans_copy_to_user(hdr, inq_response, xfer_len); + + out_free: + dma_free_coherent(&dev->pci_dev->dev, sizeof(struct nvme_id_ns), mem, + dma_addr); + out_dma: + return res; +} + +static int nvme_trans_supported_vpd_pages(struct nvme_ns *ns, + struct sg_io_hdr *hdr, u8 *inq_response, + int alloc_len) +{ + int res = SNTI_TRANSLATION_SUCCESS; + int xfer_len; + + memset(inq_response, 0, STANDARD_INQUIRY_LENGTH); + inq_response[1] = INQ_SUPPORTED_VPD_PAGES_PAGE; /* Page Code */ + inq_response[3] = INQ_NUM_SUPPORTED_VPD_PAGES; /* Page Length */ + inq_response[4] = INQ_SUPPORTED_VPD_PAGES_PAGE; + inq_response[5] = INQ_UNIT_SERIAL_NUMBER_PAGE; + inq_response[6] = INQ_DEVICE_IDENTIFICATION_PAGE; + inq_response[7] = INQ_EXTENDED_INQUIRY_DATA_PAGE; + inq_response[8] = INQ_BDEV_CHARACTERISTICS_PAGE; + + xfer_len = min(alloc_len, STANDARD_INQUIRY_LENGTH); + res = nvme_trans_copy_to_user(hdr, inq_response, xfer_len); + + return res; +} + +static int nvme_trans_unit_serial_page(struct nvme_ns *ns, + struct sg_io_hdr *hdr, u8 *inq_response, + int alloc_len) +{ + struct nvme_dev *dev = ns->dev; + int res = SNTI_TRANSLATION_SUCCESS; + int xfer_len; + + memset(inq_response, 0, STANDARD_INQUIRY_LENGTH); + inq_response[1] = INQ_UNIT_SERIAL_NUMBER_PAGE; /* Page Code */ + inq_response[3] = INQ_SERIAL_NUMBER_LENGTH; /* Page Length */ + strncpy(&inq_response[4], dev->serial, INQ_SERIAL_NUMBER_LENGTH); + + xfer_len = min(alloc_len, STANDARD_INQUIRY_LENGTH); + res = nvme_trans_copy_to_user(hdr, inq_response, xfer_len); + + return res; +} + +static int nvme_trans_device_id_page(struct nvme_ns *ns, struct sg_io_hdr *hdr, + u8 *inq_response, int alloc_len) +{ + struct nvme_dev *dev = ns->dev; + dma_addr_t dma_addr; + void *mem; + struct nvme_id_ctrl *id_ctrl; + int res = SNTI_TRANSLATION_SUCCESS; + int nvme_sc; + u8 ieee[4]; + int xfer_len; + __be32 tmp_id = cpu_to_be32(ns->ns_id); + + mem = dma_alloc_coherent(&dev->pci_dev->dev, sizeof(struct nvme_id_ns), + &dma_addr, GFP_KERNEL); + if (mem == NULL) { + res = -ENOMEM; + goto out_dma; + } + + /* nvme controller identify */ + nvme_sc = nvme_identify(dev, 0, 1, dma_addr); + res = nvme_trans_status_code(hdr, nvme_sc); + if (res) + goto out_free; + if (nvme_sc) { + res = nvme_sc; + goto out_free; + } + id_ctrl = mem; + + /* Since SCSI tried to save 4 bits... [SPC-4(r34) Table 591] */ + ieee[0] = id_ctrl->ieee[0] << 4; + ieee[1] = id_ctrl->ieee[0] >> 4 | id_ctrl->ieee[1] << 4; + ieee[2] = id_ctrl->ieee[1] >> 4 | id_ctrl->ieee[2] << 4; + ieee[3] = id_ctrl->ieee[2] >> 4; + + memset(inq_response, 0, STANDARD_INQUIRY_LENGTH); + inq_response[1] = INQ_DEVICE_IDENTIFICATION_PAGE; /* Page Code */ + inq_response[3] = 20; /* Page Length */ + /* Designation Descriptor start */ + inq_response[4] = 0x01; /* Proto ID=0h | Code set=1h */ + inq_response[5] = 0x03; /* PIV=0b | Asso=00b | Designator Type=3h */ + inq_response[6] = 0x00; /* Rsvd */ + inq_response[7] = 16; /* Designator Length */ + /* Designator start */ + inq_response[8] = 0x60 | ieee[3]; /* NAA=6h | IEEE ID MSB, High nibble*/ + inq_response[9] = ieee[2]; /* IEEE ID */ + inq_response[10] = ieee[1]; /* IEEE ID */ + inq_response[11] = ieee[0]; /* IEEE ID| Vendor Specific ID... */ + inq_response[12] = (dev->pci_dev->vendor & 0xFF00) >> 8; + inq_response[13] = (dev->pci_dev->vendor & 0x00FF); + inq_response[14] = dev->serial[0]; + inq_response[15] = dev->serial[1]; + inq_response[16] = dev->model[0]; + inq_response[17] = dev->model[1]; + memcpy(&inq_response[18], &tmp_id, sizeof(u32)); + /* Last 2 bytes are zero */ + + xfer_len = min(alloc_len, STANDARD_INQUIRY_LENGTH); + res = nvme_trans_copy_to_user(hdr, inq_response, xfer_len); + + out_free: + dma_free_coherent(&dev->pci_dev->dev, sizeof(struct nvme_id_ns), mem, + dma_addr); + out_dma: + return res; +} + +static int nvme_trans_ext_inq_page(struct nvme_ns *ns, struct sg_io_hdr *hdr, + int alloc_len) +{ + u8 *inq_response; + int res = SNTI_TRANSLATION_SUCCESS; + int nvme_sc; + struct nvme_dev *dev = ns->dev; + dma_addr_t dma_addr; + void *mem; + struct nvme_id_ctrl *id_ctrl; + struct nvme_id_ns *id_ns; + int xfer_len; + u8 microcode = 0x80; + u8 spt; + u8 spt_lut[8] = {0, 0, 2, 1, 4, 6, 5, 7}; + u8 grd_chk, app_chk, ref_chk, protect; + u8 uask_sup = 0x20; + u8 v_sup; + u8 luiclr = 0x01; + + inq_response = kmalloc(EXTENDED_INQUIRY_DATA_PAGE_LENGTH, GFP_KERNEL); + if (inq_response == NULL) { + res = -ENOMEM; + goto out_mem; + } + + mem = dma_alloc_coherent(&dev->pci_dev->dev, sizeof(struct nvme_id_ns), + &dma_addr, GFP_KERNEL); + if (mem == NULL) { + res = -ENOMEM; + goto out_dma; + } + + /* nvme ns identify */ + nvme_sc = nvme_identify(dev, ns->ns_id, 0, dma_addr); + res = nvme_trans_status_code(hdr, nvme_sc); + if (res) + goto out_free; + if (nvme_sc) { + res = nvme_sc; + goto out_free; + } + id_ns = mem; + spt = spt_lut[(id_ns->dpc) & 0x07] << 3; + (id_ns->dps) ? (protect = 0x01) : (protect = 0); + grd_chk = protect << 2; + app_chk = protect << 1; + ref_chk = protect; + + /* nvme controller identify */ + nvme_sc = nvme_identify(dev, 0, 1, dma_addr); + res = nvme_trans_status_code(hdr, nvme_sc); + if (res) + goto out_free; + if (nvme_sc) { + res = nvme_sc; + goto out_free; + } + id_ctrl = mem; + v_sup = id_ctrl->vwc; + + memset(inq_response, 0, EXTENDED_INQUIRY_DATA_PAGE_LENGTH); + inq_response[1] = INQ_EXTENDED_INQUIRY_DATA_PAGE; /* Page Code */ + inq_response[2] = 0x00; /* Page Length MSB */ + inq_response[3] = 0x3C; /* Page Length LSB */ + inq_response[4] = microcode | spt | grd_chk | app_chk | ref_chk; + inq_response[5] = uask_sup; + inq_response[6] = v_sup; + inq_response[7] = luiclr; + inq_response[8] = 0; + inq_response[9] = 0; + + xfer_len = min(alloc_len, EXTENDED_INQUIRY_DATA_PAGE_LENGTH); + res = nvme_trans_copy_to_user(hdr, inq_response, xfer_len); + + out_free: + dma_free_coherent(&dev->pci_dev->dev, sizeof(struct nvme_id_ns), mem, + dma_addr); + out_dma: + kfree(inq_response); + out_mem: + return res; +} + +static int nvme_trans_bdev_char_page(struct nvme_ns *ns, struct sg_io_hdr *hdr, + int alloc_len) +{ + u8 *inq_response; + int res = SNTI_TRANSLATION_SUCCESS; + int xfer_len; + + inq_response = kmalloc(EXTENDED_INQUIRY_DATA_PAGE_LENGTH, GFP_KERNEL); + if (inq_response == NULL) { + res = -ENOMEM; + goto out_mem; + } + + memset(inq_response, 0, EXTENDED_INQUIRY_DATA_PAGE_LENGTH); + inq_response[1] = INQ_BDEV_CHARACTERISTICS_PAGE; /* Page Code */ + inq_response[2] = 0x00; /* Page Length MSB */ + inq_response[3] = 0x3C; /* Page Length LSB */ + inq_response[4] = 0x00; /* Medium Rotation Rate MSB */ + inq_response[5] = 0x01; /* Medium Rotation Rate LSB */ + inq_response[6] = 0x00; /* Form Factor */ + + xfer_len = min(alloc_len, EXTENDED_INQUIRY_DATA_PAGE_LENGTH); + res = nvme_trans_copy_to_user(hdr, inq_response, xfer_len); + + kfree(inq_response); + out_mem: + return res; +} + +/* LOG SENSE Helper Functions */ + +static int nvme_trans_log_supp_pages(struct nvme_ns *ns, struct sg_io_hdr *hdr, + int alloc_len) +{ + int res = SNTI_TRANSLATION_SUCCESS; + int xfer_len; + u8 *log_response; + + log_response = kmalloc(LOG_PAGE_SUPPORTED_LOG_PAGES_LENGTH, GFP_KERNEL); + if (log_response == NULL) { + res = -ENOMEM; + goto out_mem; + } + memset(log_response, 0, LOG_PAGE_SUPPORTED_LOG_PAGES_LENGTH); + + log_response[0] = LOG_PAGE_SUPPORTED_LOG_PAGES_PAGE; + /* Subpage=0x00, Page Length MSB=0 */ + log_response[3] = SUPPORTED_LOG_PAGES_PAGE_LENGTH; + log_response[4] = LOG_PAGE_SUPPORTED_LOG_PAGES_PAGE; + log_response[5] = LOG_PAGE_INFORMATIONAL_EXCEPTIONS_PAGE; + log_response[6] = LOG_PAGE_TEMPERATURE_PAGE; + + xfer_len = min(alloc_len, LOG_PAGE_SUPPORTED_LOG_PAGES_LENGTH); + res = nvme_trans_copy_to_user(hdr, log_response, xfer_len); + + kfree(log_response); + out_mem: + return res; +} + +static int nvme_trans_log_info_exceptions(struct nvme_ns *ns, + struct sg_io_hdr *hdr, int alloc_len) +{ + int res = SNTI_TRANSLATION_SUCCESS; + int xfer_len; + u8 *log_response; + struct nvme_command c; + struct nvme_dev *dev = ns->dev; + struct nvme_smart_log *smart_log; + dma_addr_t dma_addr; + void *mem; + u8 temp_c; + u16 temp_k; + + log_response = kmalloc(LOG_INFO_EXCP_PAGE_LENGTH, GFP_KERNEL); + if (log_response == NULL) { + res = -ENOMEM; + goto out_mem; + } + memset(log_response, 0, LOG_INFO_EXCP_PAGE_LENGTH); + + mem = dma_alloc_coherent(&dev->pci_dev->dev, + sizeof(struct nvme_smart_log), + &dma_addr, GFP_KERNEL); + if (mem == NULL) { + res = -ENOMEM; + goto out_dma; + } + + /* Get SMART Log Page */ + memset(&c, 0, sizeof(c)); + c.common.opcode = nvme_admin_get_log_page; + c.common.nsid = cpu_to_le32(0xFFFFFFFF); + c.common.prp1 = cpu_to_le64(dma_addr); + c.common.cdw10[0] = cpu_to_le32(((sizeof(struct nvme_smart_log) / + BYTES_TO_DWORDS) << 16) | NVME_GET_SMART_LOG_PAGE); + res = nvme_submit_admin_cmd(dev, &c, NULL); + if (res != NVME_SC_SUCCESS) { + temp_c = LOG_TEMP_UNKNOWN; + } else { + smart_log = mem; + temp_k = (smart_log->temperature[1] << 8) + + (smart_log->temperature[0]); + temp_c = temp_k - KELVIN_TEMP_FACTOR; + } + + log_response[0] = LOG_PAGE_INFORMATIONAL_EXCEPTIONS_PAGE; + /* Subpage=0x00, Page Length MSB=0 */ + log_response[3] = REMAINING_INFO_EXCP_PAGE_LENGTH; + /* Informational Exceptions Log Parameter 1 Start */ + /* Parameter Code=0x0000 bytes 4,5 */ + log_response[6] = 0x23; /* DU=0, TSD=1, ETC=0, TMC=0, FMT_AND_LNK=11b */ + log_response[7] = 0x04; /* PARAMETER LENGTH */ + /* Add sense Code and qualifier = 0x00 each */ + /* Use Temperature from NVMe Get Log Page, convert to C from K */ + log_response[10] = temp_c; + + xfer_len = min(alloc_len, LOG_INFO_EXCP_PAGE_LENGTH); + res = nvme_trans_copy_to_user(hdr, log_response, xfer_len); + + dma_free_coherent(&dev->pci_dev->dev, sizeof(struct nvme_smart_log), + mem, dma_addr); + out_dma: + kfree(log_response); + out_mem: + return res; +} + +static int nvme_trans_log_temperature(struct nvme_ns *ns, struct sg_io_hdr *hdr, + int alloc_len) +{ + int res = SNTI_TRANSLATION_SUCCESS; + int xfer_len; + u8 *log_response; + struct nvme_command c; + struct nvme_dev *dev = ns->dev; + struct nvme_smart_log *smart_log; + dma_addr_t dma_addr; + void *mem; + u32 feature_resp; + u8 temp_c_cur, temp_c_thresh; + u16 temp_k; + + log_response = kmalloc(LOG_TEMP_PAGE_LENGTH, GFP_KERNEL); + if (log_response == NULL) { + res = -ENOMEM; + goto out_mem; + } + memset(log_response, 0, LOG_TEMP_PAGE_LENGTH); + + mem = dma_alloc_coherent(&dev->pci_dev->dev, + sizeof(struct nvme_smart_log), + &dma_addr, GFP_KERNEL); + if (mem == NULL) { + res = -ENOMEM; + goto out_dma; + } + + /* Get SMART Log Page */ + memset(&c, 0, sizeof(c)); + c.common.opcode = nvme_admin_get_log_page; + c.common.nsid = cpu_to_le32(0xFFFFFFFF); + c.common.prp1 = cpu_to_le64(dma_addr); + c.common.cdw10[0] = cpu_to_le32(((sizeof(struct nvme_smart_log) / + BYTES_TO_DWORDS) << 16) | NVME_GET_SMART_LOG_PAGE); + res = nvme_submit_admin_cmd(dev, &c, NULL); + if (res != NVME_SC_SUCCESS) { + temp_c_cur = LOG_TEMP_UNKNOWN; + } else { + smart_log = mem; + temp_k = (smart_log->temperature[1] << 8) + + (smart_log->temperature[0]); + temp_c_cur = temp_k - KELVIN_TEMP_FACTOR; + } + + /* Get Features for Temp Threshold */ + res = nvme_get_features(dev, NVME_FEAT_TEMP_THRESH, 0, 0, + &feature_resp); + if (res != NVME_SC_SUCCESS) + temp_c_thresh = LOG_TEMP_UNKNOWN; + else + temp_c_thresh = (feature_resp & 0xFFFF) - KELVIN_TEMP_FACTOR; + + log_response[0] = LOG_PAGE_TEMPERATURE_PAGE; + /* Subpage=0x00, Page Length MSB=0 */ + log_response[3] = REMAINING_TEMP_PAGE_LENGTH; + /* Temperature Log Parameter 1 (Temperature) Start */ + /* Parameter Code = 0x0000 */ + log_response[6] = 0x01; /* Format and Linking = 01b */ + log_response[7] = 0x02; /* Parameter Length */ + /* Use Temperature from NVMe Get Log Page, convert to C from K */ + log_response[9] = temp_c_cur; + /* Temperature Log Parameter 2 (Reference Temperature) Start */ + log_response[11] = 0x01; /* Parameter Code = 0x0001 */ + log_response[12] = 0x01; /* Format and Linking = 01b */ + log_response[13] = 0x02; /* Parameter Length */ + /* Use Temperature Thresh from NVMe Get Log Page, convert to C from K */ + log_response[15] = temp_c_thresh; + + xfer_len = min(alloc_len, LOG_TEMP_PAGE_LENGTH); + res = nvme_trans_copy_to_user(hdr, log_response, xfer_len); + + dma_free_coherent(&dev->pci_dev->dev, sizeof(struct nvme_smart_log), + mem, dma_addr); + out_dma: + kfree(log_response); + out_mem: + return res; +} + +/* MODE SENSE Helper Functions */ + +static int nvme_trans_fill_mode_parm_hdr(u8 *resp, int len, u8 cdb10, u8 llbaa, + u16 mode_data_length, u16 blk_desc_len) +{ + /* Quick check to make sure I don't stomp on my own memory... */ + if ((cdb10 && len < 8) || (!cdb10 && len < 4)) + return SNTI_INTERNAL_ERROR; + + if (cdb10) { + resp[0] = (mode_data_length & 0xFF00) >> 8; + resp[1] = (mode_data_length & 0x00FF); + /* resp[2] and [3] are zero */ + resp[4] = llbaa; + resp[5] = RESERVED_FIELD; + resp[6] = (blk_desc_len & 0xFF00) >> 8; + resp[7] = (blk_desc_len & 0x00FF); + } else { + resp[0] = (mode_data_length & 0x00FF); + /* resp[1] and [2] are zero */ + resp[3] = (blk_desc_len & 0x00FF); + } + + return SNTI_TRANSLATION_SUCCESS; +} + +static int nvme_trans_fill_blk_desc(struct nvme_ns *ns, struct sg_io_hdr *hdr, + u8 *resp, int len, u8 llbaa) +{ + int res = SNTI_TRANSLATION_SUCCESS; + int nvme_sc; + struct nvme_dev *dev = ns->dev; + dma_addr_t dma_addr; + void *mem; + struct nvme_id_ns *id_ns; + u8 flbas; + u32 lba_length; + + if (llbaa == 0 && len < MODE_PAGE_BLK_DES_LEN) + return SNTI_INTERNAL_ERROR; + else if (llbaa > 0 && len < MODE_PAGE_LLBAA_BLK_DES_LEN) + return SNTI_INTERNAL_ERROR; + + mem = dma_alloc_coherent(&dev->pci_dev->dev, sizeof(struct nvme_id_ns), + &dma_addr, GFP_KERNEL); + if (mem == NULL) { + res = -ENOMEM; + goto out; + } + + /* nvme ns identify */ + nvme_sc = nvme_identify(dev, ns->ns_id, 0, dma_addr); + res = nvme_trans_status_code(hdr, nvme_sc); + if (res) + goto out_dma; + if (nvme_sc) { + res = nvme_sc; + goto out_dma; + } + id_ns = mem; + flbas = (id_ns->flbas) & 0x0F; + lba_length = (1 << (id_ns->lbaf[flbas].ds)); + + if (llbaa == 0) { + __be32 tmp_cap = cpu_to_be32(le64_to_cpu(id_ns->ncap)); + /* Byte 4 is reserved */ + __be32 tmp_len = cpu_to_be32(lba_length & 0x00FFFFFF); + + memcpy(resp, &tmp_cap, sizeof(u32)); + memcpy(&resp[4], &tmp_len, sizeof(u32)); + } else { + __be64 tmp_cap = cpu_to_be64(le64_to_cpu(id_ns->ncap)); + __be32 tmp_len = cpu_to_be32(lba_length); + + memcpy(resp, &tmp_cap, sizeof(u64)); + /* Bytes 8, 9, 10, 11 are reserved */ + memcpy(&resp[12], &tmp_len, sizeof(u32)); + } + + out_dma: + dma_free_coherent(&dev->pci_dev->dev, sizeof(struct nvme_id_ns), mem, + dma_addr); + out: + return res; +} + +static int nvme_trans_fill_control_page(struct nvme_ns *ns, + struct sg_io_hdr *hdr, u8 *resp, + int len) +{ + if (len < MODE_PAGE_CONTROL_LEN) + return SNTI_INTERNAL_ERROR; + + resp[0] = MODE_PAGE_CONTROL; + resp[1] = MODE_PAGE_CONTROL_LEN_FIELD; + resp[2] = 0x0E; /* TST=000b, TMF_ONLY=0, DPICZ=1, + * D_SENSE=1, GLTSD=1, RLEC=0 */ + resp[3] = 0x12; /* Q_ALGO_MODIFIER=1h, NUAR=0, QERR=01b */ + /* Byte 4: VS=0, RAC=0, UA_INT=0, SWP=0 */ + resp[5] = 0x40; /* ATO=0, TAS=1, ATMPE=0, RWWP=0, AUTOLOAD=0 */ + /* resp[6] and [7] are obsolete, thus zero */ + resp[8] = 0xFF; /* Busy timeout period = 0xffff */ + resp[9] = 0xFF; + /* Bytes 10,11: Extended selftest completion time = 0x0000 */ + + return SNTI_TRANSLATION_SUCCESS; +} + +static int nvme_trans_fill_caching_page(struct nvme_ns *ns, + struct sg_io_hdr *hdr, + u8 *resp, int len) +{ + int res = SNTI_TRANSLATION_SUCCESS; + int nvme_sc; + struct nvme_dev *dev = ns->dev; + u32 feature_resp; + u8 vwc; + + if (len < MODE_PAGE_CACHING_LEN) + return SNTI_INTERNAL_ERROR; + + nvme_sc = nvme_get_features(dev, NVME_FEAT_VOLATILE_WC, 0, 0, + &feature_resp); + res = nvme_trans_status_code(hdr, nvme_sc); + if (res) + goto out; + if (nvme_sc) { + res = nvme_sc; + goto out; + } + vwc = feature_resp & 0x00000001; + + resp[0] = MODE_PAGE_CACHING; + resp[1] = MODE_PAGE_CACHING_LEN_FIELD; + resp[2] = vwc << 2; + + out: + return res; +} + +static int nvme_trans_fill_pow_cnd_page(struct nvme_ns *ns, + struct sg_io_hdr *hdr, u8 *resp, + int len) +{ + int res = SNTI_TRANSLATION_SUCCESS; + + if (len < MODE_PAGE_POW_CND_LEN) + return SNTI_INTERNAL_ERROR; + + resp[0] = MODE_PAGE_POWER_CONDITION; + resp[1] = MODE_PAGE_POW_CND_LEN_FIELD; + /* All other bytes are zero */ + + return res; +} + +static int nvme_trans_fill_inf_exc_page(struct nvme_ns *ns, + struct sg_io_hdr *hdr, u8 *resp, + int len) +{ + int res = SNTI_TRANSLATION_SUCCESS; + + if (len < MODE_PAGE_INF_EXC_LEN) + return SNTI_INTERNAL_ERROR; + + resp[0] = MODE_PAGE_INFO_EXCEP; + resp[1] = MODE_PAGE_INF_EXC_LEN_FIELD; + resp[2] = 0x88; + /* All other bytes are zero */ + + return res; +} + +static int nvme_trans_fill_all_pages(struct nvme_ns *ns, struct sg_io_hdr *hdr, + u8 *resp, int len) +{ + int res = SNTI_TRANSLATION_SUCCESS; + u16 mode_pages_offset_1 = 0; + u16 mode_pages_offset_2, mode_pages_offset_3, mode_pages_offset_4; + + mode_pages_offset_2 = mode_pages_offset_1 + MODE_PAGE_CACHING_LEN; + mode_pages_offset_3 = mode_pages_offset_2 + MODE_PAGE_CONTROL_LEN; + mode_pages_offset_4 = mode_pages_offset_3 + MODE_PAGE_POW_CND_LEN; + + res = nvme_trans_fill_caching_page(ns, hdr, &resp[mode_pages_offset_1], + MODE_PAGE_CACHING_LEN); + if (res != SNTI_TRANSLATION_SUCCESS) + goto out; + res = nvme_trans_fill_control_page(ns, hdr, &resp[mode_pages_offset_2], + MODE_PAGE_CONTROL_LEN); + if (res != SNTI_TRANSLATION_SUCCESS) + goto out; + res = nvme_trans_fill_pow_cnd_page(ns, hdr, &resp[mode_pages_offset_3], + MODE_PAGE_POW_CND_LEN); + if (res != SNTI_TRANSLATION_SUCCESS) + goto out; + res = nvme_trans_fill_inf_exc_page(ns, hdr, &resp[mode_pages_offset_4], + MODE_PAGE_INF_EXC_LEN); + if (res != SNTI_TRANSLATION_SUCCESS) + goto out; + + out: + return res; +} + +static inline int nvme_trans_get_blk_desc_len(u8 dbd, u8 llbaa) +{ + if (dbd == MODE_SENSE_BLK_DESC_ENABLED) { + /* SPC-4: len = 8 x Num_of_descriptors if llbaa = 0, 16x if 1 */ + return 8 * (llbaa + 1) * MODE_SENSE_BLK_DESC_COUNT; + } else { + return 0; + } +} + +static int nvme_trans_mode_page_create(struct nvme_ns *ns, + struct sg_io_hdr *hdr, u8 *cmd, + u16 alloc_len, u8 cdb10, + int (*mode_page_fill_func) + (struct nvme_ns *, + struct sg_io_hdr *hdr, u8 *, int), + u16 mode_pages_tot_len) +{ + int res = SNTI_TRANSLATION_SUCCESS; + int xfer_len; + u8 *response; + u8 dbd, llbaa; + u16 resp_size; + int mph_size; + u16 mode_pages_offset_1; + u16 blk_desc_len, blk_desc_offset, mode_data_length; + + dbd = GET_MODE_SENSE_DBD(cmd); + llbaa = GET_MODE_SENSE_LLBAA(cmd); + mph_size = GET_MODE_SENSE_MPH_SIZE(cdb10); + blk_desc_len = nvme_trans_get_blk_desc_len(dbd, llbaa); + + resp_size = mph_size + blk_desc_len + mode_pages_tot_len; + /* Refer spc4r34 Table 440 for calculation of Mode data Length field */ + mode_data_length = 3 + (3 * cdb10) + blk_desc_len + mode_pages_tot_len; + + blk_desc_offset = mph_size; + mode_pages_offset_1 = blk_desc_offset + blk_desc_len; + + response = kmalloc(resp_size, GFP_KERNEL); + if (response == NULL) { + res = -ENOMEM; + goto out_mem; + } + memset(response, 0, resp_size); + + res = nvme_trans_fill_mode_parm_hdr(&response[0], mph_size, cdb10, + llbaa, mode_data_length, blk_desc_len); + if (res != SNTI_TRANSLATION_SUCCESS) + goto out_free; + if (blk_desc_len > 0) { + res = nvme_trans_fill_blk_desc(ns, hdr, + &response[blk_desc_offset], + blk_desc_len, llbaa); + if (res != SNTI_TRANSLATION_SUCCESS) + goto out_free; + } + res = mode_page_fill_func(ns, hdr, &response[mode_pages_offset_1], + mode_pages_tot_len); + if (res != SNTI_TRANSLATION_SUCCESS) + goto out_free; + + xfer_len = min(alloc_len, resp_size); + res = nvme_trans_copy_to_user(hdr, response, xfer_len); + + out_free: + kfree(response); + out_mem: + return res; +} + +/* Read Capacity Helper Functions */ + +static void nvme_trans_fill_read_cap(u8 *response, struct nvme_id_ns *id_ns, + u8 cdb16) +{ + u8 flbas; + u32 lba_length; + u64 rlba; + u8 prot_en; + u8 p_type_lut[4] = {0, 0, 1, 2}; + __be64 tmp_rlba; + __be32 tmp_rlba_32; + __be32 tmp_len; + + flbas = (id_ns->flbas) & 0x0F; + lba_length = (1 << (id_ns->lbaf[flbas].ds)); + rlba = le64_to_cpup(&id_ns->nsze) - 1; + (id_ns->dps) ? (prot_en = 0x01) : (prot_en = 0); + + if (!cdb16) { + if (rlba > 0xFFFFFFFF) + rlba = 0xFFFFFFFF; + tmp_rlba_32 = cpu_to_be32(rlba); + tmp_len = cpu_to_be32(lba_length); + memcpy(response, &tmp_rlba_32, sizeof(u32)); + memcpy(&response[4], &tmp_len, sizeof(u32)); + } else { + tmp_rlba = cpu_to_be64(rlba); + tmp_len = cpu_to_be32(lba_length); + memcpy(response, &tmp_rlba, sizeof(u64)); + memcpy(&response[8], &tmp_len, sizeof(u32)); + response[12] = (p_type_lut[id_ns->dps & 0x3] << 1) | prot_en; + /* P_I_Exponent = 0x0 | LBPPBE = 0x0 */ + /* LBPME = 0 | LBPRZ = 0 | LALBA = 0x00 */ + /* Bytes 16-31 - Reserved */ + } +} + +/* Start Stop Unit Helper Functions */ + +static int nvme_trans_power_state(struct nvme_ns *ns, struct sg_io_hdr *hdr, + u8 pc, u8 pcmod, u8 start) +{ + int res = SNTI_TRANSLATION_SUCCESS; + int nvme_sc; + struct nvme_dev *dev = ns->dev; + dma_addr_t dma_addr; + void *mem; + struct nvme_id_ctrl *id_ctrl; + int lowest_pow_st; /* max npss = lowest power consumption */ + unsigned ps_desired = 0; + + /* NVMe Controller Identify */ + mem = dma_alloc_coherent(&dev->pci_dev->dev, + sizeof(struct nvme_id_ctrl), + &dma_addr, GFP_KERNEL); + if (mem == NULL) { + res = -ENOMEM; + goto out; + } + nvme_sc = nvme_identify(dev, 0, 1, dma_addr); + res = nvme_trans_status_code(hdr, nvme_sc); + if (res) + goto out_dma; + if (nvme_sc) { + res = nvme_sc; + goto out_dma; + } + id_ctrl = mem; + lowest_pow_st = id_ctrl->npss - 1; + + switch (pc) { + case NVME_POWER_STATE_START_VALID: + /* Action unspecified if POWER CONDITION MODIFIER != 0 */ + if (pcmod == 0 && start == 0x1) + ps_desired = POWER_STATE_0; + if (pcmod == 0 && start == 0x0) + ps_desired = lowest_pow_st; + break; + case NVME_POWER_STATE_ACTIVE: + /* Action unspecified if POWER CONDITION MODIFIER != 0 */ + if (pcmod == 0) + ps_desired = POWER_STATE_0; + break; + case NVME_POWER_STATE_IDLE: + /* Action unspecified if POWER CONDITION MODIFIER != [0,1,2] */ + /* min of desired state and (lps-1) because lps is STOP */ + if (pcmod == 0x0) + ps_desired = min(POWER_STATE_1, (lowest_pow_st - 1)); + else if (pcmod == 0x1) + ps_desired = min(POWER_STATE_2, (lowest_pow_st - 1)); + else if (pcmod == 0x2) + ps_desired = min(POWER_STATE_3, (lowest_pow_st - 1)); + break; + case NVME_POWER_STATE_STANDBY: + /* Action unspecified if POWER CONDITION MODIFIER != [0,1] */ + if (pcmod == 0x0) + ps_desired = max(0, (lowest_pow_st - 2)); + else if (pcmod == 0x1) + ps_desired = max(0, (lowest_pow_st - 1)); + break; + case NVME_POWER_STATE_LU_CONTROL: + default: + res = nvme_trans_completion(hdr, SAM_STAT_CHECK_CONDITION, + ILLEGAL_REQUEST, SCSI_ASC_INVALID_CDB, + SCSI_ASCQ_CAUSE_NOT_REPORTABLE); + break; + } + nvme_sc = nvme_set_features(dev, NVME_FEAT_POWER_MGMT, ps_desired, 0, + NULL); + res = nvme_trans_status_code(hdr, nvme_sc); + if (res) + goto out_dma; + if (nvme_sc) + res = nvme_sc; + out_dma: + dma_free_coherent(&dev->pci_dev->dev, sizeof(struct nvme_id_ctrl), mem, + dma_addr); + out: + return res; +} + +/* Write Buffer Helper Functions */ +/* Also using this for Format Unit with hdr passed as NULL, and buffer_id, 0 */ + +static int nvme_trans_send_fw_cmd(struct nvme_ns *ns, struct sg_io_hdr *hdr, + u8 opcode, u32 tot_len, u32 offset, + u8 buffer_id) +{ + int res = SNTI_TRANSLATION_SUCCESS; + int nvme_sc; + struct nvme_dev *dev = ns->dev; + struct nvme_command c; + struct nvme_iod *iod = NULL; + unsigned length; + + memset(&c, 0, sizeof(c)); + c.common.opcode = opcode; + if (opcode == nvme_admin_download_fw) { + if (hdr->iovec_count > 0) { + /* Assuming SGL is not allowed for this command */ + res = nvme_trans_completion(hdr, + SAM_STAT_CHECK_CONDITION, + ILLEGAL_REQUEST, + SCSI_ASC_INVALID_CDB, + SCSI_ASCQ_CAUSE_NOT_REPORTABLE); + goto out; + } + iod = nvme_map_user_pages(dev, DMA_TO_DEVICE, + (unsigned long)hdr->dxferp, tot_len); + if (IS_ERR(iod)) { + res = PTR_ERR(iod); + goto out; + } + length = nvme_setup_prps(dev, &c.common, iod, tot_len, + GFP_KERNEL); + if (length != tot_len) { + res = -ENOMEM; + goto out_unmap; + } + + c.dlfw.numd = cpu_to_le32((tot_len/BYTES_TO_DWORDS) - 1); + c.dlfw.offset = cpu_to_le32(offset/BYTES_TO_DWORDS); + } else if (opcode == nvme_admin_activate_fw) { + u32 cdw10 = buffer_id | NVME_FWACT_REPL_ACTV; + c.common.cdw10[0] = cpu_to_le32(cdw10); + } + + nvme_sc = nvme_submit_admin_cmd(dev, &c, NULL); + res = nvme_trans_status_code(hdr, nvme_sc); + if (res) + goto out_unmap; + if (nvme_sc) + res = nvme_sc; + + out_unmap: + if (opcode == nvme_admin_download_fw) { + nvme_unmap_user_pages(dev, DMA_TO_DEVICE, iod); + nvme_free_iod(dev, iod); + } + out: + return res; +} + +/* Mode Select Helper Functions */ + +static inline void nvme_trans_modesel_get_bd_len(u8 *parm_list, u8 cdb10, + u16 *bd_len, u8 *llbaa) +{ + if (cdb10) { + /* 10 Byte CDB */ + *bd_len = (parm_list[MODE_SELECT_10_BD_OFFSET] << 8) + + parm_list[MODE_SELECT_10_BD_OFFSET + 1]; + *llbaa = parm_list[MODE_SELECT_10_LLBAA_OFFSET] && + MODE_SELECT_10_LLBAA_MASK; + } else { + /* 6 Byte CDB */ + *bd_len = parm_list[MODE_SELECT_6_BD_OFFSET]; + } +} + +static void nvme_trans_modesel_save_bd(struct nvme_ns *ns, u8 *parm_list, + u16 idx, u16 bd_len, u8 llbaa) +{ + u16 bd_num; + + bd_num = bd_len / ((llbaa == 0) ? + SHORT_DESC_BLOCK : LONG_DESC_BLOCK); + /* Store block descriptor info if a FORMAT UNIT comes later */ + /* TODO Saving 1st BD info; what to do if multiple BD received? */ + if (llbaa == 0) { + /* Standard Block Descriptor - spc4r34 7.5.5.1 */ + ns->mode_select_num_blocks = + (parm_list[idx + 1] << 16) + + (parm_list[idx + 2] << 8) + + (parm_list[idx + 3]); + + ns->mode_select_block_len = + (parm_list[idx + 5] << 16) + + (parm_list[idx + 6] << 8) + + (parm_list[idx + 7]); + } else { + /* Long LBA Block Descriptor - sbc3r27 6.4.2.3 */ + ns->mode_select_num_blocks = + (((u64)parm_list[idx + 0]) << 56) + + (((u64)parm_list[idx + 1]) << 48) + + (((u64)parm_list[idx + 2]) << 40) + + (((u64)parm_list[idx + 3]) << 32) + + (((u64)parm_list[idx + 4]) << 24) + + (((u64)parm_list[idx + 5]) << 16) + + (((u64)parm_list[idx + 6]) << 8) + + ((u64)parm_list[idx + 7]); + + ns->mode_select_block_len = + (parm_list[idx + 12] << 24) + + (parm_list[idx + 13] << 16) + + (parm_list[idx + 14] << 8) + + (parm_list[idx + 15]); + } +} + +static u16 nvme_trans_modesel_get_mp(struct nvme_ns *ns, struct sg_io_hdr *hdr, + u8 *mode_page, u8 page_code) +{ + int res = SNTI_TRANSLATION_SUCCESS; + int nvme_sc; + struct nvme_dev *dev = ns->dev; + unsigned dword11; + + switch (page_code) { + case MODE_PAGE_CACHING: + dword11 = ((mode_page[2] & CACHING_MODE_PAGE_WCE_MASK) ? 1 : 0); + nvme_sc = nvme_set_features(dev, NVME_FEAT_VOLATILE_WC, dword11, + 0, NULL); + res = nvme_trans_status_code(hdr, nvme_sc); + if (res) + break; + if (nvme_sc) { + res = nvme_sc; + break; + } + break; + case MODE_PAGE_CONTROL: + break; + case MODE_PAGE_POWER_CONDITION: + /* Verify the OS is not trying to set timers */ + if ((mode_page[2] & 0x01) != 0 || (mode_page[3] & 0x0F) != 0) { + res = nvme_trans_completion(hdr, + SAM_STAT_CHECK_CONDITION, + ILLEGAL_REQUEST, + SCSI_ASC_INVALID_PARAMETER, + SCSI_ASCQ_CAUSE_NOT_REPORTABLE); + if (!res) + res = SNTI_INTERNAL_ERROR; + break; + } + break; + default: + res = nvme_trans_completion(hdr, SAM_STAT_CHECK_CONDITION, + ILLEGAL_REQUEST, SCSI_ASC_INVALID_CDB, + SCSI_ASCQ_CAUSE_NOT_REPORTABLE); + if (!res) + res = SNTI_INTERNAL_ERROR; + break; + } + + return res; +} + +static int nvme_trans_modesel_data(struct nvme_ns *ns, struct sg_io_hdr *hdr, + u8 *cmd, u16 parm_list_len, u8 pf, + u8 sp, u8 cdb10) +{ + int res = SNTI_TRANSLATION_SUCCESS; + u8 *parm_list; + u16 bd_len; + u8 llbaa = 0; + u16 index, saved_index; + u8 page_code; + u16 mp_size; + + /* Get parm list from data-in/out buffer */ + parm_list = kmalloc(parm_list_len, GFP_KERNEL); + if (parm_list == NULL) { + res = -ENOMEM; + goto out; + } + + res = nvme_trans_copy_from_user(hdr, parm_list, parm_list_len); + if (res != SNTI_TRANSLATION_SUCCESS) + goto out_mem; + + nvme_trans_modesel_get_bd_len(parm_list, cdb10, &bd_len, &llbaa); + index = (cdb10) ? (MODE_SELECT_10_MPH_SIZE) : (MODE_SELECT_6_MPH_SIZE); + + if (bd_len != 0) { + /* Block Descriptors present, parse */ + nvme_trans_modesel_save_bd(ns, parm_list, index, bd_len, llbaa); + index += bd_len; + } + saved_index = index; + + /* Multiple mode pages may be present; iterate through all */ + /* In 1st Iteration, don't do NVME Command, only check for CDB errors */ + do { + page_code = parm_list[index] & MODE_SELECT_PAGE_CODE_MASK; + mp_size = parm_list[index + 1] + 2; + if ((page_code != MODE_PAGE_CACHING) && + (page_code != MODE_PAGE_CONTROL) && + (page_code != MODE_PAGE_POWER_CONDITION)) { + res = nvme_trans_completion(hdr, + SAM_STAT_CHECK_CONDITION, + ILLEGAL_REQUEST, + SCSI_ASC_INVALID_CDB, + SCSI_ASCQ_CAUSE_NOT_REPORTABLE); + goto out_mem; + } + index += mp_size; + } while (index < parm_list_len); + + /* In 2nd Iteration, do the NVME Commands */ + index = saved_index; + do { + page_code = parm_list[index] & MODE_SELECT_PAGE_CODE_MASK; + mp_size = parm_list[index + 1] + 2; + res = nvme_trans_modesel_get_mp(ns, hdr, &parm_list[index], + page_code); + if (res != SNTI_TRANSLATION_SUCCESS) + break; + index += mp_size; + } while (index < parm_list_len); + + out_mem: + kfree(parm_list); + out: + return res; +} + +/* Format Unit Helper Functions */ + +static int nvme_trans_fmt_set_blk_size_count(struct nvme_ns *ns, + struct sg_io_hdr *hdr) +{ + int res = SNTI_TRANSLATION_SUCCESS; + int nvme_sc; + struct nvme_dev *dev = ns->dev; + dma_addr_t dma_addr; + void *mem; + struct nvme_id_ns *id_ns; + u8 flbas; + + /* + * SCSI Expects a MODE SELECT would have been issued prior to + * a FORMAT UNIT, and the block size and number would be used + * from the block descriptor in it. If a MODE SELECT had not + * been issued, FORMAT shall use the current values for both. + */ + + if (ns->mode_select_num_blocks == 0 || ns->mode_select_block_len == 0) { + mem = dma_alloc_coherent(&dev->pci_dev->dev, + sizeof(struct nvme_id_ns), &dma_addr, GFP_KERNEL); + if (mem == NULL) { + res = -ENOMEM; + goto out; + } + /* nvme ns identify */ + nvme_sc = nvme_identify(dev, ns->ns_id, 0, dma_addr); + res = nvme_trans_status_code(hdr, nvme_sc); + if (res) + goto out_dma; + if (nvme_sc) { + res = nvme_sc; + goto out_dma; + } + id_ns = mem; + + if (ns->mode_select_num_blocks == 0) + ns->mode_select_num_blocks = le64_to_cpu(id_ns->ncap); + if (ns->mode_select_block_len == 0) { + flbas = (id_ns->flbas) & 0x0F; + ns->mode_select_block_len = + (1 << (id_ns->lbaf[flbas].ds)); + } + out_dma: + dma_free_coherent(&dev->pci_dev->dev, sizeof(struct nvme_id_ns), + mem, dma_addr); + } + out: + return res; +} + +static int nvme_trans_fmt_get_parm_header(struct sg_io_hdr *hdr, u8 len, + u8 format_prot_info, u8 *nvme_pf_code) +{ + int res = SNTI_TRANSLATION_SUCCESS; + u8 *parm_list; + u8 pf_usage, pf_code; + + parm_list = kmalloc(len, GFP_KERNEL); + if (parm_list == NULL) { + res = -ENOMEM; + goto out; + } + res = nvme_trans_copy_from_user(hdr, parm_list, len); + if (res != SNTI_TRANSLATION_SUCCESS) + goto out_mem; + + if ((parm_list[FORMAT_UNIT_IMMED_OFFSET] & + FORMAT_UNIT_IMMED_MASK) != 0) { + res = nvme_trans_completion(hdr, SAM_STAT_CHECK_CONDITION, + ILLEGAL_REQUEST, SCSI_ASC_INVALID_CDB, + SCSI_ASCQ_CAUSE_NOT_REPORTABLE); + goto out_mem; + } + + if (len == FORMAT_UNIT_LONG_PARM_LIST_LEN && + (parm_list[FORMAT_UNIT_PROT_INT_OFFSET] & 0x0F) != 0) { + res = nvme_trans_completion(hdr, SAM_STAT_CHECK_CONDITION, + ILLEGAL_REQUEST, SCSI_ASC_INVALID_CDB, + SCSI_ASCQ_CAUSE_NOT_REPORTABLE); + goto out_mem; + } + pf_usage = parm_list[FORMAT_UNIT_PROT_FIELD_USAGE_OFFSET] & + FORMAT_UNIT_PROT_FIELD_USAGE_MASK; + pf_code = (pf_usage << 2) | format_prot_info; + switch (pf_code) { + case 0: + *nvme_pf_code = 0; + break; + case 2: + *nvme_pf_code = 1; + break; + case 3: + *nvme_pf_code = 2; + break; + case 7: + *nvme_pf_code = 3; + break; + default: + res = nvme_trans_completion(hdr, SAM_STAT_CHECK_CONDITION, + ILLEGAL_REQUEST, SCSI_ASC_INVALID_CDB, + SCSI_ASCQ_CAUSE_NOT_REPORTABLE); + break; + } + + out_mem: + kfree(parm_list); + out: + return res; +} + +static int nvme_trans_fmt_send_cmd(struct nvme_ns *ns, struct sg_io_hdr *hdr, + u8 prot_info) +{ + int res = SNTI_TRANSLATION_SUCCESS; + int nvme_sc; + struct nvme_dev *dev = ns->dev; + dma_addr_t dma_addr; + void *mem; + struct nvme_id_ns *id_ns; + u8 i; + u8 flbas, nlbaf; + u8 selected_lbaf = 0xFF; + u32 cdw10 = 0; + struct nvme_command c; + + /* Loop thru LBAF's in id_ns to match reqd lbaf, put in cdw10 */ + mem = dma_alloc_coherent(&dev->pci_dev->dev, sizeof(struct nvme_id_ns), + &dma_addr, GFP_KERNEL); + if (mem == NULL) { + res = -ENOMEM; + goto out; + } + /* nvme ns identify */ + nvme_sc = nvme_identify(dev, ns->ns_id, 0, dma_addr); + res = nvme_trans_status_code(hdr, nvme_sc); + if (res) + goto out_dma; + if (nvme_sc) { + res = nvme_sc; + goto out_dma; + } + id_ns = mem; + flbas = (id_ns->flbas) & 0x0F; + nlbaf = id_ns->nlbaf; + + for (i = 0; i < nlbaf; i++) { + if (ns->mode_select_block_len == (1 << (id_ns->lbaf[i].ds))) { + selected_lbaf = i; + break; + } + } + if (selected_lbaf > 0x0F) { + res = nvme_trans_completion(hdr, SAM_STAT_CHECK_CONDITION, + ILLEGAL_REQUEST, SCSI_ASC_INVALID_PARAMETER, + SCSI_ASCQ_CAUSE_NOT_REPORTABLE); + } + if (ns->mode_select_num_blocks != le64_to_cpu(id_ns->ncap)) { + res = nvme_trans_completion(hdr, SAM_STAT_CHECK_CONDITION, + ILLEGAL_REQUEST, SCSI_ASC_INVALID_PARAMETER, + SCSI_ASCQ_CAUSE_NOT_REPORTABLE); + } + + cdw10 |= prot_info << 5; + cdw10 |= selected_lbaf & 0x0F; + memset(&c, 0, sizeof(c)); + c.format.opcode = nvme_admin_format_nvm; + c.format.nsid = cpu_to_le32(ns->ns_id); + c.format.cdw10 = cpu_to_le32(cdw10); + + nvme_sc = nvme_submit_admin_cmd(dev, &c, NULL); + res = nvme_trans_status_code(hdr, nvme_sc); + if (res) + goto out_dma; + if (nvme_sc) + res = nvme_sc; + + out_dma: + dma_free_coherent(&dev->pci_dev->dev, sizeof(struct nvme_id_ns), mem, + dma_addr); + out: + return res; +} + +/* Read/Write Helper Functions */ + +static inline void nvme_trans_get_io_cdb6(u8 *cmd, + struct nvme_trans_io_cdb *cdb_info) +{ + cdb_info->fua = 0; + cdb_info->prot_info = 0; + cdb_info->lba = GET_U32_FROM_CDB(cmd, IO_6_CDB_LBA_OFFSET) & + IO_6_CDB_LBA_MASK; + cdb_info->xfer_len = GET_U8_FROM_CDB(cmd, IO_6_CDB_TX_LEN_OFFSET); + + /* sbc3r27 sec 5.32 - TRANSFER LEN of 0 implies a 256 Block transfer */ + if (cdb_info->xfer_len == 0) + cdb_info->xfer_len = IO_6_DEFAULT_TX_LEN; +} + +static inline void nvme_trans_get_io_cdb10(u8 *cmd, + struct nvme_trans_io_cdb *cdb_info) +{ + cdb_info->fua = GET_U8_FROM_CDB(cmd, IO_10_CDB_FUA_OFFSET) & + IO_CDB_FUA_MASK; + cdb_info->prot_info = GET_U8_FROM_CDB(cmd, IO_10_CDB_WP_OFFSET) & + IO_CDB_WP_MASK >> IO_CDB_WP_SHIFT; + cdb_info->lba = GET_U32_FROM_CDB(cmd, IO_10_CDB_LBA_OFFSET); + cdb_info->xfer_len = GET_U16_FROM_CDB(cmd, IO_10_CDB_TX_LEN_OFFSET); +} + +static inline void nvme_trans_get_io_cdb12(u8 *cmd, + struct nvme_trans_io_cdb *cdb_info) +{ + cdb_info->fua = GET_U8_FROM_CDB(cmd, IO_12_CDB_FUA_OFFSET) & + IO_CDB_FUA_MASK; + cdb_info->prot_info = GET_U8_FROM_CDB(cmd, IO_12_CDB_WP_OFFSET) & + IO_CDB_WP_MASK >> IO_CDB_WP_SHIFT; + cdb_info->lba = GET_U32_FROM_CDB(cmd, IO_12_CDB_LBA_OFFSET); + cdb_info->xfer_len = GET_U32_FROM_CDB(cmd, IO_12_CDB_TX_LEN_OFFSET); +} + +static inline void nvme_trans_get_io_cdb16(u8 *cmd, + struct nvme_trans_io_cdb *cdb_info) +{ + cdb_info->fua = GET_U8_FROM_CDB(cmd, IO_16_CDB_FUA_OFFSET) & + IO_CDB_FUA_MASK; + cdb_info->prot_info = GET_U8_FROM_CDB(cmd, IO_16_CDB_WP_OFFSET) & + IO_CDB_WP_MASK >> IO_CDB_WP_SHIFT; + cdb_info->lba = GET_U64_FROM_CDB(cmd, IO_16_CDB_LBA_OFFSET); + cdb_info->xfer_len = GET_U32_FROM_CDB(cmd, IO_16_CDB_TX_LEN_OFFSET); +} + +static inline u32 nvme_trans_io_get_num_cmds(struct sg_io_hdr *hdr, + struct nvme_trans_io_cdb *cdb_info, + u32 max_blocks) +{ + /* If using iovecs, send one nvme command per vector */ + if (hdr->iovec_count > 0) + return hdr->iovec_count; + else if (cdb_info->xfer_len > max_blocks) + return ((cdb_info->xfer_len - 1) / max_blocks) + 1; + else + return 1; +} + +static u16 nvme_trans_io_get_control(struct nvme_ns *ns, + struct nvme_trans_io_cdb *cdb_info) +{ + u16 control = 0; + + /* When Protection information support is added, implement here */ + + if (cdb_info->fua > 0) + control |= NVME_RW_FUA; + + return control; +} + +static int nvme_trans_do_nvme_io(struct nvme_ns *ns, struct sg_io_hdr *hdr, + struct nvme_trans_io_cdb *cdb_info, u8 is_write) +{ + int res = SNTI_TRANSLATION_SUCCESS; + int nvme_sc; + struct nvme_dev *dev = ns->dev; + struct nvme_queue *nvmeq; + u32 num_cmds; + struct nvme_iod *iod; + u64 unit_len; + u64 unit_num_blocks; /* Number of blocks to xfer in each nvme cmd */ + u32 retcode; + u32 i = 0; + u64 nvme_offset = 0; + void __user *next_mapping_addr; + struct nvme_command c; + u8 opcode = (is_write ? nvme_cmd_write : nvme_cmd_read); + u16 control; + u32 max_blocks = nvme_block_nr(ns, dev->max_hw_sectors); + + num_cmds = nvme_trans_io_get_num_cmds(hdr, cdb_info, max_blocks); + + /* + * This loop handles two cases. + * First, when an SGL is used in the form of an iovec list: + * - Use iov_base as the next mapping address for the nvme command_id + * - Use iov_len as the data transfer length for the command. + * Second, when we have a single buffer + * - If larger than max_blocks, split into chunks, offset + * each nvme command accordingly. + */ + for (i = 0; i < num_cmds; i++) { + memset(&c, 0, sizeof(c)); + if (hdr->iovec_count > 0) { + struct sg_iovec sgl; + + retcode = copy_from_user(&sgl, hdr->dxferp + + i * sizeof(struct sg_iovec), + sizeof(struct sg_iovec)); + if (retcode) + return -EFAULT; + unit_len = sgl.iov_len; + unit_num_blocks = unit_len >> ns->lba_shift; + next_mapping_addr = sgl.iov_base; + } else { + unit_num_blocks = min((u64)max_blocks, + (cdb_info->xfer_len - nvme_offset)); + unit_len = unit_num_blocks << ns->lba_shift; + next_mapping_addr = hdr->dxferp + + ((1 << ns->lba_shift) * nvme_offset); + } + + c.rw.opcode = opcode; + c.rw.nsid = cpu_to_le32(ns->ns_id); + c.rw.slba = cpu_to_le64(cdb_info->lba + nvme_offset); + c.rw.length = cpu_to_le16(unit_num_blocks - 1); + control = nvme_trans_io_get_control(ns, cdb_info); + c.rw.control = cpu_to_le16(control); + + iod = nvme_map_user_pages(dev, + (is_write) ? DMA_TO_DEVICE : DMA_FROM_DEVICE, + (unsigned long)next_mapping_addr, unit_len); + if (IS_ERR(iod)) { + res = PTR_ERR(iod); + goto out; + } + retcode = nvme_setup_prps(dev, &c.common, iod, unit_len, + GFP_KERNEL); + if (retcode != unit_len) { + nvme_unmap_user_pages(dev, + (is_write) ? DMA_TO_DEVICE : DMA_FROM_DEVICE, + iod); + nvme_free_iod(dev, iod); + res = -ENOMEM; + goto out; + } + + nvme_offset += unit_num_blocks; + + 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); + nvme_sc = nvme_submit_sync_cmd(nvmeq, &c, NULL, + NVME_IO_TIMEOUT); + if (nvme_sc != NVME_SC_SUCCESS) { + nvme_unmap_user_pages(dev, + (is_write) ? DMA_TO_DEVICE : DMA_FROM_DEVICE, + iod); + nvme_free_iod(dev, iod); + res = nvme_trans_status_code(hdr, nvme_sc); + goto out; + } + nvme_unmap_user_pages(dev, + (is_write) ? DMA_TO_DEVICE : DMA_FROM_DEVICE, + iod); + nvme_free_iod(dev, iod); + } + res = nvme_trans_status_code(hdr, NVME_SC_SUCCESS); + + out: + return res; +} + + +/* SCSI Command Translation Functions */ + +static int nvme_trans_io(struct nvme_ns *ns, struct sg_io_hdr *hdr, u8 is_write, + u8 *cmd) +{ + int res = SNTI_TRANSLATION_SUCCESS; + struct nvme_trans_io_cdb cdb_info; + u8 opcode = cmd[0]; + u64 xfer_bytes; + u64 sum_iov_len = 0; + struct sg_iovec sgl; + int i; + size_t not_copied; + + /* Extract Fields from CDB */ + switch (opcode) { + case WRITE_6: + case READ_6: + nvme_trans_get_io_cdb6(cmd, &cdb_info); + break; + case WRITE_10: + case READ_10: + nvme_trans_get_io_cdb10(cmd, &cdb_info); + break; + case WRITE_12: + case READ_12: + nvme_trans_get_io_cdb12(cmd, &cdb_info); + break; + case WRITE_16: + case READ_16: + nvme_trans_get_io_cdb16(cmd, &cdb_info); + break; + default: + /* Will never really reach here */ + res = SNTI_INTERNAL_ERROR; + goto out; + } + + /* Calculate total length of transfer (in bytes) */ + if (hdr->iovec_count > 0) { + for (i = 0; i < hdr->iovec_count; i++) { + not_copied = copy_from_user(&sgl, hdr->dxferp + + i * sizeof(struct sg_iovec), + sizeof(struct sg_iovec)); + if (not_copied) + return -EFAULT; + sum_iov_len += sgl.iov_len; + /* IO vector sizes should be multiples of block size */ + if (sgl.iov_len % (1 << ns->lba_shift) != 0) { + res = nvme_trans_completion(hdr, + SAM_STAT_CHECK_CONDITION, + ILLEGAL_REQUEST, + SCSI_ASC_INVALID_PARAMETER, + SCSI_ASCQ_CAUSE_NOT_REPORTABLE); + goto out; + } + } + } else { + sum_iov_len = hdr->dxfer_len; + } + + /* As Per sg ioctl howto, if the lengths differ, use the lower one */ + xfer_bytes = min(((u64)hdr->dxfer_len), sum_iov_len); + + /* If block count and actual data buffer size dont match, error out */ + if (xfer_bytes != (cdb_info.xfer_len << ns->lba_shift)) { + res = -EINVAL; + goto out; + } + + /* Check for 0 length transfer - it is not illegal */ + if (cdb_info.xfer_len == 0) + goto out; + + /* Send NVMe IO Command(s) */ + res = nvme_trans_do_nvme_io(ns, hdr, &cdb_info, is_write); + if (res != SNTI_TRANSLATION_SUCCESS) + goto out; + + out: + return res; +} + +static int nvme_trans_inquiry(struct nvme_ns *ns, struct sg_io_hdr *hdr, + u8 *cmd) +{ + int res = SNTI_TRANSLATION_SUCCESS; + u8 evpd; + u8 page_code; + int alloc_len; + u8 *inq_response; + + evpd = GET_INQ_EVPD_BIT(cmd); + page_code = GET_INQ_PAGE_CODE(cmd); + alloc_len = GET_INQ_ALLOC_LENGTH(cmd); + + inq_response = kmalloc(STANDARD_INQUIRY_LENGTH, GFP_KERNEL); + if (inq_response == NULL) { + res = -ENOMEM; + goto out_mem; + } + + if (evpd == 0) { + if (page_code == INQ_STANDARD_INQUIRY_PAGE) { + res = nvme_trans_standard_inquiry_page(ns, hdr, + inq_response, alloc_len); + } else { + res = nvme_trans_completion(hdr, + SAM_STAT_CHECK_CONDITION, + ILLEGAL_REQUEST, + SCSI_ASC_INVALID_CDB, + SCSI_ASCQ_CAUSE_NOT_REPORTABLE); + } + } else { + switch (page_code) { + case VPD_SUPPORTED_PAGES: + res = nvme_trans_supported_vpd_pages(ns, hdr, + inq_response, alloc_len); + break; + case VPD_SERIAL_NUMBER: + res = nvme_trans_unit_serial_page(ns, hdr, inq_response, + alloc_len); + break; + case VPD_DEVICE_IDENTIFIERS: + res = nvme_trans_device_id_page(ns, hdr, inq_response, + alloc_len); + break; + case VPD_EXTENDED_INQUIRY: + res = nvme_trans_ext_inq_page(ns, hdr, alloc_len); + break; + case VPD_BLOCK_DEV_CHARACTERISTICS: + res = nvme_trans_bdev_char_page(ns, hdr, alloc_len); + break; + default: + res = nvme_trans_completion(hdr, + SAM_STAT_CHECK_CONDITION, + ILLEGAL_REQUEST, + SCSI_ASC_INVALID_CDB, + SCSI_ASCQ_CAUSE_NOT_REPORTABLE); + break; + } + } + kfree(inq_response); + out_mem: + return res; +} + +static int nvme_trans_log_sense(struct nvme_ns *ns, struct sg_io_hdr *hdr, + u8 *cmd) +{ + int res = SNTI_TRANSLATION_SUCCESS; + u16 alloc_len; + u8 sp; + u8 pc; + u8 page_code; + + sp = GET_U8_FROM_CDB(cmd, LOG_SENSE_CDB_SP_OFFSET); + if (sp != LOG_SENSE_CDB_SP_NOT_ENABLED) { + res = nvme_trans_completion(hdr, SAM_STAT_CHECK_CONDITION, + ILLEGAL_REQUEST, SCSI_ASC_INVALID_CDB, + SCSI_ASCQ_CAUSE_NOT_REPORTABLE); + goto out; + } + pc = GET_U8_FROM_CDB(cmd, LOG_SENSE_CDB_PC_OFFSET); + page_code = pc & LOG_SENSE_CDB_PAGE_CODE_MASK; + pc = (pc & LOG_SENSE_CDB_PC_MASK) >> LOG_SENSE_CDB_PC_SHIFT; + if (pc != LOG_SENSE_CDB_PC_CUMULATIVE_VALUES) { + res = nvme_trans_completion(hdr, SAM_STAT_CHECK_CONDITION, + ILLEGAL_REQUEST, SCSI_ASC_INVALID_CDB, + SCSI_ASCQ_CAUSE_NOT_REPORTABLE); + goto out; + } + alloc_len = GET_U16_FROM_CDB(cmd, LOG_SENSE_CDB_ALLOC_LENGTH_OFFSET); + switch (page_code) { + case LOG_PAGE_SUPPORTED_LOG_PAGES_PAGE: + res = nvme_trans_log_supp_pages(ns, hdr, alloc_len); + break; + case LOG_PAGE_INFORMATIONAL_EXCEPTIONS_PAGE: + res = nvme_trans_log_info_exceptions(ns, hdr, alloc_len); + break; + case LOG_PAGE_TEMPERATURE_PAGE: + res = nvme_trans_log_temperature(ns, hdr, alloc_len); + break; + default: + res = nvme_trans_completion(hdr, SAM_STAT_CHECK_CONDITION, + ILLEGAL_REQUEST, SCSI_ASC_INVALID_CDB, + SCSI_ASCQ_CAUSE_NOT_REPORTABLE); + break; + } + + out: + return res; +} + +static int nvme_trans_mode_select(struct nvme_ns *ns, struct sg_io_hdr *hdr, + u8 *cmd) +{ + int res = SNTI_TRANSLATION_SUCCESS; + u8 cdb10 = 0; + u16 parm_list_len; + u8 page_format; + u8 save_pages; + + page_format = GET_U8_FROM_CDB(cmd, MODE_SELECT_CDB_PAGE_FORMAT_OFFSET); + page_format &= MODE_SELECT_CDB_PAGE_FORMAT_MASK; + + save_pages = GET_U8_FROM_CDB(cmd, MODE_SELECT_CDB_SAVE_PAGES_OFFSET); + save_pages &= MODE_SELECT_CDB_SAVE_PAGES_MASK; + + if (GET_OPCODE(cmd) == MODE_SELECT) { + parm_list_len = GET_U8_FROM_CDB(cmd, + MODE_SELECT_6_CDB_PARAM_LIST_LENGTH_OFFSET); + } else { + parm_list_len = GET_U16_FROM_CDB(cmd, + MODE_SELECT_10_CDB_PARAM_LIST_LENGTH_OFFSET); + cdb10 = 1; + } + + if (parm_list_len != 0) { + /* + * According to SPC-4 r24, a paramter list length field of 0 + * shall not be considered an error + */ + res = nvme_trans_modesel_data(ns, hdr, cmd, parm_list_len, + page_format, save_pages, cdb10); + } + + return res; +} + +static int nvme_trans_mode_sense(struct nvme_ns *ns, struct sg_io_hdr *hdr, + u8 *cmd) +{ + int res = SNTI_TRANSLATION_SUCCESS; + u16 alloc_len; + u8 cdb10 = 0; + u8 page_code; + u8 pc; + + if (GET_OPCODE(cmd) == MODE_SENSE) { + alloc_len = GET_U8_FROM_CDB(cmd, MODE_SENSE6_ALLOC_LEN_OFFSET); + } else { + alloc_len = GET_U16_FROM_CDB(cmd, + MODE_SENSE10_ALLOC_LEN_OFFSET); + cdb10 = 1; + } + + pc = GET_U8_FROM_CDB(cmd, MODE_SENSE_PAGE_CONTROL_OFFSET) & + MODE_SENSE_PAGE_CONTROL_MASK; + if (pc != MODE_SENSE_PC_CURRENT_VALUES) { + res = nvme_trans_completion(hdr, SAM_STAT_CHECK_CONDITION, + ILLEGAL_REQUEST, SCSI_ASC_INVALID_CDB, + SCSI_ASCQ_CAUSE_NOT_REPORTABLE); + goto out; + } + + page_code = GET_U8_FROM_CDB(cmd, MODE_SENSE_PAGE_CODE_OFFSET) & + MODE_SENSE_PAGE_CODE_MASK; + switch (page_code) { + case MODE_PAGE_CACHING: + res = nvme_trans_mode_page_create(ns, hdr, cmd, alloc_len, + cdb10, + &nvme_trans_fill_caching_page, + MODE_PAGE_CACHING_LEN); + break; + case MODE_PAGE_CONTROL: + res = nvme_trans_mode_page_create(ns, hdr, cmd, alloc_len, + cdb10, + &nvme_trans_fill_control_page, + MODE_PAGE_CONTROL_LEN); + break; + case MODE_PAGE_POWER_CONDITION: + res = nvme_trans_mode_page_create(ns, hdr, cmd, alloc_len, + cdb10, + &nvme_trans_fill_pow_cnd_page, + MODE_PAGE_POW_CND_LEN); + break; + case MODE_PAGE_INFO_EXCEP: + res = nvme_trans_mode_page_create(ns, hdr, cmd, alloc_len, + cdb10, + &nvme_trans_fill_inf_exc_page, + MODE_PAGE_INF_EXC_LEN); + break; + case MODE_PAGE_RETURN_ALL: + res = nvme_trans_mode_page_create(ns, hdr, cmd, alloc_len, + cdb10, + &nvme_trans_fill_all_pages, + MODE_PAGE_ALL_LEN); + break; + default: + res = nvme_trans_completion(hdr, SAM_STAT_CHECK_CONDITION, + ILLEGAL_REQUEST, SCSI_ASC_INVALID_CDB, + SCSI_ASCQ_CAUSE_NOT_REPORTABLE); + break; + } + + out: + return res; +} + +static int nvme_trans_read_capacity(struct nvme_ns *ns, struct sg_io_hdr *hdr, + u8 *cmd) +{ + int res = SNTI_TRANSLATION_SUCCESS; + int nvme_sc; + u32 alloc_len = READ_CAP_10_RESP_SIZE; + u32 resp_size = READ_CAP_10_RESP_SIZE; + u32 xfer_len; + u8 cdb16; + struct nvme_dev *dev = ns->dev; + dma_addr_t dma_addr; + void *mem; + struct nvme_id_ns *id_ns; + u8 *response; + + cdb16 = IS_READ_CAP_16(cmd); + if (cdb16) { + alloc_len = GET_READ_CAP_16_ALLOC_LENGTH(cmd); + resp_size = READ_CAP_16_RESP_SIZE; + } + + mem = dma_alloc_coherent(&dev->pci_dev->dev, sizeof(struct nvme_id_ns), + &dma_addr, GFP_KERNEL); + if (mem == NULL) { + res = -ENOMEM; + goto out; + } + /* nvme ns identify */ + nvme_sc = nvme_identify(dev, ns->ns_id, 0, dma_addr); + res = nvme_trans_status_code(hdr, nvme_sc); + if (res) + goto out_dma; + if (nvme_sc) { + res = nvme_sc; + goto out_dma; + } + id_ns = mem; + + response = kmalloc(resp_size, GFP_KERNEL); + if (response == NULL) { + res = -ENOMEM; + goto out_dma; + } + memset(response, 0, resp_size); + nvme_trans_fill_read_cap(response, id_ns, cdb16); + + xfer_len = min(alloc_len, resp_size); + res = nvme_trans_copy_to_user(hdr, response, xfer_len); + + kfree(response); + out_dma: + dma_free_coherent(&dev->pci_dev->dev, sizeof(struct nvme_id_ns), mem, + dma_addr); + out: + return res; +} + +static int nvme_trans_report_luns(struct nvme_ns *ns, struct sg_io_hdr *hdr, + u8 *cmd) +{ + int res = SNTI_TRANSLATION_SUCCESS; + int nvme_sc; + u32 alloc_len, xfer_len, resp_size; + u8 select_report; + u8 *response; + struct nvme_dev *dev = ns->dev; + dma_addr_t dma_addr; + void *mem; + struct nvme_id_ctrl *id_ctrl; + u32 ll_length, lun_id; + u8 lun_id_offset = REPORT_LUNS_FIRST_LUN_OFFSET; + __be32 tmp_len; + + alloc_len = GET_REPORT_LUNS_ALLOC_LENGTH(cmd); + select_report = GET_U8_FROM_CDB(cmd, REPORT_LUNS_SR_OFFSET); + + if ((select_report != ALL_LUNS_RETURNED) && + (select_report != ALL_WELL_KNOWN_LUNS_RETURNED) && + (select_report != RESTRICTED_LUNS_RETURNED)) { + res = nvme_trans_completion(hdr, SAM_STAT_CHECK_CONDITION, + ILLEGAL_REQUEST, SCSI_ASC_INVALID_CDB, + SCSI_ASCQ_CAUSE_NOT_REPORTABLE); + goto out; + } else { + /* NVMe Controller Identify */ + mem = dma_alloc_coherent(&dev->pci_dev->dev, + sizeof(struct nvme_id_ctrl), + &dma_addr, GFP_KERNEL); + if (mem == NULL) { + res = -ENOMEM; + goto out; + } + nvme_sc = nvme_identify(dev, 0, 1, dma_addr); + res = nvme_trans_status_code(hdr, nvme_sc); + if (res) + goto out_dma; + if (nvme_sc) { + res = nvme_sc; + goto out_dma; + } + id_ctrl = mem; + ll_length = le32_to_cpu(id_ctrl->nn) * LUN_ENTRY_SIZE; + resp_size = ll_length + LUN_DATA_HEADER_SIZE; + + if (alloc_len < resp_size) { + res = nvme_trans_completion(hdr, + SAM_STAT_CHECK_CONDITION, + ILLEGAL_REQUEST, SCSI_ASC_INVALID_CDB, + SCSI_ASCQ_CAUSE_NOT_REPORTABLE); + goto out_dma; + } + + response = kmalloc(resp_size, GFP_KERNEL); + if (response == NULL) { + res = -ENOMEM; + goto out_dma; + } + memset(response, 0, resp_size); + + /* The first LUN ID will always be 0 per the SAM spec */ + for (lun_id = 0; lun_id < le32_to_cpu(id_ctrl->nn); lun_id++) { + /* + * Set the LUN Id and then increment to the next LUN + * location in the parameter data. + */ + __be64 tmp_id = cpu_to_be64(lun_id); + memcpy(&response[lun_id_offset], &tmp_id, sizeof(u64)); + lun_id_offset += LUN_ENTRY_SIZE; + } + tmp_len = cpu_to_be32(ll_length); + memcpy(response, &tmp_len, sizeof(u32)); + } + + xfer_len = min(alloc_len, resp_size); + res = nvme_trans_copy_to_user(hdr, response, xfer_len); + + kfree(response); + out_dma: + dma_free_coherent(&dev->pci_dev->dev, sizeof(struct nvme_id_ctrl), mem, + dma_addr); + out: + return res; +} + +static int nvme_trans_request_sense(struct nvme_ns *ns, struct sg_io_hdr *hdr, + u8 *cmd) +{ + int res = SNTI_TRANSLATION_SUCCESS; + u8 alloc_len, xfer_len, resp_size; + u8 desc_format; + u8 *response; + + alloc_len = GET_REQUEST_SENSE_ALLOC_LENGTH(cmd); + desc_format = GET_U8_FROM_CDB(cmd, REQUEST_SENSE_DESC_OFFSET); + desc_format &= REQUEST_SENSE_DESC_MASK; + + resp_size = ((desc_format) ? (DESC_FMT_SENSE_DATA_SIZE) : + (FIXED_FMT_SENSE_DATA_SIZE)); + response = kmalloc(resp_size, GFP_KERNEL); + if (response == NULL) { + res = -ENOMEM; + goto out; + } + memset(response, 0, resp_size); + + if (desc_format == DESCRIPTOR_FORMAT_SENSE_DATA_TYPE) { + /* Descriptor Format Sense Data */ + response[0] = DESC_FORMAT_SENSE_DATA; + response[1] = NO_SENSE; + /* TODO How is LOW POWER CONDITION ON handled? (byte 2) */ + response[2] = SCSI_ASC_NO_SENSE; + response[3] = SCSI_ASCQ_CAUSE_NOT_REPORTABLE; + /* SDAT_OVFL = 0 | Additional Sense Length = 0 */ + } else { + /* Fixed Format Sense Data */ + response[0] = FIXED_SENSE_DATA; + /* Byte 1 = Obsolete */ + response[2] = NO_SENSE; /* FM, EOM, ILI, SDAT_OVFL = 0 */ + /* Bytes 3-6 - Information - set to zero */ + response[7] = FIXED_SENSE_DATA_ADD_LENGTH; + /* Bytes 8-11 - Cmd Specific Information - set to zero */ + response[12] = SCSI_ASC_NO_SENSE; + response[13] = SCSI_ASCQ_CAUSE_NOT_REPORTABLE; + /* Byte 14 = Field Replaceable Unit Code = 0 */ + /* Bytes 15-17 - SKSV=0; Sense Key Specific = 0 */ + } + + xfer_len = min(alloc_len, resp_size); + res = nvme_trans_copy_to_user(hdr, response, xfer_len); + + kfree(response); + out: + return res; +} + +static int nvme_trans_security_protocol(struct nvme_ns *ns, + struct sg_io_hdr *hdr, + u8 *cmd) +{ + return nvme_trans_completion(hdr, SAM_STAT_CHECK_CONDITION, + ILLEGAL_REQUEST, SCSI_ASC_ILLEGAL_COMMAND, + SCSI_ASCQ_CAUSE_NOT_REPORTABLE); +} + +static int nvme_trans_start_stop(struct nvme_ns *ns, struct sg_io_hdr *hdr, + u8 *cmd) +{ + int res = SNTI_TRANSLATION_SUCCESS; + int nvme_sc; + struct nvme_queue *nvmeq; + struct nvme_command c; + u8 immed, pcmod, pc, no_flush, start; + + immed = GET_U8_FROM_CDB(cmd, START_STOP_UNIT_CDB_IMMED_OFFSET); + pcmod = GET_U8_FROM_CDB(cmd, START_STOP_UNIT_CDB_POWER_COND_MOD_OFFSET); + pc = GET_U8_FROM_CDB(cmd, START_STOP_UNIT_CDB_POWER_COND_OFFSET); + no_flush = GET_U8_FROM_CDB(cmd, START_STOP_UNIT_CDB_NO_FLUSH_OFFSET); + start = GET_U8_FROM_CDB(cmd, START_STOP_UNIT_CDB_START_OFFSET); + + immed &= START_STOP_UNIT_CDB_IMMED_MASK; + pcmod &= START_STOP_UNIT_CDB_POWER_COND_MOD_MASK; + pc = (pc & START_STOP_UNIT_CDB_POWER_COND_MASK) >> NIBBLE_SHIFT; + no_flush &= START_STOP_UNIT_CDB_NO_FLUSH_MASK; + start &= START_STOP_UNIT_CDB_START_MASK; + + if (immed != 0) { + res = nvme_trans_completion(hdr, SAM_STAT_CHECK_CONDITION, + ILLEGAL_REQUEST, SCSI_ASC_INVALID_CDB, + SCSI_ASCQ_CAUSE_NOT_REPORTABLE); + } else { + if (no_flush == 0) { + /* Issue NVME FLUSH command prior to START STOP UNIT */ + memset(&c, 0, sizeof(c)); + c.common.opcode = nvme_cmd_flush; + c.common.nsid = cpu_to_le32(ns->ns_id); + + nvmeq = get_nvmeq(ns->dev); + put_nvmeq(nvmeq); + nvme_sc = nvme_submit_sync_cmd(nvmeq, &c, NULL, NVME_IO_TIMEOUT); + + res = nvme_trans_status_code(hdr, nvme_sc); + if (res) + goto out; + if (nvme_sc) { + res = nvme_sc; + goto out; + } + } + /* Setup the expected power state transition */ + res = nvme_trans_power_state(ns, hdr, pc, pcmod, start); + } + + out: + return res; +} + +static int nvme_trans_synchronize_cache(struct nvme_ns *ns, + struct sg_io_hdr *hdr, u8 *cmd) +{ + int res = SNTI_TRANSLATION_SUCCESS; + int nvme_sc; + struct nvme_command c; + struct nvme_queue *nvmeq; + + memset(&c, 0, sizeof(c)); + c.common.opcode = nvme_cmd_flush; + c.common.nsid = cpu_to_le32(ns->ns_id); + + nvmeq = get_nvmeq(ns->dev); + put_nvmeq(nvmeq); + nvme_sc = nvme_submit_sync_cmd(nvmeq, &c, NULL, NVME_IO_TIMEOUT); + + res = nvme_trans_status_code(hdr, nvme_sc); + if (res) + goto out; + if (nvme_sc) + res = nvme_sc; + + out: + return res; +} + +static int nvme_trans_format_unit(struct nvme_ns *ns, struct sg_io_hdr *hdr, + u8 *cmd) +{ + int res = SNTI_TRANSLATION_SUCCESS; + u8 parm_hdr_len = 0; + u8 nvme_pf_code = 0; + u8 format_prot_info, long_list, format_data; + + format_prot_info = GET_U8_FROM_CDB(cmd, + FORMAT_UNIT_CDB_FORMAT_PROT_INFO_OFFSET); + long_list = GET_U8_FROM_CDB(cmd, FORMAT_UNIT_CDB_LONG_LIST_OFFSET); + format_data = GET_U8_FROM_CDB(cmd, FORMAT_UNIT_CDB_FORMAT_DATA_OFFSET); + + format_prot_info = (format_prot_info & + FORMAT_UNIT_CDB_FORMAT_PROT_INFO_MASK) >> + FORMAT_UNIT_CDB_FORMAT_PROT_INFO_SHIFT; + long_list &= FORMAT_UNIT_CDB_LONG_LIST_MASK; + format_data &= FORMAT_UNIT_CDB_FORMAT_DATA_MASK; + + if (format_data != 0) { + if (format_prot_info != 0) { + if (long_list == 0) + parm_hdr_len = FORMAT_UNIT_SHORT_PARM_LIST_LEN; + else + parm_hdr_len = FORMAT_UNIT_LONG_PARM_LIST_LEN; + } + } else if (format_data == 0 && format_prot_info != 0) { + res = nvme_trans_completion(hdr, SAM_STAT_CHECK_CONDITION, + ILLEGAL_REQUEST, SCSI_ASC_INVALID_CDB, + SCSI_ASCQ_CAUSE_NOT_REPORTABLE); + goto out; + } + + /* Get parm header from data-in/out buffer */ + /* + * According to the translation spec, the only fields in the parameter + * list we are concerned with are in the header. So allocate only that. + */ + if (parm_hdr_len > 0) { + res = nvme_trans_fmt_get_parm_header(hdr, parm_hdr_len, + format_prot_info, &nvme_pf_code); + if (res != SNTI_TRANSLATION_SUCCESS) + goto out; + } + + /* Attempt to activate any previously downloaded firmware image */ + res = nvme_trans_send_fw_cmd(ns, hdr, nvme_admin_activate_fw, 0, 0, 0); + + /* Determine Block size and count and send format command */ + res = nvme_trans_fmt_set_blk_size_count(ns, hdr); + if (res != SNTI_TRANSLATION_SUCCESS) + goto out; + + res = nvme_trans_fmt_send_cmd(ns, hdr, nvme_pf_code); + + out: + return res; +} + +static int nvme_trans_test_unit_ready(struct nvme_ns *ns, + struct sg_io_hdr *hdr, + u8 *cmd) +{ + int res = SNTI_TRANSLATION_SUCCESS; + struct nvme_dev *dev = ns->dev; + + if (!(readl(&dev->bar->csts) & NVME_CSTS_RDY)) + res = nvme_trans_completion(hdr, SAM_STAT_CHECK_CONDITION, + NOT_READY, SCSI_ASC_LUN_NOT_READY, + SCSI_ASCQ_CAUSE_NOT_REPORTABLE); + else + res = nvme_trans_completion(hdr, SAM_STAT_GOOD, NO_SENSE, 0, 0); + + return res; +} + +static int nvme_trans_write_buffer(struct nvme_ns *ns, struct sg_io_hdr *hdr, + u8 *cmd) +{ + int res = SNTI_TRANSLATION_SUCCESS; + u32 buffer_offset, parm_list_length; + u8 buffer_id, mode; + + parm_list_length = + GET_U24_FROM_CDB(cmd, WRITE_BUFFER_CDB_PARM_LIST_LENGTH_OFFSET); + if (parm_list_length % BYTES_TO_DWORDS != 0) { + /* NVMe expects Firmware file to be a whole number of DWORDS */ + res = nvme_trans_completion(hdr, SAM_STAT_CHECK_CONDITION, + ILLEGAL_REQUEST, SCSI_ASC_INVALID_CDB, + SCSI_ASCQ_CAUSE_NOT_REPORTABLE); + goto out; + } + buffer_id = GET_U8_FROM_CDB(cmd, WRITE_BUFFER_CDB_BUFFER_ID_OFFSET); + if (buffer_id > NVME_MAX_FIRMWARE_SLOT) { + res = nvme_trans_completion(hdr, SAM_STAT_CHECK_CONDITION, + ILLEGAL_REQUEST, SCSI_ASC_INVALID_CDB, + SCSI_ASCQ_CAUSE_NOT_REPORTABLE); + goto out; + } + mode = GET_U8_FROM_CDB(cmd, WRITE_BUFFER_CDB_MODE_OFFSET) & + WRITE_BUFFER_CDB_MODE_MASK; + buffer_offset = + GET_U24_FROM_CDB(cmd, WRITE_BUFFER_CDB_BUFFER_OFFSET_OFFSET); + + switch (mode) { + case DOWNLOAD_SAVE_ACTIVATE: + res = nvme_trans_send_fw_cmd(ns, hdr, nvme_admin_download_fw, + parm_list_length, buffer_offset, + buffer_id); + if (res != SNTI_TRANSLATION_SUCCESS) + goto out; + res = nvme_trans_send_fw_cmd(ns, hdr, nvme_admin_activate_fw, + parm_list_length, buffer_offset, + buffer_id); + break; + case DOWNLOAD_SAVE_DEFER_ACTIVATE: + res = nvme_trans_send_fw_cmd(ns, hdr, nvme_admin_download_fw, + parm_list_length, buffer_offset, + buffer_id); + break; + case ACTIVATE_DEFERRED_MICROCODE: + res = nvme_trans_send_fw_cmd(ns, hdr, nvme_admin_activate_fw, + parm_list_length, buffer_offset, + buffer_id); + break; + default: + res = nvme_trans_completion(hdr, SAM_STAT_CHECK_CONDITION, + ILLEGAL_REQUEST, SCSI_ASC_INVALID_CDB, + SCSI_ASCQ_CAUSE_NOT_REPORTABLE); + break; + } + + out: + return res; +} + +struct scsi_unmap_blk_desc { + __be64 slba; + __be32 nlb; + u32 resv; +}; + +struct scsi_unmap_parm_list { + __be16 unmap_data_len; + __be16 unmap_blk_desc_data_len; + u32 resv; + struct scsi_unmap_blk_desc desc[0]; +}; + +static int nvme_trans_unmap(struct nvme_ns *ns, struct sg_io_hdr *hdr, + u8 *cmd) +{ + struct nvme_dev *dev = ns->dev; + struct scsi_unmap_parm_list *plist; + struct nvme_dsm_range *range; + struct nvme_queue *nvmeq; + struct nvme_command c; + int i, nvme_sc, res = -ENOMEM; + u16 ndesc, list_len; + dma_addr_t dma_addr; + + list_len = GET_U16_FROM_CDB(cmd, UNMAP_CDB_PARAM_LIST_LENGTH_OFFSET); + if (!list_len) + return -EINVAL; + + plist = kmalloc(list_len, GFP_KERNEL); + if (!plist) + return -ENOMEM; + + res = nvme_trans_copy_from_user(hdr, plist, list_len); + if (res != SNTI_TRANSLATION_SUCCESS) + goto out; + + ndesc = be16_to_cpu(plist->unmap_blk_desc_data_len) >> 4; + if (!ndesc || ndesc > 256) { + res = -EINVAL; + goto out; + } + + range = dma_alloc_coherent(&dev->pci_dev->dev, ndesc * sizeof(*range), + &dma_addr, GFP_KERNEL); + if (!range) + goto out; + + for (i = 0; i < ndesc; i++) { + range[i].nlb = cpu_to_le32(be32_to_cpu(plist->desc[i].nlb)); + range[i].slba = cpu_to_le64(be64_to_cpu(plist->desc[i].slba)); + range[i].cattr = 0; + } + + memset(&c, 0, sizeof(c)); + c.dsm.opcode = nvme_cmd_dsm; + c.dsm.nsid = cpu_to_le32(ns->ns_id); + c.dsm.prp1 = cpu_to_le64(dma_addr); + c.dsm.nr = cpu_to_le32(ndesc - 1); + c.dsm.attributes = cpu_to_le32(NVME_DSMGMT_AD); + + nvmeq = get_nvmeq(dev); + put_nvmeq(nvmeq); + + nvme_sc = nvme_submit_sync_cmd(nvmeq, &c, NULL, NVME_IO_TIMEOUT); + res = nvme_trans_status_code(hdr, nvme_sc); + + dma_free_coherent(&dev->pci_dev->dev, ndesc * sizeof(*range), + range, dma_addr); + out: + kfree(plist); + return res; +} + +static int nvme_scsi_translate(struct nvme_ns *ns, struct sg_io_hdr *hdr) +{ + u8 cmd[BLK_MAX_CDB]; + int retcode; + unsigned int opcode; + + if (hdr->cmdp == NULL) + return -EMSGSIZE; + if (copy_from_user(cmd, hdr->cmdp, hdr->cmd_len)) + return -EFAULT; + + opcode = cmd[0]; + + switch (opcode) { + case READ_6: + case READ_10: + case READ_12: + case READ_16: + retcode = nvme_trans_io(ns, hdr, 0, cmd); + break; + case WRITE_6: + case WRITE_10: + case WRITE_12: + case WRITE_16: + retcode = nvme_trans_io(ns, hdr, 1, cmd); + break; + case INQUIRY: + retcode = nvme_trans_inquiry(ns, hdr, cmd); + break; + case LOG_SENSE: + retcode = nvme_trans_log_sense(ns, hdr, cmd); + break; + case MODE_SELECT: + case MODE_SELECT_10: + retcode = nvme_trans_mode_select(ns, hdr, cmd); + break; + case MODE_SENSE: + case MODE_SENSE_10: + retcode = nvme_trans_mode_sense(ns, hdr, cmd); + break; + case READ_CAPACITY: + retcode = nvme_trans_read_capacity(ns, hdr, cmd); + break; + case SERVICE_ACTION_IN: + if (IS_READ_CAP_16(cmd)) + retcode = nvme_trans_read_capacity(ns, hdr, cmd); + else + goto out; + break; + case REPORT_LUNS: + retcode = nvme_trans_report_luns(ns, hdr, cmd); + break; + case REQUEST_SENSE: + retcode = nvme_trans_request_sense(ns, hdr, cmd); + break; + case SECURITY_PROTOCOL_IN: + case SECURITY_PROTOCOL_OUT: + retcode = nvme_trans_security_protocol(ns, hdr, cmd); + break; + case START_STOP: + retcode = nvme_trans_start_stop(ns, hdr, cmd); + break; + case SYNCHRONIZE_CACHE: + retcode = nvme_trans_synchronize_cache(ns, hdr, cmd); + break; + case FORMAT_UNIT: + retcode = nvme_trans_format_unit(ns, hdr, cmd); + break; + case TEST_UNIT_READY: + retcode = nvme_trans_test_unit_ready(ns, hdr, cmd); + break; + case WRITE_BUFFER: + retcode = nvme_trans_write_buffer(ns, hdr, cmd); + break; + case UNMAP: + retcode = nvme_trans_unmap(ns, hdr, cmd); + break; + default: + out: + retcode = nvme_trans_completion(hdr, SAM_STAT_CHECK_CONDITION, + ILLEGAL_REQUEST, SCSI_ASC_ILLEGAL_COMMAND, + SCSI_ASCQ_CAUSE_NOT_REPORTABLE); + break; + } + return retcode; +} + +int nvme_sg_io(struct nvme_ns *ns, struct sg_io_hdr __user *u_hdr) +{ + struct sg_io_hdr hdr; + int retcode; + + if (!capable(CAP_SYS_ADMIN)) + return -EACCES; + if (copy_from_user(&hdr, u_hdr, sizeof(hdr))) + return -EFAULT; + if (hdr.interface_id != 'S') + return -EINVAL; + if (hdr.cmd_len > BLK_MAX_CDB) + return -EINVAL; + + retcode = nvme_scsi_translate(ns, &hdr); + if (retcode < 0) + return retcode; + if (retcode > 0) + retcode = SNTI_TRANSLATION_SUCCESS; + if (copy_to_user(u_hdr, &hdr, sizeof(sg_io_hdr_t)) > 0) + return -EFAULT; + + return retcode; +} + +int nvme_sg_get_version_num(int __user *ip) +{ + return put_user(sg_version_num, ip); +} diff --git a/include/linux/nvme.h b/include/linux/nvme.h index 4fa3b0b9b071..f451c8d6e231 100644 --- a/include/linux/nvme.h +++ b/include/linux/nvme.h @@ -107,6 +107,12 @@ struct nvme_id_ctrl { __u8 vs[1024]; }; +enum { + NVME_CTRL_ONCS_COMPARE = 1 << 0, + NVME_CTRL_ONCS_WRITE_UNCORRECTABLE = 1 << 1, + NVME_CTRL_ONCS_DSM = 1 << 2, +}; + struct nvme_lbaf { __le16 ms; __u8 ds; @@ -201,11 +207,11 @@ struct nvme_common_command { __u8 flags; __u16 command_id; __le32 nsid; - __u32 cdw2[2]; + __le32 cdw2[2]; __le64 metadata; __le64 prp1; __le64 prp2; - __u32 cdw10[6]; + __le32 cdw10[6]; }; struct nvme_rw_command { @@ -246,6 +252,31 @@ enum { NVME_RW_DSM_COMPRESSED = 1 << 7, }; +struct nvme_dsm_cmd { + __u8 opcode; + __u8 flags; + __u16 command_id; + __le32 nsid; + __u64 rsvd2[2]; + __le64 prp1; + __le64 prp2; + __le32 nr; + __le32 attributes; + __u32 rsvd12[4]; +}; + +enum { + NVME_DSMGMT_IDR = 1 << 0, + NVME_DSMGMT_IDW = 1 << 1, + NVME_DSMGMT_AD = 1 << 2, +}; + +struct nvme_dsm_range { + __le32 cattr; + __le32 nlb; + __le64 slba; +}; + /* Admin commands */ enum nvme_admin_opcode { @@ -285,6 +316,9 @@ enum { NVME_FEAT_WRITE_ATOMIC = 0x0a, NVME_FEAT_ASYNC_EVENT = 0x0b, NVME_FEAT_SW_PROGRESS = 0x0c, + NVME_FWACT_REPL = (0 << 3), + NVME_FWACT_REPL_ACTV = (1 << 3), + NVME_FWACT_ACTV = (2 << 3), }; struct nvme_identify { @@ -362,6 +396,16 @@ struct nvme_download_firmware { __u32 rsvd12[4]; }; +struct nvme_format_cmd { + __u8 opcode; + __u8 flags; + __u16 command_id; + __le32 nsid; + __u64 rsvd2[4]; + __le32 cdw10; + __u32 rsvd11[5]; +}; + struct nvme_command { union { struct nvme_common_command common; @@ -372,6 +416,8 @@ struct nvme_command { struct nvme_create_sq create_sq; struct nvme_delete_queue delete_queue; struct nvme_download_firmware dlfw; + struct nvme_format_cmd format; + struct nvme_dsm_cmd dsm; }; }; @@ -388,6 +434,7 @@ enum { NVME_SC_FUSED_FAIL = 0x9, NVME_SC_FUSED_MISSING = 0xa, NVME_SC_INVALID_NS = 0xb, + NVME_SC_CMD_SEQ_ERROR = 0xc, NVME_SC_LBA_RANGE = 0x80, NVME_SC_CAP_EXCEEDED = 0x81, NVME_SC_NS_NOT_READY = 0x82, @@ -461,4 +508,111 @@ struct nvme_admin_cmd { #define NVME_IOCTL_ADMIN_CMD _IOWR('N', 0x41, struct nvme_admin_cmd) #define NVME_IOCTL_SUBMIT_IO _IOW('N', 0x42, struct nvme_user_io) +#ifdef __KERNEL__ +#include <linux/pci.h> +#include <linux/miscdevice.h> +#include <linux/kref.h> + +#define NVME_IO_TIMEOUT (5 * HZ) + +/* + * 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; + struct kref kref; + struct miscdevice miscdev; + char name[12]; + char serial[20]; + char model[40]; + char firmware_rev[8]; + u32 max_hw_sectors; + u32 stripe_size; + u16 oncs; +}; + +/* + * 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; + int ms; + u64 mode_select_num_blocks; + u32 mode_select_block_len; +}; + +/* + * 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 inline u64 nvme_block_nr(struct nvme_ns *ns, sector_t sector) +{ + return (sector >> (ns->lba_shift - 9)); +} + +/** + * nvme_free_iod - frees an nvme_iod + * @dev: The device that the I/O was submitted to + * @iod: The memory to free + */ +void nvme_free_iod(struct nvme_dev *dev, struct nvme_iod *iod); + +int nvme_setup_prps(struct nvme_dev *dev, struct nvme_common_command *cmd, + struct nvme_iod *iod, int total_len, gfp_t gfp); +struct nvme_iod *nvme_map_user_pages(struct nvme_dev *dev, int write, + unsigned long addr, unsigned length); +void nvme_unmap_user_pages(struct nvme_dev *dev, int write, + struct nvme_iod *iod); +struct nvme_queue *get_nvmeq(struct nvme_dev *dev); +void put_nvmeq(struct nvme_queue *nvmeq); +int nvme_submit_sync_cmd(struct nvme_queue *nvmeq, struct nvme_command *cmd, + u32 *result, unsigned timeout); +int nvme_submit_flush_data(struct nvme_queue *nvmeq, struct nvme_ns *ns); +int nvme_submit_admin_cmd(struct nvme_dev *, struct nvme_command *, + u32 *result); +int nvme_identify(struct nvme_dev *, unsigned nsid, unsigned cns, + dma_addr_t dma_addr); +int nvme_get_features(struct nvme_dev *dev, unsigned fid, unsigned nsid, + dma_addr_t dma_addr, u32 *result); +int nvme_set_features(struct nvme_dev *dev, unsigned fid, unsigned dword11, + dma_addr_t dma_addr, u32 *result); + +struct sg_io_hdr; + +int nvme_sg_io(struct nvme_ns *ns, struct sg_io_hdr __user *u_hdr); +int nvme_sg_get_version_num(int __user *ip); + +#endif + #endif /* _LINUX_NVME_H */ |