diff options
author | Linus Torvalds <torvalds@linux-foundation.org> | 2017-12-01 16:05:45 +0300 |
---|---|---|
committer | Linus Torvalds <torvalds@linux-foundation.org> | 2017-12-01 16:05:45 +0300 |
commit | 75f64f68afa165ebe139cca2adb4df0a229a06de (patch) | |
tree | 6aee1ee80863ea121c31f15ec794f2b9943ccbfd /drivers | |
parent | df8ba95c572a187ed2aa7403e97a7a7f58c01f00 (diff) | |
parent | ed565371e368f014db237aacf42b27b40b1bd247 (diff) | |
download | linux-75f64f68afa165ebe139cca2adb4df0a229a06de.tar.xz |
Merge branch 'for-linus' of git://git.kernel.dk/linux-block
Pull block fixes from Jens Axboe:
"A selection of fixes/changes that should make it into this series.
This contains:
- NVMe, two merges, containing:
- pci-e, rdma, and fc fixes
- Device quirks
- Fix for a badblocks leak in null_blk
- bcache fix from Rui Hua for a race condition regression where
-EINTR was returned to upper layers that didn't expect it.
- Regression fix for blktrace for a bug introduced in this series.
- blktrace cleanup for cgroup id.
- bdi registration error handling.
- Small series with cleanups for blk-wbt.
- Various little fixes for typos and the like.
Nothing earth shattering, most important are the NVMe and bcache fixes"
* 'for-linus' of git://git.kernel.dk/linux-block: (34 commits)
nvme-pci: fix NULL pointer dereference in nvme_free_host_mem()
nvme-rdma: fix memory leak during queue allocation
blktrace: fix trace mutex deadlock
nvme-rdma: Use mr pool
nvme-rdma: Check remotely invalidated rkey matches our expected rkey
nvme-rdma: wait for local invalidation before completing a request
nvme-rdma: don't complete requests before a send work request has completed
nvme-rdma: don't suppress send completions
bcache: check return value of register_shrinker
bcache: recover data from backing when data is clean
bcache: Fix building error on MIPS
bcache: add a comment in journal bucket reading
nvme-fc: don't use bit masks for set/test_bit() numbers
blk-wbt: fix comments typo
blk-wbt: move wbt_clear_stat to common place in wbt_done
blk-sysfs: remove NULL pointer checking in queue_wb_lat_store
blk-wbt: remove duplicated setting in wbt_init
nvme-pci: add quirk for delay before CHK RDY for WDC SN200
block: remove useless assignment in bio_split
null_blk: fix dev->badblocks leak
...
Diffstat (limited to 'drivers')
-rw-r--r-- | drivers/block/null_blk.c | 5 | ||||
-rw-r--r-- | drivers/md/bcache/alloc.c | 2 | ||||
-rw-r--r-- | drivers/md/bcache/btree.c | 5 | ||||
-rw-r--r-- | drivers/md/bcache/extents.c | 2 | ||||
-rw-r--r-- | drivers/md/bcache/journal.c | 7 | ||||
-rw-r--r-- | drivers/md/bcache/request.c | 13 | ||||
-rw-r--r-- | drivers/nvme/host/core.c | 19 | ||||
-rw-r--r-- | drivers/nvme/host/fabrics.h | 30 | ||||
-rw-r--r-- | drivers/nvme/host/fc.c | 21 | ||||
-rw-r--r-- | drivers/nvme/host/multipath.c | 2 | ||||
-rw-r--r-- | drivers/nvme/host/nvme.h | 2 | ||||
-rw-r--r-- | drivers/nvme/host/pci.c | 17 | ||||
-rw-r--r-- | drivers/nvme/host/rdma.c | 266 | ||||
-rw-r--r-- | drivers/nvme/target/fc.c | 9 | ||||
-rw-r--r-- | drivers/nvme/target/loop.c | 25 |
15 files changed, 249 insertions, 176 deletions
diff --git a/drivers/block/null_blk.c b/drivers/block/null_blk.c index c61960deb74a..ccb9975a97fa 100644 --- a/drivers/block/null_blk.c +++ b/drivers/block/null_blk.c @@ -471,7 +471,6 @@ static void nullb_device_release(struct config_item *item) { struct nullb_device *dev = to_nullb_device(item); - badblocks_exit(&dev->badblocks); null_free_device_storage(dev, false); null_free_dev(dev); } @@ -582,6 +581,10 @@ static struct nullb_device *null_alloc_dev(void) static void null_free_dev(struct nullb_device *dev) { + if (!dev) + return; + + badblocks_exit(&dev->badblocks); kfree(dev); } diff --git a/drivers/md/bcache/alloc.c b/drivers/md/bcache/alloc.c index a27d85232ce1..a0cc1bc6d884 100644 --- a/drivers/md/bcache/alloc.c +++ b/drivers/md/bcache/alloc.c @@ -490,7 +490,7 @@ int __bch_bucket_alloc_set(struct cache_set *c, unsigned reserve, if (b == -1) goto err; - k->ptr[i] = PTR(ca->buckets[b].gen, + k->ptr[i] = MAKE_PTR(ca->buckets[b].gen, bucket_to_sector(c, b), ca->sb.nr_this_dev); diff --git a/drivers/md/bcache/btree.c b/drivers/md/bcache/btree.c index 11c5503d31dc..81e8dc3dbe5e 100644 --- a/drivers/md/bcache/btree.c +++ b/drivers/md/bcache/btree.c @@ -807,7 +807,10 @@ int bch_btree_cache_alloc(struct cache_set *c) c->shrink.scan_objects = bch_mca_scan; c->shrink.seeks = 4; c->shrink.batch = c->btree_pages * 2; - register_shrinker(&c->shrink); + + if (register_shrinker(&c->shrink)) + pr_warn("bcache: %s: could not register shrinker", + __func__); return 0; } diff --git a/drivers/md/bcache/extents.c b/drivers/md/bcache/extents.c index 41c238fc3733..f9d391711595 100644 --- a/drivers/md/bcache/extents.c +++ b/drivers/md/bcache/extents.c @@ -585,7 +585,7 @@ static bool bch_extent_merge(struct btree_keys *bk, struct bkey *l, struct bkey return false; for (i = 0; i < KEY_PTRS(l); i++) - if (l->ptr[i] + PTR(0, KEY_SIZE(l), 0) != r->ptr[i] || + if (l->ptr[i] + MAKE_PTR(0, KEY_SIZE(l), 0) != r->ptr[i] || PTR_BUCKET_NR(b->c, l, i) != PTR_BUCKET_NR(b->c, r, i)) return false; diff --git a/drivers/md/bcache/journal.c b/drivers/md/bcache/journal.c index 02a98ddb592d..a87165c1d8e5 100644 --- a/drivers/md/bcache/journal.c +++ b/drivers/md/bcache/journal.c @@ -170,6 +170,11 @@ int bch_journal_read(struct cache_set *c, struct list_head *list) * find a sequence of buckets with valid journal entries */ for (i = 0; i < ca->sb.njournal_buckets; i++) { + /* + * We must try the index l with ZERO first for + * correctness due to the scenario that the journal + * bucket is circular buffer which might have wrapped + */ l = (i * 2654435769U) % ca->sb.njournal_buckets; if (test_bit(l, bitmap)) @@ -507,7 +512,7 @@ static void journal_reclaim(struct cache_set *c) continue; ja->cur_idx = next; - k->ptr[n++] = PTR(0, + k->ptr[n++] = MAKE_PTR(0, bucket_to_sector(c, ca->sb.d[ja->cur_idx]), ca->sb.nr_this_dev); } diff --git a/drivers/md/bcache/request.c b/drivers/md/bcache/request.c index 3a7aed7282b2..643c3021624f 100644 --- a/drivers/md/bcache/request.c +++ b/drivers/md/bcache/request.c @@ -708,16 +708,15 @@ static void cached_dev_read_error(struct closure *cl) { struct search *s = container_of(cl, struct search, cl); struct bio *bio = &s->bio.bio; - struct cached_dev *dc = container_of(s->d, struct cached_dev, disk); /* - * If cache device is dirty (dc->has_dirty is non-zero), then - * recovery a failed read request from cached device may get a - * stale data back. So read failure recovery is only permitted - * when cache device is clean. + * If read request hit dirty data (s->read_dirty_data is true), + * then recovery a failed read request from cached device may + * get a stale data back. So read failure recovery is only + * permitted when read request hit clean data in cache device, + * or when cache read race happened. */ - if (s->recoverable && - (dc && !atomic_read(&dc->has_dirty))) { + if (s->recoverable && !s->read_dirty_data) { /* Retry from the backing device: */ trace_bcache_read_retry(s->orig_bio); diff --git a/drivers/nvme/host/core.c b/drivers/nvme/host/core.c index 25da74d310d1..f837d666cbd4 100644 --- a/drivers/nvme/host/core.c +++ b/drivers/nvme/host/core.c @@ -1449,19 +1449,19 @@ static int nvme_pr_command(struct block_device *bdev, u32 cdw10, int srcu_idx, ret; u8 data[16] = { 0, }; + ns = nvme_get_ns_from_disk(bdev->bd_disk, &head, &srcu_idx); + if (unlikely(!ns)) + return -EWOULDBLOCK; + put_unaligned_le64(key, &data[0]); put_unaligned_le64(sa_key, &data[8]); memset(&c, 0, sizeof(c)); c.common.opcode = op; - c.common.nsid = cpu_to_le32(head->ns_id); + c.common.nsid = cpu_to_le32(ns->head->ns_id); c.common.cdw10[0] = cpu_to_le32(cdw10); - ns = nvme_get_ns_from_disk(bdev->bd_disk, &head, &srcu_idx); - if (unlikely(!ns)) - ret = -EWOULDBLOCK; - else - ret = nvme_submit_sync_cmd(ns->queue, &c, data, 16); + ret = nvme_submit_sync_cmd(ns->queue, &c, data, 16); nvme_put_ns_from_disk(head, srcu_idx); return ret; } @@ -2961,8 +2961,6 @@ static void nvme_alloc_ns(struct nvme_ctrl *ctrl, unsigned nsid) static void nvme_ns_remove(struct nvme_ns *ns) { - struct nvme_ns_head *head = ns->head; - if (test_and_set_bit(NVME_NS_REMOVING, &ns->flags)) return; @@ -2980,15 +2978,14 @@ static void nvme_ns_remove(struct nvme_ns *ns) mutex_lock(&ns->ctrl->subsys->lock); nvme_mpath_clear_current_path(ns); - if (head) - list_del_rcu(&ns->siblings); + list_del_rcu(&ns->siblings); mutex_unlock(&ns->ctrl->subsys->lock); mutex_lock(&ns->ctrl->namespaces_mutex); list_del_init(&ns->list); mutex_unlock(&ns->ctrl->namespaces_mutex); - synchronize_srcu(&head->srcu); + synchronize_srcu(&ns->head->srcu); nvme_put_ns(ns); } diff --git a/drivers/nvme/host/fabrics.h b/drivers/nvme/host/fabrics.h index 42232e731f19..9ba614953607 100644 --- a/drivers/nvme/host/fabrics.h +++ b/drivers/nvme/host/fabrics.h @@ -156,4 +156,34 @@ void nvmf_free_options(struct nvmf_ctrl_options *opts); int nvmf_get_address(struct nvme_ctrl *ctrl, char *buf, int size); bool nvmf_should_reconnect(struct nvme_ctrl *ctrl); +static inline blk_status_t nvmf_check_init_req(struct nvme_ctrl *ctrl, + struct request *rq) +{ + struct nvme_command *cmd = nvme_req(rq)->cmd; + + /* + * We cannot accept any other command until the connect command has + * completed, so only allow connect to pass. + */ + if (!blk_rq_is_passthrough(rq) || + cmd->common.opcode != nvme_fabrics_command || + cmd->fabrics.fctype != nvme_fabrics_type_connect) { + /* + * Reconnecting state means transport disruption, which can take + * a long time and even might fail permanently, fail fast to + * give upper layers a chance to failover. + * Deleting state means that the ctrl will never accept commands + * again, fail it permanently. + */ + if (ctrl->state == NVME_CTRL_RECONNECTING || + ctrl->state == NVME_CTRL_DELETING) { + nvme_req(rq)->status = NVME_SC_ABORT_REQ; + return BLK_STS_IOERR; + } + return BLK_STS_RESOURCE; /* try again later */ + } + + return BLK_STS_OK; +} + #endif /* _NVME_FABRICS_H */ diff --git a/drivers/nvme/host/fc.c b/drivers/nvme/host/fc.c index 7ab0be55c7d0..0a8af4daef89 100644 --- a/drivers/nvme/host/fc.c +++ b/drivers/nvme/host/fc.c @@ -31,7 +31,8 @@ enum nvme_fc_queue_flags { - NVME_FC_Q_CONNECTED = (1 << 0), + NVME_FC_Q_CONNECTED = 0, + NVME_FC_Q_LIVE, }; #define NVMEFC_QUEUE_DELAY 3 /* ms units */ @@ -1927,6 +1928,7 @@ nvme_fc_free_queue(struct nvme_fc_queue *queue) if (!test_and_clear_bit(NVME_FC_Q_CONNECTED, &queue->flags)) return; + clear_bit(NVME_FC_Q_LIVE, &queue->flags); /* * Current implementation never disconnects a single queue. * It always terminates a whole association. So there is never @@ -1934,7 +1936,6 @@ nvme_fc_free_queue(struct nvme_fc_queue *queue) */ queue->connection_id = 0; - clear_bit(NVME_FC_Q_CONNECTED, &queue->flags); } static void @@ -2013,6 +2014,8 @@ nvme_fc_connect_io_queues(struct nvme_fc_ctrl *ctrl, u16 qsize) ret = nvmf_connect_io_queue(&ctrl->ctrl, i); if (ret) break; + + set_bit(NVME_FC_Q_LIVE, &ctrl->queues[i].flags); } return ret; @@ -2320,6 +2323,14 @@ busy: return BLK_STS_RESOURCE; } +static inline blk_status_t nvme_fc_is_ready(struct nvme_fc_queue *queue, + struct request *rq) +{ + if (unlikely(!test_bit(NVME_FC_Q_LIVE, &queue->flags))) + return nvmf_check_init_req(&queue->ctrl->ctrl, rq); + return BLK_STS_OK; +} + static blk_status_t nvme_fc_queue_rq(struct blk_mq_hw_ctx *hctx, const struct blk_mq_queue_data *bd) @@ -2335,6 +2346,10 @@ nvme_fc_queue_rq(struct blk_mq_hw_ctx *hctx, u32 data_len; blk_status_t ret; + ret = nvme_fc_is_ready(queue, rq); + if (unlikely(ret)) + return ret; + ret = nvme_setup_cmd(ns, rq, sqe); if (ret) return ret; @@ -2727,6 +2742,8 @@ nvme_fc_create_association(struct nvme_fc_ctrl *ctrl) if (ret) goto out_disconnect_admin_queue; + set_bit(NVME_FC_Q_LIVE, &ctrl->queues[0].flags); + /* * Check controller capabilities * diff --git a/drivers/nvme/host/multipath.c b/drivers/nvme/host/multipath.c index 78d92151a904..1218a9fca846 100644 --- a/drivers/nvme/host/multipath.c +++ b/drivers/nvme/host/multipath.c @@ -131,7 +131,7 @@ static blk_qc_t nvme_ns_head_make_request(struct request_queue *q, bio->bi_opf |= REQ_NVME_MPATH; ret = direct_make_request(bio); } else if (!list_empty_careful(&head->list)) { - dev_warn_ratelimited(dev, "no path available - requeing I/O\n"); + dev_warn_ratelimited(dev, "no path available - requeuing I/O\n"); spin_lock_irq(&head->requeue_lock); bio_list_add(&head->requeue_list, bio); diff --git a/drivers/nvme/host/nvme.h b/drivers/nvme/host/nvme.h index c0873a68872f..ea1aa5283e8e 100644 --- a/drivers/nvme/host/nvme.h +++ b/drivers/nvme/host/nvme.h @@ -114,7 +114,7 @@ static inline struct nvme_request *nvme_req(struct request *req) * NVME_QUIRK_DELAY_BEFORE_CHK_RDY quirk enabled. The value (in ms) was * found empirically. */ -#define NVME_QUIRK_DELAY_AMOUNT 2000 +#define NVME_QUIRK_DELAY_AMOUNT 2300 enum nvme_ctrl_state { NVME_CTRL_NEW, diff --git a/drivers/nvme/host/pci.c b/drivers/nvme/host/pci.c index a11cfd470089..f5800c3c9082 100644 --- a/drivers/nvme/host/pci.c +++ b/drivers/nvme/host/pci.c @@ -1759,6 +1759,7 @@ static void nvme_free_host_mem(struct nvme_dev *dev) dev->nr_host_mem_descs * sizeof(*dev->host_mem_descs), dev->host_mem_descs, dev->host_mem_descs_dma); dev->host_mem_descs = NULL; + dev->nr_host_mem_descs = 0; } static int __nvme_alloc_host_mem(struct nvme_dev *dev, u64 preferred, @@ -1787,7 +1788,7 @@ static int __nvme_alloc_host_mem(struct nvme_dev *dev, u64 preferred, if (!bufs) goto out_free_descs; - for (size = 0; size < preferred; size += len) { + for (size = 0; size < preferred && i < max_entries; size += len) { dma_addr_t dma_addr; len = min_t(u64, chunk_size, preferred - size); @@ -2428,7 +2429,7 @@ static int nvme_dev_map(struct nvme_dev *dev) return -ENODEV; } -static unsigned long check_dell_samsung_bug(struct pci_dev *pdev) +static unsigned long check_vendor_combination_bug(struct pci_dev *pdev) { if (pdev->vendor == 0x144d && pdev->device == 0xa802) { /* @@ -2443,6 +2444,14 @@ static unsigned long check_dell_samsung_bug(struct pci_dev *pdev) (dmi_match(DMI_PRODUCT_NAME, "XPS 15 9550") || dmi_match(DMI_PRODUCT_NAME, "Precision 5510"))) return NVME_QUIRK_NO_DEEPEST_PS; + } else if (pdev->vendor == 0x144d && pdev->device == 0xa804) { + /* + * Samsung SSD 960 EVO drops off the PCIe bus after system + * suspend on a Ryzen board, ASUS PRIME B350M-A. + */ + if (dmi_match(DMI_BOARD_VENDOR, "ASUSTeK COMPUTER INC.") && + dmi_match(DMI_BOARD_NAME, "PRIME B350M-A")) + return NVME_QUIRK_NO_APST; } return 0; @@ -2482,7 +2491,7 @@ static int nvme_probe(struct pci_dev *pdev, const struct pci_device_id *id) if (result) goto unmap; - quirks |= check_dell_samsung_bug(pdev); + quirks |= check_vendor_combination_bug(pdev); result = nvme_init_ctrl(&dev->ctrl, &pdev->dev, &nvme_pci_ctrl_ops, quirks); @@ -2665,6 +2674,8 @@ static const struct pci_device_id nvme_id_table[] = { .driver_data = NVME_QUIRK_IDENTIFY_CNS, }, { PCI_DEVICE(0x1c58, 0x0003), /* HGST adapter */ .driver_data = NVME_QUIRK_DELAY_BEFORE_CHK_RDY, }, + { PCI_DEVICE(0x1c58, 0x0023), /* WDC SN200 adapter */ + .driver_data = NVME_QUIRK_DELAY_BEFORE_CHK_RDY, }, { PCI_DEVICE(0x1c5f, 0x0540), /* Memblaze Pblaze4 adapter */ .driver_data = NVME_QUIRK_DELAY_BEFORE_CHK_RDY, }, { PCI_DEVICE(0x144d, 0xa821), /* Samsung PM1725 */ diff --git a/drivers/nvme/host/rdma.c b/drivers/nvme/host/rdma.c index 4f9bf2f815c3..37af56596be6 100644 --- a/drivers/nvme/host/rdma.c +++ b/drivers/nvme/host/rdma.c @@ -15,6 +15,7 @@ #include <linux/module.h> #include <linux/init.h> #include <linux/slab.h> +#include <rdma/mr_pool.h> #include <linux/err.h> #include <linux/string.h> #include <linux/atomic.h> @@ -59,6 +60,9 @@ struct nvme_rdma_request { struct nvme_request req; struct ib_mr *mr; struct nvme_rdma_qe sqe; + union nvme_result result; + __le16 status; + refcount_t ref; struct ib_sge sge[1 + NVME_RDMA_MAX_INLINE_SEGMENTS]; u32 num_sge; int nents; @@ -73,11 +77,11 @@ struct nvme_rdma_request { enum nvme_rdma_queue_flags { NVME_RDMA_Q_ALLOCATED = 0, NVME_RDMA_Q_LIVE = 1, + NVME_RDMA_Q_TR_READY = 2, }; struct nvme_rdma_queue { struct nvme_rdma_qe *rsp_ring; - atomic_t sig_count; int queue_size; size_t cmnd_capsule_len; struct nvme_rdma_ctrl *ctrl; @@ -258,32 +262,6 @@ static int nvme_rdma_create_qp(struct nvme_rdma_queue *queue, const int factor) return ret; } -static int nvme_rdma_reinit_request(void *data, struct request *rq) -{ - struct nvme_rdma_ctrl *ctrl = data; - struct nvme_rdma_device *dev = ctrl->device; - struct nvme_rdma_request *req = blk_mq_rq_to_pdu(rq); - int ret = 0; - - if (WARN_ON_ONCE(!req->mr)) - return 0; - - ib_dereg_mr(req->mr); - - req->mr = ib_alloc_mr(dev->pd, IB_MR_TYPE_MEM_REG, - ctrl->max_fr_pages); - if (IS_ERR(req->mr)) { - ret = PTR_ERR(req->mr); - req->mr = NULL; - goto out; - } - - req->mr->need_inval = false; - -out: - return ret; -} - static void nvme_rdma_exit_request(struct blk_mq_tag_set *set, struct request *rq, unsigned int hctx_idx) { @@ -293,9 +271,6 @@ static void nvme_rdma_exit_request(struct blk_mq_tag_set *set, struct nvme_rdma_queue *queue = &ctrl->queues[queue_idx]; struct nvme_rdma_device *dev = queue->device; - if (req->mr) - ib_dereg_mr(req->mr); - nvme_rdma_free_qe(dev->dev, &req->sqe, sizeof(struct nvme_command), DMA_TO_DEVICE); } @@ -317,21 +292,9 @@ static int nvme_rdma_init_request(struct blk_mq_tag_set *set, if (ret) return ret; - req->mr = ib_alloc_mr(dev->pd, IB_MR_TYPE_MEM_REG, - ctrl->max_fr_pages); - if (IS_ERR(req->mr)) { - ret = PTR_ERR(req->mr); - goto out_free_qe; - } - req->queue = queue; return 0; - -out_free_qe: - nvme_rdma_free_qe(dev->dev, &req->sqe, sizeof(struct nvme_command), - DMA_TO_DEVICE); - return -ENOMEM; } static int nvme_rdma_init_hctx(struct blk_mq_hw_ctx *hctx, void *data, @@ -428,10 +391,23 @@ out_err: static void nvme_rdma_destroy_queue_ib(struct nvme_rdma_queue *queue) { - struct nvme_rdma_device *dev = queue->device; - struct ib_device *ibdev = dev->dev; + struct nvme_rdma_device *dev; + struct ib_device *ibdev; - rdma_destroy_qp(queue->cm_id); + if (!test_and_clear_bit(NVME_RDMA_Q_TR_READY, &queue->flags)) + return; + + dev = queue->device; + ibdev = dev->dev; + + ib_mr_pool_destroy(queue->qp, &queue->qp->rdma_mrs); + + /* + * The cm_id object might have been destroyed during RDMA connection + * establishment error flow to avoid getting other cma events, thus + * the destruction of the QP shouldn't use rdma_cm API. + */ + ib_destroy_qp(queue->qp); ib_free_cq(queue->ib_cq); nvme_rdma_free_ring(ibdev, queue->rsp_ring, queue->queue_size, @@ -440,6 +416,12 @@ static void nvme_rdma_destroy_queue_ib(struct nvme_rdma_queue *queue) nvme_rdma_dev_put(dev); } +static int nvme_rdma_get_max_fr_pages(struct ib_device *ibdev) +{ + return min_t(u32, NVME_RDMA_MAX_SEGMENTS, + ibdev->attrs.max_fast_reg_page_list_len); +} + static int nvme_rdma_create_queue_ib(struct nvme_rdma_queue *queue) { struct ib_device *ibdev; @@ -482,8 +464,24 @@ static int nvme_rdma_create_queue_ib(struct nvme_rdma_queue *queue) goto out_destroy_qp; } + ret = ib_mr_pool_init(queue->qp, &queue->qp->rdma_mrs, + queue->queue_size, + IB_MR_TYPE_MEM_REG, + nvme_rdma_get_max_fr_pages(ibdev)); + if (ret) { + dev_err(queue->ctrl->ctrl.device, + "failed to initialize MR pool sized %d for QID %d\n", + queue->queue_size, idx); + goto out_destroy_ring; + } + + set_bit(NVME_RDMA_Q_TR_READY, &queue->flags); + return 0; +out_destroy_ring: + nvme_rdma_free_ring(ibdev, queue->rsp_ring, queue->queue_size, + sizeof(struct nvme_completion), DMA_FROM_DEVICE); out_destroy_qp: rdma_destroy_qp(queue->cm_id); out_destroy_ib_cq: @@ -510,7 +508,6 @@ static int nvme_rdma_alloc_queue(struct nvme_rdma_ctrl *ctrl, queue->cmnd_capsule_len = sizeof(struct nvme_command); queue->queue_size = queue_size; - atomic_set(&queue->sig_count, 0); queue->cm_id = rdma_create_id(&init_net, nvme_rdma_cm_handler, queue, RDMA_PS_TCP, IB_QPT_RC); @@ -546,6 +543,7 @@ static int nvme_rdma_alloc_queue(struct nvme_rdma_ctrl *ctrl, out_destroy_cm_id: rdma_destroy_id(queue->cm_id); + nvme_rdma_destroy_queue_ib(queue); return ret; } @@ -756,8 +754,7 @@ static int nvme_rdma_configure_admin_queue(struct nvme_rdma_ctrl *ctrl, ctrl->device = ctrl->queues[0].device; - ctrl->max_fr_pages = min_t(u32, NVME_RDMA_MAX_SEGMENTS, - ctrl->device->dev->attrs.max_fast_reg_page_list_len); + ctrl->max_fr_pages = nvme_rdma_get_max_fr_pages(ctrl->device->dev); if (new) { ctrl->ctrl.admin_tagset = nvme_rdma_alloc_tagset(&ctrl->ctrl, true); @@ -771,10 +768,6 @@ static int nvme_rdma_configure_admin_queue(struct nvme_rdma_ctrl *ctrl, error = PTR_ERR(ctrl->ctrl.admin_q); goto out_free_tagset; } - } else { - error = nvme_reinit_tagset(&ctrl->ctrl, ctrl->ctrl.admin_tagset); - if (error) - goto out_free_queue; } error = nvme_rdma_start_queue(ctrl, 0); @@ -854,10 +847,6 @@ static int nvme_rdma_configure_io_queues(struct nvme_rdma_ctrl *ctrl, bool new) goto out_free_tag_set; } } else { - ret = nvme_reinit_tagset(&ctrl->ctrl, ctrl->ctrl.tagset); - if (ret) - goto out_free_io_queues; - blk_mq_update_nr_hw_queues(&ctrl->tag_set, ctrl->ctrl.queue_count - 1); } @@ -1018,8 +1007,18 @@ static void nvme_rdma_memreg_done(struct ib_cq *cq, struct ib_wc *wc) static void nvme_rdma_inv_rkey_done(struct ib_cq *cq, struct ib_wc *wc) { - if (unlikely(wc->status != IB_WC_SUCCESS)) + struct nvme_rdma_request *req = + container_of(wc->wr_cqe, struct nvme_rdma_request, reg_cqe); + struct request *rq = blk_mq_rq_from_pdu(req); + + if (unlikely(wc->status != IB_WC_SUCCESS)) { nvme_rdma_wr_error(cq, wc, "LOCAL_INV"); + return; + } + + if (refcount_dec_and_test(&req->ref)) + nvme_end_request(rq, req->status, req->result); + } static int nvme_rdma_inv_rkey(struct nvme_rdma_queue *queue, @@ -1030,7 +1029,7 @@ static int nvme_rdma_inv_rkey(struct nvme_rdma_queue *queue, .opcode = IB_WR_LOCAL_INV, .next = NULL, .num_sge = 0, - .send_flags = 0, + .send_flags = IB_SEND_SIGNALED, .ex.invalidate_rkey = req->mr->rkey, }; @@ -1044,22 +1043,15 @@ static void nvme_rdma_unmap_data(struct nvme_rdma_queue *queue, struct request *rq) { struct nvme_rdma_request *req = blk_mq_rq_to_pdu(rq); - struct nvme_rdma_ctrl *ctrl = queue->ctrl; struct nvme_rdma_device *dev = queue->device; struct ib_device *ibdev = dev->dev; - int res; if (!blk_rq_bytes(rq)) return; - if (req->mr->need_inval && test_bit(NVME_RDMA_Q_LIVE, &req->queue->flags)) { - res = nvme_rdma_inv_rkey(queue, req); - if (unlikely(res < 0)) { - dev_err(ctrl->ctrl.device, - "Queueing INV WR for rkey %#x failed (%d)\n", - req->mr->rkey, res); - nvme_rdma_error_recovery(queue->ctrl); - } + if (req->mr) { + ib_mr_pool_put(queue->qp, &queue->qp->rdma_mrs, req->mr); + req->mr = NULL; } ib_dma_unmap_sg(ibdev, req->sg_table.sgl, @@ -1118,12 +1110,18 @@ static int nvme_rdma_map_sg_fr(struct nvme_rdma_queue *queue, struct nvme_keyed_sgl_desc *sg = &c->common.dptr.ksgl; int nr; + req->mr = ib_mr_pool_get(queue->qp, &queue->qp->rdma_mrs); + if (WARN_ON_ONCE(!req->mr)) + return -EAGAIN; + /* * Align the MR to a 4K page size to match the ctrl page size and * the block virtual boundary. */ nr = ib_map_mr_sg(req->mr, req->sg_table.sgl, count, NULL, SZ_4K); if (unlikely(nr < count)) { + ib_mr_pool_put(queue->qp, &queue->qp->rdma_mrs, req->mr); + req->mr = NULL; if (nr < 0) return nr; return -EINVAL; @@ -1142,8 +1140,6 @@ static int nvme_rdma_map_sg_fr(struct nvme_rdma_queue *queue, IB_ACCESS_REMOTE_READ | IB_ACCESS_REMOTE_WRITE; - req->mr->need_inval = true; - sg->addr = cpu_to_le64(req->mr->iova); put_unaligned_le24(req->mr->length, sg->length); put_unaligned_le32(req->mr->rkey, sg->key); @@ -1163,7 +1159,7 @@ static int nvme_rdma_map_data(struct nvme_rdma_queue *queue, req->num_sge = 1; req->inline_data = false; - req->mr->need_inval = false; + refcount_set(&req->ref, 2); /* send and recv completions */ c->common.flags |= NVME_CMD_SGL_METABUF; @@ -1200,25 +1196,24 @@ static int nvme_rdma_map_data(struct nvme_rdma_queue *queue, static void nvme_rdma_send_done(struct ib_cq *cq, struct ib_wc *wc) { - if (unlikely(wc->status != IB_WC_SUCCESS)) - nvme_rdma_wr_error(cq, wc, "SEND"); -} + struct nvme_rdma_qe *qe = + container_of(wc->wr_cqe, struct nvme_rdma_qe, cqe); + struct nvme_rdma_request *req = + container_of(qe, struct nvme_rdma_request, sqe); + struct request *rq = blk_mq_rq_from_pdu(req); -/* - * We want to signal completion at least every queue depth/2. This returns the - * largest power of two that is not above half of (queue size + 1) to optimize - * (avoid divisions). - */ -static inline bool nvme_rdma_queue_sig_limit(struct nvme_rdma_queue *queue) -{ - int limit = 1 << ilog2((queue->queue_size + 1) / 2); + if (unlikely(wc->status != IB_WC_SUCCESS)) { + nvme_rdma_wr_error(cq, wc, "SEND"); + return; + } - return (atomic_inc_return(&queue->sig_count) & (limit - 1)) == 0; + if (refcount_dec_and_test(&req->ref)) + nvme_end_request(rq, req->status, req->result); } static int nvme_rdma_post_send(struct nvme_rdma_queue *queue, struct nvme_rdma_qe *qe, struct ib_sge *sge, u32 num_sge, - struct ib_send_wr *first, bool flush) + struct ib_send_wr *first) { struct ib_send_wr wr, *bad_wr; int ret; @@ -1227,31 +1222,12 @@ static int nvme_rdma_post_send(struct nvme_rdma_queue *queue, sge->length = sizeof(struct nvme_command), sge->lkey = queue->device->pd->local_dma_lkey; - qe->cqe.done = nvme_rdma_send_done; - wr.next = NULL; wr.wr_cqe = &qe->cqe; wr.sg_list = sge; wr.num_sge = num_sge; wr.opcode = IB_WR_SEND; - wr.send_flags = 0; - - /* - * Unsignalled send completions are another giant desaster in the - * IB Verbs spec: If we don't regularly post signalled sends - * the send queue will fill up and only a QP reset will rescue us. - * Would have been way to obvious to handle this in hardware or - * at least the RDMA stack.. - * - * Always signal the flushes. The magic request used for the flush - * sequencer is not allocated in our driver's tagset and it's - * triggered to be freed by blk_cleanup_queue(). So we need to - * always mark it as signaled to ensure that the "wr_cqe", which is - * embedded in request's payload, is not freed when __ib_process_cq() - * calls wr_cqe->done(). - */ - if (nvme_rdma_queue_sig_limit(queue) || flush) - wr.send_flags |= IB_SEND_SIGNALED; + wr.send_flags = IB_SEND_SIGNALED; if (first) first->next = ≀ @@ -1301,6 +1277,12 @@ static struct blk_mq_tags *nvme_rdma_tagset(struct nvme_rdma_queue *queue) return queue->ctrl->tag_set.tags[queue_idx - 1]; } +static void nvme_rdma_async_done(struct ib_cq *cq, struct ib_wc *wc) +{ + if (unlikely(wc->status != IB_WC_SUCCESS)) + nvme_rdma_wr_error(cq, wc, "ASYNC"); +} + static void nvme_rdma_submit_async_event(struct nvme_ctrl *arg) { struct nvme_rdma_ctrl *ctrl = to_rdma_ctrl(arg); @@ -1319,10 +1301,12 @@ static void nvme_rdma_submit_async_event(struct nvme_ctrl *arg) cmd->common.flags |= NVME_CMD_SGL_METABUF; nvme_rdma_set_sg_null(cmd); + sqe->cqe.done = nvme_rdma_async_done; + ib_dma_sync_single_for_device(dev, sqe->dma, sizeof(*cmd), DMA_TO_DEVICE); - ret = nvme_rdma_post_send(queue, sqe, &sge, 1, NULL, false); + ret = nvme_rdma_post_send(queue, sqe, &sge, 1, NULL); WARN_ON_ONCE(ret); } @@ -1343,14 +1327,34 @@ static int nvme_rdma_process_nvme_rsp(struct nvme_rdma_queue *queue, } req = blk_mq_rq_to_pdu(rq); - if (rq->tag == tag) - ret = 1; + req->status = cqe->status; + req->result = cqe->result; - if ((wc->wc_flags & IB_WC_WITH_INVALIDATE) && - wc->ex.invalidate_rkey == req->mr->rkey) - req->mr->need_inval = false; + if (wc->wc_flags & IB_WC_WITH_INVALIDATE) { + if (unlikely(wc->ex.invalidate_rkey != req->mr->rkey)) { + dev_err(queue->ctrl->ctrl.device, + "Bogus remote invalidation for rkey %#x\n", + req->mr->rkey); + nvme_rdma_error_recovery(queue->ctrl); + } + } else if (req->mr) { + ret = nvme_rdma_inv_rkey(queue, req); + if (unlikely(ret < 0)) { + dev_err(queue->ctrl->ctrl.device, + "Queueing INV WR for rkey %#x failed (%d)\n", + req->mr->rkey, ret); + nvme_rdma_error_recovery(queue->ctrl); + } + /* the local invalidation completion will end the request */ + return 0; + } + + if (refcount_dec_and_test(&req->ref)) { + if (rq->tag == tag) + ret = 1; + nvme_end_request(rq, req->status, req->result); + } - nvme_end_request(rq, cqe->status, cqe->result); return ret; } @@ -1591,31 +1595,11 @@ nvme_rdma_timeout(struct request *rq, bool reserved) * We cannot accept any other command until the Connect command has completed. */ static inline blk_status_t -nvme_rdma_queue_is_ready(struct nvme_rdma_queue *queue, struct request *rq) -{ - if (unlikely(!test_bit(NVME_RDMA_Q_LIVE, &queue->flags))) { - struct nvme_command *cmd = nvme_req(rq)->cmd; - - if (!blk_rq_is_passthrough(rq) || - cmd->common.opcode != nvme_fabrics_command || - cmd->fabrics.fctype != nvme_fabrics_type_connect) { - /* - * reconnecting state means transport disruption, which - * can take a long time and even might fail permanently, - * fail fast to give upper layers a chance to failover. - * deleting state means that the ctrl will never accept - * commands again, fail it permanently. - */ - if (queue->ctrl->ctrl.state == NVME_CTRL_RECONNECTING || - queue->ctrl->ctrl.state == NVME_CTRL_DELETING) { - nvme_req(rq)->status = NVME_SC_ABORT_REQ; - return BLK_STS_IOERR; - } - return BLK_STS_RESOURCE; /* try again later */ - } - } - - return 0; +nvme_rdma_is_ready(struct nvme_rdma_queue *queue, struct request *rq) +{ + if (unlikely(!test_bit(NVME_RDMA_Q_LIVE, &queue->flags))) + return nvmf_check_init_req(&queue->ctrl->ctrl, rq); + return BLK_STS_OK; } static blk_status_t nvme_rdma_queue_rq(struct blk_mq_hw_ctx *hctx, @@ -1627,14 +1611,13 @@ static blk_status_t nvme_rdma_queue_rq(struct blk_mq_hw_ctx *hctx, struct nvme_rdma_request *req = blk_mq_rq_to_pdu(rq); struct nvme_rdma_qe *sqe = &req->sqe; struct nvme_command *c = sqe->data; - bool flush = false; struct ib_device *dev; blk_status_t ret; int err; WARN_ON_ONCE(rq->tag < 0); - ret = nvme_rdma_queue_is_ready(queue, rq); + ret = nvme_rdma_is_ready(queue, rq); if (unlikely(ret)) return ret; @@ -1656,13 +1639,13 @@ static blk_status_t nvme_rdma_queue_rq(struct blk_mq_hw_ctx *hctx, goto err; } + sqe->cqe.done = nvme_rdma_send_done; + ib_dma_sync_single_for_device(dev, sqe->dma, sizeof(struct nvme_command), DMA_TO_DEVICE); - if (req_op(rq) == REQ_OP_FLUSH) - flush = true; err = nvme_rdma_post_send(queue, sqe, req->sge, req->num_sge, - req->mr->need_inval ? &req->reg_wr.wr : NULL, flush); + req->mr ? &req->reg_wr.wr : NULL); if (unlikely(err)) { nvme_rdma_unmap_data(queue, rq); goto err; @@ -1810,7 +1793,6 @@ static const struct nvme_ctrl_ops nvme_rdma_ctrl_ops = { .submit_async_event = nvme_rdma_submit_async_event, .delete_ctrl = nvme_rdma_delete_ctrl, .get_address = nvmf_get_address, - .reinit_request = nvme_rdma_reinit_request, }; static inline bool diff --git a/drivers/nvme/target/fc.c b/drivers/nvme/target/fc.c index 664d3013f68f..5fd86039e353 100644 --- a/drivers/nvme/target/fc.c +++ b/drivers/nvme/target/fc.c @@ -533,15 +533,15 @@ nvmet_fc_free_fcp_iod(struct nvmet_fc_tgt_queue *queue, tgtport->ops->fcp_req_release(&tgtport->fc_target_port, fcpreq); + /* release the queue lookup reference on the completed IO */ + nvmet_fc_tgt_q_put(queue); + spin_lock_irqsave(&queue->qlock, flags); deferfcp = list_first_entry_or_null(&queue->pending_cmd_list, struct nvmet_fc_defer_fcp_req, req_list); if (!deferfcp) { list_add_tail(&fod->fcp_list, &fod->queue->fod_list); spin_unlock_irqrestore(&queue->qlock, flags); - - /* Release reference taken at queue lookup and fod allocation */ - nvmet_fc_tgt_q_put(queue); return; } @@ -760,6 +760,9 @@ nvmet_fc_delete_target_queue(struct nvmet_fc_tgt_queue *queue) tgtport->ops->fcp_req_release(&tgtport->fc_target_port, deferfcp->fcp_req); + /* release the queue lookup reference */ + nvmet_fc_tgt_q_put(queue); + kfree(deferfcp); spin_lock_irqsave(&queue->qlock, flags); diff --git a/drivers/nvme/target/loop.c b/drivers/nvme/target/loop.c index 96d390416789..1e21b286f299 100644 --- a/drivers/nvme/target/loop.c +++ b/drivers/nvme/target/loop.c @@ -52,10 +52,15 @@ static inline struct nvme_loop_ctrl *to_loop_ctrl(struct nvme_ctrl *ctrl) return container_of(ctrl, struct nvme_loop_ctrl, ctrl); } +enum nvme_loop_queue_flags { + NVME_LOOP_Q_LIVE = 0, +}; + struct nvme_loop_queue { struct nvmet_cq nvme_cq; struct nvmet_sq nvme_sq; struct nvme_loop_ctrl *ctrl; + unsigned long flags; }; static struct nvmet_port *nvmet_loop_port; @@ -144,6 +149,14 @@ nvme_loop_timeout(struct request *rq, bool reserved) return BLK_EH_HANDLED; } +static inline blk_status_t nvme_loop_is_ready(struct nvme_loop_queue *queue, + struct request *rq) +{ + if (unlikely(!test_bit(NVME_LOOP_Q_LIVE, &queue->flags))) + return nvmf_check_init_req(&queue->ctrl->ctrl, rq); + return BLK_STS_OK; +} + static blk_status_t nvme_loop_queue_rq(struct blk_mq_hw_ctx *hctx, const struct blk_mq_queue_data *bd) { @@ -153,6 +166,10 @@ static blk_status_t nvme_loop_queue_rq(struct blk_mq_hw_ctx *hctx, struct nvme_loop_iod *iod = blk_mq_rq_to_pdu(req); blk_status_t ret; + ret = nvme_loop_is_ready(queue, req); + if (unlikely(ret)) + return ret; + ret = nvme_setup_cmd(ns, req, &iod->cmd); if (ret) return ret; @@ -267,6 +284,7 @@ static const struct blk_mq_ops nvme_loop_admin_mq_ops = { static void nvme_loop_destroy_admin_queue(struct nvme_loop_ctrl *ctrl) { + clear_bit(NVME_LOOP_Q_LIVE, &ctrl->queues[0].flags); nvmet_sq_destroy(&ctrl->queues[0].nvme_sq); blk_cleanup_queue(ctrl->ctrl.admin_q); blk_mq_free_tag_set(&ctrl->admin_tag_set); @@ -297,8 +315,10 @@ static void nvme_loop_destroy_io_queues(struct nvme_loop_ctrl *ctrl) { int i; - for (i = 1; i < ctrl->ctrl.queue_count; i++) + for (i = 1; i < ctrl->ctrl.queue_count; i++) { + clear_bit(NVME_LOOP_Q_LIVE, &ctrl->queues[i].flags); nvmet_sq_destroy(&ctrl->queues[i].nvme_sq); + } } static int nvme_loop_init_io_queues(struct nvme_loop_ctrl *ctrl) @@ -338,6 +358,7 @@ static int nvme_loop_connect_io_queues(struct nvme_loop_ctrl *ctrl) ret = nvmf_connect_io_queue(&ctrl->ctrl, i); if (ret) return ret; + set_bit(NVME_LOOP_Q_LIVE, &ctrl->queues[i].flags); } return 0; @@ -380,6 +401,8 @@ static int nvme_loop_configure_admin_queue(struct nvme_loop_ctrl *ctrl) if (error) goto out_cleanup_queue; + set_bit(NVME_LOOP_Q_LIVE, &ctrl->queues[0].flags); + error = nvmf_reg_read64(&ctrl->ctrl, NVME_REG_CAP, &ctrl->ctrl.cap); if (error) { dev_err(ctrl->ctrl.device, |