Merge branch 'for-4.15/block' of git://git.kernel.dk/linux-block

Pull core block layer updates from Jens Axboe:
 "This is the main pull request for block storage for 4.15-rc1.

  Nothing out of the ordinary in here, and no API changes or anything
  like that. Just various new features for drivers, core changes, etc.
  In particular, this pull request contains:

   - A patch series from Bart, closing the whole on blk/scsi-mq queue
     quescing.

   - A series from Christoph, building towards hidden gendisks (for
     multipath) and ability to move bio chains around.

   - NVMe
        - Support for native multipath for NVMe (Christoph).
        - Userspace notifications for AENs (Keith).
        - Command side-effects support (Keith).
        - SGL support (Chaitanya Kulkarni)
        - FC fixes and improvements (James Smart)
        - Lots of fixes and tweaks (Various)

   - bcache
        - New maintainer (Michael Lyle)
        - Writeback control improvements (Michael)
        - Various fixes (Coly, Elena, Eric, Liang, et al)

   - lightnvm updates, mostly centered around the pblk interface
     (Javier, Hans, and Rakesh).

   - Removal of unused bio/bvec kmap atomic interfaces (me, Christoph)

   - Writeback series that fix the much discussed hundreds of millions
     of sync-all units. This goes all the way, as discussed previously
     (me).

   - Fix for missing wakeup on writeback timer adjustments (Yafang
     Shao).

   - Fix laptop mode on blk-mq (me).

   - {mq,name} tupple lookup for IO schedulers, allowing us to have
     alias names. This means you can use 'deadline' on both !mq and on
     mq (where it's called mq-deadline). (me).

   - blktrace race fix, oopsing on sg load (me).

   - blk-mq optimizations (me).

   - Obscure waitqueue race fix for kyber (Omar).

   - NBD fixes (Josef).

   - Disable writeback throttling by default on bfq, like we do on cfq
     (Luca Miccio).

   - Series from Ming that enable us to treat flush requests on blk-mq
     like any other request. This is a really nice cleanup.

   - Series from Ming that improves merging on blk-mq with schedulers,
     getting us closer to flipping the switch on scsi-mq again.

   - BFQ updates (Paolo).

   - blk-mq atomic flags memory ordering fixes (Peter Z).

   - Loop cgroup support (Shaohua).

   - Lots of minor fixes from lots of different folks, both for core and
     driver code"

* 'for-4.15/block' of git://git.kernel.dk/linux-block: (294 commits)
  nvme: fix visibility of "uuid" ns attribute
  blk-mq: fixup some comment typos and lengths
  ide: ide-atapi: fix compile error with defining macro DEBUG
  blk-mq: improve tag waiting setup for non-shared tags
  brd: remove unused brd_mutex
  blk-mq: only run the hardware queue if IO is pending
  block: avoid null pointer dereference on null disk
  fs: guard_bio_eod() needs to consider partitions
  xtensa/simdisk: fix compile error
  nvme: expose subsys attribute to sysfs
  nvme: create 'slaves' and 'holders' entries for hidden controllers
  block: create 'slaves' and 'holders' entries for hidden gendisks
  nvme: also expose the namespace identification sysfs files for mpath nodes
  nvme: implement multipath access to nvme subsystems
  nvme: track shared namespaces
  nvme: introduce a nvme_ns_ids structure
  nvme: track subsystems
  block, nvme: Introduce blk_mq_req_flags_t
  block, scsi: Make SCSI quiesce and resume work reliably
  block: Add the QUEUE_FLAG_PREEMPT_ONLY request queue flag
  ...

Documentation/ABI/obsolete/proc-sys-vm-nr_pdflush_threads

@@ -1,5 +0,0 @@
-What:		/proc/sys/vm/nr_pdflush_threads
-Date:		June 2012
-Contact:	Wanpeng Li <liwp@linux.vnet.ibm.com>
-Description: Since pdflush is replaced by per-BDI flusher, the interface of old pdflush
-             exported in /proc/sys/vm/ should be removed.

Documentation/block/biodoc.txt

@@ -216,10 +216,9 @@ may need to abort DMA operations and revert to PIO for the transfer, in
 which case a virtual mapping of the page is required. For SCSI it is also
 done in some scenarios where the low level driver cannot be trusted to
 handle a single sg entry correctly. The driver is expected to perform the
-kmaps as needed on such occasions using the __bio_kmap_atomic and bio_kmap_irq
-routines as appropriate. A driver could also use the blk_queue_bounce()
-routine on its own to bounce highmem i/o to low memory for specific requests
-if so desired.
+kmaps as needed on such occasions as appropriate. A driver could also use
+the blk_queue_bounce() routine on its own to bounce highmem i/o to low
+memory for specific requests if so desired.
 
 iii. The i/o scheduler algorithm itself can be replaced/set as appropriate
 
@@ -1137,8 +1136,8 @@ use dma_map_sg for scatter gather) to be able to ship it to the driver. For
 PIO drivers (or drivers that need to revert to PIO transfer once in a
 while (IDE for example)), where the CPU is doing the actual data
 transfer a virtual mapping is needed. If the driver supports highmem I/O,
-(Sec 1.1, (ii) ) it needs to use __bio_kmap_atomic and bio_kmap_irq to
-temporarily map a bio into the virtual address space.
+(Sec 1.1, (ii) ) it needs to use kmap_atomic or similar to temporarily map
+a bio into the virtual address space.
 
 
 8. Prior/Related/Impacted patches

Documentation/block/null_blk.txt

@@ -38,7 +38,7 @@ gb=[Size in GB]: Default: 250GB
 bs=[Block size (in bytes)]: Default: 512 bytes
   The block size reported to the system.
 
-nr_devices=[Number of devices]: Default: 2
+nr_devices=[Number of devices]: Default: 1
   Number of block devices instantiated. They are instantiated as /dev/nullb0,
   etc.
 
@@ -52,13 +52,13 @@ irqmode=[0-2]: Default: 1-Soft-irq
   2: Timer: Waits a specific period (completion_nsec) for each IO before
      completion.
 
-completion_nsec=[ns]: Default: 10.000ns
+completion_nsec=[ns]: Default: 10,000ns
   Combined with irqmode=2 (timer). The time each completion event must wait.
 
-submit_queues=[0..nr_cpus]:
+submit_queues=[1..nr_cpus]:
   The number of submission queues attached to the device driver. If unset, it
-  defaults to 1 on single-queue and bio-based instances. For multi-queue,
-  it is ignored when use_per_node_hctx module parameter is 1.
+  defaults to 1. For multi-queue, it is ignored when use_per_node_hctx module
+  parameter is 1.
 
 hw_queue_depth=[0..qdepth]: Default: 64
   The hardware queue depth of the device.
@@ -73,3 +73,12 @@ use_per_node_hctx=[0/1]: Default: 0
 
 use_lightnvm=[0/1]: Default: 0
   Register device with LightNVM. Requires blk-mq and CONFIG_NVM to be enabled.
+
+no_sched=[0/1]: Default: 0
+  0: nullb* use default blk-mq io scheduler.
+  1: nullb* doesn't use io scheduler.
+
+shared_tags=[0/1]: Default: 0
+  0: Tag set is not shared.
+  1: Tag set shared between devices for blk-mq. Only makes sense with
+     nr_devices > 1, otherwise there's no tag set to share.

MAINTAINERS

@@ -2562,10 +2562,12 @@ S:	Maintained
 F:	drivers/net/hamradio/baycom*
 
 BCACHE (BLOCK LAYER CACHE)
+M:	Michael Lyle <mlyle@lyle.org>
 M:	Kent Overstreet <kent.overstreet@gmail.com>
 L:	linux-bcache@vger.kernel.org
 W:	http://bcache.evilpiepirate.org
-S:	Orphan
+C:	irc://irc.oftc.net/bcache
+S:	Maintained
 F:	drivers/md/bcache/
 
 BDISP ST MEDIA DRIVER
@@ -12085,7 +12087,6 @@ F:	drivers/mmc/host/sdhci-omap.c
 SECURE ENCRYPTING DEVICE (SED) OPAL DRIVER
 M:	Scott Bauer <scott.bauer@intel.com>
 M:	Jonathan Derrick <jonathan.derrick@intel.com>
-M:	Rafael Antognolli <rafael.antognolli@intel.com>
 L:	linux-block@vger.kernel.org
 S:	Supported
 F:	block/sed*

arch/xtensa/platforms/iss/simdisk.c

@@ -110,13 +110,13 @@ static blk_qc_t simdisk_make_request(struct request_queue *q, struct bio *bio)
 	sector_t sector = bio->bi_iter.bi_sector;
 
 	bio_for_each_segment(bvec, bio, iter) {
-		char *buffer = __bio_kmap_atomic(bio, iter);
+		char *buffer = kmap_atomic(bvec.bv_page) + bvec.bv_offset;
 		unsigned len = bvec.bv_len >> SECTOR_SHIFT;
 
 		simdisk_transfer(dev, sector, len, buffer,
 				bio_data_dir(bio) == WRITE);
 		sector += len;
-		__bio_kunmap_atomic(buffer);
+		kunmap_atomic(buffer);
 	}
 
 	bio_endio(bio);

block/bfq-iosched.c

@@ -108,6 +108,7 @@
 #include "blk-mq-tag.h"
 #include "blk-mq-sched.h"
 #include "bfq-iosched.h"
+#include "blk-wbt.h"
 
 #define BFQ_BFQQ_FNS(name)						\
 void bfq_mark_bfqq_##name(struct bfq_queue *bfqq)			\
@@ -724,6 +725,44 @@ static void bfq_updated_next_req(struct bfq_data *bfqd,
 	}
 }
 
+static unsigned int bfq_wr_duration(struct bfq_data *bfqd)
+{
+	u64 dur;
+
+	if (bfqd->bfq_wr_max_time > 0)
+		return bfqd->bfq_wr_max_time;
+
+	dur = bfqd->RT_prod;
+	do_div(dur, bfqd->peak_rate);
+
+	/*
+	 * Limit duration between 3 and 13 seconds. Tests show that
+	 * higher values than 13 seconds often yield the opposite of
+	 * the desired result, i.e., worsen responsiveness by letting
+	 * non-interactive and non-soft-real-time applications
+	 * preserve weight raising for a too long time interval.
+	 *
+	 * On the other end, lower values than 3 seconds make it
+	 * difficult for most interactive tasks to complete their jobs
+	 * before weight-raising finishes.
+	 */
+	if (dur > msecs_to_jiffies(13000))
+		dur = msecs_to_jiffies(13000);
+	else if (dur < msecs_to_jiffies(3000))
+		dur = msecs_to_jiffies(3000);
+
+	return dur;
+}
+
+/* switch back from soft real-time to interactive weight raising */
+static void switch_back_to_interactive_wr(struct bfq_queue *bfqq,
+					  struct bfq_data *bfqd)
+{
+	bfqq->wr_coeff = bfqd->bfq_wr_coeff;
+	bfqq->wr_cur_max_time = bfq_wr_duration(bfqd);
+	bfqq->last_wr_start_finish = bfqq->wr_start_at_switch_to_srt;
+}
+
 static void
 bfq_bfqq_resume_state(struct bfq_queue *bfqq, struct bfq_data *bfqd,
 		      struct bfq_io_cq *bic, bool bfq_already_existing)
@@ -750,10 +789,16 @@ bfq_bfqq_resume_state(struct bfq_queue *bfqq, struct bfq_data *bfqd,
 	if (bfqq->wr_coeff > 1 && (bfq_bfqq_in_large_burst(bfqq) ||
 	    time_is_before_jiffies(bfqq->last_wr_start_finish +
 				   bfqq->wr_cur_max_time))) {
-		bfq_log_bfqq(bfqq->bfqd, bfqq,
-		    "resume state: switching off wr");
-
-		bfqq->wr_coeff = 1;
+		if (bfqq->wr_cur_max_time == bfqd->bfq_wr_rt_max_time &&
+		    !bfq_bfqq_in_large_burst(bfqq) &&
+		    time_is_after_eq_jiffies(bfqq->wr_start_at_switch_to_srt +
+					     bfq_wr_duration(bfqd))) {
+			switch_back_to_interactive_wr(bfqq, bfqd);
+		} else {
+			bfqq->wr_coeff = 1;
+			bfq_log_bfqq(bfqq->bfqd, bfqq,
+				     "resume state: switching off wr");
+		}
 	}
 
 	/* make sure weight will be updated, however we got here */
@@ -1173,33 +1218,22 @@ static bool bfq_bfqq_update_budg_for_activation(struct bfq_data *bfqd,
 	return wr_or_deserves_wr;
 }
 
-static unsigned int bfq_wr_duration(struct bfq_data *bfqd)
+/*
+ * Return the farthest future time instant according to jiffies
+ * macros.
+ */
+static unsigned long bfq_greatest_from_now(void)
 {
-	u64 dur;
-
-	if (bfqd->bfq_wr_max_time > 0)
-		return bfqd->bfq_wr_max_time;
-
-	dur = bfqd->RT_prod;
-	do_div(dur, bfqd->peak_rate);
-
-	/*
-	 * Limit duration between 3 and 13 seconds. Tests show that
-	 * higher values than 13 seconds often yield the opposite of
-	 * the desired result, i.e., worsen responsiveness by letting
-	 * non-interactive and non-soft-real-time applications
-	 * preserve weight raising for a too long time interval.
-	 *
-	 * On the other end, lower values than 3 seconds make it
-	 * difficult for most interactive tasks to complete their jobs
-	 * before weight-raising finishes.
-	 */
-	if (dur > msecs_to_jiffies(13000))
-		dur = msecs_to_jiffies(13000);
-	else if (dur < msecs_to_jiffies(3000))
-		dur = msecs_to_jiffies(3000);
+	return jiffies + MAX_JIFFY_OFFSET;
+}
 
-	return dur;
+/*
+ * Return the farthest past time instant according to jiffies
+ * macros.
+ */
+static unsigned long bfq_smallest_from_now(void)
+{
+	return jiffies - MAX_JIFFY_OFFSET;
 }
 
 static void bfq_update_bfqq_wr_on_rq_arrival(struct bfq_data *bfqd,
@@ -1216,7 +1250,19 @@ static void bfq_update_bfqq_wr_on_rq_arrival(struct bfq_data *bfqd,
 			bfqq->wr_coeff = bfqd->bfq_wr_coeff;
 			bfqq->wr_cur_max_time = bfq_wr_duration(bfqd);
 		} else {
-			bfqq->wr_start_at_switch_to_srt = jiffies;
+			/*
+			 * No interactive weight raising in progress
+			 * here: assign minus infinity to
+			 * wr_start_at_switch_to_srt, to make sure
+			 * that, at the end of the soft-real-time
+			 * weight raising periods that is starting
+			 * now, no interactive weight-raising period
+			 * may be wrongly considered as still in
+			 * progress (and thus actually started by
+			 * mistake).
+			 */
+			bfqq->wr_start_at_switch_to_srt =
+				bfq_smallest_from_now();
 			bfqq->wr_coeff = bfqd->bfq_wr_coeff *
 				BFQ_SOFTRT_WEIGHT_FACTOR;
 			bfqq->wr_cur_max_time =
@@ -2016,10 +2062,27 @@ static void bfq_bfqq_save_state(struct bfq_queue *bfqq)
 	bic->saved_IO_bound = bfq_bfqq_IO_bound(bfqq);
 	bic->saved_in_large_burst = bfq_bfqq_in_large_burst(bfqq);
 	bic->was_in_burst_list = !hlist_unhashed(&bfqq->burst_list_node);
-	bic->saved_wr_coeff = bfqq->wr_coeff;
-	bic->saved_wr_start_at_switch_to_srt = bfqq->wr_start_at_switch_to_srt;
-	bic->saved_last_wr_start_finish = bfqq->last_wr_start_finish;
-	bic->saved_wr_cur_max_time = bfqq->wr_cur_max_time;
+	if (unlikely(bfq_bfqq_just_created(bfqq) &&
+		     !bfq_bfqq_in_large_burst(bfqq))) {
+		/*
+		 * bfqq being merged right after being created: bfqq
+		 * would have deserved interactive weight raising, but
+		 * did not make it to be set in a weight-raised state,
+		 * because of this early merge.	Store directly the
+		 * weight-raising state that would have been assigned
+		 * to bfqq, so that to avoid that bfqq unjustly fails
+		 * to enjoy weight raising if split soon.
+		 */
+		bic->saved_wr_coeff = bfqq->bfqd->bfq_wr_coeff;
+		bic->saved_wr_cur_max_time = bfq_wr_duration(bfqq->bfqd);
+		bic->saved_last_wr_start_finish = jiffies;
+	} else {
+		bic->saved_wr_coeff = bfqq->wr_coeff;
+		bic->saved_wr_start_at_switch_to_srt =
+			bfqq->wr_start_at_switch_to_srt;
+		bic->saved_last_wr_start_finish = bfqq->last_wr_start_finish;
+		bic->saved_wr_cur_max_time = bfqq->wr_cur_max_time;
+	}
 }
 
 static void
@@ -2897,24 +2960,6 @@ static unsigned long bfq_bfqq_softrt_next_start(struct bfq_data *bfqd,
 		   jiffies + nsecs_to_jiffies(bfqq->bfqd->bfq_slice_idle) + 4);
 }
 
-/*
- * Return the farthest future time instant according to jiffies
- * macros.
- */
-static unsigned long bfq_greatest_from_now(void)
-{
-	return jiffies + MAX_JIFFY_OFFSET;
-}
-
-/*
- * Return the farthest past time instant according to jiffies
- * macros.
- */
-static unsigned long bfq_smallest_from_now(void)
-{
-	return jiffies - MAX_JIFFY_OFFSET;
-}
-
 /**
  * bfq_bfqq_expire - expire a queue.
  * @bfqd: device owning the queue.
@@ -3489,11 +3534,7 @@ static void bfq_update_wr_data(struct bfq_data *bfqd, struct bfq_queue *bfqq)
 					       bfq_wr_duration(bfqd)))
 				bfq_bfqq_end_wr(bfqq);
 			else {
-				/* switch back to interactive wr */
-				bfqq->wr_coeff = bfqd->bfq_wr_coeff;
-				bfqq->wr_cur_max_time = bfq_wr_duration(bfqd);
-				bfqq->last_wr_start_finish =
-					bfqq->wr_start_at_switch_to_srt;
+				switch_back_to_interactive_wr(bfqq, bfqd);
 				bfqq->entity.prio_changed = 1;
 			}
 		}
@@ -3685,16 +3726,37 @@ void bfq_put_queue(struct bfq_queue *bfqq)
 	if (bfqq->ref)
 		return;
 
-	if (bfq_bfqq_sync(bfqq))
+	if (!hlist_unhashed(&bfqq->burst_list_node)) {
+		hlist_del_init(&bfqq->burst_list_node);
 		/*
-		 * The fact that this queue is being destroyed does not
-		 * invalidate the fact that this queue may have been
-		 * activated during the current burst. As a consequence,
-		 * although the queue does not exist anymore, and hence
-		 * needs to be removed from the burst list if there,
-		 * the burst size has not to be decremented.
+		 * Decrement also burst size after the removal, if the
+		 * process associated with bfqq is exiting, and thus
+		 * does not contribute to the burst any longer. This
+		 * decrement helps filter out false positives of large
+		 * bursts, when some short-lived process (often due to
+		 * the execution of commands by some service) happens
+		 * to start and exit while a complex application is
+		 * starting, and thus spawning several processes that
+		 * do I/O (and that *must not* be treated as a large
+		 * burst, see comments on bfq_handle_burst).
+		 *
+		 * In particular, the decrement is performed only if:
+		 * 1) bfqq is not a merged queue, because, if it is,
+		 * then this free of bfqq is not triggered by the exit
+		 * of the process bfqq is associated with, but exactly
+		 * by the fact that bfqq has just been merged.
+		 * 2) burst_size is greater than 0, to handle
+		 * unbalanced decrements. Unbalanced decrements may
+		 * happen in te following case: bfqq is inserted into
+		 * the current burst list--without incrementing
+		 * bust_size--because of a split, but the current
+		 * burst list is not the burst list bfqq belonged to
+		 * (see comments on the case of a split in
+		 * bfq_set_request).
 		 */
-		hlist_del_init(&bfqq->burst_list_node);
+		if (bfqq->bic && bfqq->bfqd->burst_size > 0)
+			bfqq->bfqd->burst_size--;
+	}
 
 	kmem_cache_free(bfq_pool, bfqq);
 #ifdef CONFIG_BFQ_GROUP_IOSCHED
@@ -4127,7 +4189,6 @@ static void __bfq_insert_request(struct bfq_data *bfqd, struct request *rq)
 		new_bfqq->allocated++;
 		bfqq->allocated--;
 		new_bfqq->ref++;
-		bfq_clear_bfqq_just_created(bfqq);
 		/*
 		 * If the bic associated with the process
 		 * issuing this request still points to bfqq
@@ -4139,6 +4200,8 @@ static void __bfq_insert_request(struct bfq_data *bfqd, struct request *rq)
 		if (bic_to_bfqq(RQ_BIC(rq), 1) == bfqq)
 			bfq_merge_bfqqs(bfqd, RQ_BIC(rq),
 					bfqq, new_bfqq);
+
+		bfq_clear_bfqq_just_created(bfqq);
 		/*
 		 * rq is about to be enqueued into new_bfqq,
 		 * release rq reference on bfqq
@@ -4424,6 +4487,34 @@ static struct bfq_queue *bfq_get_bfqq_handle_split(struct bfq_data *bfqd,
 		else {
 			bfq_clear_bfqq_in_large_burst(bfqq);
 			if (bic->was_in_burst_list)
+				/*
+				 * If bfqq was in the current
+				 * burst list before being
+				 * merged, then we have to add
+				 * it back. And we do not need
+				 * to increase burst_size, as
+				 * we did not decrement
+				 * burst_size when we removed
+				 * bfqq from the burst list as
+				 * a consequence of a merge
+				 * (see comments in
+				 * bfq_put_queue). In this
+				 * respect, it would be rather
+				 * costly to know whether the
+				 * current burst list is still
+				 * the same burst list from
+				 * which bfqq was removed on
+				 * the merge. To avoid this
+				 * cost, if bfqq was in a
+				 * burst list, then we add
+				 * bfqq to the current burst
+				 * list without any further
+				 * check. This can cause
+				 * inappropriate insertions,
+				 * but rarely enough to not
+				 * harm the detection of large
+				 * bursts significantly.
+				 */
 				hlist_add_head(&bfqq->burst_list_node,
 					       &bfqd->burst_list);
 		}
@@ -4775,7 +4866,7 @@ static int bfq_init_queue(struct request_queue *q, struct elevator_type *e)
 	bfq_init_root_group(bfqd->root_group, bfqd);
 	bfq_init_entity(&bfqd->oom_bfqq.entity, bfqd->root_group);
 
-
+	wbt_disable_default(q);
 	return 0;
 
 out_free:

block/bio-integrity.c

@@ -485,11 +485,8 @@ EXPORT_SYMBOL(bioset_integrity_create);
 
 void bioset_integrity_free(struct bio_set *bs)
 {
-	if (bs->bio_integrity_pool)
-		mempool_destroy(bs->bio_integrity_pool);
-
-	if (bs->bvec_integrity_pool)
-		mempool_destroy(bs->bvec_integrity_pool);
+	mempool_destroy(bs->bio_integrity_pool);
+	mempool_destroy(bs->bvec_integrity_pool);
 }
 EXPORT_SYMBOL(bioset_integrity_free);
 

block/bio.c

@@ -400,7 +400,7 @@ static void punt_bios_to_rescuer(struct bio_set *bs)
 
 /**
  * bio_alloc_bioset - allocate a bio for I/O
- * @gfp_mask:   the GFP_ mask given to the slab allocator
+ * @gfp_mask:   the GFP_* mask given to the slab allocator
  * @nr_iovecs:	number of iovecs to pre-allocate
  * @bs:		the bio_set to allocate from.
  *
@@ -1931,11 +1931,8 @@ void bioset_free(struct bio_set *bs)
 	if (bs->rescue_workqueue)
 		destroy_workqueue(bs->rescue_workqueue);
 
-	if (bs->bio_pool)
-		mempool_destroy(bs->bio_pool);
-
-	if (bs->bvec_pool)
-		mempool_destroy(bs->bvec_pool);
+	mempool_destroy(bs->bio_pool);
+	mempool_destroy(bs->bvec_pool);
 
 	bioset_integrity_free(bs);
 	bio_put_slab(bs);
@@ -2035,37 +2032,6 @@ int bio_associate_blkcg(struct bio *bio, struct cgroup_subsys_state *blkcg_css)
 }
 EXPORT_SYMBOL_GPL(bio_associate_blkcg);
 
-/**
- * bio_associate_current - associate a bio with %current
- * @bio: target bio
- *
- * Associate @bio with %current if it hasn't been associated yet.  Block
- * layer will treat @bio as if it were issued by %current no matter which
- * task actually issues it.
- *
- * This function takes an extra reference of @task's io_context and blkcg
- * which will be put when @bio is released.  The caller must own @bio,
- * ensure %current->io_context exists, and is responsible for synchronizing
- * calls to this function.
- */
-int bio_associate_current(struct bio *bio)
-{
-	struct io_context *ioc;
-
-	if (bio->bi_css)
-		return -EBUSY;
-
-	ioc = current->io_context;
-	if (!ioc)
-		return -ENOENT;
-
-	get_io_context_active(ioc);
-	bio->bi_ioc = ioc;
-	bio->bi_css = task_get_css(current, io_cgrp_id);
-	return 0;
-}
-EXPORT_SYMBOL_GPL(bio_associate_current);
-
 /**
  * bio_disassociate_task - undo bio_associate_current()
  * @bio: target bio

block/blk-cgroup.c

@@ -1419,6 +1419,11 @@ int blkcg_policy_register(struct blkcg_policy *pol)
 	if (i >= BLKCG_MAX_POLS)
 		goto err_unlock;
 
+	/* Make sure cpd/pd_alloc_fn and cpd/pd_free_fn in pairs */
+	if ((!pol->cpd_alloc_fn ^ !pol->cpd_free_fn) ||
+		(!pol->pd_alloc_fn ^ !pol->pd_free_fn))
+		goto err_unlock;
+
 	/* register @pol */
 	pol->plid = i;
 	blkcg_policy[pol->plid] = pol;
@@ -1452,7 +1457,7 @@ int blkcg_policy_register(struct blkcg_policy *pol)
 	return 0;
 
 err_free_cpds:
-	if (pol->cpd_alloc_fn) {
+	if (pol->cpd_free_fn) {
 		list_for_each_entry(blkcg, &all_blkcgs, all_blkcgs_node) {
 			if (blkcg->cpd[pol->plid]) {
 				pol->cpd_free_fn(blkcg->cpd[pol->plid]);
@@ -1492,7 +1497,7 @@ void blkcg_policy_unregister(struct blkcg_policy *pol)
 	/* remove cpds and unregister */
 	mutex_lock(&blkcg_pol_mutex);
 
-	if (pol->cpd_alloc_fn) {
+	if (pol->cpd_free_fn) {
 		list_for_each_entry(blkcg, &all_blkcgs, all_blkcgs_node) {
 			if (blkcg->cpd[pol->plid]) {
 				pol->cpd_free_fn(blkcg->cpd[pol->plid]);

block/blk-core.c

@@ -333,11 +333,13 @@ EXPORT_SYMBOL(blk_stop_queue);
 void blk_sync_queue(struct request_queue *q)
 {
 	del_timer_sync(&q->timeout);
+	cancel_work_sync(&q->timeout_work);
 
 	if (q->mq_ops) {
 		struct blk_mq_hw_ctx *hctx;
 		int i;
 
+		cancel_delayed_work_sync(&q->requeue_work);
 		queue_for_each_hw_ctx(q, hctx, i)
 			cancel_delayed_work_sync(&hctx->run_work);
 	} else {
@@ -346,6 +348,37 @@ void blk_sync_queue(struct request_queue *q)
 }
 EXPORT_SYMBOL(blk_sync_queue);
 
+/**
+ * blk_set_preempt_only - set QUEUE_FLAG_PREEMPT_ONLY
+ * @q: request queue pointer
+ *
+ * Returns the previous value of the PREEMPT_ONLY flag - 0 if the flag was not
+ * set and 1 if the flag was already set.
+ */
+int blk_set_preempt_only(struct request_queue *q)
+{
+	unsigned long flags;
+	int res;
+
+	spin_lock_irqsave(q->queue_lock, flags);
+	res = queue_flag_test_and_set(QUEUE_FLAG_PREEMPT_ONLY, q);
+	spin_unlock_irqrestore(q->queue_lock, flags);
+
+	return res;
+}
+EXPORT_SYMBOL_GPL(blk_set_preempt_only);
+
+void blk_clear_preempt_only(struct request_queue *q)
+{
+	unsigned long flags;
+
+	spin_lock_irqsave(q->queue_lock, flags);
+	queue_flag_clear(QUEUE_FLAG_PREEMPT_ONLY, q);
+	wake_up_all(&q->mq_freeze_wq);
+	spin_unlock_irqrestore(q->queue_lock, flags);
+}
+EXPORT_SYMBOL_GPL(blk_clear_preempt_only);
+
 /**
  * __blk_run_queue_uncond - run a queue whether or not it has been stopped
  * @q:	The queue to run
@@ -610,6 +643,9 @@ void blk_set_queue_dying(struct request_queue *q)
 		}
 		spin_unlock_irq(q->queue_lock);
 	}
+
+	/* Make blk_queue_enter() reexamine the DYING flag. */
+	wake_up_all(&q->mq_freeze_wq);
 }
 EXPORT_SYMBOL_GPL(blk_set_queue_dying);
 
@@ -718,7 +754,7 @@ static void free_request_size(void *element, void *data)
 int blk_init_rl(struct request_list *rl, struct request_queue *q,
 		gfp_t gfp_mask)
 {
-	if (unlikely(rl->rq_pool))
+	if (unlikely(rl->rq_pool) || q->mq_ops)
 		return 0;
 
 	rl->q = q;
@@ -760,15 +796,38 @@ struct request_queue *blk_alloc_queue(gfp_t gfp_mask)
 }
 EXPORT_SYMBOL(blk_alloc_queue);
 
-int blk_queue_enter(struct request_queue *q, bool nowait)
+/**
+ * blk_queue_enter() - try to increase q->q_usage_counter
+ * @q: request queue pointer
+ * @flags: BLK_MQ_REQ_NOWAIT and/or BLK_MQ_REQ_PREEMPT
+ */
+int blk_queue_enter(struct request_queue *q, blk_mq_req_flags_t flags)
 {
+	const bool preempt = flags & BLK_MQ_REQ_PREEMPT;
+
 	while (true) {
+		bool success = false;
 		int ret;
 
-		if (percpu_ref_tryget_live(&q->q_usage_counter))
+		rcu_read_lock_sched();
+		if (percpu_ref_tryget_live(&q->q_usage_counter)) {
+			/*
+			 * The code that sets the PREEMPT_ONLY flag is
+			 * responsible for ensuring that that flag is globally
+			 * visible before the queue is unfrozen.
+			 */
+			if (preempt || !blk_queue_preempt_only(q)) {
+				success = true;
+			} else {
+				percpu_ref_put(&q->q_usage_counter);
+			}
+		}
+		rcu_read_unlock_sched();
+
+		if (success)
 			return 0;
 
-		if (nowait)
+		if (flags & BLK_MQ_REQ_NOWAIT)
 			return -EBUSY;
 
 		/*
@@ -781,7 +840,8 @@ int blk_queue_enter(struct request_queue *q, bool nowait)
 		smp_rmb();
 
 		ret = wait_event_interruptible(q->mq_freeze_wq,
-				!atomic_read(&q->mq_freeze_depth) ||
+				(atomic_read(&q->mq_freeze_depth) == 0 &&
+				 (preempt || !blk_queue_preempt_only(q))) ||
 				blk_queue_dying(q));
 		if (blk_queue_dying(q))
 			return -ENODEV;
@@ -844,6 +904,7 @@ struct request_queue *blk_alloc_queue_node(gfp_t gfp_mask, int node_id)
 	setup_timer(&q->backing_dev_info->laptop_mode_wb_timer,
 		    laptop_mode_timer_fn, (unsigned long) q);
 	setup_timer(&q->timeout, blk_rq_timed_out_timer, (unsigned long) q);
+	INIT_WORK(&q->timeout_work, NULL);
 	INIT_LIST_HEAD(&q->queue_head);
 	INIT_LIST_HEAD(&q->timeout_list);
 	INIT_LIST_HEAD(&q->icq_list);
@@ -1154,7 +1215,7 @@ int blk_update_nr_requests(struct request_queue *q, unsigned int nr)
  * @rl: request list to allocate from
  * @op: operation and flags
  * @bio: bio to allocate request for (can be %NULL)
- * @gfp_mask: allocation mask
+ * @flags: BLQ_MQ_REQ_* flags
  *
  * Get a free request from @q.  This function may fail under memory
  * pressure or if @q is dead.
@@ -1164,7 +1225,7 @@ int blk_update_nr_requests(struct request_queue *q, unsigned int nr)
  * Returns request pointer on success, with @q->queue_lock *not held*.
  */
 static struct request *__get_request(struct request_list *rl, unsigned int op,
-		struct bio *bio, gfp_t gfp_mask)
+				     struct bio *bio, blk_mq_req_flags_t flags)
 {
 	struct request_queue *q = rl->q;
 	struct request *rq;
@@ -1173,6 +1234,8 @@ static struct request *__get_request(struct request_list *rl, unsigned int op,
 	struct io_cq *icq = NULL;
 	const bool is_sync = op_is_sync(op);
 	int may_queue;
+	gfp_t gfp_mask = flags & BLK_MQ_REQ_NOWAIT ? GFP_ATOMIC :
+			 __GFP_DIRECT_RECLAIM;
 	req_flags_t rq_flags = RQF_ALLOCED;
 
 	lockdep_assert_held(q->queue_lock);
@@ -1255,6 +1318,8 @@ static struct request *__get_request(struct request_list *rl, unsigned int op,
 	blk_rq_set_rl(rq, rl);
 	rq->cmd_flags = op;
 	rq->rq_flags = rq_flags;
+	if (flags & BLK_MQ_REQ_PREEMPT)
+		rq->rq_flags |= RQF_PREEMPT;
 
 	/* init elvpriv */
 	if (rq_flags & RQF_ELVPRIV) {
@@ -1333,7 +1398,7 @@ rq_starved:
  * @q: request_queue to allocate request from
  * @op: operation and flags
  * @bio: bio to allocate request for (can be %NULL)
- * @gfp_mask: allocation mask
+ * @flags: BLK_MQ_REQ_* flags.
  *
  * Get a free request from @q.  If %__GFP_DIRECT_RECLAIM is set in @gfp_mask,
  * this function keeps retrying under memory pressure and fails iff @q is dead.
@@ -1343,7 +1408,7 @@ rq_starved:
  * Returns request pointer on success, with @q->queue_lock *not held*.
  */
 static struct request *get_request(struct request_queue *q, unsigned int op,
-		struct bio *bio, gfp_t gfp_mask)
+				   struct bio *bio, blk_mq_req_flags_t flags)
 {
 	const bool is_sync = op_is_sync(op);
 	DEFINE_WAIT(wait);
@@ -1355,7 +1420,7 @@ static struct request *get_request(struct request_queue *q, unsigned int op,
 
 	rl = blk_get_rl(q, bio);	/* transferred to @rq on success */
 retry:
-	rq = __get_request(rl, op, bio, gfp_mask);
+	rq = __get_request(rl, op, bio, flags);
 	if (!IS_ERR(rq))
 		return rq;
 
@@ -1364,7 +1429,7 @@ retry:
 		return ERR_PTR(-EAGAIN);
 	}
 
-	if (!gfpflags_allow_blocking(gfp_mask) || unlikely(blk_queue_dying(q))) {
+	if ((flags & BLK_MQ_REQ_NOWAIT) || unlikely(blk_queue_dying(q))) {
 		blk_put_rl(rl);
 		return rq;
 	}
@@ -1391,20 +1456,28 @@ retry:
 	goto retry;
 }
 
+/* flags: BLK_MQ_REQ_PREEMPT and/or BLK_MQ_REQ_NOWAIT. */
 static struct request *blk_old_get_request(struct request_queue *q,
-					   unsigned int op, gfp_t gfp_mask)
+				unsigned int op, blk_mq_req_flags_t flags)
 {
 	struct request *rq;
+	gfp_t gfp_mask = flags & BLK_MQ_REQ_NOWAIT ? GFP_ATOMIC :
+			 __GFP_DIRECT_RECLAIM;
+	int ret = 0;
 
 	WARN_ON_ONCE(q->mq_ops);
 
 	/* create ioc upfront */
 	create_io_context(gfp_mask, q->node);
 
+	ret = blk_queue_enter(q, flags);
+	if (ret)
+		return ERR_PTR(ret);
 	spin_lock_irq(q->queue_lock);
-	rq = get_request(q, op, NULL, gfp_mask);
+	rq = get_request(q, op, NULL, flags);
 	if (IS_ERR(rq)) {
 		spin_unlock_irq(q->queue_lock);
+		blk_queue_exit(q);
 		return rq;
 	}
 
@@ -1415,25 +1488,40 @@ static struct request *blk_old_get_request(struct request_queue *q,
 	return rq;
 }
 
-struct request *blk_get_request(struct request_queue *q, unsigned int op,
-				gfp_t gfp_mask)
+/**
+ * blk_get_request_flags - allocate a request
+ * @q: request queue to allocate a request for
+ * @op: operation (REQ_OP_*) and REQ_* flags, e.g. REQ_SYNC.
+ * @flags: BLK_MQ_REQ_* flags, e.g. BLK_MQ_REQ_NOWAIT.
+ */
+struct request *blk_get_request_flags(struct request_queue *q, unsigned int op,
+				      blk_mq_req_flags_t flags)
 {
 	struct request *req;
 
+	WARN_ON_ONCE(op & REQ_NOWAIT);
+	WARN_ON_ONCE(flags & ~(BLK_MQ_REQ_NOWAIT | BLK_MQ_REQ_PREEMPT));
+
 	if (q->mq_ops) {
-		req = blk_mq_alloc_request(q, op,
-			(gfp_mask & __GFP_DIRECT_RECLAIM) ?
-				0 : BLK_MQ_REQ_NOWAIT);
+		req = blk_mq_alloc_request(q, op, flags);
 		if (!IS_ERR(req) && q->mq_ops->initialize_rq_fn)
 			q->mq_ops->initialize_rq_fn(req);
 	} else {
-		req = blk_old_get_request(q, op, gfp_mask);
+		req = blk_old_get_request(q, op, flags);
 		if (!IS_ERR(req) && q->initialize_rq_fn)
 			q->initialize_rq_fn(req);
 	}
 
 	return req;
 }
+EXPORT_SYMBOL(blk_get_request_flags);
+
+struct request *blk_get_request(struct request_queue *q, unsigned int op,
+				gfp_t gfp_mask)
+{
+	return blk_get_request_flags(q, op, gfp_mask & __GFP_DIRECT_RECLAIM ?
+				     0 : BLK_MQ_REQ_NOWAIT);
+}
 EXPORT_SYMBOL(blk_get_request);
 
 /**
@@ -1576,6 +1664,7 @@ void __blk_put_request(struct request_queue *q, struct request *req)
 		blk_free_request(rl, req);
 		freed_request(rl, sync, rq_flags);
 		blk_put_rl(rl);
+		blk_queue_exit(q);
 	}
 }
 EXPORT_SYMBOL_GPL(__blk_put_request);
@@ -1857,8 +1946,10 @@ get_rq:
 	 * Grab a free request. This is might sleep but can not fail.
 	 * Returns with the queue unlocked.
 	 */
-	req = get_request(q, bio->bi_opf, bio, GFP_NOIO);
+	blk_queue_enter_live(q);
+	req = get_request(q, bio->bi_opf, bio, 0);
 	if (IS_ERR(req)) {
+		blk_queue_exit(q);
 		__wbt_done(q->rq_wb, wb_acct);
 		if (PTR_ERR(req) == -ENOMEM)
 			bio->bi_status = BLK_STS_RESOURCE;
@@ -2200,8 +2291,10 @@ blk_qc_t generic_make_request(struct bio *bio)
 	current->bio_list = bio_list_on_stack;
 	do {
 		struct request_queue *q = bio->bi_disk->queue;
+		blk_mq_req_flags_t flags = bio->bi_opf & REQ_NOWAIT ?
+			BLK_MQ_REQ_NOWAIT : 0;
 
-		if (likely(blk_queue_enter(q, bio->bi_opf & REQ_NOWAIT) == 0)) {
+		if (likely(blk_queue_enter(q, flags) == 0)) {
 			struct bio_list lower, same;
 
 			/* Create a fresh bio_list for all subordinate requests */
@@ -2241,6 +2334,40 @@ out:
 }
 EXPORT_SYMBOL(generic_make_request);
 
+/**
+ * direct_make_request - hand a buffer directly to its device driver for I/O
+ * @bio:  The bio describing the location in memory and on the device.
+ *
+ * This function behaves like generic_make_request(), but does not protect
+ * against recursion.  Must only be used if the called driver is known
+ * to not call generic_make_request (or direct_make_request) again from
+ * its make_request function.  (Calling direct_make_request again from
+ * a workqueue is perfectly fine as that doesn't recurse).
+ */
+blk_qc_t direct_make_request(struct bio *bio)
+{
+	struct request_queue *q = bio->bi_disk->queue;
+	bool nowait = bio->bi_opf & REQ_NOWAIT;
+	blk_qc_t ret;
+
+	if (!generic_make_request_checks(bio))
+		return BLK_QC_T_NONE;
+
+	if (unlikely(blk_queue_enter(q, nowait ? BLK_MQ_REQ_NOWAIT : 0))) {
+		if (nowait && !blk_queue_dying(q))
+			bio->bi_status = BLK_STS_AGAIN;
+		else
+			bio->bi_status = BLK_STS_IOERR;
+		bio_endio(bio);
+		return BLK_QC_T_NONE;
+	}
+
+	ret = q->make_request_fn(q, bio);
+	blk_queue_exit(q);
+	return ret;
+}
+EXPORT_SYMBOL_GPL(direct_make_request);
+
 /**
  * submit_bio - submit a bio to the block device layer for I/O
  * @bio: The &struct bio which describes the I/O
@@ -2285,6 +2412,17 @@ blk_qc_t submit_bio(struct bio *bio)
 }
 EXPORT_SYMBOL(submit_bio);
 
+bool blk_poll(struct request_queue *q, blk_qc_t cookie)
+{
+	if (!q->poll_fn || !blk_qc_t_valid(cookie))
+		return false;
+
+	if (current->plug)
+		blk_flush_plug_list(current->plug, false);
+	return q->poll_fn(q, cookie);
+}
+EXPORT_SYMBOL_GPL(blk_poll);
+
 /**
  * blk_cloned_rq_check_limits - Helper function to check a cloned request
  *                              for new the queue limits
@@ -2350,7 +2488,7 @@ blk_status_t blk_insert_cloned_request(struct request_queue *q, struct request *
 		 * bypass a potential scheduler on the bottom device for
 		 * insert.
 		 */
-		blk_mq_request_bypass_insert(rq);
+		blk_mq_request_bypass_insert(rq, true);
 		return BLK_STS_OK;
 	}
 
@@ -2464,20 +2602,22 @@ void blk_account_io_done(struct request *req)
  * Don't process normal requests when queue is suspended
  * or in the process of suspending/resuming
  */
-static struct request *blk_pm_peek_request(struct request_queue *q,
-					   struct request *rq)
+static bool blk_pm_allow_request(struct request *rq)
 {
-	if (q->dev && (q->rpm_status == RPM_SUSPENDED ||
-	    (q->rpm_status != RPM_ACTIVE && !(rq->rq_flags & RQF_PM))))
-		return NULL;
-	else
-		return rq;
+	switch (rq->q->rpm_status) {
+	case RPM_RESUMING:
+	case RPM_SUSPENDING:
+		return rq->rq_flags & RQF_PM;
+	case RPM_SUSPENDED:
+		return false;
+	}
+
+	return true;
 }
 #else
-static inline struct request *blk_pm_peek_request(struct request_queue *q,
-						  struct request *rq)
+static bool blk_pm_allow_request(struct request *rq)
 {
-	return rq;
+	return true;
 }
 #endif
 
@@ -2517,6 +2657,48 @@ void blk_account_io_start(struct request *rq, bool new_io)
 	part_stat_unlock();
 }
 
+static struct request *elv_next_request(struct request_queue *q)
+{
+	struct request *rq;
+	struct blk_flush_queue *fq = blk_get_flush_queue(q, NULL);
+
+	WARN_ON_ONCE(q->mq_ops);
+
+	while (1) {
+		list_for_each_entry(rq, &q->queue_head, queuelist) {
+			if (blk_pm_allow_request(rq))
+				return rq;
+
+			if (rq->rq_flags & RQF_SOFTBARRIER)
+				break;
+		}
+
+		/*
+		 * Flush request is running and flush request isn't queueable
+		 * in the drive, we can hold the queue till flush request is
+		 * finished. Even we don't do this, driver can't dispatch next
+		 * requests and will requeue them. And this can improve
+		 * throughput too. For example, we have request flush1, write1,
+		 * flush 2. flush1 is dispatched, then queue is hold, write1
+		 * isn't inserted to queue. After flush1 is finished, flush2
+		 * will be dispatched. Since disk cache is already clean,
+		 * flush2 will be finished very soon, so looks like flush2 is
+		 * folded to flush1.
+		 * Since the queue is hold, a flag is set to indicate the queue
+		 * should be restarted later. Please see flush_end_io() for
+		 * details.
+		 */
+		if (fq->flush_pending_idx != fq->flush_running_idx &&
+				!queue_flush_queueable(q)) {
+			fq->flush_queue_delayed = 1;
+			return NULL;
+		}
+		if (unlikely(blk_queue_bypass(q)) ||
+		    !q->elevator->type->ops.sq.elevator_dispatch_fn(q, 0))
+			return NULL;
+	}
+}
+
 /**
  * blk_peek_request - peek at the top of a request queue
  * @q: request queue to peek at
@@ -2538,12 +2720,7 @@ struct request *blk_peek_request(struct request_queue *q)
 	lockdep_assert_held(q->queue_lock);
 	WARN_ON_ONCE(q->mq_ops);
 
-	while ((rq = __elv_next_request(q)) != NULL) {
-
-		rq = blk_pm_peek_request(q, rq);
-		if (!rq)
-			break;
-
+	while ((rq = elv_next_request(q)) != NULL) {
 		if (!(rq->rq_flags & RQF_STARTED)) {
 			/*
 			 * This is the first time the device driver
@@ -2695,6 +2872,27 @@ struct request *blk_fetch_request(struct request_queue *q)
 }
 EXPORT_SYMBOL(blk_fetch_request);
 
+/*
+ * Steal bios from a request and add them to a bio list.
+ * The request must not have been partially completed before.
+ */
+void blk_steal_bios(struct bio_list *list, struct request *rq)
+{
+	if (rq->bio) {
+		if (list->tail)
+			list->tail->bi_next = rq->bio;
+		else
+			list->head = rq->bio;
+		list->tail = rq->biotail;
+
+		rq->bio = NULL;
+		rq->biotail = NULL;
+	}
+
+	rq->__data_len = 0;
+}
+EXPORT_SYMBOL_GPL(blk_steal_bios);
+
 /**
  * blk_update_request - Special helper function for request stacking drivers
  * @req:      the request being processed

block/blk-flush.c

@@ -231,8 +231,13 @@ static void flush_end_io(struct request *flush_rq, blk_status_t error)
 		/* release the tag's ownership to the req cloned from */
 		spin_lock_irqsave(&fq->mq_flush_lock, flags);
 		hctx = blk_mq_map_queue(q, flush_rq->mq_ctx->cpu);
-		blk_mq_tag_set_rq(hctx, flush_rq->tag, fq->orig_rq);
-		flush_rq->tag = -1;
+		if (!q->elevator) {
+			blk_mq_tag_set_rq(hctx, flush_rq->tag, fq->orig_rq);
+			flush_rq->tag = -1;
+		} else {
+			blk_mq_put_driver_tag_hctx(hctx, flush_rq);
+			flush_rq->internal_tag = -1;
+		}
 	}
 
 	running = &fq->flush_queue[fq->flush_running_idx];
@@ -318,19 +323,26 @@ static bool blk_kick_flush(struct request_queue *q, struct blk_flush_queue *fq)
 	blk_rq_init(q, flush_rq);
 
 	/*
-	 * Borrow tag from the first request since they can't
-	 * be in flight at the same time. And acquire the tag's
-	 * ownership for flush req.
+	 * In case of none scheduler, borrow tag from the first request
+	 * since they can't be in flight at the same time. And acquire
+	 * the tag's ownership for flush req.
+	 *
+	 * In case of IO scheduler, flush rq need to borrow scheduler tag
+	 * just for cheating put/get driver tag.
 	 */
 	if (q->mq_ops) {
 		struct blk_mq_hw_ctx *hctx;
 
 		flush_rq->mq_ctx = first_rq->mq_ctx;
-		flush_rq->tag = first_rq->tag;
-		fq->orig_rq = first_rq;
 
-		hctx = blk_mq_map_queue(q, first_rq->mq_ctx->cpu);
-		blk_mq_tag_set_rq(hctx, first_rq->tag, flush_rq);
+		if (!q->elevator) {
+			fq->orig_rq = first_rq;
+			flush_rq->tag = first_rq->tag;
+			hctx = blk_mq_map_queue(q, first_rq->mq_ctx->cpu);
+			blk_mq_tag_set_rq(hctx, first_rq->tag, flush_rq);
+		} else {
+			flush_rq->internal_tag = first_rq->internal_tag;
+		}
 	}
 
 	flush_rq->cmd_flags = REQ_OP_FLUSH | REQ_PREFLUSH;
@@ -394,6 +406,11 @@ static void mq_flush_data_end_io(struct request *rq, blk_status_t error)
 
 	hctx = blk_mq_map_queue(q, ctx->cpu);
 
+	if (q->elevator) {
+		WARN_ON(rq->tag < 0);
+		blk_mq_put_driver_tag_hctx(hctx, rq);
+	}
+
 	/*
 	 * After populating an empty queue, kick it to avoid stall.  Read
 	 * the comment in flush_end_io().
@@ -463,7 +480,7 @@ void blk_insert_flush(struct request *rq)
 	if ((policy & REQ_FSEQ_DATA) &&
 	    !(policy & (REQ_FSEQ_PREFLUSH | REQ_FSEQ_POSTFLUSH))) {
 		if (q->mq_ops)
-			blk_mq_sched_insert_request(rq, false, true, false, false);
+			blk_mq_request_bypass_insert(rq, false);
 		else
 			list_add_tail(&rq->queuelist, &q->queue_head);
 		return;

block/blk-lib.c

@@ -275,6 +275,40 @@ static unsigned int __blkdev_sectors_to_bio_pages(sector_t nr_sects)
 	return min(pages, (sector_t)BIO_MAX_PAGES);
 }
 
+static int __blkdev_issue_zero_pages(struct block_device *bdev,
+		sector_t sector, sector_t nr_sects, gfp_t gfp_mask,
+		struct bio **biop)
+{
+	struct request_queue *q = bdev_get_queue(bdev);
+	struct bio *bio = *biop;
+	int bi_size = 0;
+	unsigned int sz;
+
+	if (!q)
+		return -ENXIO;
+
+	while (nr_sects != 0) {
+		bio = next_bio(bio, __blkdev_sectors_to_bio_pages(nr_sects),
+			       gfp_mask);
+		bio->bi_iter.bi_sector = sector;
+		bio_set_dev(bio, bdev);
+		bio_set_op_attrs(bio, REQ_OP_WRITE, 0);
+
+		while (nr_sects != 0) {
+			sz = min((sector_t) PAGE_SIZE, nr_sects << 9);
+			bi_size = bio_add_page(bio, ZERO_PAGE(0), sz, 0);
+			nr_sects -= bi_size >> 9;
+			sector += bi_size >> 9;
+			if (bi_size < sz)
+				break;
+		}
+		cond_resched();
+	}
+
+	*biop = bio;
+	return 0;
+}
+
 /**
  * __blkdev_issue_zeroout - generate number of zero filed write bios
  * @bdev:	blockdev to issue
@@ -288,12 +322,6 @@ static unsigned int __blkdev_sectors_to_bio_pages(sector_t nr_sects)
  *  Zero-fill a block range, either using hardware offload or by explicitly
  *  writing zeroes to the device.
  *
- *  Note that this function may fail with -EOPNOTSUPP if the driver signals
- *  zeroing offload support, but the device fails to process the command (for
- *  some devices there is no non-destructive way to verify whether this
- *  operation is actually supported).  In this case the caller should call
- *  retry the call to blkdev_issue_zeroout() and the fallback path will be used.
- *
  *  If a device is using logical block provisioning, the underlying space will
  *  not be released if %flags contains BLKDEV_ZERO_NOUNMAP.
  *
@@ -305,9 +333,6 @@ int __blkdev_issue_zeroout(struct block_device *bdev, sector_t sector,
 		unsigned flags)
 {
 	int ret;
-	int bi_size = 0;
-	struct bio *bio = *biop;
-	unsigned int sz;
 	sector_t bs_mask;
 
 	bs_mask = (bdev_logical_block_size(bdev) >> 9) - 1;
@@ -317,30 +342,10 @@ int __blkdev_issue_zeroout(struct block_device *bdev, sector_t sector,
 	ret = __blkdev_issue_write_zeroes(bdev, sector, nr_sects, gfp_mask,
 			biop, flags);
 	if (ret != -EOPNOTSUPP || (flags & BLKDEV_ZERO_NOFALLBACK))
-		goto out;
-
-	ret = 0;
-	while (nr_sects != 0) {
-		bio = next_bio(bio, __blkdev_sectors_to_bio_pages(nr_sects),
-			       gfp_mask);
-		bio->bi_iter.bi_sector = sector;
-		bio_set_dev(bio, bdev);
-		bio_set_op_attrs(bio, REQ_OP_WRITE, 0);
-
-		while (nr_sects != 0) {
-			sz = min((sector_t) PAGE_SIZE, nr_sects << 9);
-			bi_size = bio_add_page(bio, ZERO_PAGE(0), sz, 0);
-			nr_sects -= bi_size >> 9;
-			sector += bi_size >> 9;
-			if (bi_size < sz)
-				break;
-		}
-		cond_resched();
-	}
+		return ret;
 
-	*biop = bio;
-out:
-	return ret;
+	return __blkdev_issue_zero_pages(bdev, sector, nr_sects, gfp_mask,
+					 biop);
 }
 EXPORT_SYMBOL(__blkdev_issue_zeroout);
 
@@ -360,18 +365,49 @@ EXPORT_SYMBOL(__blkdev_issue_zeroout);
 int blkdev_issue_zeroout(struct block_device *bdev, sector_t sector,
 		sector_t nr_sects, gfp_t gfp_mask, unsigned flags)
 {
-	int ret;
-	struct bio *bio = NULL;
+	int ret = 0;
+	sector_t bs_mask;
+	struct bio *bio;
 	struct blk_plug plug;
+	bool try_write_zeroes = !!bdev_write_zeroes_sectors(bdev);
 
+	bs_mask = (bdev_logical_block_size(bdev) >> 9) - 1;
+	if ((sector | nr_sects) & bs_mask)
+		return -EINVAL;
+
+retry:
+	bio = NULL;
 	blk_start_plug(&plug);
-	ret = __blkdev_issue_zeroout(bdev, sector, nr_sects, gfp_mask,
-			&bio, flags);
+	if (try_write_zeroes) {
+		ret = __blkdev_issue_write_zeroes(bdev, sector, nr_sects,
+						  gfp_mask, &bio, flags);
+	} else if (!(flags & BLKDEV_ZERO_NOFALLBACK)) {
+		ret = __blkdev_issue_zero_pages(bdev, sector, nr_sects,
+						gfp_mask, &bio);
+	} else {
+		/* No zeroing offload support */
+		ret = -EOPNOTSUPP;
+	}
 	if (ret == 0 && bio) {
 		ret = submit_bio_wait(bio);
 		bio_put(bio);
 	}
 	blk_finish_plug(&plug);
+	if (ret && try_write_zeroes) {
+		if (!(flags & BLKDEV_ZERO_NOFALLBACK)) {
+			try_write_zeroes = false;
+			goto retry;
+		}
+		if (!bdev_write_zeroes_sectors(bdev)) {
+			/*
+			 * Zeroing offload support was indicated, but the
+			 * device reported ILLEGAL REQUEST (for some devices
+			 * there is no non-destructive way to verify whether
+			 * WRITE ZEROES is actually supported).
+			 */
+			ret = -EOPNOTSUPP;
+		}
+	}
 
 	return ret;
 }

block/blk-mq-debugfs.c

@@ -54,7 +54,6 @@ static const char *const blk_queue_flag_name[] = {
 	QUEUE_FLAG_NAME(NOMERGES),
 	QUEUE_FLAG_NAME(SAME_COMP),
 	QUEUE_FLAG_NAME(FAIL_IO),
-	QUEUE_FLAG_NAME(STACKABLE),
 	QUEUE_FLAG_NAME(NONROT),
 	QUEUE_FLAG_NAME(IO_STAT),
 	QUEUE_FLAG_NAME(DISCARD),
@@ -75,6 +74,7 @@ static const char *const blk_queue_flag_name[] = {
 	QUEUE_FLAG_NAME(REGISTERED),
 	QUEUE_FLAG_NAME(SCSI_PASSTHROUGH),
 	QUEUE_FLAG_NAME(QUIESCED),
+	QUEUE_FLAG_NAME(PREEMPT_ONLY),
 };
 #undef QUEUE_FLAG_NAME
 
@@ -180,7 +180,6 @@ static const char *const hctx_state_name[] = {
 	HCTX_STATE_NAME(STOPPED),
 	HCTX_STATE_NAME(TAG_ACTIVE),
 	HCTX_STATE_NAME(SCHED_RESTART),
-	HCTX_STATE_NAME(TAG_WAITING),
 	HCTX_STATE_NAME(START_ON_RUN),
 };
 #undef HCTX_STATE_NAME

block/blk-mq-sched.c

@@ -81,20 +81,103 @@ static bool blk_mq_sched_restart_hctx(struct blk_mq_hw_ctx *hctx)
 	} else
 		clear_bit(BLK_MQ_S_SCHED_RESTART, &hctx->state);
 
-	if (blk_mq_hctx_has_pending(hctx)) {
-		blk_mq_run_hw_queue(hctx, true);
-		return true;
-	}
+	return blk_mq_run_hw_queue(hctx, true);
+}
 
-	return false;
+/*
+ * Only SCSI implements .get_budget and .put_budget, and SCSI restarts
+ * its queue by itself in its completion handler, so we don't need to
+ * restart queue if .get_budget() returns BLK_STS_NO_RESOURCE.
+ */
+static void blk_mq_do_dispatch_sched(struct blk_mq_hw_ctx *hctx)
+{
+	struct request_queue *q = hctx->queue;
+	struct elevator_queue *e = q->elevator;
+	LIST_HEAD(rq_list);
+
+	do {
+		struct request *rq;
+
+		if (e->type->ops.mq.has_work &&
+				!e->type->ops.mq.has_work(hctx))
+			break;
+
+		if (!blk_mq_get_dispatch_budget(hctx))
+			break;
+
+		rq = e->type->ops.mq.dispatch_request(hctx);
+		if (!rq) {
+			blk_mq_put_dispatch_budget(hctx);
+			break;
+		}
+
+		/*
+		 * Now this rq owns the budget which has to be released
+		 * if this rq won't be queued to driver via .queue_rq()
+		 * in blk_mq_dispatch_rq_list().
+		 */
+		list_add(&rq->queuelist, &rq_list);
+	} while (blk_mq_dispatch_rq_list(q, &rq_list, true));
 }
 
+static struct blk_mq_ctx *blk_mq_next_ctx(struct blk_mq_hw_ctx *hctx,
+					  struct blk_mq_ctx *ctx)
+{
+	unsigned idx = ctx->index_hw;
+
+	if (++idx == hctx->nr_ctx)
+		idx = 0;
+
+	return hctx->ctxs[idx];
+}
+
+/*
+ * Only SCSI implements .get_budget and .put_budget, and SCSI restarts
+ * its queue by itself in its completion handler, so we don't need to
+ * restart queue if .get_budget() returns BLK_STS_NO_RESOURCE.
+ */
+static void blk_mq_do_dispatch_ctx(struct blk_mq_hw_ctx *hctx)
+{
+	struct request_queue *q = hctx->queue;
+	LIST_HEAD(rq_list);
+	struct blk_mq_ctx *ctx = READ_ONCE(hctx->dispatch_from);
+
+	do {
+		struct request *rq;
+
+		if (!sbitmap_any_bit_set(&hctx->ctx_map))
+			break;
+
+		if (!blk_mq_get_dispatch_budget(hctx))
+			break;
+
+		rq = blk_mq_dequeue_from_ctx(hctx, ctx);
+		if (!rq) {
+			blk_mq_put_dispatch_budget(hctx);
+			break;
+		}
+
+		/*
+		 * Now this rq owns the budget which has to be released
+		 * if this rq won't be queued to driver via .queue_rq()
+		 * in blk_mq_dispatch_rq_list().
+		 */
+		list_add(&rq->queuelist, &rq_list);
+
+		/* round robin for fair dispatch */
+		ctx = blk_mq_next_ctx(hctx, rq->mq_ctx);
+
+	} while (blk_mq_dispatch_rq_list(q, &rq_list, true));
+
+	WRITE_ONCE(hctx->dispatch_from, ctx);
+}
+
+/* return true if hw queue need to be run again */
 void blk_mq_sched_dispatch_requests(struct blk_mq_hw_ctx *hctx)
 {
 	struct request_queue *q = hctx->queue;
 	struct elevator_queue *e = q->elevator;
 	const bool has_sched_dispatch = e && e->type->ops.mq.dispatch_request;
-	bool did_work = false;
 	LIST_HEAD(rq_list);
 
 	/* RCU or SRCU read lock is needed before checking quiesced flag */
@@ -122,29 +205,34 @@ void blk_mq_sched_dispatch_requests(struct blk_mq_hw_ctx *hctx)
 	 * scheduler, we can no longer merge or sort them. So it's best to
 	 * leave them there for as long as we can. Mark the hw queue as
 	 * needing a restart in that case.
+	 *
+	 * We want to dispatch from the scheduler if there was nothing
+	 * on the dispatch list or we were able to dispatch from the
+	 * dispatch list.
 	 */
 	if (!list_empty(&rq_list)) {
 		blk_mq_sched_mark_restart_hctx(hctx);
-		did_work = blk_mq_dispatch_rq_list(q, &rq_list);
-	} else if (!has_sched_dispatch) {
+		if (blk_mq_dispatch_rq_list(q, &rq_list, false)) {
+			if (has_sched_dispatch)
+				blk_mq_do_dispatch_sched(hctx);
+			else
+				blk_mq_do_dispatch_ctx(hctx);
+		}
+	} else if (has_sched_dispatch) {
+		blk_mq_do_dispatch_sched(hctx);
+	} else if (q->mq_ops->get_budget) {
+		/*
+		 * If we need to get budget before queuing request, we
+		 * dequeue request one by one from sw queue for avoiding
+		 * to mess up I/O merge when dispatch runs out of resource.
+		 *
+		 * TODO: get more budgets, and dequeue more requests in
+		 * one time.
+		 */
+		blk_mq_do_dispatch_ctx(hctx);
+	} else {
 		blk_mq_flush_busy_ctxs(hctx, &rq_list);
-		blk_mq_dispatch_rq_list(q, &rq_list);
-	}
-
-	/*
-	 * We want to dispatch from the scheduler if we had no work left
-	 * on the dispatch list, OR if we did have work but weren't able
-	 * to make progress.
-	 */
-	if (!did_work && has_sched_dispatch) {
-		do {
-			struct request *rq;
-
-			rq = e->type->ops.mq.dispatch_request(hctx);
-			if (!rq)
-				break;
-			list_add(&rq->queuelist, &rq_list);
-		} while (blk_mq_dispatch_rq_list(q, &rq_list));
+		blk_mq_dispatch_rq_list(q, &rq_list, false);
 	}
 }
 
@@ -260,21 +348,21 @@ void blk_mq_sched_request_inserted(struct request *rq)
 EXPORT_SYMBOL_GPL(blk_mq_sched_request_inserted);
 
 static bool blk_mq_sched_bypass_insert(struct blk_mq_hw_ctx *hctx,
+				       bool has_sched,
 				       struct request *rq)
 {
-	if (rq->tag == -1) {
-		rq->rq_flags |= RQF_SORTED;
-		return false;
+	/* dispatch flush rq directly */
+	if (rq->rq_flags & RQF_FLUSH_SEQ) {
+		spin_lock(&hctx->lock);
+		list_add(&rq->queuelist, &hctx->dispatch);
+		spin_unlock(&hctx->lock);
+		return true;
 	}
 
-	/*
-	 * If we already have a real request tag, send directly to
-	 * the dispatch list.
-	 */
-	spin_lock(&hctx->lock);
-	list_add(&rq->queuelist, &hctx->dispatch);
-	spin_unlock(&hctx->lock);
-	return true;
+	if (has_sched)
+		rq->rq_flags |= RQF_SORTED;
+
+	return false;
 }
 
 /**
@@ -339,21 +427,6 @@ done:
 	}
 }
 
-/*
- * Add flush/fua to the queue. If we fail getting a driver tag, then
- * punt to the requeue list. Requeue will re-invoke us from a context
- * that's safe to block from.
- */
-static void blk_mq_sched_insert_flush(struct blk_mq_hw_ctx *hctx,
-				      struct request *rq, bool can_block)
-{
-	if (blk_mq_get_driver_tag(rq, &hctx, can_block)) {
-		blk_insert_flush(rq);
-		blk_mq_run_hw_queue(hctx, true);
-	} else
-		blk_mq_add_to_requeue_list(rq, false, true);
-}
-
 void blk_mq_sched_insert_request(struct request *rq, bool at_head,
 				 bool run_queue, bool async, bool can_block)
 {
@@ -362,12 +435,15 @@ void blk_mq_sched_insert_request(struct request *rq, bool at_head,
 	struct blk_mq_ctx *ctx = rq->mq_ctx;
 	struct blk_mq_hw_ctx *hctx = blk_mq_map_queue(q, ctx->cpu);
 
-	if (rq->tag == -1 && op_is_flush(rq->cmd_flags)) {
-		blk_mq_sched_insert_flush(hctx, rq, can_block);
-		return;
+	/* flush rq in flush machinery need to be dispatched directly */
+	if (!(rq->rq_flags & RQF_FLUSH_SEQ) && op_is_flush(rq->cmd_flags)) {
+		blk_insert_flush(rq);
+		goto run;
 	}
 
-	if (e && blk_mq_sched_bypass_insert(hctx, rq))
+	WARN_ON(e && (rq->tag != -1));
+
+	if (blk_mq_sched_bypass_insert(hctx, !!e, rq))
 		goto run;
 
 	if (e && e->type->ops.mq.insert_requests) {
@@ -393,23 +469,6 @@ void blk_mq_sched_insert_requests(struct request_queue *q,
 	struct blk_mq_hw_ctx *hctx = blk_mq_map_queue(q, ctx->cpu);
 	struct elevator_queue *e = hctx->queue->elevator;
 
-	if (e) {
-		struct request *rq, *next;
-
-		/*
-		 * We bypass requests that already have a driver tag assigned,
-		 * which should only be flushes. Flushes are only ever inserted
-		 * as single requests, so we shouldn't ever hit the
-		 * WARN_ON_ONCE() below (but let's handle it just in case).
-		 */
-		list_for_each_entry_safe(rq, next, list, queuelist) {
-			if (WARN_ON_ONCE(rq->tag != -1)) {
-				list_del_init(&rq->queuelist);
-				blk_mq_sched_bypass_insert(hctx, rq);
-			}
-		}
-	}
-
 	if (e && e->type->ops.mq.insert_requests)
 		e->type->ops.mq.insert_requests(hctx, list, false);
 	else

block/blk-mq-tag.c

@@ -298,12 +298,12 @@ void blk_mq_tagset_busy_iter(struct blk_mq_tag_set *tagset,
 }
 EXPORT_SYMBOL(blk_mq_tagset_busy_iter);
 
-int blk_mq_reinit_tagset(struct blk_mq_tag_set *set,
-			 int (reinit_request)(void *, struct request *))
+int blk_mq_tagset_iter(struct blk_mq_tag_set *set, void *data,
+			 int (fn)(void *, struct request *))
 {
 	int i, j, ret = 0;
 
-	if (WARN_ON_ONCE(!reinit_request))
+	if (WARN_ON_ONCE(!fn))
 		goto out;
 
 	for (i = 0; i < set->nr_hw_queues; i++) {
@@ -316,8 +316,7 @@ int blk_mq_reinit_tagset(struct blk_mq_tag_set *set,
 			if (!tags->static_rqs[j])
 				continue;
 
-			ret = reinit_request(set->driver_data,
-					     tags->static_rqs[j]);
+			ret = fn(data, tags->static_rqs[j]);
 			if (ret)
 				goto out;
 		}
@@ -326,7 +325,7 @@ int blk_mq_reinit_tagset(struct blk_mq_tag_set *set,
 out:
 	return ret;
 }
-EXPORT_SYMBOL_GPL(blk_mq_reinit_tagset);
+EXPORT_SYMBOL_GPL(blk_mq_tagset_iter);
 
 void blk_mq_queue_tag_busy_iter(struct request_queue *q, busy_iter_fn *fn,
 		void *priv)

block/blk-mq-tag.h

@@ -44,14 +44,9 @@ static inline struct sbq_wait_state *bt_wait_ptr(struct sbitmap_queue *bt,
 	return sbq_wait_ptr(bt, &hctx->wait_index);
 }
 
-enum {
-	BLK_MQ_TAG_CACHE_MIN	= 1,
-	BLK_MQ_TAG_CACHE_MAX	= 64,
-};
-
 enum {
 	BLK_MQ_TAG_FAIL		= -1U,
-	BLK_MQ_TAG_MIN		= BLK_MQ_TAG_CACHE_MIN,
+	BLK_MQ_TAG_MIN		= 1,
 	BLK_MQ_TAG_MAX		= BLK_MQ_TAG_FAIL - 1,
 };
 

block/blk-mq.c

@@ -37,6 +37,7 @@
 #include "blk-wbt.h"
 #include "blk-mq-sched.h"
 
+static bool blk_mq_poll(struct request_queue *q, blk_qc_t cookie);
 static void blk_mq_poll_stats_start(struct request_queue *q);
 static void blk_mq_poll_stats_fn(struct blk_stat_callback *cb);
 
@@ -60,10 +61,10 @@ static int blk_mq_poll_stats_bkt(const struct request *rq)
 /*
  * Check if any of the ctx's have pending work in this hardware queue
  */
-bool blk_mq_hctx_has_pending(struct blk_mq_hw_ctx *hctx)
+static bool blk_mq_hctx_has_pending(struct blk_mq_hw_ctx *hctx)
 {
-	return sbitmap_any_bit_set(&hctx->ctx_map) ||
-			!list_empty_careful(&hctx->dispatch) ||
+	return !list_empty_careful(&hctx->dispatch) ||
+		sbitmap_any_bit_set(&hctx->ctx_map) ||
 			blk_mq_sched_has_work(hctx);
 }
 
@@ -125,7 +126,8 @@ void blk_freeze_queue_start(struct request_queue *q)
 	freeze_depth = atomic_inc_return(&q->mq_freeze_depth);
 	if (freeze_depth == 1) {
 		percpu_ref_kill(&q->q_usage_counter);
-		blk_mq_run_hw_queues(q, false);
+		if (q->mq_ops)
+			blk_mq_run_hw_queues(q, false);
 	}
 }
 EXPORT_SYMBOL_GPL(blk_freeze_queue_start);
@@ -255,13 +257,6 @@ void blk_mq_wake_waiters(struct request_queue *q)
 	queue_for_each_hw_ctx(q, hctx, i)
 		if (blk_mq_hw_queue_mapped(hctx))
 			blk_mq_tag_wakeup_all(hctx->tags, true);
-
-	/*
-	 * If we are called because the queue has now been marked as
-	 * dying, we need to ensure that processes currently waiting on
-	 * the queue are notified as well.
-	 */
-	wake_up_all(&q->mq_freeze_wq);
 }
 
 bool blk_mq_can_queue(struct blk_mq_hw_ctx *hctx)
@@ -296,6 +291,8 @@ static struct request *blk_mq_rq_ctx_init(struct blk_mq_alloc_data *data,
 	rq->q = data->q;
 	rq->mq_ctx = data->ctx;
 	rq->cmd_flags = op;
+	if (data->flags & BLK_MQ_REQ_PREEMPT)
+		rq->rq_flags |= RQF_PREEMPT;
 	if (blk_queue_io_stat(data->q))
 		rq->rq_flags |= RQF_IO_STAT;
 	/* do not touch atomic flags, it needs atomic ops against the timer */
@@ -336,12 +333,14 @@ static struct request *blk_mq_get_request(struct request_queue *q,
 	struct elevator_queue *e = q->elevator;
 	struct request *rq;
 	unsigned int tag;
-	struct blk_mq_ctx *local_ctx = NULL;
+	bool put_ctx_on_error = false;
 
 	blk_queue_enter_live(q);
 	data->q = q;
-	if (likely(!data->ctx))
-		data->ctx = local_ctx = blk_mq_get_ctx(q);
+	if (likely(!data->ctx)) {
+		data->ctx = blk_mq_get_ctx(q);
+		put_ctx_on_error = true;
+	}
 	if (likely(!data->hctx))
 		data->hctx = blk_mq_map_queue(q, data->ctx->cpu);
 	if (op & REQ_NOWAIT)
@@ -360,8 +359,8 @@ static struct request *blk_mq_get_request(struct request_queue *q,
 
 	tag = blk_mq_get_tag(data);
 	if (tag == BLK_MQ_TAG_FAIL) {
-		if (local_ctx) {
-			blk_mq_put_ctx(local_ctx);
+		if (put_ctx_on_error) {
+			blk_mq_put_ctx(data->ctx);
 			data->ctx = NULL;
 		}
 		blk_queue_exit(q);
@@ -384,13 +383,13 @@ static struct request *blk_mq_get_request(struct request_queue *q,
 }
 
 struct request *blk_mq_alloc_request(struct request_queue *q, unsigned int op,
-		unsigned int flags)
+		blk_mq_req_flags_t flags)
 {
 	struct blk_mq_alloc_data alloc_data = { .flags = flags };
 	struct request *rq;
 	int ret;
 
-	ret = blk_queue_enter(q, flags & BLK_MQ_REQ_NOWAIT);
+	ret = blk_queue_enter(q, flags);
 	if (ret)
 		return ERR_PTR(ret);
 
@@ -410,7 +409,7 @@ struct request *blk_mq_alloc_request(struct request_queue *q, unsigned int op,
 EXPORT_SYMBOL(blk_mq_alloc_request);
 
 struct request *blk_mq_alloc_request_hctx(struct request_queue *q,
-		unsigned int op, unsigned int flags, unsigned int hctx_idx)
+	unsigned int op, blk_mq_req_flags_t flags, unsigned int hctx_idx)
 {
 	struct blk_mq_alloc_data alloc_data = { .flags = flags };
 	struct request *rq;
@@ -429,7 +428,7 @@ struct request *blk_mq_alloc_request_hctx(struct request_queue *q,
 	if (hctx_idx >= q->nr_hw_queues)
 		return ERR_PTR(-EIO);
 
-	ret = blk_queue_enter(q, true);
+	ret = blk_queue_enter(q, flags);
 	if (ret)
 		return ERR_PTR(ret);
 
@@ -476,8 +475,14 @@ void blk_mq_free_request(struct request *rq)
 	if (rq->rq_flags & RQF_MQ_INFLIGHT)
 		atomic_dec(&hctx->nr_active);
 
+	if (unlikely(laptop_mode && !blk_rq_is_passthrough(rq)))
+		laptop_io_completion(q->backing_dev_info);
+
 	wbt_done(q->rq_wb, &rq->issue_stat);
 
+	if (blk_rq_rl(rq))
+		blk_put_rl(blk_rq_rl(rq));
+
 	clear_bit(REQ_ATOM_STARTED, &rq->atomic_flags);
 	clear_bit(REQ_ATOM_POLL_SLEPT, &rq->atomic_flags);
 	if (rq->tag != -1)
@@ -593,22 +598,32 @@ void blk_mq_start_request(struct request *rq)
 
 	blk_add_timer(rq);
 
-	/*
-	 * Ensure that ->deadline is visible before set the started
-	 * flag and clear the completed flag.
-	 */
-	smp_mb__before_atomic();
+	WARN_ON_ONCE(test_bit(REQ_ATOM_STARTED, &rq->atomic_flags));
 
 	/*
 	 * Mark us as started and clear complete. Complete might have been
 	 * set if requeue raced with timeout, which then marked it as
 	 * complete. So be sure to clear complete again when we start
 	 * the request, otherwise we'll ignore the completion event.
+	 *
+	 * Ensure that ->deadline is visible before we set STARTED, such that
+	 * blk_mq_check_expired() is guaranteed to observe our ->deadline when
+	 * it observes STARTED.
 	 */
-	if (!test_bit(REQ_ATOM_STARTED, &rq->atomic_flags))
-		set_bit(REQ_ATOM_STARTED, &rq->atomic_flags);
-	if (test_bit(REQ_ATOM_COMPLETE, &rq->atomic_flags))
+	smp_wmb();
+	set_bit(REQ_ATOM_STARTED, &rq->atomic_flags);
+	if (test_bit(REQ_ATOM_COMPLETE, &rq->atomic_flags)) {
+		/*
+		 * Coherence order guarantees these consecutive stores to a
+		 * single variable propagate in the specified order. Thus the
+		 * clear_bit() is ordered _after_ the set bit. See
+		 * blk_mq_check_expired().
+		 *
+		 * (the bits must be part of the same byte for this to be
+		 * true).
+		 */
 		clear_bit(REQ_ATOM_COMPLETE, &rq->atomic_flags);
+	}
 
 	if (q->dma_drain_size && blk_rq_bytes(rq)) {
 		/*
@@ -634,6 +649,8 @@ static void __blk_mq_requeue_request(struct request *rq)
 {
 	struct request_queue *q = rq->q;
 
+	blk_mq_put_driver_tag(rq);
+
 	trace_block_rq_requeue(q, rq);
 	wbt_requeue(q->rq_wb, &rq->issue_stat);
 	blk_mq_sched_requeue_request(rq);
@@ -690,7 +707,7 @@ void blk_mq_add_to_requeue_list(struct request *rq, bool at_head,
 
 	/*
 	 * We abuse this flag that is otherwise used by the I/O scheduler to
-	 * request head insertation from the workqueue.
+	 * request head insertion from the workqueue.
 	 */
 	BUG_ON(rq->rq_flags & RQF_SOFTBARRIER);
 
@@ -778,10 +795,19 @@ static void blk_mq_check_expired(struct blk_mq_hw_ctx *hctx,
 		struct request *rq, void *priv, bool reserved)
 {
 	struct blk_mq_timeout_data *data = priv;
+	unsigned long deadline;
 
 	if (!test_bit(REQ_ATOM_STARTED, &rq->atomic_flags))
 		return;
 
+	/*
+	 * Ensures that if we see STARTED we must also see our
+	 * up-to-date deadline, see blk_mq_start_request().
+	 */
+	smp_rmb();
+
+	deadline = READ_ONCE(rq->deadline);
+
 	/*
 	 * The rq being checked may have been freed and reallocated
 	 * out already here, we avoid this race by checking rq->deadline
@@ -795,11 +821,20 @@ static void blk_mq_check_expired(struct blk_mq_hw_ctx *hctx,
 	 *   and clearing the flag in blk_mq_start_request(), so
 	 *   this rq won't be timed out too.
 	 */
-	if (time_after_eq(jiffies, rq->deadline)) {
-		if (!blk_mark_rq_complete(rq))
+	if (time_after_eq(jiffies, deadline)) {
+		if (!blk_mark_rq_complete(rq)) {
+			/*
+			 * Again coherence order ensures that consecutive reads
+			 * from the same variable must be in that order. This
+			 * ensures that if we see COMPLETE clear, we must then
+			 * see STARTED set and we'll ignore this timeout.
+			 *
+			 * (There's also the MB implied by the test_and_clear())
+			 */
 			blk_mq_rq_timed_out(rq, reserved);
-	} else if (!data->next_set || time_after(data->next, rq->deadline)) {
-		data->next = rq->deadline;
+		}
+	} else if (!data->next_set || time_after(data->next, deadline)) {
+		data->next = deadline;
 		data->next_set = 1;
 	}
 }
@@ -880,6 +915,45 @@ void blk_mq_flush_busy_ctxs(struct blk_mq_hw_ctx *hctx, struct list_head *list)
 }
 EXPORT_SYMBOL_GPL(blk_mq_flush_busy_ctxs);
 
+struct dispatch_rq_data {
+	struct blk_mq_hw_ctx *hctx;
+	struct request *rq;
+};
+
+static bool dispatch_rq_from_ctx(struct sbitmap *sb, unsigned int bitnr,
+		void *data)
+{
+	struct dispatch_rq_data *dispatch_data = data;
+	struct blk_mq_hw_ctx *hctx = dispatch_data->hctx;
+	struct blk_mq_ctx *ctx = hctx->ctxs[bitnr];
+
+	spin_lock(&ctx->lock);
+	if (unlikely(!list_empty(&ctx->rq_list))) {
+		dispatch_data->rq = list_entry_rq(ctx->rq_list.next);
+		list_del_init(&dispatch_data->rq->queuelist);
+		if (list_empty(&ctx->rq_list))
+			sbitmap_clear_bit(sb, bitnr);
+	}
+	spin_unlock(&ctx->lock);
+
+	return !dispatch_data->rq;
+}
+
+struct request *blk_mq_dequeue_from_ctx(struct blk_mq_hw_ctx *hctx,
+					struct blk_mq_ctx *start)
+{
+	unsigned off = start ? start->index_hw : 0;
+	struct dispatch_rq_data data = {
+		.hctx = hctx,
+		.rq   = NULL,
+	};
+
+	__sbitmap_for_each_set(&hctx->ctx_map, off,
+			       dispatch_rq_from_ctx, &data);
+
+	return data.rq;
+}
+
 static inline unsigned int queued_to_index(unsigned int queued)
 {
 	if (!queued)
@@ -920,109 +994,95 @@ done:
 	return rq->tag != -1;
 }
 
-static void __blk_mq_put_driver_tag(struct blk_mq_hw_ctx *hctx,
-				    struct request *rq)
-{
-	blk_mq_put_tag(hctx, hctx->tags, rq->mq_ctx, rq->tag);
-	rq->tag = -1;
-
-	if (rq->rq_flags & RQF_MQ_INFLIGHT) {
-		rq->rq_flags &= ~RQF_MQ_INFLIGHT;
-		atomic_dec(&hctx->nr_active);
-	}
-}
-
-static void blk_mq_put_driver_tag_hctx(struct blk_mq_hw_ctx *hctx,
-				       struct request *rq)
-{
-	if (rq->tag == -1 || rq->internal_tag == -1)
-		return;
-
-	__blk_mq_put_driver_tag(hctx, rq);
-}
-
-static void blk_mq_put_driver_tag(struct request *rq)
+static int blk_mq_dispatch_wake(wait_queue_entry_t *wait, unsigned mode,
+				int flags, void *key)
 {
 	struct blk_mq_hw_ctx *hctx;
 
-	if (rq->tag == -1 || rq->internal_tag == -1)
-		return;
+	hctx = container_of(wait, struct blk_mq_hw_ctx, dispatch_wait);
 
-	hctx = blk_mq_map_queue(rq->q, rq->mq_ctx->cpu);
-	__blk_mq_put_driver_tag(hctx, rq);
+	list_del_init(&wait->entry);
+	blk_mq_run_hw_queue(hctx, true);
+	return 1;
 }
 
 /*
- * If we fail getting a driver tag because all the driver tags are already
- * assigned and on the dispatch list, BUT the first entry does not have a
- * tag, then we could deadlock. For that case, move entries with assigned
- * driver tags to the front, leaving the set of tagged requests in the
- * same order, and the untagged set in the same order.
+ * Mark us waiting for a tag. For shared tags, this involves hooking us into
+ * the tag wakeups. For non-shared tags, we can simply mark us nedeing a
+ * restart. For both caes, take care to check the condition again after
+ * marking us as waiting.
  */
-static bool reorder_tags_to_front(struct list_head *list)
-{
-	struct request *rq, *tmp, *first = NULL;
-
-	list_for_each_entry_safe_reverse(rq, tmp, list, queuelist) {
-		if (rq == first)
-			break;
-		if (rq->tag != -1) {
-			list_move(&rq->queuelist, list);
-			if (!first)
-				first = rq;
-		}
-	}
-
-	return first != NULL;
-}
-
-static int blk_mq_dispatch_wake(wait_queue_entry_t *wait, unsigned mode, int flags,
-				void *key)
+static bool blk_mq_mark_tag_wait(struct blk_mq_hw_ctx **hctx,
+				 struct request *rq)
 {
-	struct blk_mq_hw_ctx *hctx;
+	struct blk_mq_hw_ctx *this_hctx = *hctx;
+	bool shared_tags = (this_hctx->flags & BLK_MQ_F_TAG_SHARED) != 0;
+	struct sbq_wait_state *ws;
+	wait_queue_entry_t *wait;
+	bool ret;
 
-	hctx = container_of(wait, struct blk_mq_hw_ctx, dispatch_wait);
+	if (!shared_tags) {
+		if (!test_bit(BLK_MQ_S_SCHED_RESTART, &this_hctx->state))
+			set_bit(BLK_MQ_S_SCHED_RESTART, &this_hctx->state);
+	} else {
+		wait = &this_hctx->dispatch_wait;
+		if (!list_empty_careful(&wait->entry))
+			return false;
 
-	list_del(&wait->entry);
-	clear_bit_unlock(BLK_MQ_S_TAG_WAITING, &hctx->state);
-	blk_mq_run_hw_queue(hctx, true);
-	return 1;
-}
+		spin_lock(&this_hctx->lock);
+		if (!list_empty(&wait->entry)) {
+			spin_unlock(&this_hctx->lock);
+			return false;
+		}
 
-static bool blk_mq_dispatch_wait_add(struct blk_mq_hw_ctx *hctx)
-{
-	struct sbq_wait_state *ws;
+		ws = bt_wait_ptr(&this_hctx->tags->bitmap_tags, this_hctx);
+		add_wait_queue(&ws->wait, wait);
+	}
 
 	/*
-	 * The TAG_WAITING bit serves as a lock protecting hctx->dispatch_wait.
-	 * The thread which wins the race to grab this bit adds the hardware
-	 * queue to the wait queue.
+	 * It's possible that a tag was freed in the window between the
+	 * allocation failure and adding the hardware queue to the wait
+	 * queue.
 	 */
-	if (test_bit(BLK_MQ_S_TAG_WAITING, &hctx->state) ||
-	    test_and_set_bit_lock(BLK_MQ_S_TAG_WAITING, &hctx->state))
-		return false;
+	ret = blk_mq_get_driver_tag(rq, hctx, false);
 
-	init_waitqueue_func_entry(&hctx->dispatch_wait, blk_mq_dispatch_wake);
-	ws = bt_wait_ptr(&hctx->tags->bitmap_tags, hctx);
+	if (!shared_tags) {
+		/*
+		 * Don't clear RESTART here, someone else could have set it.
+		 * At most this will cost an extra queue run.
+		 */
+		return ret;
+	} else {
+		if (!ret) {
+			spin_unlock(&this_hctx->lock);
+			return false;
+		}
 
-	/*
-	 * As soon as this returns, it's no longer safe to fiddle with
-	 * hctx->dispatch_wait, since a completion can wake up the wait queue
-	 * and unlock the bit.
-	 */
-	add_wait_queue(&ws->wait, &hctx->dispatch_wait);
-	return true;
+		/*
+		 * We got a tag, remove ourselves from the wait queue to ensure
+		 * someone else gets the wakeup.
+		 */
+		spin_lock_irq(&ws->wait.lock);
+		list_del_init(&wait->entry);
+		spin_unlock_irq(&ws->wait.lock);
+		spin_unlock(&this_hctx->lock);
+		return true;
+	}
 }
 
-bool blk_mq_dispatch_rq_list(struct request_queue *q, struct list_head *list)
+bool blk_mq_dispatch_rq_list(struct request_queue *q, struct list_head *list,
+			     bool got_budget)
 {
 	struct blk_mq_hw_ctx *hctx;
-	struct request *rq;
+	struct request *rq, *nxt;
+	bool no_tag = false;
 	int errors, queued;
 
 	if (list_empty(list))
 		return false;
 
+	WARN_ON(!list_is_singular(list) && got_budget);
+
 	/*
 	 * Now process all the entries, sending them to the driver.
 	 */
@@ -1033,23 +1093,29 @@ bool blk_mq_dispatch_rq_list(struct request_queue *q, struct list_head *list)
 
 		rq = list_first_entry(list, struct request, queuelist);
 		if (!blk_mq_get_driver_tag(rq, &hctx, false)) {
-			if (!queued && reorder_tags_to_front(list))
-				continue;
-
 			/*
 			 * The initial allocation attempt failed, so we need to
-			 * rerun the hardware queue when a tag is freed.
+			 * rerun the hardware queue when a tag is freed. The
+			 * waitqueue takes care of that. If the queue is run
+			 * before we add this entry back on the dispatch list,
+			 * we'll re-run it below.
 			 */
-			if (!blk_mq_dispatch_wait_add(hctx))
+			if (!blk_mq_mark_tag_wait(&hctx, rq)) {
+				if (got_budget)
+					blk_mq_put_dispatch_budget(hctx);
+				/*
+				 * For non-shared tags, the RESTART check
+				 * will suffice.
+				 */
+				if (hctx->flags & BLK_MQ_F_TAG_SHARED)
+					no_tag = true;
 				break;
+			}
+		}
 
-			/*
-			 * It's possible that a tag was freed in the window
-			 * between the allocation failure and adding the
-			 * hardware queue to the wait queue.
-			 */
-			if (!blk_mq_get_driver_tag(rq, &hctx, false))
-				break;
+		if (!got_budget && !blk_mq_get_dispatch_budget(hctx)) {
+			blk_mq_put_driver_tag(rq);
+			break;
 		}
 
 		list_del_init(&rq->queuelist);
@@ -1063,15 +1129,21 @@ bool blk_mq_dispatch_rq_list(struct request_queue *q, struct list_head *list)
 		if (list_empty(list))
 			bd.last = true;
 		else {
-			struct request *nxt;
-
 			nxt = list_first_entry(list, struct request, queuelist);
 			bd.last = !blk_mq_get_driver_tag(nxt, NULL, false);
 		}
 
 		ret = q->mq_ops->queue_rq(hctx, &bd);
 		if (ret == BLK_STS_RESOURCE) {
-			blk_mq_put_driver_tag_hctx(hctx, rq);
+			/*
+			 * If an I/O scheduler has been configured and we got a
+			 * driver tag for the next request already, free it
+			 * again.
+			 */
+			if (!list_empty(list)) {
+				nxt = list_first_entry(list, struct request, queuelist);
+				blk_mq_put_driver_tag(nxt);
+			}
 			list_add(&rq->queuelist, list);
 			__blk_mq_requeue_request(rq);
 			break;
@@ -1093,13 +1165,6 @@ bool blk_mq_dispatch_rq_list(struct request_queue *q, struct list_head *list)
 	 * that is where we will continue on next queue run.
 	 */
 	if (!list_empty(list)) {
-		/*
-		 * If an I/O scheduler has been configured and we got a driver
-		 * tag for the next request already, free it again.
-		 */
-		rq = list_first_entry(list, struct request, queuelist);
-		blk_mq_put_driver_tag(rq);
-
 		spin_lock(&hctx->lock);
 		list_splice_init(list, &hctx->dispatch);
 		spin_unlock(&hctx->lock);
@@ -1109,10 +1174,10 @@ bool blk_mq_dispatch_rq_list(struct request_queue *q, struct list_head *list)
 		 * it is no longer set that means that it was cleared by another
 		 * thread and hence that a queue rerun is needed.
 		 *
-		 * If TAG_WAITING is set that means that an I/O scheduler has
-		 * been configured and another thread is waiting for a driver
-		 * tag. To guarantee fairness, do not rerun this hardware queue
-		 * but let the other thread grab the driver tag.
+		 * If 'no_tag' is set, that means that we failed getting
+		 * a driver tag with an I/O scheduler attached. If our dispatch
+		 * waitqueue is no longer active, ensure that we run the queue
+		 * AFTER adding our entries back to the list.
 		 *
 		 * If no I/O scheduler has been configured it is possible that
 		 * the hardware queue got stopped and restarted before requests
@@ -1124,8 +1189,8 @@ bool blk_mq_dispatch_rq_list(struct request_queue *q, struct list_head *list)
 		 *   returning BLK_STS_RESOURCE. Two exceptions are scsi-mq
 		 *   and dm-rq.
 		 */
-		if (!blk_mq_sched_needs_restart(hctx) &&
-		    !test_bit(BLK_MQ_S_TAG_WAITING, &hctx->state))
+		if (!blk_mq_sched_needs_restart(hctx) ||
+		    (no_tag && list_empty_careful(&hctx->dispatch_wait.entry)))
 			blk_mq_run_hw_queue(hctx, true);
 	}
 
@@ -1218,9 +1283,14 @@ void blk_mq_delay_run_hw_queue(struct blk_mq_hw_ctx *hctx, unsigned long msecs)
 }
 EXPORT_SYMBOL(blk_mq_delay_run_hw_queue);
 
-void blk_mq_run_hw_queue(struct blk_mq_hw_ctx *hctx, bool async)
+bool blk_mq_run_hw_queue(struct blk_mq_hw_ctx *hctx, bool async)
 {
-	__blk_mq_delay_run_hw_queue(hctx, async, 0);
+	if (blk_mq_hctx_has_pending(hctx)) {
+		__blk_mq_delay_run_hw_queue(hctx, async, 0);
+		return true;
+	}
+
+	return false;
 }
 EXPORT_SYMBOL(blk_mq_run_hw_queue);
 
@@ -1230,8 +1300,7 @@ void blk_mq_run_hw_queues(struct request_queue *q, bool async)
 	int i;
 
 	queue_for_each_hw_ctx(q, hctx, i) {
-		if (!blk_mq_hctx_has_pending(hctx) ||
-		    blk_mq_hctx_stopped(hctx))
+		if (blk_mq_hctx_stopped(hctx))
 			continue;
 
 		blk_mq_run_hw_queue(hctx, async);
@@ -1405,7 +1474,7 @@ void __blk_mq_insert_request(struct blk_mq_hw_ctx *hctx, struct request *rq,
  * Should only be used carefully, when the caller knows we want to
  * bypass a potential IO scheduler on the target device.
  */
-void blk_mq_request_bypass_insert(struct request *rq)
+void blk_mq_request_bypass_insert(struct request *rq, bool run_queue)
 {
 	struct blk_mq_ctx *ctx = rq->mq_ctx;
 	struct blk_mq_hw_ctx *hctx = blk_mq_map_queue(rq->q, ctx->cpu);
@@ -1414,7 +1483,8 @@ void blk_mq_request_bypass_insert(struct request *rq)
 	list_add_tail(&rq->queuelist, &hctx->dispatch);
 	spin_unlock(&hctx->lock);
 
-	blk_mq_run_hw_queue(hctx, false);
+	if (run_queue)
+		blk_mq_run_hw_queue(hctx, false);
 }
 
 void blk_mq_insert_requests(struct blk_mq_hw_ctx *hctx, struct blk_mq_ctx *ctx,
@@ -1501,13 +1571,9 @@ static void blk_mq_bio_to_request(struct request *rq, struct bio *bio)
 {
 	blk_init_request_from_bio(rq, bio);
 
-	blk_account_io_start(rq, true);
-}
+	blk_rq_set_rl(rq, blk_get_rl(rq->q, bio));
 
-static inline bool hctx_allow_merges(struct blk_mq_hw_ctx *hctx)
-{
-	return (hctx->flags & BLK_MQ_F_SHOULD_MERGE) &&
-		!blk_queue_nomerges(hctx->queue);
+	blk_account_io_start(rq, true);
 }
 
 static inline void blk_mq_queue_io(struct blk_mq_hw_ctx *hctx,
@@ -1552,6 +1618,11 @@ static void __blk_mq_try_issue_directly(struct blk_mq_hw_ctx *hctx,
 	if (!blk_mq_get_driver_tag(rq, NULL, false))
 		goto insert;
 
+	if (!blk_mq_get_dispatch_budget(hctx)) {
+		blk_mq_put_driver_tag(rq);
+		goto insert;
+	}
+
 	new_cookie = request_to_qc_t(hctx, rq);
 
 	/*
@@ -1641,13 +1712,10 @@ static blk_qc_t blk_mq_make_request(struct request_queue *q, struct bio *bio)
 	if (unlikely(is_flush_fua)) {
 		blk_mq_put_ctx(data.ctx);
 		blk_mq_bio_to_request(rq, bio);
-		if (q->elevator) {
-			blk_mq_sched_insert_request(rq, false, true, true,
-					true);
-		} else {
-			blk_insert_flush(rq);
-			blk_mq_run_hw_queue(data.hctx, true);
-		}
+
+		/* bypass scheduler for flush rq */
+		blk_insert_flush(rq);
+		blk_mq_run_hw_queue(data.hctx, true);
 	} else if (plug && q->nr_hw_queues == 1) {
 		struct request *last = NULL;
 
@@ -1990,6 +2058,9 @@ static int blk_mq_init_hctx(struct request_queue *q,
 
 	hctx->nr_ctx = 0;
 
+	init_waitqueue_func_entry(&hctx->dispatch_wait, blk_mq_dispatch_wake);
+	INIT_LIST_HEAD(&hctx->dispatch_wait.entry);
+
 	if (set->ops->init_hctx &&
 	    set->ops->init_hctx(hctx, set->driver_data, hctx_idx))
 		goto free_bitmap;
@@ -2229,8 +2300,11 @@ static void blk_mq_add_queue_tag_set(struct blk_mq_tag_set *set,
 
 	mutex_lock(&set->tag_list_lock);
 
-	/* Check to see if we're transitioning to shared (from 1 to 2 queues). */
-	if (!list_empty(&set->tag_list) && !(set->flags & BLK_MQ_F_TAG_SHARED)) {
+	/*
+	 * Check to see if we're transitioning to shared (from 1 to 2 queues).
+	 */
+	if (!list_empty(&set->tag_list) &&
+	    !(set->flags & BLK_MQ_F_TAG_SHARED)) {
 		set->flags |= BLK_MQ_F_TAG_SHARED;
 		/* update existing queue */
 		blk_mq_update_tag_set_depth(set, true);
@@ -2404,6 +2478,8 @@ struct request_queue *blk_mq_init_allocated_queue(struct blk_mq_tag_set *set,
 	spin_lock_init(&q->requeue_lock);
 
 	blk_queue_make_request(q, blk_mq_make_request);
+	if (q->mq_ops->poll)
+		q->poll_fn = blk_mq_poll;
 
 	/*
 	 * Do this after blk_queue_make_request() overrides it...
@@ -2460,10 +2536,9 @@ static void blk_mq_queue_reinit(struct request_queue *q)
 
 	/*
 	 * redo blk_mq_init_cpu_queues and blk_mq_init_hw_queues. FIXME: maybe
-	 * we should change hctx numa_node according to new topology (this
-	 * involves free and re-allocate memory, worthy doing?)
+	 * we should change hctx numa_node according to the new topology (this
+	 * involves freeing and re-allocating memory, worth doing?)
 	 */
-
 	blk_mq_map_swqueue(q);
 
 	blk_mq_sysfs_register(q);
@@ -2552,6 +2627,9 @@ int blk_mq_alloc_tag_set(struct blk_mq_tag_set *set)
 	if (!set->ops->queue_rq)
 		return -EINVAL;
 
+	if (!set->ops->get_budget ^ !set->ops->put_budget)
+		return -EINVAL;
+
 	if (set->queue_depth > BLK_MQ_MAX_DEPTH) {
 		pr_info("blk-mq: reduced tag depth to %u\n",
 			BLK_MQ_MAX_DEPTH);
@@ -2642,8 +2720,7 @@ int blk_mq_update_nr_requests(struct request_queue *q, unsigned int nr)
 		 * queue depth. This is similar to what the old code would do.
 		 */
 		if (!hctx->sched_tags) {
-			ret = blk_mq_tag_update_depth(hctx, &hctx->tags,
-							min(nr, set->queue_depth),
+			ret = blk_mq_tag_update_depth(hctx, &hctx->tags, nr,
 							false);
 		} else {
 			ret = blk_mq_tag_update_depth(hctx, &hctx->sched_tags,
@@ -2863,20 +2940,14 @@ static bool __blk_mq_poll(struct blk_mq_hw_ctx *hctx, struct request *rq)
 	return false;
 }
 
-bool blk_mq_poll(struct request_queue *q, blk_qc_t cookie)
+static bool blk_mq_poll(struct request_queue *q, blk_qc_t cookie)
 {
 	struct blk_mq_hw_ctx *hctx;
-	struct blk_plug *plug;
 	struct request *rq;
 
-	if (!q->mq_ops || !q->mq_ops->poll || !blk_qc_t_valid(cookie) ||
-	    !test_bit(QUEUE_FLAG_POLL, &q->queue_flags))
+	if (!test_bit(QUEUE_FLAG_POLL, &q->queue_flags))
 		return false;
 
-	plug = current->plug;
-	if (plug)
-		blk_flush_plug_list(plug, false);
-
 	hctx = q->queue_hw_ctx[blk_qc_t_to_queue_num(cookie)];
 	if (!blk_qc_t_is_internal(cookie))
 		rq = blk_mq_tag_to_rq(hctx->tags, blk_qc_t_to_tag(cookie));
@@ -2894,10 +2965,15 @@ bool blk_mq_poll(struct request_queue *q, blk_qc_t cookie)
 
 	return __blk_mq_poll(hctx, rq);
 }
-EXPORT_SYMBOL_GPL(blk_mq_poll);
 
 static int __init blk_mq_init(void)
 {
+	/*
+	 * See comment in block/blk.h rq_atomic_flags enum
+	 */
+	BUILD_BUG_ON((REQ_ATOM_STARTED / BITS_PER_BYTE) !=
+			(REQ_ATOM_COMPLETE / BITS_PER_BYTE));
+
 	cpuhp_setup_state_multi(CPUHP_BLK_MQ_DEAD, "block/mq:dead", NULL,
 				blk_mq_hctx_notify_dead);
 	return 0;

block/blk-mq.h

@@ -3,6 +3,7 @@
 #define INT_BLK_MQ_H
 
 #include "blk-stat.h"
+#include "blk-mq-tag.h"
 
 struct blk_mq_tag_set;
 
@@ -26,16 +27,16 @@ struct blk_mq_ctx {
 	struct kobject		kobj;
 } ____cacheline_aligned_in_smp;
 
-void blk_mq_run_hw_queue(struct blk_mq_hw_ctx *hctx, bool async);
 void blk_mq_freeze_queue(struct request_queue *q);
 void blk_mq_free_queue(struct request_queue *q);
 int blk_mq_update_nr_requests(struct request_queue *q, unsigned int nr);
 void blk_mq_wake_waiters(struct request_queue *q);
-bool blk_mq_dispatch_rq_list(struct request_queue *, struct list_head *);
+bool blk_mq_dispatch_rq_list(struct request_queue *, struct list_head *, bool);
 void blk_mq_flush_busy_ctxs(struct blk_mq_hw_ctx *hctx, struct list_head *list);
-bool blk_mq_hctx_has_pending(struct blk_mq_hw_ctx *hctx);
 bool blk_mq_get_driver_tag(struct request *rq, struct blk_mq_hw_ctx **hctx,
 				bool wait);
+struct request *blk_mq_dequeue_from_ctx(struct blk_mq_hw_ctx *hctx,
+					struct blk_mq_ctx *start);
 
 /*
  * Internal helpers for allocating/freeing the request map
@@ -55,7 +56,7 @@ int blk_mq_alloc_rqs(struct blk_mq_tag_set *set, struct blk_mq_tags *tags,
  */
 void __blk_mq_insert_request(struct blk_mq_hw_ctx *hctx, struct request *rq,
 				bool at_head);
-void blk_mq_request_bypass_insert(struct request *rq);
+void blk_mq_request_bypass_insert(struct request *rq, bool run_queue);
 void blk_mq_insert_requests(struct blk_mq_hw_ctx *hctx, struct blk_mq_ctx *ctx,
 				struct list_head *list);
 
@@ -109,7 +110,7 @@ static inline void blk_mq_put_ctx(struct blk_mq_ctx *ctx)
 struct blk_mq_alloc_data {
 	/* input parameter */
 	struct request_queue *q;
-	unsigned int flags;
+	blk_mq_req_flags_t flags;
 	unsigned int shallow_depth;
 
 	/* input & output parameter */
@@ -138,4 +139,53 @@ static inline bool blk_mq_hw_queue_mapped(struct blk_mq_hw_ctx *hctx)
 void blk_mq_in_flight(struct request_queue *q, struct hd_struct *part,
 			unsigned int inflight[2]);
 
+static inline void blk_mq_put_dispatch_budget(struct blk_mq_hw_ctx *hctx)
+{
+	struct request_queue *q = hctx->queue;
+
+	if (q->mq_ops->put_budget)
+		q->mq_ops->put_budget(hctx);
+}
+
+static inline bool blk_mq_get_dispatch_budget(struct blk_mq_hw_ctx *hctx)
+{
+	struct request_queue *q = hctx->queue;
+
+	if (q->mq_ops->get_budget)
+		return q->mq_ops->get_budget(hctx);
+	return true;
+}
+
+static inline void __blk_mq_put_driver_tag(struct blk_mq_hw_ctx *hctx,
+					   struct request *rq)
+{
+	blk_mq_put_tag(hctx, hctx->tags, rq->mq_ctx, rq->tag);
+	rq->tag = -1;
+
+	if (rq->rq_flags & RQF_MQ_INFLIGHT) {
+		rq->rq_flags &= ~RQF_MQ_INFLIGHT;
+		atomic_dec(&hctx->nr_active);
+	}
+}
+
+static inline void blk_mq_put_driver_tag_hctx(struct blk_mq_hw_ctx *hctx,
+				       struct request *rq)
+{
+	if (rq->tag == -1 || rq->internal_tag == -1)
+		return;
+
+	__blk_mq_put_driver_tag(hctx, rq);
+}
+
+static inline void blk_mq_put_driver_tag(struct request *rq)
+{
+	struct blk_mq_hw_ctx *hctx;
+
+	if (rq->tag == -1 || rq->internal_tag == -1)
+		return;
+
+	hctx = blk_mq_map_queue(rq->q, rq->mq_ctx->cpu);
+	__blk_mq_put_driver_tag(hctx, rq);
+}
+
 #endif

block/blk-settings.c

@@ -157,7 +157,7 @@ EXPORT_SYMBOL(blk_set_stacking_limits);
  * Caveat:
  *    The driver that does this *must* be able to deal appropriately
  *    with buffers in "highmemory". This can be accomplished by either calling
- *    __bio_kmap_atomic() to get a temporary kernel mapping, or by calling
+ *    kmap_atomic() to get a temporary kernel mapping, or by calling
  *    blk_queue_bounce() to create a buffer in normal memory.
  **/
 void blk_queue_make_request(struct request_queue *q, make_request_fn *mfn)

block/blk-stat.c

@@ -11,8 +11,6 @@
 #include "blk-mq.h"
 #include "blk.h"
 
-#define BLK_RQ_STAT_BATCH	64
-
 struct blk_queue_stats {
 	struct list_head callbacks;
 	spinlock_t lock;
@@ -23,45 +21,21 @@ static void blk_stat_init(struct blk_rq_stat *stat)
 {
 	stat->min = -1ULL;
 	stat->max = stat->nr_samples = stat->mean = 0;
-	stat->batch = stat->nr_batch = 0;
-}
-
-static void blk_stat_flush_batch(struct blk_rq_stat *stat)
-{
-	const s32 nr_batch = READ_ONCE(stat->nr_batch);
-	const s32 nr_samples = READ_ONCE(stat->nr_samples);
-
-	if (!nr_batch)
-		return;
-	if (!nr_samples)
-		stat->mean = div64_s64(stat->batch, nr_batch);
-	else {
-		stat->mean = div64_s64((stat->mean * nr_samples) +
-					stat->batch,
-					nr_batch + nr_samples);
-	}
-
-	stat->nr_samples += nr_batch;
-	stat->nr_batch = stat->batch = 0;
+	stat->batch = 0;
 }
 
+/* src is a per-cpu stat, mean isn't initialized */
 static void blk_stat_sum(struct blk_rq_stat *dst, struct blk_rq_stat *src)
 {
-	blk_stat_flush_batch(src);
-
 	if (!src->nr_samples)
 		return;
 
 	dst->min = min(dst->min, src->min);
 	dst->max = max(dst->max, src->max);
 
-	if (!dst->nr_samples)
-		dst->mean = src->mean;
-	else {
-		dst->mean = div64_s64((src->mean * src->nr_samples) +
-					(dst->mean * dst->nr_samples),
-					dst->nr_samples + src->nr_samples);
-	}
+	dst->mean = div_u64(src->batch + dst->mean * dst->nr_samples,
+				dst->nr_samples + src->nr_samples);
+
 	dst->nr_samples += src->nr_samples;
 }
 
@@ -69,13 +43,8 @@ static void __blk_stat_add(struct blk_rq_stat *stat, u64 value)
 {
 	stat->min = min(stat->min, value);
 	stat->max = max(stat->max, value);
-
-	if (stat->batch + value < stat->batch ||
-	    stat->nr_batch + 1 == BLK_RQ_STAT_BATCH)
-		blk_stat_flush_batch(stat);
-
 	stat->batch += value;
-	stat->nr_batch++;
+	stat->nr_samples++;
 }
 
 void blk_stat_add(struct request *rq)
@@ -84,7 +53,7 @@ void blk_stat_add(struct request *rq)
 	struct blk_stat_callback *cb;
 	struct blk_rq_stat *stat;
 	int bucket;
-	s64 now, value;
+	u64 now, value;
 
 	now = __blk_stat_time(ktime_to_ns(ktime_get()));
 	if (now < blk_stat_time(&rq->issue_stat))

block/blk-throttle.c

@@ -2113,8 +2113,12 @@ static inline void throtl_update_latency_buckets(struct throtl_data *td)
 static void blk_throtl_assoc_bio(struct throtl_grp *tg, struct bio *bio)
 {
 #ifdef CONFIG_BLK_DEV_THROTTLING_LOW
-	if (bio->bi_css)
+	if (bio->bi_css) {
+		if (bio->bi_cg_private)
+			blkg_put(tg_to_blkg(bio->bi_cg_private));
 		bio->bi_cg_private = tg;
+		blkg_get(tg_to_blkg(tg));
+	}
 	blk_stat_set_issue(&bio->bi_issue_stat, bio_sectors(bio));
 #endif
 }
@@ -2284,8 +2288,10 @@ void blk_throtl_bio_endio(struct bio *bio)
 
 	start_time = blk_stat_time(&bio->bi_issue_stat) >> 10;
 	finish_time = __blk_stat_time(finish_time_ns) >> 10;
-	if (!start_time || finish_time <= start_time)
+	if (!start_time || finish_time <= start_time) {
+		blkg_put(tg_to_blkg(tg));
 		return;
+	}
 
 	lat = finish_time - start_time;
 	/* this is only for bio based driver */
@@ -2315,6 +2321,8 @@ void blk_throtl_bio_endio(struct bio *bio)
 		tg->bio_cnt /= 2;
 		tg->bad_bio_cnt /= 2;
 	}
+
+	blkg_put(tg_to_blkg(tg));
 }
 #endif
 

block/blk-timeout.c

@@ -134,8 +134,6 @@ void blk_timeout_work(struct work_struct *work)
 	struct request *rq, *tmp;
 	int next_set = 0;
 
-	if (blk_queue_enter(q, true))
-		return;
 	spin_lock_irqsave(q->queue_lock, flags);
 
 	list_for_each_entry_safe(rq, tmp, &q->timeout_list, timeout_list)
@@ -145,7 +143,6 @@ void blk_timeout_work(struct work_struct *work)
 		mod_timer(&q->timeout, round_jiffies_up(next));
 
 	spin_unlock_irqrestore(q->queue_lock, flags);
-	blk_queue_exit(q);
 }
 
 /**
@@ -211,7 +208,7 @@ void blk_add_timer(struct request *req)
 	if (!req->timeout)
 		req->timeout = q->rq_timeout;
 
-	req->deadline = jiffies + req->timeout;
+	WRITE_ONCE(req->deadline, jiffies + req->timeout);
 
 	/*
 	 * Only the non-mq case needs to add the request to a protected list.

block/blk-wbt.c

@@ -654,7 +654,7 @@ void wbt_set_write_cache(struct rq_wb *rwb, bool write_cache_on)
 }
 
 /*
- * Disable wbt, if enabled by default. Only called from CFQ.
+ * Disable wbt, if enabled by default.
  */
 void wbt_disable_default(struct request_queue *q)
 {

block/blk.h

@@ -123,8 +123,15 @@ void blk_account_io_done(struct request *req);
  * Internal atomic flags for request handling
  */
 enum rq_atomic_flags {
+	/*
+	 * Keep these two bits first - not because we depend on the
+	 * value of them, but we do depend on them being in the same
+	 * byte of storage to ensure ordering on writes. Keeping them
+	 * first will achieve that nicely.
+	 */
 	REQ_ATOM_COMPLETE = 0,
 	REQ_ATOM_STARTED,
+
 	REQ_ATOM_POLL_SLEPT,
 };
 
@@ -149,45 +156,6 @@ static inline void blk_clear_rq_complete(struct request *rq)
 
 void blk_insert_flush(struct request *rq);
 
-static inline struct request *__elv_next_request(struct request_queue *q)
-{
-	struct request *rq;
-	struct blk_flush_queue *fq = blk_get_flush_queue(q, NULL);
-
-	WARN_ON_ONCE(q->mq_ops);
-
-	while (1) {
-		if (!list_empty(&q->queue_head)) {
-			rq = list_entry_rq(q->queue_head.next);
-			return rq;
-		}
-
-		/*
-		 * Flush request is running and flush request isn't queueable
-		 * in the drive, we can hold the queue till flush request is
-		 * finished. Even we don't do this, driver can't dispatch next
-		 * requests and will requeue them. And this can improve
-		 * throughput too. For example, we have request flush1, write1,
-		 * flush 2. flush1 is dispatched, then queue is hold, write1
-		 * isn't inserted to queue. After flush1 is finished, flush2
-		 * will be dispatched. Since disk cache is already clean,
-		 * flush2 will be finished very soon, so looks like flush2 is
-		 * folded to flush1.
-		 * Since the queue is hold, a flag is set to indicate the queue
-		 * should be restarted later. Please see flush_end_io() for
-		 * details.
-		 */
-		if (fq->flush_pending_idx != fq->flush_running_idx &&
-				!queue_flush_queueable(q)) {
-			fq->flush_queue_delayed = 1;
-			return NULL;
-		}
-		if (unlikely(blk_queue_bypass(q)) ||
-		    !q->elevator->type->ops.sq.elevator_dispatch_fn(q, 0))
-			return NULL;
-	}
-}
-
 static inline void elv_activate_rq(struct request_queue *q, struct request *rq)
 {
 	struct elevator_queue *e = q->elevator;

block/bsg.c

@@ -137,7 +137,7 @@ static inline struct hlist_head *bsg_dev_idx_hash(int index)
 
 static int blk_fill_sgv4_hdr_rq(struct request_queue *q, struct request *rq,
 				struct sg_io_v4 *hdr, struct bsg_device *bd,
-				fmode_t has_write_perm)
+				fmode_t mode)
 {
 	struct scsi_request *req = scsi_req(rq);
 
@@ -152,7 +152,7 @@ static int blk_fill_sgv4_hdr_rq(struct request_queue *q, struct request *rq,
 		return -EFAULT;
 
 	if (hdr->subprotocol == BSG_SUB_PROTOCOL_SCSI_CMD) {
-		if (blk_verify_command(req->cmd, has_write_perm))
+		if (blk_verify_command(req->cmd, mode))
 			return -EPERM;
 	} else if (!capable(CAP_SYS_RAWIO))
 		return -EPERM;
@@ -206,7 +206,7 @@ bsg_validate_sgv4_hdr(struct sg_io_v4 *hdr, int *op)
  * map sg_io_v4 to a request.
  */
 static struct request *
-bsg_map_hdr(struct bsg_device *bd, struct sg_io_v4 *hdr, fmode_t has_write_perm)
+bsg_map_hdr(struct bsg_device *bd, struct sg_io_v4 *hdr, fmode_t mode)
 {
 	struct request_queue *q = bd->queue;
 	struct request *rq, *next_rq = NULL;
@@ -237,7 +237,7 @@ bsg_map_hdr(struct bsg_device *bd, struct sg_io_v4 *hdr, fmode_t has_write_perm)
 	if (IS_ERR(rq))
 		return rq;
 
-	ret = blk_fill_sgv4_hdr_rq(q, rq, hdr, bd, has_write_perm);
+	ret = blk_fill_sgv4_hdr_rq(q, rq, hdr, bd, mode);
 	if (ret)
 		goto out;
 
@@ -587,8 +587,7 @@ bsg_read(struct file *file, char __user *buf, size_t count, loff_t *ppos)
 }
 
 static int __bsg_write(struct bsg_device *bd, const char __user *buf,
-		       size_t count, ssize_t *bytes_written,
-		       fmode_t has_write_perm)
+		       size_t count, ssize_t *bytes_written, fmode_t mode)
 {
 	struct bsg_command *bc;
 	struct request *rq;
@@ -619,7 +618,7 @@ static int __bsg_write(struct bsg_device *bd, const char __user *buf,
 		/*
 		 * get a request, fill in the blanks, and add to request queue
 		 */
-		rq = bsg_map_hdr(bd, &bc->hdr, has_write_perm);
+		rq = bsg_map_hdr(bd, &bc->hdr, mode);
 		if (IS_ERR(rq)) {
 			ret = PTR_ERR(rq);
 			rq = NULL;
@@ -655,8 +654,7 @@ bsg_write(struct file *file, const char __user *buf, size_t count, loff_t *ppos)
 	bsg_set_block(bd, file);
 
 	bytes_written = 0;
-	ret = __bsg_write(bd, buf, count, &bytes_written,
-			  file->f_mode & FMODE_WRITE);
+	ret = __bsg_write(bd, buf, count, &bytes_written, file->f_mode);
 
 	*ppos = bytes_written;
 
@@ -915,7 +913,7 @@ static long bsg_ioctl(struct file *file, unsigned int cmd, unsigned long arg)
 		if (copy_from_user(&hdr, uarg, sizeof(hdr)))
 			return -EFAULT;
 
-		rq = bsg_map_hdr(bd, &hdr, file->f_mode & FMODE_WRITE);
+		rq = bsg_map_hdr(bd, &hdr, file->f_mode);
 		if (IS_ERR(rq))
 			return PTR_ERR(rq);
 

block/elevator.c

@@ -83,12 +83,25 @@ bool elv_bio_merge_ok(struct request *rq, struct bio *bio)
 }
 EXPORT_SYMBOL(elv_bio_merge_ok);
 
-static struct elevator_type *elevator_find(const char *name)
+static bool elevator_match(const struct elevator_type *e, const char *name)
+{
+	if (!strcmp(e->elevator_name, name))
+		return true;
+	if (e->elevator_alias && !strcmp(e->elevator_alias, name))
+		return true;
+
+	return false;
+}
+
+/*
+ * Return scheduler with name 'name' and with matching 'mq capability
+ */
+static struct elevator_type *elevator_find(const char *name, bool mq)
 {
 	struct elevator_type *e;
 
 	list_for_each_entry(e, &elv_list, list) {
-		if (!strcmp(e->elevator_name, name))
+		if (elevator_match(e, name) && (mq == e->uses_mq))
 			return e;
 	}
 
@@ -100,25 +113,25 @@ static void elevator_put(struct elevator_type *e)
 	module_put(e->elevator_owner);
 }
 
-static struct elevator_type *elevator_get(const char *name, bool try_loading)
+static struct elevator_type *elevator_get(struct request_queue *q,
+					  const char *name, bool try_loading)
 {
 	struct elevator_type *e;
 
 	spin_lock(&elv_list_lock);
 
-	e = elevator_find(name);
+	e = elevator_find(name, q->mq_ops != NULL);
 	if (!e && try_loading) {
 		spin_unlock(&elv_list_lock);
 		request_module("%s-iosched", name);
 		spin_lock(&elv_list_lock);
-		e = elevator_find(name);
+		e = elevator_find(name, q->mq_ops != NULL);
 	}
 
 	if (e && !try_module_get(e->elevator_owner))
 		e = NULL;
 
 	spin_unlock(&elv_list_lock);
-
 	return e;
 }
 
@@ -144,8 +157,12 @@ void __init load_default_elevator_module(void)
 	if (!chosen_elevator[0])
 		return;
 
+	/*
+	 * Boot parameter is deprecated, we haven't supported that for MQ.
+	 * Only look for non-mq schedulers from here.
+	 */
 	spin_lock(&elv_list_lock);
-	e = elevator_find(chosen_elevator);
+	e = elevator_find(chosen_elevator, false);
 	spin_unlock(&elv_list_lock);
 
 	if (!e)
@@ -202,7 +219,7 @@ int elevator_init(struct request_queue *q, char *name)
 	q->boundary_rq = NULL;
 
 	if (name) {
-		e = elevator_get(name, true);
+		e = elevator_get(q, name, true);
 		if (!e)
 			return -EINVAL;
 	}
@@ -214,7 +231,7 @@ int elevator_init(struct request_queue *q, char *name)
 	 * allowed from async.
 	 */
 	if (!e && !q->mq_ops && *chosen_elevator) {
-		e = elevator_get(chosen_elevator, false);
+		e = elevator_get(q, chosen_elevator, false);
 		if (!e)
 			printk(KERN_ERR "I/O scheduler %s not found\n",
 							chosen_elevator);
@@ -229,17 +246,17 @@ int elevator_init(struct request_queue *q, char *name)
 		 */
 		if (q->mq_ops) {
 			if (q->nr_hw_queues == 1)
-				e = elevator_get("mq-deadline", false);
+				e = elevator_get(q, "mq-deadline", false);
 			if (!e)
 				return 0;
 		} else
-			e = elevator_get(CONFIG_DEFAULT_IOSCHED, false);
+			e = elevator_get(q, CONFIG_DEFAULT_IOSCHED, false);
 
 		if (!e) {
 			printk(KERN_ERR
 				"Default I/O scheduler not found. " \
 				"Using noop.\n");
-			e = elevator_get("noop", false);
+			e = elevator_get(q, "noop", false);
 		}
 	}
 
@@ -905,7 +922,7 @@ int elv_register(struct elevator_type *e)
 
 	/* register, don't allow duplicate names */
 	spin_lock(&elv_list_lock);
-	if (elevator_find(e->elevator_name)) {
+	if (elevator_find(e->elevator_name, e->uses_mq)) {
 		spin_unlock(&elv_list_lock);
 		if (e->icq_cache)
 			kmem_cache_destroy(e->icq_cache);
@@ -915,9 +932,9 @@ int elv_register(struct elevator_type *e)
 	spin_unlock(&elv_list_lock);
 
 	/* print pretty message */
-	if (!strcmp(e->elevator_name, chosen_elevator) ||
+	if (elevator_match(e, chosen_elevator) ||
 			(!*chosen_elevator &&
-			 !strcmp(e->elevator_name, CONFIG_DEFAULT_IOSCHED)))
+			 elevator_match(e, CONFIG_DEFAULT_IOSCHED)))
 				def = " (default)";
 
 	printk(KERN_INFO "io scheduler %s registered%s\n", e->elevator_name,
@@ -1066,25 +1083,15 @@ static int __elevator_change(struct request_queue *q, const char *name)
 		return elevator_switch(q, NULL);
 
 	strlcpy(elevator_name, name, sizeof(elevator_name));
-	e = elevator_get(strstrip(elevator_name), true);
+	e = elevator_get(q, strstrip(elevator_name), true);
 	if (!e)
 		return -EINVAL;
 
-	if (q->elevator &&
-	    !strcmp(elevator_name, q->elevator->type->elevator_name)) {
+	if (q->elevator && elevator_match(q->elevator->type, elevator_name)) {
 		elevator_put(e);
 		return 0;
 	}
 
-	if (!e->uses_mq && q->mq_ops) {
-		elevator_put(e);
-		return -EINVAL;
-	}
-	if (e->uses_mq && !q->mq_ops) {
-		elevator_put(e);
-		return -EINVAL;
-	}
-
 	return elevator_switch(q, e);
 }
 
@@ -1116,9 +1123,10 @@ ssize_t elv_iosched_show(struct request_queue *q, char *name)
 	struct elevator_queue *e = q->elevator;
 	struct elevator_type *elv = NULL;
 	struct elevator_type *__e;
+	bool uses_mq = q->mq_ops != NULL;
 	int len = 0;
 
-	if (!blk_queue_stackable(q))
+	if (!queue_is_rq_based(q))
 		return sprintf(name, "none\n");
 
 	if (!q->elevator)
@@ -1128,7 +1136,8 @@ ssize_t elv_iosched_show(struct request_queue *q, char *name)
 
 	spin_lock(&elv_list_lock);
 	list_for_each_entry(__e, &elv_list, list) {
-		if (elv && !strcmp(elv->elevator_name, __e->elevator_name)) {
+		if (elv && elevator_match(elv, __e->elevator_name) &&
+		    (__e->uses_mq == uses_mq)) {
 			len += sprintf(name+len, "[%s] ", elv->elevator_name);
 			continue;
 		}

block/genhd.c

@@ -588,6 +588,11 @@ static void register_disk(struct device *parent, struct gendisk *disk)
 	disk->part0.holder_dir = kobject_create_and_add("holders", &ddev->kobj);
 	disk->slave_dir = kobject_create_and_add("slaves", &ddev->kobj);
 
+	if (disk->flags & GENHD_FL_HIDDEN) {
+		dev_set_uevent_suppress(ddev, 0);
+		return;
+	}
+
 	/* No minors to use for partitions */
 	if (!disk_part_scan_enabled(disk))
 		goto exit;
@@ -616,6 +621,11 @@ exit:
 	while ((part = disk_part_iter_next(&piter)))
 		kobject_uevent(&part_to_dev(part)->kobj, KOBJ_ADD);
 	disk_part_iter_exit(&piter);
+
+	err = sysfs_create_link(&ddev->kobj,
+				&disk->queue->backing_dev_info->dev->kobj,
+				"bdi");
+	WARN_ON(err);
 }
 
 /**
@@ -630,7 +640,6 @@ exit:
  */
 void device_add_disk(struct device *parent, struct gendisk *disk)
 {
-	struct backing_dev_info *bdi;
 	dev_t devt;
 	int retval;
 
@@ -639,7 +648,8 @@ void device_add_disk(struct device *parent, struct gendisk *disk)
 	 * parameters make sense.
 	 */
 	WARN_ON(disk->minors && !(disk->major || disk->first_minor));
-	WARN_ON(!disk->minors && !(disk->flags & GENHD_FL_EXT_DEVT));
+	WARN_ON(!disk->minors &&
+		!(disk->flags & (GENHD_FL_EXT_DEVT | GENHD_FL_HIDDEN)));
 
 	disk->flags |= GENHD_FL_UP;
 
@@ -648,22 +658,26 @@ void device_add_disk(struct device *parent, struct gendisk *disk)
 		WARN_ON(1);
 		return;
 	}
-	disk_to_dev(disk)->devt = devt;
-
-	/* ->major and ->first_minor aren't supposed to be
-	 * dereferenced from here on, but set them just in case.
-	 */
 	disk->major = MAJOR(devt);
 	disk->first_minor = MINOR(devt);
 
 	disk_alloc_events(disk);
 
-	/* Register BDI before referencing it from bdev */
-	bdi = disk->queue->backing_dev_info;
-	bdi_register_owner(bdi, disk_to_dev(disk));
-
-	blk_register_region(disk_devt(disk), disk->minors, NULL,
-			    exact_match, exact_lock, disk);
+	if (disk->flags & GENHD_FL_HIDDEN) {
+		/*
+		 * Don't let hidden disks show up in /proc/partitions,
+		 * and don't bother scanning for partitions either.
+		 */
+		disk->flags |= GENHD_FL_SUPPRESS_PARTITION_INFO;
+		disk->flags |= GENHD_FL_NO_PART_SCAN;
+	} else {
+		/* Register BDI before referencing it from bdev */
+		disk_to_dev(disk)->devt = devt;
+		bdi_register_owner(disk->queue->backing_dev_info,
+				disk_to_dev(disk));
+		blk_register_region(disk_devt(disk), disk->minors, NULL,
+				    exact_match, exact_lock, disk);
+	}
 	register_disk(parent, disk);
 	blk_register_queue(disk);
 
@@ -673,10 +687,6 @@ void device_add_disk(struct device *parent, struct gendisk *disk)
 	 */
 	WARN_ON_ONCE(!blk_get_queue(disk->queue));
 
-	retval = sysfs_create_link(&disk_to_dev(disk)->kobj, &bdi->dev->kobj,
-				   "bdi");
-	WARN_ON(retval);
-
 	disk_add_events(disk);
 	blk_integrity_add(disk);
 }
@@ -705,7 +715,8 @@ void del_gendisk(struct gendisk *disk)
 	set_capacity(disk, 0);
 	disk->flags &= ~GENHD_FL_UP;
 
-	sysfs_remove_link(&disk_to_dev(disk)->kobj, "bdi");
+	if (!(disk->flags & GENHD_FL_HIDDEN))
+		sysfs_remove_link(&disk_to_dev(disk)->kobj, "bdi");
 	if (disk->queue) {
 		/*
 		 * Unregister bdi before releasing device numbers (as they can
@@ -716,13 +727,15 @@ void del_gendisk(struct gendisk *disk)
 	} else {
 		WARN_ON(1);
 	}
-	blk_unregister_region(disk_devt(disk), disk->minors);
 
-	part_stat_set_all(&disk->part0, 0);
-	disk->part0.stamp = 0;
+	if (!(disk->flags & GENHD_FL_HIDDEN))
+		blk_unregister_region(disk_devt(disk), disk->minors);
 
 	kobject_put(disk->part0.holder_dir);
 	kobject_put(disk->slave_dir);
+
+	part_stat_set_all(&disk->part0, 0);
+	disk->part0.stamp = 0;
 	if (!sysfs_deprecated)
 		sysfs_remove_link(block_depr, dev_name(disk_to_dev(disk)));
 	pm_runtime_set_memalloc_noio(disk_to_dev(disk), false);
@@ -785,6 +798,10 @@ struct gendisk *get_gendisk(dev_t devt, int *partno)
 		spin_unlock_bh(&ext_devt_lock);
 	}
 
+	if (disk && unlikely(disk->flags & GENHD_FL_HIDDEN)) {
+		put_disk(disk);
+		disk = NULL;
+	}
 	return disk;
 }
 EXPORT_SYMBOL(get_gendisk);
@@ -1028,6 +1045,15 @@ static ssize_t disk_removable_show(struct device *dev,
 		       (disk->flags & GENHD_FL_REMOVABLE ? 1 : 0));
 }
 
+static ssize_t disk_hidden_show(struct device *dev,
+				   struct device_attribute *attr, char *buf)
+{
+	struct gendisk *disk = dev_to_disk(dev);
+
+	return sprintf(buf, "%d\n",
+		       (disk->flags & GENHD_FL_HIDDEN ? 1 : 0));
+}
+
 static ssize_t disk_ro_show(struct device *dev,
 				   struct device_attribute *attr, char *buf)
 {
@@ -1065,6 +1091,7 @@ static ssize_t disk_discard_alignment_show(struct device *dev,
 static DEVICE_ATTR(range, S_IRUGO, disk_range_show, NULL);
 static DEVICE_ATTR(ext_range, S_IRUGO, disk_ext_range_show, NULL);
 static DEVICE_ATTR(removable, S_IRUGO, disk_removable_show, NULL);
+static DEVICE_ATTR(hidden, S_IRUGO, disk_hidden_show, NULL);
 static DEVICE_ATTR(ro, S_IRUGO, disk_ro_show, NULL);
 static DEVICE_ATTR(size, S_IRUGO, part_size_show, NULL);
 static DEVICE_ATTR(alignment_offset, S_IRUGO, disk_alignment_offset_show, NULL);
@@ -1089,6 +1116,7 @@ static struct attribute *disk_attrs[] = {
 	&dev_attr_range.attr,
 	&dev_attr_ext_range.attr,
 	&dev_attr_removable.attr,
+	&dev_attr_hidden.attr,
 	&dev_attr_ro.attr,
 	&dev_attr_size.attr,
 	&dev_attr_alignment_offset.attr,

block/ioctl.c

@@ -202,10 +202,16 @@ static int blk_ioctl_discard(struct block_device *bdev, fmode_t mode,
 {
 	uint64_t range[2];
 	uint64_t start, len;
+	struct request_queue *q = bdev_get_queue(bdev);
+	struct address_space *mapping = bdev->bd_inode->i_mapping;
+
 
 	if (!(mode & FMODE_WRITE))
 		return -EBADF;
 
+	if (!blk_queue_discard(q))
+		return -EOPNOTSUPP;
+
 	if (copy_from_user(range, (void __user *)arg, sizeof(range)))
 		return -EFAULT;
 
@@ -216,12 +222,12 @@ static int blk_ioctl_discard(struct block_device *bdev, fmode_t mode,
 		return -EINVAL;
 	if (len & 511)
 		return -EINVAL;
-	start >>= 9;
-	len >>= 9;
 
-	if (start + len > (i_size_read(bdev->bd_inode) >> 9))
+	if (start + len > i_size_read(bdev->bd_inode))
 		return -EINVAL;
-	return blkdev_issue_discard(bdev, start, len, GFP_KERNEL, flags);
+	truncate_inode_pages_range(mapping, start, start + len);
+	return blkdev_issue_discard(bdev, start >> 9, len >> 9,
+				    GFP_KERNEL, flags);
 }
 
 static int blk_ioctl_zeroout(struct block_device *bdev, fmode_t mode,
@@ -437,11 +443,12 @@ static int blkdev_roset(struct block_device *bdev, fmode_t mode,
 {
 	int ret, n;
 
+	if (!capable(CAP_SYS_ADMIN))
+		return -EACCES;
+
 	ret = __blkdev_driver_ioctl(bdev, mode, cmd, arg);
 	if (!is_unrecognized_ioctl(ret))
 		return ret;
-	if (!capable(CAP_SYS_ADMIN))
-		return -EACCES;
 	if (get_user(n, (int __user *)arg))
 		return -EFAULT;
 	set_device_ro(bdev, n);

block/kyber-iosched.c

@@ -541,9 +541,17 @@ static int kyber_get_domain_token(struct kyber_queue_data *kqd,
 
 		/*
 		 * Try again in case a token was freed before we got on the wait
-		 * queue.
+		 * queue. The waker may have already removed the entry from the
+		 * wait queue, but list_del_init() is okay with that.
 		 */
 		nr = __sbitmap_queue_get(domain_tokens);
+		if (nr >= 0) {
+			unsigned long flags;
+
+			spin_lock_irqsave(&ws->wait.lock, flags);
+			list_del_init(&wait->entry);
+			spin_unlock_irqrestore(&ws->wait.lock, flags);
+		}
 	}
 	return nr;
 }
@@ -641,7 +649,7 @@ static bool kyber_has_work(struct blk_mq_hw_ctx *hctx)
 		if (!list_empty_careful(&khd->rqs[i]))
 			return true;
 	}
-	return false;
+	return sbitmap_any_bit_set(&hctx->ctx_map);
 }
 
 #define KYBER_LAT_SHOW_STORE(op)					\

block/mq-deadline.c

@@ -657,6 +657,7 @@ static struct elevator_type mq_deadline = {
 #endif
 	.elevator_attrs = deadline_attrs,
 	.elevator_name = "mq-deadline",
+	.elevator_alias = "deadline",
 	.elevator_owner = THIS_MODULE,
 };
 MODULE_ALIAS("mq-deadline-iosched");

block/scsi_ioctl.c

@@ -207,7 +207,7 @@ static void blk_set_cmd_filter_defaults(struct blk_cmd_filter *filter)
 	__set_bit(GPCMD_SET_READ_AHEAD, filter->write_ok);
 }
 
-int blk_verify_command(unsigned char *cmd, fmode_t has_write_perm)
+int blk_verify_command(unsigned char *cmd, fmode_t mode)
 {
 	struct blk_cmd_filter *filter = &blk_default_cmd_filter;
 
@@ -220,7 +220,7 @@ int blk_verify_command(unsigned char *cmd, fmode_t has_write_perm)
 		return 0;
 
 	/* Write-safe commands require a writable open */
-	if (test_bit(cmd[0], filter->write_ok) && has_write_perm)
+	if (test_bit(cmd[0], filter->write_ok) && (mode & FMODE_WRITE))
 		return 0;
 
 	return -EPERM;
@@ -234,7 +234,7 @@ static int blk_fill_sghdr_rq(struct request_queue *q, struct request *rq,
 
 	if (copy_from_user(req->cmd, hdr->cmdp, hdr->cmd_len))
 		return -EFAULT;
-	if (blk_verify_command(req->cmd, mode & FMODE_WRITE))
+	if (blk_verify_command(req->cmd, mode))
 		return -EPERM;
 
 	/*
@@ -469,7 +469,7 @@ int sg_scsi_ioctl(struct request_queue *q, struct gendisk *disk, fmode_t mode,
 	if (in_len && copy_from_user(buffer, sic->data + cmdlen, in_len))
 		goto error;
 
-	err = blk_verify_command(req->cmd, mode & FMODE_WRITE);
+	err = blk_verify_command(req->cmd, mode);
 	if (err)
 		goto error;
 

drivers/block/Kconfig

@@ -68,9 +68,13 @@ config AMIGA_Z2RAM
 	  To compile this driver as a module, choose M here: the
 	  module will be called z2ram.
 
+config CDROM
+	tristate
+
 config GDROM
 	tristate "SEGA Dreamcast GD-ROM drive"
 	depends on SH_DREAMCAST
+	select CDROM
 	select BLK_SCSI_REQUEST # only for the generic cdrom code
 	help
 	  A standard SEGA Dreamcast comes with a modified CD ROM drive called a
@@ -348,6 +352,7 @@ config BLK_DEV_RAM_DAX
 config CDROM_PKTCDVD
 	tristate "Packet writing on CD/DVD media (DEPRECATED)"
 	depends on !UML
+	select CDROM
 	select BLK_SCSI_REQUEST
 	help
 	  Note: This driver is deprecated and will be removed from the

drivers/block/brd.c

@@ -60,7 +60,6 @@ struct brd_device {
 /*
  * Look up and return a brd's page for a given sector.
  */
-static DEFINE_MUTEX(brd_mutex);
 static struct page *brd_lookup_page(struct brd_device *brd, sector_t sector)
 {
 	pgoff_t idx;

drivers/block/cryptoloop.c

@@ -43,7 +43,6 @@ cryptoloop_init(struct loop_device *lo, const struct loop_info64 *info)
 	int cipher_len;
 	int mode_len;
 	char cms[LO_NAME_SIZE];			/* cipher-mode string */
-	char *cipher;
 	char *mode;
 	char *cmsp = cms;			/* c-m string pointer */
 	struct crypto_skcipher *tfm;
@@ -56,7 +55,6 @@ cryptoloop_init(struct loop_device *lo, const struct loop_info64 *info)
 	strncpy(cms, info->lo_crypt_name, LO_NAME_SIZE);
 	cms[LO_NAME_SIZE - 1] = 0;
 
-	cipher = cmsp;
 	cipher_len = strcspn(cmsp, "-");
 
 	mode = cmsp + cipher_len;

drivers/block/loop.c

@@ -476,6 +476,8 @@ static void lo_rw_aio_complete(struct kiocb *iocb, long ret, long ret2)
 {
 	struct loop_cmd *cmd = container_of(iocb, struct loop_cmd, iocb);
 
+	if (cmd->css)
+		css_put(cmd->css);
 	cmd->ret = ret;
 	lo_rw_aio_do_completion(cmd);
 }
@@ -535,6 +537,8 @@ static int lo_rw_aio(struct loop_device *lo, struct loop_cmd *cmd,
 	cmd->iocb.ki_filp = file;
 	cmd->iocb.ki_complete = lo_rw_aio_complete;
 	cmd->iocb.ki_flags = IOCB_DIRECT;
+	if (cmd->css)
+		kthread_associate_blkcg(cmd->css);
 
 	if (rw == WRITE)
 		ret = call_write_iter(file, &cmd->iocb, &iter);
@@ -542,6 +546,7 @@ static int lo_rw_aio(struct loop_device *lo, struct loop_cmd *cmd,
 		ret = call_read_iter(file, &cmd->iocb, &iter);
 
 	lo_rw_aio_do_completion(cmd);
+	kthread_associate_blkcg(NULL);
 
 	if (ret != -EIOCBQUEUED)
 		cmd->iocb.ki_complete(&cmd->iocb, ret, 0);
@@ -1686,6 +1691,14 @@ static blk_status_t loop_queue_rq(struct blk_mq_hw_ctx *hctx,
 		break;
 	}
 
+	/* always use the first bio's css */
+#ifdef CONFIG_BLK_CGROUP
+	if (cmd->use_aio && cmd->rq->bio && cmd->rq->bio->bi_css) {
+		cmd->css = cmd->rq->bio->bi_css;
+		css_get(cmd->css);
+	} else
+#endif
+		cmd->css = NULL;
 	kthread_queue_work(&lo->worker, &cmd->work);
 
 	return BLK_STS_OK;

drivers/block/loop.h

@@ -72,6 +72,7 @@ struct loop_cmd {
 	long ret;
 	struct kiocb iocb;
 	struct bio_vec *bvec;
+	struct cgroup_subsys_state *css;
 };
 
 /* Support for loadable transfer modules */

drivers/block/mtip32xx/mtip32xx.c

@@ -887,12 +887,9 @@ static void mtip_issue_non_ncq_command(struct mtip_port *port, int tag)
 static bool mtip_pause_ncq(struct mtip_port *port,
 				struct host_to_dev_fis *fis)
 {
-	struct host_to_dev_fis *reply;
 	unsigned long task_file_data;
 
-	reply = port->rxfis + RX_FIS_D2H_REG;
 	task_file_data = readl(port->mmio+PORT_TFDATA);
-
 	if ((task_file_data & 1))
 		return false;
 
@@ -1020,7 +1017,6 @@ static int mtip_exec_internal_command(struct mtip_port *port,
 		.opts = opts
 	};
 	int rv = 0;
-	unsigned long start;
 
 	/* Make sure the buffer is 8 byte aligned. This is asic specific. */
 	if (buffer & 0x00000007) {
@@ -1057,7 +1053,6 @@ static int mtip_exec_internal_command(struct mtip_port *port,
 	/* Copy the command to the command table */
 	memcpy(int_cmd->command, fis, fis_len*4);
 
-	start = jiffies;
 	rq->timeout = timeout;
 
 	/* insert request and run queue */
@@ -3015,7 +3010,6 @@ static int mtip_hw_init(struct driver_data *dd)
 {
 	int i;
 	int rv;
-	unsigned int num_command_slots;
 	unsigned long timeout, timetaken;
 
 	dd->mmio = pcim_iomap_table(dd->pdev)[MTIP_ABAR];
@@ -3025,7 +3019,6 @@ static int mtip_hw_init(struct driver_data *dd)
 		rv = -EIO;
 		goto out1;
 	}
-	num_command_slots = dd->slot_groups * 32;
 
 	hba_setup(dd);
 

drivers/block/nbd.c

@@ -288,15 +288,6 @@ static enum blk_eh_timer_return nbd_xmit_timeout(struct request *req,
 		cmd->status = BLK_STS_TIMEOUT;
 		return BLK_EH_HANDLED;
 	}
-
-	/* If we are waiting on our dead timer then we could get timeout
-	 * callbacks for our request.  For this we just want to reset the timer
-	 * and let the queue side take care of everything.
-	 */
-	if (!completion_done(&cmd->send_complete)) {
-		nbd_config_put(nbd);
-		return BLK_EH_RESET_TIMER;
-	}
 	config = nbd->config;
 
 	if (config->num_connections > 1) {
@@ -723,9 +714,9 @@ static int wait_for_reconnect(struct nbd_device *nbd)
 		return 0;
 	if (test_bit(NBD_DISCONNECTED, &config->runtime_flags))
 		return 0;
-	wait_event_interruptible_timeout(config->conn_wait,
-					 atomic_read(&config->live_connections),
-					 config->dead_conn_timeout);
+	wait_event_timeout(config->conn_wait,
+			   atomic_read(&config->live_connections),
+			   config->dead_conn_timeout);
 	return atomic_read(&config->live_connections);
 }
 
@@ -740,6 +731,7 @@ static int nbd_handle_cmd(struct nbd_cmd *cmd, int index)
 	if (!refcount_inc_not_zero(&nbd->config_refs)) {
 		dev_err_ratelimited(disk_to_dev(nbd->disk),
 				    "Socks array is empty\n");
+		blk_mq_start_request(req);
 		return -EINVAL;
 	}
 	config = nbd->config;
@@ -748,6 +740,7 @@ static int nbd_handle_cmd(struct nbd_cmd *cmd, int index)
 		dev_err_ratelimited(disk_to_dev(nbd->disk),
 				    "Attempted send on invalid socket\n");
 		nbd_config_put(nbd);
+		blk_mq_start_request(req);
 		return -EINVAL;
 	}
 	cmd->status = BLK_STS_OK;
@@ -771,6 +764,7 @@ again:
 			 */
 			sock_shutdown(nbd);
 			nbd_config_put(nbd);
+			blk_mq_start_request(req);
 			return -EIO;
 		}
 		goto again;
@@ -781,6 +775,7 @@ again:
 	 * here so that it gets put _after_ the request that is already on the
 	 * dispatch list.
 	 */
+	blk_mq_start_request(req);
 	if (unlikely(nsock->pending && nsock->pending != req)) {
 		blk_mq_requeue_request(req, true);
 		ret = 0;
@@ -793,10 +788,10 @@ again:
 	ret = nbd_send_cmd(nbd, cmd, index);
 	if (ret == -EAGAIN) {
 		dev_err_ratelimited(disk_to_dev(nbd->disk),
-				    "Request send failed trying another connection\n");
+				    "Request send failed, requeueing\n");
 		nbd_mark_nsock_dead(nbd, nsock, 1);
-		mutex_unlock(&nsock->tx_lock);
-		goto again;
+		blk_mq_requeue_request(req, true);
+		ret = 0;
 	}
 out:
 	mutex_unlock(&nsock->tx_lock);
@@ -820,7 +815,6 @@ static blk_status_t nbd_queue_rq(struct blk_mq_hw_ctx *hctx,
 	 * done sending everything over the wire.
 	 */
 	init_completion(&cmd->send_complete);
-	blk_mq_start_request(bd->rq);
 
 	/* We can be called directly from the user space process, which means we
 	 * could possibly have signals pending so our sendmsg will fail.  In

drivers/block/null_blk.c

@@ -154,6 +154,10 @@ enum {
 	NULL_Q_MQ		= 2,
 };
 
+static int g_no_sched;
+module_param_named(no_sched, g_no_sched, int, S_IRUGO);
+MODULE_PARM_DESC(no_sched, "No io scheduler");
+
 static int g_submit_queues = 1;
 module_param_named(submit_queues, g_submit_queues, int, S_IRUGO);
 MODULE_PARM_DESC(submit_queues, "Number of submission queues");
@@ -1754,6 +1758,8 @@ static int null_init_tag_set(struct nullb *nullb, struct blk_mq_tag_set *set)
 	set->numa_node = nullb ? nullb->dev->home_node : g_home_node;
 	set->cmd_size	= sizeof(struct nullb_cmd);
 	set->flags = BLK_MQ_F_SHOULD_MERGE;
+	if (g_no_sched)
+		set->flags |= BLK_MQ_F_NO_SCHED;
 	set->driver_data = NULL;
 
 	if ((nullb && nullb->dev->blocking) || g_blocking)
@@ -1985,8 +1991,10 @@ static int __init null_init(void)
 
 	for (i = 0; i < nr_devices; i++) {
 		dev = null_alloc_dev();
-		if (!dev)
+		if (!dev) {
+			ret = -ENOMEM;
 			goto err_dev;
+		}
 		ret = null_add_dev(dev);
 		if (ret) {
 			null_free_dev(dev);

drivers/block/paride/Kconfig

@@ -26,6 +26,7 @@ config PARIDE_PD
 config PARIDE_PCD
 	tristate "Parallel port ATAPI CD-ROMs"
 	depends on PARIDE
+	select CDROM
 	select BLK_SCSI_REQUEST # only for the generic cdrom code
 	---help---
 	  This option enables the high-level driver for ATAPI CD-ROM devices

drivers/block/skd_main.c

@@ -1967,7 +1967,8 @@ static void skd_isr_msg_from_dev(struct skd_device *skdev)
 		break;
 
 	case FIT_MTD_CMD_LOG_HOST_ID:
-		skdev->connect_time_stamp = get_seconds();
+		/* hardware interface overflows in y2106 */
+		skdev->connect_time_stamp = (u32)ktime_get_real_seconds();
 		data = skdev->connect_time_stamp & 0xFFFF;
 		mtd = FIT_MXD_CONS(FIT_MTD_CMD_LOG_TIME_STAMP_LO, 0, data);
 		SKD_WRITEL(skdev, mtd, FIT_MSG_TO_DEVICE);

drivers/cdrom/Makefile

@@ -1,14 +1,3 @@
 # SPDX-License-Identifier: GPL-2.0
-# Makefile for the kernel cdrom device drivers.
-#
-# 30 Jan 1998, Michael Elizabeth Chastain, <mailto:mec@shout.net>
-# Rewritten to use lists instead of if-statements.
-
-# Each configuration option enables a list of files.
-
-obj-$(CONFIG_BLK_DEV_IDECD)	+=              cdrom.o
-obj-$(CONFIG_BLK_DEV_SR)	+=              cdrom.o
-obj-$(CONFIG_PARIDE_PCD)	+=		cdrom.o
-obj-$(CONFIG_CDROM_PKTCDVD)	+=		cdrom.o
-
-obj-$(CONFIG_GDROM)		+= gdrom.o      cdrom.o
+obj-$(CONFIG_CDROM)	+= cdrom.o
+obj-$(CONFIG_GDROM)	+= gdrom.o

drivers/ide/Kconfig

@@ -117,7 +117,9 @@ config BLK_DEV_DELKIN
 
 config BLK_DEV_IDECD
 	tristate "Include IDE/ATAPI CDROM support"
+	depends on BLK_DEV
 	select IDE_ATAPI
+	select CDROM
 	---help---
 	  If you have a CD-ROM drive using the ATAPI protocol, say Y. ATAPI is
 	  a newer protocol used by IDE CD-ROM and TAPE drives, similar to the

drivers/ide/ide-atapi.c

@@ -282,7 +282,7 @@ int ide_cd_expiry(ide_drive_t *drive)
 	struct request *rq = drive->hwif->rq;
 	unsigned long wait = 0;
 
-	debug_log("%s: rq->cmd[0]: 0x%x\n", __func__, rq->cmd[0]);
+	debug_log("%s: scsi_req(rq)->cmd[0]: 0x%x\n", __func__, scsi_req(rq)->cmd[0]);
 
 	/*
 	 * Some commands are *slow* and normally take a long time to complete.
@@ -463,7 +463,7 @@ static ide_startstop_t ide_pc_intr(ide_drive_t *drive)
 				return ide_do_reset(drive);
 			}
 
-			debug_log("[cmd %x]: check condition\n", rq->cmd[0]);
+			debug_log("[cmd %x]: check condition\n", scsi_req(rq)->cmd[0]);
 
 			/* Retry operation */
 			ide_retry_pc(drive);
@@ -531,7 +531,7 @@ static ide_startstop_t ide_pc_intr(ide_drive_t *drive)
 		ide_pad_transfer(drive, write, bcount);
 
 	debug_log("[cmd %x] transferred %d bytes, padded %d bytes, resid: %u\n",
-		  rq->cmd[0], done, bcount, scsi_req(rq)->resid_len);
+		  scsi_req(rq)->cmd[0], done, bcount, scsi_req(rq)->resid_len);
 
 	/* And set the interrupt handler again */
 	ide_set_handler(drive, ide_pc_intr, timeout);

drivers/ide/ide-pm.c

@@ -90,9 +90,9 @@ int generic_ide_resume(struct device *dev)
 	}
 
 	memset(&rqpm, 0, sizeof(rqpm));
-	rq = blk_get_request(drive->queue, REQ_OP_DRV_IN, __GFP_RECLAIM);
+	rq = blk_get_request_flags(drive->queue, REQ_OP_DRV_IN,
+				   BLK_MQ_REQ_PREEMPT);
 	ide_req(rq)->type = ATA_PRIV_PM_RESUME;
-	rq->rq_flags |= RQF_PREEMPT;
 	rq->special = &rqpm;
 	rqpm.pm_step = IDE_PM_START_RESUME;
 	rqpm.pm_state = PM_EVENT_ON;

drivers/lightnvm/Kconfig

@@ -4,7 +4,8 @@
 
 menuconfig NVM
 	bool "Open-Channel SSD target support"
-	depends on BLOCK && HAS_DMA
+	depends on BLOCK && HAS_DMA && PCI
+	select BLK_DEV_NVME
 	help
 	  Say Y here to get to enable Open-channel SSDs.
 

drivers/lightnvm/core.c

@@ -22,6 +22,7 @@
 #include <linux/types.h>
 #include <linux/sem.h>
 #include <linux/bitmap.h>
+#include <linux/module.h>
 #include <linux/moduleparam.h>
 #include <linux/miscdevice.h>
 #include <linux/lightnvm.h>
@@ -138,7 +139,6 @@ static struct nvm_tgt_dev *nvm_create_tgt_dev(struct nvm_dev *dev,
 	int prev_nr_luns;
 	int i, j;
 
-	nr_chnls = nr_luns / dev->geo.luns_per_chnl;
 	nr_chnls = (nr_chnls_mod == 0) ? nr_chnls : nr_chnls + 1;
 
 	dev_map = kmalloc(sizeof(struct nvm_dev_map), GFP_KERNEL);
@@ -226,6 +226,24 @@ static const struct block_device_operations nvm_fops = {
 	.owner		= THIS_MODULE,
 };
 
+static struct nvm_tgt_type *nvm_find_target_type(const char *name, int lock)
+{
+	struct nvm_tgt_type *tmp, *tt = NULL;
+
+	if (lock)
+		down_write(&nvm_tgtt_lock);
+
+	list_for_each_entry(tmp, &nvm_tgt_types, list)
+		if (!strcmp(name, tmp->name)) {
+			tt = tmp;
+			break;
+		}
+
+	if (lock)
+		up_write(&nvm_tgtt_lock);
+	return tt;
+}
+
 static int nvm_create_tgt(struct nvm_dev *dev, struct nvm_ioctl_create *create)
 {
 	struct nvm_ioctl_create_simple *s = &create->conf.s;
@@ -316,6 +334,8 @@ static int nvm_create_tgt(struct nvm_dev *dev, struct nvm_ioctl_create *create)
 	list_add_tail(&t->list, &dev->targets);
 	mutex_unlock(&dev->mlock);
 
+	__module_get(tt->owner);
+
 	return 0;
 err_sysfs:
 	if (tt->exit)
@@ -351,6 +371,7 @@ static void __nvm_remove_target(struct nvm_target *t)
 
 	nvm_remove_tgt_dev(t->dev, 1);
 	put_disk(tdisk);
+	module_put(t->type->owner);
 
 	list_del(&t->list);
 	kfree(t);
@@ -532,25 +553,6 @@ void nvm_part_to_tgt(struct nvm_dev *dev, sector_t *entries,
 }
 EXPORT_SYMBOL(nvm_part_to_tgt);
 
-struct nvm_tgt_type *nvm_find_target_type(const char *name, int lock)
-{
-	struct nvm_tgt_type *tmp, *tt = NULL;
-
-	if (lock)
-		down_write(&nvm_tgtt_lock);
-
-	list_for_each_entry(tmp, &nvm_tgt_types, list)
-		if (!strcmp(name, tmp->name)) {
-			tt = tmp;
-			break;
-		}
-
-	if (lock)
-		up_write(&nvm_tgtt_lock);
-	return tt;
-}
-EXPORT_SYMBOL(nvm_find_target_type);
-
 int nvm_register_tgt_type(struct nvm_tgt_type *tt)
 {
 	int ret = 0;
@@ -571,9 +573,9 @@ void nvm_unregister_tgt_type(struct nvm_tgt_type *tt)
 	if (!tt)
 		return;
 
-	down_write(&nvm_lock);
+	down_write(&nvm_tgtt_lock);
 	list_del(&tt->list);
-	up_write(&nvm_lock);
+	up_write(&nvm_tgtt_lock);
 }
 EXPORT_SYMBOL(nvm_unregister_tgt_type);
 
@@ -602,6 +604,52 @@ static struct nvm_dev *nvm_find_nvm_dev(const char *name)
 	return NULL;
 }
 
+static int nvm_set_rqd_ppalist(struct nvm_tgt_dev *tgt_dev, struct nvm_rq *rqd,
+			const struct ppa_addr *ppas, int nr_ppas)
+{
+	struct nvm_dev *dev = tgt_dev->parent;
+	struct nvm_geo *geo = &tgt_dev->geo;
+	int i, plane_cnt, pl_idx;
+	struct ppa_addr ppa;
+
+	if (geo->plane_mode == NVM_PLANE_SINGLE && nr_ppas == 1) {
+		rqd->nr_ppas = nr_ppas;
+		rqd->ppa_addr = ppas[0];
+
+		return 0;
+	}
+
+	rqd->nr_ppas = nr_ppas;
+	rqd->ppa_list = nvm_dev_dma_alloc(dev, GFP_KERNEL, &rqd->dma_ppa_list);
+	if (!rqd->ppa_list) {
+		pr_err("nvm: failed to allocate dma memory\n");
+		return -ENOMEM;
+	}
+
+	plane_cnt = geo->plane_mode;
+	rqd->nr_ppas *= plane_cnt;
+
+	for (i = 0; i < nr_ppas; i++) {
+		for (pl_idx = 0; pl_idx < plane_cnt; pl_idx++) {
+			ppa = ppas[i];
+			ppa.g.pl = pl_idx;
+			rqd->ppa_list[(pl_idx * nr_ppas) + i] = ppa;
+		}
+	}
+
+	return 0;
+}
+
+static void nvm_free_rqd_ppalist(struct nvm_tgt_dev *tgt_dev,
+			struct nvm_rq *rqd)
+{
+	if (!rqd->ppa_list)
+		return;
+
+	nvm_dev_dma_free(tgt_dev->parent, rqd->ppa_list, rqd->dma_ppa_list);
+}
+
+
 int nvm_set_tgt_bb_tbl(struct nvm_tgt_dev *tgt_dev, struct ppa_addr *ppas,
 		       int nr_ppas, int type)
 {
@@ -616,7 +664,7 @@ int nvm_set_tgt_bb_tbl(struct nvm_tgt_dev *tgt_dev, struct ppa_addr *ppas,
 
 	memset(&rqd, 0, sizeof(struct nvm_rq));
 
-	nvm_set_rqd_ppalist(tgt_dev, &rqd, ppas, nr_ppas, 1);
+	nvm_set_rqd_ppalist(tgt_dev, &rqd, ppas, nr_ppas);
 	nvm_rq_tgt_to_dev(tgt_dev, &rqd);
 
 	ret = dev->ops->set_bb_tbl(dev, &rqd.ppa_addr, rqd.nr_ppas, type);
@@ -658,12 +706,25 @@ int nvm_submit_io(struct nvm_tgt_dev *tgt_dev, struct nvm_rq *rqd)
 }
 EXPORT_SYMBOL(nvm_submit_io);
 
-static void nvm_end_io_sync(struct nvm_rq *rqd)
+int nvm_submit_io_sync(struct nvm_tgt_dev *tgt_dev, struct nvm_rq *rqd)
 {
-	struct completion *waiting = rqd->private;
+	struct nvm_dev *dev = tgt_dev->parent;
+	int ret;
 
-	complete(waiting);
+	if (!dev->ops->submit_io_sync)
+		return -ENODEV;
+
+	nvm_rq_tgt_to_dev(tgt_dev, rqd);
+
+	rqd->dev = tgt_dev;
+
+	/* In case of error, fail with right address format */
+	ret = dev->ops->submit_io_sync(dev, rqd);
+	nvm_rq_dev_to_tgt(tgt_dev, rqd);
+
+	return ret;
 }
+EXPORT_SYMBOL(nvm_submit_io_sync);
 
 int nvm_erase_sync(struct nvm_tgt_dev *tgt_dev, struct ppa_addr *ppas,
 								int nr_ppas)
@@ -671,25 +732,21 @@ int nvm_erase_sync(struct nvm_tgt_dev *tgt_dev, struct ppa_addr *ppas,
 	struct nvm_geo *geo = &tgt_dev->geo;
 	struct nvm_rq rqd;
 	int ret;
-	DECLARE_COMPLETION_ONSTACK(wait);
 
 	memset(&rqd, 0, sizeof(struct nvm_rq));
 
 	rqd.opcode = NVM_OP_ERASE;
-	rqd.end_io = nvm_end_io_sync;
-	rqd.private = &wait;
 	rqd.flags = geo->plane_mode >> 1;
 
-	ret = nvm_set_rqd_ppalist(tgt_dev, &rqd, ppas, nr_ppas, 1);
+	ret = nvm_set_rqd_ppalist(tgt_dev, &rqd, ppas, nr_ppas);
 	if (ret)
 		return ret;
 
-	ret = nvm_submit_io(tgt_dev, &rqd);
+	ret = nvm_submit_io_sync(tgt_dev, &rqd);
 	if (ret) {
 		pr_err("rrpr: erase I/O submission failed: %d\n", ret);
 		goto free_ppa_list;
 	}
-	wait_for_completion_io(&wait);
 
 free_ppa_list:
 	nvm_free_rqd_ppalist(tgt_dev, &rqd);
@@ -775,57 +832,6 @@ void nvm_put_area(struct nvm_tgt_dev *tgt_dev, sector_t begin)
 }
 EXPORT_SYMBOL(nvm_put_area);
 
-int nvm_set_rqd_ppalist(struct nvm_tgt_dev *tgt_dev, struct nvm_rq *rqd,
-			const struct ppa_addr *ppas, int nr_ppas, int vblk)
-{
-	struct nvm_dev *dev = tgt_dev->parent;
-	struct nvm_geo *geo = &tgt_dev->geo;
-	int i, plane_cnt, pl_idx;
-	struct ppa_addr ppa;
-
-	if ((!vblk || geo->plane_mode == NVM_PLANE_SINGLE) && nr_ppas == 1) {
-		rqd->nr_ppas = nr_ppas;
-		rqd->ppa_addr = ppas[0];
-
-		return 0;
-	}
-
-	rqd->nr_ppas = nr_ppas;
-	rqd->ppa_list = nvm_dev_dma_alloc(dev, GFP_KERNEL, &rqd->dma_ppa_list);
-	if (!rqd->ppa_list) {
-		pr_err("nvm: failed to allocate dma memory\n");
-		return -ENOMEM;
-	}
-
-	if (!vblk) {
-		for (i = 0; i < nr_ppas; i++)
-			rqd->ppa_list[i] = ppas[i];
-	} else {
-		plane_cnt = geo->plane_mode;
-		rqd->nr_ppas *= plane_cnt;
-
-		for (i = 0; i < nr_ppas; i++) {
-			for (pl_idx = 0; pl_idx < plane_cnt; pl_idx++) {
-				ppa = ppas[i];
-				ppa.g.pl = pl_idx;
-				rqd->ppa_list[(pl_idx * nr_ppas) + i] = ppa;
-			}
-		}
-	}
-
-	return 0;
-}
-EXPORT_SYMBOL(nvm_set_rqd_ppalist);
-
-void nvm_free_rqd_ppalist(struct nvm_tgt_dev *tgt_dev, struct nvm_rq *rqd)
-{
-	if (!rqd->ppa_list)
-		return;
-
-	nvm_dev_dma_free(tgt_dev->parent, rqd->ppa_list, rqd->dma_ppa_list);
-}
-EXPORT_SYMBOL(nvm_free_rqd_ppalist);
-
 void nvm_end_io(struct nvm_rq *rqd)
 {
 	struct nvm_tgt_dev *tgt_dev = rqd->dev;
@@ -1177,7 +1183,7 @@ static long nvm_ioctl_info(struct file *file, void __user *arg)
 	info->version[1] = NVM_VERSION_MINOR;
 	info->version[2] = NVM_VERSION_PATCH;
 
-	down_write(&nvm_lock);
+	down_write(&nvm_tgtt_lock);
 	list_for_each_entry(tt, &nvm_tgt_types, list) {
 		struct nvm_ioctl_info_tgt *tgt = &info->tgts[tgt_iter];
 
@@ -1190,7 +1196,7 @@ static long nvm_ioctl_info(struct file *file, void __user *arg)
 	}
 
 	info->tgtsize = tgt_iter;
-	up_write(&nvm_lock);
+	up_write(&nvm_tgtt_lock);
 
 	if (copy_to_user(arg, info, sizeof(struct nvm_ioctl_info))) {
 		kfree(info);

drivers/lightnvm/pblk-cache.c

@@ -43,8 +43,10 @@ retry:
 	if (unlikely(!bio_has_data(bio)))
 		goto out;
 
-	w_ctx.flags = flags;
 	pblk_ppa_set_empty(&w_ctx.ppa);
+	w_ctx.flags = flags;
+	if (bio->bi_opf & REQ_PREFLUSH)
+		w_ctx.flags |= PBLK_FLUSH_ENTRY;
 
 	for (i = 0; i < nr_entries; i++) {
 		void *data = bio_data(bio);
@@ -73,12 +75,11 @@ out:
  * On GC the incoming lbas are not necessarily sequential. Also, some of the
  * lbas might not be valid entries, which are marked as empty by the GC thread
  */
-int pblk_write_gc_to_cache(struct pblk *pblk, void *data, u64 *lba_list,
-			   unsigned int nr_entries, unsigned int nr_rec_entries,
-			   struct pblk_line *gc_line, unsigned long flags)
+int pblk_write_gc_to_cache(struct pblk *pblk, struct pblk_gc_rq *gc_rq)
 {
 	struct pblk_w_ctx w_ctx;
 	unsigned int bpos, pos;
+	void *data = gc_rq->data;
 	int i, valid_entries;
 
 	/* Update the write buffer head (mem) with the entries that we can
@@ -86,28 +87,29 @@ int pblk_write_gc_to_cache(struct pblk *pblk, void *data, u64 *lba_list,
 	 * rollback from here on.
 	 */
 retry:
-	if (!pblk_rb_may_write_gc(&pblk->rwb, nr_rec_entries, &bpos)) {
+	if (!pblk_rb_may_write_gc(&pblk->rwb, gc_rq->secs_to_gc, &bpos)) {
 		io_schedule();
 		goto retry;
 	}
 
-	w_ctx.flags = flags;
+	w_ctx.flags = PBLK_IOTYPE_GC;
 	pblk_ppa_set_empty(&w_ctx.ppa);
 
-	for (i = 0, valid_entries = 0; i < nr_entries; i++) {
-		if (lba_list[i] == ADDR_EMPTY)
+	for (i = 0, valid_entries = 0; i < gc_rq->nr_secs; i++) {
+		if (gc_rq->lba_list[i] == ADDR_EMPTY)
 			continue;
 
-		w_ctx.lba = lba_list[i];
+		w_ctx.lba = gc_rq->lba_list[i];
 
 		pos = pblk_rb_wrap_pos(&pblk->rwb, bpos + valid_entries);
-		pblk_rb_write_entry_gc(&pblk->rwb, data, w_ctx, gc_line, pos);
+		pblk_rb_write_entry_gc(&pblk->rwb, data, w_ctx, gc_rq->line,
+						gc_rq->paddr_list[i], pos);
 
 		data += PBLK_EXPOSED_PAGE_SIZE;
 		valid_entries++;
 	}
 
-	WARN_ONCE(nr_rec_entries != valid_entries,
+	WARN_ONCE(gc_rq->secs_to_gc != valid_entries,
 					"pblk: inconsistent GC write\n");
 
 #ifdef CONFIG_NVM_DEBUG

drivers/lightnvm/pblk-core.c

@@ -18,6 +18,31 @@
 
 #include "pblk.h"
 
+static void pblk_line_mark_bb(struct work_struct *work)
+{
+	struct pblk_line_ws *line_ws = container_of(work, struct pblk_line_ws,
+									ws);
+	struct pblk *pblk = line_ws->pblk;
+	struct nvm_tgt_dev *dev = pblk->dev;
+	struct ppa_addr *ppa = line_ws->priv;
+	int ret;
+
+	ret = nvm_set_tgt_bb_tbl(dev, ppa, 1, NVM_BLK_T_GRWN_BAD);
+	if (ret) {
+		struct pblk_line *line;
+		int pos;
+
+		line = &pblk->lines[pblk_dev_ppa_to_line(*ppa)];
+		pos = pblk_dev_ppa_to_pos(&dev->geo, *ppa);
+
+		pr_err("pblk: failed to mark bb, line:%d, pos:%d\n",
+				line->id, pos);
+	}
+
+	kfree(ppa);
+	mempool_free(line_ws, pblk->gen_ws_pool);
+}
+
 static void pblk_mark_bb(struct pblk *pblk, struct pblk_line *line,
 			 struct ppa_addr *ppa)
 {
@@ -33,7 +58,8 @@ static void pblk_mark_bb(struct pblk *pblk, struct pblk_line *line,
 		pr_err("pblk: attempted to erase bb: line:%d, pos:%d\n",
 							line->id, pos);
 
-	pblk_line_run_ws(pblk, NULL, ppa, pblk_line_mark_bb, pblk->bb_wq);
+	pblk_gen_run_ws(pblk, NULL, ppa, pblk_line_mark_bb,
+						GFP_ATOMIC, pblk->bb_wq);
 }
 
 static void __pblk_end_io_erase(struct pblk *pblk, struct nvm_rq *rqd)
@@ -63,7 +89,7 @@ static void pblk_end_io_erase(struct nvm_rq *rqd)
 	struct pblk *pblk = rqd->private;
 
 	__pblk_end_io_erase(pblk, rqd);
-	mempool_free(rqd, pblk->g_rq_pool);
+	mempool_free(rqd, pblk->e_rq_pool);
 }
 
 void __pblk_map_invalidate(struct pblk *pblk, struct pblk_line *line,
@@ -77,11 +103,7 @@ void __pblk_map_invalidate(struct pblk *pblk, struct pblk_line *line,
 	 * that newer updates are not overwritten.
 	 */
 	spin_lock(&line->lock);
-	if (line->state == PBLK_LINESTATE_GC ||
-					line->state == PBLK_LINESTATE_FREE) {
-		spin_unlock(&line->lock);
-		return;
-	}
+	WARN_ON(line->state == PBLK_LINESTATE_FREE);
 
 	if (test_and_set_bit(paddr, line->invalid_bitmap)) {
 		WARN_ONCE(1, "pblk: double invalidate\n");
@@ -98,8 +120,7 @@ void __pblk_map_invalidate(struct pblk *pblk, struct pblk_line *line,
 		spin_lock(&l_mg->gc_lock);
 		spin_lock(&line->lock);
 		/* Prevent moving a line that has just been chosen for GC */
-		if (line->state == PBLK_LINESTATE_GC ||
-					line->state == PBLK_LINESTATE_FREE) {
+		if (line->state == PBLK_LINESTATE_GC) {
 			spin_unlock(&line->lock);
 			spin_unlock(&l_mg->gc_lock);
 			return;
@@ -150,17 +171,25 @@ static void pblk_invalidate_range(struct pblk *pblk, sector_t slba,
 	spin_unlock(&pblk->trans_lock);
 }
 
-struct nvm_rq *pblk_alloc_rqd(struct pblk *pblk, int rw)
+/* Caller must guarantee that the request is a valid type */
+struct nvm_rq *pblk_alloc_rqd(struct pblk *pblk, int type)
 {
 	mempool_t *pool;
 	struct nvm_rq *rqd;
 	int rq_size;
 
-	if (rw == WRITE) {
+	switch (type) {
+	case PBLK_WRITE:
+	case PBLK_WRITE_INT:
 		pool = pblk->w_rq_pool;
 		rq_size = pblk_w_rq_size;
-	} else {
-		pool = pblk->g_rq_pool;
+		break;
+	case PBLK_READ:
+		pool = pblk->r_rq_pool;
+		rq_size = pblk_g_rq_size;
+		break;
+	default:
+		pool = pblk->e_rq_pool;
 		rq_size = pblk_g_rq_size;
 	}
 
@@ -170,15 +199,30 @@ struct nvm_rq *pblk_alloc_rqd(struct pblk *pblk, int rw)
 	return rqd;
 }
 
-void pblk_free_rqd(struct pblk *pblk, struct nvm_rq *rqd, int rw)
+/* Typically used on completion path. Cannot guarantee request consistency */
+void pblk_free_rqd(struct pblk *pblk, struct nvm_rq *rqd, int type)
 {
+	struct nvm_tgt_dev *dev = pblk->dev;
 	mempool_t *pool;
 
-	if (rw == WRITE)
+	switch (type) {
+	case PBLK_WRITE:
+		kfree(((struct pblk_c_ctx *)nvm_rq_to_pdu(rqd))->lun_bitmap);
+	case PBLK_WRITE_INT:
 		pool = pblk->w_rq_pool;
-	else
-		pool = pblk->g_rq_pool;
+		break;
+	case PBLK_READ:
+		pool = pblk->r_rq_pool;
+		break;
+	case PBLK_ERASE:
+		pool = pblk->e_rq_pool;
+		break;
+	default:
+		pr_err("pblk: trying to free unknown rqd type\n");
+		return;
+	}
 
+	nvm_dev_dma_free(dev->parent, rqd->meta_list, rqd->dma_meta_list);
 	mempool_free(rqd, pool);
 }
 
@@ -190,10 +234,9 @@ void pblk_bio_free_pages(struct pblk *pblk, struct bio *bio, int off,
 
 	WARN_ON(off + nr_pages != bio->bi_vcnt);
 
-	bio_advance(bio, off * PBLK_EXPOSED_PAGE_SIZE);
 	for (i = off; i < nr_pages + off; i++) {
 		bv = bio->bi_io_vec[i];
-		mempool_free(bv.bv_page, pblk->page_pool);
+		mempool_free(bv.bv_page, pblk->page_bio_pool);
 	}
 }
 
@@ -205,14 +248,12 @@ int pblk_bio_add_pages(struct pblk *pblk, struct bio *bio, gfp_t flags,
 	int i, ret;
 
 	for (i = 0; i < nr_pages; i++) {
-		page = mempool_alloc(pblk->page_pool, flags);
-		if (!page)
-			goto err;
+		page = mempool_alloc(pblk->page_bio_pool, flags);
 
 		ret = bio_add_pc_page(q, bio, page, PBLK_EXPOSED_PAGE_SIZE, 0);
 		if (ret != PBLK_EXPOSED_PAGE_SIZE) {
 			pr_err("pblk: could not add page to bio\n");
-			mempool_free(page, pblk->page_pool);
+			mempool_free(page, pblk->page_bio_pool);
 			goto err;
 		}
 	}
@@ -245,13 +286,6 @@ void pblk_write_should_kick(struct pblk *pblk)
 		pblk_write_kick(pblk);
 }
 
-void pblk_end_bio_sync(struct bio *bio)
-{
-	struct completion *waiting = bio->bi_private;
-
-	complete(waiting);
-}
-
 void pblk_end_io_sync(struct nvm_rq *rqd)
 {
 	struct completion *waiting = rqd->private;
@@ -259,7 +293,7 @@ void pblk_end_io_sync(struct nvm_rq *rqd)
 	complete(waiting);
 }
 
-void pblk_wait_for_meta(struct pblk *pblk)
+static void pblk_wait_for_meta(struct pblk *pblk)
 {
 	do {
 		if (!atomic_read(&pblk->inflight_io))
@@ -336,17 +370,6 @@ void pblk_discard(struct pblk *pblk, struct bio *bio)
 	pblk_invalidate_range(pblk, slba, nr_secs);
 }
 
-struct ppa_addr pblk_get_lba_map(struct pblk *pblk, sector_t lba)
-{
-	struct ppa_addr ppa;
-
-	spin_lock(&pblk->trans_lock);
-	ppa = pblk_trans_map_get(pblk, lba);
-	spin_unlock(&pblk->trans_lock);
-
-	return ppa;
-}
-
 void pblk_log_write_err(struct pblk *pblk, struct nvm_rq *rqd)
 {
 	atomic_long_inc(&pblk->write_failed);
@@ -389,39 +412,38 @@ int pblk_submit_io(struct pblk *pblk, struct nvm_rq *rqd)
 	struct nvm_tgt_dev *dev = pblk->dev;
 
 #ifdef CONFIG_NVM_DEBUG
-	struct ppa_addr *ppa_list;
+	int ret;
 
-	ppa_list = (rqd->nr_ppas > 1) ? rqd->ppa_list : &rqd->ppa_addr;
-	if (pblk_boundary_ppa_checks(dev, ppa_list, rqd->nr_ppas)) {
-		WARN_ON(1);
-		return -EINVAL;
-	}
+	ret = pblk_check_io(pblk, rqd);
+	if (ret)
+		return ret;
+#endif
 
-	if (rqd->opcode == NVM_OP_PWRITE) {
-		struct pblk_line *line;
-		struct ppa_addr ppa;
-		int i;
+	atomic_inc(&pblk->inflight_io);
 
-		for (i = 0; i < rqd->nr_ppas; i++) {
-			ppa = ppa_list[i];
-			line = &pblk->lines[pblk_dev_ppa_to_line(ppa)];
+	return nvm_submit_io(dev, rqd);
+}
 
-			spin_lock(&line->lock);
-			if (line->state != PBLK_LINESTATE_OPEN) {
-				pr_err("pblk: bad ppa: line:%d,state:%d\n",
-							line->id, line->state);
-				WARN_ON(1);
-				spin_unlock(&line->lock);
-				return -EINVAL;
-			}
-			spin_unlock(&line->lock);
-		}
-	}
+int pblk_submit_io_sync(struct pblk *pblk, struct nvm_rq *rqd)
+{
+	struct nvm_tgt_dev *dev = pblk->dev;
+
+#ifdef CONFIG_NVM_DEBUG
+	int ret;
+
+	ret = pblk_check_io(pblk, rqd);
+	if (ret)
+		return ret;
 #endif
 
 	atomic_inc(&pblk->inflight_io);
 
-	return nvm_submit_io(dev, rqd);
+	return nvm_submit_io_sync(dev, rqd);
+}
+
+static void pblk_bio_map_addr_endio(struct bio *bio)
+{
+	bio_put(bio);
 }
 
 struct bio *pblk_bio_map_addr(struct pblk *pblk, void *data,
@@ -460,6 +482,8 @@ struct bio *pblk_bio_map_addr(struct pblk *pblk, void *data,
 
 		kaddr += PAGE_SIZE;
 	}
+
+	bio->bi_end_io = pblk_bio_map_addr_endio;
 out:
 	return bio;
 }
@@ -486,12 +510,14 @@ void pblk_dealloc_page(struct pblk *pblk, struct pblk_line *line, int nr_secs)
 	u64 addr;
 	int i;
 
+	spin_lock(&line->lock);
 	addr = find_next_zero_bit(line->map_bitmap,
 					pblk->lm.sec_per_line, line->cur_sec);
 	line->cur_sec = addr - nr_secs;
 
 	for (i = 0; i < nr_secs; i++, line->cur_sec--)
 		WARN_ON(!test_and_clear_bit(line->cur_sec, line->map_bitmap));
+	spin_unlock(&line->lock);
 }
 
 u64 __pblk_alloc_page(struct pblk *pblk, struct pblk_line *line, int nr_secs)
@@ -565,12 +591,11 @@ static int pblk_line_submit_emeta_io(struct pblk *pblk, struct pblk_line *line,
 	int cmd_op, bio_op;
 	int i, j;
 	int ret;
-	DECLARE_COMPLETION_ONSTACK(wait);
 
-	if (dir == WRITE) {
+	if (dir == PBLK_WRITE) {
 		bio_op = REQ_OP_WRITE;
 		cmd_op = NVM_OP_PWRITE;
-	} else if (dir == READ) {
+	} else if (dir == PBLK_READ) {
 		bio_op = REQ_OP_READ;
 		cmd_op = NVM_OP_PREAD;
 	} else
@@ -607,13 +632,11 @@ next_rq:
 	rqd.dma_ppa_list = dma_ppa_list;
 	rqd.opcode = cmd_op;
 	rqd.nr_ppas = rq_ppas;
-	rqd.end_io = pblk_end_io_sync;
-	rqd.private = &wait;
 
-	if (dir == WRITE) {
+	if (dir == PBLK_WRITE) {
 		struct pblk_sec_meta *meta_list = rqd.meta_list;
 
-		rqd.flags = pblk_set_progr_mode(pblk, WRITE);
+		rqd.flags = pblk_set_progr_mode(pblk, PBLK_WRITE);
 		for (i = 0; i < rqd.nr_ppas; ) {
 			spin_lock(&line->lock);
 			paddr = __pblk_alloc_page(pblk, line, min);
@@ -662,25 +685,17 @@ next_rq:
 		}
 	}
 
-	ret = pblk_submit_io(pblk, &rqd);
+	ret = pblk_submit_io_sync(pblk, &rqd);
 	if (ret) {
 		pr_err("pblk: emeta I/O submission failed: %d\n", ret);
 		bio_put(bio);
 		goto free_rqd_dma;
 	}
 
-	if (!wait_for_completion_io_timeout(&wait,
-				msecs_to_jiffies(PBLK_COMMAND_TIMEOUT_MS))) {
-		pr_err("pblk: emeta I/O timed out\n");
-	}
 	atomic_dec(&pblk->inflight_io);
-	reinit_completion(&wait);
-
-	if (likely(pblk->l_mg.emeta_alloc_type == PBLK_VMALLOC_META))
-		bio_put(bio);
 
 	if (rqd.error) {
-		if (dir == WRITE)
+		if (dir == PBLK_WRITE)
 			pblk_log_write_err(pblk, &rqd);
 		else
 			pblk_log_read_err(pblk, &rqd);
@@ -721,14 +736,13 @@ static int pblk_line_submit_smeta_io(struct pblk *pblk, struct pblk_line *line,
 	int i, ret;
 	int cmd_op, bio_op;
 	int flags;
-	DECLARE_COMPLETION_ONSTACK(wait);
 
-	if (dir == WRITE) {
+	if (dir == PBLK_WRITE) {
 		bio_op = REQ_OP_WRITE;
 		cmd_op = NVM_OP_PWRITE;
-		flags = pblk_set_progr_mode(pblk, WRITE);
+		flags = pblk_set_progr_mode(pblk, PBLK_WRITE);
 		lba_list = emeta_to_lbas(pblk, line->emeta->buf);
-	} else if (dir == READ) {
+	} else if (dir == PBLK_READ) {
 		bio_op = REQ_OP_READ;
 		cmd_op = NVM_OP_PREAD;
 		flags = pblk_set_read_mode(pblk, PBLK_READ_SEQUENTIAL);
@@ -758,15 +772,13 @@ static int pblk_line_submit_smeta_io(struct pblk *pblk, struct pblk_line *line,
 	rqd.opcode = cmd_op;
 	rqd.flags = flags;
 	rqd.nr_ppas = lm->smeta_sec;
-	rqd.end_io = pblk_end_io_sync;
-	rqd.private = &wait;
 
 	for (i = 0; i < lm->smeta_sec; i++, paddr++) {
 		struct pblk_sec_meta *meta_list = rqd.meta_list;
 
 		rqd.ppa_list[i] = addr_to_gen_ppa(pblk, paddr, line->id);
 
-		if (dir == WRITE) {
+		if (dir == PBLK_WRITE) {
 			__le64 addr_empty = cpu_to_le64(ADDR_EMPTY);
 
 			meta_list[i].lba = lba_list[paddr] = addr_empty;
@@ -778,21 +790,17 @@ static int pblk_line_submit_smeta_io(struct pblk *pblk, struct pblk_line *line,
 	 * the write thread is the only one sending write and erase commands,
 	 * there is no need to take the LUN semaphore.
 	 */
-	ret = pblk_submit_io(pblk, &rqd);
+	ret = pblk_submit_io_sync(pblk, &rqd);
 	if (ret) {
 		pr_err("pblk: smeta I/O submission failed: %d\n", ret);
 		bio_put(bio);
 		goto free_ppa_list;
 	}
 
-	if (!wait_for_completion_io_timeout(&wait,
-				msecs_to_jiffies(PBLK_COMMAND_TIMEOUT_MS))) {
-		pr_err("pblk: smeta I/O timed out\n");
-	}
 	atomic_dec(&pblk->inflight_io);
 
 	if (rqd.error) {
-		if (dir == WRITE)
+		if (dir == PBLK_WRITE)
 			pblk_log_write_err(pblk, &rqd);
 		else
 			pblk_log_read_err(pblk, &rqd);
@@ -808,14 +816,14 @@ int pblk_line_read_smeta(struct pblk *pblk, struct pblk_line *line)
 {
 	u64 bpaddr = pblk_line_smeta_start(pblk, line);
 
-	return pblk_line_submit_smeta_io(pblk, line, bpaddr, READ);
+	return pblk_line_submit_smeta_io(pblk, line, bpaddr, PBLK_READ);
 }
 
 int pblk_line_read_emeta(struct pblk *pblk, struct pblk_line *line,
 			 void *emeta_buf)
 {
 	return pblk_line_submit_emeta_io(pblk, line, emeta_buf,
-						line->emeta_ssec, READ);
+						line->emeta_ssec, PBLK_READ);
 }
 
 static void pblk_setup_e_rq(struct pblk *pblk, struct nvm_rq *rqd,
@@ -824,7 +832,7 @@ static void pblk_setup_e_rq(struct pblk *pblk, struct nvm_rq *rqd,
 	rqd->opcode = NVM_OP_ERASE;
 	rqd->ppa_addr = ppa;
 	rqd->nr_ppas = 1;
-	rqd->flags = pblk_set_progr_mode(pblk, ERASE);
+	rqd->flags = pblk_set_progr_mode(pblk, PBLK_ERASE);
 	rqd->bio = NULL;
 }
 
@@ -832,19 +840,15 @@ static int pblk_blk_erase_sync(struct pblk *pblk, struct ppa_addr ppa)
 {
 	struct nvm_rq rqd;
 	int ret = 0;
-	DECLARE_COMPLETION_ONSTACK(wait);
 
 	memset(&rqd, 0, sizeof(struct nvm_rq));
 
 	pblk_setup_e_rq(pblk, &rqd, ppa);
 
-	rqd.end_io = pblk_end_io_sync;
-	rqd.private = &wait;
-
 	/* The write thread schedules erases so that it minimizes disturbances
 	 * with writes. Thus, there is no need to take the LUN semaphore.
 	 */
-	ret = pblk_submit_io(pblk, &rqd);
+	ret = pblk_submit_io_sync(pblk, &rqd);
 	if (ret) {
 		struct nvm_tgt_dev *dev = pblk->dev;
 		struct nvm_geo *geo = &dev->geo;
@@ -857,11 +861,6 @@ static int pblk_blk_erase_sync(struct pblk *pblk, struct ppa_addr ppa)
 		goto out;
 	}
 
-	if (!wait_for_completion_io_timeout(&wait,
-				msecs_to_jiffies(PBLK_COMMAND_TIMEOUT_MS))) {
-		pr_err("pblk: sync erase timed out\n");
-	}
-
 out:
 	rqd.private = pblk;
 	__pblk_end_io_erase(pblk, &rqd);
@@ -976,7 +975,7 @@ static int pblk_line_init_metadata(struct pblk *pblk, struct pblk_line *line,
 	memcpy(smeta_buf->header.uuid, pblk->instance_uuid, 16);
 	smeta_buf->header.id = cpu_to_le32(line->id);
 	smeta_buf->header.type = cpu_to_le16(line->type);
-	smeta_buf->header.version = cpu_to_le16(1);
+	smeta_buf->header.version = SMETA_VERSION;
 
 	/* Start metadata */
 	smeta_buf->seq_nr = cpu_to_le64(line->seq_nr);
@@ -1046,7 +1045,7 @@ static int pblk_line_init_bb(struct pblk *pblk, struct pblk_line *line,
 	line->smeta_ssec = off;
 	line->cur_sec = off + lm->smeta_sec;
 
-	if (init && pblk_line_submit_smeta_io(pblk, line, off, WRITE)) {
+	if (init && pblk_line_submit_smeta_io(pblk, line, off, PBLK_WRITE)) {
 		pr_debug("pblk: line smeta I/O failed. Retry\n");
 		return 1;
 	}
@@ -1056,7 +1055,6 @@ static int pblk_line_init_bb(struct pblk *pblk, struct pblk_line *line,
 	/* Mark emeta metadata sectors as bad sectors. We need to consider bad
 	 * blocks to make sure that there are enough sectors to store emeta
 	 */
-	bit = lm->sec_per_line;
 	off = lm->sec_per_line - lm->emeta_sec[0];
 	bitmap_set(line->invalid_bitmap, off, lm->emeta_sec[0]);
 	while (nr_bb) {
@@ -1093,25 +1091,21 @@ static int pblk_line_prepare(struct pblk *pblk, struct pblk_line *line)
 	struct pblk_line_meta *lm = &pblk->lm;
 	int blk_in_line = atomic_read(&line->blk_in_line);
 
-	line->map_bitmap = mempool_alloc(pblk->line_meta_pool, GFP_ATOMIC);
+	line->map_bitmap = kzalloc(lm->sec_bitmap_len, GFP_ATOMIC);
 	if (!line->map_bitmap)
 		return -ENOMEM;
-	memset(line->map_bitmap, 0, lm->sec_bitmap_len);
 
-	/* invalid_bitmap is special since it is used when line is closed. No
-	 * need to zeroized; it will be initialized using bb info form
-	 * map_bitmap
-	 */
-	line->invalid_bitmap = mempool_alloc(pblk->line_meta_pool, GFP_ATOMIC);
+	/* will be initialized using bb info from map_bitmap */
+	line->invalid_bitmap = kmalloc(lm->sec_bitmap_len, GFP_ATOMIC);
 	if (!line->invalid_bitmap) {
-		mempool_free(line->map_bitmap, pblk->line_meta_pool);
+		kfree(line->map_bitmap);
 		return -ENOMEM;
 	}
 
 	spin_lock(&line->lock);
 	if (line->state != PBLK_LINESTATE_FREE) {
-		mempool_free(line->invalid_bitmap, pblk->line_meta_pool);
-		mempool_free(line->map_bitmap, pblk->line_meta_pool);
+		kfree(line->map_bitmap);
+		kfree(line->invalid_bitmap);
 		spin_unlock(&line->lock);
 		WARN(1, "pblk: corrupted line %d, state %d\n",
 							line->id, line->state);
@@ -1163,7 +1157,7 @@ int pblk_line_recov_alloc(struct pblk *pblk, struct pblk_line *line)
 
 void pblk_line_recov_close(struct pblk *pblk, struct pblk_line *line)
 {
-	mempool_free(line->map_bitmap, pblk->line_meta_pool);
+	kfree(line->map_bitmap);
 	line->map_bitmap = NULL;
 	line->smeta = NULL;
 	line->emeta = NULL;
@@ -1328,6 +1322,41 @@ static void pblk_stop_writes(struct pblk *pblk, struct pblk_line *line)
 	pblk->state = PBLK_STATE_STOPPING;
 }
 
+static void pblk_line_close_meta_sync(struct pblk *pblk)
+{
+	struct pblk_line_mgmt *l_mg = &pblk->l_mg;
+	struct pblk_line_meta *lm = &pblk->lm;
+	struct pblk_line *line, *tline;
+	LIST_HEAD(list);
+
+	spin_lock(&l_mg->close_lock);
+	if (list_empty(&l_mg->emeta_list)) {
+		spin_unlock(&l_mg->close_lock);
+		return;
+	}
+
+	list_cut_position(&list, &l_mg->emeta_list, l_mg->emeta_list.prev);
+	spin_unlock(&l_mg->close_lock);
+
+	list_for_each_entry_safe(line, tline, &list, list) {
+		struct pblk_emeta *emeta = line->emeta;
+
+		while (emeta->mem < lm->emeta_len[0]) {
+			int ret;
+
+			ret = pblk_submit_meta_io(pblk, line);
+			if (ret) {
+				pr_err("pblk: sync meta line %d failed (%d)\n",
+							line->id, ret);
+				return;
+			}
+		}
+	}
+
+	pblk_wait_for_meta(pblk);
+	flush_workqueue(pblk->close_wq);
+}
+
 void pblk_pipeline_stop(struct pblk *pblk)
 {
 	struct pblk_line_mgmt *l_mg = &pblk->l_mg;
@@ -1361,17 +1390,17 @@ void pblk_pipeline_stop(struct pblk *pblk)
 	spin_unlock(&l_mg->free_lock);
 }
 
-void pblk_line_replace_data(struct pblk *pblk)
+struct pblk_line *pblk_line_replace_data(struct pblk *pblk)
 {
 	struct pblk_line_mgmt *l_mg = &pblk->l_mg;
-	struct pblk_line *cur, *new;
+	struct pblk_line *cur, *new = NULL;
 	unsigned int left_seblks;
 	int is_next = 0;
 
 	cur = l_mg->data_line;
 	new = l_mg->data_next;
 	if (!new)
-		return;
+		goto out;
 	l_mg->data_line = new;
 
 	spin_lock(&l_mg->free_lock);
@@ -1379,7 +1408,7 @@ void pblk_line_replace_data(struct pblk *pblk)
 		l_mg->data_line = NULL;
 		l_mg->data_next = NULL;
 		spin_unlock(&l_mg->free_lock);
-		return;
+		goto out;
 	}
 
 	pblk_line_setup_metadata(new, l_mg, &pblk->lm);
@@ -1391,7 +1420,7 @@ retry_erase:
 		/* If line is not fully erased, erase it */
 		if (atomic_read(&new->left_eblks)) {
 			if (pblk_line_erase(pblk, new))
-				return;
+				goto out;
 		} else {
 			io_schedule();
 		}
@@ -1402,7 +1431,7 @@ retry_setup:
 	if (!pblk_line_init_metadata(pblk, new, cur)) {
 		new = pblk_line_retry(pblk, new);
 		if (!new)
-			return;
+			goto out;
 
 		goto retry_setup;
 	}
@@ -1410,7 +1439,7 @@ retry_setup:
 	if (!pblk_line_init_bb(pblk, new, 1)) {
 		new = pblk_line_retry(pblk, new);
 		if (!new)
-			return;
+			goto out;
 
 		goto retry_setup;
 	}
@@ -1434,14 +1463,15 @@ retry_setup:
 
 	if (is_next)
 		pblk_rl_free_lines_dec(&pblk->rl, l_mg->data_next);
+
+out:
+	return new;
 }
 
 void pblk_line_free(struct pblk *pblk, struct pblk_line *line)
 {
-	if (line->map_bitmap)
-		mempool_free(line->map_bitmap, pblk->line_meta_pool);
-	if (line->invalid_bitmap)
-		mempool_free(line->invalid_bitmap, pblk->line_meta_pool);
+	kfree(line->map_bitmap);
+	kfree(line->invalid_bitmap);
 
 	*line->vsc = cpu_to_le32(EMPTY_ENTRY);
 
@@ -1451,11 +1481,10 @@ void pblk_line_free(struct pblk *pblk, struct pblk_line *line)
 	line->emeta = NULL;
 }
 
-void pblk_line_put(struct kref *ref)
+static void __pblk_line_put(struct pblk *pblk, struct pblk_line *line)
 {
-	struct pblk_line *line = container_of(ref, struct pblk_line, ref);
-	struct pblk *pblk = line->pblk;
 	struct pblk_line_mgmt *l_mg = &pblk->l_mg;
+	struct pblk_gc *gc = &pblk->gc;
 
 	spin_lock(&line->lock);
 	WARN_ON(line->state != PBLK_LINESTATE_GC);
@@ -1464,6 +1493,8 @@ void pblk_line_put(struct kref *ref)
 	pblk_line_free(pblk, line);
 	spin_unlock(&line->lock);
 
+	atomic_dec(&gc->pipeline_gc);
+
 	spin_lock(&l_mg->free_lock);
 	list_add_tail(&line->list, &l_mg->free_list);
 	l_mg->nr_free_lines++;
@@ -1472,13 +1503,49 @@ void pblk_line_put(struct kref *ref)
 	pblk_rl_free_lines_inc(&pblk->rl, line);
 }
 
+static void pblk_line_put_ws(struct work_struct *work)
+{
+	struct pblk_line_ws *line_put_ws = container_of(work,
+						struct pblk_line_ws, ws);
+	struct pblk *pblk = line_put_ws->pblk;
+	struct pblk_line *line = line_put_ws->line;
+
+	__pblk_line_put(pblk, line);
+	mempool_free(line_put_ws, pblk->gen_ws_pool);
+}
+
+void pblk_line_put(struct kref *ref)
+{
+	struct pblk_line *line = container_of(ref, struct pblk_line, ref);
+	struct pblk *pblk = line->pblk;
+
+	__pblk_line_put(pblk, line);
+}
+
+void pblk_line_put_wq(struct kref *ref)
+{
+	struct pblk_line *line = container_of(ref, struct pblk_line, ref);
+	struct pblk *pblk = line->pblk;
+	struct pblk_line_ws *line_put_ws;
+
+	line_put_ws = mempool_alloc(pblk->gen_ws_pool, GFP_ATOMIC);
+	if (!line_put_ws)
+		return;
+
+	line_put_ws->pblk = pblk;
+	line_put_ws->line = line;
+	line_put_ws->priv = NULL;
+
+	INIT_WORK(&line_put_ws->ws, pblk_line_put_ws);
+	queue_work(pblk->r_end_wq, &line_put_ws->ws);
+}
+
 int pblk_blk_erase_async(struct pblk *pblk, struct ppa_addr ppa)
 {
 	struct nvm_rq *rqd;
 	int err;
 
-	rqd = mempool_alloc(pblk->g_rq_pool, GFP_KERNEL);
-	memset(rqd, 0, pblk_g_rq_size);
+	rqd = pblk_alloc_rqd(pblk, PBLK_ERASE);
 
 	pblk_setup_e_rq(pblk, rqd, ppa);
 
@@ -1517,41 +1584,6 @@ int pblk_line_is_full(struct pblk_line *line)
 	return (line->left_msecs == 0);
 }
 
-void pblk_line_close_meta_sync(struct pblk *pblk)
-{
-	struct pblk_line_mgmt *l_mg = &pblk->l_mg;
-	struct pblk_line_meta *lm = &pblk->lm;
-	struct pblk_line *line, *tline;
-	LIST_HEAD(list);
-
-	spin_lock(&l_mg->close_lock);
-	if (list_empty(&l_mg->emeta_list)) {
-		spin_unlock(&l_mg->close_lock);
-		return;
-	}
-
-	list_cut_position(&list, &l_mg->emeta_list, l_mg->emeta_list.prev);
-	spin_unlock(&l_mg->close_lock);
-
-	list_for_each_entry_safe(line, tline, &list, list) {
-		struct pblk_emeta *emeta = line->emeta;
-
-		while (emeta->mem < lm->emeta_len[0]) {
-			int ret;
-
-			ret = pblk_submit_meta_io(pblk, line);
-			if (ret) {
-				pr_err("pblk: sync meta line %d failed (%d)\n",
-							line->id, ret);
-				return;
-			}
-		}
-	}
-
-	pblk_wait_for_meta(pblk);
-	flush_workqueue(pblk->close_wq);
-}
-
 static void pblk_line_should_sync_meta(struct pblk *pblk)
 {
 	if (pblk_rl_is_limit(&pblk->rl))
@@ -1582,15 +1614,13 @@ void pblk_line_close(struct pblk *pblk, struct pblk_line *line)
 
 	list_add_tail(&line->list, move_list);
 
-	mempool_free(line->map_bitmap, pblk->line_meta_pool);
+	kfree(line->map_bitmap);
 	line->map_bitmap = NULL;
 	line->smeta = NULL;
 	line->emeta = NULL;
 
 	spin_unlock(&line->lock);
 	spin_unlock(&l_mg->gc_lock);
-
-	pblk_gc_should_kick(pblk);
 }
 
 void pblk_line_close_meta(struct pblk *pblk, struct pblk_line *line)
@@ -1624,43 +1654,16 @@ void pblk_line_close_ws(struct work_struct *work)
 	struct pblk_line *line = line_ws->line;
 
 	pblk_line_close(pblk, line);
-	mempool_free(line_ws, pblk->line_ws_pool);
-}
-
-void pblk_line_mark_bb(struct work_struct *work)
-{
-	struct pblk_line_ws *line_ws = container_of(work, struct pblk_line_ws,
-									ws);
-	struct pblk *pblk = line_ws->pblk;
-	struct nvm_tgt_dev *dev = pblk->dev;
-	struct ppa_addr *ppa = line_ws->priv;
-	int ret;
-
-	ret = nvm_set_tgt_bb_tbl(dev, ppa, 1, NVM_BLK_T_GRWN_BAD);
-	if (ret) {
-		struct pblk_line *line;
-		int pos;
-
-		line = &pblk->lines[pblk_dev_ppa_to_line(*ppa)];
-		pos = pblk_dev_ppa_to_pos(&dev->geo, *ppa);
-
-		pr_err("pblk: failed to mark bb, line:%d, pos:%d\n",
-				line->id, pos);
-	}
-
-	kfree(ppa);
-	mempool_free(line_ws, pblk->line_ws_pool);
+	mempool_free(line_ws, pblk->gen_ws_pool);
 }
 
-void pblk_line_run_ws(struct pblk *pblk, struct pblk_line *line, void *priv,
-		      void (*work)(struct work_struct *),
+void pblk_gen_run_ws(struct pblk *pblk, struct pblk_line *line, void *priv,
+		      void (*work)(struct work_struct *), gfp_t gfp_mask,
 		      struct workqueue_struct *wq)
 {
 	struct pblk_line_ws *line_ws;
 
-	line_ws = mempool_alloc(pblk->line_ws_pool, GFP_ATOMIC);
-	if (!line_ws)
-		return;
+	line_ws = mempool_alloc(pblk->gen_ws_pool, gfp_mask);
 
 	line_ws->pblk = pblk;
 	line_ws->line = line;
@@ -1689,16 +1692,8 @@ static void __pblk_down_page(struct pblk *pblk, struct ppa_addr *ppa_list,
 #endif
 
 	ret = down_timeout(&rlun->wr_sem, msecs_to_jiffies(30000));
-	if (ret) {
-		switch (ret) {
-		case -ETIME:
-			pr_err("pblk: lun semaphore timed out\n");
-			break;
-		case -EINTR:
-			pr_err("pblk: lun semaphore timed out\n");
-			break;
-		}
-	}
+	if (ret == -ETIME || ret == -EINTR)
+		pr_err("pblk: taking lun semaphore timed out: err %d\n", -ret);
 }
 
 void pblk_down_page(struct pblk *pblk, struct ppa_addr *ppa_list, int nr_ppas)
@@ -1758,13 +1753,11 @@ void pblk_up_rq(struct pblk *pblk, struct ppa_addr *ppa_list, int nr_ppas,
 		rlun = &pblk->luns[bit];
 		up(&rlun->wr_sem);
 	}
-
-	kfree(lun_bitmap);
 }
 
 void pblk_update_map(struct pblk *pblk, sector_t lba, struct ppa_addr ppa)
 {
-	struct ppa_addr l2p_ppa;
+	struct ppa_addr ppa_l2p;
 
 	/* logic error: lba out-of-bounds. Ignore update */
 	if (!(lba < pblk->rl.nr_secs)) {
@@ -1773,10 +1766,10 @@ void pblk_update_map(struct pblk *pblk, sector_t lba, struct ppa_addr ppa)
 	}
 
 	spin_lock(&pblk->trans_lock);
-	l2p_ppa = pblk_trans_map_get(pblk, lba);
+	ppa_l2p = pblk_trans_map_get(pblk, lba);
 
-	if (!pblk_addr_in_cache(l2p_ppa) && !pblk_ppa_empty(l2p_ppa))
-		pblk_map_invalidate(pblk, l2p_ppa);
+	if (!pblk_addr_in_cache(ppa_l2p) && !pblk_ppa_empty(ppa_l2p))
+		pblk_map_invalidate(pblk, ppa_l2p);
 
 	pblk_trans_map_set(pblk, lba, ppa);
 	spin_unlock(&pblk->trans_lock);
@@ -1784,6 +1777,7 @@ void pblk_update_map(struct pblk *pblk, sector_t lba, struct ppa_addr ppa)
 
 void pblk_update_map_cache(struct pblk *pblk, sector_t lba, struct ppa_addr ppa)
 {
+
 #ifdef CONFIG_NVM_DEBUG
 	/* Callers must ensure that the ppa points to a cache address */
 	BUG_ON(!pblk_addr_in_cache(ppa));
@@ -1793,16 +1787,16 @@ void pblk_update_map_cache(struct pblk *pblk, sector_t lba, struct ppa_addr ppa)
 	pblk_update_map(pblk, lba, ppa);
 }
 
-int pblk_update_map_gc(struct pblk *pblk, sector_t lba, struct ppa_addr ppa,
-		       struct pblk_line *gc_line)
+int pblk_update_map_gc(struct pblk *pblk, sector_t lba, struct ppa_addr ppa_new,
+		       struct pblk_line *gc_line, u64 paddr_gc)
 {
-	struct ppa_addr l2p_ppa;
+	struct ppa_addr ppa_l2p, ppa_gc;
 	int ret = 1;
 
 #ifdef CONFIG_NVM_DEBUG
 	/* Callers must ensure that the ppa points to a cache address */
-	BUG_ON(!pblk_addr_in_cache(ppa));
-	BUG_ON(pblk_rb_pos_oob(&pblk->rwb, pblk_addr_to_cacheline(ppa)));
+	BUG_ON(!pblk_addr_in_cache(ppa_new));
+	BUG_ON(pblk_rb_pos_oob(&pblk->rwb, pblk_addr_to_cacheline(ppa_new)));
 #endif
 
 	/* logic error: lba out-of-bounds. Ignore update */
@@ -1812,36 +1806,41 @@ int pblk_update_map_gc(struct pblk *pblk, sector_t lba, struct ppa_addr ppa,
 	}
 
 	spin_lock(&pblk->trans_lock);
-	l2p_ppa = pblk_trans_map_get(pblk, lba);
+	ppa_l2p = pblk_trans_map_get(pblk, lba);
+	ppa_gc = addr_to_gen_ppa(pblk, paddr_gc, gc_line->id);
+
+	if (!pblk_ppa_comp(ppa_l2p, ppa_gc)) {
+		spin_lock(&gc_line->lock);
+		WARN(!test_bit(paddr_gc, gc_line->invalid_bitmap),
+						"pblk: corrupted GC update");
+		spin_unlock(&gc_line->lock);
 
-	/* Prevent updated entries to be overwritten by GC */
-	if (pblk_addr_in_cache(l2p_ppa) || pblk_ppa_empty(l2p_ppa) ||
-				pblk_tgt_ppa_to_line(l2p_ppa) != gc_line->id) {
 		ret = 0;
 		goto out;
 	}
 
-	pblk_trans_map_set(pblk, lba, ppa);
+	pblk_trans_map_set(pblk, lba, ppa_new);
 out:
 	spin_unlock(&pblk->trans_lock);
 	return ret;
 }
 
-void pblk_update_map_dev(struct pblk *pblk, sector_t lba, struct ppa_addr ppa,
-			 struct ppa_addr entry_line)
+void pblk_update_map_dev(struct pblk *pblk, sector_t lba,
+			 struct ppa_addr ppa_mapped, struct ppa_addr ppa_cache)
 {
-	struct ppa_addr l2p_line;
+	struct ppa_addr ppa_l2p;
 
 #ifdef CONFIG_NVM_DEBUG
 	/* Callers must ensure that the ppa points to a device address */
-	BUG_ON(pblk_addr_in_cache(ppa));
+	BUG_ON(pblk_addr_in_cache(ppa_mapped));
 #endif
 	/* Invalidate and discard padded entries */
 	if (lba == ADDR_EMPTY) {
 #ifdef CONFIG_NVM_DEBUG
 		atomic_long_inc(&pblk->padded_wb);
 #endif
-		pblk_map_invalidate(pblk, ppa);
+		if (!pblk_ppa_empty(ppa_mapped))
+			pblk_map_invalidate(pblk, ppa_mapped);
 		return;
 	}
 
@@ -1852,22 +1851,22 @@ void pblk_update_map_dev(struct pblk *pblk, sector_t lba, struct ppa_addr ppa,
 	}
 
 	spin_lock(&pblk->trans_lock);
-	l2p_line = pblk_trans_map_get(pblk, lba);
+	ppa_l2p = pblk_trans_map_get(pblk, lba);
 
 	/* Do not update L2P if the cacheline has been updated. In this case,
 	 * the mapped ppa must be invalidated
 	 */
-	if (l2p_line.ppa != entry_line.ppa) {
-		if (!pblk_ppa_empty(ppa))
-			pblk_map_invalidate(pblk, ppa);
+	if (!pblk_ppa_comp(ppa_l2p, ppa_cache)) {
+		if (!pblk_ppa_empty(ppa_mapped))
+			pblk_map_invalidate(pblk, ppa_mapped);
 		goto out;
 	}
 
 #ifdef CONFIG_NVM_DEBUG
-	WARN_ON(!pblk_addr_in_cache(l2p_line) && !pblk_ppa_empty(l2p_line));
+	WARN_ON(!pblk_addr_in_cache(ppa_l2p) && !pblk_ppa_empty(ppa_l2p));
 #endif
 
-	pblk_trans_map_set(pblk, lba, ppa);
+	pblk_trans_map_set(pblk, lba, ppa_mapped);
 out:
 	spin_unlock(&pblk->trans_lock);
 }
@@ -1878,23 +1877,32 @@ void pblk_lookup_l2p_seq(struct pblk *pblk, struct ppa_addr *ppas,
 	int i;
 
 	spin_lock(&pblk->trans_lock);
-	for (i = 0; i < nr_secs; i++)
-		ppas[i] = pblk_trans_map_get(pblk, blba + i);
+	for (i = 0; i < nr_secs; i++) {
+		struct ppa_addr ppa;
+
+		ppa = ppas[i] = pblk_trans_map_get(pblk, blba + i);
+
+		/* If the L2P entry maps to a line, the reference is valid */
+		if (!pblk_ppa_empty(ppa) && !pblk_addr_in_cache(ppa)) {
+			int line_id = pblk_dev_ppa_to_line(ppa);
+			struct pblk_line *line = &pblk->lines[line_id];
+
+			kref_get(&line->ref);
+		}
+	}
 	spin_unlock(&pblk->trans_lock);
 }
 
 void pblk_lookup_l2p_rand(struct pblk *pblk, struct ppa_addr *ppas,
 			  u64 *lba_list, int nr_secs)
 {
-	sector_t lba;
+	u64 lba;
 	int i;
 
 	spin_lock(&pblk->trans_lock);
 	for (i = 0; i < nr_secs; i++) {
 		lba = lba_list[i];
-		if (lba == ADDR_EMPTY) {
-			ppas[i].ppa = ADDR_EMPTY;
-		} else {
+		if (lba != ADDR_EMPTY) {
 			/* logic error: lba out-of-bounds. Ignore update */
 			if (!(lba < pblk->rl.nr_secs)) {
 				WARN(1, "pblk: corrupted L2P map request\n");

drivers/lightnvm/pblk-gc.c

@@ -20,7 +20,8 @@
 
 static void pblk_gc_free_gc_rq(struct pblk_gc_rq *gc_rq)
 {
-	vfree(gc_rq->data);
+	if (gc_rq->data)
+		vfree(gc_rq->data);
 	kfree(gc_rq);
 }
 
@@ -41,10 +42,7 @@ static int pblk_gc_write(struct pblk *pblk)
 	spin_unlock(&gc->w_lock);
 
 	list_for_each_entry_safe(gc_rq, tgc_rq, &w_list, list) {
-		pblk_write_gc_to_cache(pblk, gc_rq->data, gc_rq->lba_list,
-				gc_rq->nr_secs, gc_rq->secs_to_gc,
-				gc_rq->line, PBLK_IOTYPE_GC);
-
+		pblk_write_gc_to_cache(pblk, gc_rq);
 		list_del(&gc_rq->list);
 		kref_put(&gc_rq->line->ref, pblk_line_put);
 		pblk_gc_free_gc_rq(gc_rq);
@@ -58,42 +56,59 @@ static void pblk_gc_writer_kick(struct pblk_gc *gc)
 	wake_up_process(gc->gc_writer_ts);
 }
 
-/*
- * Responsible for managing all memory related to a gc request. Also in case of
- * failure
- */
-static int pblk_gc_move_valid_secs(struct pblk *pblk, struct pblk_gc_rq *gc_rq)
+static void pblk_put_line_back(struct pblk *pblk, struct pblk_line *line)
+{
+	struct pblk_line_mgmt *l_mg = &pblk->l_mg;
+	struct list_head *move_list;
+
+	spin_lock(&line->lock);
+	WARN_ON(line->state != PBLK_LINESTATE_GC);
+	line->state = PBLK_LINESTATE_CLOSED;
+	move_list = pblk_line_gc_list(pblk, line);
+	spin_unlock(&line->lock);
+
+	if (move_list) {
+		spin_lock(&l_mg->gc_lock);
+		list_add_tail(&line->list, move_list);
+		spin_unlock(&l_mg->gc_lock);
+	}
+}
+
+static void pblk_gc_line_ws(struct work_struct *work)
 {
+	struct pblk_line_ws *gc_rq_ws = container_of(work,
+						struct pblk_line_ws, ws);
+	struct pblk *pblk = gc_rq_ws->pblk;
 	struct nvm_tgt_dev *dev = pblk->dev;
 	struct nvm_geo *geo = &dev->geo;
 	struct pblk_gc *gc = &pblk->gc;
-	struct pblk_line *line = gc_rq->line;
-	void *data;
-	unsigned int secs_to_gc;
-	int ret = 0;
+	struct pblk_line *line = gc_rq_ws->line;
+	struct pblk_gc_rq *gc_rq = gc_rq_ws->priv;
+	int ret;
 
-	data = vmalloc(gc_rq->nr_secs * geo->sec_size);
-	if (!data) {
-		ret = -ENOMEM;
+	up(&gc->gc_sem);
+
+	gc_rq->data = vmalloc(gc_rq->nr_secs * geo->sec_size);
+	if (!gc_rq->data) {
+		pr_err("pblk: could not GC line:%d (%d/%d)\n",
+					line->id, *line->vsc, gc_rq->nr_secs);
 		goto out;
 	}
 
 	/* Read from GC victim block */
-	if (pblk_submit_read_gc(pblk, gc_rq->lba_list, data, gc_rq->nr_secs,
-							&secs_to_gc, line)) {
-		ret = -EFAULT;
-		goto free_data;
+	ret = pblk_submit_read_gc(pblk, gc_rq);
+	if (ret) {
+		pr_err("pblk: failed GC read in line:%d (err:%d)\n",
+								line->id, ret);
+		goto out;
 	}
 
-	if (!secs_to_gc)
-		goto free_rq;
-
-	gc_rq->data = data;
-	gc_rq->secs_to_gc = secs_to_gc;
+	if (!gc_rq->secs_to_gc)
+		goto out;
 
 retry:
 	spin_lock(&gc->w_lock);
-	if (gc->w_entries >= PBLK_GC_W_QD) {
+	if (gc->w_entries >= PBLK_GC_RQ_QD) {
 		spin_unlock(&gc->w_lock);
 		pblk_gc_writer_kick(&pblk->gc);
 		usleep_range(128, 256);
@@ -105,53 +120,13 @@ retry:
 
 	pblk_gc_writer_kick(&pblk->gc);
 
-	return 0;
+	kfree(gc_rq_ws);
+	return;
 
-free_rq:
-	kfree(gc_rq);
-free_data:
-	vfree(data);
 out:
+	pblk_gc_free_gc_rq(gc_rq);
 	kref_put(&line->ref, pblk_line_put);
-	return ret;
-}
-
-static void pblk_put_line_back(struct pblk *pblk, struct pblk_line *line)
-{
-	struct pblk_line_mgmt *l_mg = &pblk->l_mg;
-	struct list_head *move_list;
-
-	spin_lock(&line->lock);
-	WARN_ON(line->state != PBLK_LINESTATE_GC);
-	line->state = PBLK_LINESTATE_CLOSED;
-	move_list = pblk_line_gc_list(pblk, line);
-	spin_unlock(&line->lock);
-
-	if (move_list) {
-		spin_lock(&l_mg->gc_lock);
-		list_add_tail(&line->list, move_list);
-		spin_unlock(&l_mg->gc_lock);
-	}
-}
-
-static void pblk_gc_line_ws(struct work_struct *work)
-{
-	struct pblk_line_ws *line_rq_ws = container_of(work,
-						struct pblk_line_ws, ws);
-	struct pblk *pblk = line_rq_ws->pblk;
-	struct pblk_gc *gc = &pblk->gc;
-	struct pblk_line *line = line_rq_ws->line;
-	struct pblk_gc_rq *gc_rq = line_rq_ws->priv;
-
-	up(&gc->gc_sem);
-
-	if (pblk_gc_move_valid_secs(pblk, gc_rq)) {
-		pr_err("pblk: could not GC all sectors: line:%d (%d/%d)\n",
-						line->id, *line->vsc,
-						gc_rq->nr_secs);
-	}
-
-	mempool_free(line_rq_ws, pblk->line_ws_pool);
+	kfree(gc_rq_ws);
 }
 
 static void pblk_gc_line_prepare_ws(struct work_struct *work)
@@ -164,17 +139,24 @@ static void pblk_gc_line_prepare_ws(struct work_struct *work)
 	struct pblk_line_meta *lm = &pblk->lm;
 	struct pblk_gc *gc = &pblk->gc;
 	struct line_emeta *emeta_buf;
-	struct pblk_line_ws *line_rq_ws;
+	struct pblk_line_ws *gc_rq_ws;
 	struct pblk_gc_rq *gc_rq;
 	__le64 *lba_list;
+	unsigned long *invalid_bitmap;
 	int sec_left, nr_secs, bit;
 	int ret;
 
+	invalid_bitmap = kmalloc(lm->sec_bitmap_len, GFP_KERNEL);
+	if (!invalid_bitmap) {
+		pr_err("pblk: could not allocate GC invalid bitmap\n");
+		goto fail_free_ws;
+	}
+
 	emeta_buf = pblk_malloc(lm->emeta_len[0], l_mg->emeta_alloc_type,
 								GFP_KERNEL);
 	if (!emeta_buf) {
 		pr_err("pblk: cannot use GC emeta\n");
-		return;
+		goto fail_free_bitmap;
 	}
 
 	ret = pblk_line_read_emeta(pblk, line, emeta_buf);
@@ -193,7 +175,11 @@ static void pblk_gc_line_prepare_ws(struct work_struct *work)
 		goto fail_free_emeta;
 	}
 
+	spin_lock(&line->lock);
+	bitmap_copy(invalid_bitmap, line->invalid_bitmap, lm->sec_per_line);
 	sec_left = pblk_line_vsc(line);
+	spin_unlock(&line->lock);
+
 	if (sec_left < 0) {
 		pr_err("pblk: corrupted GC line (%d)\n", line->id);
 		goto fail_free_emeta;
@@ -207,11 +193,12 @@ next_rq:
 
 	nr_secs = 0;
 	do {
-		bit = find_next_zero_bit(line->invalid_bitmap, lm->sec_per_line,
+		bit = find_next_zero_bit(invalid_bitmap, lm->sec_per_line,
 								bit + 1);
 		if (bit > line->emeta_ssec)
 			break;
 
+		gc_rq->paddr_list[nr_secs] = bit;
 		gc_rq->lba_list[nr_secs++] = le64_to_cpu(lba_list[bit]);
 	} while (nr_secs < pblk->max_write_pgs);
 
@@ -223,19 +210,25 @@ next_rq:
 	gc_rq->nr_secs = nr_secs;
 	gc_rq->line = line;
 
-	line_rq_ws = mempool_alloc(pblk->line_ws_pool, GFP_KERNEL);
-	if (!line_rq_ws)
+	gc_rq_ws = kmalloc(sizeof(struct pblk_line_ws), GFP_KERNEL);
+	if (!gc_rq_ws)
 		goto fail_free_gc_rq;
 
-	line_rq_ws->pblk = pblk;
-	line_rq_ws->line = line;
-	line_rq_ws->priv = gc_rq;
+	gc_rq_ws->pblk = pblk;
+	gc_rq_ws->line = line;
+	gc_rq_ws->priv = gc_rq;
+
+	/* The write GC path can be much slower than the read GC one due to
+	 * the budget imposed by the rate-limiter. Balance in case that we get
+	 * back pressure from the write GC path.
+	 */
+	while (down_timeout(&gc->gc_sem, msecs_to_jiffies(30000)))
+		io_schedule();
 
-	down(&gc->gc_sem);
 	kref_get(&line->ref);
 
-	INIT_WORK(&line_rq_ws->ws, pblk_gc_line_ws);
-	queue_work(gc->gc_line_reader_wq, &line_rq_ws->ws);
+	INIT_WORK(&gc_rq_ws->ws, pblk_gc_line_ws);
+	queue_work(gc->gc_line_reader_wq, &gc_rq_ws->ws);
 
 	sec_left -= nr_secs;
 	if (sec_left > 0)
@@ -243,10 +236,11 @@ next_rq:
 
 out:
 	pblk_mfree(emeta_buf, l_mg->emeta_alloc_type);
-	mempool_free(line_ws, pblk->line_ws_pool);
+	kfree(line_ws);
+	kfree(invalid_bitmap);
 
 	kref_put(&line->ref, pblk_line_put);
-	atomic_dec(&gc->inflight_gc);
+	atomic_dec(&gc->read_inflight_gc);
 
 	return;
 
@@ -254,10 +248,14 @@ fail_free_gc_rq:
 	kfree(gc_rq);
 fail_free_emeta:
 	pblk_mfree(emeta_buf, l_mg->emeta_alloc_type);
+fail_free_bitmap:
+	kfree(invalid_bitmap);
+fail_free_ws:
+	kfree(line_ws);
+
 	pblk_put_line_back(pblk, line);
 	kref_put(&line->ref, pblk_line_put);
-	mempool_free(line_ws, pblk->line_ws_pool);
-	atomic_dec(&gc->inflight_gc);
+	atomic_dec(&gc->read_inflight_gc);
 
 	pr_err("pblk: Failed to GC line %d\n", line->id);
 }
@@ -269,19 +267,40 @@ static int pblk_gc_line(struct pblk *pblk, struct pblk_line *line)
 
 	pr_debug("pblk: line '%d' being reclaimed for GC\n", line->id);
 
-	line_ws = mempool_alloc(pblk->line_ws_pool, GFP_KERNEL);
+	line_ws = kmalloc(sizeof(struct pblk_line_ws), GFP_KERNEL);
 	if (!line_ws)
 		return -ENOMEM;
 
 	line_ws->pblk = pblk;
 	line_ws->line = line;
 
+	atomic_inc(&gc->pipeline_gc);
 	INIT_WORK(&line_ws->ws, pblk_gc_line_prepare_ws);
 	queue_work(gc->gc_reader_wq, &line_ws->ws);
 
 	return 0;
 }
 
+static void pblk_gc_reader_kick(struct pblk_gc *gc)
+{
+	wake_up_process(gc->gc_reader_ts);
+}
+
+static void pblk_gc_kick(struct pblk *pblk)
+{
+	struct pblk_gc *gc = &pblk->gc;
+
+	pblk_gc_writer_kick(gc);
+	pblk_gc_reader_kick(gc);
+
+	/* If we're shutting down GC, let's not start it up again */
+	if (gc->gc_enabled) {
+		wake_up_process(gc->gc_ts);
+		mod_timer(&gc->gc_timer,
+			  jiffies + msecs_to_jiffies(GC_TIME_MSECS));
+	}
+}
+
 static int pblk_gc_read(struct pblk *pblk)
 {
 	struct pblk_gc *gc = &pblk->gc;
@@ -305,11 +324,6 @@ static int pblk_gc_read(struct pblk *pblk)
 	return 0;
 }
 
-static void pblk_gc_reader_kick(struct pblk_gc *gc)
-{
-	wake_up_process(gc->gc_reader_ts);
-}
-
 static struct pblk_line *pblk_gc_get_victim_line(struct pblk *pblk,
 						 struct list_head *group_list)
 {
@@ -338,26 +352,17 @@ static bool pblk_gc_should_run(struct pblk_gc *gc, struct pblk_rl *rl)
 	return ((gc->gc_active) && (nr_blocks_need > nr_blocks_free));
 }
 
-/*
- * Lines with no valid sectors will be returned to the free list immediately. If
- * GC is activated - either because the free block count is under the determined
- * threshold, or because it is being forced from user space - only lines with a
- * high count of invalid sectors will be recycled.
- */
-static void pblk_gc_run(struct pblk *pblk)
+void pblk_gc_free_full_lines(struct pblk *pblk)
 {
 	struct pblk_line_mgmt *l_mg = &pblk->l_mg;
 	struct pblk_gc *gc = &pblk->gc;
 	struct pblk_line *line;
-	struct list_head *group_list;
-	bool run_gc;
-	int inflight_gc, gc_group = 0, prev_group = 0;
 
 	do {
 		spin_lock(&l_mg->gc_lock);
 		if (list_empty(&l_mg->gc_full_list)) {
 			spin_unlock(&l_mg->gc_lock);
-			break;
+			return;
 		}
 
 		line = list_first_entry(&l_mg->gc_full_list,
@@ -371,11 +376,30 @@ static void pblk_gc_run(struct pblk *pblk)
 		list_del(&line->list);
 		spin_unlock(&l_mg->gc_lock);
 
+		atomic_inc(&gc->pipeline_gc);
 		kref_put(&line->ref, pblk_line_put);
 	} while (1);
+}
+
+/*
+ * Lines with no valid sectors will be returned to the free list immediately. If
+ * GC is activated - either because the free block count is under the determined
+ * threshold, or because it is being forced from user space - only lines with a
+ * high count of invalid sectors will be recycled.
+ */
+static void pblk_gc_run(struct pblk *pblk)
+{
+	struct pblk_line_mgmt *l_mg = &pblk->l_mg;
+	struct pblk_gc *gc = &pblk->gc;
+	struct pblk_line *line;
+	struct list_head *group_list;
+	bool run_gc;
+	int read_inflight_gc, gc_group = 0, prev_group = 0;
+
+	pblk_gc_free_full_lines(pblk);
 
 	run_gc = pblk_gc_should_run(&pblk->gc, &pblk->rl);
-	if (!run_gc || (atomic_read(&gc->inflight_gc) >= PBLK_GC_L_QD))
+	if (!run_gc || (atomic_read(&gc->read_inflight_gc) >= PBLK_GC_L_QD))
 		return;
 
 next_gc_group:
@@ -402,14 +426,14 @@ next_gc_group:
 		list_add_tail(&line->list, &gc->r_list);
 		spin_unlock(&gc->r_lock);
 
-		inflight_gc = atomic_inc_return(&gc->inflight_gc);
+		read_inflight_gc = atomic_inc_return(&gc->read_inflight_gc);
 		pblk_gc_reader_kick(gc);
 
 		prev_group = 1;
 
 		/* No need to queue up more GC lines than we can handle */
 		run_gc = pblk_gc_should_run(&pblk->gc, &pblk->rl);
-		if (!run_gc || inflight_gc >= PBLK_GC_L_QD)
+		if (!run_gc || read_inflight_gc >= PBLK_GC_L_QD)
 			break;
 	} while (1);
 
@@ -418,16 +442,6 @@ next_gc_group:
 		goto next_gc_group;
 }
 
-void pblk_gc_kick(struct pblk *pblk)
-{
-	struct pblk_gc *gc = &pblk->gc;
-
-	wake_up_process(gc->gc_ts);
-	pblk_gc_writer_kick(gc);
-	pblk_gc_reader_kick(gc);
-	mod_timer(&gc->gc_timer, jiffies + msecs_to_jiffies(GC_TIME_MSECS));
-}
-
 static void pblk_gc_timer(unsigned long data)
 {
 	struct pblk *pblk = (struct pblk *)data;
@@ -465,6 +479,7 @@ static int pblk_gc_writer_ts(void *data)
 static int pblk_gc_reader_ts(void *data)
 {
 	struct pblk *pblk = data;
+	struct pblk_gc *gc = &pblk->gc;
 
 	while (!kthread_should_stop()) {
 		if (!pblk_gc_read(pblk))
@@ -473,6 +488,18 @@ static int pblk_gc_reader_ts(void *data)
 		io_schedule();
 	}
 
+#ifdef CONFIG_NVM_DEBUG
+	pr_info("pblk: flushing gc pipeline, %d lines left\n",
+		atomic_read(&gc->pipeline_gc));
+#endif
+
+	do {
+		if (!atomic_read(&gc->pipeline_gc))
+			break;
+
+		schedule();
+	} while (1);
+
 	return 0;
 }
 
@@ -486,10 +513,10 @@ void pblk_gc_should_start(struct pblk *pblk)
 {
 	struct pblk_gc *gc = &pblk->gc;
 
-	if (gc->gc_enabled && !gc->gc_active)
+	if (gc->gc_enabled && !gc->gc_active) {
 		pblk_gc_start(pblk);
-
-	pblk_gc_kick(pblk);
+		pblk_gc_kick(pblk);
+	}
 }
 
 /*
@@ -510,6 +537,11 @@ void pblk_gc_should_stop(struct pblk *pblk)
 		pblk_gc_stop(pblk, 0);
 }
 
+void pblk_gc_should_kick(struct pblk *pblk)
+{
+	pblk_rl_update_rates(&pblk->rl);
+}
+
 void pblk_gc_sysfs_state_show(struct pblk *pblk, int *gc_enabled,
 			      int *gc_active)
 {
@@ -576,7 +608,8 @@ int pblk_gc_init(struct pblk *pblk)
 	gc->gc_forced = 0;
 	gc->gc_enabled = 1;
 	gc->w_entries = 0;
-	atomic_set(&gc->inflight_gc, 0);
+	atomic_set(&gc->read_inflight_gc, 0);
+	atomic_set(&gc->pipeline_gc, 0);
 
 	/* Workqueue that reads valid sectors from a line and submit them to the
 	 * GC writer to be recycled.
@@ -602,7 +635,7 @@ int pblk_gc_init(struct pblk *pblk)
 	spin_lock_init(&gc->w_lock);
 	spin_lock_init(&gc->r_lock);
 
-	sema_init(&gc->gc_sem, 128);
+	sema_init(&gc->gc_sem, PBLK_GC_RQ_QD);
 
 	INIT_LIST_HEAD(&gc->w_list);
 	INIT_LIST_HEAD(&gc->r_list);
@@ -625,24 +658,24 @@ void pblk_gc_exit(struct pblk *pblk)
 {
 	struct pblk_gc *gc = &pblk->gc;
 
-	flush_workqueue(gc->gc_reader_wq);
-	flush_workqueue(gc->gc_line_reader_wq);
-
-	del_timer(&gc->gc_timer);
+	gc->gc_enabled = 0;
+	del_timer_sync(&gc->gc_timer);
 	pblk_gc_stop(pblk, 1);
 
 	if (gc->gc_ts)
 		kthread_stop(gc->gc_ts);
 
+	if (gc->gc_reader_ts)
+		kthread_stop(gc->gc_reader_ts);
+
+	flush_workqueue(gc->gc_reader_wq);
 	if (gc->gc_reader_wq)
 		destroy_workqueue(gc->gc_reader_wq);
 
+	flush_workqueue(gc->gc_line_reader_wq);
 	if (gc->gc_line_reader_wq)
 		destroy_workqueue(gc->gc_line_reader_wq);
 
 	if (gc->gc_writer_ts)
 		kthread_stop(gc->gc_writer_ts);
-
-	if (gc->gc_reader_ts)
-		kthread_stop(gc->gc_reader_ts);
 }

drivers/lightnvm/pblk-init.c

@@ -20,8 +20,8 @@
 
 #include "pblk.h"
 
-static struct kmem_cache *pblk_blk_ws_cache, *pblk_rec_cache, *pblk_g_rq_cache,
-				*pblk_w_rq_cache, *pblk_line_meta_cache;
+static struct kmem_cache *pblk_ws_cache, *pblk_rec_cache, *pblk_g_rq_cache,
+				*pblk_w_rq_cache;
 static DECLARE_RWSEM(pblk_lock);
 struct bio_set *pblk_bio_set;
 
@@ -46,7 +46,7 @@ static int pblk_rw_io(struct request_queue *q, struct pblk *pblk,
 	 * user I/Os. Unless stalled, the rate limiter leaves at least 256KB
 	 * available for user I/O.
 	 */
-	if (unlikely(pblk_get_secs(bio) >= pblk_rl_sysfs_rate_show(&pblk->rl)))
+	if (pblk_get_secs(bio) > pblk_rl_max_io(&pblk->rl))
 		blk_queue_split(q, &bio);
 
 	return pblk_write_to_cache(pblk, bio, PBLK_IOTYPE_USER);
@@ -76,6 +76,28 @@ static blk_qc_t pblk_make_rq(struct request_queue *q, struct bio *bio)
 	return BLK_QC_T_NONE;
 }
 
+static size_t pblk_trans_map_size(struct pblk *pblk)
+{
+	int entry_size = 8;
+
+	if (pblk->ppaf_bitsize < 32)
+		entry_size = 4;
+
+	return entry_size * pblk->rl.nr_secs;
+}
+
+#ifdef CONFIG_NVM_DEBUG
+static u32 pblk_l2p_crc(struct pblk *pblk)
+{
+	size_t map_size;
+	u32 crc = ~(u32)0;
+
+	map_size = pblk_trans_map_size(pblk);
+	crc = crc32_le(crc, pblk->trans_map, map_size);
+	return crc;
+}
+#endif
+
 static void pblk_l2p_free(struct pblk *pblk)
 {
 	vfree(pblk->trans_map);
@@ -85,12 +107,10 @@ static int pblk_l2p_init(struct pblk *pblk)
 {
 	sector_t i;
 	struct ppa_addr ppa;
-	int entry_size = 8;
+	size_t map_size;
 
-	if (pblk->ppaf_bitsize < 32)
-		entry_size = 4;
-
-	pblk->trans_map = vmalloc(entry_size * pblk->rl.nr_secs);
+	map_size = pblk_trans_map_size(pblk);
+	pblk->trans_map = vmalloc(map_size);
 	if (!pblk->trans_map)
 		return -ENOMEM;
 
@@ -132,7 +152,6 @@ static int pblk_rwb_init(struct pblk *pblk)
 }
 
 /* Minimum pages needed within a lun */
-#define PAGE_POOL_SIZE 16
 #define ADDR_POOL_SIZE 64
 
 static int pblk_set_ppaf(struct pblk *pblk)
@@ -182,12 +201,10 @@ static int pblk_set_ppaf(struct pblk *pblk)
 
 static int pblk_init_global_caches(struct pblk *pblk)
 {
-	char cache_name[PBLK_CACHE_NAME_LEN];
-
 	down_write(&pblk_lock);
-	pblk_blk_ws_cache = kmem_cache_create("pblk_blk_ws",
+	pblk_ws_cache = kmem_cache_create("pblk_blk_ws",
 				sizeof(struct pblk_line_ws), 0, 0, NULL);
-	if (!pblk_blk_ws_cache) {
+	if (!pblk_ws_cache) {
 		up_write(&pblk_lock);
 		return -ENOMEM;
 	}
@@ -195,7 +212,7 @@ static int pblk_init_global_caches(struct pblk *pblk)
 	pblk_rec_cache = kmem_cache_create("pblk_rec",
 				sizeof(struct pblk_rec_ctx), 0, 0, NULL);
 	if (!pblk_rec_cache) {
-		kmem_cache_destroy(pblk_blk_ws_cache);
+		kmem_cache_destroy(pblk_ws_cache);
 		up_write(&pblk_lock);
 		return -ENOMEM;
 	}
@@ -203,7 +220,7 @@ static int pblk_init_global_caches(struct pblk *pblk)
 	pblk_g_rq_cache = kmem_cache_create("pblk_g_rq", pblk_g_rq_size,
 				0, 0, NULL);
 	if (!pblk_g_rq_cache) {
-		kmem_cache_destroy(pblk_blk_ws_cache);
+		kmem_cache_destroy(pblk_ws_cache);
 		kmem_cache_destroy(pblk_rec_cache);
 		up_write(&pblk_lock);
 		return -ENOMEM;
@@ -212,30 +229,25 @@ static int pblk_init_global_caches(struct pblk *pblk)
 	pblk_w_rq_cache = kmem_cache_create("pblk_w_rq", pblk_w_rq_size,
 				0, 0, NULL);
 	if (!pblk_w_rq_cache) {
-		kmem_cache_destroy(pblk_blk_ws_cache);
+		kmem_cache_destroy(pblk_ws_cache);
 		kmem_cache_destroy(pblk_rec_cache);
 		kmem_cache_destroy(pblk_g_rq_cache);
 		up_write(&pblk_lock);
 		return -ENOMEM;
 	}
-
-	snprintf(cache_name, sizeof(cache_name), "pblk_line_m_%s",
-							pblk->disk->disk_name);
-	pblk_line_meta_cache = kmem_cache_create(cache_name,
-				pblk->lm.sec_bitmap_len, 0, 0, NULL);
-	if (!pblk_line_meta_cache) {
-		kmem_cache_destroy(pblk_blk_ws_cache);
-		kmem_cache_destroy(pblk_rec_cache);
-		kmem_cache_destroy(pblk_g_rq_cache);
-		kmem_cache_destroy(pblk_w_rq_cache);
-		up_write(&pblk_lock);
-		return -ENOMEM;
-	}
 	up_write(&pblk_lock);
 
 	return 0;
 }
 
+static void pblk_free_global_caches(struct pblk *pblk)
+{
+	kmem_cache_destroy(pblk_ws_cache);
+	kmem_cache_destroy(pblk_rec_cache);
+	kmem_cache_destroy(pblk_g_rq_cache);
+	kmem_cache_destroy(pblk_w_rq_cache);
+}
+
 static int pblk_core_init(struct pblk *pblk)
 {
 	struct nvm_tgt_dev *dev = pblk->dev;
@@ -247,70 +259,80 @@ static int pblk_core_init(struct pblk *pblk)
 	if (pblk_init_global_caches(pblk))
 		return -ENOMEM;
 
-	pblk->page_pool = mempool_create_page_pool(PAGE_POOL_SIZE, 0);
-	if (!pblk->page_pool)
-		return -ENOMEM;
+	/* Internal bios can be at most the sectors signaled by the device. */
+	pblk->page_bio_pool = mempool_create_page_pool(nvm_max_phys_sects(dev),
+									0);
+	if (!pblk->page_bio_pool)
+		goto free_global_caches;
 
-	pblk->line_ws_pool = mempool_create_slab_pool(PBLK_WS_POOL_SIZE,
-							pblk_blk_ws_cache);
-	if (!pblk->line_ws_pool)
-		goto free_page_pool;
+	pblk->gen_ws_pool = mempool_create_slab_pool(PBLK_GEN_WS_POOL_SIZE,
+							pblk_ws_cache);
+	if (!pblk->gen_ws_pool)
+		goto free_page_bio_pool;
 
 	pblk->rec_pool = mempool_create_slab_pool(geo->nr_luns, pblk_rec_cache);
 	if (!pblk->rec_pool)
-		goto free_blk_ws_pool;
+		goto free_gen_ws_pool;
 
-	pblk->g_rq_pool = mempool_create_slab_pool(PBLK_READ_REQ_POOL_SIZE,
+	pblk->r_rq_pool = mempool_create_slab_pool(geo->nr_luns,
 							pblk_g_rq_cache);
-	if (!pblk->g_rq_pool)
+	if (!pblk->r_rq_pool)
 		goto free_rec_pool;
 
-	pblk->w_rq_pool = mempool_create_slab_pool(geo->nr_luns * 2,
+	pblk->e_rq_pool = mempool_create_slab_pool(geo->nr_luns,
+							pblk_g_rq_cache);
+	if (!pblk->e_rq_pool)
+		goto free_r_rq_pool;
+
+	pblk->w_rq_pool = mempool_create_slab_pool(geo->nr_luns,
 							pblk_w_rq_cache);
 	if (!pblk->w_rq_pool)
-		goto free_g_rq_pool;
-
-	pblk->line_meta_pool =
-			mempool_create_slab_pool(PBLK_META_POOL_SIZE,
-							pblk_line_meta_cache);
-	if (!pblk->line_meta_pool)
-		goto free_w_rq_pool;
+		goto free_e_rq_pool;
 
 	pblk->close_wq = alloc_workqueue("pblk-close-wq",
 			WQ_MEM_RECLAIM | WQ_UNBOUND, PBLK_NR_CLOSE_JOBS);
 	if (!pblk->close_wq)
-		goto free_line_meta_pool;
+		goto free_w_rq_pool;
 
 	pblk->bb_wq = alloc_workqueue("pblk-bb-wq",
 			WQ_MEM_RECLAIM | WQ_UNBOUND, 0);
 	if (!pblk->bb_wq)
 		goto free_close_wq;
 
-	if (pblk_set_ppaf(pblk))
+	pblk->r_end_wq = alloc_workqueue("pblk-read-end-wq",
+			WQ_MEM_RECLAIM | WQ_UNBOUND, 0);
+	if (!pblk->r_end_wq)
 		goto free_bb_wq;
 
+	if (pblk_set_ppaf(pblk))
+		goto free_r_end_wq;
+
 	if (pblk_rwb_init(pblk))
-		goto free_bb_wq;
+		goto free_r_end_wq;
 
 	INIT_LIST_HEAD(&pblk->compl_list);
 	return 0;
 
+free_r_end_wq:
+	destroy_workqueue(pblk->r_end_wq);
 free_bb_wq:
 	destroy_workqueue(pblk->bb_wq);
 free_close_wq:
 	destroy_workqueue(pblk->close_wq);
-free_line_meta_pool:
-	mempool_destroy(pblk->line_meta_pool);
 free_w_rq_pool:
 	mempool_destroy(pblk->w_rq_pool);
-free_g_rq_pool:
-	mempool_destroy(pblk->g_rq_pool);
+free_e_rq_pool:
+	mempool_destroy(pblk->e_rq_pool);
+free_r_rq_pool:
+	mempool_destroy(pblk->r_rq_pool);
 free_rec_pool:
 	mempool_destroy(pblk->rec_pool);
-free_blk_ws_pool:
-	mempool_destroy(pblk->line_ws_pool);
-free_page_pool:
-	mempool_destroy(pblk->page_pool);
+free_gen_ws_pool:
+	mempool_destroy(pblk->gen_ws_pool);
+free_page_bio_pool:
+	mempool_destroy(pblk->page_bio_pool);
+free_global_caches:
+	pblk_free_global_caches(pblk);
 	return -ENOMEM;
 }
 
@@ -319,21 +341,20 @@ static void pblk_core_free(struct pblk *pblk)
 	if (pblk->close_wq)
 		destroy_workqueue(pblk->close_wq);
 
+	if (pblk->r_end_wq)
+		destroy_workqueue(pblk->r_end_wq);
+
 	if (pblk->bb_wq)
 		destroy_workqueue(pblk->bb_wq);
 
-	mempool_destroy(pblk->page_pool);
-	mempool_destroy(pblk->line_ws_pool);
+	mempool_destroy(pblk->page_bio_pool);
+	mempool_destroy(pblk->gen_ws_pool);
 	mempool_destroy(pblk->rec_pool);
-	mempool_destroy(pblk->g_rq_pool);
+	mempool_destroy(pblk->r_rq_pool);
+	mempool_destroy(pblk->e_rq_pool);
 	mempool_destroy(pblk->w_rq_pool);
-	mempool_destroy(pblk->line_meta_pool);
 
-	kmem_cache_destroy(pblk_blk_ws_cache);
-	kmem_cache_destroy(pblk_rec_cache);
-	kmem_cache_destroy(pblk_g_rq_cache);
-	kmem_cache_destroy(pblk_w_rq_cache);
-	kmem_cache_destroy(pblk_line_meta_cache);
+	pblk_free_global_caches(pblk);
 }
 
 static void pblk_luns_free(struct pblk *pblk)
@@ -372,13 +393,11 @@ static void pblk_line_meta_free(struct pblk *pblk)
 	kfree(l_mg->bb_aux);
 	kfree(l_mg->vsc_list);
 
-	spin_lock(&l_mg->free_lock);
 	for (i = 0; i < PBLK_DATA_LINES; i++) {
 		kfree(l_mg->sline_meta[i]);
 		pblk_mfree(l_mg->eline_meta[i]->buf, l_mg->emeta_alloc_type);
 		kfree(l_mg->eline_meta[i]);
 	}
-	spin_unlock(&l_mg->free_lock);
 
 	kfree(pblk->lines);
 }
@@ -507,6 +526,13 @@ static int pblk_lines_configure(struct pblk *pblk, int flags)
 		}
 	}
 
+#ifdef CONFIG_NVM_DEBUG
+	pr_info("pblk init: L2P CRC: %x\n", pblk_l2p_crc(pblk));
+#endif
+
+	/* Free full lines directly as GC has not been started yet */
+	pblk_gc_free_full_lines(pblk);
+
 	if (!line) {
 		/* Configure next line for user data */
 		line = pblk_line_get_first_data(pblk);
@@ -630,7 +656,10 @@ static int pblk_lines_alloc_metadata(struct pblk *pblk)
 
 fail_free_emeta:
 	while (--i >= 0) {
-		vfree(l_mg->eline_meta[i]->buf);
+		if (l_mg->emeta_alloc_type == PBLK_VMALLOC_META)
+			vfree(l_mg->eline_meta[i]->buf);
+		else
+			kfree(l_mg->eline_meta[i]->buf);
 		kfree(l_mg->eline_meta[i]);
 	}
 
@@ -681,8 +710,8 @@ static int pblk_lines_init(struct pblk *pblk)
 	lm->blk_bitmap_len = BITS_TO_LONGS(geo->nr_luns) * sizeof(long);
 	lm->sec_bitmap_len = BITS_TO_LONGS(lm->sec_per_line) * sizeof(long);
 	lm->lun_bitmap_len = BITS_TO_LONGS(geo->nr_luns) * sizeof(long);
-	lm->high_thrs = lm->sec_per_line / 2;
-	lm->mid_thrs = lm->sec_per_line / 4;
+	lm->mid_thrs = lm->sec_per_line / 2;
+	lm->high_thrs = lm->sec_per_line / 4;
 	lm->meta_distance = (geo->nr_luns / 2) * pblk->min_write_pgs;
 
 	/* Calculate necessary pages for smeta. See comment over struct
@@ -713,9 +742,13 @@ add_emeta_page:
 		goto add_emeta_page;
 	}
 
-	lm->emeta_bb = geo->nr_luns - i;
-	lm->min_blk_line = 1 + DIV_ROUND_UP(lm->smeta_sec + lm->emeta_sec[0],
-							geo->sec_per_blk);
+	lm->emeta_bb = geo->nr_luns > i ? geo->nr_luns - i : 0;
+
+	lm->min_blk_line = 1;
+	if (geo->nr_luns > 1)
+		lm->min_blk_line += DIV_ROUND_UP(lm->smeta_sec +
+					lm->emeta_sec[0], geo->sec_per_blk);
+
 	if (lm->min_blk_line > lm->blk_per_line) {
 		pr_err("pblk: config. not supported. Min. LUN in line:%d\n",
 							lm->blk_per_line);
@@ -890,6 +923,11 @@ static void pblk_exit(void *private)
 	down_write(&pblk_lock);
 	pblk_gc_exit(pblk);
 	pblk_tear_down(pblk);
+
+#ifdef CONFIG_NVM_DEBUG
+	pr_info("pblk exit: L2P CRC: %x\n", pblk_l2p_crc(pblk));
+#endif
+
 	pblk_free(pblk);
 	up_write(&pblk_lock);
 }
@@ -911,7 +949,7 @@ static void *pblk_init(struct nvm_tgt_dev *dev, struct gendisk *tdisk,
 	int ret;
 
 	if (dev->identity.dom & NVM_RSP_L2P) {
-		pr_err("pblk: device-side L2P table not supported. (%x)\n",
+		pr_err("pblk: host-side L2P table not supported. (%x)\n",
 							dev->identity.dom);
 		return ERR_PTR(-EINVAL);
 	}
@@ -923,6 +961,7 @@ static void *pblk_init(struct nvm_tgt_dev *dev, struct gendisk *tdisk,
 	pblk->dev = dev;
 	pblk->disk = tdisk;
 	pblk->state = PBLK_STATE_RUNNING;
+	pblk->gc.gc_enabled = 0;
 
 	spin_lock_init(&pblk->trans_lock);
 	spin_lock_init(&pblk->lock);
@@ -944,6 +983,7 @@ static void *pblk_init(struct nvm_tgt_dev *dev, struct gendisk *tdisk,
 	atomic_long_set(&pblk->recov_writes, 0);
 	atomic_long_set(&pblk->recov_writes, 0);
 	atomic_long_set(&pblk->recov_gc_writes, 0);
+	atomic_long_set(&pblk->recov_gc_reads, 0);
 #endif
 
 	atomic_long_set(&pblk->read_failed, 0);
@@ -1012,6 +1052,10 @@ static void *pblk_init(struct nvm_tgt_dev *dev, struct gendisk *tdisk,
 			pblk->rwb.nr_entries);
 
 	wake_up_process(pblk->writer_ts);
+
+	/* Check if we need to start GC */
+	pblk_gc_should_kick(pblk);
+
 	return pblk;
 
 fail_stop_writer:
@@ -1044,6 +1088,7 @@ static struct nvm_tgt_type tt_pblk = {
 
 	.sysfs_init	= pblk_sysfs_init,
 	.sysfs_exit	= pblk_sysfs_exit,
+	.owner		= THIS_MODULE,
 };
 
 static int __init pblk_module_init(void)

drivers/lightnvm/pblk-map.c

@@ -25,16 +25,28 @@ static void pblk_map_page_data(struct pblk *pblk, unsigned int sentry,
 			       unsigned int valid_secs)
 {
 	struct pblk_line *line = pblk_line_get_data(pblk);
-	struct pblk_emeta *emeta = line->emeta;
+	struct pblk_emeta *emeta;
 	struct pblk_w_ctx *w_ctx;
-	__le64 *lba_list = emeta_to_lbas(pblk, emeta->buf);
+	__le64 *lba_list;
 	u64 paddr;
 	int nr_secs = pblk->min_write_pgs;
 	int i;
 
+	if (pblk_line_is_full(line)) {
+		struct pblk_line *prev_line = line;
+
+		line = pblk_line_replace_data(pblk);
+		pblk_line_close_meta(pblk, prev_line);
+	}
+
+	emeta = line->emeta;
+	lba_list = emeta_to_lbas(pblk, emeta->buf);
+
 	paddr = pblk_alloc_page(pblk, line, nr_secs);
 
 	for (i = 0; i < nr_secs; i++, paddr++) {
+		__le64 addr_empty = cpu_to_le64(ADDR_EMPTY);
+
 		/* ppa to be sent to the device */
 		ppa_list[i] = addr_to_gen_ppa(pblk, paddr, line->id);
 
@@ -51,22 +63,14 @@ static void pblk_map_page_data(struct pblk *pblk, unsigned int sentry,
 			w_ctx->ppa = ppa_list[i];
 			meta_list[i].lba = cpu_to_le64(w_ctx->lba);
 			lba_list[paddr] = cpu_to_le64(w_ctx->lba);
-			line->nr_valid_lbas++;
+			if (lba_list[paddr] != addr_empty)
+				line->nr_valid_lbas++;
 		} else {
-			__le64 addr_empty = cpu_to_le64(ADDR_EMPTY);
-
 			lba_list[paddr] = meta_list[i].lba = addr_empty;
 			__pblk_map_invalidate(pblk, line, paddr);
 		}
 	}
 
-	if (pblk_line_is_full(line)) {
-		struct pblk_line *prev_line = line;
-
-		pblk_line_replace_data(pblk);
-		pblk_line_close_meta(pblk, prev_line);
-	}
-
 	pblk_down_rq(pblk, ppa_list, nr_secs, lun_bitmap);
 }
 

drivers/lightnvm/pblk-rb.c

@@ -201,8 +201,7 @@ unsigned int pblk_rb_read_commit(struct pblk_rb *rb, unsigned int nr_entries)
 	return subm;
 }
 
-static int __pblk_rb_update_l2p(struct pblk_rb *rb, unsigned int *l2p_upd,
-				unsigned int to_update)
+static int __pblk_rb_update_l2p(struct pblk_rb *rb, unsigned int to_update)
 {
 	struct pblk *pblk = container_of(rb, struct pblk, rwb);
 	struct pblk_line *line;
@@ -213,7 +212,7 @@ static int __pblk_rb_update_l2p(struct pblk_rb *rb, unsigned int *l2p_upd,
 	int flags;
 
 	for (i = 0; i < to_update; i++) {
-		entry = &rb->entries[*l2p_upd];
+		entry = &rb->entries[rb->l2p_update];
 		w_ctx = &entry->w_ctx;
 
 		flags = READ_ONCE(entry->w_ctx.flags);
@@ -230,7 +229,7 @@ static int __pblk_rb_update_l2p(struct pblk_rb *rb, unsigned int *l2p_upd,
 		line = &pblk->lines[pblk_tgt_ppa_to_line(w_ctx->ppa)];
 		kref_put(&line->ref, pblk_line_put);
 		clean_wctx(w_ctx);
-		*l2p_upd = (*l2p_upd + 1) & (rb->nr_entries - 1);
+		rb->l2p_update = (rb->l2p_update + 1) & (rb->nr_entries - 1);
 	}
 
 	pblk_rl_out(&pblk->rl, user_io, gc_io);
@@ -258,7 +257,7 @@ static int pblk_rb_update_l2p(struct pblk_rb *rb, unsigned int nr_entries,
 
 	count = nr_entries - space;
 	/* l2p_update used exclusively under rb->w_lock */
-	ret = __pblk_rb_update_l2p(rb, &rb->l2p_update, count);
+	ret = __pblk_rb_update_l2p(rb, count);
 
 out:
 	return ret;
@@ -280,7 +279,7 @@ void pblk_rb_sync_l2p(struct pblk_rb *rb)
 	sync = smp_load_acquire(&rb->sync);
 
 	to_update = pblk_rb_ring_count(sync, rb->l2p_update, rb->nr_entries);
-	__pblk_rb_update_l2p(rb, &rb->l2p_update, to_update);
+	__pblk_rb_update_l2p(rb, to_update);
 
 	spin_unlock(&rb->w_lock);
 }
@@ -325,8 +324,8 @@ void pblk_rb_write_entry_user(struct pblk_rb *rb, void *data,
 }
 
 void pblk_rb_write_entry_gc(struct pblk_rb *rb, void *data,
-			    struct pblk_w_ctx w_ctx, struct pblk_line *gc_line,
-			    unsigned int ring_pos)
+			    struct pblk_w_ctx w_ctx, struct pblk_line *line,
+			    u64 paddr, unsigned int ring_pos)
 {
 	struct pblk *pblk = container_of(rb, struct pblk, rwb);
 	struct pblk_rb_entry *entry;
@@ -341,7 +340,7 @@ void pblk_rb_write_entry_gc(struct pblk_rb *rb, void *data,
 
 	__pblk_rb_write_entry(rb, data, w_ctx, entry);
 
-	if (!pblk_update_map_gc(pblk, w_ctx.lba, entry->cacheline, gc_line))
+	if (!pblk_update_map_gc(pblk, w_ctx.lba, entry->cacheline, line, paddr))
 		entry->w_ctx.lba = ADDR_EMPTY;
 
 	flags = w_ctx.flags | PBLK_WRITTEN_DATA;
@@ -355,7 +354,6 @@ static int pblk_rb_sync_point_set(struct pblk_rb *rb, struct bio *bio,
 {
 	struct pblk_rb_entry *entry;
 	unsigned int subm, sync_point;
-	int flags;
 
 	subm = READ_ONCE(rb->subm);
 
@@ -369,12 +367,6 @@ static int pblk_rb_sync_point_set(struct pblk_rb *rb, struct bio *bio,
 	sync_point = (pos == 0) ? (rb->nr_entries - 1) : (pos - 1);
 	entry = &rb->entries[sync_point];
 
-	flags = READ_ONCE(entry->w_ctx.flags);
-	flags |= PBLK_FLUSH_ENTRY;
-
-	/* Release flags on context. Protect from writes */
-	smp_store_release(&entry->w_ctx.flags, flags);
-
 	/* Protect syncs */
 	smp_store_release(&rb->sync_point, sync_point);
 
@@ -454,6 +446,7 @@ static int pblk_rb_may_write_flush(struct pblk_rb *rb, unsigned int nr_entries,
 
 	/* Protect from read count */
 	smp_store_release(&rb->mem, mem);
+
 	return 1;
 }
 
@@ -558,12 +551,13 @@ out:
  * persist data on the write buffer to the media.
  */
 unsigned int pblk_rb_read_to_bio(struct pblk_rb *rb, struct nvm_rq *rqd,
-				 struct bio *bio, unsigned int pos,
-				 unsigned int nr_entries, unsigned int count)
+				 unsigned int pos, unsigned int nr_entries,
+				 unsigned int count)
 {
 	struct pblk *pblk = container_of(rb, struct pblk, rwb);
 	struct request_queue *q = pblk->dev->q;
 	struct pblk_c_ctx *c_ctx = nvm_rq_to_pdu(rqd);
+	struct bio *bio = rqd->bio;
 	struct pblk_rb_entry *entry;
 	struct page *page;
 	unsigned int pad = 0, to_read = nr_entries;

drivers/lightnvm/pblk-read.c

@@ -39,21 +39,15 @@ static int pblk_read_from_cache(struct pblk *pblk, struct bio *bio,
 }
 
 static void pblk_read_ppalist_rq(struct pblk *pblk, struct nvm_rq *rqd,
-				 unsigned long *read_bitmap)
+				 sector_t blba, unsigned long *read_bitmap)
 {
+	struct pblk_sec_meta *meta_list = rqd->meta_list;
 	struct bio *bio = rqd->bio;
 	struct ppa_addr ppas[PBLK_MAX_REQ_ADDRS];
-	sector_t blba = pblk_get_lba(bio);
 	int nr_secs = rqd->nr_ppas;
 	bool advanced_bio = false;
 	int i, j = 0;
 
-	/* logic error: lba out-of-bounds. Ignore read request */
-	if (blba + nr_secs >= pblk->rl.nr_secs) {
-		WARN(1, "pblk: read lbas out of bounds\n");
-		return;
-	}
-
 	pblk_lookup_l2p_seq(pblk, ppas, blba, nr_secs);
 
 	for (i = 0; i < nr_secs; i++) {
@@ -63,6 +57,7 @@ static void pblk_read_ppalist_rq(struct pblk *pblk, struct nvm_rq *rqd,
 retry:
 		if (pblk_ppa_empty(p)) {
 			WARN_ON(test_and_set_bit(i, read_bitmap));
+			meta_list[i].lba = cpu_to_le64(ADDR_EMPTY);
 
 			if (unlikely(!advanced_bio)) {
 				bio_advance(bio, (i) * PBLK_EXPOSED_PAGE_SIZE);
@@ -82,6 +77,7 @@ retry:
 				goto retry;
 			}
 			WARN_ON(test_and_set_bit(i, read_bitmap));
+			meta_list[i].lba = cpu_to_le64(lba);
 			advanced_bio = true;
 #ifdef CONFIG_NVM_DEBUG
 			atomic_long_inc(&pblk->cache_reads);
@@ -117,10 +113,51 @@ static int pblk_submit_read_io(struct pblk *pblk, struct nvm_rq *rqd)
 	return NVM_IO_OK;
 }
 
-static void pblk_end_io_read(struct nvm_rq *rqd)
+static void pblk_read_check(struct pblk *pblk, struct nvm_rq *rqd,
+			   sector_t blba)
+{
+	struct pblk_sec_meta *meta_list = rqd->meta_list;
+	int nr_lbas = rqd->nr_ppas;
+	int i;
+
+	for (i = 0; i < nr_lbas; i++) {
+		u64 lba = le64_to_cpu(meta_list[i].lba);
+
+		if (lba == ADDR_EMPTY)
+			continue;
+
+		WARN(lba != blba + i, "pblk: corrupted read LBA\n");
+	}
+}
+
+static void pblk_read_put_rqd_kref(struct pblk *pblk, struct nvm_rq *rqd)
+{
+	struct ppa_addr *ppa_list;
+	int i;
+
+	ppa_list = (rqd->nr_ppas > 1) ? rqd->ppa_list : &rqd->ppa_addr;
+
+	for (i = 0; i < rqd->nr_ppas; i++) {
+		struct ppa_addr ppa = ppa_list[i];
+		struct pblk_line *line;
+
+		line = &pblk->lines[pblk_dev_ppa_to_line(ppa)];
+		kref_put(&line->ref, pblk_line_put_wq);
+	}
+}
+
+static void pblk_end_user_read(struct bio *bio)
+{
+#ifdef CONFIG_NVM_DEBUG
+	WARN_ONCE(bio->bi_status, "pblk: corrupted read bio\n");
+#endif
+	bio_endio(bio);
+	bio_put(bio);
+}
+
+static void __pblk_end_io_read(struct pblk *pblk, struct nvm_rq *rqd,
+			       bool put_line)
 {
-	struct pblk *pblk = rqd->private;
-	struct nvm_tgt_dev *dev = pblk->dev;
 	struct pblk_g_ctx *r_ctx = nvm_rq_to_pdu(rqd);
 	struct bio *bio = rqd->bio;
 
@@ -131,47 +168,51 @@ static void pblk_end_io_read(struct nvm_rq *rqd)
 		WARN_ONCE(bio->bi_status, "pblk: corrupted read error\n");
 #endif
 
-	nvm_dev_dma_free(dev->parent, rqd->meta_list, rqd->dma_meta_list);
+	pblk_read_check(pblk, rqd, r_ctx->lba);
 
 	bio_put(bio);
-	if (r_ctx->private) {
-		struct bio *orig_bio = r_ctx->private;
+	if (r_ctx->private)
+		pblk_end_user_read((struct bio *)r_ctx->private);
 
-#ifdef CONFIG_NVM_DEBUG
-		WARN_ONCE(orig_bio->bi_status, "pblk: corrupted read bio\n");
-#endif
-		bio_endio(orig_bio);
-		bio_put(orig_bio);
-	}
+	if (put_line)
+		pblk_read_put_rqd_kref(pblk, rqd);
 
 #ifdef CONFIG_NVM_DEBUG
 	atomic_long_add(rqd->nr_ppas, &pblk->sync_reads);
 	atomic_long_sub(rqd->nr_ppas, &pblk->inflight_reads);
 #endif
 
-	pblk_free_rqd(pblk, rqd, READ);
+	pblk_free_rqd(pblk, rqd, PBLK_READ);
 	atomic_dec(&pblk->inflight_io);
 }
 
+static void pblk_end_io_read(struct nvm_rq *rqd)
+{
+	struct pblk *pblk = rqd->private;
+
+	__pblk_end_io_read(pblk, rqd, true);
+}
+
 static int pblk_fill_partial_read_bio(struct pblk *pblk, struct nvm_rq *rqd,
 				      unsigned int bio_init_idx,
 				      unsigned long *read_bitmap)
 {
 	struct bio *new_bio, *bio = rqd->bio;
+	struct pblk_sec_meta *meta_list = rqd->meta_list;
 	struct bio_vec src_bv, dst_bv;
 	void *ppa_ptr = NULL;
 	void *src_p, *dst_p;
 	dma_addr_t dma_ppa_list = 0;
+	__le64 *lba_list_mem, *lba_list_media;
 	int nr_secs = rqd->nr_ppas;
 	int nr_holes = nr_secs - bitmap_weight(read_bitmap, nr_secs);
 	int i, ret, hole;
-	DECLARE_COMPLETION_ONSTACK(wait);
+
+	/* Re-use allocated memory for intermediate lbas */
+	lba_list_mem = (((void *)rqd->ppa_list) + pblk_dma_ppa_size);
+	lba_list_media = (((void *)rqd->ppa_list) + 2 * pblk_dma_ppa_size);
 
 	new_bio = bio_alloc(GFP_KERNEL, nr_holes);
-	if (!new_bio) {
-		pr_err("pblk: could not alloc read bio\n");
-		return NVM_IO_ERR;
-	}
 
 	if (pblk_bio_add_pages(pblk, new_bio, GFP_KERNEL, nr_holes))
 		goto err;
@@ -181,34 +222,29 @@ static int pblk_fill_partial_read_bio(struct pblk *pblk, struct nvm_rq *rqd,
 		goto err;
 	}
 
+	for (i = 0; i < nr_secs; i++)
+		lba_list_mem[i] = meta_list[i].lba;
+
 	new_bio->bi_iter.bi_sector = 0; /* internal bio */
 	bio_set_op_attrs(new_bio, REQ_OP_READ, 0);
-	new_bio->bi_private = &wait;
-	new_bio->bi_end_io = pblk_end_bio_sync;
 
 	rqd->bio = new_bio;
 	rqd->nr_ppas = nr_holes;
 	rqd->flags = pblk_set_read_mode(pblk, PBLK_READ_RANDOM);
-	rqd->end_io = NULL;
 
-	if (unlikely(nr_secs > 1 && nr_holes == 1)) {
+	if (unlikely(nr_holes == 1)) {
 		ppa_ptr = rqd->ppa_list;
 		dma_ppa_list = rqd->dma_ppa_list;
 		rqd->ppa_addr = rqd->ppa_list[0];
 	}
 
-	ret = pblk_submit_read_io(pblk, rqd);
+	ret = pblk_submit_io_sync(pblk, rqd);
 	if (ret) {
 		bio_put(rqd->bio);
-		pr_err("pblk: read IO submission failed\n");
+		pr_err("pblk: sync read IO submission failed\n");
 		goto err;
 	}
 
-	if (!wait_for_completion_io_timeout(&wait,
-				msecs_to_jiffies(PBLK_COMMAND_TIMEOUT_MS))) {
-		pr_err("pblk: partial read I/O timed out\n");
-	}
-
 	if (rqd->error) {
 		atomic_long_inc(&pblk->read_failed);
 #ifdef CONFIG_NVM_DEBUG
@@ -216,15 +252,31 @@ static int pblk_fill_partial_read_bio(struct pblk *pblk, struct nvm_rq *rqd,
 #endif
 	}
 
-	if (unlikely(nr_secs > 1 && nr_holes == 1)) {
+	if (unlikely(nr_holes == 1)) {
+		struct ppa_addr ppa;
+
+		ppa = rqd->ppa_addr;
 		rqd->ppa_list = ppa_ptr;
 		rqd->dma_ppa_list = dma_ppa_list;
+		rqd->ppa_list[0] = ppa;
+	}
+
+	for (i = 0; i < nr_secs; i++) {
+		lba_list_media[i] = meta_list[i].lba;
+		meta_list[i].lba = lba_list_mem[i];
 	}
 
 	/* Fill the holes in the original bio */
 	i = 0;
 	hole = find_first_zero_bit(read_bitmap, nr_secs);
 	do {
+		int line_id = pblk_dev_ppa_to_line(rqd->ppa_list[i]);
+		struct pblk_line *line = &pblk->lines[line_id];
+
+		kref_put(&line->ref, pblk_line_put);
+
+		meta_list[hole].lba = lba_list_media[i];
+
 		src_bv = new_bio->bi_io_vec[i++];
 		dst_bv = bio->bi_io_vec[bio_init_idx + hole];
 
@@ -238,7 +290,7 @@ static int pblk_fill_partial_read_bio(struct pblk *pblk, struct nvm_rq *rqd,
 		kunmap_atomic(src_p);
 		kunmap_atomic(dst_p);
 
-		mempool_free(src_bv.bv_page, pblk->page_pool);
+		mempool_free(src_bv.bv_page, pblk->page_bio_pool);
 
 		hole = find_next_zero_bit(read_bitmap, nr_secs, hole + 1);
 	} while (hole < nr_secs);
@@ -246,34 +298,26 @@ static int pblk_fill_partial_read_bio(struct pblk *pblk, struct nvm_rq *rqd,
 	bio_put(new_bio);
 
 	/* Complete the original bio and associated request */
+	bio_endio(bio);
 	rqd->bio = bio;
 	rqd->nr_ppas = nr_secs;
-	rqd->private = pblk;
 
-	bio_endio(bio);
-	pblk_end_io_read(rqd);
+	__pblk_end_io_read(pblk, rqd, false);
 	return NVM_IO_OK;
 
 err:
 	/* Free allocated pages in new bio */
 	pblk_bio_free_pages(pblk, bio, 0, new_bio->bi_vcnt);
-	rqd->private = pblk;
-	pblk_end_io_read(rqd);
+	__pblk_end_io_read(pblk, rqd, false);
 	return NVM_IO_ERR;
 }
 
 static void pblk_read_rq(struct pblk *pblk, struct nvm_rq *rqd,
-			 unsigned long *read_bitmap)
+			 sector_t lba, unsigned long *read_bitmap)
 {
+	struct pblk_sec_meta *meta_list = rqd->meta_list;
 	struct bio *bio = rqd->bio;
 	struct ppa_addr ppa;
-	sector_t lba = pblk_get_lba(bio);
-
-	/* logic error: lba out-of-bounds. Ignore read request */
-	if (lba >= pblk->rl.nr_secs) {
-		WARN(1, "pblk: read lba out of bounds\n");
-		return;
-	}
 
 	pblk_lookup_l2p_seq(pblk, &ppa, lba, 1);
 
@@ -284,6 +328,7 @@ static void pblk_read_rq(struct pblk *pblk, struct nvm_rq *rqd,
 retry:
 	if (pblk_ppa_empty(ppa)) {
 		WARN_ON(test_and_set_bit(0, read_bitmap));
+		meta_list[0].lba = cpu_to_le64(ADDR_EMPTY);
 		return;
 	}
 
@@ -295,9 +340,12 @@ retry:
 			pblk_lookup_l2p_seq(pblk, &ppa, lba, 1);
 			goto retry;
 		}
+
 		WARN_ON(test_and_set_bit(0, read_bitmap));
+		meta_list[0].lba = cpu_to_le64(lba);
+
 #ifdef CONFIG_NVM_DEBUG
-			atomic_long_inc(&pblk->cache_reads);
+		atomic_long_inc(&pblk->cache_reads);
 #endif
 	} else {
 		rqd->ppa_addr = ppa;
@@ -309,22 +357,24 @@ retry:
 int pblk_submit_read(struct pblk *pblk, struct bio *bio)
 {
 	struct nvm_tgt_dev *dev = pblk->dev;
+	sector_t blba = pblk_get_lba(bio);
 	unsigned int nr_secs = pblk_get_secs(bio);
+	struct pblk_g_ctx *r_ctx;
 	struct nvm_rq *rqd;
-	unsigned long read_bitmap; /* Max 64 ppas per request */
 	unsigned int bio_init_idx;
+	unsigned long read_bitmap; /* Max 64 ppas per request */
 	int ret = NVM_IO_ERR;
 
-	if (nr_secs > PBLK_MAX_REQ_ADDRS)
+	/* logic error: lba out-of-bounds. Ignore read request */
+	if (blba >= pblk->rl.nr_secs || nr_secs > PBLK_MAX_REQ_ADDRS) {
+		WARN(1, "pblk: read lba out of bounds (lba:%llu, nr:%d)\n",
+					(unsigned long long)blba, nr_secs);
 		return NVM_IO_ERR;
+	}
 
 	bitmap_zero(&read_bitmap, nr_secs);
 
-	rqd = pblk_alloc_rqd(pblk, READ);
-	if (IS_ERR(rqd)) {
-		pr_err_ratelimited("pblk: not able to alloc rqd");
-		return NVM_IO_ERR;
-	}
+	rqd = pblk_alloc_rqd(pblk, PBLK_READ);
 
 	rqd->opcode = NVM_OP_PREAD;
 	rqd->bio = bio;
@@ -332,6 +382,9 @@ int pblk_submit_read(struct pblk *pblk, struct bio *bio)
 	rqd->private = pblk;
 	rqd->end_io = pblk_end_io_read;
 
+	r_ctx = nvm_rq_to_pdu(rqd);
+	r_ctx->lba = blba;
+
 	/* Save the index for this bio's start. This is needed in case
 	 * we need to fill a partial read.
 	 */
@@ -348,23 +401,22 @@ int pblk_submit_read(struct pblk *pblk, struct bio *bio)
 		rqd->ppa_list = rqd->meta_list + pblk_dma_meta_size;
 		rqd->dma_ppa_list = rqd->dma_meta_list + pblk_dma_meta_size;
 
-		pblk_read_ppalist_rq(pblk, rqd, &read_bitmap);
+		pblk_read_ppalist_rq(pblk, rqd, blba, &read_bitmap);
 	} else {
-		pblk_read_rq(pblk, rqd, &read_bitmap);
+		pblk_read_rq(pblk, rqd, blba, &read_bitmap);
 	}
 
 	bio_get(bio);
 	if (bitmap_full(&read_bitmap, nr_secs)) {
 		bio_endio(bio);
 		atomic_inc(&pblk->inflight_io);
-		pblk_end_io_read(rqd);
+		__pblk_end_io_read(pblk, rqd, false);
 		return NVM_IO_OK;
 	}
 
 	/* All sectors are to be read from the device */
 	if (bitmap_empty(&read_bitmap, rqd->nr_ppas)) {
 		struct bio *int_bio = NULL;
-		struct pblk_g_ctx *r_ctx = nvm_rq_to_pdu(rqd);
 
 		/* Clone read bio to deal with read errors internally */
 		int_bio = bio_clone_fast(bio, GFP_KERNEL, pblk_bio_set);
@@ -399,40 +451,46 @@ int pblk_submit_read(struct pblk *pblk, struct bio *bio)
 	return NVM_IO_OK;
 
 fail_rqd_free:
-	pblk_free_rqd(pblk, rqd, READ);
+	pblk_free_rqd(pblk, rqd, PBLK_READ);
 	return ret;
 }
 
 static int read_ppalist_rq_gc(struct pblk *pblk, struct nvm_rq *rqd,
 			      struct pblk_line *line, u64 *lba_list,
-			      unsigned int nr_secs)
+			      u64 *paddr_list_gc, unsigned int nr_secs)
 {
-	struct ppa_addr ppas[PBLK_MAX_REQ_ADDRS];
+	struct ppa_addr ppa_list_l2p[PBLK_MAX_REQ_ADDRS];
+	struct ppa_addr ppa_gc;
 	int valid_secs = 0;
 	int i;
 
-	pblk_lookup_l2p_rand(pblk, ppas, lba_list, nr_secs);
+	pblk_lookup_l2p_rand(pblk, ppa_list_l2p, lba_list, nr_secs);
 
 	for (i = 0; i < nr_secs; i++) {
-		if (pblk_addr_in_cache(ppas[i]) || ppas[i].g.blk != line->id ||
-						pblk_ppa_empty(ppas[i])) {
-			lba_list[i] = ADDR_EMPTY;
+		if (lba_list[i] == ADDR_EMPTY)
+			continue;
+
+		ppa_gc = addr_to_gen_ppa(pblk, paddr_list_gc[i], line->id);
+		if (!pblk_ppa_comp(ppa_list_l2p[i], ppa_gc)) {
+			paddr_list_gc[i] = lba_list[i] = ADDR_EMPTY;
 			continue;
 		}
 
-		rqd->ppa_list[valid_secs++] = ppas[i];
+		rqd->ppa_list[valid_secs++] = ppa_list_l2p[i];
 	}
 
 #ifdef CONFIG_NVM_DEBUG
 	atomic_long_add(valid_secs, &pblk->inflight_reads);
 #endif
+
 	return valid_secs;
 }
 
 static int read_rq_gc(struct pblk *pblk, struct nvm_rq *rqd,
-		      struct pblk_line *line, sector_t lba)
+		      struct pblk_line *line, sector_t lba,
+		      u64 paddr_gc)
 {
-	struct ppa_addr ppa;
+	struct ppa_addr ppa_l2p, ppa_gc;
 	int valid_secs = 0;
 
 	if (lba == ADDR_EMPTY)
@@ -445,15 +503,14 @@ static int read_rq_gc(struct pblk *pblk, struct nvm_rq *rqd,
 	}
 
 	spin_lock(&pblk->trans_lock);
-	ppa = pblk_trans_map_get(pblk, lba);
+	ppa_l2p = pblk_trans_map_get(pblk, lba);
 	spin_unlock(&pblk->trans_lock);
 
-	/* Ignore updated values until the moment */
-	if (pblk_addr_in_cache(ppa) || ppa.g.blk != line->id ||
-							pblk_ppa_empty(ppa))
+	ppa_gc = addr_to_gen_ppa(pblk, paddr_gc, line->id);
+	if (!pblk_ppa_comp(ppa_l2p, ppa_gc))
 		goto out;
 
-	rqd->ppa_addr = ppa;
+	rqd->ppa_addr = ppa_l2p;
 	valid_secs = 1;
 
 #ifdef CONFIG_NVM_DEBUG
@@ -464,42 +521,44 @@ out:
 	return valid_secs;
 }
 
-int pblk_submit_read_gc(struct pblk *pblk, u64 *lba_list, void *data,
-			unsigned int nr_secs, unsigned int *secs_to_gc,
-			struct pblk_line *line)
+int pblk_submit_read_gc(struct pblk *pblk, struct pblk_gc_rq *gc_rq)
 {
 	struct nvm_tgt_dev *dev = pblk->dev;
 	struct nvm_geo *geo = &dev->geo;
 	struct bio *bio;
 	struct nvm_rq rqd;
-	int ret, data_len;
-	DECLARE_COMPLETION_ONSTACK(wait);
+	int data_len;
+	int ret = NVM_IO_OK;
 
 	memset(&rqd, 0, sizeof(struct nvm_rq));
 
 	rqd.meta_list = nvm_dev_dma_alloc(dev->parent, GFP_KERNEL,
 							&rqd.dma_meta_list);
 	if (!rqd.meta_list)
-		return NVM_IO_ERR;
+		return -ENOMEM;
 
-	if (nr_secs > 1) {
+	if (gc_rq->nr_secs > 1) {
 		rqd.ppa_list = rqd.meta_list + pblk_dma_meta_size;
 		rqd.dma_ppa_list = rqd.dma_meta_list + pblk_dma_meta_size;
 
-		*secs_to_gc = read_ppalist_rq_gc(pblk, &rqd, line, lba_list,
-								nr_secs);
-		if (*secs_to_gc == 1)
+		gc_rq->secs_to_gc = read_ppalist_rq_gc(pblk, &rqd, gc_rq->line,
+							gc_rq->lba_list,
+							gc_rq->paddr_list,
+							gc_rq->nr_secs);
+		if (gc_rq->secs_to_gc == 1)
 			rqd.ppa_addr = rqd.ppa_list[0];
 	} else {
-		*secs_to_gc = read_rq_gc(pblk, &rqd, line, lba_list[0]);
+		gc_rq->secs_to_gc = read_rq_gc(pblk, &rqd, gc_rq->line,
+							gc_rq->lba_list[0],
+							gc_rq->paddr_list[0]);
 	}
 
-	if (!(*secs_to_gc))
+	if (!(gc_rq->secs_to_gc))
 		goto out;
 
-	data_len = (*secs_to_gc) * geo->sec_size;
-	bio = pblk_bio_map_addr(pblk, data, *secs_to_gc, data_len,
-						PBLK_KMALLOC_META, GFP_KERNEL);
+	data_len = (gc_rq->secs_to_gc) * geo->sec_size;
+	bio = pblk_bio_map_addr(pblk, gc_rq->data, gc_rq->secs_to_gc, data_len,
+						PBLK_VMALLOC_META, GFP_KERNEL);
 	if (IS_ERR(bio)) {
 		pr_err("pblk: could not allocate GC bio (%lu)\n", PTR_ERR(bio));
 		goto err_free_dma;
@@ -509,23 +568,16 @@ int pblk_submit_read_gc(struct pblk *pblk, u64 *lba_list, void *data,
 	bio_set_op_attrs(bio, REQ_OP_READ, 0);
 
 	rqd.opcode = NVM_OP_PREAD;
-	rqd.end_io = pblk_end_io_sync;
-	rqd.private = &wait;
-	rqd.nr_ppas = *secs_to_gc;
+	rqd.nr_ppas = gc_rq->secs_to_gc;
 	rqd.flags = pblk_set_read_mode(pblk, PBLK_READ_RANDOM);
 	rqd.bio = bio;
 
-	ret = pblk_submit_read_io(pblk, &rqd);
-	if (ret) {
-		bio_endio(bio);
+	if (pblk_submit_io_sync(pblk, &rqd)) {
+		ret = -EIO;
 		pr_err("pblk: GC read request failed\n");
-		goto err_free_dma;
+		goto err_free_bio;
 	}
 
-	if (!wait_for_completion_io_timeout(&wait,
-				msecs_to_jiffies(PBLK_COMMAND_TIMEOUT_MS))) {
-		pr_err("pblk: GC read I/O timed out\n");
-	}
 	atomic_dec(&pblk->inflight_io);
 
 	if (rqd.error) {
@@ -536,16 +588,18 @@ int pblk_submit_read_gc(struct pblk *pblk, u64 *lba_list, void *data,
 	}
 
 #ifdef CONFIG_NVM_DEBUG
-	atomic_long_add(*secs_to_gc, &pblk->sync_reads);
-	atomic_long_add(*secs_to_gc, &pblk->recov_gc_reads);
-	atomic_long_sub(*secs_to_gc, &pblk->inflight_reads);
+	atomic_long_add(gc_rq->secs_to_gc, &pblk->sync_reads);
+	atomic_long_add(gc_rq->secs_to_gc, &pblk->recov_gc_reads);
+	atomic_long_sub(gc_rq->secs_to_gc, &pblk->inflight_reads);
 #endif
 
 out:
 	nvm_dev_dma_free(dev->parent, rqd.meta_list, rqd.dma_meta_list);
-	return NVM_IO_OK;
+	return ret;
 
+err_free_bio:
+	bio_put(bio);
 err_free_dma:
 	nvm_dev_dma_free(dev->parent, rqd.meta_list, rqd.dma_meta_list);
-	return NVM_IO_ERR;
+	return ret;
 }

drivers/lightnvm/pblk-recovery.c

@@ -34,10 +34,6 @@ void pblk_submit_rec(struct work_struct *work)
 								max_secs);
 
 	bio = bio_alloc(GFP_KERNEL, nr_rec_secs);
-	if (!bio) {
-		pr_err("pblk: not able to create recovery bio\n");
-		return;
-	}
 
 	bio->bi_iter.bi_sector = 0;
 	bio_set_op_attrs(bio, REQ_OP_WRITE, 0);
@@ -71,7 +67,7 @@ void pblk_submit_rec(struct work_struct *work)
 
 err:
 	bio_put(bio);
-	pblk_free_rqd(pblk, rqd, WRITE);
+	pblk_free_rqd(pblk, rqd, PBLK_WRITE);
 }
 
 int pblk_recov_setup_rq(struct pblk *pblk, struct pblk_c_ctx *c_ctx,
@@ -84,12 +80,7 @@ int pblk_recov_setup_rq(struct pblk *pblk, struct pblk_c_ctx *c_ctx,
 	struct pblk_c_ctx *rec_ctx;
 	int nr_entries = c_ctx->nr_valid + c_ctx->nr_padded;
 
-	rec_rqd = pblk_alloc_rqd(pblk, WRITE);
-	if (IS_ERR(rec_rqd)) {
-		pr_err("pblk: could not create recovery req.\n");
-		return -ENOMEM;
-	}
-
+	rec_rqd = pblk_alloc_rqd(pblk, PBLK_WRITE);
 	rec_ctx = nvm_rq_to_pdu(rec_rqd);
 
 	/* Copy completion bitmap, but exclude the first X completed entries */
@@ -142,19 +133,19 @@ static int pblk_recov_l2p_from_emeta(struct pblk *pblk, struct pblk_line *line)
 	struct pblk_emeta *emeta = line->emeta;
 	struct line_emeta *emeta_buf = emeta->buf;
 	__le64 *lba_list;
-	int data_start;
-	int nr_data_lbas, nr_valid_lbas, nr_lbas = 0;
-	int i;
+	u64 data_start, data_end;
+	u64 nr_valid_lbas, nr_lbas = 0;
+	u64 i;
 
 	lba_list = pblk_recov_get_lba_list(pblk, emeta_buf);
 	if (!lba_list)
 		return 1;
 
 	data_start = pblk_line_smeta_start(pblk, line) + lm->smeta_sec;
-	nr_data_lbas = lm->sec_per_line - lm->emeta_sec[0];
+	data_end = line->emeta_ssec;
 	nr_valid_lbas = le64_to_cpu(emeta_buf->nr_valid_lbas);
 
-	for (i = data_start; i < nr_data_lbas && nr_lbas < nr_valid_lbas; i++) {
+	for (i = data_start; i < data_end; i++) {
 		struct ppa_addr ppa;
 		int pos;
 
@@ -181,8 +172,8 @@ static int pblk_recov_l2p_from_emeta(struct pblk *pblk, struct pblk_line *line)
 	}
 
 	if (nr_valid_lbas != nr_lbas)
-		pr_err("pblk: line %d - inconsistent lba list(%llu/%d)\n",
-				line->id, emeta_buf->nr_valid_lbas, nr_lbas);
+		pr_err("pblk: line %d - inconsistent lba list(%llu/%llu)\n",
+				line->id, nr_valid_lbas, nr_lbas);
 
 	line->left_msecs = 0;
 
@@ -225,7 +216,6 @@ static int pblk_recov_read_oob(struct pblk *pblk, struct pblk_line *line,
 	int rq_ppas, rq_len;
 	int i, j;
 	int ret = 0;
-	DECLARE_COMPLETION_ONSTACK(wait);
 
 	ppa_list = p.ppa_list;
 	meta_list = p.meta_list;
@@ -262,8 +252,6 @@ next_read_rq:
 	rqd->ppa_list = ppa_list;
 	rqd->dma_ppa_list = dma_ppa_list;
 	rqd->dma_meta_list = dma_meta_list;
-	rqd->end_io = pblk_end_io_sync;
-	rqd->private = &wait;
 
 	if (pblk_io_aligned(pblk, rq_ppas))
 		rqd->flags = pblk_set_read_mode(pblk, PBLK_READ_SEQUENTIAL);
@@ -289,19 +277,13 @@ next_read_rq:
 	}
 
 	/* If read fails, more padding is needed */
-	ret = pblk_submit_io(pblk, rqd);
+	ret = pblk_submit_io_sync(pblk, rqd);
 	if (ret) {
 		pr_err("pblk: I/O submission failed: %d\n", ret);
 		return ret;
 	}
 
-	if (!wait_for_completion_io_timeout(&wait,
-				msecs_to_jiffies(PBLK_COMMAND_TIMEOUT_MS))) {
-		pr_err("pblk: L2P recovery read timed out\n");
-		return -EINTR;
-	}
 	atomic_dec(&pblk->inflight_io);
-	reinit_completion(&wait);
 
 	/* At this point, the read should not fail. If it does, it is a problem
 	 * we cannot recover from here. Need FTL log.
@@ -338,13 +320,10 @@ static void pblk_end_io_recov(struct nvm_rq *rqd)
 {
 	struct pblk_pad_rq *pad_rq = rqd->private;
 	struct pblk *pblk = pad_rq->pblk;
-	struct nvm_tgt_dev *dev = pblk->dev;
 
 	pblk_up_page(pblk, rqd->ppa_list, rqd->nr_ppas);
 
-	bio_put(rqd->bio);
-	nvm_dev_dma_free(dev->parent, rqd->meta_list, rqd->dma_meta_list);
-	pblk_free_rqd(pblk, rqd, WRITE);
+	pblk_free_rqd(pblk, rqd, PBLK_WRITE_INT);
 
 	atomic_dec(&pblk->inflight_io);
 	kref_put(&pad_rq->ref, pblk_recov_complete);
@@ -404,25 +383,21 @@ next_pad_rq:
 	ppa_list = (void *)(meta_list) + pblk_dma_meta_size;
 	dma_ppa_list = dma_meta_list + pblk_dma_meta_size;
 
-	rqd = pblk_alloc_rqd(pblk, WRITE);
-	if (IS_ERR(rqd)) {
-		ret = PTR_ERR(rqd);
-		goto fail_free_meta;
-	}
-
 	bio = pblk_bio_map_addr(pblk, data, rq_ppas, rq_len,
 						PBLK_VMALLOC_META, GFP_KERNEL);
 	if (IS_ERR(bio)) {
 		ret = PTR_ERR(bio);
-		goto fail_free_rqd;
+		goto fail_free_meta;
 	}
 
 	bio->bi_iter.bi_sector = 0; /* internal bio */
 	bio_set_op_attrs(bio, REQ_OP_WRITE, 0);
 
+	rqd = pblk_alloc_rqd(pblk, PBLK_WRITE_INT);
+
 	rqd->bio = bio;
 	rqd->opcode = NVM_OP_PWRITE;
-	rqd->flags = pblk_set_progr_mode(pblk, WRITE);
+	rqd->flags = pblk_set_progr_mode(pblk, PBLK_WRITE);
 	rqd->meta_list = meta_list;
 	rqd->nr_ppas = rq_ppas;
 	rqd->ppa_list = ppa_list;
@@ -490,8 +465,6 @@ free_rq:
 
 fail_free_bio:
 	bio_put(bio);
-fail_free_rqd:
-	pblk_free_rqd(pblk, rqd, WRITE);
 fail_free_meta:
 	nvm_dev_dma_free(dev->parent, meta_list, dma_meta_list);
 fail_free_pad:
@@ -522,7 +495,6 @@ static int pblk_recov_scan_all_oob(struct pblk *pblk, struct pblk_line *line,
 	int ret = 0;
 	int rec_round;
 	int left_ppas = pblk_calc_sec_in_line(pblk, line) - line->cur_sec;
-	DECLARE_COMPLETION_ONSTACK(wait);
 
 	ppa_list = p.ppa_list;
 	meta_list = p.meta_list;
@@ -557,8 +529,6 @@ next_rq:
 	rqd->ppa_list = ppa_list;
 	rqd->dma_ppa_list = dma_ppa_list;
 	rqd->dma_meta_list = dma_meta_list;
-	rqd->end_io = pblk_end_io_sync;
-	rqd->private = &wait;
 
 	if (pblk_io_aligned(pblk, rq_ppas))
 		rqd->flags = pblk_set_read_mode(pblk, PBLK_READ_SEQUENTIAL);
@@ -584,18 +554,13 @@ next_rq:
 				addr_to_gen_ppa(pblk, w_ptr, line->id);
 	}
 
-	ret = pblk_submit_io(pblk, rqd);
+	ret = pblk_submit_io_sync(pblk, rqd);
 	if (ret) {
 		pr_err("pblk: I/O submission failed: %d\n", ret);
 		return ret;
 	}
 
-	if (!wait_for_completion_io_timeout(&wait,
-				msecs_to_jiffies(PBLK_COMMAND_TIMEOUT_MS))) {
-		pr_err("pblk: L2P recovery read timed out\n");
-	}
 	atomic_dec(&pblk->inflight_io);
-	reinit_completion(&wait);
 
 	/* This should not happen since the read failed during normal recovery,
 	 * but the media works funny sometimes...
@@ -663,7 +628,6 @@ static int pblk_recov_scan_oob(struct pblk *pblk, struct pblk_line *line,
 	int i, j;
 	int ret = 0;
 	int left_ppas = pblk_calc_sec_in_line(pblk, line);
-	DECLARE_COMPLETION_ONSTACK(wait);
 
 	ppa_list = p.ppa_list;
 	meta_list = p.meta_list;
@@ -696,8 +660,6 @@ next_rq:
 	rqd->ppa_list = ppa_list;
 	rqd->dma_ppa_list = dma_ppa_list;
 	rqd->dma_meta_list = dma_meta_list;
-	rqd->end_io = pblk_end_io_sync;
-	rqd->private = &wait;
 
 	if (pblk_io_aligned(pblk, rq_ppas))
 		rqd->flags = pblk_set_read_mode(pblk, PBLK_READ_SEQUENTIAL);
@@ -723,19 +685,14 @@ next_rq:
 				addr_to_gen_ppa(pblk, paddr, line->id);
 	}
 
-	ret = pblk_submit_io(pblk, rqd);
+	ret = pblk_submit_io_sync(pblk, rqd);
 	if (ret) {
 		pr_err("pblk: I/O submission failed: %d\n", ret);
 		bio_put(bio);
 		return ret;
 	}
 
-	if (!wait_for_completion_io_timeout(&wait,
-				msecs_to_jiffies(PBLK_COMMAND_TIMEOUT_MS))) {
-		pr_err("pblk: L2P recovery read timed out\n");
-	}
 	atomic_dec(&pblk->inflight_io);
-	reinit_completion(&wait);
 
 	/* Reached the end of the written line */
 	if (rqd->error) {
@@ -785,15 +742,9 @@ static int pblk_recov_l2p_from_oob(struct pblk *pblk, struct pblk_line *line)
 	dma_addr_t dma_ppa_list, dma_meta_list;
 	int done, ret = 0;
 
-	rqd = pblk_alloc_rqd(pblk, READ);
-	if (IS_ERR(rqd))
-		return PTR_ERR(rqd);
-
 	meta_list = nvm_dev_dma_alloc(dev->parent, GFP_KERNEL, &dma_meta_list);
-	if (!meta_list) {
-		ret = -ENOMEM;
-		goto free_rqd;
-	}
+	if (!meta_list)
+		return -ENOMEM;
 
 	ppa_list = (void *)(meta_list) + pblk_dma_meta_size;
 	dma_ppa_list = dma_meta_list + pblk_dma_meta_size;
@@ -804,6 +755,8 @@ static int pblk_recov_l2p_from_oob(struct pblk *pblk, struct pblk_line *line)
 		goto free_meta_list;
 	}
 
+	rqd = pblk_alloc_rqd(pblk, PBLK_READ);
+
 	p.ppa_list = ppa_list;
 	p.meta_list = meta_list;
 	p.rqd = rqd;
@@ -832,8 +785,6 @@ out:
 	kfree(data);
 free_meta_list:
 	nvm_dev_dma_free(dev->parent, meta_list, dma_meta_list);
-free_rqd:
-	pblk_free_rqd(pblk, rqd, READ);
 
 	return ret;
 }
@@ -851,10 +802,32 @@ static void pblk_recov_line_add_ordered(struct list_head *head,
 	__list_add(&line->list, t->list.prev, &t->list);
 }
 
-struct pblk_line *pblk_recov_l2p(struct pblk *pblk)
+static u64 pblk_line_emeta_start(struct pblk *pblk, struct pblk_line *line)
 {
 	struct nvm_tgt_dev *dev = pblk->dev;
 	struct nvm_geo *geo = &dev->geo;
+	struct pblk_line_meta *lm = &pblk->lm;
+	unsigned int emeta_secs;
+	u64 emeta_start;
+	struct ppa_addr ppa;
+	int pos;
+
+	emeta_secs = lm->emeta_sec[0];
+	emeta_start = lm->sec_per_line;
+
+	while (emeta_secs) {
+		emeta_start--;
+		ppa = addr_to_pblk_ppa(pblk, emeta_start, line->id);
+		pos = pblk_ppa_to_pos(geo, ppa);
+		if (!test_bit(pos, line->blk_bitmap))
+			emeta_secs--;
+	}
+
+	return emeta_start;
+}
+
+struct pblk_line *pblk_recov_l2p(struct pblk *pblk)
+{
 	struct pblk_line_meta *lm = &pblk->lm;
 	struct pblk_line_mgmt *l_mg = &pblk->l_mg;
 	struct pblk_line *line, *tline, *data_line = NULL;
@@ -900,9 +873,9 @@ struct pblk_line *pblk_recov_l2p(struct pblk *pblk)
 		if (le32_to_cpu(smeta_buf->header.identifier) != PBLK_MAGIC)
 			continue;
 
-		if (le16_to_cpu(smeta_buf->header.version) != 1) {
+		if (smeta_buf->header.version != SMETA_VERSION) {
 			pr_err("pblk: found incompatible line version %u\n",
-					smeta_buf->header.version);
+					le16_to_cpu(smeta_buf->header.version));
 			return ERR_PTR(-EINVAL);
 		}
 
@@ -954,15 +927,9 @@ struct pblk_line *pblk_recov_l2p(struct pblk *pblk)
 
 	/* Verify closed blocks and recover this portion of L2P table*/
 	list_for_each_entry_safe(line, tline, &recov_list, list) {
-		int off, nr_bb;
-
 		recovered_lines++;
-		/* Calculate where emeta starts based on the line bb */
-		off = lm->sec_per_line - lm->emeta_sec[0];
-		nr_bb = bitmap_weight(line->blk_bitmap, lm->blk_per_line);
-		off -= nr_bb * geo->sec_per_pl;
 
-		line->emeta_ssec = off;
+		line->emeta_ssec = pblk_line_emeta_start(pblk, line);
 		line->emeta = emeta;
 		memset(line->emeta->buf, 0, lm->emeta_len[0]);
 
@@ -987,7 +954,7 @@ next:
 			list_move_tail(&line->list, move_list);
 			spin_unlock(&l_mg->gc_lock);
 
-			mempool_free(line->map_bitmap, pblk->line_meta_pool);
+			kfree(line->map_bitmap);
 			line->map_bitmap = NULL;
 			line->smeta = NULL;
 			line->emeta = NULL;

drivers/lightnvm/pblk-rl.c

@@ -96,9 +96,11 @@ unsigned long pblk_rl_nr_free_blks(struct pblk_rl *rl)
  *
  * Only the total number of free blocks is used to configure the rate limiter.
  */
-static int pblk_rl_update_rates(struct pblk_rl *rl, unsigned long max)
+void pblk_rl_update_rates(struct pblk_rl *rl)
 {
+	struct pblk *pblk = container_of(rl, struct pblk, rl);
 	unsigned long free_blocks = pblk_rl_nr_free_blks(rl);
+	int max = rl->rb_budget;
 
 	if (free_blocks >= rl->high) {
 		rl->rb_user_max = max;
@@ -124,23 +126,18 @@ static int pblk_rl_update_rates(struct pblk_rl *rl, unsigned long max)
 		rl->rb_state = PBLK_RL_LOW;
 	}
 
-	return rl->rb_state;
+	if (rl->rb_state == (PBLK_RL_MID | PBLK_RL_LOW))
+		pblk_gc_should_start(pblk);
+	else
+		pblk_gc_should_stop(pblk);
 }
 
 void pblk_rl_free_lines_inc(struct pblk_rl *rl, struct pblk_line *line)
 {
-	struct pblk *pblk = container_of(rl, struct pblk, rl);
 	int blk_in_line = atomic_read(&line->blk_in_line);
-	int ret;
 
 	atomic_add(blk_in_line, &rl->free_blocks);
-	/* Rates will not change that often - no need to lock update */
-	ret = pblk_rl_update_rates(rl, rl->rb_budget);
-
-	if (ret == (PBLK_RL_MID | PBLK_RL_LOW))
-		pblk_gc_should_start(pblk);
-	else
-		pblk_gc_should_stop(pblk);
+	pblk_rl_update_rates(rl);
 }
 
 void pblk_rl_free_lines_dec(struct pblk_rl *rl, struct pblk_line *line)
@@ -148,19 +145,7 @@ void pblk_rl_free_lines_dec(struct pblk_rl *rl, struct pblk_line *line)
 	int blk_in_line = atomic_read(&line->blk_in_line);
 
 	atomic_sub(blk_in_line, &rl->free_blocks);
-}
-
-void pblk_gc_should_kick(struct pblk *pblk)
-{
-	struct pblk_rl *rl = &pblk->rl;
-	int ret;
-
-	/* Rates will not change that often - no need to lock update */
-	ret = pblk_rl_update_rates(rl, rl->rb_budget);
-	if (ret == (PBLK_RL_MID | PBLK_RL_LOW))
-		pblk_gc_should_start(pblk);
-	else
-		pblk_gc_should_stop(pblk);
+	pblk_rl_update_rates(rl);
 }
 
 int pblk_rl_high_thrs(struct pblk_rl *rl)
@@ -168,14 +153,9 @@ int pblk_rl_high_thrs(struct pblk_rl *rl)
 	return rl->high;
 }
 
-int pblk_rl_low_thrs(struct pblk_rl *rl)
-{
-	return rl->low;
-}
-
-int pblk_rl_sysfs_rate_show(struct pblk_rl *rl)
+int pblk_rl_max_io(struct pblk_rl *rl)
 {
-	return rl->rb_user_max;
+	return rl->rb_max_io;
 }
 
 static void pblk_rl_u_timer(unsigned long data)
@@ -214,6 +194,7 @@ void pblk_rl_init(struct pblk_rl *rl, int budget)
 	/* To start with, all buffer is available to user I/O writers */
 	rl->rb_budget = budget;
 	rl->rb_user_max = budget;
+	rl->rb_max_io = budget >> 1;
 	rl->rb_gc_max = 0;
 	rl->rb_state = PBLK_RL_HIGH;
 

drivers/lightnvm/pblk-sysfs.c

@@ -253,7 +253,7 @@ static ssize_t pblk_sysfs_lines(struct pblk *pblk, char *page)
 	sz += snprintf(page + sz, PAGE_SIZE - sz,
 		"GC: full:%d, high:%d, mid:%d, low:%d, empty:%d, queue:%d\n",
 			gc_full, gc_high, gc_mid, gc_low, gc_empty,
-			atomic_read(&pblk->gc.inflight_gc));
+			atomic_read(&pblk->gc.read_inflight_gc));
 
 	sz += snprintf(page + sz, PAGE_SIZE - sz,
 		"data (%d) cur:%d, left:%d, vsc:%d, s:%d, map:%d/%d (%d)\n",

drivers/lightnvm/pblk-write.c

@@ -20,7 +20,6 @@
 static unsigned long pblk_end_w_bio(struct pblk *pblk, struct nvm_rq *rqd,
 				    struct pblk_c_ctx *c_ctx)
 {
-	struct nvm_tgt_dev *dev = pblk->dev;
 	struct bio *original_bio;
 	unsigned long ret;
 	int i;
@@ -33,16 +32,18 @@ static unsigned long pblk_end_w_bio(struct pblk *pblk, struct nvm_rq *rqd,
 			bio_endio(original_bio);
 	}
 
+	if (c_ctx->nr_padded)
+		pblk_bio_free_pages(pblk, rqd->bio, c_ctx->nr_valid,
+							c_ctx->nr_padded);
+
 #ifdef CONFIG_NVM_DEBUG
-	atomic_long_add(c_ctx->nr_valid, &pblk->sync_writes);
+	atomic_long_add(rqd->nr_ppas, &pblk->sync_writes);
 #endif
 
 	ret = pblk_rb_sync_advance(&pblk->rwb, c_ctx->nr_valid);
 
-	nvm_dev_dma_free(dev->parent, rqd->meta_list, rqd->dma_meta_list);
-
 	bio_put(rqd->bio);
-	pblk_free_rqd(pblk, rqd, WRITE);
+	pblk_free_rqd(pblk, rqd, PBLK_WRITE);
 
 	return ret;
 }
@@ -107,10 +108,7 @@ static void pblk_end_w_fail(struct pblk *pblk, struct nvm_rq *rqd)
 		ppa_list = &rqd->ppa_addr;
 
 	recovery = mempool_alloc(pblk->rec_pool, GFP_ATOMIC);
-	if (!recovery) {
-		pr_err("pblk: could not allocate recovery context\n");
-		return;
-	}
+
 	INIT_LIST_HEAD(&recovery->failed);
 
 	bit = -1;
@@ -175,7 +173,6 @@ static void pblk_end_io_write(struct nvm_rq *rqd)
 static void pblk_end_io_write_meta(struct nvm_rq *rqd)
 {
 	struct pblk *pblk = rqd->private;
-	struct nvm_tgt_dev *dev = pblk->dev;
 	struct pblk_g_ctx *m_ctx = nvm_rq_to_pdu(rqd);
 	struct pblk_line *line = m_ctx->private;
 	struct pblk_emeta *emeta = line->emeta;
@@ -187,19 +184,13 @@ static void pblk_end_io_write_meta(struct nvm_rq *rqd)
 		pblk_log_write_err(pblk, rqd);
 		pr_err("pblk: metadata I/O failed. Line %d\n", line->id);
 	}
-#ifdef CONFIG_NVM_DEBUG
-	else
-		WARN_ONCE(rqd->bio->bi_status, "pblk: corrupted write error\n");
-#endif
 
 	sync = atomic_add_return(rqd->nr_ppas, &emeta->sync);
 	if (sync == emeta->nr_entries)
-		pblk_line_run_ws(pblk, line, NULL, pblk_line_close_ws,
-								pblk->close_wq);
+		pblk_gen_run_ws(pblk, line, NULL, pblk_line_close_ws,
+						GFP_ATOMIC, pblk->close_wq);
 
-	bio_put(rqd->bio);
-	nvm_dev_dma_free(dev->parent, rqd->meta_list, rqd->dma_meta_list);
-	pblk_free_rqd(pblk, rqd, READ);
+	pblk_free_rqd(pblk, rqd, PBLK_WRITE_INT);
 
 	atomic_dec(&pblk->inflight_io);
 }
@@ -213,7 +204,7 @@ static int pblk_alloc_w_rq(struct pblk *pblk, struct nvm_rq *rqd,
 	/* Setup write request */
 	rqd->opcode = NVM_OP_PWRITE;
 	rqd->nr_ppas = nr_secs;
-	rqd->flags = pblk_set_progr_mode(pblk, WRITE);
+	rqd->flags = pblk_set_progr_mode(pblk, PBLK_WRITE);
 	rqd->private = pblk;
 	rqd->end_io = end_io;
 
@@ -229,15 +220,16 @@ static int pblk_alloc_w_rq(struct pblk *pblk, struct nvm_rq *rqd,
 }
 
 static int pblk_setup_w_rq(struct pblk *pblk, struct nvm_rq *rqd,
-			   struct pblk_c_ctx *c_ctx, struct ppa_addr *erase_ppa)
+			   struct ppa_addr *erase_ppa)
 {
 	struct pblk_line_meta *lm = &pblk->lm;
 	struct pblk_line *e_line = pblk_line_get_erase(pblk);
+	struct pblk_c_ctx *c_ctx = nvm_rq_to_pdu(rqd);
 	unsigned int valid = c_ctx->nr_valid;
 	unsigned int padded = c_ctx->nr_padded;
 	unsigned int nr_secs = valid + padded;
 	unsigned long *lun_bitmap;
-	int ret = 0;
+	int ret;
 
 	lun_bitmap = kzalloc(lm->lun_bitmap_len, GFP_KERNEL);
 	if (!lun_bitmap)
@@ -279,7 +271,7 @@ int pblk_setup_w_rec_rq(struct pblk *pblk, struct nvm_rq *rqd,
 	pblk_map_rq(pblk, rqd, c_ctx->sentry, lun_bitmap, c_ctx->nr_valid, 0);
 
 	rqd->ppa_status = (u64)0;
-	rqd->flags = pblk_set_progr_mode(pblk, WRITE);
+	rqd->flags = pblk_set_progr_mode(pblk, PBLK_WRITE);
 
 	return ret;
 }
@@ -303,55 +295,6 @@ static int pblk_calc_secs_to_sync(struct pblk *pblk, unsigned int secs_avail,
 	return secs_to_sync;
 }
 
-static inline int pblk_valid_meta_ppa(struct pblk *pblk,
-				      struct pblk_line *meta_line,
-				      struct ppa_addr *ppa_list, int nr_ppas)
-{
-	struct nvm_tgt_dev *dev = pblk->dev;
-	struct nvm_geo *geo = &dev->geo;
-	struct pblk_line *data_line;
-	struct ppa_addr ppa, ppa_opt;
-	u64 paddr;
-	int i;
-
-	data_line = &pblk->lines[pblk_dev_ppa_to_line(ppa_list[0])];
-	paddr = pblk_lookup_page(pblk, meta_line);
-	ppa = addr_to_gen_ppa(pblk, paddr, 0);
-
-	if (test_bit(pblk_ppa_to_pos(geo, ppa), data_line->blk_bitmap))
-		return 1;
-
-	/* Schedule a metadata I/O that is half the distance from the data I/O
-	 * with regards to the number of LUNs forming the pblk instance. This
-	 * balances LUN conflicts across every I/O.
-	 *
-	 * When the LUN configuration changes (e.g., due to GC), this distance
-	 * can align, which would result on a LUN deadlock. In this case, modify
-	 * the distance to not be optimal, but allow metadata I/Os to succeed.
-	 */
-	ppa_opt = addr_to_gen_ppa(pblk, paddr + data_line->meta_distance, 0);
-	if (unlikely(ppa_opt.ppa == ppa.ppa)) {
-		data_line->meta_distance--;
-		return 0;
-	}
-
-	for (i = 0; i < nr_ppas; i += pblk->min_write_pgs)
-		if (ppa_list[i].g.ch == ppa_opt.g.ch &&
-					ppa_list[i].g.lun == ppa_opt.g.lun)
-			return 1;
-
-	if (test_bit(pblk_ppa_to_pos(geo, ppa_opt), data_line->blk_bitmap)) {
-		for (i = 0; i < nr_ppas; i += pblk->min_write_pgs)
-			if (ppa_list[i].g.ch == ppa.g.ch &&
-						ppa_list[i].g.lun == ppa.g.lun)
-				return 0;
-
-		return 1;
-	}
-
-	return 0;
-}
-
 int pblk_submit_meta_io(struct pblk *pblk, struct pblk_line *meta_line)
 {
 	struct nvm_tgt_dev *dev = pblk->dev;
@@ -370,11 +313,8 @@ int pblk_submit_meta_io(struct pblk *pblk, struct pblk_line *meta_line)
 	int i, j;
 	int ret;
 
-	rqd = pblk_alloc_rqd(pblk, READ);
-	if (IS_ERR(rqd)) {
-		pr_err("pblk: cannot allocate write req.\n");
-		return PTR_ERR(rqd);
-	}
+	rqd = pblk_alloc_rqd(pblk, PBLK_WRITE_INT);
+
 	m_ctx = nvm_rq_to_pdu(rqd);
 	m_ctx->private = meta_line;
 
@@ -407,8 +347,6 @@ int pblk_submit_meta_io(struct pblk *pblk, struct pblk_line *meta_line)
 	if (emeta->mem >= lm->emeta_len[0]) {
 		spin_lock(&l_mg->close_lock);
 		list_del(&meta_line->list);
-		WARN(!bitmap_full(meta_line->map_bitmap, lm->sec_per_line),
-				"pblk: corrupt meta line %d\n", meta_line->id);
 		spin_unlock(&l_mg->close_lock);
 	}
 
@@ -428,18 +366,51 @@ fail_rollback:
 	pblk_dealloc_page(pblk, meta_line, rq_ppas);
 	list_add(&meta_line->list, &meta_line->list);
 	spin_unlock(&l_mg->close_lock);
-
-	nvm_dev_dma_free(dev->parent, rqd->meta_list, rqd->dma_meta_list);
 fail_free_bio:
-	if (likely(l_mg->emeta_alloc_type == PBLK_VMALLOC_META))
-		bio_put(bio);
+	bio_put(bio);
 fail_free_rqd:
-	pblk_free_rqd(pblk, rqd, READ);
+	pblk_free_rqd(pblk, rqd, PBLK_WRITE_INT);
 	return ret;
 }
 
-static int pblk_sched_meta_io(struct pblk *pblk, struct ppa_addr *prev_list,
-			       int prev_n)
+static inline bool pblk_valid_meta_ppa(struct pblk *pblk,
+				       struct pblk_line *meta_line,
+				       struct nvm_rq *data_rqd)
+{
+	struct nvm_tgt_dev *dev = pblk->dev;
+	struct nvm_geo *geo = &dev->geo;
+	struct pblk_c_ctx *data_c_ctx = nvm_rq_to_pdu(data_rqd);
+	struct pblk_line *data_line = pblk_line_get_data(pblk);
+	struct ppa_addr ppa, ppa_opt;
+	u64 paddr;
+	int pos_opt;
+
+	/* Schedule a metadata I/O that is half the distance from the data I/O
+	 * with regards to the number of LUNs forming the pblk instance. This
+	 * balances LUN conflicts across every I/O.
+	 *
+	 * When the LUN configuration changes (e.g., due to GC), this distance
+	 * can align, which would result on metadata and data I/Os colliding. In
+	 * this case, modify the distance to not be optimal, but move the
+	 * optimal in the right direction.
+	 */
+	paddr = pblk_lookup_page(pblk, meta_line);
+	ppa = addr_to_gen_ppa(pblk, paddr, 0);
+	ppa_opt = addr_to_gen_ppa(pblk, paddr + data_line->meta_distance, 0);
+	pos_opt = pblk_ppa_to_pos(geo, ppa_opt);
+
+	if (test_bit(pos_opt, data_c_ctx->lun_bitmap) ||
+				test_bit(pos_opt, data_line->blk_bitmap))
+		return true;
+
+	if (unlikely(pblk_ppa_comp(ppa_opt, ppa)))
+		data_line->meta_distance--;
+
+	return false;
+}
+
+static struct pblk_line *pblk_should_submit_meta_io(struct pblk *pblk,
+						    struct nvm_rq *data_rqd)
 {
 	struct pblk_line_meta *lm = &pblk->lm;
 	struct pblk_line_mgmt *l_mg = &pblk->l_mg;
@@ -449,57 +420,45 @@ static int pblk_sched_meta_io(struct pblk *pblk, struct ppa_addr *prev_list,
 retry:
 	if (list_empty(&l_mg->emeta_list)) {
 		spin_unlock(&l_mg->close_lock);
-		return 0;
+		return NULL;
 	}
 	meta_line = list_first_entry(&l_mg->emeta_list, struct pblk_line, list);
-	if (bitmap_full(meta_line->map_bitmap, lm->sec_per_line))
+	if (meta_line->emeta->mem >= lm->emeta_len[0])
 		goto retry;
 	spin_unlock(&l_mg->close_lock);
 
-	if (!pblk_valid_meta_ppa(pblk, meta_line, prev_list, prev_n))
-		return 0;
+	if (!pblk_valid_meta_ppa(pblk, meta_line, data_rqd))
+		return NULL;
 
-	return pblk_submit_meta_io(pblk, meta_line);
+	return meta_line;
 }
 
 static int pblk_submit_io_set(struct pblk *pblk, struct nvm_rq *rqd)
 {
-	struct pblk_c_ctx *c_ctx = nvm_rq_to_pdu(rqd);
 	struct ppa_addr erase_ppa;
+	struct pblk_line *meta_line;
 	int err;
 
 	ppa_set_empty(&erase_ppa);
 
 	/* Assign lbas to ppas and populate request structure */
-	err = pblk_setup_w_rq(pblk, rqd, c_ctx, &erase_ppa);
+	err = pblk_setup_w_rq(pblk, rqd, &erase_ppa);
 	if (err) {
 		pr_err("pblk: could not setup write request: %d\n", err);
 		return NVM_IO_ERR;
 	}
 
-	if (likely(ppa_empty(erase_ppa))) {
-		/* Submit metadata write for previous data line */
-		err = pblk_sched_meta_io(pblk, rqd->ppa_list, rqd->nr_ppas);
-		if (err) {
-			pr_err("pblk: metadata I/O submission failed: %d", err);
-			return NVM_IO_ERR;
-		}
+	meta_line = pblk_should_submit_meta_io(pblk, rqd);
 
-		/* Submit data write for current data line */
-		err = pblk_submit_io(pblk, rqd);
-		if (err) {
-			pr_err("pblk: data I/O submission failed: %d\n", err);
-			return NVM_IO_ERR;
-		}
-	} else {
-		/* Submit data write for current data line */
-		err = pblk_submit_io(pblk, rqd);
-		if (err) {
-			pr_err("pblk: data I/O submission failed: %d\n", err);
-			return NVM_IO_ERR;
-		}
+	/* Submit data write for current data line */
+	err = pblk_submit_io(pblk, rqd);
+	if (err) {
+		pr_err("pblk: data I/O submission failed: %d\n", err);
+		return NVM_IO_ERR;
+	}
 
-		/* Submit available erase for next data line */
+	if (!ppa_empty(erase_ppa)) {
+		/* Submit erase for next data line */
 		if (pblk_blk_erase_async(pblk, erase_ppa)) {
 			struct pblk_line *e_line = pblk_line_get_erase(pblk);
 			struct nvm_tgt_dev *dev = pblk->dev;
@@ -512,6 +471,15 @@ static int pblk_submit_io_set(struct pblk *pblk, struct nvm_rq *rqd)
 		}
 	}
 
+	if (meta_line) {
+		/* Submit metadata write for previous data line */
+		err = pblk_submit_meta_io(pblk, meta_line);
+		if (err) {
+			pr_err("pblk: metadata I/O submission failed: %d", err);
+			return NVM_IO_ERR;
+		}
+	}
+
 	return NVM_IO_OK;
 }
 
@@ -521,7 +489,8 @@ static void pblk_free_write_rqd(struct pblk *pblk, struct nvm_rq *rqd)
 	struct bio *bio = rqd->bio;
 
 	if (c_ctx->nr_padded)
-		pblk_bio_free_pages(pblk, bio, rqd->nr_ppas, c_ctx->nr_padded);
+		pblk_bio_free_pages(pblk, bio, c_ctx->nr_valid,
+							c_ctx->nr_padded);
 }
 
 static int pblk_submit_write(struct pblk *pblk)
@@ -543,31 +512,24 @@ static int pblk_submit_write(struct pblk *pblk)
 	if (!secs_to_flush && secs_avail < pblk->min_write_pgs)
 		return 1;
 
-	rqd = pblk_alloc_rqd(pblk, WRITE);
-	if (IS_ERR(rqd)) {
-		pr_err("pblk: cannot allocate write req.\n");
-		return 1;
-	}
-
-	bio = bio_alloc(GFP_KERNEL, pblk->max_write_pgs);
-	if (!bio) {
-		pr_err("pblk: cannot allocate write bio\n");
-		goto fail_free_rqd;
-	}
-	bio->bi_iter.bi_sector = 0; /* internal bio */
-	bio_set_op_attrs(bio, REQ_OP_WRITE, 0);
-	rqd->bio = bio;
-
 	secs_to_sync = pblk_calc_secs_to_sync(pblk, secs_avail, secs_to_flush);
 	if (secs_to_sync > pblk->max_write_pgs) {
 		pr_err("pblk: bad buffer sync calculation\n");
-		goto fail_put_bio;
+		return 1;
 	}
 
 	secs_to_com = (secs_to_sync > secs_avail) ? secs_avail : secs_to_sync;
 	pos = pblk_rb_read_commit(&pblk->rwb, secs_to_com);
 
-	if (pblk_rb_read_to_bio(&pblk->rwb, rqd, bio, pos, secs_to_sync,
+	bio = bio_alloc(GFP_KERNEL, secs_to_sync);
+
+	bio->bi_iter.bi_sector = 0; /* internal bio */
+	bio_set_op_attrs(bio, REQ_OP_WRITE, 0);
+
+	rqd = pblk_alloc_rqd(pblk, PBLK_WRITE);
+	rqd->bio = bio;
+
+	if (pblk_rb_read_to_bio(&pblk->rwb, rqd, pos, secs_to_sync,
 								secs_avail)) {
 		pr_err("pblk: corrupted write bio\n");
 		goto fail_put_bio;
@@ -586,8 +548,7 @@ fail_free_bio:
 	pblk_free_write_rqd(pblk, rqd);
 fail_put_bio:
 	bio_put(bio);
-fail_free_rqd:
-	pblk_free_rqd(pblk, rqd, WRITE);
+	pblk_free_rqd(pblk, rqd, PBLK_WRITE);
 
 	return 1;
 }

drivers/lightnvm/pblk.h

@@ -40,10 +40,6 @@
 #define PBLK_MAX_REQ_ADDRS (64)
 #define PBLK_MAX_REQ_ADDRS_PW (6)
 
-#define PBLK_WS_POOL_SIZE (128)
-#define PBLK_META_POOL_SIZE (128)
-#define PBLK_READ_REQ_POOL_SIZE (1024)
-
 #define PBLK_NR_CLOSE_JOBS (4)
 
 #define PBLK_CACHE_NAME_LEN (DISK_NAME_LEN + 16)
@@ -59,7 +55,15 @@
 		for ((i) = 0, rlun = &(pblk)->luns[0]; \
 			(i) < (pblk)->nr_luns; (i)++, rlun = &(pblk)->luns[(i)])
 
-#define ERASE 2 /* READ = 0, WRITE = 1 */
+/* Static pool sizes */
+#define PBLK_GEN_WS_POOL_SIZE (2)
+
+enum {
+	PBLK_READ		= READ,
+	PBLK_WRITE		= WRITE,/* Write from write buffer */
+	PBLK_WRITE_INT,			/* Internal write - no write buffer */
+	PBLK_ERASE,
+};
 
 enum {
 	/* IO Types */
@@ -95,6 +99,7 @@ enum {
 };
 
 #define pblk_dma_meta_size (sizeof(struct pblk_sec_meta) * PBLK_MAX_REQ_ADDRS)
+#define pblk_dma_ppa_size (sizeof(u64) * PBLK_MAX_REQ_ADDRS)
 
 /* write buffer completion context */
 struct pblk_c_ctx {
@@ -106,9 +111,10 @@ struct pblk_c_ctx {
 	unsigned int nr_padded;
 };
 
-/* generic context */
+/* read context */
 struct pblk_g_ctx {
 	void *private;
+	u64 lba;
 };
 
 /* Pad context */
@@ -207,6 +213,7 @@ struct pblk_lun {
 struct pblk_gc_rq {
 	struct pblk_line *line;
 	void *data;
+	u64 paddr_list[PBLK_MAX_REQ_ADDRS];
 	u64 lba_list[PBLK_MAX_REQ_ADDRS];
 	int nr_secs;
 	int secs_to_gc;
@@ -231,7 +238,10 @@ struct pblk_gc {
 	struct timer_list gc_timer;
 
 	struct semaphore gc_sem;
-	atomic_t inflight_gc;
+	atomic_t read_inflight_gc; /* Number of lines with inflight GC reads */
+	atomic_t pipeline_gc;	   /* Number of lines in the GC pipeline -
+				    * started reads to finished writes
+				    */
 	int w_entries;
 
 	struct list_head w_list;
@@ -267,6 +277,7 @@ struct pblk_rl {
 	int rb_gc_max;		/* Max buffer entries available for GC I/O */
 	int rb_gc_rsv;		/* Reserved buffer entries for GC I/O */
 	int rb_state;		/* Rate-limiter current state */
+	int rb_max_io;		/* Maximum size for an I/O giving the config */
 
 	atomic_t rb_user_cnt;	/* User I/O buffer counter */
 	atomic_t rb_gc_cnt;	/* GC I/O buffer counter */
@@ -310,6 +321,7 @@ enum {
 };
 
 #define PBLK_MAGIC 0x70626c6b /*pblk*/
+#define SMETA_VERSION cpu_to_le16(1)
 
 struct line_header {
 	__le32 crc;
@@ -618,15 +630,16 @@ struct pblk {
 
 	struct list_head compl_list;
 
-	mempool_t *page_pool;
-	mempool_t *line_ws_pool;
+	mempool_t *page_bio_pool;
+	mempool_t *gen_ws_pool;
 	mempool_t *rec_pool;
-	mempool_t *g_rq_pool;
+	mempool_t *r_rq_pool;
 	mempool_t *w_rq_pool;
-	mempool_t *line_meta_pool;
+	mempool_t *e_rq_pool;
 
 	struct workqueue_struct *close_wq;
 	struct workqueue_struct *bb_wq;
+	struct workqueue_struct *r_end_wq;
 
 	struct timer_list wtimer;
 
@@ -657,15 +670,15 @@ int pblk_rb_may_write_gc(struct pblk_rb *rb, unsigned int nr_entries,
 void pblk_rb_write_entry_user(struct pblk_rb *rb, void *data,
 			      struct pblk_w_ctx w_ctx, unsigned int pos);
 void pblk_rb_write_entry_gc(struct pblk_rb *rb, void *data,
-			    struct pblk_w_ctx w_ctx, struct pblk_line *gc_line,
-			    unsigned int pos);
+			    struct pblk_w_ctx w_ctx, struct pblk_line *line,
+			    u64 paddr, unsigned int pos);
 struct pblk_w_ctx *pblk_rb_w_ctx(struct pblk_rb *rb, unsigned int pos);
 void pblk_rb_flush(struct pblk_rb *rb);
 
 void pblk_rb_sync_l2p(struct pblk_rb *rb);
 unsigned int pblk_rb_read_to_bio(struct pblk_rb *rb, struct nvm_rq *rqd,
-				 struct bio *bio, unsigned int pos,
-				 unsigned int nr_entries, unsigned int count);
+				 unsigned int pos, unsigned int nr_entries,
+				 unsigned int count);
 unsigned int pblk_rb_read_to_bio_list(struct pblk_rb *rb, struct bio *bio,
 				      struct list_head *list,
 				      unsigned int max);
@@ -692,24 +705,23 @@ ssize_t pblk_rb_sysfs(struct pblk_rb *rb, char *buf);
 /*
  * pblk core
  */
-struct nvm_rq *pblk_alloc_rqd(struct pblk *pblk, int rw);
+struct nvm_rq *pblk_alloc_rqd(struct pblk *pblk, int type);
+void pblk_free_rqd(struct pblk *pblk, struct nvm_rq *rqd, int type);
 void pblk_set_sec_per_write(struct pblk *pblk, int sec_per_write);
 int pblk_setup_w_rec_rq(struct pblk *pblk, struct nvm_rq *rqd,
 			struct pblk_c_ctx *c_ctx);
-void pblk_free_rqd(struct pblk *pblk, struct nvm_rq *rqd, int rw);
-void pblk_wait_for_meta(struct pblk *pblk);
-struct ppa_addr pblk_get_lba_map(struct pblk *pblk, sector_t lba);
 void pblk_discard(struct pblk *pblk, struct bio *bio);
 void pblk_log_write_err(struct pblk *pblk, struct nvm_rq *rqd);
 void pblk_log_read_err(struct pblk *pblk, struct nvm_rq *rqd);
 int pblk_submit_io(struct pblk *pblk, struct nvm_rq *rqd);
+int pblk_submit_io_sync(struct pblk *pblk, struct nvm_rq *rqd);
 int pblk_submit_meta_io(struct pblk *pblk, struct pblk_line *meta_line);
 struct bio *pblk_bio_map_addr(struct pblk *pblk, void *data,
 			      unsigned int nr_secs, unsigned int len,
 			      int alloc_type, gfp_t gfp_mask);
 struct pblk_line *pblk_line_get(struct pblk *pblk);
 struct pblk_line *pblk_line_get_first_data(struct pblk *pblk);
-void pblk_line_replace_data(struct pblk *pblk);
+struct pblk_line *pblk_line_replace_data(struct pblk *pblk);
 int pblk_line_recov_alloc(struct pblk *pblk, struct pblk_line *line);
 void pblk_line_recov_close(struct pblk *pblk, struct pblk_line *line);
 struct pblk_line *pblk_line_get_data(struct pblk *pblk);
@@ -719,19 +731,18 @@ int pblk_line_is_full(struct pblk_line *line);
 void pblk_line_free(struct pblk *pblk, struct pblk_line *line);
 void pblk_line_close_meta(struct pblk *pblk, struct pblk_line *line);
 void pblk_line_close(struct pblk *pblk, struct pblk_line *line);
-void pblk_line_close_meta_sync(struct pblk *pblk);
 void pblk_line_close_ws(struct work_struct *work);
 void pblk_pipeline_stop(struct pblk *pblk);
-void pblk_line_mark_bb(struct work_struct *work);
-void pblk_line_run_ws(struct pblk *pblk, struct pblk_line *line, void *priv,
-		      void (*work)(struct work_struct *),
-		      struct workqueue_struct *wq);
+void pblk_gen_run_ws(struct pblk *pblk, struct pblk_line *line, void *priv,
+		     void (*work)(struct work_struct *), gfp_t gfp_mask,
+		     struct workqueue_struct *wq);
 u64 pblk_line_smeta_start(struct pblk *pblk, struct pblk_line *line);
 int pblk_line_read_smeta(struct pblk *pblk, struct pblk_line *line);
 int pblk_line_read_emeta(struct pblk *pblk, struct pblk_line *line,
 			 void *emeta_buf);
 int pblk_blk_erase_async(struct pblk *pblk, struct ppa_addr erase_ppa);
 void pblk_line_put(struct kref *ref);
+void pblk_line_put_wq(struct kref *ref);
 struct list_head *pblk_line_gc_list(struct pblk *pblk, struct pblk_line *line);
 u64 pblk_lookup_page(struct pblk *pblk, struct pblk_line *line);
 void pblk_dealloc_page(struct pblk *pblk, struct pblk_line *line, int nr_secs);
@@ -745,7 +756,6 @@ void pblk_down_rq(struct pblk *pblk, struct ppa_addr *ppa_list, int nr_ppas,
 void pblk_down_page(struct pblk *pblk, struct ppa_addr *ppa_list, int nr_ppas);
 void pblk_up_rq(struct pblk *pblk, struct ppa_addr *ppa_list, int nr_ppas,
 		unsigned long *lun_bitmap);
-void pblk_end_bio_sync(struct bio *bio);
 void pblk_end_io_sync(struct nvm_rq *rqd);
 int pblk_bio_add_pages(struct pblk *pblk, struct bio *bio, gfp_t flags,
 		       int nr_pages);
@@ -760,7 +770,7 @@ void pblk_update_map_cache(struct pblk *pblk, sector_t lba,
 void pblk_update_map_dev(struct pblk *pblk, sector_t lba,
 			 struct ppa_addr ppa, struct ppa_addr entry_line);
 int pblk_update_map_gc(struct pblk *pblk, sector_t lba, struct ppa_addr ppa,
-		       struct pblk_line *gc_line);
+		       struct pblk_line *gc_line, u64 paddr);
 void pblk_lookup_l2p_rand(struct pblk *pblk, struct ppa_addr *ppas,
 			  u64 *lba_list, int nr_secs);
 void pblk_lookup_l2p_seq(struct pblk *pblk, struct ppa_addr *ppas,
@@ -771,9 +781,7 @@ void pblk_lookup_l2p_seq(struct pblk *pblk, struct ppa_addr *ppas,
  */
 int pblk_write_to_cache(struct pblk *pblk, struct bio *bio,
 			unsigned long flags);
-int pblk_write_gc_to_cache(struct pblk *pblk, void *data, u64 *lba_list,
-			   unsigned int nr_entries, unsigned int nr_rec_entries,
-			   struct pblk_line *gc_line, unsigned long flags);
+int pblk_write_gc_to_cache(struct pblk *pblk, struct pblk_gc_rq *gc_rq);
 
 /*
  * pblk map
@@ -797,9 +805,7 @@ void pblk_write_should_kick(struct pblk *pblk);
  */
 extern struct bio_set *pblk_bio_set;
 int pblk_submit_read(struct pblk *pblk, struct bio *bio);
-int pblk_submit_read_gc(struct pblk *pblk, u64 *lba_list, void *data,
-			unsigned int nr_secs, unsigned int *secs_to_gc,
-			struct pblk_line *line);
+int pblk_submit_read_gc(struct pblk *pblk, struct pblk_gc_rq *gc_rq);
 /*
  * pblk recovery
  */
@@ -815,7 +821,7 @@ int pblk_recov_setup_rq(struct pblk *pblk, struct pblk_c_ctx *c_ctx,
  * pblk gc
  */
 #define PBLK_GC_MAX_READERS 8	/* Max number of outstanding GC reader jobs */
-#define PBLK_GC_W_QD 128	/* Queue depth for inflight GC write I/Os */
+#define PBLK_GC_RQ_QD 128	/* Queue depth for inflight GC requests */
 #define PBLK_GC_L_QD 4		/* Queue depth for inflight GC lines */
 #define PBLK_GC_RSV_LINE 1	/* Reserved lines for GC */
 
@@ -824,7 +830,7 @@ void pblk_gc_exit(struct pblk *pblk);
 void pblk_gc_should_start(struct pblk *pblk);
 void pblk_gc_should_stop(struct pblk *pblk);
 void pblk_gc_should_kick(struct pblk *pblk);
-void pblk_gc_kick(struct pblk *pblk);
+void pblk_gc_free_full_lines(struct pblk *pblk);
 void pblk_gc_sysfs_state_show(struct pblk *pblk, int *gc_enabled,
 			      int *gc_active);
 int pblk_gc_sysfs_force(struct pblk *pblk, int force);
@@ -834,8 +840,8 @@ int pblk_gc_sysfs_force(struct pblk *pblk, int force);
  */
 void pblk_rl_init(struct pblk_rl *rl, int budget);
 void pblk_rl_free(struct pblk_rl *rl);
+void pblk_rl_update_rates(struct pblk_rl *rl);
 int pblk_rl_high_thrs(struct pblk_rl *rl);
-int pblk_rl_low_thrs(struct pblk_rl *rl);
 unsigned long pblk_rl_nr_free_blks(struct pblk_rl *rl);
 int pblk_rl_user_may_insert(struct pblk_rl *rl, int nr_entries);
 void pblk_rl_inserted(struct pblk_rl *rl, int nr_entries);
@@ -843,10 +849,9 @@ void pblk_rl_user_in(struct pblk_rl *rl, int nr_entries);
 int pblk_rl_gc_may_insert(struct pblk_rl *rl, int nr_entries);
 void pblk_rl_gc_in(struct pblk_rl *rl, int nr_entries);
 void pblk_rl_out(struct pblk_rl *rl, int nr_user, int nr_gc);
-int pblk_rl_sysfs_rate_show(struct pblk_rl *rl);
+int pblk_rl_max_io(struct pblk_rl *rl);
 void pblk_rl_free_lines_inc(struct pblk_rl *rl, struct pblk_line *line);
 void pblk_rl_free_lines_dec(struct pblk_rl *rl, struct pblk_line *line);
-void pblk_rl_set_space_limit(struct pblk_rl *rl, int entries_left);
 int pblk_rl_is_limit(struct pblk_rl *rl);
 
 /*
@@ -892,13 +897,7 @@ static inline void *emeta_to_vsc(struct pblk *pblk, struct line_emeta *emeta)
 
 static inline int pblk_line_vsc(struct pblk_line *line)
 {
-	int vsc;
-
-	spin_lock(&line->lock);
-	vsc = le32_to_cpu(*line->vsc);
-	spin_unlock(&line->lock);
-
-	return vsc;
+	return le32_to_cpu(*line->vsc);
 }
 
 #define NVM_MEM_PAGE_WRITE (8)
@@ -1140,7 +1139,7 @@ static inline int pblk_set_progr_mode(struct pblk *pblk, int type)
 
 	flags = geo->plane_mode >> 1;
 
-	if (type == WRITE)
+	if (type == PBLK_WRITE)
 		flags |= NVM_IO_SCRAMBLE_ENABLE;
 
 	return flags;
@@ -1200,7 +1199,6 @@ static inline void pblk_print_failed_rqd(struct pblk *pblk, struct nvm_rq *rqd,
 
 	pr_err("error:%d, ppa_status:%llx\n", error, rqd->ppa_status);
 }
-#endif
 
 static inline int pblk_boundary_ppa_checks(struct nvm_tgt_dev *tgt_dev,
 				       struct ppa_addr *ppas, int nr_ppas)
@@ -1221,14 +1219,50 @@ static inline int pblk_boundary_ppa_checks(struct nvm_tgt_dev *tgt_dev,
 				ppa->g.sec < geo->sec_per_pg)
 			continue;
 
-#ifdef CONFIG_NVM_DEBUG
 		print_ppa(ppa, "boundary", i);
-#endif
+
 		return 1;
 	}
 	return 0;
 }
 
+static inline int pblk_check_io(struct pblk *pblk, struct nvm_rq *rqd)
+{
+	struct nvm_tgt_dev *dev = pblk->dev;
+	struct ppa_addr *ppa_list;
+
+	ppa_list = (rqd->nr_ppas > 1) ? rqd->ppa_list : &rqd->ppa_addr;
+
+	if (pblk_boundary_ppa_checks(dev, ppa_list, rqd->nr_ppas)) {
+		WARN_ON(1);
+		return -EINVAL;
+	}
+
+	if (rqd->opcode == NVM_OP_PWRITE) {
+		struct pblk_line *line;
+		struct ppa_addr ppa;
+		int i;
+
+		for (i = 0; i < rqd->nr_ppas; i++) {
+			ppa = ppa_list[i];
+			line = &pblk->lines[pblk_dev_ppa_to_line(ppa)];
+
+			spin_lock(&line->lock);
+			if (line->state != PBLK_LINESTATE_OPEN) {
+				pr_err("pblk: bad ppa: line:%d,state:%d\n",
+							line->id, line->state);
+				WARN_ON(1);
+				spin_unlock(&line->lock);
+				return -EINVAL;
+			}
+			spin_unlock(&line->lock);
+		}
+	}
+
+	return 0;
+}
+#endif
+
 static inline int pblk_boundary_paddr_checks(struct pblk *pblk, u64 paddr)
 {
 	struct pblk_line_meta *lm = &pblk->lm;

drivers/md/bcache/alloc.c

@@ -407,7 +407,8 @@ long bch_bucket_alloc(struct cache *ca, unsigned reserve, bool wait)
 
 	finish_wait(&ca->set->bucket_wait, &w);
 out:
-	wake_up_process(ca->alloc_thread);
+	if (ca->alloc_thread)
+		wake_up_process(ca->alloc_thread);
 
 	trace_bcache_alloc(ca, reserve);
 
@@ -442,6 +443,11 @@ out:
 		b->prio = INITIAL_PRIO;
 	}
 
+	if (ca->set->avail_nbuckets > 0) {
+		ca->set->avail_nbuckets--;
+		bch_update_bucket_in_use(ca->set, &ca->set->gc_stats);
+	}
+
 	return r;
 }
 
@@ -449,6 +455,11 @@ void __bch_bucket_free(struct cache *ca, struct bucket *b)
 {
 	SET_GC_MARK(b, 0);
 	SET_GC_SECTORS_USED(b, 0);
+
+	if (ca->set->avail_nbuckets < ca->set->nbuckets) {
+		ca->set->avail_nbuckets++;
+		bch_update_bucket_in_use(ca->set, &ca->set->gc_stats);
+	}
 }
 
 void bch_bucket_free(struct cache_set *c, struct bkey *k)
@@ -601,7 +612,7 @@ bool bch_alloc_sectors(struct cache_set *c, struct bkey *k, unsigned sectors,
 
 	/*
 	 * If we had to allocate, we might race and not need to allocate the
-	 * second time we call find_data_bucket(). If we allocated a bucket but
+	 * second time we call pick_data_bucket(). If we allocated a bucket but
 	 * didn't use it, drop the refcount bch_bucket_alloc_set() took:
 	 */
 	if (KEY_PTRS(&alloc.key))

drivers/md/bcache/bcache.h

@@ -185,6 +185,7 @@
 #include <linux/mutex.h>
 #include <linux/rbtree.h>
 #include <linux/rwsem.h>
+#include <linux/refcount.h>
 #include <linux/types.h>
 #include <linux/workqueue.h>
 
@@ -266,9 +267,6 @@ struct bcache_device {
 	atomic_t		*stripe_sectors_dirty;
 	unsigned long		*full_dirty_stripes;
 
-	unsigned long		sectors_dirty_last;
-	long			sectors_dirty_derivative;
-
 	struct bio_set		*bio_split;
 
 	unsigned		data_csum:1;
@@ -300,7 +298,7 @@ struct cached_dev {
 	struct semaphore	sb_write_mutex;
 
 	/* Refcount on the cache set. Always nonzero when we're caching. */
-	atomic_t		count;
+	refcount_t		count;
 	struct work_struct	detach;
 
 	/*
@@ -363,12 +361,14 @@ struct cached_dev {
 
 	uint64_t		writeback_rate_target;
 	int64_t			writeback_rate_proportional;
-	int64_t			writeback_rate_derivative;
-	int64_t			writeback_rate_change;
+	int64_t			writeback_rate_integral;
+	int64_t			writeback_rate_integral_scaled;
+	int32_t			writeback_rate_change;
 
 	unsigned		writeback_rate_update_seconds;
-	unsigned		writeback_rate_d_term;
+	unsigned		writeback_rate_i_term_inverse;
 	unsigned		writeback_rate_p_term_inverse;
+	unsigned		writeback_rate_minimum;
 };
 
 enum alloc_reserve {
@@ -582,6 +582,7 @@ struct cache_set {
 	uint8_t			need_gc;
 	struct gc_stat		gc_stats;
 	size_t			nbuckets;
+	size_t			avail_nbuckets;
 
 	struct task_struct	*gc_thread;
 	/* Where in the btree gc currently is */
@@ -807,13 +808,13 @@ do {									\
 
 static inline void cached_dev_put(struct cached_dev *dc)
 {
-	if (atomic_dec_and_test(&dc->count))
+	if (refcount_dec_and_test(&dc->count))
 		schedule_work(&dc->detach);
 }
 
 static inline bool cached_dev_get(struct cached_dev *dc)
 {
-	if (!atomic_inc_not_zero(&dc->count))
+	if (!refcount_inc_not_zero(&dc->count))
 		return false;
 
 	/* Paired with the mb in cached_dev_attach */

drivers/md/bcache/btree.c

@@ -1241,6 +1241,11 @@ void bch_initial_mark_key(struct cache_set *c, int level, struct bkey *k)
 	__bch_btree_mark_key(c, level, k);
 }
 
+void bch_update_bucket_in_use(struct cache_set *c, struct gc_stat *stats)
+{
+	stats->in_use = (c->nbuckets - c->avail_nbuckets) * 100 / c->nbuckets;
+}
+
 static bool btree_gc_mark_node(struct btree *b, struct gc_stat *gc)
 {
 	uint8_t stale = 0;
@@ -1652,9 +1657,8 @@ static void btree_gc_start(struct cache_set *c)
 	mutex_unlock(&c->bucket_lock);
 }
 
-static size_t bch_btree_gc_finish(struct cache_set *c)
+static void bch_btree_gc_finish(struct cache_set *c)
 {
-	size_t available = 0;
 	struct bucket *b;
 	struct cache *ca;
 	unsigned i;
@@ -1691,6 +1695,7 @@ static size_t bch_btree_gc_finish(struct cache_set *c)
 	}
 	rcu_read_unlock();
 
+	c->avail_nbuckets = 0;
 	for_each_cache(ca, c, i) {
 		uint64_t *i;
 
@@ -1712,18 +1717,16 @@ static size_t bch_btree_gc_finish(struct cache_set *c)
 			BUG_ON(!GC_MARK(b) && GC_SECTORS_USED(b));
 
 			if (!GC_MARK(b) || GC_MARK(b) == GC_MARK_RECLAIMABLE)
-				available++;
+				c->avail_nbuckets++;
 		}
 	}
 
 	mutex_unlock(&c->bucket_lock);
-	return available;
 }
 
 static void bch_btree_gc(struct cache_set *c)
 {
 	int ret;
-	unsigned long available;
 	struct gc_stat stats;
 	struct closure writes;
 	struct btree_op op;
@@ -1746,14 +1749,14 @@ static void bch_btree_gc(struct cache_set *c)
 			pr_warn("gc failed!");
 	} while (ret);
 
-	available = bch_btree_gc_finish(c);
+	bch_btree_gc_finish(c);
 	wake_up_allocators(c);
 
 	bch_time_stats_update(&c->btree_gc_time, start_time);
 
 	stats.key_bytes *= sizeof(uint64_t);
 	stats.data	<<= 9;
-	stats.in_use	= (c->nbuckets - available) * 100 / c->nbuckets;
+	bch_update_bucket_in_use(c, &stats);
 	memcpy(&c->gc_stats, &stats, sizeof(struct gc_stat));
 
 	trace_bcache_gc_end(c);

drivers/md/bcache/btree.h

@@ -306,5 +306,5 @@ void bch_keybuf_del(struct keybuf *, struct keybuf_key *);
 struct keybuf_key *bch_keybuf_next(struct keybuf *);
 struct keybuf_key *bch_keybuf_next_rescan(struct cache_set *, struct keybuf *,
 					  struct bkey *, keybuf_pred_fn *);
-
+void bch_update_bucket_in_use(struct cache_set *c, struct gc_stat *stats);
 #endif

drivers/md/bcache/closure.h

@@ -252,6 +252,12 @@ static inline void set_closure_fn(struct closure *cl, closure_fn *fn,
 static inline void closure_queue(struct closure *cl)
 {
 	struct workqueue_struct *wq = cl->wq;
+	/**
+	 * Changes made to closure, work_struct, or a couple of other structs
+	 * may cause work.func not pointing to the right location.
+	 */
+	BUILD_BUG_ON(offsetof(struct closure, fn)
+		     != offsetof(struct work_struct, func));
 	if (wq) {
 		INIT_WORK(&cl->work, cl->work.func);
 		BUG_ON(!queue_work(wq, &cl->work));

drivers/md/bcache/request.c

@@ -27,12 +27,12 @@ struct kmem_cache *bch_search_cache;
 
 static void bch_data_insert_start(struct closure *);
 
-static unsigned cache_mode(struct cached_dev *dc, struct bio *bio)
+static unsigned cache_mode(struct cached_dev *dc)
 {
 	return BDEV_CACHE_MODE(&dc->sb);
 }
 
-static bool verify(struct cached_dev *dc, struct bio *bio)
+static bool verify(struct cached_dev *dc)
 {
 	return dc->verify;
 }
@@ -370,7 +370,7 @@ static struct hlist_head *iohash(struct cached_dev *dc, uint64_t k)
 static bool check_should_bypass(struct cached_dev *dc, struct bio *bio)
 {
 	struct cache_set *c = dc->disk.c;
-	unsigned mode = cache_mode(dc, bio);
+	unsigned mode = cache_mode(dc);
 	unsigned sectors, congested = bch_get_congested(c);
 	struct task_struct *task = current;
 	struct io *i;
@@ -385,6 +385,14 @@ static bool check_should_bypass(struct cached_dev *dc, struct bio *bio)
 	     op_is_write(bio_op(bio))))
 		goto skip;
 
+	/*
+	 * Flag for bypass if the IO is for read-ahead or background,
+	 * unless the read-ahead request is for metadata (eg, for gfs2).
+	 */
+	if (bio->bi_opf & (REQ_RAHEAD|REQ_BACKGROUND) &&
+	    !(bio->bi_opf & REQ_META))
+		goto skip;
+
 	if (bio->bi_iter.bi_sector & (c->sb.block_size - 1) ||
 	    bio_sectors(bio) & (c->sb.block_size - 1)) {
 		pr_debug("skipping unaligned io");
@@ -463,6 +471,7 @@ struct search {
 	unsigned		recoverable:1;
 	unsigned		write:1;
 	unsigned		read_dirty_data:1;
+	unsigned		cache_missed:1;
 
 	unsigned long		start_time;
 
@@ -649,6 +658,7 @@ static inline struct search *search_alloc(struct bio *bio,
 
 	s->orig_bio		= bio;
 	s->cache_miss		= NULL;
+	s->cache_missed		= 0;
 	s->d			= d;
 	s->recoverable		= 1;
 	s->write		= op_is_write(bio_op(bio));
@@ -698,8 +708,16 @@ 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 (s->recoverable) {
+	/*
+	 * 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 (s->recoverable &&
+	    (dc && !atomic_read(&dc->has_dirty))) {
 		/* Retry from the backing device: */
 		trace_bcache_read_retry(s->orig_bio);
 
@@ -740,7 +758,7 @@ static void cached_dev_read_done(struct closure *cl)
 		s->cache_miss = NULL;
 	}
 
-	if (verify(dc, &s->bio.bio) && s->recoverable && !s->read_dirty_data)
+	if (verify(dc) && s->recoverable && !s->read_dirty_data)
 		bch_data_verify(dc, s->orig_bio);
 
 	bio_complete(s);
@@ -760,12 +778,12 @@ static void cached_dev_read_done_bh(struct closure *cl)
 	struct cached_dev *dc = container_of(s->d, struct cached_dev, disk);
 
 	bch_mark_cache_accounting(s->iop.c, s->d,
-				  !s->cache_miss, s->iop.bypass);
+				  !s->cache_missed, s->iop.bypass);
 	trace_bcache_read(s->orig_bio, !s->cache_miss, s->iop.bypass);
 
 	if (s->iop.status)
 		continue_at_nobarrier(cl, cached_dev_read_error, bcache_wq);
-	else if (s->iop.bio || verify(dc, &s->bio.bio))
+	else if (s->iop.bio || verify(dc))
 		continue_at_nobarrier(cl, cached_dev_read_done, bcache_wq);
 	else
 		continue_at_nobarrier(cl, cached_dev_bio_complete, NULL);
@@ -779,6 +797,8 @@ static int cached_dev_cache_miss(struct btree *b, struct search *s,
 	struct cached_dev *dc = container_of(s->d, struct cached_dev, disk);
 	struct bio *miss, *cache_bio;
 
+	s->cache_missed = 1;
+
 	if (s->cache_miss || s->iop.bypass) {
 		miss = bio_next_split(bio, sectors, GFP_NOIO, s->d->bio_split);
 		ret = miss == bio ? MAP_DONE : MAP_CONTINUE;
@@ -892,7 +912,7 @@ static void cached_dev_write(struct cached_dev *dc, struct search *s)
 		s->iop.bypass = true;
 
 	if (should_writeback(dc, s->orig_bio,
-			     cache_mode(dc, bio),
+			     cache_mode(dc),
 			     s->iop.bypass)) {
 		s->iop.bypass = false;
 		s->iop.writeback = true;

drivers/md/bcache/super.c

@@ -53,12 +53,15 @@ LIST_HEAD(bch_cache_sets);
 static LIST_HEAD(uncached_devices);
 
 static int bcache_major;
-static DEFINE_IDA(bcache_minor);
+static DEFINE_IDA(bcache_device_idx);
 static wait_queue_head_t unregister_wait;
 struct workqueue_struct *bcache_wq;
 
 #define BTREE_MAX_PAGES		(256 * 1024 / PAGE_SIZE)
-#define BCACHE_MINORS		16 /* partition support */
+/* limitation of partitions number on single bcache device */
+#define BCACHE_MINORS		128
+/* limitation of bcache devices number on single system */
+#define BCACHE_DEVICE_IDX_MAX	((1U << MINORBITS)/BCACHE_MINORS)
 
 /* Superblock */
 
@@ -721,6 +724,16 @@ static void bcache_device_attach(struct bcache_device *d, struct cache_set *c,
 	closure_get(&c->caching);
 }
 
+static inline int first_minor_to_idx(int first_minor)
+{
+	return (first_minor/BCACHE_MINORS);
+}
+
+static inline int idx_to_first_minor(int idx)
+{
+	return (idx * BCACHE_MINORS);
+}
+
 static void bcache_device_free(struct bcache_device *d)
 {
 	lockdep_assert_held(&bch_register_lock);
@@ -734,7 +747,8 @@ static void bcache_device_free(struct bcache_device *d)
 	if (d->disk && d->disk->queue)
 		blk_cleanup_queue(d->disk->queue);
 	if (d->disk) {
-		ida_simple_remove(&bcache_minor, d->disk->first_minor);
+		ida_simple_remove(&bcache_device_idx,
+				  first_minor_to_idx(d->disk->first_minor));
 		put_disk(d->disk);
 	}
 
@@ -751,7 +765,7 @@ static int bcache_device_init(struct bcache_device *d, unsigned block_size,
 {
 	struct request_queue *q;
 	size_t n;
-	int minor;
+	int idx;
 
 	if (!d->stripe_size)
 		d->stripe_size = 1 << 31;
@@ -776,25 +790,24 @@ static int bcache_device_init(struct bcache_device *d, unsigned block_size,
 	if (!d->full_dirty_stripes)
 		return -ENOMEM;
 
-	minor = ida_simple_get(&bcache_minor, 0, MINORMASK + 1, GFP_KERNEL);
-	if (minor < 0)
-		return minor;
-
-	minor *= BCACHE_MINORS;
+	idx = ida_simple_get(&bcache_device_idx, 0,
+				BCACHE_DEVICE_IDX_MAX, GFP_KERNEL);
+	if (idx < 0)
+		return idx;
 
 	if (!(d->bio_split = bioset_create(4, offsetof(struct bbio, bio),
 					   BIOSET_NEED_BVECS |
 					   BIOSET_NEED_RESCUER)) ||
 	    !(d->disk = alloc_disk(BCACHE_MINORS))) {
-		ida_simple_remove(&bcache_minor, minor);
+		ida_simple_remove(&bcache_device_idx, idx);
 		return -ENOMEM;
 	}
 
 	set_capacity(d->disk, sectors);
-	snprintf(d->disk->disk_name, DISK_NAME_LEN, "bcache%i", minor);
+	snprintf(d->disk->disk_name, DISK_NAME_LEN, "bcache%i", idx);
 
 	d->disk->major		= bcache_major;
-	d->disk->first_minor	= minor;
+	d->disk->first_minor	= idx_to_first_minor(idx);
 	d->disk->fops		= &bcache_ops;
 	d->disk->private_data	= d;
 
@@ -889,7 +902,7 @@ static void cached_dev_detach_finish(struct work_struct *w)
 	closure_init_stack(&cl);
 
 	BUG_ON(!test_bit(BCACHE_DEV_DETACHING, &dc->disk.flags));
-	BUG_ON(atomic_read(&dc->count));
+	BUG_ON(refcount_read(&dc->count));
 
 	mutex_lock(&bch_register_lock);
 
@@ -1016,7 +1029,7 @@ int bch_cached_dev_attach(struct cached_dev *dc, struct cache_set *c)
 	 * dc->c must be set before dc->count != 0 - paired with the mb in
 	 * cached_dev_get()
 	 */
-	atomic_set(&dc->count, 1);
+	refcount_set(&dc->count, 1);
 
 	/* Block writeback thread, but spawn it */
 	down_write(&dc->writeback_lock);
@@ -1028,7 +1041,7 @@ int bch_cached_dev_attach(struct cached_dev *dc, struct cache_set *c)
 	if (BDEV_STATE(&dc->sb) == BDEV_STATE_DIRTY) {
 		bch_sectors_dirty_init(&dc->disk);
 		atomic_set(&dc->has_dirty, 1);
-		atomic_inc(&dc->count);
+		refcount_inc(&dc->count);
 		bch_writeback_queue(dc);
 	}
 
@@ -1129,9 +1142,6 @@ static int cached_dev_init(struct cached_dev *dc, unsigned block_size)
 	if (ret)
 		return ret;
 
-	set_capacity(dc->disk.disk,
-		     dc->bdev->bd_part->nr_sects - dc->sb.data_offset);
-
 	dc->disk.disk->queue->backing_dev_info->ra_pages =
 		max(dc->disk.disk->queue->backing_dev_info->ra_pages,
 		    q->backing_dev_info->ra_pages);
@@ -2085,6 +2095,7 @@ static void bcache_exit(void)
 	if (bcache_major)
 		unregister_blkdev(bcache_major, "bcache");
 	unregister_reboot_notifier(&reboot);
+	mutex_destroy(&bch_register_lock);
 }
 
 static int __init bcache_init(void)
@@ -2103,14 +2114,15 @@ static int __init bcache_init(void)
 	bcache_major = register_blkdev(0, "bcache");
 	if (bcache_major < 0) {
 		unregister_reboot_notifier(&reboot);
+		mutex_destroy(&bch_register_lock);
 		return bcache_major;
 	}
 
 	if (!(bcache_wq = alloc_workqueue("bcache", WQ_MEM_RECLAIM, 0)) ||
 	    !(bcache_kobj = kobject_create_and_add("bcache", fs_kobj)) ||
-	    sysfs_create_files(bcache_kobj, files) ||
 	    bch_request_init() ||
-	    bch_debug_init(bcache_kobj))
+	    bch_debug_init(bcache_kobj) ||
+	    sysfs_create_files(bcache_kobj, files))
 		goto err;
 
 	return 0;

drivers/md/bcache/sysfs.c

@@ -82,8 +82,9 @@ rw_attribute(writeback_delay);
 rw_attribute(writeback_rate);
 
 rw_attribute(writeback_rate_update_seconds);
-rw_attribute(writeback_rate_d_term);
+rw_attribute(writeback_rate_i_term_inverse);
 rw_attribute(writeback_rate_p_term_inverse);
+rw_attribute(writeback_rate_minimum);
 read_attribute(writeback_rate_debug);
 
 read_attribute(stripe_size);
@@ -131,15 +132,16 @@ SHOW(__bch_cached_dev)
 	sysfs_hprint(writeback_rate,	dc->writeback_rate.rate << 9);
 
 	var_print(writeback_rate_update_seconds);
-	var_print(writeback_rate_d_term);
+	var_print(writeback_rate_i_term_inverse);
 	var_print(writeback_rate_p_term_inverse);
+	var_print(writeback_rate_minimum);
 
 	if (attr == &sysfs_writeback_rate_debug) {
 		char rate[20];
 		char dirty[20];
 		char target[20];
 		char proportional[20];
-		char derivative[20];
+		char integral[20];
 		char change[20];
 		s64 next_io;
 
@@ -147,7 +149,7 @@ SHOW(__bch_cached_dev)
 		bch_hprint(dirty,	bcache_dev_sectors_dirty(&dc->disk) << 9);
 		bch_hprint(target,	dc->writeback_rate_target << 9);
 		bch_hprint(proportional,dc->writeback_rate_proportional << 9);
-		bch_hprint(derivative,	dc->writeback_rate_derivative << 9);
+		bch_hprint(integral,	dc->writeback_rate_integral_scaled << 9);
 		bch_hprint(change,	dc->writeback_rate_change << 9);
 
 		next_io = div64_s64(dc->writeback_rate.next - local_clock(),
@@ -158,11 +160,11 @@ SHOW(__bch_cached_dev)
 			       "dirty:\t\t%s\n"
 			       "target:\t\t%s\n"
 			       "proportional:\t%s\n"
-			       "derivative:\t%s\n"
+			       "integral:\t%s\n"
 			       "change:\t\t%s/sec\n"
 			       "next io:\t%llims\n",
 			       rate, dirty, target, proportional,
-			       derivative, change, next_io);
+			       integral, change, next_io);
 	}
 
 	sysfs_hprint(dirty_data,
@@ -214,7 +216,7 @@ STORE(__cached_dev)
 			    dc->writeback_rate.rate, 1, INT_MAX);
 
 	d_strtoul_nonzero(writeback_rate_update_seconds);
-	d_strtoul(writeback_rate_d_term);
+	d_strtoul(writeback_rate_i_term_inverse);
 	d_strtoul_nonzero(writeback_rate_p_term_inverse);
 
 	d_strtoi_h(sequential_cutoff);
@@ -320,7 +322,7 @@ static struct attribute *bch_cached_dev_files[] = {
 	&sysfs_writeback_percent,
 	&sysfs_writeback_rate,
 	&sysfs_writeback_rate_update_seconds,
-	&sysfs_writeback_rate_d_term,
+	&sysfs_writeback_rate_i_term_inverse,
 	&sysfs_writeback_rate_p_term_inverse,
 	&sysfs_writeback_rate_debug,
 	&sysfs_dirty_data,
@@ -746,6 +748,11 @@ static struct attribute *bch_cache_set_internal_files[] = {
 };
 KTYPE(bch_cache_set_internal);
 
+static int __bch_cache_cmp(const void *l, const void *r)
+{
+	return *((uint16_t *)r) - *((uint16_t *)l);
+}
+
 SHOW(__bch_cache)
 {
 	struct cache *ca = container_of(kobj, struct cache, kobj);
@@ -770,9 +777,6 @@ SHOW(__bch_cache)
 					       CACHE_REPLACEMENT(&ca->sb));
 
 	if (attr == &sysfs_priority_stats) {
-		int cmp(const void *l, const void *r)
-		{	return *((uint16_t *) r) - *((uint16_t *) l); }
-
 		struct bucket *b;
 		size_t n = ca->sb.nbuckets, i;
 		size_t unused = 0, available = 0, dirty = 0, meta = 0;
@@ -801,7 +805,7 @@ SHOW(__bch_cache)
 			p[i] = ca->buckets[i].prio;
 		mutex_unlock(&ca->set->bucket_lock);
 
-		sort(p, n, sizeof(uint16_t), cmp, NULL);
+		sort(p, n, sizeof(uint16_t), __bch_cache_cmp, NULL);
 
 		while (n &&
 		       !cached[n - 1])

drivers/md/bcache/util.c

@@ -232,8 +232,14 @@ uint64_t bch_next_delay(struct bch_ratelimit *d, uint64_t done)
 
 	d->next += div_u64(done * NSEC_PER_SEC, d->rate);
 
-	if (time_before64(now + NSEC_PER_SEC, d->next))
-		d->next = now + NSEC_PER_SEC;
+	/* Bound the time.  Don't let us fall further than 2 seconds behind
+	 * (this prevents unnecessary backlog that would make it impossible
+	 * to catch up).  If we're ahead of the desired writeback rate,
+	 * don't let us sleep more than 2.5 seconds (so we can notice/respond
+	 * if the control system tells us to speed up!).
+	 */
+	if (time_before64(now + NSEC_PER_SEC * 5LLU / 2LLU, d->next))
+		d->next = now + NSEC_PER_SEC * 5LLU / 2LLU;
 
 	if (time_after64(now - NSEC_PER_SEC * 2, d->next))
 		d->next = now - NSEC_PER_SEC * 2;

drivers/md/bcache/util.h

@@ -442,10 +442,10 @@ struct bch_ratelimit {
 	uint64_t		next;
 
 	/*
-	 * Rate at which we want to do work, in units per nanosecond
+	 * Rate at which we want to do work, in units per second
 	 * The units here correspond to the units passed to bch_next_delay()
 	 */
-	unsigned		rate;
+	uint32_t		rate;
 };
 
 static inline void bch_ratelimit_reset(struct bch_ratelimit *d)

drivers/md/bcache/writeback.c

@@ -26,48 +26,63 @@ static void __update_writeback_rate(struct cached_dev *dc)
 				bcache_flash_devs_sectors_dirty(c);
 	uint64_t cache_dirty_target =
 		div_u64(cache_sectors * dc->writeback_percent, 100);
-
 	int64_t target = div64_u64(cache_dirty_target * bdev_sectors(dc->bdev),
 				   c->cached_dev_sectors);
 
-	/* PD controller */
-
+	/*
+	 * PI controller:
+	 * Figures out the amount that should be written per second.
+	 *
+	 * First, the error (number of sectors that are dirty beyond our
+	 * target) is calculated.  The error is accumulated (numerically
+	 * integrated).
+	 *
+	 * Then, the proportional value and integral value are scaled
+	 * based on configured values.  These are stored as inverses to
+	 * avoid fixed point math and to make configuration easy-- e.g.
+	 * the default value of 40 for writeback_rate_p_term_inverse
+	 * attempts to write at a rate that would retire all the dirty
+	 * blocks in 40 seconds.
+	 *
+	 * The writeback_rate_i_inverse value of 10000 means that 1/10000th
+	 * of the error is accumulated in the integral term per second.
+	 * This acts as a slow, long-term average that is not subject to
+	 * variations in usage like the p term.
+	 */
 	int64_t dirty = bcache_dev_sectors_dirty(&dc->disk);
-	int64_t derivative = dirty - dc->disk.sectors_dirty_last;
-	int64_t proportional = dirty - target;
-	int64_t change;
-
-	dc->disk.sectors_dirty_last = dirty;
-
-	/* Scale to sectors per second */
-
-	proportional *= dc->writeback_rate_update_seconds;
-	proportional = div_s64(proportional, dc->writeback_rate_p_term_inverse);
-
-	derivative = div_s64(derivative, dc->writeback_rate_update_seconds);
-
-	derivative = ewma_add(dc->disk.sectors_dirty_derivative, derivative,
-			      (dc->writeback_rate_d_term /
-			       dc->writeback_rate_update_seconds) ?: 1, 0);
-
-	derivative *= dc->writeback_rate_d_term;
-	derivative = div_s64(derivative, dc->writeback_rate_p_term_inverse);
-
-	change = proportional + derivative;
+	int64_t error = dirty - target;
+	int64_t proportional_scaled =
+		div_s64(error, dc->writeback_rate_p_term_inverse);
+	int64_t integral_scaled;
+	uint32_t new_rate;
+
+	if ((error < 0 && dc->writeback_rate_integral > 0) ||
+	    (error > 0 && time_before64(local_clock(),
+			 dc->writeback_rate.next + NSEC_PER_MSEC))) {
+		/*
+		 * Only decrease the integral term if it's more than
+		 * zero.  Only increase the integral term if the device
+		 * is keeping up.  (Don't wind up the integral
+		 * ineffectively in either case).
+		 *
+		 * It's necessary to scale this by
+		 * writeback_rate_update_seconds to keep the integral
+		 * term dimensioned properly.
+		 */
+		dc->writeback_rate_integral += error *
+			dc->writeback_rate_update_seconds;
+	}
 
-	/* Don't increase writeback rate if the device isn't keeping up */
-	if (change > 0 &&
-	    time_after64(local_clock(),
-			 dc->writeback_rate.next + NSEC_PER_MSEC))
-		change = 0;
+	integral_scaled = div_s64(dc->writeback_rate_integral,
+			dc->writeback_rate_i_term_inverse);
 
-	dc->writeback_rate.rate =
-		clamp_t(int64_t, (int64_t) dc->writeback_rate.rate + change,
-			1, NSEC_PER_MSEC);
+	new_rate = clamp_t(int32_t, (proportional_scaled + integral_scaled),
+			dc->writeback_rate_minimum, NSEC_PER_SEC);
 
-	dc->writeback_rate_proportional = proportional;
-	dc->writeback_rate_derivative = derivative;
-	dc->writeback_rate_change = change;
+	dc->writeback_rate_proportional = proportional_scaled;
+	dc->writeback_rate_integral_scaled = integral_scaled;
+	dc->writeback_rate_change = new_rate - dc->writeback_rate.rate;
+	dc->writeback_rate.rate = new_rate;
 	dc->writeback_rate_target = target;
 }
 
@@ -180,13 +195,21 @@ static void write_dirty(struct closure *cl)
 	struct dirty_io *io = container_of(cl, struct dirty_io, cl);
 	struct keybuf_key *w = io->bio.bi_private;
 
-	dirty_init(w);
-	bio_set_op_attrs(&io->bio, REQ_OP_WRITE, 0);
-	io->bio.bi_iter.bi_sector = KEY_START(&w->key);
-	bio_set_dev(&io->bio, io->dc->bdev);
-	io->bio.bi_end_io	= dirty_endio;
+	/*
+	 * IO errors are signalled using the dirty bit on the key.
+	 * If we failed to read, we should not attempt to write to the
+	 * backing device.  Instead, immediately go to write_dirty_finish
+	 * to clean up.
+	 */
+	if (KEY_DIRTY(&w->key)) {
+		dirty_init(w);
+		bio_set_op_attrs(&io->bio, REQ_OP_WRITE, 0);
+		io->bio.bi_iter.bi_sector = KEY_START(&w->key);
+		bio_set_dev(&io->bio, io->dc->bdev);
+		io->bio.bi_end_io	= dirty_endio;
 
-	closure_bio_submit(&io->bio, cl);
+		closure_bio_submit(&io->bio, cl);
+	}
 
 	continue_at(cl, write_dirty_finish, io->dc->writeback_write_wq);
 }
@@ -418,6 +441,8 @@ static int bch_writeback_thread(void *arg)
 	struct cached_dev *dc = arg;
 	bool searched_full_index;
 
+	bch_ratelimit_reset(&dc->writeback_rate);
+
 	while (!kthread_should_stop()) {
 		down_write(&dc->writeback_lock);
 		if (!atomic_read(&dc->has_dirty) ||
@@ -445,7 +470,6 @@ static int bch_writeback_thread(void *arg)
 
 		up_write(&dc->writeback_lock);
 
-		bch_ratelimit_reset(&dc->writeback_rate);
 		read_dirty(dc);
 
 		if (searched_full_index) {
@@ -455,6 +479,8 @@ static int bch_writeback_thread(void *arg)
 			       !kthread_should_stop() &&
 			       !test_bit(BCACHE_DEV_DETACHING, &dc->disk.flags))
 				delay = schedule_timeout_interruptible(delay);
+
+			bch_ratelimit_reset(&dc->writeback_rate);
 		}
 	}
 
@@ -492,8 +518,6 @@ void bch_sectors_dirty_init(struct bcache_device *d)
 
 	bch_btree_map_keys(&op.op, d->c, &KEY(op.inode, 0, 0),
 			   sectors_dirty_init_fn, 0);
-
-	d->sectors_dirty_last = bcache_dev_sectors_dirty(d);
 }
 
 void bch_cached_dev_writeback_init(struct cached_dev *dc)
@@ -507,10 +531,11 @@ void bch_cached_dev_writeback_init(struct cached_dev *dc)
 	dc->writeback_percent		= 10;
 	dc->writeback_delay		= 30;
 	dc->writeback_rate.rate		= 1024;
+	dc->writeback_rate_minimum	= 8;
 
 	dc->writeback_rate_update_seconds = 5;
-	dc->writeback_rate_d_term	= 30;
-	dc->writeback_rate_p_term_inverse = 6000;
+	dc->writeback_rate_p_term_inverse = 40;
+	dc->writeback_rate_i_term_inverse = 10000;
 
 	INIT_DELAYED_WORK(&dc->writeback_rate_update, update_writeback_rate);
 }

drivers/md/bcache/writeback.h

@@ -77,7 +77,9 @@ static inline bool should_writeback(struct cached_dev *dc, struct bio *bio,
 	if (would_skip)
 		return false;
 
-	return op_is_sync(bio->bi_opf) || in_use <= CUTOFF_WRITEBACK;
+	return (op_is_sync(bio->bi_opf) ||
+		bio->bi_opf & (REQ_META|REQ_PRIO) ||
+		in_use <= CUTOFF_WRITEBACK);
 }
 
 static inline void bch_writeback_queue(struct cached_dev *dc)
@@ -90,7 +92,7 @@ static inline void bch_writeback_add(struct cached_dev *dc)
 {
 	if (!atomic_read(&dc->has_dirty) &&
 	    !atomic_xchg(&dc->has_dirty, 1)) {
-		atomic_inc(&dc->count);
+		refcount_inc(&dc->count);
 
 		if (BDEV_STATE(&dc->sb) != BDEV_STATE_DIRTY) {
 			SET_BDEV_STATE(&dc->sb, BDEV_STATE_DIRTY);

drivers/md/bitmap.c

@@ -368,7 +368,7 @@ static int read_page(struct file *file, unsigned long index,
 	pr_debug("read bitmap file (%dB @ %llu)\n", (int)PAGE_SIZE,
 		 (unsigned long long)index << PAGE_SHIFT);
 
-	bh = alloc_page_buffers(page, 1<<inode->i_blkbits, 0);
+	bh = alloc_page_buffers(page, 1<<inode->i_blkbits, false);
 	if (!bh) {
 		ret = -ENOMEM;
 		goto out;

drivers/md/dm-rq.c

@@ -56,7 +56,7 @@ static unsigned dm_get_blk_mq_queue_depth(void)
 
 int dm_request_based(struct mapped_device *md)
 {
-	return blk_queue_stackable(md->queue);
+	return queue_is_rq_based(md->queue);
 }
 
 static void dm_old_start_queue(struct request_queue *q)

drivers/md/dm-table.c

@@ -1000,7 +1000,7 @@ verify_rq_based:
 	list_for_each_entry(dd, devices, list) {
 		struct request_queue *q = bdev_get_queue(dd->dm_dev->bdev);
 
-		if (!blk_queue_stackable(q)) {
+		if (!queue_is_rq_based(q)) {
 			DMERR("table load rejected: including"
 			      " non-request-stackable devices");
 			return -EINVAL;
@@ -1847,19 +1847,6 @@ void dm_table_set_restrictions(struct dm_table *t, struct request_queue *q,
 	 */
 	if (blk_queue_add_random(q) && dm_table_all_devices_attribute(t, device_is_not_random))
 		queue_flag_clear_unlocked(QUEUE_FLAG_ADD_RANDOM, q);
-
-	/*
-	 * QUEUE_FLAG_STACKABLE must be set after all queue settings are
-	 * visible to other CPUs because, once the flag is set, incoming bios
-	 * are processed by request-based dm, which refers to the queue
-	 * settings.
-	 * Until the flag set, bios are passed to bio-based dm and queued to
-	 * md->deferred where queue settings are not needed yet.
-	 * Those bios are passed to request-based dm at the resume time.
-	 */
-	smp_mb();
-	if (dm_table_request_based(t))
-		queue_flag_set_unlocked(QUEUE_FLAG_STACKABLE, q);
 }
 
 unsigned int dm_table_get_num_targets(struct dm_table *t)

drivers/md/dm.c

@@ -1618,17 +1618,6 @@ static void dm_wq_work(struct work_struct *work);
 
 void dm_init_md_queue(struct mapped_device *md)
 {
-	/*
-	 * Request-based dm devices cannot be stacked on top of bio-based dm
-	 * devices.  The type of this dm device may not have been decided yet.
-	 * The type is decided at the first table loading time.
-	 * To prevent problematic device stacking, clear the queue flag
-	 * for request stacking support until then.
-	 *
-	 * This queue is new, so no concurrency on the queue_flags.
-	 */
-	queue_flag_clear_unlocked(QUEUE_FLAG_STACKABLE, md->queue);
-
 	/*
 	 * Initialize data that will only be used by a non-blk-mq DM queue
 	 * - must do so here (in alloc_dev callchain) before queue is used

drivers/nvme/Kconfig

@@ -1,2 +1,6 @@
+menu "NVME Support"
+
 source "drivers/nvme/host/Kconfig"
 source "drivers/nvme/target/Kconfig"
+
+endmenu

drivers/nvme/host/Kconfig

@@ -13,6 +13,15 @@ config BLK_DEV_NVME
 	  To compile this driver as a module, choose M here: the
 	  module will be called nvme.
 
+config NVME_MULTIPATH
+	bool "NVMe multipath support"
+	depends on NVME_CORE
+	---help---
+	   This option enables support for multipath access to NVMe
+	   subsystems.  If this option is enabled only a single
+	   /dev/nvmeXnY device will show up for each NVMe namespaces,
+	   even if it is accessible through multiple controllers.
+
 config NVME_FABRICS
 	tristate
 

drivers/nvme/host/Makefile

@@ -6,6 +6,7 @@ obj-$(CONFIG_NVME_RDMA)			+= nvme-rdma.o
 obj-$(CONFIG_NVME_FC)			+= nvme-fc.o
 
 nvme-core-y				:= core.o
+nvme-core-$(CONFIG_NVME_MULTIPATH)	+= multipath.o
 nvme-core-$(CONFIG_NVM)			+= lightnvm.o
 
 nvme-y					+= pci.o

drivers/nvme/host/core.c

@@ -34,13 +34,13 @@
 
 #define NVME_MINORS		(1U << MINORBITS)
 
-unsigned char admin_timeout = 60;
-module_param(admin_timeout, byte, 0644);
+unsigned int admin_timeout = 60;
+module_param(admin_timeout, uint, 0644);
 MODULE_PARM_DESC(admin_timeout, "timeout in seconds for admin commands");
 EXPORT_SYMBOL_GPL(admin_timeout);
 
-unsigned char nvme_io_timeout = 30;
-module_param_named(io_timeout, nvme_io_timeout, byte, 0644);
+unsigned int nvme_io_timeout = 30;
+module_param_named(io_timeout, nvme_io_timeout, uint, 0644);
 MODULE_PARM_DESC(io_timeout, "timeout in seconds for I/O");
 EXPORT_SYMBOL_GPL(nvme_io_timeout);
 
@@ -52,9 +52,6 @@ static u8 nvme_max_retries = 5;
 module_param_named(max_retries, nvme_max_retries, byte, 0644);
 MODULE_PARM_DESC(max_retries, "max number of retries a command may have");
 
-static int nvme_char_major;
-module_param(nvme_char_major, int, 0);
-
 static unsigned long default_ps_max_latency_us = 100000;
 module_param(default_ps_max_latency_us, ulong, 0644);
 MODULE_PARM_DESC(default_ps_max_latency_us,
@@ -71,10 +68,17 @@ MODULE_PARM_DESC(streams, "turn on support for Streams write directives");
 struct workqueue_struct *nvme_wq;
 EXPORT_SYMBOL_GPL(nvme_wq);
 
-static LIST_HEAD(nvme_ctrl_list);
-static DEFINE_SPINLOCK(dev_list_lock);
+static DEFINE_IDA(nvme_subsystems_ida);
+static LIST_HEAD(nvme_subsystems);
+static DEFINE_MUTEX(nvme_subsystems_lock);
 
+static DEFINE_IDA(nvme_instance_ida);
+static dev_t nvme_chr_devt;
 static struct class *nvme_class;
+static struct class *nvme_subsys_class;
+
+static void nvme_ns_remove(struct nvme_ns *ns);
+static int nvme_revalidate_disk(struct gendisk *disk);
 
 static __le32 nvme_get_log_dw10(u8 lid, size_t size)
 {
@@ -101,6 +105,51 @@ static int nvme_reset_ctrl_sync(struct nvme_ctrl *ctrl)
 	return ret;
 }
 
+static void nvme_delete_ctrl_work(struct work_struct *work)
+{
+	struct nvme_ctrl *ctrl =
+		container_of(work, struct nvme_ctrl, delete_work);
+
+	flush_work(&ctrl->reset_work);
+	nvme_stop_ctrl(ctrl);
+	nvme_remove_namespaces(ctrl);
+	ctrl->ops->delete_ctrl(ctrl);
+	nvme_uninit_ctrl(ctrl);
+	nvme_put_ctrl(ctrl);
+}
+
+int nvme_delete_ctrl(struct nvme_ctrl *ctrl)
+{
+	if (!nvme_change_ctrl_state(ctrl, NVME_CTRL_DELETING))
+		return -EBUSY;
+	if (!queue_work(nvme_wq, &ctrl->delete_work))
+		return -EBUSY;
+	return 0;
+}
+EXPORT_SYMBOL_GPL(nvme_delete_ctrl);
+
+int nvme_delete_ctrl_sync(struct nvme_ctrl *ctrl)
+{
+	int ret = 0;
+
+	/*
+	 * Keep a reference until the work is flushed since ->delete_ctrl
+	 * can free the controller.
+	 */
+	nvme_get_ctrl(ctrl);
+	ret = nvme_delete_ctrl(ctrl);
+	if (!ret)
+		flush_work(&ctrl->delete_work);
+	nvme_put_ctrl(ctrl);
+	return ret;
+}
+EXPORT_SYMBOL_GPL(nvme_delete_ctrl_sync);
+
+static inline bool nvme_ns_has_pi(struct nvme_ns *ns)
+{
+	return ns->pi_type && ns->ms == sizeof(struct t10_pi_tuple);
+}
+
 static blk_status_t nvme_error_status(struct request *req)
 {
 	switch (nvme_req(req)->status & 0x7ff) {
@@ -142,9 +191,16 @@ static inline bool nvme_req_needs_retry(struct request *req)
 void nvme_complete_rq(struct request *req)
 {
 	if (unlikely(nvme_req(req)->status && nvme_req_needs_retry(req))) {
-		nvme_req(req)->retries++;
-		blk_mq_requeue_request(req, true);
-		return;
+		if (nvme_req_needs_failover(req)) {
+			nvme_failover_req(req);
+			return;
+		}
+
+		if (!blk_queue_dying(req->q)) {
+			nvme_req(req)->retries++;
+			blk_mq_requeue_request(req, true);
+			return;
+		}
 	}
 
 	blk_mq_end_request(req, nvme_error_status(req));
@@ -153,18 +209,13 @@ EXPORT_SYMBOL_GPL(nvme_complete_rq);
 
 void nvme_cancel_request(struct request *req, void *data, bool reserved)
 {
-	int status;
-
 	if (!blk_mq_request_started(req))
 		return;
 
 	dev_dbg_ratelimited(((struct nvme_ctrl *) data)->device,
 				"Cancelling I/O %d", req->tag);
 
-	status = NVME_SC_ABORT_REQ;
-	if (blk_queue_dying(req->q))
-		status |= NVME_SC_DNR;
-	nvme_req(req)->status = status;
+	nvme_req(req)->status = NVME_SC_ABORT_REQ;
 	blk_mq_complete_request(req);
 
 }
@@ -205,6 +256,7 @@ bool nvme_change_ctrl_state(struct nvme_ctrl *ctrl,
 	case NVME_CTRL_RECONNECTING:
 		switch (old_state) {
 		case NVME_CTRL_LIVE:
+		case NVME_CTRL_RESETTING:
 			changed = true;
 			/* FALLTHRU */
 		default:
@@ -239,11 +291,29 @@ bool nvme_change_ctrl_state(struct nvme_ctrl *ctrl,
 		ctrl->state = new_state;
 
 	spin_unlock_irqrestore(&ctrl->lock, flags);
-
+	if (changed && ctrl->state == NVME_CTRL_LIVE)
+		nvme_kick_requeue_lists(ctrl);
 	return changed;
 }
 EXPORT_SYMBOL_GPL(nvme_change_ctrl_state);
 
+static void nvme_free_ns_head(struct kref *ref)
+{
+	struct nvme_ns_head *head =
+		container_of(ref, struct nvme_ns_head, ref);
+
+	nvme_mpath_remove_disk(head);
+	ida_simple_remove(&head->subsys->ns_ida, head->instance);
+	list_del_init(&head->entry);
+	cleanup_srcu_struct(&head->srcu);
+	kfree(head);
+}
+
+static void nvme_put_ns_head(struct nvme_ns_head *head)
+{
+	kref_put(&head->ref, nvme_free_ns_head);
+}
+
 static void nvme_free_ns(struct kref *kref)
 {
 	struct nvme_ns *ns = container_of(kref, struct nvme_ns, kref);
@@ -251,14 +321,8 @@ static void nvme_free_ns(struct kref *kref)
 	if (ns->ndev)
 		nvme_nvm_unregister(ns);
 
-	if (ns->disk) {
-		spin_lock(&dev_list_lock);
-		ns->disk->private_data = NULL;
-		spin_unlock(&dev_list_lock);
-	}
-
 	put_disk(ns->disk);
-	ida_simple_remove(&ns->ctrl->ns_ida, ns->instance);
+	nvme_put_ns_head(ns->head);
 	nvme_put_ctrl(ns->ctrl);
 	kfree(ns);
 }
@@ -268,31 +332,8 @@ static void nvme_put_ns(struct nvme_ns *ns)
 	kref_put(&ns->kref, nvme_free_ns);
 }
 
-static struct nvme_ns *nvme_get_ns_from_disk(struct gendisk *disk)
-{
-	struct nvme_ns *ns;
-
-	spin_lock(&dev_list_lock);
-	ns = disk->private_data;
-	if (ns) {
-		if (!kref_get_unless_zero(&ns->kref))
-			goto fail;
-		if (!try_module_get(ns->ctrl->ops->module))
-			goto fail_put_ns;
-	}
-	spin_unlock(&dev_list_lock);
-
-	return ns;
-
-fail_put_ns:
-	kref_put(&ns->kref, nvme_free_ns);
-fail:
-	spin_unlock(&dev_list_lock);
-	return NULL;
-}
-
 struct request *nvme_alloc_request(struct request_queue *q,
-		struct nvme_command *cmd, unsigned int flags, int qid)
+		struct nvme_command *cmd, blk_mq_req_flags_t flags, int qid)
 {
 	unsigned op = nvme_is_write(cmd) ? REQ_OP_DRV_OUT : REQ_OP_DRV_IN;
 	struct request *req;
@@ -417,7 +458,7 @@ static inline void nvme_setup_flush(struct nvme_ns *ns,
 {
 	memset(cmnd, 0, sizeof(*cmnd));
 	cmnd->common.opcode = nvme_cmd_flush;
-	cmnd->common.nsid = cpu_to_le32(ns->ns_id);
+	cmnd->common.nsid = cpu_to_le32(ns->head->ns_id);
 }
 
 static blk_status_t nvme_setup_discard(struct nvme_ns *ns, struct request *req,
@@ -448,7 +489,7 @@ static blk_status_t nvme_setup_discard(struct nvme_ns *ns, struct request *req,
 
 	memset(cmnd, 0, sizeof(*cmnd));
 	cmnd->dsm.opcode = nvme_cmd_dsm;
-	cmnd->dsm.nsid = cpu_to_le32(ns->ns_id);
+	cmnd->dsm.nsid = cpu_to_le32(ns->head->ns_id);
 	cmnd->dsm.nr = cpu_to_le32(segments - 1);
 	cmnd->dsm.attributes = cpu_to_le32(NVME_DSMGMT_AD);
 
@@ -467,16 +508,6 @@ static inline blk_status_t nvme_setup_rw(struct nvme_ns *ns,
 	u16 control = 0;
 	u32 dsmgmt = 0;
 
-	/*
-	 * If formated with metadata, require the block layer provide a buffer
-	 * unless this namespace is formated such that the metadata can be
-	 * stripped/generated by the controller with PRACT=1.
-	 */
-	if (ns && ns->ms &&
-	    (!ns->pi_type || ns->ms != sizeof(struct t10_pi_tuple)) &&
-	    !blk_integrity_rq(req) && !blk_rq_is_passthrough(req))
-		return BLK_STS_NOTSUPP;
-
 	if (req->cmd_flags & REQ_FUA)
 		control |= NVME_RW_FUA;
 	if (req->cmd_flags & (REQ_FAILFAST_DEV | REQ_RAHEAD))
@@ -487,7 +518,7 @@ static inline blk_status_t nvme_setup_rw(struct nvme_ns *ns,
 
 	memset(cmnd, 0, sizeof(*cmnd));
 	cmnd->rw.opcode = (rq_data_dir(req) ? nvme_cmd_write : nvme_cmd_read);
-	cmnd->rw.nsid = cpu_to_le32(ns->ns_id);
+	cmnd->rw.nsid = cpu_to_le32(ns->head->ns_id);
 	cmnd->rw.slba = cpu_to_le64(nvme_block_nr(ns, blk_rq_pos(req)));
 	cmnd->rw.length = cpu_to_le16((blk_rq_bytes(req) >> ns->lba_shift) - 1);
 
@@ -495,6 +526,18 @@ static inline blk_status_t nvme_setup_rw(struct nvme_ns *ns,
 		nvme_assign_write_stream(ctrl, req, &control, &dsmgmt);
 
 	if (ns->ms) {
+		/*
+		 * If formated with metadata, the block layer always provides a
+		 * metadata buffer if CONFIG_BLK_DEV_INTEGRITY is enabled.  Else
+		 * we enable the PRACT bit for protection information or set the
+		 * namespace capacity to zero to prevent any I/O.
+		 */
+		if (!blk_integrity_rq(req)) {
+			if (WARN_ON_ONCE(!nvme_ns_has_pi(ns)))
+				return BLK_STS_NOTSUPP;
+			control |= NVME_RW_PRINFO_PRACT;
+		}
+
 		switch (ns->pi_type) {
 		case NVME_NS_DPS_PI_TYPE3:
 			control |= NVME_RW_PRINFO_PRCHK_GUARD;
@@ -507,8 +550,6 @@ static inline blk_status_t nvme_setup_rw(struct nvme_ns *ns,
 					nvme_block_nr(ns, blk_rq_pos(req)));
 			break;
 		}
-		if (!blk_integrity_rq(req))
-			control |= NVME_RW_PRINFO_PRACT;
 	}
 
 	cmnd->rw.control = cpu_to_le16(control);
@@ -560,7 +601,8 @@ EXPORT_SYMBOL_GPL(nvme_setup_cmd);
  */
 int __nvme_submit_sync_cmd(struct request_queue *q, struct nvme_command *cmd,
 		union nvme_result *result, void *buffer, unsigned bufflen,
-		unsigned timeout, int qid, int at_head, int flags)
+		unsigned timeout, int qid, int at_head,
+		blk_mq_req_flags_t flags)
 {
 	struct request *req;
 	int ret;
@@ -778,7 +820,7 @@ static int nvme_identify_ctrl(struct nvme_ctrl *dev, struct nvme_id_ctrl **id)
 }
 
 static int nvme_identify_ns_descs(struct nvme_ctrl *ctrl, unsigned nsid,
-		u8 *eui64, u8 *nguid, uuid_t *uuid)
+		struct nvme_ns_ids *ids)
 {
 	struct nvme_command c = { };
 	int status;
@@ -814,7 +856,7 @@ static int nvme_identify_ns_descs(struct nvme_ctrl *ctrl, unsigned nsid,
 				goto free_data;
 			}
 			len = NVME_NIDT_EUI64_LEN;
-			memcpy(eui64, data + pos + sizeof(*cur), len);
+			memcpy(ids->eui64, data + pos + sizeof(*cur), len);
 			break;
 		case NVME_NIDT_NGUID:
 			if (cur->nidl != NVME_NIDT_NGUID_LEN) {
@@ -824,7 +866,7 @@ static int nvme_identify_ns_descs(struct nvme_ctrl *ctrl, unsigned nsid,
 				goto free_data;
 			}
 			len = NVME_NIDT_NGUID_LEN;
-			memcpy(nguid, data + pos + sizeof(*cur), len);
+			memcpy(ids->nguid, data + pos + sizeof(*cur), len);
 			break;
 		case NVME_NIDT_UUID:
 			if (cur->nidl != NVME_NIDT_UUID_LEN) {
@@ -834,7 +876,7 @@ static int nvme_identify_ns_descs(struct nvme_ctrl *ctrl, unsigned nsid,
 				goto free_data;
 			}
 			len = NVME_NIDT_UUID_LEN;
-			uuid_copy(uuid, data + pos + sizeof(*cur));
+			uuid_copy(&ids->uuid, data + pos + sizeof(*cur));
 			break;
 		default:
 			/* Skip unnkown types */
@@ -968,7 +1010,7 @@ static int nvme_submit_io(struct nvme_ns *ns, struct nvme_user_io __user *uio)
 	memset(&c, 0, sizeof(c));
 	c.rw.opcode = io.opcode;
 	c.rw.flags = io.flags;
-	c.rw.nsid = cpu_to_le32(ns->ns_id);
+	c.rw.nsid = cpu_to_le32(ns->head->ns_id);
 	c.rw.slba = cpu_to_le64(io.slba);
 	c.rw.length = cpu_to_le16(io.nblocks);
 	c.rw.control = cpu_to_le16(io.control);
@@ -982,12 +1024,87 @@ static int nvme_submit_io(struct nvme_ns *ns, struct nvme_user_io __user *uio)
 			metadata, meta_len, io.slba, NULL, 0);
 }
 
+static u32 nvme_known_admin_effects(u8 opcode)
+{
+	switch (opcode) {
+	case nvme_admin_format_nvm:
+		return NVME_CMD_EFFECTS_CSUPP | NVME_CMD_EFFECTS_LBCC |
+					NVME_CMD_EFFECTS_CSE_MASK;
+	case nvme_admin_sanitize_nvm:
+		return NVME_CMD_EFFECTS_CSE_MASK;
+	default:
+		break;
+	}
+	return 0;
+}
+
+static u32 nvme_passthru_start(struct nvme_ctrl *ctrl, struct nvme_ns *ns,
+								u8 opcode)
+{
+	u32 effects = 0;
+
+	if (ns) {
+		if (ctrl->effects)
+			effects = le32_to_cpu(ctrl->effects->iocs[opcode]);
+		if (effects & ~NVME_CMD_EFFECTS_CSUPP)
+			dev_warn(ctrl->device,
+				 "IO command:%02x has unhandled effects:%08x\n",
+				 opcode, effects);
+		return 0;
+	}
+
+	if (ctrl->effects)
+		effects = le32_to_cpu(ctrl->effects->iocs[opcode]);
+	else
+		effects = nvme_known_admin_effects(opcode);
+
+	/*
+	 * For simplicity, IO to all namespaces is quiesced even if the command
+	 * effects say only one namespace is affected.
+	 */
+	if (effects & (NVME_CMD_EFFECTS_LBCC | NVME_CMD_EFFECTS_CSE_MASK)) {
+		nvme_start_freeze(ctrl);
+		nvme_wait_freeze(ctrl);
+	}
+	return effects;
+}
+
+static void nvme_update_formats(struct nvme_ctrl *ctrl)
+{
+	struct nvme_ns *ns;
+
+	mutex_lock(&ctrl->namespaces_mutex);
+	list_for_each_entry(ns, &ctrl->namespaces, list) {
+		if (ns->disk && nvme_revalidate_disk(ns->disk))
+			nvme_ns_remove(ns);
+	}
+	mutex_unlock(&ctrl->namespaces_mutex);
+}
+
+static void nvme_passthru_end(struct nvme_ctrl *ctrl, u32 effects)
+{
+	/*
+	 * Revalidate LBA changes prior to unfreezing. This is necessary to
+	 * prevent memory corruption if a logical block size was changed by
+	 * this command.
+	 */
+	if (effects & NVME_CMD_EFFECTS_LBCC)
+		nvme_update_formats(ctrl);
+	if (effects & (NVME_CMD_EFFECTS_LBCC | NVME_CMD_EFFECTS_CSE_MASK))
+		nvme_unfreeze(ctrl);
+	if (effects & NVME_CMD_EFFECTS_CCC)
+		nvme_init_identify(ctrl);
+	if (effects & (NVME_CMD_EFFECTS_NIC | NVME_CMD_EFFECTS_NCC))
+		nvme_queue_scan(ctrl);
+}
+
 static int nvme_user_cmd(struct nvme_ctrl *ctrl, struct nvme_ns *ns,
 			struct nvme_passthru_cmd __user *ucmd)
 {
 	struct nvme_passthru_cmd cmd;
 	struct nvme_command c;
 	unsigned timeout = 0;
+	u32 effects;
 	int status;
 
 	if (!capable(CAP_SYS_ADMIN))
@@ -1013,10 +1130,13 @@ static int nvme_user_cmd(struct nvme_ctrl *ctrl, struct nvme_ns *ns,
 	if (cmd.timeout_ms)
 		timeout = msecs_to_jiffies(cmd.timeout_ms);
 
+	effects = nvme_passthru_start(ctrl, ns, cmd.opcode);
 	status = nvme_submit_user_cmd(ns ? ns->queue : ctrl->admin_q, &c,
 			(void __user *)(uintptr_t)cmd.addr, cmd.data_len,
 			(void __user *)(uintptr_t)cmd.metadata, cmd.metadata,
 			0, &cmd.result, timeout);
+	nvme_passthru_end(ctrl, effects);
+
 	if (status >= 0) {
 		if (put_user(cmd.result, &ucmd->result))
 			return -EFAULT;
@@ -1025,15 +1145,37 @@ static int nvme_user_cmd(struct nvme_ctrl *ctrl, struct nvme_ns *ns,
 	return status;
 }
 
-static int nvme_ioctl(struct block_device *bdev, fmode_t mode,
-		unsigned int cmd, unsigned long arg)
+/*
+ * Issue ioctl requests on the first available path.  Note that unlike normal
+ * block layer requests we will not retry failed request on another controller.
+ */
+static struct nvme_ns *nvme_get_ns_from_disk(struct gendisk *disk,
+		struct nvme_ns_head **head, int *srcu_idx)
 {
-	struct nvme_ns *ns = bdev->bd_disk->private_data;
+#ifdef CONFIG_NVME_MULTIPATH
+	if (disk->fops == &nvme_ns_head_ops) {
+		*head = disk->private_data;
+		*srcu_idx = srcu_read_lock(&(*head)->srcu);
+		return nvme_find_path(*head);
+	}
+#endif
+	*head = NULL;
+	*srcu_idx = -1;
+	return disk->private_data;
+}
 
+static void nvme_put_ns_from_disk(struct nvme_ns_head *head, int idx)
+{
+	if (head)
+		srcu_read_unlock(&head->srcu, idx);
+}
+
+static int nvme_ns_ioctl(struct nvme_ns *ns, unsigned cmd, unsigned long arg)
+{
 	switch (cmd) {
 	case NVME_IOCTL_ID:
 		force_successful_syscall_return();
-		return ns->ns_id;
+		return ns->head->ns_id;
 	case NVME_IOCTL_ADMIN_CMD:
 		return nvme_user_cmd(ns->ctrl, NULL, (void __user *)arg);
 	case NVME_IOCTL_IO_CMD:
@@ -1052,27 +1194,39 @@ static int nvme_ioctl(struct block_device *bdev, fmode_t mode,
 	}
 }
 
-#ifdef CONFIG_COMPAT
-static int nvme_compat_ioctl(struct block_device *bdev, fmode_t mode,
-			unsigned int cmd, unsigned long arg)
+static int nvme_ioctl(struct block_device *bdev, fmode_t mode,
+		unsigned int cmd, unsigned long arg)
 {
-	return nvme_ioctl(bdev, mode, cmd, arg);
+	struct nvme_ns_head *head = NULL;
+	struct nvme_ns *ns;
+	int srcu_idx, ret;
+
+	ns = nvme_get_ns_from_disk(bdev->bd_disk, &head, &srcu_idx);
+	if (unlikely(!ns))
+		ret = -EWOULDBLOCK;
+	else
+		ret = nvme_ns_ioctl(ns, cmd, arg);
+	nvme_put_ns_from_disk(head, srcu_idx);
+	return ret;
 }
-#else
-#define nvme_compat_ioctl	NULL
-#endif
 
 static int nvme_open(struct block_device *bdev, fmode_t mode)
 {
-	return nvme_get_ns_from_disk(bdev->bd_disk) ? 0 : -ENXIO;
+	struct nvme_ns *ns = bdev->bd_disk->private_data;
+
+#ifdef CONFIG_NVME_MULTIPATH
+	/* should never be called due to GENHD_FL_HIDDEN */
+	if (WARN_ON_ONCE(ns->head->disk))
+		return -ENXIO;
+#endif
+	if (!kref_get_unless_zero(&ns->kref))
+		return -ENXIO;
+	return 0;
 }
 
 static void nvme_release(struct gendisk *disk, fmode_t mode)
 {
-	struct nvme_ns *ns = disk->private_data;
-
-	module_put(ns->ctrl->ops->module);
-	nvme_put_ns(ns);
+	nvme_put_ns(disk->private_data);
 }
 
 static int nvme_getgeo(struct block_device *bdev, struct hd_geometry *geo)
@@ -1085,35 +1239,12 @@ static int nvme_getgeo(struct block_device *bdev, struct hd_geometry *geo)
 }
 
 #ifdef CONFIG_BLK_DEV_INTEGRITY
-static void nvme_prep_integrity(struct gendisk *disk, struct nvme_id_ns *id,
-		u16 bs)
-{
-	struct nvme_ns *ns = disk->private_data;
-	u16 old_ms = ns->ms;
-	u8 pi_type = 0;
-
-	ns->ms = le16_to_cpu(id->lbaf[id->flbas & NVME_NS_FLBAS_LBA_MASK].ms);
-	ns->ext = ns->ms && (id->flbas & NVME_NS_FLBAS_META_EXT);
-
-	/* PI implementation requires metadata equal t10 pi tuple size */
-	if (ns->ms == sizeof(struct t10_pi_tuple))
-		pi_type = id->dps & NVME_NS_DPS_PI_MASK;
-
-	if (blk_get_integrity(disk) &&
-	    (ns->pi_type != pi_type || ns->ms != old_ms ||
-	     bs != queue_logical_block_size(disk->queue) ||
-	     (ns->ms && ns->ext)))
-		blk_integrity_unregister(disk);
-
-	ns->pi_type = pi_type;
-}
-
-static void nvme_init_integrity(struct nvme_ns *ns)
+static void nvme_init_integrity(struct gendisk *disk, u16 ms, u8 pi_type)
 {
 	struct blk_integrity integrity;
 
 	memset(&integrity, 0, sizeof(integrity));
-	switch (ns->pi_type) {
+	switch (pi_type) {
 	case NVME_NS_DPS_PI_TYPE3:
 		integrity.profile = &t10_pi_type3_crc;
 		integrity.tag_size = sizeof(u16) + sizeof(u32);
@@ -1129,16 +1260,12 @@ static void nvme_init_integrity(struct nvme_ns *ns)
 		integrity.profile = NULL;
 		break;
 	}
-	integrity.tuple_size = ns->ms;
-	blk_integrity_register(ns->disk, &integrity);
-	blk_queue_max_integrity_segments(ns->queue, 1);
+	integrity.tuple_size = ms;
+	blk_integrity_register(disk, &integrity);
+	blk_queue_max_integrity_segments(disk->queue, 1);
 }
 #else
-static void nvme_prep_integrity(struct gendisk *disk, struct nvme_id_ns *id,
-		u16 bs)
-{
-}
-static void nvme_init_integrity(struct nvme_ns *ns)
+static void nvme_init_integrity(struct gendisk *disk, u16 ms, u8 pi_type)
 {
 }
 #endif /* CONFIG_BLK_DEV_INTEGRITY */
@@ -1149,53 +1276,89 @@ static void nvme_set_chunk_size(struct nvme_ns *ns)
 	blk_queue_chunk_sectors(ns->queue, rounddown_pow_of_two(chunk_size));
 }
 
-static void nvme_config_discard(struct nvme_ns *ns)
+static void nvme_config_discard(struct nvme_ctrl *ctrl,
+		unsigned stream_alignment, struct request_queue *queue)
 {
-	struct nvme_ctrl *ctrl = ns->ctrl;
-	u32 logical_block_size = queue_logical_block_size(ns->queue);
+	u32 size = queue_logical_block_size(queue);
+
+	if (stream_alignment)
+		size *= stream_alignment;
 
 	BUILD_BUG_ON(PAGE_SIZE / sizeof(struct nvme_dsm_range) <
 			NVME_DSM_MAX_RANGES);
 
-	if (ctrl->nr_streams && ns->sws && ns->sgs) {
-		unsigned int sz = logical_block_size * ns->sws * ns->sgs;
+	queue->limits.discard_alignment = size;
+	queue->limits.discard_granularity = size;
 
-		ns->queue->limits.discard_alignment = sz;
-		ns->queue->limits.discard_granularity = sz;
-	} else {
-		ns->queue->limits.discard_alignment = logical_block_size;
-		ns->queue->limits.discard_granularity = logical_block_size;
-	}
-	blk_queue_max_discard_sectors(ns->queue, UINT_MAX);
-	blk_queue_max_discard_segments(ns->queue, NVME_DSM_MAX_RANGES);
-	queue_flag_set_unlocked(QUEUE_FLAG_DISCARD, ns->queue);
+	blk_queue_max_discard_sectors(queue, UINT_MAX);
+	blk_queue_max_discard_segments(queue, NVME_DSM_MAX_RANGES);
+	queue_flag_set_unlocked(QUEUE_FLAG_DISCARD, queue);
 
 	if (ctrl->quirks & NVME_QUIRK_DEALLOCATE_ZEROES)
-		blk_queue_max_write_zeroes_sectors(ns->queue, UINT_MAX);
+		blk_queue_max_write_zeroes_sectors(queue, UINT_MAX);
 }
 
 static void nvme_report_ns_ids(struct nvme_ctrl *ctrl, unsigned int nsid,
-		struct nvme_id_ns *id, u8 *eui64, u8 *nguid, uuid_t *uuid)
+		struct nvme_id_ns *id, struct nvme_ns_ids *ids)
 {
+	memset(ids, 0, sizeof(*ids));
+
 	if (ctrl->vs >= NVME_VS(1, 1, 0))
-		memcpy(eui64, id->eui64, sizeof(id->eui64));
+		memcpy(ids->eui64, id->eui64, sizeof(id->eui64));
 	if (ctrl->vs >= NVME_VS(1, 2, 0))
-		memcpy(nguid, id->nguid, sizeof(id->nguid));
+		memcpy(ids->nguid, id->nguid, sizeof(id->nguid));
 	if (ctrl->vs >= NVME_VS(1, 3, 0)) {
 		 /* Don't treat error as fatal we potentially
 		  * already have a NGUID or EUI-64
 		  */
-		if (nvme_identify_ns_descs(ctrl, nsid, eui64, nguid, uuid))
+		if (nvme_identify_ns_descs(ctrl, nsid, ids))
 			dev_warn(ctrl->device,
 				 "%s: Identify Descriptors failed\n", __func__);
 	}
 }
 
+static bool nvme_ns_ids_valid(struct nvme_ns_ids *ids)
+{
+	return !uuid_is_null(&ids->uuid) ||
+		memchr_inv(ids->nguid, 0, sizeof(ids->nguid)) ||
+		memchr_inv(ids->eui64, 0, sizeof(ids->eui64));
+}
+
+static bool nvme_ns_ids_equal(struct nvme_ns_ids *a, struct nvme_ns_ids *b)
+{
+	return uuid_equal(&a->uuid, &b->uuid) &&
+		memcmp(&a->nguid, &b->nguid, sizeof(a->nguid)) == 0 &&
+		memcmp(&a->eui64, &b->eui64, sizeof(a->eui64)) == 0;
+}
+
+static void nvme_update_disk_info(struct gendisk *disk,
+		struct nvme_ns *ns, struct nvme_id_ns *id)
+{
+	sector_t capacity = le64_to_cpup(&id->nsze) << (ns->lba_shift - 9);
+	unsigned stream_alignment = 0;
+
+	if (ns->ctrl->nr_streams && ns->sws && ns->sgs)
+		stream_alignment = ns->sws * ns->sgs;
+
+	blk_mq_freeze_queue(disk->queue);
+	blk_integrity_unregister(disk);
+
+	blk_queue_logical_block_size(disk->queue, 1 << ns->lba_shift);
+	if (ns->ms && !ns->ext &&
+	    (ns->ctrl->ops->flags & NVME_F_METADATA_SUPPORTED))
+		nvme_init_integrity(disk, ns->ms, ns->pi_type);
+	if (ns->ms && !nvme_ns_has_pi(ns) && !blk_get_integrity(disk))
+		capacity = 0;
+	set_capacity(disk, capacity);
+
+	if (ns->ctrl->oncs & NVME_CTRL_ONCS_DSM)
+		nvme_config_discard(ns->ctrl, stream_alignment, disk->queue);
+	blk_mq_unfreeze_queue(disk->queue);
+}
+
 static void __nvme_revalidate_disk(struct gendisk *disk, struct nvme_id_ns *id)
 {
 	struct nvme_ns *ns = disk->private_data;
-	struct nvme_ctrl *ctrl = ns->ctrl;
-	u16 bs;
 
 	/*
 	 * If identify namespace failed, use default 512 byte block size so
@@ -1204,26 +1367,22 @@ static void __nvme_revalidate_disk(struct gendisk *disk, struct nvme_id_ns *id)
 	ns->lba_shift = id->lbaf[id->flbas & NVME_NS_FLBAS_LBA_MASK].ds;
 	if (ns->lba_shift == 0)
 		ns->lba_shift = 9;
-	bs = 1 << ns->lba_shift;
 	ns->noiob = le16_to_cpu(id->noiob);
+	ns->ext = ns->ms && (id->flbas & NVME_NS_FLBAS_META_EXT);
+	ns->ms = le16_to_cpu(id->lbaf[id->flbas & NVME_NS_FLBAS_LBA_MASK].ms);
+	/* the PI implementation requires metadata equal t10 pi tuple size */
+	if (ns->ms == sizeof(struct t10_pi_tuple))
+		ns->pi_type = id->dps & NVME_NS_DPS_PI_MASK;
+	else
+		ns->pi_type = 0;
 
-	blk_mq_freeze_queue(disk->queue);
-
-	if (ctrl->ops->flags & NVME_F_METADATA_SUPPORTED)
-		nvme_prep_integrity(disk, id, bs);
-	blk_queue_logical_block_size(ns->queue, bs);
 	if (ns->noiob)
 		nvme_set_chunk_size(ns);
-	if (ns->ms && !blk_get_integrity(disk) && !ns->ext)
-		nvme_init_integrity(ns);
-	if (ns->ms && !(ns->ms == 8 && ns->pi_type) && !blk_get_integrity(disk))
-		set_capacity(disk, 0);
-	else
-		set_capacity(disk, le64_to_cpup(&id->nsze) << (ns->lba_shift - 9));
-
-	if (ctrl->oncs & NVME_CTRL_ONCS_DSM)
-		nvme_config_discard(ns);
-	blk_mq_unfreeze_queue(disk->queue);
+	nvme_update_disk_info(disk, ns, id);
+#ifdef CONFIG_NVME_MULTIPATH
+	if (ns->head->disk)
+		nvme_update_disk_info(ns->head->disk, ns, id);
+#endif
 }
 
 static int nvme_revalidate_disk(struct gendisk *disk)
@@ -1231,8 +1390,7 @@ static int nvme_revalidate_disk(struct gendisk *disk)
 	struct nvme_ns *ns = disk->private_data;
 	struct nvme_ctrl *ctrl = ns->ctrl;
 	struct nvme_id_ns *id;
-	u8 eui64[8] = { 0 }, nguid[16] = { 0 };
-	uuid_t uuid = uuid_null;
+	struct nvme_ns_ids ids;
 	int ret = 0;
 
 	if (test_bit(NVME_NS_DEAD, &ns->flags)) {
@@ -1240,7 +1398,7 @@ static int nvme_revalidate_disk(struct gendisk *disk)
 		return -ENODEV;
 	}
 
-	id = nvme_identify_ns(ctrl, ns->ns_id);
+	id = nvme_identify_ns(ctrl, ns->head->ns_id);
 	if (!id)
 		return -ENODEV;
 
@@ -1250,12 +1408,10 @@ static int nvme_revalidate_disk(struct gendisk *disk)
 	}
 
 	__nvme_revalidate_disk(disk, id);
-	nvme_report_ns_ids(ctrl, ns->ns_id, id, eui64, nguid, &uuid);
-	if (!uuid_equal(&ns->uuid, &uuid) ||
-	    memcmp(&ns->nguid, &nguid, sizeof(ns->nguid)) ||
-	    memcmp(&ns->eui, &eui64, sizeof(ns->eui))) {
+	nvme_report_ns_ids(ctrl, ns->head->ns_id, id, &ids);
+	if (!nvme_ns_ids_equal(&ns->head->ids, &ids)) {
 		dev_err(ctrl->device,
-			"identifiers changed for nsid %d\n", ns->ns_id);
+			"identifiers changed for nsid %d\n", ns->head->ns_id);
 		ret = -ENODEV;
 	}
 
@@ -1287,8 +1443,10 @@ static char nvme_pr_type(enum pr_type type)
 static int nvme_pr_command(struct block_device *bdev, u32 cdw10,
 				u64 key, u64 sa_key, u8 op)
 {
-	struct nvme_ns *ns = bdev->bd_disk->private_data;
+	struct nvme_ns_head *head = NULL;
+	struct nvme_ns *ns;
 	struct nvme_command c;
+	int srcu_idx, ret;
 	u8 data[16] = { 0, };
 
 	put_unaligned_le64(key, &data[0]);
@@ -1296,10 +1454,16 @@ static int nvme_pr_command(struct block_device *bdev, u32 cdw10,
 
 	memset(&c, 0, sizeof(c));
 	c.common.opcode = op;
-	c.common.nsid = cpu_to_le32(ns->ns_id);
+	c.common.nsid = cpu_to_le32(head->ns_id);
 	c.common.cdw10[0] = cpu_to_le32(cdw10);
 
-	return nvme_submit_sync_cmd(ns->queue, &c, data, 16);
+	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);
+	nvme_put_ns_from_disk(head, srcu_idx);
+	return ret;
 }
 
 static int nvme_pr_register(struct block_device *bdev, u64 old,
@@ -1381,7 +1545,7 @@ EXPORT_SYMBOL_GPL(nvme_sec_submit);
 static const struct block_device_operations nvme_fops = {
 	.owner		= THIS_MODULE,
 	.ioctl		= nvme_ioctl,
-	.compat_ioctl	= nvme_compat_ioctl,
+	.compat_ioctl	= nvme_ioctl,
 	.open		= nvme_open,
 	.release	= nvme_release,
 	.getgeo		= nvme_getgeo,
@@ -1389,6 +1553,32 @@ static const struct block_device_operations nvme_fops = {
 	.pr_ops		= &nvme_pr_ops,
 };
 
+#ifdef CONFIG_NVME_MULTIPATH
+static int nvme_ns_head_open(struct block_device *bdev, fmode_t mode)
+{
+	struct nvme_ns_head *head = bdev->bd_disk->private_data;
+
+	if (!kref_get_unless_zero(&head->ref))
+		return -ENXIO;
+	return 0;
+}
+
+static void nvme_ns_head_release(struct gendisk *disk, fmode_t mode)
+{
+	nvme_put_ns_head(disk->private_data);
+}
+
+const struct block_device_operations nvme_ns_head_ops = {
+	.owner		= THIS_MODULE,
+	.open		= nvme_ns_head_open,
+	.release	= nvme_ns_head_release,
+	.ioctl		= nvme_ioctl,
+	.compat_ioctl	= nvme_ioctl,
+	.getgeo		= nvme_getgeo,
+	.pr_ops		= &nvme_pr_ops,
+};
+#endif /* CONFIG_NVME_MULTIPATH */
+
 static int nvme_wait_ready(struct nvme_ctrl *ctrl, u64 cap, bool enabled)
 {
 	unsigned long timeout =
@@ -1737,14 +1927,15 @@ static bool quirk_matches(const struct nvme_id_ctrl *id,
 		string_matches(id->fr, q->fr, sizeof(id->fr));
 }
 
-static void nvme_init_subnqn(struct nvme_ctrl *ctrl, struct nvme_id_ctrl *id)
+static void nvme_init_subnqn(struct nvme_subsystem *subsys, struct nvme_ctrl *ctrl,
+		struct nvme_id_ctrl *id)
 {
 	size_t nqnlen;
 	int off;
 
 	nqnlen = strnlen(id->subnqn, NVMF_NQN_SIZE);
 	if (nqnlen > 0 && nqnlen < NVMF_NQN_SIZE) {
-		strcpy(ctrl->subnqn, id->subnqn);
+		strncpy(subsys->subnqn, id->subnqn, NVMF_NQN_SIZE);
 		return;
 	}
 
@@ -1752,14 +1943,222 @@ static void nvme_init_subnqn(struct nvme_ctrl *ctrl, struct nvme_id_ctrl *id)
 		dev_warn(ctrl->device, "missing or invalid SUBNQN field.\n");
 
 	/* Generate a "fake" NQN per Figure 254 in NVMe 1.3 + ECN 001 */
-	off = snprintf(ctrl->subnqn, NVMF_NQN_SIZE,
+	off = snprintf(subsys->subnqn, NVMF_NQN_SIZE,
 			"nqn.2014.08.org.nvmexpress:%4x%4x",
 			le16_to_cpu(id->vid), le16_to_cpu(id->ssvid));
-	memcpy(ctrl->subnqn + off, id->sn, sizeof(id->sn));
+	memcpy(subsys->subnqn + off, id->sn, sizeof(id->sn));
 	off += sizeof(id->sn);
-	memcpy(ctrl->subnqn + off, id->mn, sizeof(id->mn));
+	memcpy(subsys->subnqn + off, id->mn, sizeof(id->mn));
 	off += sizeof(id->mn);
-	memset(ctrl->subnqn + off, 0, sizeof(ctrl->subnqn) - off);
+	memset(subsys->subnqn + off, 0, sizeof(subsys->subnqn) - off);
+}
+
+static void __nvme_release_subsystem(struct nvme_subsystem *subsys)
+{
+	ida_simple_remove(&nvme_subsystems_ida, subsys->instance);
+	kfree(subsys);
+}
+
+static void nvme_release_subsystem(struct device *dev)
+{
+	__nvme_release_subsystem(container_of(dev, struct nvme_subsystem, dev));
+}
+
+static void nvme_destroy_subsystem(struct kref *ref)
+{
+	struct nvme_subsystem *subsys =
+			container_of(ref, struct nvme_subsystem, ref);
+
+	mutex_lock(&nvme_subsystems_lock);
+	list_del(&subsys->entry);
+	mutex_unlock(&nvme_subsystems_lock);
+
+	ida_destroy(&subsys->ns_ida);
+	device_del(&subsys->dev);
+	put_device(&subsys->dev);
+}
+
+static void nvme_put_subsystem(struct nvme_subsystem *subsys)
+{
+	kref_put(&subsys->ref, nvme_destroy_subsystem);
+}
+
+static struct nvme_subsystem *__nvme_find_get_subsystem(const char *subsysnqn)
+{
+	struct nvme_subsystem *subsys;
+
+	lockdep_assert_held(&nvme_subsystems_lock);
+
+	list_for_each_entry(subsys, &nvme_subsystems, entry) {
+		if (strcmp(subsys->subnqn, subsysnqn))
+			continue;
+		if (!kref_get_unless_zero(&subsys->ref))
+			continue;
+		return subsys;
+	}
+
+	return NULL;
+}
+
+#define SUBSYS_ATTR_RO(_name, _mode, _show)			\
+	struct device_attribute subsys_attr_##_name = \
+		__ATTR(_name, _mode, _show, NULL)
+
+static ssize_t nvme_subsys_show_nqn(struct device *dev,
+				    struct device_attribute *attr,
+				    char *buf)
+{
+	struct nvme_subsystem *subsys =
+		container_of(dev, struct nvme_subsystem, dev);
+
+	return snprintf(buf, PAGE_SIZE, "%s\n", subsys->subnqn);
+}
+static SUBSYS_ATTR_RO(subsysnqn, S_IRUGO, nvme_subsys_show_nqn);
+
+#define nvme_subsys_show_str_function(field)				\
+static ssize_t subsys_##field##_show(struct device *dev,		\
+			    struct device_attribute *attr, char *buf)	\
+{									\
+	struct nvme_subsystem *subsys =					\
+		container_of(dev, struct nvme_subsystem, dev);		\
+	return sprintf(buf, "%.*s\n",					\
+		       (int)sizeof(subsys->field), subsys->field);	\
+}									\
+static SUBSYS_ATTR_RO(field, S_IRUGO, subsys_##field##_show);
+
+nvme_subsys_show_str_function(model);
+nvme_subsys_show_str_function(serial);
+nvme_subsys_show_str_function(firmware_rev);
+
+static struct attribute *nvme_subsys_attrs[] = {
+	&subsys_attr_model.attr,
+	&subsys_attr_serial.attr,
+	&subsys_attr_firmware_rev.attr,
+	&subsys_attr_subsysnqn.attr,
+	NULL,
+};
+
+static struct attribute_group nvme_subsys_attrs_group = {
+	.attrs = nvme_subsys_attrs,
+};
+
+static const struct attribute_group *nvme_subsys_attrs_groups[] = {
+	&nvme_subsys_attrs_group,
+	NULL,
+};
+
+static int nvme_init_subsystem(struct nvme_ctrl *ctrl, struct nvme_id_ctrl *id)
+{
+	struct nvme_subsystem *subsys, *found;
+	int ret;
+
+	subsys = kzalloc(sizeof(*subsys), GFP_KERNEL);
+	if (!subsys)
+		return -ENOMEM;
+	ret = ida_simple_get(&nvme_subsystems_ida, 0, 0, GFP_KERNEL);
+	if (ret < 0) {
+		kfree(subsys);
+		return ret;
+	}
+	subsys->instance = ret;
+	mutex_init(&subsys->lock);
+	kref_init(&subsys->ref);
+	INIT_LIST_HEAD(&subsys->ctrls);
+	INIT_LIST_HEAD(&subsys->nsheads);
+	nvme_init_subnqn(subsys, ctrl, id);
+	memcpy(subsys->serial, id->sn, sizeof(subsys->serial));
+	memcpy(subsys->model, id->mn, sizeof(subsys->model));
+	memcpy(subsys->firmware_rev, id->fr, sizeof(subsys->firmware_rev));
+	subsys->vendor_id = le16_to_cpu(id->vid);
+	subsys->cmic = id->cmic;
+
+	subsys->dev.class = nvme_subsys_class;
+	subsys->dev.release = nvme_release_subsystem;
+	subsys->dev.groups = nvme_subsys_attrs_groups;
+	dev_set_name(&subsys->dev, "nvme-subsys%d", subsys->instance);
+	device_initialize(&subsys->dev);
+
+	mutex_lock(&nvme_subsystems_lock);
+	found = __nvme_find_get_subsystem(subsys->subnqn);
+	if (found) {
+		/*
+		 * Verify that the subsystem actually supports multiple
+		 * controllers, else bail out.
+		 */
+		if (!(id->cmic & (1 << 1))) {
+			dev_err(ctrl->device,
+				"ignoring ctrl due to duplicate subnqn (%s).\n",
+				found->subnqn);
+			nvme_put_subsystem(found);
+			ret = -EINVAL;
+			goto out_unlock;
+		}
+
+		__nvme_release_subsystem(subsys);
+		subsys = found;
+	} else {
+		ret = device_add(&subsys->dev);
+		if (ret) {
+			dev_err(ctrl->device,
+				"failed to register subsystem device.\n");
+			goto out_unlock;
+		}
+		ida_init(&subsys->ns_ida);
+		list_add_tail(&subsys->entry, &nvme_subsystems);
+	}
+
+	ctrl->subsys = subsys;
+	mutex_unlock(&nvme_subsystems_lock);
+
+	if (sysfs_create_link(&subsys->dev.kobj, &ctrl->device->kobj,
+			dev_name(ctrl->device))) {
+		dev_err(ctrl->device,
+			"failed to create sysfs link from subsystem.\n");
+		/* the transport driver will eventually put the subsystem */
+		return -EINVAL;
+	}
+
+	mutex_lock(&subsys->lock);
+	list_add_tail(&ctrl->subsys_entry, &subsys->ctrls);
+	mutex_unlock(&subsys->lock);
+
+	return 0;
+
+out_unlock:
+	mutex_unlock(&nvme_subsystems_lock);
+	put_device(&subsys->dev);
+	return ret;
+}
+
+static int nvme_get_log(struct nvme_ctrl *ctrl, u8 log_page, void *log,
+			size_t size)
+{
+	struct nvme_command c = { };
+
+	c.common.opcode = nvme_admin_get_log_page;
+	c.common.nsid = cpu_to_le32(NVME_NSID_ALL);
+	c.common.cdw10[0] = nvme_get_log_dw10(log_page, size);
+
+	return nvme_submit_sync_cmd(ctrl->admin_q, &c, log, size);
+}
+
+static int nvme_get_effects_log(struct nvme_ctrl *ctrl)
+{
+	int ret;
+
+	if (!ctrl->effects)
+		ctrl->effects = kzalloc(sizeof(*ctrl->effects), GFP_KERNEL);
+
+	if (!ctrl->effects)
+		return 0;
+
+	ret = nvme_get_log(ctrl, NVME_LOG_CMD_EFFECTS, ctrl->effects,
+					sizeof(*ctrl->effects));
+	if (ret) {
+		kfree(ctrl->effects);
+		ctrl->effects = NULL;
+	}
+	return ret;
 }
 
 /*
@@ -1797,9 +2196,19 @@ int nvme_init_identify(struct nvme_ctrl *ctrl)
 		return -EIO;
 	}
 
-	nvme_init_subnqn(ctrl, id);
+	if (id->lpa & NVME_CTRL_LPA_CMD_EFFECTS_LOG) {
+		ret = nvme_get_effects_log(ctrl);
+		if (ret < 0)
+			return ret;
+	}
 
 	if (!ctrl->identified) {
+		int i;
+
+		ret = nvme_init_subsystem(ctrl, id);
+		if (ret)
+			goto out_free;
+
 		/*
 		 * Check for quirks.  Quirk can depend on firmware version,
 		 * so, in principle, the set of quirks present can change
@@ -1808,9 +2217,6 @@ int nvme_init_identify(struct nvme_ctrl *ctrl)
 		 * the device, but we'd have to make sure that the driver
 		 * behaves intelligently if the quirks change.
 		 */
-
-		int i;
-
 		for (i = 0; i < ARRAY_SIZE(core_quirks); i++) {
 			if (quirk_matches(id, &core_quirks[i]))
 				ctrl->quirks |= core_quirks[i].quirks;
@@ -1823,14 +2229,10 @@ int nvme_init_identify(struct nvme_ctrl *ctrl)
 	}
 
 	ctrl->oacs = le16_to_cpu(id->oacs);
-	ctrl->vid = le16_to_cpu(id->vid);
 	ctrl->oncs = le16_to_cpup(&id->oncs);
 	atomic_set(&ctrl->abort_limit, id->acl + 1);
 	ctrl->vwc = id->vwc;
 	ctrl->cntlid = le16_to_cpup(&id->cntlid);
-	memcpy(ctrl->serial, id->sn, sizeof(id->sn));
-	memcpy(ctrl->model, id->mn, sizeof(id->mn));
-	memcpy(ctrl->firmware_rev, id->fr, sizeof(id->fr));
 	if (id->mdts)
 		max_hw_sectors = 1 << (id->mdts + page_shift - 9);
 	else
@@ -1931,33 +2333,12 @@ EXPORT_SYMBOL_GPL(nvme_init_identify);
 
 static int nvme_dev_open(struct inode *inode, struct file *file)
 {
-	struct nvme_ctrl *ctrl;
-	int instance = iminor(inode);
-	int ret = -ENODEV;
-
-	spin_lock(&dev_list_lock);
-	list_for_each_entry(ctrl, &nvme_ctrl_list, node) {
-		if (ctrl->instance != instance)
-			continue;
-
-		if (!ctrl->admin_q) {
-			ret = -EWOULDBLOCK;
-			break;
-		}
-		if (!kref_get_unless_zero(&ctrl->kref))
-			break;
-		file->private_data = ctrl;
-		ret = 0;
-		break;
-	}
-	spin_unlock(&dev_list_lock);
-
-	return ret;
-}
+	struct nvme_ctrl *ctrl =
+		container_of(inode->i_cdev, struct nvme_ctrl, cdev);
 
-static int nvme_dev_release(struct inode *inode, struct file *file)
-{
-	nvme_put_ctrl(file->private_data);
+	if (ctrl->state != NVME_CTRL_LIVE)
+		return -EWOULDBLOCK;
+	file->private_data = ctrl;
 	return 0;
 }
 
@@ -2021,7 +2402,6 @@ static long nvme_dev_ioctl(struct file *file, unsigned int cmd,
 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,
 };
@@ -2051,77 +2431,86 @@ static ssize_t nvme_sysfs_rescan(struct device *dev,
 }
 static DEVICE_ATTR(rescan_controller, S_IWUSR, NULL, nvme_sysfs_rescan);
 
+static inline struct nvme_ns_head *dev_to_ns_head(struct device *dev)
+{
+	struct gendisk *disk = dev_to_disk(dev);
+
+	if (disk->fops == &nvme_fops)
+		return nvme_get_ns_from_dev(dev)->head;
+	else
+		return disk->private_data;
+}
+
 static ssize_t wwid_show(struct device *dev, struct device_attribute *attr,
-								char *buf)
+		char *buf)
 {
-	struct nvme_ns *ns = nvme_get_ns_from_dev(dev);
-	struct nvme_ctrl *ctrl = ns->ctrl;
-	int serial_len = sizeof(ctrl->serial);
-	int model_len = sizeof(ctrl->model);
+	struct nvme_ns_head *head = dev_to_ns_head(dev);
+	struct nvme_ns_ids *ids = &head->ids;
+	struct nvme_subsystem *subsys = head->subsys;
+	int serial_len = sizeof(subsys->serial);
+	int model_len = sizeof(subsys->model);
 
-	if (!uuid_is_null(&ns->uuid))
-		return sprintf(buf, "uuid.%pU\n", &ns->uuid);
+	if (!uuid_is_null(&ids->uuid))
+		return sprintf(buf, "uuid.%pU\n", &ids->uuid);
 
-	if (memchr_inv(ns->nguid, 0, sizeof(ns->nguid)))
-		return sprintf(buf, "eui.%16phN\n", ns->nguid);
+	if (memchr_inv(ids->nguid, 0, sizeof(ids->nguid)))
+		return sprintf(buf, "eui.%16phN\n", ids->nguid);
 
-	if (memchr_inv(ns->eui, 0, sizeof(ns->eui)))
-		return sprintf(buf, "eui.%8phN\n", ns->eui);
+	if (memchr_inv(ids->eui64, 0, sizeof(ids->eui64)))
+		return sprintf(buf, "eui.%8phN\n", ids->eui64);
 
-	while (serial_len > 0 && (ctrl->serial[serial_len - 1] == ' ' ||
-				  ctrl->serial[serial_len - 1] == '\0'))
+	while (serial_len > 0 && (subsys->serial[serial_len - 1] == ' ' ||
+				  subsys->serial[serial_len - 1] == '\0'))
 		serial_len--;
-	while (model_len > 0 && (ctrl->model[model_len - 1] == ' ' ||
-				 ctrl->model[model_len - 1] == '\0'))
+	while (model_len > 0 && (subsys->model[model_len - 1] == ' ' ||
+				 subsys->model[model_len - 1] == '\0'))
 		model_len--;
 
-	return sprintf(buf, "nvme.%04x-%*phN-%*phN-%08x\n", ctrl->vid,
-		serial_len, ctrl->serial, model_len, ctrl->model, ns->ns_id);
+	return sprintf(buf, "nvme.%04x-%*phN-%*phN-%08x\n", subsys->vendor_id,
+		serial_len, subsys->serial, model_len, subsys->model,
+		head->ns_id);
 }
 static DEVICE_ATTR(wwid, S_IRUGO, wwid_show, NULL);
 
 static ssize_t nguid_show(struct device *dev, struct device_attribute *attr,
-			  char *buf)
+		char *buf)
 {
-	struct nvme_ns *ns = nvme_get_ns_from_dev(dev);
-	return sprintf(buf, "%pU\n", ns->nguid);
+	return sprintf(buf, "%pU\n", dev_to_ns_head(dev)->ids.nguid);
 }
 static DEVICE_ATTR(nguid, S_IRUGO, nguid_show, NULL);
 
 static ssize_t uuid_show(struct device *dev, struct device_attribute *attr,
-								char *buf)
+		char *buf)
 {
-	struct nvme_ns *ns = nvme_get_ns_from_dev(dev);
+	struct nvme_ns_ids *ids = &dev_to_ns_head(dev)->ids;
 
 	/* For backward compatibility expose the NGUID to userspace if
 	 * we have no UUID set
 	 */
-	if (uuid_is_null(&ns->uuid)) {
+	if (uuid_is_null(&ids->uuid)) {
 		printk_ratelimited(KERN_WARNING
 				   "No UUID available providing old NGUID\n");
-		return sprintf(buf, "%pU\n", ns->nguid);
+		return sprintf(buf, "%pU\n", ids->nguid);
 	}
-	return sprintf(buf, "%pU\n", &ns->uuid);
+	return sprintf(buf, "%pU\n", &ids->uuid);
 }
 static DEVICE_ATTR(uuid, S_IRUGO, uuid_show, NULL);
 
 static ssize_t eui_show(struct device *dev, struct device_attribute *attr,
-								char *buf)
+		char *buf)
 {
-	struct nvme_ns *ns = nvme_get_ns_from_dev(dev);
-	return sprintf(buf, "%8phd\n", ns->eui);
+	return sprintf(buf, "%8ph\n", dev_to_ns_head(dev)->ids.eui64);
 }
 static DEVICE_ATTR(eui, S_IRUGO, eui_show, NULL);
 
 static ssize_t nsid_show(struct device *dev, struct device_attribute *attr,
-								char *buf)
+		char *buf)
 {
-	struct nvme_ns *ns = nvme_get_ns_from_dev(dev);
-	return sprintf(buf, "%d\n", ns->ns_id);
+	return sprintf(buf, "%d\n", dev_to_ns_head(dev)->ns_id);
 }
 static DEVICE_ATTR(nsid, S_IRUGO, nsid_show, NULL);
 
-static struct attribute *nvme_ns_attrs[] = {
+static struct attribute *nvme_ns_id_attrs[] = {
 	&dev_attr_wwid.attr,
 	&dev_attr_uuid.attr,
 	&dev_attr_nguid.attr,
@@ -2130,31 +2519,31 @@ static struct attribute *nvme_ns_attrs[] = {
 	NULL,
 };
 
-static umode_t nvme_ns_attrs_are_visible(struct kobject *kobj,
+static umode_t nvme_ns_id_attrs_are_visible(struct kobject *kobj,
 		struct attribute *a, int n)
 {
 	struct device *dev = container_of(kobj, struct device, kobj);
-	struct nvme_ns *ns = nvme_get_ns_from_dev(dev);
+	struct nvme_ns_ids *ids = &dev_to_ns_head(dev)->ids;
 
 	if (a == &dev_attr_uuid.attr) {
-		if (uuid_is_null(&ns->uuid) &&
-		    !memchr_inv(ns->nguid, 0, sizeof(ns->nguid)))
+		if (uuid_is_null(&ids->uuid) &&
+		    !memchr_inv(ids->nguid, 0, sizeof(ids->nguid)))
 			return 0;
 	}
 	if (a == &dev_attr_nguid.attr) {
-		if (!memchr_inv(ns->nguid, 0, sizeof(ns->nguid)))
+		if (!memchr_inv(ids->nguid, 0, sizeof(ids->nguid)))
 			return 0;
 	}
 	if (a == &dev_attr_eui.attr) {
-		if (!memchr_inv(ns->eui, 0, sizeof(ns->eui)))
+		if (!memchr_inv(ids->eui64, 0, sizeof(ids->eui64)))
 			return 0;
 	}
 	return a->mode;
 }
 
-static const struct attribute_group nvme_ns_attr_group = {
-	.attrs		= nvme_ns_attrs,
-	.is_visible	= nvme_ns_attrs_are_visible,
+const struct attribute_group nvme_ns_id_attr_group = {
+	.attrs		= nvme_ns_id_attrs,
+	.is_visible	= nvme_ns_id_attrs_are_visible,
 };
 
 #define nvme_show_str_function(field)						\
@@ -2162,10 +2551,15 @@ static ssize_t  field##_show(struct device *dev,				\
 			    struct device_attribute *attr, char *buf)		\
 {										\
         struct nvme_ctrl *ctrl = dev_get_drvdata(dev);				\
-        return sprintf(buf, "%.*s\n", (int)sizeof(ctrl->field), ctrl->field);	\
+        return sprintf(buf, "%.*s\n",						\
+		(int)sizeof(ctrl->subsys->field), ctrl->subsys->field);		\
 }										\
 static DEVICE_ATTR(field, S_IRUGO, field##_show, NULL);
 
+nvme_show_str_function(model);
+nvme_show_str_function(serial);
+nvme_show_str_function(firmware_rev);
+
 #define nvme_show_int_function(field)						\
 static ssize_t  field##_show(struct device *dev,				\
 			    struct device_attribute *attr, char *buf)		\
@@ -2175,9 +2569,6 @@ static ssize_t  field##_show(struct device *dev,				\
 }										\
 static DEVICE_ATTR(field, S_IRUGO, field##_show, NULL);
 
-nvme_show_str_function(model);
-nvme_show_str_function(serial);
-nvme_show_str_function(firmware_rev);
 nvme_show_int_function(cntlid);
 
 static ssize_t nvme_sysfs_delete(struct device *dev,
@@ -2187,7 +2578,7 @@ static ssize_t nvme_sysfs_delete(struct device *dev,
 	struct nvme_ctrl *ctrl = dev_get_drvdata(dev);
 
 	if (device_remove_file_self(dev, attr))
-		ctrl->ops->delete_ctrl(ctrl);
+		nvme_delete_ctrl_sync(ctrl);
 	return count;
 }
 static DEVICE_ATTR(delete_controller, S_IWUSR, NULL, nvme_sysfs_delete);
@@ -2231,7 +2622,7 @@ static ssize_t nvme_sysfs_show_subsysnqn(struct device *dev,
 {
 	struct nvme_ctrl *ctrl = dev_get_drvdata(dev);
 
-	return snprintf(buf, PAGE_SIZE, "%s\n", ctrl->subnqn);
+	return snprintf(buf, PAGE_SIZE, "%s\n", ctrl->subsys->subnqn);
 }
 static DEVICE_ATTR(subsysnqn, S_IRUGO, nvme_sysfs_show_subsysnqn, NULL);
 
@@ -2284,12 +2675,128 @@ static const struct attribute_group *nvme_dev_attr_groups[] = {
 	NULL,
 };
 
+static struct nvme_ns_head *__nvme_find_ns_head(struct nvme_subsystem *subsys,
+		unsigned nsid)
+{
+	struct nvme_ns_head *h;
+
+	lockdep_assert_held(&subsys->lock);
+
+	list_for_each_entry(h, &subsys->nsheads, entry) {
+		if (h->ns_id == nsid && kref_get_unless_zero(&h->ref))
+			return h;
+	}
+
+	return NULL;
+}
+
+static int __nvme_check_ids(struct nvme_subsystem *subsys,
+		struct nvme_ns_head *new)
+{
+	struct nvme_ns_head *h;
+
+	lockdep_assert_held(&subsys->lock);
+
+	list_for_each_entry(h, &subsys->nsheads, entry) {
+		if (nvme_ns_ids_valid(&new->ids) &&
+		    nvme_ns_ids_equal(&new->ids, &h->ids))
+			return -EINVAL;
+	}
+
+	return 0;
+}
+
+static struct nvme_ns_head *nvme_alloc_ns_head(struct nvme_ctrl *ctrl,
+		unsigned nsid, struct nvme_id_ns *id)
+{
+	struct nvme_ns_head *head;
+	int ret = -ENOMEM;
+
+	head = kzalloc(sizeof(*head), GFP_KERNEL);
+	if (!head)
+		goto out;
+	ret = ida_simple_get(&ctrl->subsys->ns_ida, 1, 0, GFP_KERNEL);
+	if (ret < 0)
+		goto out_free_head;
+	head->instance = ret;
+	INIT_LIST_HEAD(&head->list);
+	init_srcu_struct(&head->srcu);
+	head->subsys = ctrl->subsys;
+	head->ns_id = nsid;
+	kref_init(&head->ref);
+
+	nvme_report_ns_ids(ctrl, nsid, id, &head->ids);
+
+	ret = __nvme_check_ids(ctrl->subsys, head);
+	if (ret) {
+		dev_err(ctrl->device,
+			"duplicate IDs for nsid %d\n", nsid);
+		goto out_cleanup_srcu;
+	}
+
+	ret = nvme_mpath_alloc_disk(ctrl, head);
+	if (ret)
+		goto out_cleanup_srcu;
+
+	list_add_tail(&head->entry, &ctrl->subsys->nsheads);
+	return head;
+out_cleanup_srcu:
+	cleanup_srcu_struct(&head->srcu);
+	ida_simple_remove(&ctrl->subsys->ns_ida, head->instance);
+out_free_head:
+	kfree(head);
+out:
+	return ERR_PTR(ret);
+}
+
+static int nvme_init_ns_head(struct nvme_ns *ns, unsigned nsid,
+		struct nvme_id_ns *id, bool *new)
+{
+	struct nvme_ctrl *ctrl = ns->ctrl;
+	bool is_shared = id->nmic & (1 << 0);
+	struct nvme_ns_head *head = NULL;
+	int ret = 0;
+
+	mutex_lock(&ctrl->subsys->lock);
+	if (is_shared)
+		head = __nvme_find_ns_head(ctrl->subsys, nsid);
+	if (!head) {
+		head = nvme_alloc_ns_head(ctrl, nsid, id);
+		if (IS_ERR(head)) {
+			ret = PTR_ERR(head);
+			goto out_unlock;
+		}
+
+		*new = true;
+	} else {
+		struct nvme_ns_ids ids;
+
+		nvme_report_ns_ids(ctrl, nsid, id, &ids);
+		if (!nvme_ns_ids_equal(&head->ids, &ids)) {
+			dev_err(ctrl->device,
+				"IDs don't match for shared namespace %d\n",
+					nsid);
+			ret = -EINVAL;
+			goto out_unlock;
+		}
+
+		*new = false;
+	}
+
+	list_add_tail(&ns->siblings, &head->list);
+	ns->head = head;
+
+out_unlock:
+	mutex_unlock(&ctrl->subsys->lock);
+	return ret;
+}
+
 static int ns_cmp(void *priv, struct list_head *a, struct list_head *b)
 {
 	struct nvme_ns *nsa = container_of(a, struct nvme_ns, list);
 	struct nvme_ns *nsb = container_of(b, struct nvme_ns, list);
 
-	return nsa->ns_id - nsb->ns_id;
+	return nsa->head->ns_id - nsb->head->ns_id;
 }
 
 static struct nvme_ns *nvme_find_get_ns(struct nvme_ctrl *ctrl, unsigned nsid)
@@ -2298,12 +2805,13 @@ static struct nvme_ns *nvme_find_get_ns(struct nvme_ctrl *ctrl, unsigned nsid)
 
 	mutex_lock(&ctrl->namespaces_mutex);
 	list_for_each_entry(ns, &ctrl->namespaces, list) {
-		if (ns->ns_id == nsid) {
-			kref_get(&ns->kref);
+		if (ns->head->ns_id == nsid) {
+			if (!kref_get_unless_zero(&ns->kref))
+				continue;
 			ret = ns;
 			break;
 		}
-		if (ns->ns_id > nsid)
+		if (ns->head->ns_id > nsid)
 			break;
 	}
 	mutex_unlock(&ctrl->namespaces_mutex);
@@ -2318,7 +2826,7 @@ static int nvme_setup_streams_ns(struct nvme_ctrl *ctrl, struct nvme_ns *ns)
 	if (!ctrl->nr_streams)
 		return 0;
 
-	ret = nvme_get_stream_params(ctrl, &s, ns->ns_id);
+	ret = nvme_get_stream_params(ctrl, &s, ns->head->ns_id);
 	if (ret)
 		return ret;
 
@@ -2342,33 +2850,27 @@ static void nvme_alloc_ns(struct nvme_ctrl *ctrl, unsigned nsid)
 	struct gendisk *disk;
 	struct nvme_id_ns *id;
 	char disk_name[DISK_NAME_LEN];
-	int node = dev_to_node(ctrl->dev);
+	int node = dev_to_node(ctrl->dev), flags = GENHD_FL_EXT_DEVT;
+	bool new = true;
 
 	ns = kzalloc_node(sizeof(*ns), GFP_KERNEL, node);
 	if (!ns)
 		return;
 
-	ns->instance = ida_simple_get(&ctrl->ns_ida, 1, 0, GFP_KERNEL);
-	if (ns->instance < 0)
-		goto out_free_ns;
-
 	ns->queue = blk_mq_init_queue(ctrl->tagset);
 	if (IS_ERR(ns->queue))
-		goto out_release_instance;
+		goto out_free_ns;
 	queue_flag_set_unlocked(QUEUE_FLAG_NONROT, ns->queue);
 	ns->queue->queuedata = ns;
 	ns->ctrl = ctrl;
 
 	kref_init(&ns->kref);
-	ns->ns_id = nsid;
 	ns->lba_shift = 9; /* set to a default value for 512 until disk is validated */
 
 	blk_queue_logical_block_size(ns->queue, 1 << ns->lba_shift);
 	nvme_set_queue_limits(ctrl, ns->queue);
 	nvme_setup_streams_ns(ctrl, ns);
 
-	sprintf(disk_name, "nvme%dn%d", ctrl->instance, ns->instance);
-
 	id = nvme_identify_ns(ctrl, nsid);
 	if (!id)
 		goto out_free_queue;
@@ -2376,23 +2878,49 @@ static void nvme_alloc_ns(struct nvme_ctrl *ctrl, unsigned nsid)
 	if (id->ncap == 0)
 		goto out_free_id;
 
-	nvme_report_ns_ids(ctrl, ns->ns_id, id, ns->eui, ns->nguid, &ns->uuid);
+	if (nvme_init_ns_head(ns, nsid, id, &new))
+		goto out_free_id;
+	
+#ifdef CONFIG_NVME_MULTIPATH
+	/*
+	 * If multipathing is enabled we need to always use the subsystem
+	 * instance number for numbering our devices to avoid conflicts
+	 * between subsystems that have multiple controllers and thus use
+	 * the multipath-aware subsystem node and those that have a single
+	 * controller and use the controller node directly.
+	 */
+	if (ns->head->disk) {
+		sprintf(disk_name, "nvme%dc%dn%d", ctrl->subsys->instance,
+				ctrl->cntlid, ns->head->instance);
+		flags = GENHD_FL_HIDDEN;
+	} else {
+		sprintf(disk_name, "nvme%dn%d", ctrl->subsys->instance,
+				ns->head->instance);
+	}
+#else
+	/*
+	 * But without the multipath code enabled, multiple controller per
+	 * subsystems are visible as devices and thus we cannot use the
+	 * subsystem instance.
+	 */
+	sprintf(disk_name, "nvme%dn%d", ctrl->instance, ns->head->instance);
+#endif
 
 	if ((ctrl->quirks & NVME_QUIRK_LIGHTNVM) && id->vs[0] == 0x1) {
 		if (nvme_nvm_register(ns, disk_name, node)) {
 			dev_warn(ctrl->device, "LightNVM init failure\n");
-			goto out_free_id;
+			goto out_unlink_ns;
 		}
 	}
 
 	disk = alloc_disk_node(0, node);
 	if (!disk)
-		goto out_free_id;
+		goto out_unlink_ns;
 
 	disk->fops = &nvme_fops;
 	disk->private_data = ns;
 	disk->queue = ns->queue;
-	disk->flags = GENHD_FL_EXT_DEVT;
+	disk->flags = flags;
 	memcpy(disk->disk_name, disk_name, DISK_NAME_LEN);
 	ns->disk = disk;
 
@@ -2402,49 +2930,65 @@ static void nvme_alloc_ns(struct nvme_ctrl *ctrl, unsigned nsid)
 	list_add_tail(&ns->list, &ctrl->namespaces);
 	mutex_unlock(&ctrl->namespaces_mutex);
 
-	kref_get(&ctrl->kref);
+	nvme_get_ctrl(ctrl);
 
 	kfree(id);
 
 	device_add_disk(ctrl->device, ns->disk);
 	if (sysfs_create_group(&disk_to_dev(ns->disk)->kobj,
-					&nvme_ns_attr_group))
+					&nvme_ns_id_attr_group))
 		pr_warn("%s: failed to create sysfs group for identification\n",
 			ns->disk->disk_name);
 	if (ns->ndev && nvme_nvm_register_sysfs(ns))
 		pr_warn("%s: failed to register lightnvm sysfs group for identification\n",
 			ns->disk->disk_name);
+
+	if (new)
+		nvme_mpath_add_disk(ns->head);
+	nvme_mpath_add_disk_links(ns);
 	return;
+ out_unlink_ns:
+	mutex_lock(&ctrl->subsys->lock);
+	list_del_rcu(&ns->siblings);
+	mutex_unlock(&ctrl->subsys->lock);
  out_free_id:
 	kfree(id);
  out_free_queue:
 	blk_cleanup_queue(ns->queue);
- out_release_instance:
-	ida_simple_remove(&ctrl->ns_ida, ns->instance);
  out_free_ns:
 	kfree(ns);
 }
 
 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;
 
 	if (ns->disk && ns->disk->flags & GENHD_FL_UP) {
 		if (blk_get_integrity(ns->disk))
 			blk_integrity_unregister(ns->disk);
+		nvme_mpath_remove_disk_links(ns);
 		sysfs_remove_group(&disk_to_dev(ns->disk)->kobj,
-					&nvme_ns_attr_group);
+					&nvme_ns_id_attr_group);
 		if (ns->ndev)
 			nvme_nvm_unregister_sysfs(ns);
 		del_gendisk(ns->disk);
 		blk_cleanup_queue(ns->queue);
 	}
 
+	mutex_lock(&ns->ctrl->subsys->lock);
+	nvme_mpath_clear_current_path(ns);
+	if (head)
+		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);
 	nvme_put_ns(ns);
 }
 
@@ -2467,7 +3011,7 @@ static void nvme_remove_invalid_namespaces(struct nvme_ctrl *ctrl,
 	struct nvme_ns *ns, *next;
 
 	list_for_each_entry_safe(ns, next, &ctrl->namespaces, list) {
-		if (ns->ns_id > nsid)
+		if (ns->head->ns_id > nsid)
 			nvme_ns_remove(ns);
 	}
 }
@@ -2583,20 +3127,29 @@ void nvme_remove_namespaces(struct nvme_ctrl *ctrl)
 }
 EXPORT_SYMBOL_GPL(nvme_remove_namespaces);
 
+static void nvme_aen_uevent(struct nvme_ctrl *ctrl)
+{
+	char *envp[2] = { NULL, NULL };
+	u32 aen_result = ctrl->aen_result;
+
+	ctrl->aen_result = 0;
+	if (!aen_result)
+		return;
+
+	envp[0] = kasprintf(GFP_KERNEL, "NVME_AEN=%#08x", aen_result);
+	if (!envp[0])
+		return;
+	kobject_uevent_env(&ctrl->device->kobj, KOBJ_CHANGE, envp);
+	kfree(envp[0]);
+}
+
 static void nvme_async_event_work(struct work_struct *work)
 {
 	struct nvme_ctrl *ctrl =
 		container_of(work, struct nvme_ctrl, async_event_work);
 
-	spin_lock_irq(&ctrl->lock);
-	while (ctrl->state == NVME_CTRL_LIVE && ctrl->event_limit > 0) {
-		int aer_idx = --ctrl->event_limit;
-
-		spin_unlock_irq(&ctrl->lock);
-		ctrl->ops->submit_async_event(ctrl, aer_idx);
-		spin_lock_irq(&ctrl->lock);
-	}
-	spin_unlock_irq(&ctrl->lock);
+	nvme_aen_uevent(ctrl);
+	ctrl->ops->submit_async_event(ctrl);
 }
 
 static bool nvme_ctrl_pp_status(struct nvme_ctrl *ctrl)
@@ -2615,18 +3168,13 @@ static bool nvme_ctrl_pp_status(struct nvme_ctrl *ctrl)
 
 static void nvme_get_fw_slot_info(struct nvme_ctrl *ctrl)
 {
-	struct nvme_command c = { };
 	struct nvme_fw_slot_info_log *log;
 
 	log = kmalloc(sizeof(*log), GFP_KERNEL);
 	if (!log)
 		return;
 
-	c.common.opcode = nvme_admin_get_log_page;
-	c.common.nsid = cpu_to_le32(NVME_NSID_ALL);
-	c.common.cdw10[0] = nvme_get_log_dw10(NVME_LOG_FW_SLOT, sizeof(*log));
-
-	if (!nvme_submit_sync_cmd(ctrl->admin_q, &c, log, sizeof(*log)))
+	if (nvme_get_log(ctrl, NVME_LOG_FW_SLOT, log, sizeof(*log)))
 		dev_warn(ctrl->device,
 				"Get FW SLOT INFO log error\n");
 	kfree(log);
@@ -2660,7 +3208,7 @@ static void nvme_fw_act_work(struct work_struct *work)
 		return;
 
 	nvme_start_queues(ctrl);
-	/* read FW slot informationi to clear the AER*/
+	/* read FW slot information to clear the AER */
 	nvme_get_fw_slot_info(ctrl);
 }
 
@@ -2668,24 +3216,21 @@ void nvme_complete_async_event(struct nvme_ctrl *ctrl, __le16 status,
 		union nvme_result *res)
 {
 	u32 result = le32_to_cpu(res->u32);
-	bool done = true;
 
-	switch (le16_to_cpu(status) >> 1) {
-	case NVME_SC_SUCCESS:
-		done = false;
-		/*FALLTHRU*/
-	case NVME_SC_ABORT_REQ:
-		++ctrl->event_limit;
-		if (ctrl->state == NVME_CTRL_LIVE)
-			queue_work(nvme_wq, &ctrl->async_event_work);
+	if (le16_to_cpu(status) >> 1 != NVME_SC_SUCCESS)
+		return;
+
+	switch (result & 0x7) {
+	case NVME_AER_ERROR:
+	case NVME_AER_SMART:
+	case NVME_AER_CSS:
+	case NVME_AER_VS:
+		ctrl->aen_result = result;
 		break;
 	default:
 		break;
 	}
 
-	if (done)
-		return;
-
 	switch (result & 0xff07) {
 	case NVME_AER_NOTICE_NS_CHANGED:
 		dev_info(ctrl->device, "rescanning\n");
@@ -2697,44 +3242,9 @@ void nvme_complete_async_event(struct nvme_ctrl *ctrl, __le16 status,
 	default:
 		dev_warn(ctrl->device, "async event result %08x\n", result);
 	}
-}
-EXPORT_SYMBOL_GPL(nvme_complete_async_event);
-
-void nvme_queue_async_events(struct nvme_ctrl *ctrl)
-{
-	ctrl->event_limit = NVME_NR_AERS;
 	queue_work(nvme_wq, &ctrl->async_event_work);
 }
-EXPORT_SYMBOL_GPL(nvme_queue_async_events);
-
-static DEFINE_IDA(nvme_instance_ida);
-
-static int nvme_set_instance(struct nvme_ctrl *ctrl)
-{
-	int instance, error;
-
-	do {
-		if (!ida_pre_get(&nvme_instance_ida, GFP_KERNEL))
-			return -ENODEV;
-
-		spin_lock(&dev_list_lock);
-		error = ida_get_new(&nvme_instance_ida, &instance);
-		spin_unlock(&dev_list_lock);
-	} while (error == -EAGAIN);
-
-	if (error)
-		return -ENODEV;
-
-	ctrl->instance = instance;
-	return 0;
-}
-
-static void nvme_release_instance(struct nvme_ctrl *ctrl)
-{
-	spin_lock(&dev_list_lock);
-	ida_remove(&nvme_instance_ida, ctrl->instance);
-	spin_unlock(&dev_list_lock);
-}
+EXPORT_SYMBOL_GPL(nvme_complete_async_event);
 
 void nvme_stop_ctrl(struct nvme_ctrl *ctrl)
 {
@@ -2752,7 +3262,7 @@ void nvme_start_ctrl(struct nvme_ctrl *ctrl)
 
 	if (ctrl->queue_count > 1) {
 		nvme_queue_scan(ctrl);
-		nvme_queue_async_events(ctrl);
+		queue_work(nvme_wq, &ctrl->async_event_work);
 		nvme_start_queues(ctrl);
 	}
 }
@@ -2760,30 +3270,31 @@ EXPORT_SYMBOL_GPL(nvme_start_ctrl);
 
 void nvme_uninit_ctrl(struct nvme_ctrl *ctrl)
 {
-	device_destroy(nvme_class, MKDEV(nvme_char_major, ctrl->instance));
-
-	spin_lock(&dev_list_lock);
-	list_del(&ctrl->node);
-	spin_unlock(&dev_list_lock);
+	cdev_device_del(&ctrl->cdev, ctrl->device);
 }
 EXPORT_SYMBOL_GPL(nvme_uninit_ctrl);
 
-static void nvme_free_ctrl(struct kref *kref)
+static void nvme_free_ctrl(struct device *dev)
 {
-	struct nvme_ctrl *ctrl = container_of(kref, struct nvme_ctrl, kref);
+	struct nvme_ctrl *ctrl =
+		container_of(dev, struct nvme_ctrl, ctrl_device);
+	struct nvme_subsystem *subsys = ctrl->subsys;
 
-	put_device(ctrl->device);
-	nvme_release_instance(ctrl);
-	ida_destroy(&ctrl->ns_ida);
+	ida_simple_remove(&nvme_instance_ida, ctrl->instance);
+	kfree(ctrl->effects);
+
+	if (subsys) {
+		mutex_lock(&subsys->lock);
+		list_del(&ctrl->subsys_entry);
+		mutex_unlock(&subsys->lock);
+		sysfs_remove_link(&subsys->dev.kobj, dev_name(ctrl->device));
+	}
 
 	ctrl->ops->free_ctrl(ctrl);
-}
 
-void nvme_put_ctrl(struct nvme_ctrl *ctrl)
-{
-	kref_put(&ctrl->kref, nvme_free_ctrl);
+	if (subsys)
+		nvme_put_subsystem(subsys);
 }
-EXPORT_SYMBOL_GPL(nvme_put_ctrl);
 
 /*
  * Initialize a NVMe controller structures.  This needs to be called during
@@ -2799,32 +3310,36 @@ int nvme_init_ctrl(struct nvme_ctrl *ctrl, struct device *dev,
 	spin_lock_init(&ctrl->lock);
 	INIT_LIST_HEAD(&ctrl->namespaces);
 	mutex_init(&ctrl->namespaces_mutex);
-	kref_init(&ctrl->kref);
 	ctrl->dev = dev;
 	ctrl->ops = ops;
 	ctrl->quirks = quirks;
 	INIT_WORK(&ctrl->scan_work, nvme_scan_work);
 	INIT_WORK(&ctrl->async_event_work, nvme_async_event_work);
 	INIT_WORK(&ctrl->fw_act_work, nvme_fw_act_work);
+	INIT_WORK(&ctrl->delete_work, nvme_delete_ctrl_work);
 
-	ret = nvme_set_instance(ctrl);
-	if (ret)
+	ret = ida_simple_get(&nvme_instance_ida, 0, 0, GFP_KERNEL);
+	if (ret < 0)
 		goto out;
-
-	ctrl->device = device_create_with_groups(nvme_class, ctrl->dev,
-				MKDEV(nvme_char_major, ctrl->instance),
-				ctrl, nvme_dev_attr_groups,
-				"nvme%d", ctrl->instance);
-	if (IS_ERR(ctrl->device)) {
-		ret = PTR_ERR(ctrl->device);
+	ctrl->instance = ret;
+
+	device_initialize(&ctrl->ctrl_device);
+	ctrl->device = &ctrl->ctrl_device;
+	ctrl->device->devt = MKDEV(MAJOR(nvme_chr_devt), ctrl->instance);
+	ctrl->device->class = nvme_class;
+	ctrl->device->parent = ctrl->dev;
+	ctrl->device->groups = nvme_dev_attr_groups;
+	ctrl->device->release = nvme_free_ctrl;
+	dev_set_drvdata(ctrl->device, ctrl);
+	ret = dev_set_name(ctrl->device, "nvme%d", ctrl->instance);
+	if (ret)
 		goto out_release_instance;
-	}
-	get_device(ctrl->device);
-	ida_init(&ctrl->ns_ida);
 
-	spin_lock(&dev_list_lock);
-	list_add_tail(&ctrl->node, &nvme_ctrl_list);
-	spin_unlock(&dev_list_lock);
+	cdev_init(&ctrl->cdev, &nvme_dev_fops);
+	ctrl->cdev.owner = ops->module;
+	ret = cdev_device_add(&ctrl->cdev, ctrl->device);
+	if (ret)
+		goto out_free_name;
 
 	/*
 	 * Initialize latency tolerance controls.  The sysfs files won't
@@ -2835,8 +3350,10 @@ int nvme_init_ctrl(struct nvme_ctrl *ctrl, struct device *dev,
 		min(default_ps_max_latency_us, (unsigned long)S32_MAX));
 
 	return 0;
+out_free_name:
+	kfree_const(dev->kobj.name);
 out_release_instance:
-	nvme_release_instance(ctrl);
+	ida_simple_remove(&nvme_instance_ida, ctrl->instance);
 out:
 	return ret;
 }
@@ -2945,6 +3462,16 @@ void nvme_start_queues(struct nvme_ctrl *ctrl)
 }
 EXPORT_SYMBOL_GPL(nvme_start_queues);
 
+int nvme_reinit_tagset(struct nvme_ctrl *ctrl, struct blk_mq_tag_set *set)
+{
+	if (!ctrl->ops->reinit_request)
+		return 0;
+
+	return blk_mq_tagset_iter(set, set->driver_data,
+			ctrl->ops->reinit_request);
+}
+EXPORT_SYMBOL_GPL(nvme_reinit_tagset);
+
 int __init nvme_core_init(void)
 {
 	int result;
@@ -2954,12 +3481,9 @@ int __init nvme_core_init(void)
 	if (!nvme_wq)
 		return -ENOMEM;
 
-	result = __register_chrdev(nvme_char_major, 0, NVME_MINORS, "nvme",
-							&nvme_dev_fops);
+	result = alloc_chrdev_region(&nvme_chr_devt, 0, NVME_MINORS, "nvme");
 	if (result < 0)
 		goto destroy_wq;
-	else if (result > 0)
-		nvme_char_major = result;
 
 	nvme_class = class_create(THIS_MODULE, "nvme");
 	if (IS_ERR(nvme_class)) {
@@ -2967,10 +3491,17 @@ int __init nvme_core_init(void)
 		goto unregister_chrdev;
 	}
 
+	nvme_subsys_class = class_create(THIS_MODULE, "nvme-subsystem");
+	if (IS_ERR(nvme_subsys_class)) {
+		result = PTR_ERR(nvme_subsys_class);
+		goto destroy_class;
+	}
 	return 0;
 
+destroy_class:
+	class_destroy(nvme_class);
 unregister_chrdev:
-	__unregister_chrdev(nvme_char_major, 0, NVME_MINORS, "nvme");
+	unregister_chrdev_region(nvme_chr_devt, NVME_MINORS);
 destroy_wq:
 	destroy_workqueue(nvme_wq);
 	return result;
@@ -2978,8 +3509,10 @@ destroy_wq:
 
 void nvme_core_exit(void)
 {
+	ida_destroy(&nvme_subsystems_ida);
+	class_destroy(nvme_subsys_class);
 	class_destroy(nvme_class);
-	__unregister_chrdev(nvme_char_major, 0, NVME_MINORS, "nvme");
+	unregister_chrdev_region(nvme_chr_devt, NVME_MINORS);
 	destroy_workqueue(nvme_wq);
 }
 

drivers/nvme/host/fabrics.c

@@ -548,6 +548,7 @@ static const match_table_t opt_tokens = {
 	{ NVMF_OPT_HOSTNQN,		"hostnqn=%s"		},
 	{ NVMF_OPT_HOST_TRADDR,		"host_traddr=%s"	},
 	{ NVMF_OPT_HOST_ID,		"hostid=%s"		},
+	{ NVMF_OPT_DUP_CONNECT,		"duplicate_connect"	},
 	{ NVMF_OPT_ERR,			NULL			}
 };
 
@@ -566,6 +567,7 @@ static int nvmf_parse_options(struct nvmf_ctrl_options *opts,
 	opts->nr_io_queues = num_online_cpus();
 	opts->reconnect_delay = NVMF_DEF_RECONNECT_DELAY;
 	opts->kato = NVME_DEFAULT_KATO;
+	opts->duplicate_connect = false;
 
 	options = o = kstrdup(buf, GFP_KERNEL);
 	if (!options)
@@ -742,6 +744,9 @@ static int nvmf_parse_options(struct nvmf_ctrl_options *opts,
 				goto out;
 			}
 			break;
+		case NVMF_OPT_DUP_CONNECT:
+			opts->duplicate_connect = true;
+			break;
 		default:
 			pr_warn("unknown parameter or missing value '%s' in ctrl creation request\n",
 				p);
@@ -823,7 +828,7 @@ EXPORT_SYMBOL_GPL(nvmf_free_options);
 #define NVMF_REQUIRED_OPTS	(NVMF_OPT_TRANSPORT | NVMF_OPT_NQN)
 #define NVMF_ALLOWED_OPTS	(NVMF_OPT_QUEUE_SIZE | NVMF_OPT_NR_IO_QUEUES | \
 				 NVMF_OPT_KATO | NVMF_OPT_HOSTNQN | \
-				 NVMF_OPT_HOST_ID)
+				 NVMF_OPT_HOST_ID | NVMF_OPT_DUP_CONNECT)
 
 static struct nvme_ctrl *
 nvmf_create_ctrl(struct device *dev, const char *buf, size_t count)
@@ -841,6 +846,9 @@ nvmf_create_ctrl(struct device *dev, const char *buf, size_t count)
 	if (ret)
 		goto out_free_opts;
 
+
+	request_module("nvme-%s", opts->transport);
+
 	/*
 	 * Check the generic options first as we need a valid transport for
 	 * the lookup below.  Then clear the generic flags so that transport
@@ -874,12 +882,12 @@ nvmf_create_ctrl(struct device *dev, const char *buf, size_t count)
 		goto out_unlock;
 	}
 
-	if (strcmp(ctrl->subnqn, opts->subsysnqn)) {
+	if (strcmp(ctrl->subsys->subnqn, opts->subsysnqn)) {
 		dev_warn(ctrl->device,
 			"controller returned incorrect NQN: \"%s\".\n",
-			ctrl->subnqn);
+			ctrl->subsys->subnqn);
 		up_read(&nvmf_transports_rwsem);
-		ctrl->ops->delete_ctrl(ctrl);
+		nvme_delete_ctrl_sync(ctrl);
 		return ERR_PTR(-EINVAL);
 	}
 

drivers/nvme/host/fabrics.h

@@ -57,6 +57,7 @@ enum {
 	NVMF_OPT_HOST_TRADDR	= 1 << 10,
 	NVMF_OPT_CTRL_LOSS_TMO	= 1 << 11,
 	NVMF_OPT_HOST_ID	= 1 << 12,
+	NVMF_OPT_DUP_CONNECT	= 1 << 13,
 };
 
 /**
@@ -96,6 +97,7 @@ struct nvmf_ctrl_options {
 	unsigned int		nr_io_queues;
 	unsigned int		reconnect_delay;
 	bool			discovery_nqn;
+	bool			duplicate_connect;
 	unsigned int		kato;
 	struct nvmf_host	*host;
 	int			max_reconnects;
@@ -131,6 +133,18 @@ struct nvmf_transport_ops {
 					struct nvmf_ctrl_options *opts);
 };
 
+static inline bool
+nvmf_ctlr_matches_baseopts(struct nvme_ctrl *ctrl,
+			struct nvmf_ctrl_options *opts)
+{
+	if (strcmp(opts->subsysnqn, ctrl->opts->subsysnqn) ||
+	    strcmp(opts->host->nqn, ctrl->opts->host->nqn) ||
+	    memcmp(&opts->host->id, &ctrl->opts->host->id, sizeof(uuid_t)))
+		return false;
+
+	return true;
+}
+
 int nvmf_reg_read32(struct nvme_ctrl *ctrl, u32 off, u32 *val);
 int nvmf_reg_read64(struct nvme_ctrl *ctrl, u32 off, u64 *val);
 int nvmf_reg_write32(struct nvme_ctrl *ctrl, u32 off, u32 val);

drivers/nvme/host/fc.c

@@ -30,27 +30,19 @@
 /* *************************** Data Structures/Defines ****************** */
 
 
-/*
- * We handle AEN commands ourselves and don't even let the
- * block layer know about them.
- */
-#define NVME_FC_NR_AEN_COMMANDS	1
-#define NVME_FC_AQ_BLKMQ_DEPTH	\
-	(NVME_AQ_DEPTH - NVME_FC_NR_AEN_COMMANDS)
-#define AEN_CMDID_BASE		(NVME_FC_AQ_BLKMQ_DEPTH + 1)
-
 enum nvme_fc_queue_flags {
 	NVME_FC_Q_CONNECTED = (1 << 0),
 };
 
 #define NVMEFC_QUEUE_DELAY	3		/* ms units */
 
+#define NVME_FC_DEFAULT_DEV_LOSS_TMO	60	/* seconds */
+
 struct nvme_fc_queue {
 	struct nvme_fc_ctrl	*ctrl;
 	struct device		*dev;
 	struct blk_mq_hw_ctx	*hctx;
 	void			*lldd_handle;
-	int			queue_size;
 	size_t			cmnd_capsule_len;
 	u32			qnum;
 	u32			rqcnt;
@@ -124,6 +116,7 @@ struct nvme_fc_lport {
 	struct device			*dev;	/* physical device for dma */
 	struct nvme_fc_port_template	*ops;
 	struct kref			ref;
+	atomic_t                        act_rport_cnt;
 } __aligned(sizeof(u64));	/* alignment for other things alloc'd with */
 
 struct nvme_fc_rport {
@@ -136,6 +129,8 @@ struct nvme_fc_rport {
 	struct nvme_fc_lport		*lport;
 	spinlock_t			lock;
 	struct kref			ref;
+	atomic_t                        act_ctrl_cnt;
+	unsigned long			dev_loss_end;
 } __aligned(sizeof(u64));	/* alignment for other things alloc'd with */
 
 enum nvme_fcctrl_flags {
@@ -150,6 +145,7 @@ struct nvme_fc_ctrl {
 	struct nvme_fc_rport	*rport;
 	u32			cnum;
 
+	bool			assoc_active;
 	u64			association_id;
 
 	struct list_head	ctrl_list;	/* rport->ctrl_list */
@@ -157,7 +153,6 @@ struct nvme_fc_ctrl {
 	struct blk_mq_tag_set	admin_tag_set;
 	struct blk_mq_tag_set	tag_set;
 
-	struct work_struct	delete_work;
 	struct delayed_work	connect_work;
 
 	struct kref		ref;
@@ -165,7 +160,7 @@ struct nvme_fc_ctrl {
 	u32			iocnt;
 	wait_queue_head_t	ioabort_wait;
 
-	struct nvme_fc_fcp_op	aen_ops[NVME_FC_NR_AEN_COMMANDS];
+	struct nvme_fc_fcp_op	aen_ops[NVME_NR_AEN_COMMANDS];
 
 	struct nvme_ctrl	ctrl;
 };
@@ -213,10 +208,16 @@ static DEFINE_IDA(nvme_fc_ctrl_cnt);
 
 
 
+/*
+ * These items are short-term. They will eventually be moved into
+ * a generic FC class. See comments in module init.
+ */
+static struct class *fc_class;
+static struct device *fc_udev_device;
+
 
 /* *********************** FC-NVME Port Management ************************ */
 
-static int __nvme_fc_del_ctrl(struct nvme_fc_ctrl *);
 static void __nvme_fc_delete_hw_queue(struct nvme_fc_ctrl *,
 			struct nvme_fc_queue *, unsigned int);
 
@@ -235,9 +236,6 @@ nvme_fc_free_lport(struct kref *ref)
 	list_del(&lport->port_list);
 	spin_unlock_irqrestore(&nvme_fc_lock, flags);
 
-	/* let the LLDD know we've finished tearing it down */
-	lport->ops->localport_delete(&lport->localport);
-
 	ida_simple_remove(&nvme_fc_local_port_cnt, lport->localport.port_num);
 	ida_destroy(&lport->endp_cnt);
 
@@ -260,7 +258,9 @@ nvme_fc_lport_get(struct nvme_fc_lport *lport)
 
 
 static struct nvme_fc_lport *
-nvme_fc_attach_to_unreg_lport(struct nvme_fc_port_info *pinfo)
+nvme_fc_attach_to_unreg_lport(struct nvme_fc_port_info *pinfo,
+			struct nvme_fc_port_template *ops,
+			struct device *dev)
 {
 	struct nvme_fc_lport *lport;
 	unsigned long flags;
@@ -272,6 +272,11 @@ nvme_fc_attach_to_unreg_lport(struct nvme_fc_port_info *pinfo)
 		    lport->localport.port_name != pinfo->port_name)
 			continue;
 
+		if (lport->dev != dev) {
+			lport = ERR_PTR(-EXDEV);
+			goto out_done;
+		}
+
 		if (lport->localport.port_state != FC_OBJSTATE_DELETED) {
 			lport = ERR_PTR(-EEXIST);
 			goto out_done;
@@ -288,6 +293,7 @@ nvme_fc_attach_to_unreg_lport(struct nvme_fc_port_info *pinfo)
 
 		/* resume the lport */
 
+		lport->ops = ops;
 		lport->localport.port_role = pinfo->port_role;
 		lport->localport.port_id = pinfo->port_id;
 		lport->localport.port_state = FC_OBJSTATE_ONLINE;
@@ -348,7 +354,7 @@ nvme_fc_register_localport(struct nvme_fc_port_info *pinfo,
 	 * expired, we can simply re-enable the localport. Remoteports
 	 * and controller reconnections should resume naturally.
 	 */
-	newrec = nvme_fc_attach_to_unreg_lport(pinfo);
+	newrec = nvme_fc_attach_to_unreg_lport(pinfo, template, dev);
 
 	/* found an lport, but something about its state is bad */
 	if (IS_ERR(newrec)) {
@@ -384,6 +390,7 @@ nvme_fc_register_localport(struct nvme_fc_port_info *pinfo,
 	INIT_LIST_HEAD(&newrec->port_list);
 	INIT_LIST_HEAD(&newrec->endp_list);
 	kref_init(&newrec->ref);
+	atomic_set(&newrec->act_rport_cnt, 0);
 	newrec->ops = template;
 	newrec->dev = dev;
 	ida_init(&newrec->endp_cnt);
@@ -446,12 +453,177 @@ nvme_fc_unregister_localport(struct nvme_fc_local_port *portptr)
 
 	spin_unlock_irqrestore(&nvme_fc_lock, flags);
 
+	if (atomic_read(&lport->act_rport_cnt) == 0)
+		lport->ops->localport_delete(&lport->localport);
+
 	nvme_fc_lport_put(lport);
 
 	return 0;
 }
 EXPORT_SYMBOL_GPL(nvme_fc_unregister_localport);
 
+/*
+ * TRADDR strings, per FC-NVME are fixed format:
+ *   "nn-0x<16hexdigits>:pn-0x<16hexdigits>" - 43 characters
+ * udev event will only differ by prefix of what field is
+ * being specified:
+ *    "NVMEFC_HOST_TRADDR=" or "NVMEFC_TRADDR=" - 19 max characters
+ *  19 + 43 + null_fudge = 64 characters
+ */
+#define FCNVME_TRADDR_LENGTH		64
+
+static void
+nvme_fc_signal_discovery_scan(struct nvme_fc_lport *lport,
+		struct nvme_fc_rport *rport)
+{
+	char hostaddr[FCNVME_TRADDR_LENGTH];	/* NVMEFC_HOST_TRADDR=...*/
+	char tgtaddr[FCNVME_TRADDR_LENGTH];	/* NVMEFC_TRADDR=...*/
+	char *envp[4] = { "FC_EVENT=nvmediscovery", hostaddr, tgtaddr, NULL };
+
+	if (!(rport->remoteport.port_role & FC_PORT_ROLE_NVME_DISCOVERY))
+		return;
+
+	snprintf(hostaddr, sizeof(hostaddr),
+		"NVMEFC_HOST_TRADDR=nn-0x%016llx:pn-0x%016llx",
+		lport->localport.node_name, lport->localport.port_name);
+	snprintf(tgtaddr, sizeof(tgtaddr),
+		"NVMEFC_TRADDR=nn-0x%016llx:pn-0x%016llx",
+		rport->remoteport.node_name, rport->remoteport.port_name);
+	kobject_uevent_env(&fc_udev_device->kobj, KOBJ_CHANGE, envp);
+}
+
+static void
+nvme_fc_free_rport(struct kref *ref)
+{
+	struct nvme_fc_rport *rport =
+		container_of(ref, struct nvme_fc_rport, ref);
+	struct nvme_fc_lport *lport =
+			localport_to_lport(rport->remoteport.localport);
+	unsigned long flags;
+
+	WARN_ON(rport->remoteport.port_state != FC_OBJSTATE_DELETED);
+	WARN_ON(!list_empty(&rport->ctrl_list));
+
+	/* remove from lport list */
+	spin_lock_irqsave(&nvme_fc_lock, flags);
+	list_del(&rport->endp_list);
+	spin_unlock_irqrestore(&nvme_fc_lock, flags);
+
+	ida_simple_remove(&lport->endp_cnt, rport->remoteport.port_num);
+
+	kfree(rport);
+
+	nvme_fc_lport_put(lport);
+}
+
+static void
+nvme_fc_rport_put(struct nvme_fc_rport *rport)
+{
+	kref_put(&rport->ref, nvme_fc_free_rport);
+}
+
+static int
+nvme_fc_rport_get(struct nvme_fc_rport *rport)
+{
+	return kref_get_unless_zero(&rport->ref);
+}
+
+static void
+nvme_fc_resume_controller(struct nvme_fc_ctrl *ctrl)
+{
+	switch (ctrl->ctrl.state) {
+	case NVME_CTRL_NEW:
+	case NVME_CTRL_RECONNECTING:
+		/*
+		 * As all reconnects were suppressed, schedule a
+		 * connect.
+		 */
+		dev_info(ctrl->ctrl.device,
+			"NVME-FC{%d}: connectivity re-established. "
+			"Attempting reconnect\n", ctrl->cnum);
+
+		queue_delayed_work(nvme_wq, &ctrl->connect_work, 0);
+		break;
+
+	case NVME_CTRL_RESETTING:
+		/*
+		 * Controller is already in the process of terminating the
+		 * association. No need to do anything further. The reconnect
+		 * step will naturally occur after the reset completes.
+		 */
+		break;
+
+	default:
+		/* no action to take - let it delete */
+		break;
+	}
+}
+
+static struct nvme_fc_rport *
+nvme_fc_attach_to_suspended_rport(struct nvme_fc_lport *lport,
+				struct nvme_fc_port_info *pinfo)
+{
+	struct nvme_fc_rport *rport;
+	struct nvme_fc_ctrl *ctrl;
+	unsigned long flags;
+
+	spin_lock_irqsave(&nvme_fc_lock, flags);
+
+	list_for_each_entry(rport, &lport->endp_list, endp_list) {
+		if (rport->remoteport.node_name != pinfo->node_name ||
+		    rport->remoteport.port_name != pinfo->port_name)
+			continue;
+
+		if (!nvme_fc_rport_get(rport)) {
+			rport = ERR_PTR(-ENOLCK);
+			goto out_done;
+		}
+
+		spin_unlock_irqrestore(&nvme_fc_lock, flags);
+
+		spin_lock_irqsave(&rport->lock, flags);
+
+		/* has it been unregistered */
+		if (rport->remoteport.port_state != FC_OBJSTATE_DELETED) {
+			/* means lldd called us twice */
+			spin_unlock_irqrestore(&rport->lock, flags);
+			nvme_fc_rport_put(rport);
+			return ERR_PTR(-ESTALE);
+		}
+
+		rport->remoteport.port_state = FC_OBJSTATE_ONLINE;
+		rport->dev_loss_end = 0;
+
+		/*
+		 * kick off a reconnect attempt on all associations to the
+		 * remote port. A successful reconnects will resume i/o.
+		 */
+		list_for_each_entry(ctrl, &rport->ctrl_list, ctrl_list)
+			nvme_fc_resume_controller(ctrl);
+
+		spin_unlock_irqrestore(&rport->lock, flags);
+
+		return rport;
+	}
+
+	rport = NULL;
+
+out_done:
+	spin_unlock_irqrestore(&nvme_fc_lock, flags);
+
+	return rport;
+}
+
+static inline void
+__nvme_fc_set_dev_loss_tmo(struct nvme_fc_rport *rport,
+			struct nvme_fc_port_info *pinfo)
+{
+	if (pinfo->dev_loss_tmo)
+		rport->remoteport.dev_loss_tmo = pinfo->dev_loss_tmo;
+	else
+		rport->remoteport.dev_loss_tmo = NVME_FC_DEFAULT_DEV_LOSS_TMO;
+}
+
 /**
  * nvme_fc_register_remoteport - transport entry point called by an
  *                              LLDD to register the existence of a NVME
@@ -478,28 +650,52 @@ nvme_fc_register_remoteport(struct nvme_fc_local_port *localport,
 	unsigned long flags;
 	int ret, idx;
 
+	if (!nvme_fc_lport_get(lport)) {
+		ret = -ESHUTDOWN;
+		goto out_reghost_failed;
+	}
+
+	/*
+	 * look to see if there is already a remoteport that is waiting
+	 * for a reconnect (within dev_loss_tmo) with the same WWN's.
+	 * If so, transition to it and reconnect.
+	 */
+	newrec = nvme_fc_attach_to_suspended_rport(lport, pinfo);
+
+	/* found an rport, but something about its state is bad */
+	if (IS_ERR(newrec)) {
+		ret = PTR_ERR(newrec);
+		goto out_lport_put;
+
+	/* found existing rport, which was resumed */
+	} else if (newrec) {
+		nvme_fc_lport_put(lport);
+		__nvme_fc_set_dev_loss_tmo(newrec, pinfo);
+		nvme_fc_signal_discovery_scan(lport, newrec);
+		*portptr = &newrec->remoteport;
+		return 0;
+	}
+
+	/* nothing found - allocate a new remoteport struct */
+
 	newrec = kmalloc((sizeof(*newrec) + lport->ops->remote_priv_sz),
 			 GFP_KERNEL);
 	if (!newrec) {
 		ret = -ENOMEM;
-		goto out_reghost_failed;
-	}
-
-	if (!nvme_fc_lport_get(lport)) {
-		ret = -ESHUTDOWN;
-		goto out_kfree_rport;
+		goto out_lport_put;
 	}
 
 	idx = ida_simple_get(&lport->endp_cnt, 0, 0, GFP_KERNEL);
 	if (idx < 0) {
 		ret = -ENOSPC;
-		goto out_lport_put;
+		goto out_kfree_rport;
 	}
 
 	INIT_LIST_HEAD(&newrec->endp_list);
 	INIT_LIST_HEAD(&newrec->ctrl_list);
 	INIT_LIST_HEAD(&newrec->ls_req_list);
 	kref_init(&newrec->ref);
+	atomic_set(&newrec->act_ctrl_cnt, 0);
 	spin_lock_init(&newrec->lock);
 	newrec->remoteport.localport = &lport->localport;
 	newrec->dev = lport->dev;
@@ -511,63 +707,27 @@ nvme_fc_register_remoteport(struct nvme_fc_local_port *localport,
 	newrec->remoteport.port_id = pinfo->port_id;
 	newrec->remoteport.port_state = FC_OBJSTATE_ONLINE;
 	newrec->remoteport.port_num = idx;
+	__nvme_fc_set_dev_loss_tmo(newrec, pinfo);
 
 	spin_lock_irqsave(&nvme_fc_lock, flags);
 	list_add_tail(&newrec->endp_list, &lport->endp_list);
 	spin_unlock_irqrestore(&nvme_fc_lock, flags);
 
+	nvme_fc_signal_discovery_scan(lport, newrec);
+
 	*portptr = &newrec->remoteport;
 	return 0;
 
-out_lport_put:
-	nvme_fc_lport_put(lport);
 out_kfree_rport:
 	kfree(newrec);
+out_lport_put:
+	nvme_fc_lport_put(lport);
 out_reghost_failed:
 	*portptr = NULL;
 	return ret;
 }
 EXPORT_SYMBOL_GPL(nvme_fc_register_remoteport);
 
-static void
-nvme_fc_free_rport(struct kref *ref)
-{
-	struct nvme_fc_rport *rport =
-		container_of(ref, struct nvme_fc_rport, ref);
-	struct nvme_fc_lport *lport =
-			localport_to_lport(rport->remoteport.localport);
-	unsigned long flags;
-
-	WARN_ON(rport->remoteport.port_state != FC_OBJSTATE_DELETED);
-	WARN_ON(!list_empty(&rport->ctrl_list));
-
-	/* remove from lport list */
-	spin_lock_irqsave(&nvme_fc_lock, flags);
-	list_del(&rport->endp_list);
-	spin_unlock_irqrestore(&nvme_fc_lock, flags);
-
-	/* let the LLDD know we've finished tearing it down */
-	lport->ops->remoteport_delete(&rport->remoteport);
-
-	ida_simple_remove(&lport->endp_cnt, rport->remoteport.port_num);
-
-	kfree(rport);
-
-	nvme_fc_lport_put(lport);
-}
-
-static void
-nvme_fc_rport_put(struct nvme_fc_rport *rport)
-{
-	kref_put(&rport->ref, nvme_fc_free_rport);
-}
-
-static int
-nvme_fc_rport_get(struct nvme_fc_rport *rport)
-{
-	return kref_get_unless_zero(&rport->ref);
-}
-
 static int
 nvme_fc_abort_lsops(struct nvme_fc_rport *rport)
 {
@@ -592,6 +752,58 @@ restart:
 	return 0;
 }
 
+static void
+nvme_fc_ctrl_connectivity_loss(struct nvme_fc_ctrl *ctrl)
+{
+	dev_info(ctrl->ctrl.device,
+		"NVME-FC{%d}: controller connectivity lost. Awaiting "
+		"Reconnect", ctrl->cnum);
+
+	switch (ctrl->ctrl.state) {
+	case NVME_CTRL_NEW:
+	case NVME_CTRL_LIVE:
+		/*
+		 * Schedule a controller reset. The reset will terminate the
+		 * association and schedule the reconnect timer.  Reconnects
+		 * will be attempted until either the ctlr_loss_tmo
+		 * (max_retries * connect_delay) expires or the remoteport's
+		 * dev_loss_tmo expires.
+		 */
+		if (nvme_reset_ctrl(&ctrl->ctrl)) {
+			dev_warn(ctrl->ctrl.device,
+				"NVME-FC{%d}: Couldn't schedule reset. "
+				"Deleting controller.\n",
+				ctrl->cnum);
+			nvme_delete_ctrl(&ctrl->ctrl);
+		}
+		break;
+
+	case NVME_CTRL_RECONNECTING:
+		/*
+		 * The association has already been terminated and the
+		 * controller is attempting reconnects.  No need to do anything
+		 * futher.  Reconnects will be attempted until either the
+		 * ctlr_loss_tmo (max_retries * connect_delay) expires or the
+		 * remoteport's dev_loss_tmo expires.
+		 */
+		break;
+
+	case NVME_CTRL_RESETTING:
+		/*
+		 * Controller is already in the process of terminating the
+		 * association.  No need to do anything further. The reconnect
+		 * step will kick in naturally after the association is
+		 * terminated.
+		 */
+		break;
+
+	case NVME_CTRL_DELETING:
+	default:
+		/* no action to take - let it delete */
+		break;
+	}
+}
+
 /**
  * nvme_fc_unregister_remoteport - transport entry point called by an
  *                              LLDD to deregister/remove a previously
@@ -621,19 +833,78 @@ nvme_fc_unregister_remoteport(struct nvme_fc_remote_port *portptr)
 	}
 	portptr->port_state = FC_OBJSTATE_DELETED;
 
-	/* tear down all associations to the remote port */
-	list_for_each_entry(ctrl, &rport->ctrl_list, ctrl_list)
-		__nvme_fc_del_ctrl(ctrl);
+	rport->dev_loss_end = jiffies + (portptr->dev_loss_tmo * HZ);
+
+	list_for_each_entry(ctrl, &rport->ctrl_list, ctrl_list) {
+		/* if dev_loss_tmo==0, dev loss is immediate */
+		if (!portptr->dev_loss_tmo) {
+			dev_warn(ctrl->ctrl.device,
+				"NVME-FC{%d}: controller connectivity lost. "
+				"Deleting controller.\n",
+				ctrl->cnum);
+			nvme_delete_ctrl(&ctrl->ctrl);
+		} else
+			nvme_fc_ctrl_connectivity_loss(ctrl);
+	}
 
 	spin_unlock_irqrestore(&rport->lock, flags);
 
 	nvme_fc_abort_lsops(rport);
 
+	if (atomic_read(&rport->act_ctrl_cnt) == 0)
+		rport->lport->ops->remoteport_delete(portptr);
+
+	/*
+	 * release the reference, which will allow, if all controllers
+	 * go away, which should only occur after dev_loss_tmo occurs,
+	 * for the rport to be torn down.
+	 */
 	nvme_fc_rport_put(rport);
+
 	return 0;
 }
 EXPORT_SYMBOL_GPL(nvme_fc_unregister_remoteport);
 
+/**
+ * nvme_fc_rescan_remoteport - transport entry point called by an
+ *                              LLDD to request a nvme device rescan.
+ * @remoteport: pointer to the (registered) remote port that is to be
+ *              rescanned.
+ *
+ * Returns: N/A
+ */
+void
+nvme_fc_rescan_remoteport(struct nvme_fc_remote_port *remoteport)
+{
+	struct nvme_fc_rport *rport = remoteport_to_rport(remoteport);
+
+	nvme_fc_signal_discovery_scan(rport->lport, rport);
+}
+EXPORT_SYMBOL_GPL(nvme_fc_rescan_remoteport);
+
+int
+nvme_fc_set_remoteport_devloss(struct nvme_fc_remote_port *portptr,
+			u32 dev_loss_tmo)
+{
+	struct nvme_fc_rport *rport = remoteport_to_rport(portptr);
+	unsigned long flags;
+
+	spin_lock_irqsave(&rport->lock, flags);
+
+	if (portptr->port_state != FC_OBJSTATE_ONLINE) {
+		spin_unlock_irqrestore(&rport->lock, flags);
+		return -EINVAL;
+	}
+
+	/* a dev_loss_tmo of 0 (immediate) is allowed to be set */
+	rport->remoteport.dev_loss_tmo = dev_loss_tmo;
+
+	spin_unlock_irqrestore(&rport->lock, flags);
+
+	return 0;
+}
+EXPORT_SYMBOL_GPL(nvme_fc_set_remoteport_devloss);
+
 
 /* *********************** FC-NVME DMA Handling **************************** */
 
@@ -723,7 +994,6 @@ fc_dma_unmap_sg(struct device *dev, struct scatterlist *sg, int nents,
 		dma_unmap_sg(dev, sg, nents, dir);
 }
 
-
 /* *********************** FC-NVME LS Handling **************************** */
 
 static void nvme_fc_ctrl_put(struct nvme_fc_ctrl *);
@@ -1266,7 +1536,7 @@ nvme_fc_abort_aen_ops(struct nvme_fc_ctrl *ctrl)
 	unsigned long flags;
 	int i, ret;
 
-	for (i = 0; i < NVME_FC_NR_AEN_COMMANDS; i++, aen_op++) {
+	for (i = 0; i < NVME_NR_AEN_COMMANDS; i++, aen_op++) {
 		if (atomic_read(&aen_op->state) != FCPOP_STATE_ACTIVE)
 			continue;
 
@@ -1331,7 +1601,7 @@ nvme_fc_fcpio_done(struct nvmefc_fcp_req *req)
 	struct nvme_command *sqe = &op->cmd_iu.sqe;
 	__le16 status = cpu_to_le16(NVME_SC_SUCCESS << 1);
 	union nvme_result result;
-	bool complete_rq, terminate_assoc = true;
+	bool terminate_assoc = true;
 
 	/*
 	 * WARNING:
@@ -1373,8 +1643,9 @@ nvme_fc_fcpio_done(struct nvmefc_fcp_req *req)
 	fc_dma_sync_single_for_cpu(ctrl->lport->dev, op->fcp_req.rspdma,
 				sizeof(op->rsp_iu), DMA_FROM_DEVICE);
 
-	if (atomic_read(&op->state) == FCPOP_STATE_ABORTED)
-		status = cpu_to_le16((NVME_SC_ABORT_REQ | NVME_SC_DNR) << 1);
+	if (atomic_read(&op->state) == FCPOP_STATE_ABORTED ||
+			op->flags & FCOP_FLAGS_TERMIO)
+		status = cpu_to_le16(NVME_SC_ABORT_REQ << 1);
 	else if (freq->status)
 		status = cpu_to_le16(NVME_SC_INTERNAL << 1);
 
@@ -1438,23 +1709,27 @@ nvme_fc_fcpio_done(struct nvmefc_fcp_req *req)
 done:
 	if (op->flags & FCOP_FLAGS_AEN) {
 		nvme_complete_async_event(&queue->ctrl->ctrl, status, &result);
-		complete_rq = __nvme_fc_fcpop_chk_teardowns(ctrl, op);
+		__nvme_fc_fcpop_chk_teardowns(ctrl, op);
 		atomic_set(&op->state, FCPOP_STATE_IDLE);
 		op->flags = FCOP_FLAGS_AEN;	/* clear other flags */
 		nvme_fc_ctrl_put(ctrl);
 		goto check_error;
 	}
 
-	complete_rq = __nvme_fc_fcpop_chk_teardowns(ctrl, op);
-	if (!complete_rq) {
-		if (unlikely(op->flags & FCOP_FLAGS_TERMIO)) {
-			status = cpu_to_le16(NVME_SC_ABORT_REQ << 1);
-			if (blk_queue_dying(rq->q))
-				status |= cpu_to_le16(NVME_SC_DNR << 1);
-		}
-		nvme_end_request(rq, status, result);
-	} else
+	/*
+	 * Force failures of commands if we're killing the controller
+	 * or have an error on a command used to create an new association
+	 */
+	if (status &&
+	    (blk_queue_dying(rq->q) ||
+	     ctrl->ctrl.state == NVME_CTRL_NEW ||
+	     ctrl->ctrl.state == NVME_CTRL_RECONNECTING))
+		status |= cpu_to_le16(NVME_SC_DNR << 1);
+
+	if (__nvme_fc_fcpop_chk_teardowns(ctrl, op))
 		__nvme_fc_final_op_cleanup(rq);
+	else
+		nvme_end_request(rq, status, result);
 
 check_error:
 	if (terminate_assoc)
@@ -1531,7 +1806,7 @@ nvme_fc_init_aen_ops(struct nvme_fc_ctrl *ctrl)
 	int i, ret;
 
 	aen_op = ctrl->aen_ops;
-	for (i = 0; i < NVME_FC_NR_AEN_COMMANDS; i++, aen_op++) {
+	for (i = 0; i < NVME_NR_AEN_COMMANDS; i++, aen_op++) {
 		private = kzalloc(ctrl->lport->ops->fcprqst_priv_sz,
 						GFP_KERNEL);
 		if (!private)
@@ -1541,7 +1816,7 @@ nvme_fc_init_aen_ops(struct nvme_fc_ctrl *ctrl)
 		sqe = &cmdiu->sqe;
 		ret = __nvme_fc_init_request(ctrl, &ctrl->queues[0],
 				aen_op, (struct request *)NULL,
-				(AEN_CMDID_BASE + i));
+				(NVME_AQ_BLK_MQ_DEPTH + i));
 		if (ret) {
 			kfree(private);
 			return ret;
@@ -1554,7 +1829,7 @@ nvme_fc_init_aen_ops(struct nvme_fc_ctrl *ctrl)
 		memset(sqe, 0, sizeof(*sqe));
 		sqe->common.opcode = nvme_admin_async_event;
 		/* Note: core layer may overwrite the sqe.command_id value */
-		sqe->common.command_id = AEN_CMDID_BASE + i;
+		sqe->common.command_id = NVME_AQ_BLK_MQ_DEPTH + i;
 	}
 	return 0;
 }
@@ -1566,7 +1841,7 @@ nvme_fc_term_aen_ops(struct nvme_fc_ctrl *ctrl)
 	int i;
 
 	aen_op = ctrl->aen_ops;
-	for (i = 0; i < NVME_FC_NR_AEN_COMMANDS; i++, aen_op++) {
+	for (i = 0; i < NVME_NR_AEN_COMMANDS; i++, aen_op++) {
 		if (!aen_op->fcp_req.private)
 			continue;
 
@@ -1610,7 +1885,7 @@ nvme_fc_init_admin_hctx(struct blk_mq_hw_ctx *hctx, void *data,
 }
 
 static void
-nvme_fc_init_queue(struct nvme_fc_ctrl *ctrl, int idx, size_t queue_size)
+nvme_fc_init_queue(struct nvme_fc_ctrl *ctrl, int idx)
 {
 	struct nvme_fc_queue *queue;
 
@@ -1626,8 +1901,6 @@ nvme_fc_init_queue(struct nvme_fc_ctrl *ctrl, int idx, size_t queue_size)
 	else
 		queue->cmnd_capsule_len = sizeof(struct nvme_command);
 
-	queue->queue_size = queue_size;
-
 	/*
 	 * Considered whether we should allocate buffers for all SQEs
 	 * and CQEs and dma map them - mapping their respective entries
@@ -1751,7 +2024,7 @@ nvme_fc_init_io_queues(struct nvme_fc_ctrl *ctrl)
 	int i;
 
 	for (i = 1; i < ctrl->ctrl.queue_count; i++)
-		nvme_fc_init_queue(ctrl, i, ctrl->ctrl.sqsize);
+		nvme_fc_init_queue(ctrl, i);
 }
 
 static void
@@ -1825,13 +2098,6 @@ nvme_fc_error_recovery(struct nvme_fc_ctrl *ctrl, char *errmsg)
 	dev_warn(ctrl->ctrl.device,
 		"NVME-FC{%d}: resetting controller\n", ctrl->cnum);
 
-	if (!nvme_change_ctrl_state(&ctrl->ctrl, NVME_CTRL_RECONNECTING)) {
-		dev_err(ctrl->ctrl.device,
-			"NVME-FC{%d}: error_recovery: Couldn't change state "
-			"to RECONNECTING\n", ctrl->cnum);
-		return;
-	}
-
 	nvme_reset_ctrl(&ctrl->ctrl);
 }
 
@@ -1842,13 +2108,14 @@ nvme_fc_timeout(struct request *rq, bool reserved)
 	struct nvme_fc_ctrl *ctrl = op->ctrl;
 	int ret;
 
-	if (reserved)
+	if (ctrl->rport->remoteport.port_state != FC_OBJSTATE_ONLINE ||
+			atomic_read(&op->state) == FCPOP_STATE_ABORTED)
 		return BLK_EH_RESET_TIMER;
 
 	ret = __nvme_fc_abort_op(ctrl, op);
 	if (ret)
-		/* io wasn't active to abort consider it done */
-		return BLK_EH_HANDLED;
+		/* io wasn't active to abort */
+		return BLK_EH_NOT_HANDLED;
 
 	/*
 	 * we can't individually ABTS an io without affecting the queue,
@@ -1859,7 +2126,12 @@ nvme_fc_timeout(struct request *rq, bool reserved)
 	 */
 	nvme_fc_error_recovery(ctrl, "io timeout error");
 
-	return BLK_EH_HANDLED;
+	/*
+	 * the io abort has been initiated. Have the reset timer
+	 * restarted and the abort completion will complete the io
+	 * shortly. Avoids a synchronous wait while the abort finishes.
+	 */
+	return BLK_EH_RESET_TIMER;
 }
 
 static int
@@ -2110,7 +2382,7 @@ nvme_fc_poll(struct blk_mq_hw_ctx *hctx, unsigned int tag)
 }
 
 static void
-nvme_fc_submit_async_event(struct nvme_ctrl *arg, int aer_idx)
+nvme_fc_submit_async_event(struct nvme_ctrl *arg)
 {
 	struct nvme_fc_ctrl *ctrl = to_fc_ctrl(arg);
 	struct nvme_fc_fcp_op *aen_op;
@@ -2118,9 +2390,6 @@ nvme_fc_submit_async_event(struct nvme_ctrl *arg, int aer_idx)
 	bool terminating = false;
 	blk_status_t ret;
 
-	if (aer_idx > NVME_FC_NR_AEN_COMMANDS)
-		return;
-
 	spin_lock_irqsave(&ctrl->lock, flags);
 	if (ctrl->flags & FCCTRL_TERMIO)
 		terminating = true;
@@ -2129,13 +2398,13 @@ nvme_fc_submit_async_event(struct nvme_ctrl *arg, int aer_idx)
 	if (terminating)
 		return;
 
-	aen_op = &ctrl->aen_ops[aer_idx];
+	aen_op = &ctrl->aen_ops[0];
 
 	ret = nvme_fc_start_fcp_op(ctrl, aen_op->queue, aen_op, 0,
 					NVMEFC_FCP_NODATA);
 	if (ret)
 		dev_err(ctrl->ctrl.device,
-			"failed async event work [%d]\n", aer_idx);
+			"failed async event work\n");
 }
 
 static void
@@ -2337,7 +2606,7 @@ nvme_fc_reinit_io_queues(struct nvme_fc_ctrl *ctrl)
 
 	nvme_fc_init_io_queues(ctrl);
 
-	ret = blk_mq_reinit_tagset(&ctrl->tag_set, nvme_fc_reinit_request);
+	ret = nvme_reinit_tagset(&ctrl->ctrl, ctrl->ctrl.tagset);
 	if (ret)
 		goto out_free_io_queues;
 
@@ -2360,6 +2629,61 @@ out_free_io_queues:
 	return ret;
 }
 
+static void
+nvme_fc_rport_active_on_lport(struct nvme_fc_rport *rport)
+{
+	struct nvme_fc_lport *lport = rport->lport;
+
+	atomic_inc(&lport->act_rport_cnt);
+}
+
+static void
+nvme_fc_rport_inactive_on_lport(struct nvme_fc_rport *rport)
+{
+	struct nvme_fc_lport *lport = rport->lport;
+	u32 cnt;
+
+	cnt = atomic_dec_return(&lport->act_rport_cnt);
+	if (cnt == 0 && lport->localport.port_state == FC_OBJSTATE_DELETED)
+		lport->ops->localport_delete(&lport->localport);
+}
+
+static int
+nvme_fc_ctlr_active_on_rport(struct nvme_fc_ctrl *ctrl)
+{
+	struct nvme_fc_rport *rport = ctrl->rport;
+	u32 cnt;
+
+	if (ctrl->assoc_active)
+		return 1;
+
+	ctrl->assoc_active = true;
+	cnt = atomic_inc_return(&rport->act_ctrl_cnt);
+	if (cnt == 1)
+		nvme_fc_rport_active_on_lport(rport);
+
+	return 0;
+}
+
+static int
+nvme_fc_ctlr_inactive_on_rport(struct nvme_fc_ctrl *ctrl)
+{
+	struct nvme_fc_rport *rport = ctrl->rport;
+	struct nvme_fc_lport *lport = rport->lport;
+	u32 cnt;
+
+	/* ctrl->assoc_active=false will be set independently */
+
+	cnt = atomic_dec_return(&rport->act_ctrl_cnt);
+	if (cnt == 0) {
+		if (rport->remoteport.port_state == FC_OBJSTATE_DELETED)
+			lport->ops->remoteport_delete(&rport->remoteport);
+		nvme_fc_rport_inactive_on_lport(rport);
+	}
+
+	return 0;
+}
+
 /*
  * This routine restarts the controller on the host side, and
  * on the link side, recreates the controller association.
@@ -2368,26 +2692,31 @@ static int
 nvme_fc_create_association(struct nvme_fc_ctrl *ctrl)
 {
 	struct nvmf_ctrl_options *opts = ctrl->ctrl.opts;
-	u32 segs;
 	int ret;
 	bool changed;
 
 	++ctrl->ctrl.nr_reconnects;
 
+	if (ctrl->rport->remoteport.port_state != FC_OBJSTATE_ONLINE)
+		return -ENODEV;
+
+	if (nvme_fc_ctlr_active_on_rport(ctrl))
+		return -ENOTUNIQ;
+
 	/*
 	 * Create the admin queue
 	 */
 
-	nvme_fc_init_queue(ctrl, 0, NVME_FC_AQ_BLKMQ_DEPTH);
+	nvme_fc_init_queue(ctrl, 0);
 
 	ret = __nvme_fc_create_hw_queue(ctrl, &ctrl->queues[0], 0,
-				NVME_FC_AQ_BLKMQ_DEPTH);
+				NVME_AQ_BLK_MQ_DEPTH);
 	if (ret)
 		goto out_free_queue;
 
 	ret = nvme_fc_connect_admin_queue(ctrl, &ctrl->queues[0],
-				NVME_FC_AQ_BLKMQ_DEPTH,
-				(NVME_FC_AQ_BLKMQ_DEPTH / 4));
+				NVME_AQ_BLK_MQ_DEPTH,
+				(NVME_AQ_BLK_MQ_DEPTH / 4));
 	if (ret)
 		goto out_delete_hw_queue;
 
@@ -2419,9 +2748,8 @@ nvme_fc_create_association(struct nvme_fc_ctrl *ctrl)
 	if (ret)
 		goto out_disconnect_admin_queue;
 
-	segs = min_t(u32, NVME_FC_MAX_SEGMENTS,
-			ctrl->lport->ops->max_sgl_segments);
-	ctrl->ctrl.max_hw_sectors = (segs - 1) << (PAGE_SHIFT - 9);
+	ctrl->ctrl.max_hw_sectors =
+		(ctrl->lport->ops->max_sgl_segments - 1) << (PAGE_SHIFT - 9);
 
 	ret = nvme_init_identify(&ctrl->ctrl);
 	if (ret)
@@ -2465,11 +2793,11 @@ nvme_fc_create_association(struct nvme_fc_ctrl *ctrl)
 	}
 
 	changed = nvme_change_ctrl_state(&ctrl->ctrl, NVME_CTRL_LIVE);
-	WARN_ON_ONCE(!changed);
 
 	ctrl->ctrl.nr_reconnects = 0;
 
-	nvme_start_ctrl(&ctrl->ctrl);
+	if (changed)
+		nvme_start_ctrl(&ctrl->ctrl);
 
 	return 0;	/* Success */
 
@@ -2482,6 +2810,8 @@ out_delete_hw_queue:
 	__nvme_fc_delete_hw_queue(ctrl, &ctrl->queues[0], 0);
 out_free_queue:
 	nvme_fc_free_queue(&ctrl->queues[0]);
+	ctrl->assoc_active = false;
+	nvme_fc_ctlr_inactive_on_rport(ctrl);
 
 	return ret;
 }
@@ -2497,6 +2827,10 @@ nvme_fc_delete_association(struct nvme_fc_ctrl *ctrl)
 {
 	unsigned long flags;
 
+	if (!ctrl->assoc_active)
+		return;
+	ctrl->assoc_active = false;
+
 	spin_lock_irqsave(&ctrl->lock, flags);
 	ctrl->flags |= FCCTRL_TERMIO;
 	ctrl->iocnt = 0;
@@ -2537,7 +2871,8 @@ nvme_fc_delete_association(struct nvme_fc_ctrl *ctrl)
 	 * use blk_mq_tagset_busy_itr() and the transport routine to
 	 * terminate the exchanges.
 	 */
-	blk_mq_quiesce_queue(ctrl->ctrl.admin_q);
+	if (ctrl->ctrl.state != NVME_CTRL_NEW)
+		blk_mq_quiesce_queue(ctrl->ctrl.admin_q);
 	blk_mq_tagset_busy_iter(&ctrl->admin_tag_set,
 				nvme_fc_terminate_exchange, &ctrl->ctrl);
 
@@ -2568,102 +2903,64 @@ nvme_fc_delete_association(struct nvme_fc_ctrl *ctrl)
 
 	__nvme_fc_delete_hw_queue(ctrl, &ctrl->queues[0], 0);
 	nvme_fc_free_queue(&ctrl->queues[0]);
+
+	nvme_fc_ctlr_inactive_on_rport(ctrl);
 }
 
 static void
-nvme_fc_delete_ctrl_work(struct work_struct *work)
+nvme_fc_delete_ctrl(struct nvme_ctrl *nctrl)
 {
-	struct nvme_fc_ctrl *ctrl =
-		container_of(work, struct nvme_fc_ctrl, delete_work);
+	struct nvme_fc_ctrl *ctrl = to_fc_ctrl(nctrl);
 
-	cancel_work_sync(&ctrl->ctrl.reset_work);
 	cancel_delayed_work_sync(&ctrl->connect_work);
-	nvme_stop_ctrl(&ctrl->ctrl);
-	nvme_remove_namespaces(&ctrl->ctrl);
 	/*
 	 * kill the association on the link side.  this will block
 	 * waiting for io to terminate
 	 */
 	nvme_fc_delete_association(ctrl);
-
-	/*
-	 * tear down the controller
-	 * After the last reference on the nvme ctrl is removed,
-	 * the transport nvme_fc_nvme_ctrl_freed() callback will be
-	 * invoked. From there, the transport will tear down it's
-	 * logical queues and association.
-	 */
-	nvme_uninit_ctrl(&ctrl->ctrl);
-
-	nvme_put_ctrl(&ctrl->ctrl);
-}
-
-static bool
-__nvme_fc_schedule_delete_work(struct nvme_fc_ctrl *ctrl)
-{
-	if (!nvme_change_ctrl_state(&ctrl->ctrl, NVME_CTRL_DELETING))
-		return true;
-
-	if (!queue_work(nvme_wq, &ctrl->delete_work))
-		return true;
-
-	return false;
-}
-
-static int
-__nvme_fc_del_ctrl(struct nvme_fc_ctrl *ctrl)
-{
-	return __nvme_fc_schedule_delete_work(ctrl) ? -EBUSY : 0;
-}
-
-/*
- * Request from nvme core layer to delete the controller
- */
-static int
-nvme_fc_del_nvme_ctrl(struct nvme_ctrl *nctrl)
-{
-	struct nvme_fc_ctrl *ctrl = to_fc_ctrl(nctrl);
-	int ret;
-
-	if (!kref_get_unless_zero(&ctrl->ctrl.kref))
-		return -EBUSY;
-
-	ret = __nvme_fc_del_ctrl(ctrl);
-
-	if (!ret)
-		flush_workqueue(nvme_wq);
-
-	nvme_put_ctrl(&ctrl->ctrl);
-
-	return ret;
 }
 
 static void
 nvme_fc_reconnect_or_delete(struct nvme_fc_ctrl *ctrl, int status)
 {
-	/* If we are resetting/deleting then do nothing */
-	if (ctrl->ctrl.state != NVME_CTRL_RECONNECTING) {
-		WARN_ON_ONCE(ctrl->ctrl.state == NVME_CTRL_NEW ||
-			ctrl->ctrl.state == NVME_CTRL_LIVE);
-		return;
-	}
+	struct nvme_fc_rport *rport = ctrl->rport;
+	struct nvme_fc_remote_port *portptr = &rport->remoteport;
+	unsigned long recon_delay = ctrl->ctrl.opts->reconnect_delay * HZ;
+	bool recon = true;
 
-	dev_info(ctrl->ctrl.device,
-		"NVME-FC{%d}: reset: Reconnect attempt failed (%d)\n",
-		ctrl->cnum, status);
+	if (ctrl->ctrl.state != NVME_CTRL_RECONNECTING)
+		return;
 
-	if (nvmf_should_reconnect(&ctrl->ctrl)) {
+	if (portptr->port_state == FC_OBJSTATE_ONLINE)
 		dev_info(ctrl->ctrl.device,
-			"NVME-FC{%d}: Reconnect attempt in %d seconds.\n",
-			ctrl->cnum, ctrl->ctrl.opts->reconnect_delay);
-		queue_delayed_work(nvme_wq, &ctrl->connect_work,
-				ctrl->ctrl.opts->reconnect_delay * HZ);
+			"NVME-FC{%d}: reset: Reconnect attempt failed (%d)\n",
+			ctrl->cnum, status);
+	else if (time_after_eq(jiffies, rport->dev_loss_end))
+		recon = false;
+
+	if (recon && nvmf_should_reconnect(&ctrl->ctrl)) {
+		if (portptr->port_state == FC_OBJSTATE_ONLINE)
+			dev_info(ctrl->ctrl.device,
+				"NVME-FC{%d}: Reconnect attempt in %ld "
+				"seconds\n",
+				ctrl->cnum, recon_delay / HZ);
+		else if (time_after(jiffies + recon_delay, rport->dev_loss_end))
+			recon_delay = rport->dev_loss_end - jiffies;
+
+		queue_delayed_work(nvme_wq, &ctrl->connect_work, recon_delay);
 	} else {
-		dev_warn(ctrl->ctrl.device,
+		if (portptr->port_state == FC_OBJSTATE_ONLINE)
+			dev_warn(ctrl->ctrl.device,
 				"NVME-FC{%d}: Max reconnect attempts (%d) "
 				"reached. Removing controller\n",
 				ctrl->cnum, ctrl->ctrl.nr_reconnects);
-		WARN_ON(__nvme_fc_schedule_delete_work(ctrl));
+		else
+			dev_warn(ctrl->ctrl.device,
+				"NVME-FC{%d}: dev_loss_tmo (%d) expired "
+				"while waiting for remoteport connectivity. "
+				"Removing controller\n", ctrl->cnum,
+				portptr->dev_loss_tmo);
+		WARN_ON(nvme_delete_ctrl(&ctrl->ctrl));
 	}
 }
 
@@ -2675,15 +2972,28 @@ nvme_fc_reset_ctrl_work(struct work_struct *work)
 	int ret;
 
 	nvme_stop_ctrl(&ctrl->ctrl);
+
 	/* will block will waiting for io to terminate */
 	nvme_fc_delete_association(ctrl);
 
-	ret = nvme_fc_create_association(ctrl);
+	if (!nvme_change_ctrl_state(&ctrl->ctrl, NVME_CTRL_RECONNECTING)) {
+		dev_err(ctrl->ctrl.device,
+			"NVME-FC{%d}: error_recovery: Couldn't change state "
+			"to RECONNECTING\n", ctrl->cnum);
+		return;
+	}
+
+	if (ctrl->rport->remoteport.port_state == FC_OBJSTATE_ONLINE)
+		ret = nvme_fc_create_association(ctrl);
+	else
+		ret = -ENOTCONN;
+
 	if (ret)
 		nvme_fc_reconnect_or_delete(ctrl, ret);
 	else
 		dev_info(ctrl->ctrl.device,
-			"NVME-FC{%d}: controller reset complete\n", ctrl->cnum);
+			"NVME-FC{%d}: controller reset complete\n",
+			ctrl->cnum);
 }
 
 static const struct nvme_ctrl_ops nvme_fc_ctrl_ops = {
@@ -2695,8 +3005,9 @@ static const struct nvme_ctrl_ops nvme_fc_ctrl_ops = {
 	.reg_write32		= nvmf_reg_write32,
 	.free_ctrl		= nvme_fc_nvme_ctrl_freed,
 	.submit_async_event	= nvme_fc_submit_async_event,
-	.delete_ctrl		= nvme_fc_del_nvme_ctrl,
+	.delete_ctrl		= nvme_fc_delete_ctrl,
 	.get_address		= nvmf_get_address,
+	.reinit_request		= nvme_fc_reinit_request,
 };
 
 static void
@@ -2728,6 +3039,33 @@ static const struct blk_mq_ops nvme_fc_admin_mq_ops = {
 };
 
 
+/*
+ * Fails a controller request if it matches an existing controller
+ * (association) with the same tuple:
+ * <Host NQN, Host ID, local FC port, remote FC port, SUBSYS NQN>
+ *
+ * The ports don't need to be compared as they are intrinsically
+ * already matched by the port pointers supplied.
+ */
+static bool
+nvme_fc_existing_controller(struct nvme_fc_rport *rport,
+		struct nvmf_ctrl_options *opts)
+{
+	struct nvme_fc_ctrl *ctrl;
+	unsigned long flags;
+	bool found = false;
+
+	spin_lock_irqsave(&rport->lock, flags);
+	list_for_each_entry(ctrl, &rport->ctrl_list, ctrl_list) {
+		found = nvmf_ctlr_matches_baseopts(&ctrl->ctrl, opts);
+		if (found)
+			break;
+	}
+	spin_unlock_irqrestore(&rport->lock, flags);
+
+	return found;
+}
+
 static struct nvme_ctrl *
 nvme_fc_init_ctrl(struct device *dev, struct nvmf_ctrl_options *opts,
 	struct nvme_fc_lport *lport, struct nvme_fc_rport *rport)
@@ -2742,6 +3080,12 @@ nvme_fc_init_ctrl(struct device *dev, struct nvmf_ctrl_options *opts,
 		goto out_fail;
 	}
 
+	if (!opts->duplicate_connect &&
+	    nvme_fc_existing_controller(rport, opts)) {
+		ret = -EALREADY;
+		goto out_fail;
+	}
+
 	ctrl = kzalloc(sizeof(*ctrl), GFP_KERNEL);
 	if (!ctrl) {
 		ret = -ENOMEM;
@@ -2760,12 +3104,12 @@ nvme_fc_init_ctrl(struct device *dev, struct nvmf_ctrl_options *opts,
 	ctrl->rport = rport;
 	ctrl->dev = lport->dev;
 	ctrl->cnum = idx;
+	ctrl->assoc_active = false;
 	init_waitqueue_head(&ctrl->ioabort_wait);
 
 	get_device(ctrl->dev);
 	kref_init(&ctrl->ref);
 
-	INIT_WORK(&ctrl->delete_work, nvme_fc_delete_ctrl_work);
 	INIT_WORK(&ctrl->ctrl.reset_work, nvme_fc_reset_ctrl_work);
 	INIT_DELAYED_WORK(&ctrl->connect_work, nvme_fc_connect_ctrl_work);
 	spin_lock_init(&ctrl->lock);
@@ -2787,7 +3131,7 @@ nvme_fc_init_ctrl(struct device *dev, struct nvmf_ctrl_options *opts,
 
 	memset(&ctrl->admin_tag_set, 0, sizeof(ctrl->admin_tag_set));
 	ctrl->admin_tag_set.ops = &nvme_fc_admin_mq_ops;
-	ctrl->admin_tag_set.queue_depth = NVME_FC_AQ_BLKMQ_DEPTH;
+	ctrl->admin_tag_set.queue_depth = NVME_AQ_MQ_TAG_DEPTH;
 	ctrl->admin_tag_set.reserved_tags = 2; /* fabric connect + Keep-Alive */
 	ctrl->admin_tag_set.numa_node = NUMA_NO_NODE;
 	ctrl->admin_tag_set.cmd_size = sizeof(struct nvme_fc_fcp_op) +
@@ -2797,6 +3141,7 @@ nvme_fc_init_ctrl(struct device *dev, struct nvmf_ctrl_options *opts,
 	ctrl->admin_tag_set.driver_data = ctrl;
 	ctrl->admin_tag_set.nr_hw_queues = 1;
 	ctrl->admin_tag_set.timeout = ADMIN_TIMEOUT;
+	ctrl->admin_tag_set.flags = BLK_MQ_F_NO_SCHED;
 
 	ret = blk_mq_alloc_tag_set(&ctrl->admin_tag_set);
 	if (ret)
@@ -2878,7 +3223,7 @@ nvme_fc_init_ctrl(struct device *dev, struct nvmf_ctrl_options *opts,
 		return ERR_PTR(ret);
 	}
 
-	kref_get(&ctrl->ctrl.kref);
+	nvme_get_ctrl(&ctrl->ctrl);
 
 	dev_info(ctrl->ctrl.device,
 		"NVME-FC{%d}: new ctrl: NQN \"%s\"\n",
@@ -3026,7 +3371,50 @@ static struct nvmf_transport_ops nvme_fc_transport = {
 
 static int __init nvme_fc_init_module(void)
 {
-	return nvmf_register_transport(&nvme_fc_transport);
+	int ret;
+
+	/*
+	 * NOTE:
+	 * It is expected that in the future the kernel will combine
+	 * the FC-isms that are currently under scsi and now being
+	 * added to by NVME into a new standalone FC class. The SCSI
+	 * and NVME protocols and their devices would be under this
+	 * new FC class.
+	 *
+	 * As we need something to post FC-specific udev events to,
+	 * specifically for nvme probe events, start by creating the
+	 * new device class.  When the new standalone FC class is
+	 * put in place, this code will move to a more generic
+	 * location for the class.
+	 */
+	fc_class = class_create(THIS_MODULE, "fc");
+	if (IS_ERR(fc_class)) {
+		pr_err("couldn't register class fc\n");
+		return PTR_ERR(fc_class);
+	}
+
+	/*
+	 * Create a device for the FC-centric udev events
+	 */
+	fc_udev_device = device_create(fc_class, NULL, MKDEV(0, 0), NULL,
+				"fc_udev_device");
+	if (IS_ERR(fc_udev_device)) {
+		pr_err("couldn't create fc_udev device!\n");
+		ret = PTR_ERR(fc_udev_device);
+		goto out_destroy_class;
+	}
+
+	ret = nvmf_register_transport(&nvme_fc_transport);
+	if (ret)
+		goto out_destroy_device;
+
+	return 0;
+
+out_destroy_device:
+	device_destroy(fc_class, MKDEV(0, 0));
+out_destroy_class:
+	class_destroy(fc_class);
+	return ret;
 }
 
 static void __exit nvme_fc_exit_module(void)
@@ -3039,6 +3427,9 @@ static void __exit nvme_fc_exit_module(void)
 
 	ida_destroy(&nvme_fc_local_port_cnt);
 	ida_destroy(&nvme_fc_ctrl_cnt);
+
+	device_destroy(fc_class, MKDEV(0, 0));
+	class_destroy(fc_class);
 }
 
 module_init(nvme_fc_init_module);

drivers/nvme/host/lightnvm.c

@@ -305,7 +305,7 @@ static int nvme_nvm_identity(struct nvm_dev *nvmdev, struct nvm_id *nvm_id)
 	int ret;
 
 	c.identity.opcode = nvme_nvm_admin_identity;
-	c.identity.nsid = cpu_to_le32(ns->ns_id);
+	c.identity.nsid = cpu_to_le32(ns->head->ns_id);
 	c.identity.chnl_off = 0;
 
 	nvme_nvm_id = kmalloc(sizeof(struct nvme_nvm_id), GFP_KERNEL);
@@ -344,7 +344,7 @@ static int nvme_nvm_get_l2p_tbl(struct nvm_dev *nvmdev, u64 slba, u32 nlb,
 	int ret = 0;
 
 	c.l2p.opcode = nvme_nvm_admin_get_l2p_tbl;
-	c.l2p.nsid = cpu_to_le32(ns->ns_id);
+	c.l2p.nsid = cpu_to_le32(ns->head->ns_id);
 	entries = kmalloc(len, GFP_KERNEL);
 	if (!entries)
 		return -ENOMEM;
@@ -402,7 +402,7 @@ static int nvme_nvm_get_bb_tbl(struct nvm_dev *nvmdev, struct ppa_addr ppa,
 	int ret = 0;
 
 	c.get_bb.opcode = nvme_nvm_admin_get_bb_tbl;
-	c.get_bb.nsid = cpu_to_le32(ns->ns_id);
+	c.get_bb.nsid = cpu_to_le32(ns->head->ns_id);
 	c.get_bb.spba = cpu_to_le64(ppa.ppa);
 
 	bb_tbl = kzalloc(tblsz, GFP_KERNEL);
@@ -452,7 +452,7 @@ static int nvme_nvm_set_bb_tbl(struct nvm_dev *nvmdev, struct ppa_addr *ppas,
 	int ret = 0;
 
 	c.set_bb.opcode = nvme_nvm_admin_set_bb_tbl;
-	c.set_bb.nsid = cpu_to_le32(ns->ns_id);
+	c.set_bb.nsid = cpu_to_le32(ns->head->ns_id);
 	c.set_bb.spba = cpu_to_le64(ppas->ppa);
 	c.set_bb.nlb = cpu_to_le16(nr_ppas - 1);
 	c.set_bb.value = type;
@@ -469,7 +469,7 @@ static inline void nvme_nvm_rqtocmd(struct nvm_rq *rqd, struct nvme_ns *ns,
 				    struct nvme_nvm_command *c)
 {
 	c->ph_rw.opcode = rqd->opcode;
-	c->ph_rw.nsid = cpu_to_le32(ns->ns_id);
+	c->ph_rw.nsid = cpu_to_le32(ns->head->ns_id);
 	c->ph_rw.spba = cpu_to_le64(rqd->ppa_addr.ppa);
 	c->ph_rw.metadata = cpu_to_le64(rqd->dma_meta_list);
 	c->ph_rw.control = cpu_to_le16(rqd->flags);
@@ -492,34 +492,47 @@ static void nvme_nvm_end_io(struct request *rq, blk_status_t status)
 	blk_mq_free_request(rq);
 }
 
-static int nvme_nvm_submit_io(struct nvm_dev *dev, struct nvm_rq *rqd)
+static struct request *nvme_nvm_alloc_request(struct request_queue *q,
+					      struct nvm_rq *rqd,
+					      struct nvme_nvm_command *cmd)
 {
-	struct request_queue *q = dev->q;
 	struct nvme_ns *ns = q->queuedata;
 	struct request *rq;
-	struct bio *bio = rqd->bio;
-	struct nvme_nvm_command *cmd;
-
-	cmd = kzalloc(sizeof(struct nvme_nvm_command), GFP_KERNEL);
-	if (!cmd)
-		return -ENOMEM;
 
 	nvme_nvm_rqtocmd(rqd, ns, cmd);
 
 	rq = nvme_alloc_request(q, (struct nvme_command *)cmd, 0, NVME_QID_ANY);
-	if (IS_ERR(rq)) {
-		kfree(cmd);
-		return PTR_ERR(rq);
-	}
+	if (IS_ERR(rq))
+		return rq;
+
 	rq->cmd_flags &= ~REQ_FAILFAST_DRIVER;
 
-	if (bio) {
-		blk_init_request_from_bio(rq, bio);
+	if (rqd->bio) {
+		blk_init_request_from_bio(rq, rqd->bio);
 	} else {
 		rq->ioprio = IOPRIO_PRIO_VALUE(IOPRIO_CLASS_BE, IOPRIO_NORM);
 		rq->__data_len = 0;
 	}
 
+	return rq;
+}
+
+static int nvme_nvm_submit_io(struct nvm_dev *dev, struct nvm_rq *rqd)
+{
+	struct request_queue *q = dev->q;
+	struct nvme_nvm_command *cmd;
+	struct request *rq;
+
+	cmd = kzalloc(sizeof(struct nvme_nvm_command), GFP_KERNEL);
+	if (!cmd)
+		return -ENOMEM;
+
+	rq = nvme_nvm_alloc_request(q, rqd, cmd);
+	if (IS_ERR(rq)) {
+		kfree(cmd);
+		return PTR_ERR(rq);
+	}
+
 	rq->end_io_data = rqd;
 
 	blk_execute_rq_nowait(q, NULL, rq, 0, nvme_nvm_end_io);
@@ -527,6 +540,34 @@ static int nvme_nvm_submit_io(struct nvm_dev *dev, struct nvm_rq *rqd)
 	return 0;
 }
 
+static int nvme_nvm_submit_io_sync(struct nvm_dev *dev, struct nvm_rq *rqd)
+{
+	struct request_queue *q = dev->q;
+	struct request *rq;
+	struct nvme_nvm_command cmd;
+	int ret = 0;
+
+	memset(&cmd, 0, sizeof(struct nvme_nvm_command));
+
+	rq = nvme_nvm_alloc_request(q, rqd, &cmd);
+	if (IS_ERR(rq))
+		return PTR_ERR(rq);
+
+	/* I/Os can fail and the error is signaled through rqd. Callers must
+	 * handle the error accordingly.
+	 */
+	blk_execute_rq(q, NULL, rq, 0);
+	if (nvme_req(rq)->flags & NVME_REQ_CANCELLED)
+		ret = -EINTR;
+
+	rqd->ppa_status = le64_to_cpu(nvme_req(rq)->result.u64);
+	rqd->error = nvme_req(rq)->status;
+
+	blk_mq_free_request(rq);
+
+	return ret;
+}
+
 static void *nvme_nvm_create_dma_pool(struct nvm_dev *nvmdev, char *name)
 {
 	struct nvme_ns *ns = nvmdev->q->queuedata;
@@ -562,6 +603,7 @@ static struct nvm_dev_ops nvme_nvm_dev_ops = {
 	.set_bb_tbl		= nvme_nvm_set_bb_tbl,
 
 	.submit_io		= nvme_nvm_submit_io,
+	.submit_io_sync		= nvme_nvm_submit_io_sync,
 
 	.create_dma_pool	= nvme_nvm_create_dma_pool,
 	.destroy_dma_pool	= nvme_nvm_destroy_dma_pool,
@@ -600,8 +642,6 @@ static int nvme_nvm_submit_user_cmd(struct request_queue *q,
 
 	rq->timeout = timeout ? timeout : ADMIN_TIMEOUT;
 
-	rq->cmd_flags &= ~REQ_FAILFAST_DRIVER;
-
 	if (ppa_buf && ppa_len) {
 		ppa_list = dma_pool_alloc(dev->dma_pool, GFP_KERNEL, &ppa_dma);
 		if (!ppa_list) {
@@ -691,7 +731,7 @@ static int nvme_nvm_submit_vio(struct nvme_ns *ns,
 
 	memset(&c, 0, sizeof(c));
 	c.ph_rw.opcode = vio.opcode;
-	c.ph_rw.nsid = cpu_to_le32(ns->ns_id);
+	c.ph_rw.nsid = cpu_to_le32(ns->head->ns_id);
 	c.ph_rw.control = cpu_to_le16(vio.control);
 	c.ph_rw.length = cpu_to_le16(vio.nppas);
 
@@ -728,7 +768,7 @@ static int nvme_nvm_user_vcmd(struct nvme_ns *ns, int admin,
 
 	memset(&c, 0, sizeof(c));
 	c.common.opcode = vcmd.opcode;
-	c.common.nsid = cpu_to_le32(ns->ns_id);
+	c.common.nsid = cpu_to_le32(ns->head->ns_id);
 	c.common.cdw2[0] = cpu_to_le32(vcmd.cdw2);
 	c.common.cdw2[1] = cpu_to_le32(vcmd.cdw3);
 	/* cdw11-12 */

drivers/nvme/host/multipath.c

@@ -0,0 +1,291 @@
+/*
+ * Copyright (c) 2017 Christoph Hellwig.
+ *
+ * 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.
+ */
+
+#include <linux/moduleparam.h>
+#include "nvme.h"
+
+static bool multipath = true;
+module_param(multipath, bool, 0644);
+MODULE_PARM_DESC(multipath,
+	"turn on native support for multiple controllers per subsystem");
+
+void nvme_failover_req(struct request *req)
+{
+	struct nvme_ns *ns = req->q->queuedata;
+	unsigned long flags;
+
+	spin_lock_irqsave(&ns->head->requeue_lock, flags);
+	blk_steal_bios(&ns->head->requeue_list, req);
+	spin_unlock_irqrestore(&ns->head->requeue_lock, flags);
+	blk_mq_end_request(req, 0);
+
+	nvme_reset_ctrl(ns->ctrl);
+	kblockd_schedule_work(&ns->head->requeue_work);
+}
+
+bool nvme_req_needs_failover(struct request *req)
+{
+	if (!(req->cmd_flags & REQ_NVME_MPATH))
+		return false;
+
+	switch (nvme_req(req)->status & 0x7ff) {
+	/*
+	 * Generic command status:
+	 */
+	case NVME_SC_INVALID_OPCODE:
+	case NVME_SC_INVALID_FIELD:
+	case NVME_SC_INVALID_NS:
+	case NVME_SC_LBA_RANGE:
+	case NVME_SC_CAP_EXCEEDED:
+	case NVME_SC_RESERVATION_CONFLICT:
+		return false;
+
+	/*
+	 * I/O command set specific error.  Unfortunately these values are
+	 * reused for fabrics commands, but those should never get here.
+	 */
+	case NVME_SC_BAD_ATTRIBUTES:
+	case NVME_SC_INVALID_PI:
+	case NVME_SC_READ_ONLY:
+	case NVME_SC_ONCS_NOT_SUPPORTED:
+		WARN_ON_ONCE(nvme_req(req)->cmd->common.opcode ==
+			nvme_fabrics_command);
+		return false;
+
+	/*
+	 * Media and Data Integrity Errors:
+	 */
+	case NVME_SC_WRITE_FAULT:
+	case NVME_SC_READ_ERROR:
+	case NVME_SC_GUARD_CHECK:
+	case NVME_SC_APPTAG_CHECK:
+	case NVME_SC_REFTAG_CHECK:
+	case NVME_SC_COMPARE_FAILED:
+	case NVME_SC_ACCESS_DENIED:
+	case NVME_SC_UNWRITTEN_BLOCK:
+		return false;
+	}
+
+	/* Everything else could be a path failure, so should be retried */
+	return true;
+}
+
+void nvme_kick_requeue_lists(struct nvme_ctrl *ctrl)
+{
+	struct nvme_ns *ns;
+
+	mutex_lock(&ctrl->namespaces_mutex);
+	list_for_each_entry(ns, &ctrl->namespaces, list) {
+		if (ns->head->disk)
+			kblockd_schedule_work(&ns->head->requeue_work);
+	}
+	mutex_unlock(&ctrl->namespaces_mutex);
+}
+
+static struct nvme_ns *__nvme_find_path(struct nvme_ns_head *head)
+{
+	struct nvme_ns *ns;
+
+	list_for_each_entry_rcu(ns, &head->list, siblings) {
+		if (ns->ctrl->state == NVME_CTRL_LIVE) {
+			rcu_assign_pointer(head->current_path, ns);
+			return ns;
+		}
+	}
+
+	return NULL;
+}
+
+inline struct nvme_ns *nvme_find_path(struct nvme_ns_head *head)
+{
+	struct nvme_ns *ns = srcu_dereference(head->current_path, &head->srcu);
+
+	if (unlikely(!ns || ns->ctrl->state != NVME_CTRL_LIVE))
+		ns = __nvme_find_path(head);
+	return ns;
+}
+
+static blk_qc_t nvme_ns_head_make_request(struct request_queue *q,
+		struct bio *bio)
+{
+	struct nvme_ns_head *head = q->queuedata;
+	struct device *dev = disk_to_dev(head->disk);
+	struct nvme_ns *ns;
+	blk_qc_t ret = BLK_QC_T_NONE;
+	int srcu_idx;
+
+	srcu_idx = srcu_read_lock(&head->srcu);
+	ns = nvme_find_path(head);
+	if (likely(ns)) {
+		bio->bi_disk = ns->disk;
+		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");
+
+		spin_lock_irq(&head->requeue_lock);
+		bio_list_add(&head->requeue_list, bio);
+		spin_unlock_irq(&head->requeue_lock);
+	} else {
+		dev_warn_ratelimited(dev, "no path - failing I/O\n");
+
+		bio->bi_status = BLK_STS_IOERR;
+		bio_endio(bio);
+	}
+
+	srcu_read_unlock(&head->srcu, srcu_idx);
+	return ret;
+}
+
+static bool nvme_ns_head_poll(struct request_queue *q, blk_qc_t qc)
+{
+	struct nvme_ns_head *head = q->queuedata;
+	struct nvme_ns *ns;
+	bool found = false;
+	int srcu_idx;
+
+	srcu_idx = srcu_read_lock(&head->srcu);
+	ns = srcu_dereference(head->current_path, &head->srcu);
+	if (likely(ns && ns->ctrl->state == NVME_CTRL_LIVE))
+		found = ns->queue->poll_fn(q, qc);
+	srcu_read_unlock(&head->srcu, srcu_idx);
+	return found;
+}
+
+static void nvme_requeue_work(struct work_struct *work)
+{
+	struct nvme_ns_head *head =
+		container_of(work, struct nvme_ns_head, requeue_work);
+	struct bio *bio, *next;
+
+	spin_lock_irq(&head->requeue_lock);
+	next = bio_list_get(&head->requeue_list);
+	spin_unlock_irq(&head->requeue_lock);
+
+	while ((bio = next) != NULL) {
+		next = bio->bi_next;
+		bio->bi_next = NULL;
+
+		/*
+		 * Reset disk to the mpath node and resubmit to select a new
+		 * path.
+		 */
+		bio->bi_disk = head->disk;
+		generic_make_request(bio);
+	}
+}
+
+int nvme_mpath_alloc_disk(struct nvme_ctrl *ctrl, struct nvme_ns_head *head)
+{
+	struct request_queue *q;
+	bool vwc = false;
+
+	bio_list_init(&head->requeue_list);
+	spin_lock_init(&head->requeue_lock);
+	INIT_WORK(&head->requeue_work, nvme_requeue_work);
+
+	/*
+	 * Add a multipath node if the subsystems supports multiple controllers.
+	 * We also do this for private namespaces as the namespace sharing data could
+	 * change after a rescan.
+	 */
+	if (!(ctrl->subsys->cmic & (1 << 1)) || !multipath)
+		return 0;
+
+	q = blk_alloc_queue_node(GFP_KERNEL, NUMA_NO_NODE);
+	if (!q)
+		goto out;
+	q->queuedata = head;
+	blk_queue_make_request(q, nvme_ns_head_make_request);
+	q->poll_fn = nvme_ns_head_poll;
+	queue_flag_set_unlocked(QUEUE_FLAG_NONROT, q);
+	/* set to a default value for 512 until disk is validated */
+	blk_queue_logical_block_size(q, 512);
+
+	/* we need to propagate up the VMC settings */
+	if (ctrl->vwc & NVME_CTRL_VWC_PRESENT)
+		vwc = true;
+	blk_queue_write_cache(q, vwc, vwc);
+
+	head->disk = alloc_disk(0);
+	if (!head->disk)
+		goto out_cleanup_queue;
+	head->disk->fops = &nvme_ns_head_ops;
+	head->disk->private_data = head;
+	head->disk->queue = q;
+	head->disk->flags = GENHD_FL_EXT_DEVT;
+	sprintf(head->disk->disk_name, "nvme%dn%d",
+			ctrl->subsys->instance, head->instance);
+	return 0;
+
+out_cleanup_queue:
+	blk_cleanup_queue(q);
+out:
+	return -ENOMEM;
+}
+
+void nvme_mpath_add_disk(struct nvme_ns_head *head)
+{
+	if (!head->disk)
+		return;
+	device_add_disk(&head->subsys->dev, head->disk);
+	if (sysfs_create_group(&disk_to_dev(head->disk)->kobj,
+			&nvme_ns_id_attr_group))
+		pr_warn("%s: failed to create sysfs group for identification\n",
+			head->disk->disk_name);
+}
+
+void nvme_mpath_add_disk_links(struct nvme_ns *ns)
+{
+	struct kobject *slave_disk_kobj, *holder_disk_kobj;
+
+	if (!ns->head->disk)
+		return;
+
+	slave_disk_kobj = &disk_to_dev(ns->disk)->kobj;
+	if (sysfs_create_link(ns->head->disk->slave_dir, slave_disk_kobj,
+			kobject_name(slave_disk_kobj)))
+		return;
+
+	holder_disk_kobj = &disk_to_dev(ns->head->disk)->kobj;
+	if (sysfs_create_link(ns->disk->part0.holder_dir, holder_disk_kobj,
+			kobject_name(holder_disk_kobj)))
+		sysfs_remove_link(ns->head->disk->slave_dir,
+			kobject_name(slave_disk_kobj));
+}
+
+void nvme_mpath_remove_disk(struct nvme_ns_head *head)
+{
+	if (!head->disk)
+		return;
+	sysfs_remove_group(&disk_to_dev(head->disk)->kobj,
+			   &nvme_ns_id_attr_group);
+	del_gendisk(head->disk);
+	blk_set_queue_dying(head->disk->queue);
+	/* make sure all pending bios are cleaned up */
+	kblockd_schedule_work(&head->requeue_work);
+	flush_work(&head->requeue_work);
+	blk_cleanup_queue(head->disk->queue);
+	put_disk(head->disk);
+}
+
+void nvme_mpath_remove_disk_links(struct nvme_ns *ns)
+{
+	if (!ns->head->disk)
+		return;
+
+	sysfs_remove_link(ns->disk->part0.holder_dir,
+			kobject_name(&disk_to_dev(ns->head->disk)->kobj));
+	sysfs_remove_link(ns->head->disk->slave_dir,
+			kobject_name(&disk_to_dev(ns->disk)->kobj));
+}

drivers/nvme/host/nvme.h

@@ -15,16 +15,17 @@
 #define _NVME_H
 
 #include <linux/nvme.h>
+#include <linux/cdev.h>
 #include <linux/pci.h>
 #include <linux/kref.h>
 #include <linux/blk-mq.h>
 #include <linux/lightnvm.h>
 #include <linux/sed-opal.h>
 
-extern unsigned char nvme_io_timeout;
+extern unsigned int nvme_io_timeout;
 #define NVME_IO_TIMEOUT	(nvme_io_timeout * HZ)
 
-extern unsigned char admin_timeout;
+extern unsigned int admin_timeout;
 #define ADMIN_TIMEOUT	(admin_timeout * HZ)
 
 #define NVME_DEFAULT_KATO	5
@@ -94,6 +95,11 @@ struct nvme_request {
 	u16			status;
 };
 
+/*
+ * Mark a bio as coming in through the mpath node.
+ */
+#define REQ_NVME_MPATH		REQ_DRV
+
 enum {
 	NVME_REQ_CANCELLED		= (1 << 0),
 };
@@ -127,24 +133,23 @@ struct nvme_ctrl {
 	struct request_queue *admin_q;
 	struct request_queue *connect_q;
 	struct device *dev;
-	struct kref kref;
 	int instance;
 	struct blk_mq_tag_set *tagset;
 	struct blk_mq_tag_set *admin_tagset;
 	struct list_head namespaces;
 	struct mutex namespaces_mutex;
+	struct device ctrl_device;
 	struct device *device;	/* char device */
-	struct list_head node;
-	struct ida ns_ida;
+	struct cdev cdev;
 	struct work_struct reset_work;
+	struct work_struct delete_work;
+
+	struct nvme_subsystem *subsys;
+	struct list_head subsys_entry;
 
 	struct opal_dev *opal_dev;
 
 	char name[12];
-	char serial[20];
-	char model[40];
-	char firmware_rev[8];
-	char subnqn[NVMF_NQN_SIZE];
 	u16 cntlid;
 
 	u32 ctrl_config;
@@ -155,23 +160,23 @@ struct nvme_ctrl {
 	u32 page_size;
 	u32 max_hw_sectors;
 	u16 oncs;
-	u16 vid;
 	u16 oacs;
 	u16 nssa;
 	u16 nr_streams;
 	atomic_t abort_limit;
-	u8 event_limit;
 	u8 vwc;
 	u32 vs;
 	u32 sgls;
 	u16 kas;
 	u8 npss;
 	u8 apsta;
+	u32 aen_result;
 	unsigned int shutdown_timeout;
 	unsigned int kato;
 	bool subsystem;
 	unsigned long quirks;
 	struct nvme_id_power_state psd[32];
+	struct nvme_effects_log *effects;
 	struct work_struct scan_work;
 	struct work_struct async_event_work;
 	struct delayed_work ka_work;
@@ -197,21 +202,72 @@ struct nvme_ctrl {
 	struct nvmf_ctrl_options *opts;
 };
 
+struct nvme_subsystem {
+	int			instance;
+	struct device		dev;
+	/*
+	 * Because we unregister the device on the last put we need
+	 * a separate refcount.
+	 */
+	struct kref		ref;
+	struct list_head	entry;
+	struct mutex		lock;
+	struct list_head	ctrls;
+	struct list_head	nsheads;
+	char			subnqn[NVMF_NQN_SIZE];
+	char			serial[20];
+	char			model[40];
+	char			firmware_rev[8];
+	u8			cmic;
+	u16			vendor_id;
+	struct ida		ns_ida;
+};
+
+/*
+ * Container structure for uniqueue namespace identifiers.
+ */
+struct nvme_ns_ids {
+	u8	eui64[8];
+	u8	nguid[16];
+	uuid_t	uuid;
+};
+
+/*
+ * Anchor structure for namespaces.  There is one for each namespace in a
+ * NVMe subsystem that any of our controllers can see, and the namespace
+ * structure for each controller is chained of it.  For private namespaces
+ * there is a 1:1 relation to our namespace structures, that is ->list
+ * only ever has a single entry for private namespaces.
+ */
+struct nvme_ns_head {
+#ifdef CONFIG_NVME_MULTIPATH
+	struct gendisk		*disk;
+	struct nvme_ns __rcu	*current_path;
+	struct bio_list		requeue_list;
+	spinlock_t		requeue_lock;
+	struct work_struct	requeue_work;
+#endif
+	struct list_head	list;
+	struct srcu_struct      srcu;
+	struct nvme_subsystem	*subsys;
+	unsigned		ns_id;
+	struct nvme_ns_ids	ids;
+	struct list_head	entry;
+	struct kref		ref;
+	int			instance;
+};
+
 struct nvme_ns {
 	struct list_head list;
 
 	struct nvme_ctrl *ctrl;
 	struct request_queue *queue;
 	struct gendisk *disk;
+	struct list_head siblings;
 	struct nvm_dev *ndev;
 	struct kref kref;
-	int instance;
+	struct nvme_ns_head *head;
 
-	u8 eui[8];
-	u8 nguid[16];
-	uuid_t uuid;
-
-	unsigned ns_id;
 	int lba_shift;
 	u16 ms;
 	u16 sgs;
@@ -234,9 +290,10 @@ struct nvme_ctrl_ops {
 	int (*reg_write32)(struct nvme_ctrl *ctrl, u32 off, u32 val);
 	int (*reg_read64)(struct nvme_ctrl *ctrl, u32 off, u64 *val);
 	void (*free_ctrl)(struct nvme_ctrl *ctrl);
-	void (*submit_async_event)(struct nvme_ctrl *ctrl, int aer_idx);
-	int (*delete_ctrl)(struct nvme_ctrl *ctrl);
+	void (*submit_async_event)(struct nvme_ctrl *ctrl);
+	void (*delete_ctrl)(struct nvme_ctrl *ctrl);
 	int (*get_address)(struct nvme_ctrl *ctrl, char *buf, int size);
+	int (*reinit_request)(void *data, struct request *rq);
 };
 
 static inline bool nvme_ctrl_ready(struct nvme_ctrl *ctrl)
@@ -278,6 +335,16 @@ static inline void nvme_end_request(struct request *req, __le16 status,
 	blk_mq_complete_request(req);
 }
 
+static inline void nvme_get_ctrl(struct nvme_ctrl *ctrl)
+{
+	get_device(ctrl->device);
+}
+
+static inline void nvme_put_ctrl(struct nvme_ctrl *ctrl)
+{
+	put_device(ctrl->device);
+}
+
 void nvme_complete_rq(struct request *req);
 void nvme_cancel_request(struct request *req, void *data, bool reserved);
 bool nvme_change_ctrl_state(struct nvme_ctrl *ctrl,
@@ -299,10 +366,8 @@ void nvme_remove_namespaces(struct nvme_ctrl *ctrl);
 int nvme_sec_submit(void *data, u16 spsp, u8 secp, void *buffer, size_t len,
 		bool send);
 
-#define NVME_NR_AERS	1
 void nvme_complete_async_event(struct nvme_ctrl *ctrl, __le16 status,
 		union nvme_result *res);
-void nvme_queue_async_events(struct nvme_ctrl *ctrl);
 
 void nvme_stop_queues(struct nvme_ctrl *ctrl);
 void nvme_start_queues(struct nvme_ctrl *ctrl);
@@ -311,21 +376,79 @@ void nvme_unfreeze(struct nvme_ctrl *ctrl);
 void nvme_wait_freeze(struct nvme_ctrl *ctrl);
 void nvme_wait_freeze_timeout(struct nvme_ctrl *ctrl, long timeout);
 void nvme_start_freeze(struct nvme_ctrl *ctrl);
+int nvme_reinit_tagset(struct nvme_ctrl *ctrl, struct blk_mq_tag_set *set);
 
 #define NVME_QID_ANY -1
 struct request *nvme_alloc_request(struct request_queue *q,
-		struct nvme_command *cmd, unsigned int flags, int qid);
+		struct nvme_command *cmd, blk_mq_req_flags_t flags, int qid);
 blk_status_t nvme_setup_cmd(struct nvme_ns *ns, struct request *req,
 		struct nvme_command *cmd);
 int nvme_submit_sync_cmd(struct request_queue *q, struct nvme_command *cmd,
 		void *buf, unsigned bufflen);
 int __nvme_submit_sync_cmd(struct request_queue *q, struct nvme_command *cmd,
 		union nvme_result *result, void *buffer, unsigned bufflen,
-		unsigned timeout, int qid, int at_head, int flags);
+		unsigned timeout, int qid, int at_head,
+		blk_mq_req_flags_t flags);
 int nvme_set_queue_count(struct nvme_ctrl *ctrl, int *count);
 void nvme_start_keep_alive(struct nvme_ctrl *ctrl);
 void nvme_stop_keep_alive(struct nvme_ctrl *ctrl);
 int nvme_reset_ctrl(struct nvme_ctrl *ctrl);
+int nvme_delete_ctrl(struct nvme_ctrl *ctrl);
+int nvme_delete_ctrl_sync(struct nvme_ctrl *ctrl);
+
+extern const struct attribute_group nvme_ns_id_attr_group;
+extern const struct block_device_operations nvme_ns_head_ops;
+
+#ifdef CONFIG_NVME_MULTIPATH
+void nvme_failover_req(struct request *req);
+bool nvme_req_needs_failover(struct request *req);
+void nvme_kick_requeue_lists(struct nvme_ctrl *ctrl);
+int nvme_mpath_alloc_disk(struct nvme_ctrl *ctrl,struct nvme_ns_head *head);
+void nvme_mpath_add_disk(struct nvme_ns_head *head);
+void nvme_mpath_add_disk_links(struct nvme_ns *ns);
+void nvme_mpath_remove_disk(struct nvme_ns_head *head);
+void nvme_mpath_remove_disk_links(struct nvme_ns *ns);
+
+static inline void nvme_mpath_clear_current_path(struct nvme_ns *ns)
+{
+	struct nvme_ns_head *head = ns->head;
+
+	if (head && ns == srcu_dereference(head->current_path, &head->srcu))
+		rcu_assign_pointer(head->current_path, NULL);
+}
+struct nvme_ns *nvme_find_path(struct nvme_ns_head *head);
+#else
+static inline void nvme_failover_req(struct request *req)
+{
+}
+static inline bool nvme_req_needs_failover(struct request *req)
+{
+	return false;
+}
+static inline void nvme_kick_requeue_lists(struct nvme_ctrl *ctrl)
+{
+}
+static inline int nvme_mpath_alloc_disk(struct nvme_ctrl *ctrl,
+		struct nvme_ns_head *head)
+{
+	return 0;
+}
+static inline void nvme_mpath_add_disk(struct nvme_ns_head *head)
+{
+}
+static inline void nvme_mpath_remove_disk(struct nvme_ns_head *head)
+{
+}
+static inline void nvme_mpath_add_disk_links(struct nvme_ns *ns)
+{
+}
+static inline void nvme_mpath_remove_disk_links(struct nvme_ns *ns)
+{
+}
+static inline void nvme_mpath_clear_current_path(struct nvme_ns *ns)
+{
+}
+#endif /* CONFIG_NVME_MULTIPATH */
 
 #ifdef CONFIG_NVM
 int nvme_nvm_register(struct nvme_ns *ns, char *disk_name, int node);

drivers/nvme/host/pci.c

@@ -13,7 +13,6 @@
  */
 
 #include <linux/aer.h>
-#include <linux/bitops.h>
 #include <linux/blkdev.h>
 #include <linux/blk-mq.h>
 #include <linux/blk-mq-pci.h>
@@ -26,12 +25,9 @@
 #include <linux/mutex.h>
 #include <linux/once.h>
 #include <linux/pci.h>
-#include <linux/poison.h>
 #include <linux/t10-pi.h>
-#include <linux/timer.h>
 #include <linux/types.h>
 #include <linux/io-64-nonatomic-lo-hi.h>
-#include <asm/unaligned.h>
 #include <linux/sed-opal.h>
 
 #include "nvme.h"
@@ -39,11 +35,7 @@
 #define SQ_SIZE(depth)		(depth * sizeof(struct nvme_command))
 #define CQ_SIZE(depth)		(depth * sizeof(struct nvme_completion))
 
-/*
- * We handle AEN commands ourselves and don't even let the
- * block layer know about them.
- */
-#define NVME_AQ_BLKMQ_DEPTH	(NVME_AQ_DEPTH - NVME_NR_AERS)
+#define SGES_PER_PAGE	(PAGE_SIZE / sizeof(struct nvme_sgl_desc))
 
 static int use_threaded_interrupts;
 module_param(use_threaded_interrupts, int, 0);
@@ -57,6 +49,12 @@ module_param(max_host_mem_size_mb, uint, 0444);
 MODULE_PARM_DESC(max_host_mem_size_mb,
 	"Maximum Host Memory Buffer (HMB) size per controller (in MiB)");
 
+static unsigned int sgl_threshold = SZ_32K;
+module_param(sgl_threshold, uint, 0644);
+MODULE_PARM_DESC(sgl_threshold,
+		"Use SGLs when average request segment size is larger or equal to "
+		"this size. Use 0 to disable SGLs.");
+
 static int io_queue_depth_set(const char *val, const struct kernel_param *kp);
 static const struct kernel_param_ops io_queue_depth_ops = {
 	.set = io_queue_depth_set,
@@ -178,6 +176,7 @@ struct nvme_queue {
 struct nvme_iod {
 	struct nvme_request req;
 	struct nvme_queue *nvmeq;
+	bool use_sgl;
 	int aborted;
 	int npages;		/* In the PRP list. 0 means small pool in use */
 	int nents;		/* Used in scatterlist */
@@ -331,17 +330,35 @@ static int nvme_npages(unsigned size, struct nvme_dev *dev)
 	return DIV_ROUND_UP(8 * nprps, PAGE_SIZE - 8);
 }
 
-static unsigned int nvme_iod_alloc_size(struct nvme_dev *dev,
-		unsigned int size, unsigned int nseg)
+/*
+ * Calculates the number of pages needed for the SGL segments. For example a 4k
+ * page can accommodate 256 SGL descriptors.
+ */
+static int nvme_pci_npages_sgl(unsigned int num_seg)
 {
-	return sizeof(__le64 *) * nvme_npages(size, dev) +
-			sizeof(struct scatterlist) * nseg;
+	return DIV_ROUND_UP(num_seg * sizeof(struct nvme_sgl_desc), PAGE_SIZE);
 }
 
-static unsigned int nvme_cmd_size(struct nvme_dev *dev)
+static unsigned int nvme_pci_iod_alloc_size(struct nvme_dev *dev,
+		unsigned int size, unsigned int nseg, bool use_sgl)
 {
-	return sizeof(struct nvme_iod) +
-		nvme_iod_alloc_size(dev, NVME_INT_BYTES(dev), NVME_INT_PAGES);
+	size_t alloc_size;
+
+	if (use_sgl)
+		alloc_size = sizeof(__le64 *) * nvme_pci_npages_sgl(nseg);
+	else
+		alloc_size = sizeof(__le64 *) * nvme_npages(size, dev);
+
+	return alloc_size + sizeof(struct scatterlist) * nseg;
+}
+
+static unsigned int nvme_pci_cmd_size(struct nvme_dev *dev, bool use_sgl)
+{
+	unsigned int alloc_size = nvme_pci_iod_alloc_size(dev,
+				    NVME_INT_BYTES(dev), NVME_INT_PAGES,
+				    use_sgl);
+
+	return sizeof(struct nvme_iod) + alloc_size;
 }
 
 static int nvme_admin_init_hctx(struct blk_mq_hw_ctx *hctx, void *data,
@@ -425,10 +442,10 @@ static void __nvme_submit_cmd(struct nvme_queue *nvmeq,
 	nvmeq->sq_tail = tail;
 }
 
-static __le64 **iod_list(struct request *req)
+static void **nvme_pci_iod_list(struct request *req)
 {
 	struct nvme_iod *iod = blk_mq_rq_to_pdu(req);
-	return (__le64 **)(iod->sg + blk_rq_nr_phys_segments(req));
+	return (void **)(iod->sg + blk_rq_nr_phys_segments(req));
 }
 
 static blk_status_t nvme_init_iod(struct request *rq, struct nvme_dev *dev)
@@ -438,7 +455,10 @@ static blk_status_t nvme_init_iod(struct request *rq, struct nvme_dev *dev)
 	unsigned int size = blk_rq_payload_bytes(rq);
 
 	if (nseg > NVME_INT_PAGES || size > NVME_INT_BYTES(dev)) {
-		iod->sg = kmalloc(nvme_iod_alloc_size(dev, size, nseg), GFP_ATOMIC);
+		size_t alloc_size = nvme_pci_iod_alloc_size(dev, size, nseg,
+				iod->use_sgl);
+
+		iod->sg = kmalloc(alloc_size, GFP_ATOMIC);
 		if (!iod->sg)
 			return BLK_STS_RESOURCE;
 	} else {
@@ -456,18 +476,31 @@ static blk_status_t nvme_init_iod(struct request *rq, struct nvme_dev *dev)
 static void nvme_free_iod(struct nvme_dev *dev, struct request *req)
 {
 	struct nvme_iod *iod = blk_mq_rq_to_pdu(req);
-	const int last_prp = dev->ctrl.page_size / 8 - 1;
+	const int last_prp = dev->ctrl.page_size / sizeof(__le64) - 1;
+	dma_addr_t dma_addr = iod->first_dma, next_dma_addr;
+
 	int i;
-	__le64 **list = iod_list(req);
-	dma_addr_t prp_dma = iod->first_dma;
 
 	if (iod->npages == 0)
-		dma_pool_free(dev->prp_small_pool, list[0], prp_dma);
+		dma_pool_free(dev->prp_small_pool, nvme_pci_iod_list(req)[0],
+			dma_addr);
+
 	for (i = 0; i < iod->npages; i++) {
-		__le64 *prp_list = list[i];
-		dma_addr_t next_prp_dma = le64_to_cpu(prp_list[last_prp]);
-		dma_pool_free(dev->prp_page_pool, prp_list, prp_dma);
-		prp_dma = next_prp_dma;
+		void *addr = nvme_pci_iod_list(req)[i];
+
+		if (iod->use_sgl) {
+			struct nvme_sgl_desc *sg_list = addr;
+
+			next_dma_addr =
+			    le64_to_cpu((sg_list[SGES_PER_PAGE - 1]).addr);
+		} else {
+			__le64 *prp_list = addr;
+
+			next_dma_addr = le64_to_cpu(prp_list[last_prp]);
+		}
+
+		dma_pool_free(dev->prp_page_pool, addr, dma_addr);
+		dma_addr = next_dma_addr;
 	}
 
 	if (iod->sg != iod->inline_sg)
@@ -555,7 +588,8 @@ static void nvme_print_sgl(struct scatterlist *sgl, int nents)
 	}
 }
 
-static blk_status_t nvme_setup_prps(struct nvme_dev *dev, struct request *req)
+static blk_status_t nvme_pci_setup_prps(struct nvme_dev *dev,
+		struct request *req, struct nvme_rw_command *cmnd)
 {
 	struct nvme_iod *iod = blk_mq_rq_to_pdu(req);
 	struct dma_pool *pool;
@@ -566,14 +600,16 @@ static blk_status_t nvme_setup_prps(struct nvme_dev *dev, struct request *req)
 	u32 page_size = dev->ctrl.page_size;
 	int offset = dma_addr & (page_size - 1);
 	__le64 *prp_list;
-	__le64 **list = iod_list(req);
+	void **list = nvme_pci_iod_list(req);
 	dma_addr_t prp_dma;
 	int nprps, i;
 
+	iod->use_sgl = false;
+
 	length -= (page_size - offset);
 	if (length <= 0) {
 		iod->first_dma = 0;
-		return BLK_STS_OK;
+		goto done;
 	}
 
 	dma_len -= (page_size - offset);
@@ -587,7 +623,7 @@ static blk_status_t nvme_setup_prps(struct nvme_dev *dev, struct request *req)
 
 	if (length <= page_size) {
 		iod->first_dma = dma_addr;
-		return BLK_STS_OK;
+		goto done;
 	}
 
 	nprps = DIV_ROUND_UP(length, page_size);
@@ -634,6 +670,10 @@ static blk_status_t nvme_setup_prps(struct nvme_dev *dev, struct request *req)
 		dma_len = sg_dma_len(sg);
 	}
 
+done:
+	cmnd->dptr.prp1 = cpu_to_le64(sg_dma_address(iod->sg));
+	cmnd->dptr.prp2 = cpu_to_le64(iod->first_dma);
+
 	return BLK_STS_OK;
 
  bad_sgl:
@@ -643,6 +683,110 @@ static blk_status_t nvme_setup_prps(struct nvme_dev *dev, struct request *req)
 	return BLK_STS_IOERR;
 }
 
+static void nvme_pci_sgl_set_data(struct nvme_sgl_desc *sge,
+		struct scatterlist *sg)
+{
+	sge->addr = cpu_to_le64(sg_dma_address(sg));
+	sge->length = cpu_to_le32(sg_dma_len(sg));
+	sge->type = NVME_SGL_FMT_DATA_DESC << 4;
+}
+
+static void nvme_pci_sgl_set_seg(struct nvme_sgl_desc *sge,
+		dma_addr_t dma_addr, int entries)
+{
+	sge->addr = cpu_to_le64(dma_addr);
+	if (entries < SGES_PER_PAGE) {
+		sge->length = cpu_to_le32(entries * sizeof(*sge));
+		sge->type = NVME_SGL_FMT_LAST_SEG_DESC << 4;
+	} else {
+		sge->length = cpu_to_le32(PAGE_SIZE);
+		sge->type = NVME_SGL_FMT_SEG_DESC << 4;
+	}
+}
+
+static blk_status_t nvme_pci_setup_sgls(struct nvme_dev *dev,
+		struct request *req, struct nvme_rw_command *cmd)
+{
+	struct nvme_iod *iod = blk_mq_rq_to_pdu(req);
+	int length = blk_rq_payload_bytes(req);
+	struct dma_pool *pool;
+	struct nvme_sgl_desc *sg_list;
+	struct scatterlist *sg = iod->sg;
+	int entries = iod->nents, i = 0;
+	dma_addr_t sgl_dma;
+
+	iod->use_sgl = true;
+
+	/* setting the transfer type as SGL */
+	cmd->flags = NVME_CMD_SGL_METABUF;
+
+	if (length == sg_dma_len(sg)) {
+		nvme_pci_sgl_set_data(&cmd->dptr.sgl, sg);
+		return BLK_STS_OK;
+	}
+
+	if (entries <= (256 / sizeof(struct nvme_sgl_desc))) {
+		pool = dev->prp_small_pool;
+		iod->npages = 0;
+	} else {
+		pool = dev->prp_page_pool;
+		iod->npages = 1;
+	}
+
+	sg_list = dma_pool_alloc(pool, GFP_ATOMIC, &sgl_dma);
+	if (!sg_list) {
+		iod->npages = -1;
+		return BLK_STS_RESOURCE;
+	}
+
+	nvme_pci_iod_list(req)[0] = sg_list;
+	iod->first_dma = sgl_dma;
+
+	nvme_pci_sgl_set_seg(&cmd->dptr.sgl, sgl_dma, entries);
+
+	do {
+		if (i == SGES_PER_PAGE) {
+			struct nvme_sgl_desc *old_sg_desc = sg_list;
+			struct nvme_sgl_desc *link = &old_sg_desc[i - 1];
+
+			sg_list = dma_pool_alloc(pool, GFP_ATOMIC, &sgl_dma);
+			if (!sg_list)
+				return BLK_STS_RESOURCE;
+
+			i = 0;
+			nvme_pci_iod_list(req)[iod->npages++] = sg_list;
+			sg_list[i++] = *link;
+			nvme_pci_sgl_set_seg(link, sgl_dma, entries);
+		}
+
+		nvme_pci_sgl_set_data(&sg_list[i++], sg);
+
+		length -= sg_dma_len(sg);
+		sg = sg_next(sg);
+		entries--;
+	} while (length > 0);
+
+	WARN_ON(entries > 0);
+	return BLK_STS_OK;
+}
+
+static inline bool nvme_pci_use_sgls(struct nvme_dev *dev, struct request *req)
+{
+	struct nvme_iod *iod = blk_mq_rq_to_pdu(req);
+	unsigned int avg_seg_size;
+
+	avg_seg_size = DIV_ROUND_UP(blk_rq_payload_bytes(req),
+			blk_rq_nr_phys_segments(req));
+
+	if (!(dev->ctrl.sgls & ((1 << 0) | (1 << 1))))
+		return false;
+	if (!iod->nvmeq->qid)
+		return false;
+	if (!sgl_threshold || avg_seg_size < sgl_threshold)
+		return false;
+	return true;
+}
+
 static blk_status_t nvme_map_data(struct nvme_dev *dev, struct request *req,
 		struct nvme_command *cmnd)
 {
@@ -662,7 +806,11 @@ static blk_status_t nvme_map_data(struct nvme_dev *dev, struct request *req,
 				DMA_ATTR_NO_WARN))
 		goto out;
 
-	ret = nvme_setup_prps(dev, req);
+	if (nvme_pci_use_sgls(dev, req))
+		ret = nvme_pci_setup_sgls(dev, req, &cmnd->rw);
+	else
+		ret = nvme_pci_setup_prps(dev, req, &cmnd->rw);
+
 	if (ret != BLK_STS_OK)
 		goto out_unmap;
 
@@ -682,8 +830,6 @@ static blk_status_t nvme_map_data(struct nvme_dev *dev, struct request *req,
 			goto out_unmap;
 	}
 
-	cmnd->rw.dptr.prp1 = cpu_to_le64(sg_dma_address(iod->sg));
-	cmnd->rw.dptr.prp2 = cpu_to_le64(iod->first_dma);
 	if (blk_integrity_rq(req))
 		cmnd->rw.metadata = cpu_to_le64(sg_dma_address(&iod->meta_sg));
 	return BLK_STS_OK;
@@ -804,7 +950,7 @@ static inline void nvme_handle_cqe(struct nvme_queue *nvmeq,
 	 * for them but rather special case them here.
 	 */
 	if (unlikely(nvmeq->qid == 0 &&
-			cqe->command_id >= NVME_AQ_BLKMQ_DEPTH)) {
+			cqe->command_id >= NVME_AQ_BLK_MQ_DEPTH)) {
 		nvme_complete_async_event(&nvmeq->dev->ctrl,
 				cqe->status, &cqe->result);
 		return;
@@ -897,7 +1043,7 @@ static int nvme_poll(struct blk_mq_hw_ctx *hctx, unsigned int tag)
 	return __nvme_poll(nvmeq, tag);
 }
 
-static void nvme_pci_submit_async_event(struct nvme_ctrl *ctrl, int aer_idx)
+static void nvme_pci_submit_async_event(struct nvme_ctrl *ctrl)
 {
 	struct nvme_dev *dev = to_nvme_dev(ctrl);
 	struct nvme_queue *nvmeq = dev->queues[0];
@@ -905,7 +1051,7 @@ static void nvme_pci_submit_async_event(struct nvme_ctrl *ctrl, int aer_idx)
 
 	memset(&c, 0, sizeof(c));
 	c.common.opcode = nvme_admin_async_event;
-	c.common.command_id = NVME_AQ_BLKMQ_DEPTH + aer_idx;
+	c.common.command_id = NVME_AQ_BLK_MQ_DEPTH;
 
 	spin_lock_irq(&nvmeq->q_lock);
 	__nvme_submit_cmd(nvmeq, &c);
@@ -930,7 +1076,7 @@ static int adapter_alloc_cq(struct nvme_dev *dev, u16 qid,
 	int flags = NVME_QUEUE_PHYS_CONTIG | NVME_CQ_IRQ_ENABLED;
 
 	/*
-	 * Note: we (ab)use the fact the the prp fields survive if no data
+	 * Note: we (ab)use the fact that the prp fields survive if no data
 	 * is attached to the request.
 	 */
 	memset(&c, 0, sizeof(c));
@@ -951,7 +1097,7 @@ static int adapter_alloc_sq(struct nvme_dev *dev, u16 qid,
 	int flags = NVME_QUEUE_PHYS_CONTIG;
 
 	/*
-	 * Note: we (ab)use the fact the the prp fields survive if no data
+	 * Note: we (ab)use the fact that the prp fields survive if no data
 	 * is attached to the request.
 	 */
 	memset(&c, 0, sizeof(c));
@@ -1372,14 +1518,10 @@ static int nvme_alloc_admin_tags(struct nvme_dev *dev)
 		dev->admin_tagset.ops = &nvme_mq_admin_ops;
 		dev->admin_tagset.nr_hw_queues = 1;
 
-		/*
-		 * Subtract one to leave an empty queue entry for 'Full Queue'
-		 * condition. See NVM-Express 1.2 specification, section 4.1.2.
-		 */
-		dev->admin_tagset.queue_depth = NVME_AQ_BLKMQ_DEPTH - 1;
+		dev->admin_tagset.queue_depth = NVME_AQ_MQ_TAG_DEPTH;
 		dev->admin_tagset.timeout = ADMIN_TIMEOUT;
 		dev->admin_tagset.numa_node = dev_to_node(dev->dev);
-		dev->admin_tagset.cmd_size = nvme_cmd_size(dev);
+		dev->admin_tagset.cmd_size = nvme_pci_cmd_size(dev, false);
 		dev->admin_tagset.flags = BLK_MQ_F_NO_SCHED;
 		dev->admin_tagset.driver_data = dev;
 
@@ -1906,7 +2048,11 @@ static int nvme_dev_add(struct nvme_dev *dev)
 		dev->tagset.numa_node = dev_to_node(dev->dev);
 		dev->tagset.queue_depth =
 				min_t(int, dev->q_depth, BLK_MQ_MAX_DEPTH) - 1;
-		dev->tagset.cmd_size = nvme_cmd_size(dev);
+		dev->tagset.cmd_size = nvme_pci_cmd_size(dev, false);
+		if ((dev->ctrl.sgls & ((1 << 0) | (1 << 1))) && sgl_threshold) {
+			dev->tagset.cmd_size = max(dev->tagset.cmd_size,
+					nvme_pci_cmd_size(dev, true));
+		}
 		dev->tagset.flags = BLK_MQ_F_SHOULD_MERGE;
 		dev->tagset.driver_data = dev;
 
@@ -2132,9 +2278,9 @@ static void nvme_remove_dead_ctrl(struct nvme_dev *dev, int status)
 {
 	dev_warn(dev->ctrl.device, "Removing after probe failure status: %d\n", status);
 
-	kref_get(&dev->ctrl.kref);
+	nvme_get_ctrl(&dev->ctrl);
 	nvme_dev_disable(dev, false);
-	if (!schedule_work(&dev->remove_work))
+	if (!queue_work(nvme_wq, &dev->remove_work))
 		nvme_put_ctrl(&dev->ctrl);
 }
 
@@ -2557,6 +2703,7 @@ static int __init nvme_init(void)
 static void __exit nvme_exit(void)
 {
 	pci_unregister_driver(&nvme_driver);
+	flush_workqueue(nvme_wq);
 	_nvme_check_size();
 }
 

drivers/nvme/host/rdma.c

@@ -41,17 +41,9 @@
 
 #define NVME_RDMA_MAX_INLINE_SEGMENTS	1
 
-/*
- * We handle AEN commands ourselves and don't even let the
- * block layer know about them.
- */
-#define NVME_RDMA_NR_AEN_COMMANDS      1
-#define NVME_RDMA_AQ_BLKMQ_DEPTH       \
-	(NVME_AQ_DEPTH - NVME_RDMA_NR_AEN_COMMANDS)
-
 struct nvme_rdma_device {
-	struct ib_device       *dev;
-	struct ib_pd	       *pd;
+	struct ib_device	*dev;
+	struct ib_pd		*pd;
 	struct kref		ref;
 	struct list_head	entry;
 };
@@ -79,8 +71,8 @@ struct nvme_rdma_request {
 };
 
 enum nvme_rdma_queue_flags {
-	NVME_RDMA_Q_LIVE		= 0,
-	NVME_RDMA_Q_DELETING		= 1,
+	NVME_RDMA_Q_ALLOCATED		= 0,
+	NVME_RDMA_Q_LIVE		= 1,
 };
 
 struct nvme_rdma_queue {
@@ -105,7 +97,6 @@ struct nvme_rdma_ctrl {
 
 	/* other member variables */
 	struct blk_mq_tag_set	tag_set;
-	struct work_struct	delete_work;
 	struct work_struct	err_work;
 
 	struct nvme_rdma_qe	async_event_sqe;
@@ -274,6 +265,9 @@ static int nvme_rdma_reinit_request(void *data, struct request *rq)
 	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,
@@ -434,11 +428,9 @@ out_err:
 
 static void nvme_rdma_destroy_queue_ib(struct nvme_rdma_queue *queue)
 {
-	struct nvme_rdma_device *dev;
-	struct ib_device *ibdev;
+	struct nvme_rdma_device *dev = queue->device;
+	struct ib_device *ibdev = dev->dev;
 
-	dev = queue->device;
-	ibdev = dev->dev;
 	rdma_destroy_qp(queue->cm_id);
 	ib_free_cq(queue->ib_cq);
 
@@ -493,7 +485,7 @@ static int nvme_rdma_create_queue_ib(struct nvme_rdma_queue *queue)
 	return 0;
 
 out_destroy_qp:
-	ib_destroy_qp(queue->qp);
+	rdma_destroy_qp(queue->cm_id);
 out_destroy_ib_cq:
 	ib_free_cq(queue->ib_cq);
 out_put_dev:
@@ -544,11 +536,11 @@ static int nvme_rdma_alloc_queue(struct nvme_rdma_ctrl *ctrl,
 	ret = nvme_rdma_wait_for_cm(queue);
 	if (ret) {
 		dev_info(ctrl->ctrl.device,
-			"rdma_resolve_addr wait failed (%d).\n", ret);
+			"rdma connection establishment failed (%d)\n", ret);
 		goto out_destroy_cm_id;
 	}
 
-	clear_bit(NVME_RDMA_Q_DELETING, &queue->flags);
+	set_bit(NVME_RDMA_Q_ALLOCATED, &queue->flags);
 
 	return 0;
 
@@ -568,7 +560,7 @@ static void nvme_rdma_stop_queue(struct nvme_rdma_queue *queue)
 
 static void nvme_rdma_free_queue(struct nvme_rdma_queue *queue)
 {
-	if (test_and_set_bit(NVME_RDMA_Q_DELETING, &queue->flags))
+	if (!test_and_clear_bit(NVME_RDMA_Q_ALLOCATED, &queue->flags))
 		return;
 
 	if (nvme_rdma_queue_idx(queue) == 0) {
@@ -676,11 +668,10 @@ out_free_queues:
 	return ret;
 }
 
-static void nvme_rdma_free_tagset(struct nvme_ctrl *nctrl, bool admin)
+static void nvme_rdma_free_tagset(struct nvme_ctrl *nctrl,
+		struct blk_mq_tag_set *set)
 {
 	struct nvme_rdma_ctrl *ctrl = to_rdma_ctrl(nctrl);
-	struct blk_mq_tag_set *set = admin ?
-			&ctrl->admin_tag_set : &ctrl->tag_set;
 
 	blk_mq_free_tag_set(set);
 	nvme_rdma_dev_put(ctrl->device);
@@ -697,7 +688,7 @@ static struct blk_mq_tag_set *nvme_rdma_alloc_tagset(struct nvme_ctrl *nctrl,
 		set = &ctrl->admin_tag_set;
 		memset(set, 0, sizeof(*set));
 		set->ops = &nvme_rdma_admin_mq_ops;
-		set->queue_depth = NVME_RDMA_AQ_BLKMQ_DEPTH;
+		set->queue_depth = NVME_AQ_MQ_TAG_DEPTH;
 		set->reserved_tags = 2; /* connect + keep-alive */
 		set->numa_node = NUMA_NO_NODE;
 		set->cmd_size = sizeof(struct nvme_rdma_request) +
@@ -705,6 +696,7 @@ static struct blk_mq_tag_set *nvme_rdma_alloc_tagset(struct nvme_ctrl *nctrl,
 		set->driver_data = ctrl;
 		set->nr_hw_queues = 1;
 		set->timeout = ADMIN_TIMEOUT;
+		set->flags = BLK_MQ_F_NO_SCHED;
 	} else {
 		set = &ctrl->tag_set;
 		memset(set, 0, sizeof(*set));
@@ -748,7 +740,7 @@ static void nvme_rdma_destroy_admin_queue(struct nvme_rdma_ctrl *ctrl,
 	nvme_rdma_stop_queue(&ctrl->queues[0]);
 	if (remove) {
 		blk_cleanup_queue(ctrl->ctrl.admin_q);
-		nvme_rdma_free_tagset(&ctrl->ctrl, true);
+		nvme_rdma_free_tagset(&ctrl->ctrl, ctrl->ctrl.admin_tagset);
 	}
 	nvme_rdma_free_queue(&ctrl->queues[0]);
 }
@@ -780,8 +772,7 @@ static int nvme_rdma_configure_admin_queue(struct nvme_rdma_ctrl *ctrl,
 			goto out_free_tagset;
 		}
 	} else {
-		error = blk_mq_reinit_tagset(&ctrl->admin_tag_set,
-					     nvme_rdma_reinit_request);
+		error = nvme_reinit_tagset(&ctrl->ctrl, ctrl->ctrl.admin_tagset);
 		if (error)
 			goto out_free_queue;
 	}
@@ -825,7 +816,7 @@ out_cleanup_queue:
 		blk_cleanup_queue(ctrl->ctrl.admin_q);
 out_free_tagset:
 	if (new)
-		nvme_rdma_free_tagset(&ctrl->ctrl, true);
+		nvme_rdma_free_tagset(&ctrl->ctrl, ctrl->ctrl.admin_tagset);
 out_free_queue:
 	nvme_rdma_free_queue(&ctrl->queues[0]);
 	return error;
@@ -837,7 +828,7 @@ static void nvme_rdma_destroy_io_queues(struct nvme_rdma_ctrl *ctrl,
 	nvme_rdma_stop_io_queues(ctrl);
 	if (remove) {
 		blk_cleanup_queue(ctrl->ctrl.connect_q);
-		nvme_rdma_free_tagset(&ctrl->ctrl, false);
+		nvme_rdma_free_tagset(&ctrl->ctrl, ctrl->ctrl.tagset);
 	}
 	nvme_rdma_free_io_queues(ctrl);
 }
@@ -863,8 +854,7 @@ static int nvme_rdma_configure_io_queues(struct nvme_rdma_ctrl *ctrl, bool new)
 			goto out_free_tag_set;
 		}
 	} else {
-		ret = blk_mq_reinit_tagset(&ctrl->tag_set,
-					   nvme_rdma_reinit_request);
+		ret = nvme_reinit_tagset(&ctrl->ctrl, ctrl->ctrl.tagset);
 		if (ret)
 			goto out_free_io_queues;
 
@@ -883,7 +873,7 @@ out_cleanup_connect_q:
 		blk_cleanup_queue(ctrl->ctrl.connect_q);
 out_free_tag_set:
 	if (new)
-		nvme_rdma_free_tagset(&ctrl->ctrl, false);
+		nvme_rdma_free_tagset(&ctrl->ctrl, ctrl->ctrl.tagset);
 out_free_io_queues:
 	nvme_rdma_free_io_queues(ctrl);
 	return ret;
@@ -922,7 +912,7 @@ static void nvme_rdma_reconnect_or_remove(struct nvme_rdma_ctrl *ctrl)
 				ctrl->ctrl.opts->reconnect_delay * HZ);
 	} else {
 		dev_info(ctrl->ctrl.device, "Removing controller...\n");
-		queue_work(nvme_wq, &ctrl->delete_work);
+		nvme_delete_ctrl(&ctrl->ctrl);
 	}
 }
 
@@ -935,10 +925,6 @@ static void nvme_rdma_reconnect_ctrl_work(struct work_struct *work)
 
 	++ctrl->ctrl.nr_reconnects;
 
-	if (ctrl->ctrl.queue_count > 1)
-		nvme_rdma_destroy_io_queues(ctrl, false);
-
-	nvme_rdma_destroy_admin_queue(ctrl, false);
 	ret = nvme_rdma_configure_admin_queue(ctrl, false);
 	if (ret)
 		goto requeue;
@@ -946,7 +932,7 @@ static void nvme_rdma_reconnect_ctrl_work(struct work_struct *work)
 	if (ctrl->ctrl.queue_count > 1) {
 		ret = nvme_rdma_configure_io_queues(ctrl, false);
 		if (ret)
-			goto requeue;
+			goto destroy_admin;
 	}
 
 	changed = nvme_change_ctrl_state(&ctrl->ctrl, NVME_CTRL_LIVE);
@@ -956,14 +942,17 @@ static void nvme_rdma_reconnect_ctrl_work(struct work_struct *work)
 		return;
 	}
 
-	ctrl->ctrl.nr_reconnects = 0;
-
 	nvme_start_ctrl(&ctrl->ctrl);
 
-	dev_info(ctrl->ctrl.device, "Successfully reconnected\n");
+	dev_info(ctrl->ctrl.device, "Successfully reconnected (%d attempts)\n",
+			ctrl->ctrl.nr_reconnects);
+
+	ctrl->ctrl.nr_reconnects = 0;
 
 	return;
 
+destroy_admin:
+	nvme_rdma_destroy_admin_queue(ctrl, false);
 requeue:
 	dev_info(ctrl->ctrl.device, "Failed reconnect attempt %d\n",
 			ctrl->ctrl.nr_reconnects);
@@ -979,17 +968,15 @@ static void nvme_rdma_error_recovery_work(struct work_struct *work)
 
 	if (ctrl->ctrl.queue_count > 1) {
 		nvme_stop_queues(&ctrl->ctrl);
-		nvme_rdma_stop_io_queues(ctrl);
-	}
-	blk_mq_quiesce_queue(ctrl->ctrl.admin_q);
-	nvme_rdma_stop_queue(&ctrl->queues[0]);
-
-	/* We must take care of fastfail/requeue all our inflight requests */
-	if (ctrl->ctrl.queue_count > 1)
 		blk_mq_tagset_busy_iter(&ctrl->tag_set,
 					nvme_cancel_request, &ctrl->ctrl);
+		nvme_rdma_destroy_io_queues(ctrl, false);
+	}
+
+	blk_mq_quiesce_queue(ctrl->ctrl.admin_q);
 	blk_mq_tagset_busy_iter(&ctrl->admin_tag_set,
 				nvme_cancel_request, &ctrl->ctrl);
+	nvme_rdma_destroy_admin_queue(ctrl, false);
 
 	/*
 	 * queues are not a live anymore, so restart the queues to fail fast
@@ -1065,7 +1052,7 @@ static void nvme_rdma_unmap_data(struct nvme_rdma_queue *queue,
 	if (!blk_rq_bytes(rq))
 		return;
 
-	if (req->mr->need_inval) {
+	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,
@@ -1314,7 +1301,7 @@ 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_submit_async_event(struct nvme_ctrl *arg, int aer_idx)
+static void nvme_rdma_submit_async_event(struct nvme_ctrl *arg)
 {
 	struct nvme_rdma_ctrl *ctrl = to_rdma_ctrl(arg);
 	struct nvme_rdma_queue *queue = &ctrl->queues[0];
@@ -1324,14 +1311,11 @@ static void nvme_rdma_submit_async_event(struct nvme_ctrl *arg, int aer_idx)
 	struct ib_sge sge;
 	int ret;
 
-	if (WARN_ON_ONCE(aer_idx != 0))
-		return;
-
 	ib_dma_sync_single_for_cpu(dev, sqe->dma, sizeof(*cmd), DMA_TO_DEVICE);
 
 	memset(cmd, 0, sizeof(*cmd));
 	cmd->common.opcode = nvme_admin_async_event;
-	cmd->common.command_id = NVME_RDMA_AQ_BLKMQ_DEPTH;
+	cmd->common.command_id = NVME_AQ_BLK_MQ_DEPTH;
 	cmd->common.flags |= NVME_CMD_SGL_METABUF;
 	nvme_rdma_set_sg_null(cmd);
 
@@ -1393,7 +1377,7 @@ static int __nvme_rdma_recv_done(struct ib_cq *cq, struct ib_wc *wc, int tag)
 	 * for them but rather special case them here.
 	 */
 	if (unlikely(nvme_rdma_queue_idx(queue) == 0 &&
-			cqe->command_id >= NVME_RDMA_AQ_BLKMQ_DEPTH))
+			cqe->command_id >= NVME_AQ_BLK_MQ_DEPTH))
 		nvme_complete_async_event(&queue->ctrl->ctrl, cqe->status,
 				&cqe->result);
 	else
@@ -1590,6 +1574,10 @@ nvme_rdma_timeout(struct request *rq, bool reserved)
 {
 	struct nvme_rdma_request *req = blk_mq_rq_to_pdu(rq);
 
+	dev_warn(req->queue->ctrl->ctrl.device,
+		 "I/O %d QID %d timeout, reset controller\n",
+		 rq->tag, nvme_rdma_queue_idx(req->queue));
+
 	/* queue error recovery */
 	nvme_rdma_error_recovery(req->queue->ctrl);
 
@@ -1767,50 +1755,9 @@ static void nvme_rdma_shutdown_ctrl(struct nvme_rdma_ctrl *ctrl, bool shutdown)
 	nvme_rdma_destroy_admin_queue(ctrl, shutdown);
 }
 
-static void nvme_rdma_remove_ctrl(struct nvme_rdma_ctrl *ctrl)
+static void nvme_rdma_delete_ctrl(struct nvme_ctrl *ctrl)
 {
-	nvme_remove_namespaces(&ctrl->ctrl);
-	nvme_rdma_shutdown_ctrl(ctrl, true);
-	nvme_uninit_ctrl(&ctrl->ctrl);
-	nvme_put_ctrl(&ctrl->ctrl);
-}
-
-static void nvme_rdma_del_ctrl_work(struct work_struct *work)
-{
-	struct nvme_rdma_ctrl *ctrl = container_of(work,
-				struct nvme_rdma_ctrl, delete_work);
-
-	nvme_stop_ctrl(&ctrl->ctrl);
-	nvme_rdma_remove_ctrl(ctrl);
-}
-
-static int __nvme_rdma_del_ctrl(struct nvme_rdma_ctrl *ctrl)
-{
-	if (!nvme_change_ctrl_state(&ctrl->ctrl, NVME_CTRL_DELETING))
-		return -EBUSY;
-
-	if (!queue_work(nvme_wq, &ctrl->delete_work))
-		return -EBUSY;
-
-	return 0;
-}
-
-static int nvme_rdma_del_ctrl(struct nvme_ctrl *nctrl)
-{
-	struct nvme_rdma_ctrl *ctrl = to_rdma_ctrl(nctrl);
-	int ret = 0;
-
-	/*
-	 * Keep a reference until all work is flushed since
-	 * __nvme_rdma_del_ctrl can free the ctrl mem
-	 */
-	if (!kref_get_unless_zero(&ctrl->ctrl.kref))
-		return -EBUSY;
-	ret = __nvme_rdma_del_ctrl(ctrl);
-	if (!ret)
-		flush_work(&ctrl->delete_work);
-	nvme_put_ctrl(&ctrl->ctrl);
-	return ret;
+	nvme_rdma_shutdown_ctrl(to_rdma_ctrl(ctrl), true);
 }
 
 static void nvme_rdma_reset_ctrl_work(struct work_struct *work)
@@ -1834,7 +1781,11 @@ static void nvme_rdma_reset_ctrl_work(struct work_struct *work)
 	}
 
 	changed = nvme_change_ctrl_state(&ctrl->ctrl, NVME_CTRL_LIVE);
-	WARN_ON_ONCE(!changed);
+	if (!changed) {
+		/* state change failure is ok if we're in DELETING state */
+		WARN_ON_ONCE(ctrl->ctrl.state != NVME_CTRL_DELETING);
+		return;
+	}
 
 	nvme_start_ctrl(&ctrl->ctrl);
 
@@ -1842,7 +1793,10 @@ static void nvme_rdma_reset_ctrl_work(struct work_struct *work)
 
 out_fail:
 	dev_warn(ctrl->ctrl.device, "Removing after reset failure\n");
-	nvme_rdma_remove_ctrl(ctrl);
+	nvme_remove_namespaces(&ctrl->ctrl);
+	nvme_rdma_shutdown_ctrl(ctrl, true);
+	nvme_uninit_ctrl(&ctrl->ctrl);
+	nvme_put_ctrl(&ctrl->ctrl);
 }
 
 static const struct nvme_ctrl_ops nvme_rdma_ctrl_ops = {
@@ -1854,10 +1808,88 @@ static const struct nvme_ctrl_ops nvme_rdma_ctrl_ops = {
 	.reg_write32		= nvmf_reg_write32,
 	.free_ctrl		= nvme_rdma_free_ctrl,
 	.submit_async_event	= nvme_rdma_submit_async_event,
-	.delete_ctrl		= nvme_rdma_del_ctrl,
+	.delete_ctrl		= nvme_rdma_delete_ctrl,
 	.get_address		= nvmf_get_address,
+	.reinit_request		= nvme_rdma_reinit_request,
 };
 
+static inline bool
+__nvme_rdma_options_match(struct nvme_rdma_ctrl *ctrl,
+	struct nvmf_ctrl_options *opts)
+{
+	char *stdport = __stringify(NVME_RDMA_IP_PORT);
+
+
+	if (!nvmf_ctlr_matches_baseopts(&ctrl->ctrl, opts) ||
+	    strcmp(opts->traddr, ctrl->ctrl.opts->traddr))
+		return false;
+
+	if (opts->mask & NVMF_OPT_TRSVCID &&
+	    ctrl->ctrl.opts->mask & NVMF_OPT_TRSVCID) {
+		if (strcmp(opts->trsvcid, ctrl->ctrl.opts->trsvcid))
+			return false;
+	} else if (opts->mask & NVMF_OPT_TRSVCID) {
+		if (strcmp(opts->trsvcid, stdport))
+			return false;
+	} else if (ctrl->ctrl.opts->mask & NVMF_OPT_TRSVCID) {
+		if (strcmp(stdport, ctrl->ctrl.opts->trsvcid))
+			return false;
+	}
+	/* else, it's a match as both have stdport. Fall to next checks */
+
+	/*
+	 * checking the local address is rough. In most cases, one
+	 * is not specified and the host port is selected by the stack.
+	 *
+	 * Assume no match if:
+	 *  local address is specified and address is not the same
+	 *  local address is not specified but remote is, or vice versa
+	 *    (admin using specific host_traddr when it matters).
+	 */
+	if (opts->mask & NVMF_OPT_HOST_TRADDR &&
+	    ctrl->ctrl.opts->mask & NVMF_OPT_HOST_TRADDR) {
+		if (strcmp(opts->host_traddr, ctrl->ctrl.opts->host_traddr))
+			return false;
+	} else if (opts->mask & NVMF_OPT_HOST_TRADDR ||
+		   ctrl->ctrl.opts->mask & NVMF_OPT_HOST_TRADDR)
+		return false;
+	/*
+	 * if neither controller had an host port specified, assume it's
+	 * a match as everything else matched.
+	 */
+
+	return true;
+}
+
+/*
+ * Fails a connection request if it matches an existing controller
+ * (association) with the same tuple:
+ * <Host NQN, Host ID, local address, remote address, remote port, SUBSYS NQN>
+ *
+ * if local address is not specified in the request, it will match an
+ * existing controller with all the other parameters the same and no
+ * local port address specified as well.
+ *
+ * The ports don't need to be compared as they are intrinsically
+ * already matched by the port pointers supplied.
+ */
+static bool
+nvme_rdma_existing_controller(struct nvmf_ctrl_options *opts)
+{
+	struct nvme_rdma_ctrl *ctrl;
+	bool found = false;
+
+	mutex_lock(&nvme_rdma_ctrl_mutex);
+	list_for_each_entry(ctrl, &nvme_rdma_ctrl_list, list) {
+		found = __nvme_rdma_options_match(ctrl, opts);
+		if (found)
+			break;
+	}
+	mutex_unlock(&nvme_rdma_ctrl_mutex);
+
+	return found;
+}
+
 static struct nvme_ctrl *nvme_rdma_create_ctrl(struct device *dev,
 		struct nvmf_ctrl_options *opts)
 {
@@ -1894,6 +1926,11 @@ static struct nvme_ctrl *nvme_rdma_create_ctrl(struct device *dev,
 		}
 	}
 
+	if (!opts->duplicate_connect && nvme_rdma_existing_controller(opts)) {
+		ret = -EALREADY;
+		goto out_free_ctrl;
+	}
+
 	ret = nvme_init_ctrl(&ctrl->ctrl, dev, &nvme_rdma_ctrl_ops,
 				0 /* no quirks, we're perfect! */);
 	if (ret)
@@ -1902,7 +1939,6 @@ static struct nvme_ctrl *nvme_rdma_create_ctrl(struct device *dev,
 	INIT_DELAYED_WORK(&ctrl->reconnect_work,
 			nvme_rdma_reconnect_ctrl_work);
 	INIT_WORK(&ctrl->err_work, nvme_rdma_error_recovery_work);
-	INIT_WORK(&ctrl->delete_work, nvme_rdma_del_ctrl_work);
 	INIT_WORK(&ctrl->ctrl.reset_work, nvme_rdma_reset_ctrl_work);
 
 	ctrl->ctrl.queue_count = opts->nr_io_queues + 1; /* +1 for admin queue */
@@ -1961,7 +1997,7 @@ static struct nvme_ctrl *nvme_rdma_create_ctrl(struct device *dev,
 	dev_info(ctrl->ctrl.device, "new ctrl: NQN \"%s\", addr %pISpcs\n",
 		ctrl->ctrl.opts->subsysnqn, &ctrl->addr);
 
-	kref_get(&ctrl->ctrl.kref);
+	nvme_get_ctrl(&ctrl->ctrl);
 
 	mutex_lock(&nvme_rdma_ctrl_mutex);
 	list_add_tail(&ctrl->list, &nvme_rdma_ctrl_list);
@@ -2006,7 +2042,7 @@ static void nvme_rdma_remove_one(struct ib_device *ib_device, void *client_data)
 		dev_info(ctrl->ctrl.device,
 			"Removing ctrl: NQN \"%s\", addr %pISp\n",
 			ctrl->ctrl.opts->subsysnqn, &ctrl->addr);
-		__nvme_rdma_del_ctrl(ctrl);
+		nvme_delete_ctrl(&ctrl->ctrl);
 	}
 	mutex_unlock(&nvme_rdma_ctrl_mutex);
 

drivers/nvme/target/admin-cmd.c

@@ -35,17 +35,14 @@ u32 nvmet_get_log_page_len(struct nvme_command *cmd)
 static u16 nvmet_get_smart_log_nsid(struct nvmet_req *req,
 		struct nvme_smart_log *slog)
 {
-	u16 status;
 	struct nvmet_ns *ns;
 	u64 host_reads, host_writes, data_units_read, data_units_written;
 
-	status = NVME_SC_SUCCESS;
 	ns = nvmet_find_namespace(req->sq->ctrl, req->cmd->get_log_page.nsid);
 	if (!ns) {
-		status = NVME_SC_INVALID_NS;
 		pr_err("nvmet : Could not find namespace id : %d\n",
 				le32_to_cpu(req->cmd->get_log_page.nsid));
-		goto out;
+		return NVME_SC_INVALID_NS;
 	}
 
 	host_reads = part_stat_read(ns->bdev->bd_part, ios[READ]);
@@ -58,20 +55,18 @@ static u16 nvmet_get_smart_log_nsid(struct nvmet_req *req,
 	put_unaligned_le64(host_writes, &slog->host_writes[0]);
 	put_unaligned_le64(data_units_written, &slog->data_units_written[0]);
 	nvmet_put_namespace(ns);
-out:
-	return status;
+
+	return NVME_SC_SUCCESS;
 }
 
 static u16 nvmet_get_smart_log_all(struct nvmet_req *req,
 		struct nvme_smart_log *slog)
 {
-	u16 status;
 	u64 host_reads = 0, host_writes = 0;
 	u64 data_units_read = 0, data_units_written = 0;
 	struct nvmet_ns *ns;
 	struct nvmet_ctrl *ctrl;
 
-	status = NVME_SC_SUCCESS;
 	ctrl = req->sq->ctrl;
 
 	rcu_read_lock();
@@ -91,7 +86,7 @@ static u16 nvmet_get_smart_log_all(struct nvmet_req *req,
 	put_unaligned_le64(host_writes, &slog->host_writes[0]);
 	put_unaligned_le64(data_units_written, &slog->data_units_written[0]);
 
-	return status;
+	return NVME_SC_SUCCESS;
 }
 
 static u16 nvmet_get_smart_log(struct nvmet_req *req,
@@ -144,10 +139,8 @@ static void nvmet_execute_get_log_page(struct nvmet_req *req)
 		}
 		smart_log = buf;
 		status = nvmet_get_smart_log(req, smart_log);
-		if (status) {
-			memset(buf, '\0', data_len);
+		if (status)
 			goto err;
-		}
 		break;
 	case NVME_LOG_FW_SLOT:
 		/*
@@ -300,7 +293,7 @@ static void nvmet_execute_identify_ns(struct nvmet_req *req)
 	}
 
 	/*
-	 * nuse = ncap = nsze isn't aways true, but we have no way to find
+	 * nuse = ncap = nsze isn't always true, but we have no way to find
 	 * that out from the underlying device.
 	 */
 	id->ncap = id->nuse = id->nsze =
@@ -424,7 +417,7 @@ out:
 }
 
 /*
- * A "mimimum viable" abort implementation: the command is mandatory in the
+ * A "minimum viable" abort implementation: the command is mandatory in the
  * spec, but we are not required to do any useful work.  We couldn't really
  * do a useful abort, so don't bother even with waiting for the command
  * to be exectuted and return immediately telling the command to abort

drivers/nvme/target/core.c

@@ -57,6 +57,17 @@ u16 nvmet_copy_from_sgl(struct nvmet_req *req, off_t off, void *buf, size_t len)
 	return 0;
 }
 
+static unsigned int nvmet_max_nsid(struct nvmet_subsys *subsys)
+{
+	struct nvmet_ns *ns;
+
+	if (list_empty(&subsys->namespaces))
+		return 0;
+
+	ns = list_last_entry(&subsys->namespaces, struct nvmet_ns, dev_link);
+	return ns->nsid;
+}
+
 static u32 nvmet_async_event_result(struct nvmet_async_event *aen)
 {
 	return aen->event_type | (aen->event_info << 8) | (aen->log_page << 16);
@@ -334,6 +345,8 @@ void nvmet_ns_disable(struct nvmet_ns *ns)
 
 	ns->enabled = false;
 	list_del_rcu(&ns->dev_link);
+	if (ns->nsid == subsys->max_nsid)
+		subsys->max_nsid = nvmet_max_nsid(subsys);
 	mutex_unlock(&subsys->lock);
 
 	/*
@@ -497,6 +510,7 @@ bool nvmet_req_init(struct nvmet_req *req, struct nvmet_cq *cq,
 	req->ops = ops;
 	req->sg = NULL;
 	req->sg_cnt = 0;
+	req->transfer_len = 0;
 	req->rsp->status = 0;
 
 	/* no support for fused commands yet */
@@ -546,6 +560,15 @@ void nvmet_req_uninit(struct nvmet_req *req)
 }
 EXPORT_SYMBOL_GPL(nvmet_req_uninit);
 
+void nvmet_req_execute(struct nvmet_req *req)
+{
+	if (unlikely(req->data_len != req->transfer_len))
+		nvmet_req_complete(req, NVME_SC_SGL_INVALID_DATA | NVME_SC_DNR);
+	else
+		req->execute(req);
+}
+EXPORT_SYMBOL_GPL(nvmet_req_execute);
+
 static inline bool nvmet_cc_en(u32 cc)
 {
 	return (cc >> NVME_CC_EN_SHIFT) & 0x1;

drivers/nvme/target/fc.c

@@ -76,7 +76,6 @@ struct nvmet_fc_fcp_iod {
 	dma_addr_t			rspdma;
 	struct scatterlist		*data_sg;
 	int				data_sg_cnt;
-	u32				total_length;
 	u32				offset;
 	enum nvmet_fcp_datadir		io_dir;
 	bool				active;
@@ -150,6 +149,7 @@ struct nvmet_fc_tgt_assoc {
 	struct list_head		a_list;
 	struct nvmet_fc_tgt_queue	*queues[NVMET_NR_QUEUES + 1];
 	struct kref			ref;
+	struct work_struct		del_work;
 };
 
 
@@ -232,6 +232,7 @@ static void nvmet_fc_tgtport_put(struct nvmet_fc_tgtport *tgtport);
 static int nvmet_fc_tgtport_get(struct nvmet_fc_tgtport *tgtport);
 static void nvmet_fc_handle_fcp_rqst(struct nvmet_fc_tgtport *tgtport,
 					struct nvmet_fc_fcp_iod *fod);
+static void nvmet_fc_delete_target_assoc(struct nvmet_fc_tgt_assoc *assoc);
 
 
 /* *********************** FC-NVME DMA Handling **************************** */
@@ -802,6 +803,16 @@ nvmet_fc_find_target_queue(struct nvmet_fc_tgtport *tgtport,
 	return NULL;
 }
 
+static void
+nvmet_fc_delete_assoc(struct work_struct *work)
+{
+	struct nvmet_fc_tgt_assoc *assoc =
+		container_of(work, struct nvmet_fc_tgt_assoc, del_work);
+
+	nvmet_fc_delete_target_assoc(assoc);
+	nvmet_fc_tgt_a_put(assoc);
+}
+
 static struct nvmet_fc_tgt_assoc *
 nvmet_fc_alloc_target_assoc(struct nvmet_fc_tgtport *tgtport)
 {
@@ -826,6 +837,7 @@ nvmet_fc_alloc_target_assoc(struct nvmet_fc_tgtport *tgtport)
 	assoc->a_id = idx;
 	INIT_LIST_HEAD(&assoc->a_list);
 	kref_init(&assoc->ref);
+	INIT_WORK(&assoc->del_work, nvmet_fc_delete_assoc);
 
 	while (needrandom) {
 		get_random_bytes(&ran, sizeof(ran) - BYTES_FOR_QID);
@@ -1118,8 +1130,7 @@ nvmet_fc_delete_ctrl(struct nvmet_ctrl *ctrl)
 		nvmet_fc_tgtport_put(tgtport);
 
 		if (found_ctrl) {
-			nvmet_fc_delete_target_assoc(assoc);
-			nvmet_fc_tgt_a_put(assoc);
+			schedule_work(&assoc->del_work);
 			return;
 		}
 
@@ -1688,7 +1699,7 @@ nvmet_fc_alloc_tgt_pgs(struct nvmet_fc_fcp_iod *fod)
 	u32 page_len, length;
 	int i = 0;
 
-	length = fod->total_length;
+	length = fod->req.transfer_len;
 	nent = DIV_ROUND_UP(length, PAGE_SIZE);
 	sg = kmalloc_array(nent, sizeof(struct scatterlist), GFP_KERNEL);
 	if (!sg)
@@ -1777,7 +1788,7 @@ nvmet_fc_prep_fcp_rsp(struct nvmet_fc_tgtport *tgtport,
 	u32 rsn, rspcnt, xfr_length;
 
 	if (fod->fcpreq->op == NVMET_FCOP_READDATA_RSP)
-		xfr_length = fod->total_length;
+		xfr_length = fod->req.transfer_len;
 	else
 		xfr_length = fod->offset;
 
@@ -1803,7 +1814,7 @@ nvmet_fc_prep_fcp_rsp(struct nvmet_fc_tgtport *tgtport,
 	rspcnt = atomic_inc_return(&fod->queue->zrspcnt);
 	if (!(rspcnt % fod->queue->ersp_ratio) ||
 	    sqe->opcode == nvme_fabrics_command ||
-	    xfr_length != fod->total_length ||
+	    xfr_length != fod->req.transfer_len ||
 	    (le16_to_cpu(cqe->status) & 0xFFFE) || cqewd[0] || cqewd[1] ||
 	    (sqe->flags & (NVME_CMD_FUSE_FIRST | NVME_CMD_FUSE_SECOND)) ||
 	    queue_90percent_full(fod->queue, le16_to_cpu(cqe->sq_head)))
@@ -1880,7 +1891,7 @@ nvmet_fc_transfer_fcp_data(struct nvmet_fc_tgtport *tgtport,
 	fcpreq->timeout = NVME_FC_TGTOP_TIMEOUT_SEC;
 
 	tlen = min_t(u32, tgtport->max_sg_cnt * PAGE_SIZE,
-			(fod->total_length - fod->offset));
+			(fod->req.transfer_len - fod->offset));
 	fcpreq->transfer_length = tlen;
 	fcpreq->transferred_length = 0;
 	fcpreq->fcp_error = 0;
@@ -1894,7 +1905,7 @@ nvmet_fc_transfer_fcp_data(struct nvmet_fc_tgtport *tgtport,
 	 * combined xfr with response.
 	 */
 	if ((op == NVMET_FCOP_READDATA) &&
-	    ((fod->offset + fcpreq->transfer_length) == fod->total_length) &&
+	    ((fod->offset + fcpreq->transfer_length) == fod->req.transfer_len) &&
 	    (tgtport->ops->target_features & NVMET_FCTGTFEAT_READDATA_RSP)) {
 		fcpreq->op = NVMET_FCOP_READDATA_RSP;
 		nvmet_fc_prep_fcp_rsp(tgtport, fod);
@@ -1974,7 +1985,7 @@ nvmet_fc_fod_op_done(struct nvmet_fc_fcp_iod *fod)
 		}
 
 		fod->offset += fcpreq->transferred_length;
-		if (fod->offset != fod->total_length) {
+		if (fod->offset != fod->req.transfer_len) {
 			spin_lock_irqsave(&fod->flock, flags);
 			fod->writedataactive = true;
 			spin_unlock_irqrestore(&fod->flock, flags);
@@ -1986,9 +1997,7 @@ nvmet_fc_fod_op_done(struct nvmet_fc_fcp_iod *fod)
 		}
 
 		/* data transfer complete, resume with nvmet layer */
-
-		fod->req.execute(&fod->req);
-
+		nvmet_req_execute(&fod->req);
 		break;
 
 	case NVMET_FCOP_READDATA:
@@ -2011,7 +2020,7 @@ nvmet_fc_fod_op_done(struct nvmet_fc_fcp_iod *fod)
 		}
 
 		fod->offset += fcpreq->transferred_length;
-		if (fod->offset != fod->total_length) {
+		if (fod->offset != fod->req.transfer_len) {
 			/* transfer the next chunk */
 			nvmet_fc_transfer_fcp_data(tgtport, fod,
 						NVMET_FCOP_READDATA);
@@ -2148,7 +2157,7 @@ nvmet_fc_handle_fcp_rqst(struct nvmet_fc_tgtport *tgtport,
 
 	fod->fcpreq->done = nvmet_fc_xmt_fcp_op_done;
 
-	fod->total_length = be32_to_cpu(cmdiu->data_len);
+	fod->req.transfer_len = be32_to_cpu(cmdiu->data_len);
 	if (cmdiu->flags & FCNVME_CMD_FLAGS_WRITE) {
 		fod->io_dir = NVMET_FCP_WRITE;
 		if (!nvme_is_write(&cmdiu->sqe))
@@ -2159,7 +2168,7 @@ nvmet_fc_handle_fcp_rqst(struct nvmet_fc_tgtport *tgtport,
 			goto transport_error;
 	} else {
 		fod->io_dir = NVMET_FCP_NODATA;
-		if (fod->total_length)
+		if (fod->req.transfer_len)
 			goto transport_error;
 	}
 
@@ -2167,9 +2176,6 @@ nvmet_fc_handle_fcp_rqst(struct nvmet_fc_tgtport *tgtport,
 	fod->req.rsp = &fod->rspiubuf.cqe;
 	fod->req.port = fod->queue->port;
 
-	/* ensure nvmet handlers will set cmd handler callback */
-	fod->req.execute = NULL;
-
 	/* clear any response payload */
 	memset(&fod->rspiubuf, 0, sizeof(fod->rspiubuf));
 
@@ -2189,7 +2195,7 @@ nvmet_fc_handle_fcp_rqst(struct nvmet_fc_tgtport *tgtport,
 	/* keep a running counter of tail position */
 	atomic_inc(&fod->queue->sqtail);
 
-	if (fod->total_length) {
+	if (fod->req.transfer_len) {
 		ret = nvmet_fc_alloc_tgt_pgs(fod);
 		if (ret) {
 			nvmet_req_complete(&fod->req, ret);
@@ -2212,9 +2218,7 @@ nvmet_fc_handle_fcp_rqst(struct nvmet_fc_tgtport *tgtport,
 	 * can invoke the nvmet_layer now. If read data, cmd completion will
 	 * push the data
 	 */
-
-	fod->req.execute(&fod->req);
-
+	nvmet_req_execute(&fod->req);
 	return;
 
 transport_error:

drivers/nvme/target/io-cmd.c

@@ -33,18 +33,11 @@ static inline u32 nvmet_rw_len(struct nvmet_req *req)
 			req->ns->blksize_shift;
 }
 
-static void nvmet_inline_bio_init(struct nvmet_req *req)
-{
-	struct bio *bio = &req->inline_bio;
-
-	bio_init(bio, req->inline_bvec, NVMET_MAX_INLINE_BIOVEC);
-}
-
 static void nvmet_execute_rw(struct nvmet_req *req)
 {
 	int sg_cnt = req->sg_cnt;
+	struct bio *bio = &req->inline_bio;
 	struct scatterlist *sg;
-	struct bio *bio;
 	sector_t sector;
 	blk_qc_t cookie;
 	int op, op_flags = 0, i;
@@ -66,8 +59,7 @@ static void nvmet_execute_rw(struct nvmet_req *req)
 	sector = le64_to_cpu(req->cmd->rw.slba);
 	sector <<= (req->ns->blksize_shift - 9);
 
-	nvmet_inline_bio_init(req);
-	bio = &req->inline_bio;
+	bio_init(bio, req->inline_bvec, ARRAY_SIZE(req->inline_bvec));
 	bio_set_dev(bio, req->ns->bdev);
 	bio->bi_iter.bi_sector = sector;
 	bio->bi_private = req;
@@ -94,16 +86,14 @@ static void nvmet_execute_rw(struct nvmet_req *req)
 
 	cookie = submit_bio(bio);
 
-	blk_mq_poll(bdev_get_queue(req->ns->bdev), cookie);
+	blk_poll(bdev_get_queue(req->ns->bdev), cookie);
 }
 
 static void nvmet_execute_flush(struct nvmet_req *req)
 {
-	struct bio *bio;
-
-	nvmet_inline_bio_init(req);
-	bio = &req->inline_bio;
+	struct bio *bio = &req->inline_bio;
 
+	bio_init(bio, req->inline_bvec, ARRAY_SIZE(req->inline_bvec));
 	bio_set_dev(bio, req->ns->bdev);
 	bio->bi_private = req;
 	bio->bi_end_io = nvmet_bio_done;

drivers/nvme/target/loop.c

@@ -23,14 +23,6 @@
 
 #define NVME_LOOP_MAX_SEGMENTS		256
 
-/*
- * We handle AEN commands ourselves and don't even let the
- * block layer know about them.
- */
-#define NVME_LOOP_NR_AEN_COMMANDS	1
-#define NVME_LOOP_AQ_BLKMQ_DEPTH	\
-	(NVME_AQ_DEPTH - NVME_LOOP_NR_AEN_COMMANDS)
-
 struct nvme_loop_iod {
 	struct nvme_request	nvme_req;
 	struct nvme_command	cmd;
@@ -53,7 +45,6 @@ struct nvme_loop_ctrl {
 	struct nvme_ctrl	ctrl;
 
 	struct nvmet_ctrl	*target_ctrl;
-	struct work_struct	delete_work;
 };
 
 static inline struct nvme_loop_ctrl *to_loop_ctrl(struct nvme_ctrl *ctrl)
@@ -113,7 +104,7 @@ static void nvme_loop_queue_response(struct nvmet_req *req)
 	 * for them but rather special case them here.
 	 */
 	if (unlikely(nvme_loop_queue_idx(queue) == 0 &&
-			cqe->command_id >= NVME_LOOP_AQ_BLKMQ_DEPTH)) {
+			cqe->command_id >= NVME_AQ_BLK_MQ_DEPTH)) {
 		nvme_complete_async_event(&queue->ctrl->ctrl, cqe->status,
 				&cqe->result);
 	} else {
@@ -136,7 +127,7 @@ static void nvme_loop_execute_work(struct work_struct *work)
 	struct nvme_loop_iod *iod =
 		container_of(work, struct nvme_loop_iod, work);
 
-	iod->req.execute(&iod->req);
+	nvmet_req_execute(&iod->req);
 }
 
 static enum blk_eh_timer_return
@@ -185,6 +176,7 @@ static blk_status_t nvme_loop_queue_rq(struct blk_mq_hw_ctx *hctx,
 
 		iod->req.sg = iod->sg_table.sgl;
 		iod->req.sg_cnt = blk_rq_map_sg(req->q, req, iod->sg_table.sgl);
+		iod->req.transfer_len = blk_rq_bytes(req);
 	}
 
 	blk_mq_start_request(req);
@@ -193,7 +185,7 @@ static blk_status_t nvme_loop_queue_rq(struct blk_mq_hw_ctx *hctx,
 	return BLK_STS_OK;
 }
 
-static void nvme_loop_submit_async_event(struct nvme_ctrl *arg, int aer_idx)
+static void nvme_loop_submit_async_event(struct nvme_ctrl *arg)
 {
 	struct nvme_loop_ctrl *ctrl = to_loop_ctrl(arg);
 	struct nvme_loop_queue *queue = &ctrl->queues[0];
@@ -201,7 +193,7 @@ static void nvme_loop_submit_async_event(struct nvme_ctrl *arg, int aer_idx)
 
 	memset(&iod->cmd, 0, sizeof(iod->cmd));
 	iod->cmd.common.opcode = nvme_admin_async_event;
-	iod->cmd.common.command_id = NVME_LOOP_AQ_BLKMQ_DEPTH;
+	iod->cmd.common.command_id = NVME_AQ_BLK_MQ_DEPTH;
 	iod->cmd.common.flags |= NVME_CMD_SGL_METABUF;
 
 	if (!nvmet_req_init(&iod->req, &queue->nvme_cq, &queue->nvme_sq,
@@ -357,7 +349,7 @@ static int nvme_loop_configure_admin_queue(struct nvme_loop_ctrl *ctrl)
 
 	memset(&ctrl->admin_tag_set, 0, sizeof(ctrl->admin_tag_set));
 	ctrl->admin_tag_set.ops = &nvme_loop_admin_mq_ops;
-	ctrl->admin_tag_set.queue_depth = NVME_LOOP_AQ_BLKMQ_DEPTH;
+	ctrl->admin_tag_set.queue_depth = NVME_AQ_MQ_TAG_DEPTH;
 	ctrl->admin_tag_set.reserved_tags = 2; /* connect + keep-alive */
 	ctrl->admin_tag_set.numa_node = NUMA_NO_NODE;
 	ctrl->admin_tag_set.cmd_size = sizeof(struct nvme_loop_iod) +
@@ -365,6 +357,7 @@ static int nvme_loop_configure_admin_queue(struct nvme_loop_ctrl *ctrl)
 	ctrl->admin_tag_set.driver_data = ctrl;
 	ctrl->admin_tag_set.nr_hw_queues = 1;
 	ctrl->admin_tag_set.timeout = ADMIN_TIMEOUT;
+	ctrl->admin_tag_set.flags = BLK_MQ_F_NO_SCHED;
 
 	ctrl->queues[0].ctrl = ctrl;
 	error = nvmet_sq_init(&ctrl->queues[0].nvme_sq);
@@ -438,41 +431,9 @@ static void nvme_loop_shutdown_ctrl(struct nvme_loop_ctrl *ctrl)
 	nvme_loop_destroy_admin_queue(ctrl);
 }
 
-static void nvme_loop_del_ctrl_work(struct work_struct *work)
+static void nvme_loop_delete_ctrl_host(struct nvme_ctrl *ctrl)
 {
-	struct nvme_loop_ctrl *ctrl = container_of(work,
-				struct nvme_loop_ctrl, delete_work);
-
-	nvme_stop_ctrl(&ctrl->ctrl);
-	nvme_remove_namespaces(&ctrl->ctrl);
-	nvme_loop_shutdown_ctrl(ctrl);
-	nvme_uninit_ctrl(&ctrl->ctrl);
-	nvme_put_ctrl(&ctrl->ctrl);
-}
-
-static int __nvme_loop_del_ctrl(struct nvme_loop_ctrl *ctrl)
-{
-	if (!nvme_change_ctrl_state(&ctrl->ctrl, NVME_CTRL_DELETING))
-		return -EBUSY;
-
-	if (!queue_work(nvme_wq, &ctrl->delete_work))
-		return -EBUSY;
-
-	return 0;
-}
-
-static int nvme_loop_del_ctrl(struct nvme_ctrl *nctrl)
-{
-	struct nvme_loop_ctrl *ctrl = to_loop_ctrl(nctrl);
-	int ret;
-
-	ret = __nvme_loop_del_ctrl(ctrl);
-	if (ret)
-		return ret;
-
-	flush_work(&ctrl->delete_work);
-
-	return 0;
+	nvme_loop_shutdown_ctrl(to_loop_ctrl(ctrl));
 }
 
 static void nvme_loop_delete_ctrl(struct nvmet_ctrl *nctrl)
@@ -482,7 +443,7 @@ static void nvme_loop_delete_ctrl(struct nvmet_ctrl *nctrl)
 	mutex_lock(&nvme_loop_ctrl_mutex);
 	list_for_each_entry(ctrl, &nvme_loop_ctrl_list, list) {
 		if (ctrl->ctrl.cntlid == nctrl->cntlid)
-			__nvme_loop_del_ctrl(ctrl);
+			nvme_delete_ctrl(&ctrl->ctrl);
 	}
 	mutex_unlock(&nvme_loop_ctrl_mutex);
 }
@@ -538,7 +499,7 @@ static const struct nvme_ctrl_ops nvme_loop_ctrl_ops = {
 	.reg_write32		= nvmf_reg_write32,
 	.free_ctrl		= nvme_loop_free_ctrl,
 	.submit_async_event	= nvme_loop_submit_async_event,
-	.delete_ctrl		= nvme_loop_del_ctrl,
+	.delete_ctrl		= nvme_loop_delete_ctrl_host,
 };
 
 static int nvme_loop_create_io_queues(struct nvme_loop_ctrl *ctrl)
@@ -600,7 +561,6 @@ static struct nvme_ctrl *nvme_loop_create_ctrl(struct device *dev,
 	ctrl->ctrl.opts = opts;
 	INIT_LIST_HEAD(&ctrl->list);
 
-	INIT_WORK(&ctrl->delete_work, nvme_loop_del_ctrl_work);
 	INIT_WORK(&ctrl->ctrl.reset_work, nvme_loop_reset_ctrl_work);
 
 	ret = nvme_init_ctrl(&ctrl->ctrl, dev, &nvme_loop_ctrl_ops,
@@ -641,7 +601,7 @@ static struct nvme_ctrl *nvme_loop_create_ctrl(struct device *dev,
 	dev_info(ctrl->ctrl.device,
 		 "new ctrl: \"%s\"\n", ctrl->ctrl.opts->subsysnqn);
 
-	kref_get(&ctrl->ctrl.kref);
+	nvme_get_ctrl(&ctrl->ctrl);
 
 	changed = nvme_change_ctrl_state(&ctrl->ctrl, NVME_CTRL_LIVE);
 	WARN_ON_ONCE(!changed);
@@ -730,7 +690,7 @@ static void __exit nvme_loop_cleanup_module(void)
 
 	mutex_lock(&nvme_loop_ctrl_mutex);
 	list_for_each_entry_safe(ctrl, next, &nvme_loop_ctrl_list, list)
-		__nvme_loop_del_ctrl(ctrl);
+		nvme_delete_ctrl(&ctrl->ctrl);
 	mutex_unlock(&nvme_loop_ctrl_mutex);
 
 	flush_workqueue(nvme_wq);

drivers/nvme/target/nvmet.h

@@ -223,7 +223,10 @@ struct nvmet_req {
 	struct bio		inline_bio;
 	struct bio_vec		inline_bvec[NVMET_MAX_INLINE_BIOVEC];
 	int			sg_cnt;
+	/* data length as parsed from the command: */
 	size_t			data_len;
+	/* data length as parsed from the SGL descriptor: */
+	size_t			transfer_len;
 
 	struct nvmet_port	*port;
 
@@ -266,6 +269,7 @@ u16 nvmet_parse_fabrics_cmd(struct nvmet_req *req);
 bool nvmet_req_init(struct nvmet_req *req, struct nvmet_cq *cq,
 		struct nvmet_sq *sq, struct nvmet_fabrics_ops *ops);
 void nvmet_req_uninit(struct nvmet_req *req);
+void nvmet_req_execute(struct nvmet_req *req);
 void nvmet_req_complete(struct nvmet_req *req, u16 status);
 
 void nvmet_cq_setup(struct nvmet_ctrl *ctrl, struct nvmet_cq *cq, u16 qid,
@@ -314,7 +318,7 @@ u16 nvmet_copy_from_sgl(struct nvmet_req *req, off_t off, void *buf,
 u32 nvmet_get_log_page_len(struct nvme_command *cmd);
 
 #define NVMET_QUEUE_SIZE	1024
-#define NVMET_NR_QUEUES		64
+#define NVMET_NR_QUEUES		128
 #define NVMET_MAX_CMD		NVMET_QUEUE_SIZE
 #define NVMET_KAS		10
 #define NVMET_DISC_KATO		120

drivers/nvme/target/rdma.c

@@ -148,14 +148,14 @@ static inline u32 get_unaligned_le24(const u8 *p)
 static inline bool nvmet_rdma_need_data_in(struct nvmet_rdma_rsp *rsp)
 {
 	return nvme_is_write(rsp->req.cmd) &&
-		rsp->req.data_len &&
+		rsp->req.transfer_len &&
 		!(rsp->flags & NVMET_RDMA_REQ_INLINE_DATA);
 }
 
 static inline bool nvmet_rdma_need_data_out(struct nvmet_rdma_rsp *rsp)
 {
 	return !nvme_is_write(rsp->req.cmd) &&
-		rsp->req.data_len &&
+		rsp->req.transfer_len &&
 		!rsp->req.rsp->status &&
 		!(rsp->flags & NVMET_RDMA_REQ_INLINE_DATA);
 }
@@ -577,7 +577,7 @@ static void nvmet_rdma_read_data_done(struct ib_cq *cq, struct ib_wc *wc)
 		return;
 	}
 
-	rsp->req.execute(&rsp->req);
+	nvmet_req_execute(&rsp->req);
 }
 
 static void nvmet_rdma_use_inline_sg(struct nvmet_rdma_rsp *rsp, u32 len,
@@ -609,6 +609,7 @@ static u16 nvmet_rdma_map_sgl_inline(struct nvmet_rdma_rsp *rsp)
 
 	nvmet_rdma_use_inline_sg(rsp, len, off);
 	rsp->flags |= NVMET_RDMA_REQ_INLINE_DATA;
+	rsp->req.transfer_len += len;
 	return 0;
 }
 
@@ -636,6 +637,7 @@ static u16 nvmet_rdma_map_sgl_keyed(struct nvmet_rdma_rsp *rsp,
 			nvmet_data_dir(&rsp->req));
 	if (ret < 0)
 		return NVME_SC_INTERNAL;
+	rsp->req.transfer_len += len;
 	rsp->n_rdma += ret;
 
 	if (invalidate) {
@@ -693,7 +695,7 @@ static bool nvmet_rdma_execute_command(struct nvmet_rdma_rsp *rsp)
 				queue->cm_id->port_num, &rsp->read_cqe, NULL))
 			nvmet_req_complete(&rsp->req, NVME_SC_DATA_XFER_ERROR);
 	} else {
-		rsp->req.execute(&rsp->req);
+		nvmet_req_execute(&rsp->req);
 	}
 
 	return true;
@@ -1512,15 +1514,17 @@ static struct nvmet_fabrics_ops nvmet_rdma_ops = {
 
 static void nvmet_rdma_remove_one(struct ib_device *ib_device, void *client_data)
 {
-	struct nvmet_rdma_queue *queue;
+	struct nvmet_rdma_queue *queue, *tmp;
 
 	/* Device is being removed, delete all queues using this device */
 	mutex_lock(&nvmet_rdma_queue_mutex);
-	list_for_each_entry(queue, &nvmet_rdma_queue_list, queue_list) {
+	list_for_each_entry_safe(queue, tmp, &nvmet_rdma_queue_list,
+				 queue_list) {
 		if (queue->dev->device != ib_device)
 			continue;
 
 		pr_info("Removing queue %d\n", queue->idx);
+		list_del_init(&queue->queue_list);
 		__nvmet_rdma_queue_disconnect(queue);
 	}
 	mutex_unlock(&nvmet_rdma_queue_mutex);

drivers/scsi/Kconfig

@@ -130,7 +130,8 @@ config CHR_DEV_OSST
 
 config BLK_DEV_SR
 	tristate "SCSI CDROM support"
-	depends on SCSI
+	depends on SCSI && BLK_DEV
+	select CDROM
 	---help---
 	  If you want to use a CD or DVD drive attached to your computer
 	  by SCSI, FireWire, USB or ATAPI, say Y and read the SCSI-HOWTO

drivers/scsi/lpfc/lpfc_attr.c

@@ -3246,6 +3246,11 @@ lpfc_update_rport_devloss_tmo(struct lpfc_vport *vport)
 			continue;
 		if (ndlp->rport)
 			ndlp->rport->dev_loss_tmo = vport->cfg_devloss_tmo;
+#if (IS_ENABLED(CONFIG_NVME_FC))
+		if (ndlp->nrport)
+			nvme_fc_set_remoteport_devloss(ndlp->nrport->remoteport,
+						       vport->cfg_devloss_tmo);
+#endif
 	}
 	spin_unlock_irq(shost->host_lock);
 }

drivers/scsi/scsi_lib.c

@@ -252,9 +252,9 @@ int scsi_execute(struct scsi_device *sdev, const unsigned char *cmd,
 	struct scsi_request *rq;
 	int ret = DRIVER_ERROR << 24;
 
-	req = blk_get_request(sdev->request_queue,
+	req = blk_get_request_flags(sdev->request_queue,
 			data_direction == DMA_TO_DEVICE ?
-			REQ_OP_SCSI_OUT : REQ_OP_SCSI_IN, __GFP_RECLAIM);
+			REQ_OP_SCSI_OUT : REQ_OP_SCSI_IN, BLK_MQ_REQ_PREEMPT);
 	if (IS_ERR(req))
 		return ret;
 	rq = scsi_req(req);
@@ -268,7 +268,7 @@ int scsi_execute(struct scsi_device *sdev, const unsigned char *cmd,
 	rq->retries = retries;
 	req->timeout = timeout;
 	req->cmd_flags |= flags;
-	req->rq_flags |= rq_flags | RQF_QUIET | RQF_PREEMPT;
+	req->rq_flags |= rq_flags | RQF_QUIET;
 
 	/*
 	 * head injection *required* here otherwise quiesce won't work
@@ -1301,7 +1301,7 @@ scsi_prep_state_check(struct scsi_device *sdev, struct request *req)
 			/*
 			 * If the devices is blocked we defer normal commands.
 			 */
-			if (!(req->rq_flags & RQF_PREEMPT))
+			if (req && !(req->rq_flags & RQF_PREEMPT))
 				ret = BLKPREP_DEFER;
 			break;
 		default:
@@ -1310,7 +1310,7 @@ scsi_prep_state_check(struct scsi_device *sdev, struct request *req)
 			 * special commands.  In particular any user initiated
 			 * command is not allowed.
 			 */
-			if (!(req->rq_flags & RQF_PREEMPT))
+			if (req && !(req->rq_flags & RQF_PREEMPT))
 				ret = BLKPREP_KILL;
 			break;
 		}
@@ -1940,6 +1940,33 @@ static void scsi_mq_done(struct scsi_cmnd *cmd)
 	blk_mq_complete_request(cmd->request);
 }
 
+static void scsi_mq_put_budget(struct blk_mq_hw_ctx *hctx)
+{
+	struct request_queue *q = hctx->queue;
+	struct scsi_device *sdev = q->queuedata;
+
+	atomic_dec(&sdev->device_busy);
+	put_device(&sdev->sdev_gendev);
+}
+
+static bool scsi_mq_get_budget(struct blk_mq_hw_ctx *hctx)
+{
+	struct request_queue *q = hctx->queue;
+	struct scsi_device *sdev = q->queuedata;
+
+	if (!get_device(&sdev->sdev_gendev))
+		goto out;
+	if (!scsi_dev_queue_ready(q, sdev))
+		goto out_put_device;
+
+	return true;
+
+out_put_device:
+	put_device(&sdev->sdev_gendev);
+out:
+	return false;
+}
+
 static blk_status_t scsi_queue_rq(struct blk_mq_hw_ctx *hctx,
 			 const struct blk_mq_queue_data *bd)
 {
@@ -1953,16 +1980,11 @@ static blk_status_t scsi_queue_rq(struct blk_mq_hw_ctx *hctx,
 
 	ret = prep_to_mq(scsi_prep_state_check(sdev, req));
 	if (ret != BLK_STS_OK)
-		goto out;
+		goto out_put_budget;
 
 	ret = BLK_STS_RESOURCE;
-	if (!get_device(&sdev->sdev_gendev))
-		goto out;
-
-	if (!scsi_dev_queue_ready(q, sdev))
-		goto out_put_device;
 	if (!scsi_target_queue_ready(shost, sdev))
-		goto out_dec_device_busy;
+		goto out_put_budget;
 	if (!scsi_host_queue_ready(q, shost, sdev))
 		goto out_dec_target_busy;
 
@@ -1993,15 +2015,12 @@ static blk_status_t scsi_queue_rq(struct blk_mq_hw_ctx *hctx,
 	return BLK_STS_OK;
 
 out_dec_host_busy:
-	atomic_dec(&shost->host_busy);
+       atomic_dec(&shost->host_busy);
 out_dec_target_busy:
 	if (scsi_target(sdev)->can_queue > 0)
 		atomic_dec(&scsi_target(sdev)->target_busy);
-out_dec_device_busy:
-	atomic_dec(&sdev->device_busy);
-out_put_device:
-	put_device(&sdev->sdev_gendev);
-out:
+out_put_budget:
+	scsi_mq_put_budget(hctx);
 	switch (ret) {
 	case BLK_STS_OK:
 		break;
@@ -2205,6 +2224,8 @@ struct request_queue *scsi_old_alloc_queue(struct scsi_device *sdev)
 }
 
 static const struct blk_mq_ops scsi_mq_ops = {
+	.get_budget	= scsi_mq_get_budget,
+	.put_budget	= scsi_mq_put_budget,
 	.queue_rq	= scsi_queue_rq,
 	.complete	= scsi_softirq_done,
 	.timeout	= scsi_timeout,
@@ -2919,21 +2940,37 @@ static void scsi_wait_for_queuecommand(struct scsi_device *sdev)
 int
 scsi_device_quiesce(struct scsi_device *sdev)
 {
+	struct request_queue *q = sdev->request_queue;
 	int err;
 
+	/*
+	 * It is allowed to call scsi_device_quiesce() multiple times from
+	 * the same context but concurrent scsi_device_quiesce() calls are
+	 * not allowed.
+	 */
+	WARN_ON_ONCE(sdev->quiesced_by && sdev->quiesced_by != current);
+
+	blk_set_preempt_only(q);
+
+	blk_mq_freeze_queue(q);
+	/*
+	 * Ensure that the effect of blk_set_preempt_only() will be visible
+	 * for percpu_ref_tryget() callers that occur after the queue
+	 * unfreeze even if the queue was already frozen before this function
+	 * was called. See also https://lwn.net/Articles/573497/.
+	 */
+	synchronize_rcu();
+	blk_mq_unfreeze_queue(q);
+
 	mutex_lock(&sdev->state_mutex);
 	err = scsi_device_set_state(sdev, SDEV_QUIESCE);
+	if (err == 0)
+		sdev->quiesced_by = current;
+	else
+		blk_clear_preempt_only(q);
 	mutex_unlock(&sdev->state_mutex);
 
-	if (err)
-		return err;
-
-	scsi_run_queue(sdev->request_queue);
-	while (atomic_read(&sdev->device_busy)) {
-		msleep_interruptible(200);
-		scsi_run_queue(sdev->request_queue);
-	}
-	return 0;
+	return err;
 }
 EXPORT_SYMBOL(scsi_device_quiesce);
 
@@ -2953,9 +2990,11 @@ void scsi_device_resume(struct scsi_device *sdev)
 	 * device deleted during suspend)
 	 */
 	mutex_lock(&sdev->state_mutex);
-	if (sdev->sdev_state == SDEV_QUIESCE &&
-	    scsi_device_set_state(sdev, SDEV_RUNNING) == 0)
-		scsi_run_queue(sdev->request_queue);
+	WARN_ON_ONCE(!sdev->quiesced_by);
+	sdev->quiesced_by = NULL;
+	blk_clear_preempt_only(sdev->request_queue);
+	if (sdev->sdev_state == SDEV_QUIESCE)
+		scsi_device_set_state(sdev, SDEV_RUNNING);
 	mutex_unlock(&sdev->state_mutex);
 }
 EXPORT_SYMBOL(scsi_device_resume);

drivers/scsi/sg.c

@@ -217,7 +217,7 @@ static int sg_allow_access(struct file *filp, unsigned char *cmd)
 	if (sfp->parentdp->device->type == TYPE_SCANNER)
 		return 0;
 
-	return blk_verify_command(cmd, filp->f_mode & FMODE_WRITE);
+	return blk_verify_command(cmd, filp->f_mode);
 }
 
 static int

fs/block_dev.c

@@ -54,18 +54,6 @@ struct block_device *I_BDEV(struct inode *inode)
 }
 EXPORT_SYMBOL(I_BDEV);
 
-void __vfs_msg(struct super_block *sb, const char *prefix, const char *fmt, ...)
-{
-	struct va_format vaf;
-	va_list args;
-
-	va_start(args, fmt);
-	vaf.fmt = fmt;
-	vaf.va = &args;
-	printk_ratelimited("%sVFS (%s): %pV\n", prefix, sb->s_id, &vaf);
-	va_end(args);
-}
-
 static void bdev_write_inode(struct block_device *bdev)
 {
 	struct inode *inode = bdev->bd_inode;
@@ -249,7 +237,7 @@ __blkdev_direct_IO_simple(struct kiocb *iocb, struct iov_iter *iter,
 		if (!READ_ONCE(bio.bi_private))
 			break;
 		if (!(iocb->ki_flags & IOCB_HIPRI) ||
-		    !blk_mq_poll(bdev_get_queue(bdev), qc))
+		    !blk_poll(bdev_get_queue(bdev), qc))
 			io_schedule();
 	}
 	__set_current_state(TASK_RUNNING);
@@ -414,7 +402,7 @@ __blkdev_direct_IO(struct kiocb *iocb, struct iov_iter *iter, int nr_pages)
 			break;
 
 		if (!(iocb->ki_flags & IOCB_HIPRI) ||
-		    !blk_mq_poll(bdev_get_queue(bdev), qc))
+		    !blk_poll(bdev_get_queue(bdev), qc))
 			io_schedule();
 	}
 	__set_current_state(TASK_RUNNING);
@@ -674,7 +662,7 @@ int bdev_read_page(struct block_device *bdev, sector_t sector,
 	if (!ops->rw_page || bdev_get_integrity(bdev))
 		return result;
 
-	result = blk_queue_enter(bdev->bd_queue, false);
+	result = blk_queue_enter(bdev->bd_queue, 0);
 	if (result)
 		return result;
 	result = ops->rw_page(bdev, sector + get_start_sect(bdev), page, false);
@@ -710,7 +698,7 @@ int bdev_write_page(struct block_device *bdev, sector_t sector,
 
 	if (!ops->rw_page || bdev_get_integrity(bdev))
 		return -EOPNOTSUPP;
-	result = blk_queue_enter(bdev->bd_queue, false);
+	result = blk_queue_enter(bdev->bd_queue, 0);
 	if (result)
 		return result;
 

fs/buffer.c

@@ -252,27 +252,6 @@ out:
 	return ret;
 }
 
-/*
- * Kick the writeback threads then try to free up some ZONE_NORMAL memory.
- */
-static void free_more_memory(void)
-{
-	struct zoneref *z;
-	int nid;
-
-	wakeup_flusher_threads(1024, WB_REASON_FREE_MORE_MEM);
-	yield();
-
-	for_each_online_node(nid) {
-
-		z = first_zones_zonelist(node_zonelist(nid, GFP_NOFS),
-						gfp_zone(GFP_NOFS), NULL);
-		if (z->zone)
-			try_to_free_pages(node_zonelist(nid, GFP_NOFS), 0,
-						GFP_NOFS, NULL);
-	}
-}
-
 /*
  * I/O completion handler for block_read_full_page() - pages
  * which come unlocked at the end of I/O.
@@ -861,16 +840,19 @@ int remove_inode_buffers(struct inode *inode)
  * which may not fail from ordinary buffer allocations.
  */
 struct buffer_head *alloc_page_buffers(struct page *page, unsigned long size,
-		int retry)
+		bool retry)
 {
 	struct buffer_head *bh, *head;
+	gfp_t gfp = GFP_NOFS;
 	long offset;
 
-try_again:
+	if (retry)
+		gfp |= __GFP_NOFAIL;
+
 	head = NULL;
 	offset = PAGE_SIZE;
 	while ((offset -= size) >= 0) {
-		bh = alloc_buffer_head(GFP_NOFS);
+		bh = alloc_buffer_head(gfp);
 		if (!bh)
 			goto no_grow;
 
@@ -896,23 +878,7 @@ no_grow:
 		} while (head);
 	}
 
-	/*
-	 * Return failure for non-async IO requests.  Async IO requests
-	 * are not allowed to fail, so we have to wait until buffer heads
-	 * become available.  But we don't want tasks sleeping with 
-	 * partially complete buffers, so all were released above.
-	 */
-	if (!retry)
-		return NULL;
-
-	/* We're _really_ low on memory. Now we just
-	 * wait for old buffer heads to become free due to
-	 * finishing IO.  Since this is an async request and
-	 * the reserve list is empty, we're sure there are 
-	 * async buffer heads in use.
-	 */
-	free_more_memory();
-	goto try_again;
+	return NULL;
 }
 EXPORT_SYMBOL_GPL(alloc_page_buffers);
 
@@ -1001,8 +967,6 @@ grow_dev_page(struct block_device *bdev, sector_t block,
 	gfp_mask |= __GFP_NOFAIL;
 
 	page = find_or_create_page(inode->i_mapping, index, gfp_mask);
-	if (!page)
-		return ret;
 
 	BUG_ON(!PageLocked(page));
 
@@ -1021,9 +985,7 @@ grow_dev_page(struct block_device *bdev, sector_t block,
 	/*
 	 * Allocate some buffers for this page
 	 */
-	bh = alloc_page_buffers(page, size, 0);
-	if (!bh)
-		goto failed;
+	bh = alloc_page_buffers(page, size, true);
 
 	/*
 	 * Link the page to the buffers and initialise them.  Take the
@@ -1103,8 +1065,6 @@ __getblk_slow(struct block_device *bdev, sector_t block,
 		ret = grow_buffers(bdev, block, size, gfp);
 		if (ret < 0)
 			return NULL;
-		if (ret == 0)
-			free_more_memory();
 	}
 }
 
@@ -1575,7 +1535,7 @@ void create_empty_buffers(struct page *page,
 {
 	struct buffer_head *bh, *head, *tail;
 
-	head = alloc_page_buffers(page, blocksize, 1);
+	head = alloc_page_buffers(page, blocksize, true);
 	bh = head;
 	do {
 		bh->b_state |= b_state;
@@ -2639,7 +2599,7 @@ int nobh_write_begin(struct address_space *mapping,
 	 * Be careful: the buffer linked list is a NULL terminated one, rather
 	 * than the circular one we're used to.
 	 */
-	head = alloc_page_buffers(page, blocksize, 0);
+	head = alloc_page_buffers(page, blocksize, false);
 	if (!head) {
 		ret = -ENOMEM;
 		goto out_release;
@@ -3056,8 +3016,16 @@ void guard_bio_eod(int op, struct bio *bio)
 	sector_t maxsector;
 	struct bio_vec *bvec = &bio->bi_io_vec[bio->bi_vcnt - 1];
 	unsigned truncated_bytes;
+	struct hd_struct *part;
+
+	rcu_read_lock();
+	part = __disk_get_part(bio->bi_disk, bio->bi_partno);
+	if (part)
+		maxsector = part_nr_sects_read(part);
+	else
+		maxsector = get_capacity(bio->bi_disk);
+	rcu_read_unlock();
 
-	maxsector = get_capacity(bio->bi_disk);
 	if (!maxsector)
 		return;