block: let blkcg_gq grab request queue's refcnt

In the whole lifetime of blkcg_gq instance, ->q will be referred, such
as, ->pd_free_fn() is called in blkg_free, and throtl_pd_free() still
may touch the request queue via &tg->service_queue.pending_timer which
is handled by throtl_pending_timer_fn(), so it is reasonable to grab
request queue's refcnt by blkcg_gq instance.

Previously blkcg_exit_queue() is called from blk_release_queue, and it
is hard to avoid the use-after-free. But recently commit 1059699f87eb ("block:
move blkcg initialization/destroy into disk allocation/release handler")
is merged to for-5.18/block, it becomes simple to fix the issue by simply
grabbing request queue's refcnt.

Reported-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Ming Lei <ming.lei@redhat.com>
Link: https://lore.kernel.org/r/20220318130144.1066064-3-ming.lei@redhat.com
Signed-off-by: Jens Axboe <axboe@kernel.dk>

block: avoid use-after-free on throttle data

In throtl_pending_timer_fn(), request queue is retrieved from throttle
data. And tg's pending timer is deleted synchronously when releasing the
associated blkg, at that time, throttle data may have been freed since
commit 1059699f87eb ("block: move blkcg initialization/destroy into disk
allocation/release handler") moves freeing q->td to disk_release() from
blk_release_queue(). So use-after-free on q->td in throtl_pending_timer_fn
can be triggered.

Fixes the issue by:

- do nothing in case that disk is released, when there isn't any bio to
  dispatch

- retrieve request queue from blkg instead of throttle data for
non top-level pending timer.

Signed-off-by: Ming Lei <ming.lei@redhat.com>
Link: https://lore.kernel.org/r/20220318130144.1066064-2-ming.lei@redhat.com
Signed-off-by: Jens Axboe <axboe@kernel.dk>

block: limit request dispatch loop duration

When IO requests are made continuously and the target block device
handles requests faster than request arrival, the request dispatch loop
keeps on repeating to dispatch the arriving requests very long time,
more than a minute. Since the loop runs as a workqueue worker task, the
very long loop duration triggers workqueue watchdog timeout and BUG [1].

To avoid the very long loop duration, break the loop periodically. When
opportunity to dispatch requests still exists, check need_resched(). If
need_resched() returns true, the dispatch loop already consumed its time
slice, then reschedule the dispatch work and break the loop. With heavy
IO load, need_resched() does not return true for 20~30 seconds. To cover
such case, check time spent in the dispatch loop with jiffies. If more
than 1 second is spent, reschedule the dispatch work and break the loop.

[1]

[  609.691437] BUG: workqueue lockup - pool cpus=10 node=1 flags=0x0 nice=-20 stuck for 35s!
[  609.701820] Showing busy workqueues and worker pools:
[  609.707915] workqueue events: flags=0x0
[  609.712615]   pwq 0: cpus=0 node=0 flags=0x0 nice=0 active=1/256 refcnt=2
[  609.712626]     pending: drm_fb_helper_damage_work [drm_kms_helper]
[  609.712687] workqueue events_freezable: flags=0x4
[  609.732943]   pwq 0: cpus=0 node=0 flags=0x0 nice=0 active=1/256 refcnt=2
[  609.732952]     pending: pci_pme_list_scan
[  609.732968] workqueue events_power_efficient: flags=0x80
[  609.751947]   pwq 0: cpus=0 node=0 flags=0x0 nice=0 active=1/256 refcnt=2
[  609.751955]     pending: neigh_managed_work
[  609.752018] workqueue kblockd: flags=0x18
[  609.769480]   pwq 21: cpus=10 node=1 flags=0x0 nice=-20 active=3/256 refcnt=4
[  609.769488]     in-flight: 1020:blk_mq_run_work_fn
[  609.769498]     pending: blk_mq_timeout_work, blk_mq_run_work_fn
[  609.769744] pool 21: cpus=10 node=1 flags=0x0 nice=-20 hung=35s workers=2 idle: 67
[  639.899730] BUG: workqueue lockup - pool cpus=10 node=1 flags=0x0 nice=-20 stuck for 66s!
[  639.909513] Showing busy workqueues and worker pools:
[  639.915404] workqueue events: flags=0x0
[  639.920197]   pwq 0: cpus=0 node=0 flags=0x0 nice=0 active=1/256 refcnt=2
[  639.920215]     pending: drm_fb_helper_damage_work [drm_kms_helper]
[  639.920365] workqueue kblockd: flags=0x18
[  639.939932]   pwq 21: cpus=10 node=1 flags=0x0 nice=-20 active=3/256 refcnt=4
[  639.939942]     in-flight: 1020:blk_mq_run_work_fn
[  639.939955]     pending: blk_mq_timeout_work, blk_mq_run_work_fn
[  639.940212] pool 21: cpus=10 node=1 flags=0x0 nice=-20 hung=66s workers=2 idle: 67

Fixes: 6e6fcbc27e778 ("blk-mq: support batching dispatch in case of io")
Signed-off-by: Shin'ichiro Kawasaki <shinichiro.kawasaki@wdc.com>
Cc: stable@vger.kernel.org # v5.10+
Link: https://lore.kernel.org/linux-block/20220310091649.zypaem5lkyfadymg@shindev/
Link: https://lore.kernel.org/r/20220318022641.133484-1-shinichiro.kawasaki@wdc.com
Signed-off-by: Jens Axboe <axboe@kernel.dk>

block/bfq-iosched: Fix spelling mistake "tenative" -> "tentative"

There is a spelling mistake in a bfq_log_bfqq message. Fix it.

Signed-off-by: Colin Ian King <colin.i.king@gmail.com>
Reviewed-by: Chaitanya Kulkarni <kch@nvidia.com>
Link: https://lore.kernel.org/r/20220315221539.2959167-1-colin.i.king@gmail.com
Signed-off-by: Jens Axboe <axboe@kernel.dk>

sr: simplify the local variable initialization in sr_block_open()

Commit 01d0c698536f ("sr: implement ->free_disk to simplify refcounting")
refactored sr_block_open(), initialized one variable with a duplicate
assignment (probably an unintended copy & paste duplication) and turned one
error case into an early return, which makes the initialization of the
return variable needless.

So, simplify the local variable initialization in sr_block_open() to make
the code a bit more clear.

No functional change. No change in resulting object code.

Signed-off-by: Lukas Bulwahn <lukas.bulwahn@gmail.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Link: https://lore.kernel.org/r/20220314150321.17720-1-lukas.bulwahn@gmail.com
Signed-off-by: Jens Axboe <axboe@kernel.dk>

block: don't merge across cgroup boundaries if blkcg is enabled

blk-iocost and iolatency are cgroup aware rq-qos policies but they didn't
disable merges across different cgroups. This obviously can lead to
accounting and control errors but more importantly to priority inversions -
e.g. an IO which belongs to a higher priority cgroup or IO class may end up
getting throttled incorrectly because it gets merged to an IO issued from a
low priority cgroup.

Fix it by adding blk_cgroup_mergeable() which is called from merge paths and
rejects cross-cgroup and cross-issue_as_root merges.

Signed-off-by: Tejun Heo <tj@kernel.org>
Fixes: d70675121546 ("block: introduce blk-iolatency io controller")
Cc: stable@vger.kernel.org # v4.19+
Cc: Josef Bacik <jbacik@fb.com>
Link: https://lore.kernel.org/r/Yi/eE/6zFNyWJ+qd@slm.duckdns.org
Signed-off-by: Jens Axboe <axboe@kernel.dk>

block: fix rq-qos breakage from skipping rq_qos_done_bio()

a647a524a467 ("block: don't call rq_qos_ops->done_bio if the bio isn't
tracked") made bio_endio() skip rq_qos_done_bio() if BIO_TRACKED is not set.
While this fixed a potential oops, it also broke blk-iocost by skipping the
done_bio callback for merged bios.

Before, whether a bio goes through rq_qos_throttle() or rq_qos_merge(),
rq_qos_done_bio() would be called on the bio on completion with BIO_TRACKED
distinguishing the former from the latter. rq_qos_done_bio() is not called
for bios which wenth through rq_qos_merge(). This royally confuses
blk-iocost as the merged bios never finish and are considered perpetually
in-flight.

One reliably reproducible failure mode is an intermediate cgroup geting
stuck active preventing its children from being activated due to the
leaf-only rule, leading to loss of control. The following is from
resctl-bench protection scenario which emulates isolating a web server like
workload from a memory bomb run on an iocost configuration which should
yield a reasonable level of protection.

  # cat /sys/block/nvme2n1/device/model
  Samsung SSD 970 PRO 512GB
  # cat /sys/fs/cgroup/io.cost.model
  259:0 ctrl=user model=linear rbps=834913556 rseqiops=93622 rrandiops=102913 wbps=618985353 wseqiops=72325 wrandiops=71025
  # cat /sys/fs/cgroup/io.cost.qos
  259:0 enable=1 ctrl=user rpct=95.00 rlat=18776 wpct=95.00 wlat=8897 min=60.00 max=100.00
  # resctl-bench -m 29.6G -r out.json run protection::scenario=mem-hog,loops=1
  ...
  Memory Hog Summary
  ==================

  IO Latency: R p50=242u:336u/2.5m p90=794u:1.4m/7.5m p99=2.7m:8.0m/62.5m max=8.0m:36.4m/350m
              W p50=221u:323u/1.5m p90=709u:1.2m/5.5m p99=1.5m:2.5m/9.5m max=6.9m:35.9m/350m

  Isolation and Request Latency Impact Distributions:

                min   p01   p05   p10   p25   p50   p75   p90   p95   p99   max  mean stdev
  isol%       15.90 15.90 15.90 40.05 57.24 59.07 60.01 74.63 74.63 90.35 90.35 58.12 15.82
  lat-imp%        0     0     0     0     0  4.55 14.68 15.54 233.5 548.1 548.1 53.88 143.6

  Result: isol=58.12:15.82% lat_imp=53.88%:143.6 work_csv=100.0% missing=3.96%

The isolation result of 58.12% is close to what this device would show
without any IO control.

Fix it by introducing a new flag BIO_QOS_MERGED to mark merged bios and
calling rq_qos_done_bio() on them too. For consistency and clarity, rename
BIO_TRACKED to BIO_QOS_THROTTLED. The flag checks are moved into
rq_qos_done_bio() so that it's next to the code paths that set the flags.

With the patch applied, the above same benchmark shows:

  # resctl-bench -m 29.6G -r out.json run protection::scenario=mem-hog,loops=1
  ...
  Memory Hog Summary
  ==================

  IO Latency: R p50=123u:84.4u/985u p90=322u:256u/2.5m p99=1.6m:1.4m/9.5m max=11.1m:36.0m/350m
              W p50=429u:274u/995u p90=1.7m:1.3m/4.5m p99=3.4m:2.7m/11.5m max=7.9m:5.9m/26.5m

  Isolation and Request Latency Impact Distributions:

                min   p01   p05   p10   p25   p50   p75   p90   p95   p99   max  mean stdev
  isol%       84.91 84.91 89.51 90.73 92.31 94.49 96.36 98.04 98.71 100.0 100.0 94.42  2.81
  lat-imp%        0     0     0     0     0  2.81  5.73 11.11 13.92 17.53 22.61  4.10  4.68

  Result: isol=94.42:2.81% lat_imp=4.10%:4.68 work_csv=58.34% missing=0%

Signed-off-by: Tejun Heo <tj@kernel.org>
Fixes: a647a524a467 ("block: don't call rq_qos_ops->done_bio if the bio isn't tracked")
Cc: stable@vger.kernel.org # v5.15+
Cc: Ming Lei <ming.lei@redhat.com>
Cc: Yu Kuai <yukuai3@huawei.com>
Reviewed-by: Ming Lei <ming.lei@redhat.com>
Link: https://lore.kernel.org/r/Yi7rdrzQEHjJLGKB@slm.duckdns.org
Signed-off-by: Jens Axboe <axboe@kernel.dk>

block: flush plug based on hardware and software queue order

We used to sort the plug list if we had multiple queues before dispatching
requests to the IO scheduler. This usually isn't needed, but for certain
workloads that interleave requests to disks, it's a less efficient to
process the plug list one-by-one if everything is interleaved.

Don't sort the list, but skip through it and flush out entries that have
the same target at the same time.

Fixes: df87eb0fce8f ("block: get rid of plug list sorting")
Reported-and-tested-by: Song Liu <song@kernel.org>
Reviewed-by: Song Liu <songliubraving@fb.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>

block: ensure plug merging checks the correct queue at least once

Song reports that a RAID rebuild workload runs much slower recently,
and it is seeing a lot less merging than it did previously. The reason
is that a previous commit reduced the amount of work we do for plug
merging. RAID rebuild interleaves requests between disks, so a last-entry
check in plug merging always misses a merge opportunity since we always
find a different disk than what we are looking for.

Modify the logic such that it's still a one-hit cache, but ensure that
we check enough to find the right target before giving up.

Fixes: d38a9c04c0d5 ("block: only check previous entry for plug merge attempt")
Reported-and-tested-by: Song Liu <song@kernel.org>
Reviewed-by: Song Liu <songliubraving@fb.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>

block: move rq_qos_exit() into disk_release()

Keep all teardown of file system I/O related functionality in one place.
There can't be file system I/O in disk_release(), so it is safe to move
rq_qos_exit() there.

Signed-off-by: Ming Lei <ming.lei@redhat.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>
Link: https://lore.kernel.org/r/20220308055200.735835-15-hch@lst.de
Signed-off-by: Jens Axboe <axboe@kernel.dk>

block: do more work in elevator_exit

Move the calls to ioc_clear_queue and blk_mq_sched_free_rqs into
elevator_exit.  Except for one call where we know we can't have io_cq
structures yet these always go together, and that extra call in an
error path is harmless.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Link: https://lore.kernel.org/r/20220308055200.735835-14-hch@lst.de
Signed-off-by: Jens Axboe <axboe@kernel.dk>

block: move blk_exit_queue into disk_release

There can't be file system I/O in disk_release(), so move the call to
blk_exit_queue() there, preparing to have the teardown of file system I/O
only functionality in one place, when the gendisk that is needed for it
is torn down.

We still need to freeze queue here since the request is freed after the
bio is completed and passthrough request rely on scheduler tags as well.

The disk can be released before or after queue is cleaned up, and we have
to free the scheduler request pool before blk_cleanup_queue returns,
while the static request pool has to be freed before exiting the
I/O scheduler.

Signed-off-by: Ming Lei <ming.lei@redhat.com>
[hch: rebased, updated the commit log]
Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Bart Van Assche <bvanassche@acm.org>
Link: https://lore.kernel.org/r/20220308055200.735835-13-hch@lst.de
Signed-off-by: Jens Axboe <axboe@kernel.dk>

block: move q_usage_counter release into blk_queue_release

After blk_cleanup_queue() returns, disk may not be released yet, so
probably bio may still be submitted and ->q_usage_counter may be
touched, so far this way seems safe, but not good from API's viewpoint.

Move the release q_usage_counter into blk_queue_release().

Signed-off-by: Ming Lei <ming.lei@redhat.com>
Reviewed-by: Bart Van Assche <bvanassche@acm.org>
Signed-off-by: Christoph Hellwig <hch@lst.de>
Link: https://lore.kernel.org/r/20220308055200.735835-12-hch@lst.de
Signed-off-by: Jens Axboe <axboe@kernel.dk>

block: don't remove hctx debugfs dir from blk_mq_exit_queue

The queue's top debugfs dir is removed from blk_release_queue(), so all
hctx's debugfs dirs are removed from there. Given blk_mq_exit_queue()
is only called from blk_cleanup_queue(), it isn't necessary to remove
hctx debugfs from blk_mq_exit_queue().

So remove it from blk_mq_exit_queue().

Signed-off-by: Ming Lei <ming.lei@redhat.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>
Link: https://lore.kernel.org/r/20220308055200.735835-11-hch@lst.de
Signed-off-by: Jens Axboe <axboe@kernel.dk>

block: move blkcg initialization/destroy into disk allocation/release handler

blkcg works on FS bio level, so it is reasonable to make both blkcg and
gendisk sharing same lifetime. Meantime there won't be any FS IO when
releasing disk, so safe to move blkcg initialization/destroy into disk
allocation/release handler

Long term, we can move blkcg into gendisk completely.

Signed-off-by: Ming Lei <ming.lei@redhat.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Bart Van Assche <bvanassche@acm.org>
Reviewed-by: Chaitanya Kulkarni <kch@nvidia.com>
Link: https://lore.kernel.org/r/20220308055200.735835-10-hch@lst.de
[axboe: fixup missing blk-cgroup.h include]
Signed-off-by: Jens Axboe <axboe@kernel.dk>

sr: implement ->free_disk to simplify refcounting

Simplify the refcounting and remove the need to clear disk->private_data
by implementing the ->free_disk method.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Bart Van Assche <bvanassche@acm.org>
Reviewed-by: Ming Lei <ming.lei@redhat.com>
Reviewed-by: Martin K. Petersen <martin.petersen@oracle.com>
Link: https://lore.kernel.org/r/20220308055200.735835-9-hch@lst.de
Signed-off-by: Jens Axboe <axboe@kernel.dk>

sd: implement ->free_disk to simplify refcounting

Implement the ->free_disk method to to put struct scsi_disk when the last
gendisk reference count goes away.  This removes the need to clear
->private_data and thus freeze the queue on unbind.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Ming Lei <ming.lei@redhat.com>
Reviewed-by: Martin K. Petersen <martin.petersen@oracle.com>
Link: https://lore.kernel.org/r/20220308055200.735835-8-hch@lst.de
Signed-off-by: Jens Axboe <axboe@kernel.dk>

sd: delay calling free_opal_dev

Call free_opal_dev from scsi_disk_release as the opal_dev field is accessed
from the ioctl handler, which isn't synchronized vs sd_release and thus
can be accessed during or after sd_release was called.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Bart Van Assche <bvanassche@acm.org>
Reviewed-by: Chaitanya Kulkarni <kch@nvidia.com>
Reviewed-by: Ming Lei <ming.lei@redhat.com>
Reviewed-by: Martin K. Petersen <martin.petersen@oracle.com>
Link: https://lore.kernel.org/r/20220308055200.735835-7-hch@lst.de
Signed-off-by: Jens Axboe <axboe@kernel.dk>

sd: call sd_zbc_release_disk before releasing the scsi_device reference

sd_zbc_release_disk accesses disk->device, so ensure that actually still has
a valid reference.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Bart Van Assche <bvanassche@acm.org>
Reviewed-by: Ming Lei <ming.lei@redhat.com>
Reviewed-by: Martin K. Petersen <martin.petersen@oracle.com>
Link: https://lore.kernel.org/r/20220308055200.735835-6-hch@lst.de
Signed-off-by: Jens Axboe <axboe@kernel.dk>

sd: rename the scsi_disk.dev field

dev is very hard to grep for.  Give the field a more descriptive name and
documents its purpose.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Bart Van Assche <bvanassche@acm.org>
Reviewed-by: Chaitanya Kulkarni <kch@nvidia.com>
Reviewed-by: Ming Lei <ming.lei@redhat.com>
Reviewed-by: Martin K. Petersen <martin.petersen@oracle.com>
Link: https://lore.kernel.org/r/20220308055200.735835-5-hch@lst.de
Signed-off-by: Jens Axboe <axboe@kernel.dk>

scsi: don't use disk->private_data to find the scsi_driver

Requiring every ULP to have the scsi_drive as first member of the
private data is rather fragile and not necessary anyway.  Just use
the driver hanging off the SCSI device instead.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Bart Van Assche <bvanassche@acm.org>
Reviewed-by: Chaitanya Kulkarni <kch@nvidia.com>
Reviewed-by: Ming Lei <ming.lei@redhat.com>
Reviewed-by: Martin K. Petersen <martin.petersen@oracle.com>
Link: https://lore.kernel.org/r/20220308055200.735835-4-hch@lst.de
Signed-off-by: Jens Axboe <axboe@kernel.dk>

blk-mq: handle already freed tags gracefully in blk_mq_free_rqs

To simplify further changes allow for double calling blk_mq_free_rqs on
a queue.

Signed-off-by: Ming Lei <ming.lei@redhat.com>
[hch: split out from a larger patch]
Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Bart Van Assche <bvanassche@acm.org>
Reviewed-by: Martin K. Petersen <martin.petersen@oracle.com>
Link: https://lore.kernel.org/r/20220308055200.735835-3-hch@lst.de
Signed-off-by: Jens Axboe <axboe@kernel.dk>

blk-mq: do not include passthrough requests in I/O accounting

I/O accounting buckets I/O into the read/write/discard categories into
which passthrough I/O does not fit at all.  It also accounts to the
block_device, which may not even exist for passthrough I/O.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Bart Van Assche <bvanassche@acm.org>
Reviewed-by: Chaitanya Kulkarni <kch@nvidia.com>
Reviewed-by: Martin K. Petersen <martin.petersen@oracle.com>
Link: https://lore.kernel.org/r/20220308055200.735835-2-hch@lst.de
Signed-off-by: Jens Axboe <axboe@kernel.dk>

blk-mq: manage hctx map via xarray

First code becomes more clean by switching to xarray from plain array.

Second use-after-free on q->queue_hw_ctx can be fixed because
queue_for_each_hw_ctx() may be run when updating nr_hw_queues is
in-progress. With this patch, q->hctx_table is defined as xarray, and
this structure will share same lifetime with request queue, so
queue_for_each_hw_ctx() can use q->hctx_table to lookup hctx reliably.

Reported-by: Yu Kuai <yukuai3@huawei.com>
Signed-off-by: Ming Lei <ming.lei@redhat.com>
Reviewed-by: Hannes Reinecke <hare@suse.de>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Link: https://lore.kernel.org/r/20220308073219.91173-7-ming.lei@redhat.com
[axboe: fix blk_mq_hw_ctx forward declaration]
Signed-off-by: Jens Axboe <axboe@kernel.dk>

blk-mq: prepare for implementing hctx table via xarray

It is inevitable to cause use-after-free on q->queue_hw_ctx between
queue_for_each_hw_ctx() and blk_mq_update_nr_hw_queues(). And converting
to xarray can fix the uaf, meantime code gets cleaner.

Prepare for converting q->queue_hctx_ctx into xarray, one thing is that
xa_for_each() can only accept 'unsigned long' as index, so changes type
of hctx index of queue_for_each_hw_ctx() into 'unsigned long'.

Reviewed-by: Hannes Reinecke <hare@suse.de>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Ming Lei <ming.lei@redhat.com>
Link: https://lore.kernel.org/r/20220308073219.91173-6-ming.lei@redhat.com
Signed-off-by: Jens Axboe <axboe@kernel.dk>

block: mtip32xx: don't touch q->queue_hw_ctx

q->queue_hw_ctx is really one blk-mq internal structure for retrieving
hctx via its index, not supposed to be used by drivers. Meantime drivers
can get the tags structure easily from tagset.

Reviewed-by: Hannes Reinecke <hare@suse.de>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Ming Lei <ming.lei@redhat.com>
Link: https://lore.kernel.org/r/20220308073219.91173-5-ming.lei@redhat.com
Signed-off-by: Jens Axboe <axboe@kernel.dk>

blk-mq: reconfigure poll after queue map is changed

queue map can be changed when updating nr_hw_queues, so we need to
reconfigure queue's poll capability. Add one helper for doing this job.

Reviewed-by: Hannes Reinecke <hare@suse.de>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Ming Lei <ming.lei@redhat.com>
Link: https://lore.kernel.org/r/20220308073219.91173-4-ming.lei@redhat.com
Signed-off-by: Jens Axboe <axboe@kernel.dk>

blk-mq: simplify reallocation of hw ctxs a bit

blk_mq_alloc_and_init_hctx() has already taken reuse into account, so
no need to do it outside, then we can simplify blk_mq_realloc_hw_ctxs().

Reviewed-by: Hannes Reinecke <hare@suse.de>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Ming Lei <ming.lei@redhat.com>
Link: https://lore.kernel.org/r/20220308073219.91173-3-ming.lei@redhat.com
Signed-off-by: Jens Axboe <axboe@kernel.dk>

blk-mq: figure out correct numa node for hw queue

The current code always uses default queue map and hw queue index
for figuring out the numa node for hw queue, this way isn't correct
because blk-mq supports three queue maps, and the correct queue map
should be used for the specified hw queue.

Reviewed-by: Hannes Reinecke <hare@suse.de>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Ming Lei <ming.lei@redhat.com>
Link: https://lore.kernel.org/r/20220308073219.91173-2-ming.lei@redhat.com
Signed-off-by: Jens Axboe <axboe@kernel.dk>

Revert "Revert "block, bfq: honor already-setup queue merges""

A crash [1] happened to be triggered in conjunction with commit
2d52c58b9c9b ("block, bfq: honor already-setup queue merges"). The
latter was then reverted by commit ebc69e897e17 ("Revert "block, bfq:
honor already-setup queue merges""). Yet, the reverted commit was not
the one introducing the bug. In fact, it actually triggered a UAF
introduced by a different commit, and now fixed by commit d29bd41428cf
("block, bfq: reset last_bfqq_created on group change").

So, there is no point in keeping commit 2d52c58b9c9b ("block, bfq:
honor already-setup queue merges") out. This commit restores it.

[1] https://bugzilla.kernel.org/show_bug.cgi?id=214503

Reported-by: Holger Hoffstätte <holger@applied-asynchrony.com>
Signed-off-by: Paolo Valente <paolo.valente@linaro.org>
Link: https://lore.kernel.org/r/20211125181510.15004-1-paolo.valente@linaro.org
Signed-off-by: Jens Axboe <axboe@kernel.dk>

block: remove bio_devname

All callers are gone, so remove this wrapper.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Chaitanya Kulkarni <kch@nvidia.com>
Reviewed-by: Johannes Thumshirn <johannes.thumshirn@wdc.com>
Link: https://lore.kernel.org/r/20220304180105.409765-11-hch@lst.de
Signed-off-by: Jens Axboe <axboe@kernel.dk>

ext4: stop using bio_devname

Use the %pg format specifier to save on stack consuption and code size.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Chaitanya Kulkarni <kch@nvidia.com>
Reviewed-by: Johannes Thumshirn <johannes.thumshirn@wdc.com>
Link: https://lore.kernel.org/r/20220304180105.409765-10-hch@lst.de
Signed-off-by: Jens Axboe <axboe@kernel.dk>

raid5-ppl: stop using bio_devname

Use the %pg format specifier to save on stack consuption and code size.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Acked-by: Song Liu <song@kernel.org>
Reviewed-by: Johannes Thumshirn <johannes.thumshirn@wdc.com>
Link: https://lore.kernel.org/r/20220304180105.409765-9-hch@lst.de
Signed-off-by: Jens Axboe <axboe@kernel.dk>

raid1: stop using bio_devname

Use the %pg format specifier to save on stack consuption and code size.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Acked-by: Song Liu <song@kernel.org>
Reviewed-by: Chaitanya Kulkarni <kch@nvidia.com>
Reviewed-by: Johannes Thumshirn <johannes.thumshirn@wdc.com>
Link: https://lore.kernel.org/r/20220304180105.409765-8-hch@lst.de
Signed-off-by: Jens Axboe <axboe@kernel.dk>

md-multipath: stop using bio_devname

Use the %pg format specifier to save on stack consuption and code size.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Acked-by: Song Liu <song@kernel.org>
Reviewed-by: Chaitanya Kulkarni <kch@nvidia.com>
Reviewed-by: Johannes Thumshirn <johannes.thumshirn@wdc.com>
Link: https://lore.kernel.org/r/20220304180105.409765-7-hch@lst.de
Signed-off-by: Jens Axboe <axboe@kernel.dk>

dm-integrity: stop using bio_devname

Use the %pg format specifier to save on stack consuption and code size.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Chaitanya Kulkarni <kch@nvidia.com>
Reviewed-by: Johannes Thumshirn <johannes.thumshirn@wdc.com>
Link: https://lore.kernel.org/r/20220304180105.409765-6-hch@lst.de
Signed-off-by: Jens Axboe <axboe@kernel.dk>

dm-crypt: stop using bio_devname

Use the %pg format specifier to save on stack consuption and code size.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Chaitanya Kulkarni <kch@nvidia.com>
Reviewed-by: Johannes Thumshirn <johannes.thumshirn@wdc.com>
Link: https://lore.kernel.org/r/20220304180105.409765-5-hch@lst.de
Signed-off-by: Jens Axboe <axboe@kernel.dk>

pktcdvd: remove a pointless debug check in pkt_submit_bio

->queuedata is set up in pkt_init_queue, so it can't be NULL here.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Johannes Thumshirn <johannes.thumshirn@wdc.com>
Link: https://lore.kernel.org/r/20220304180105.409765-4-hch@lst.de
Signed-off-by: Jens Axboe <axboe@kernel.dk>

block: remove handle_bad_sector

Use the %pg format specifier instead of the stack hungry bdevname
function, and remove handle_bad_sector given that it is not pointless.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Chaitanya Kulkarni <kch@nvidia.com>
Reviewed-by: Johannes Thumshirn <johannes.thumshirn@wdc.com>
Link: https://lore.kernel.org/r/20220304180105.409765-3-hch@lst.de
Signed-off-by: Jens Axboe <axboe@kernel.dk>

block: fix and cleanup bio_check_ro

Don't use a WARN_ON when printing a potentially user triggered
condition.  Also don't print the partno when the block device name
already includes it, and use the %pg specifier to simplify printing
the block device name.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Chaitanya Kulkarni <kch@nvidia.com>
Reviewed-by: Johannes Thumshirn <johannes.thumshirn@wdc.com>
Link: https://lore.kernel.org/r/20220304180105.409765-2-hch@lst.de
Signed-off-by: Jens Axboe <axboe@kernel.dk>

bfq: fix use-after-free in bfq_dispatch_request

KASAN reports a use-after-free report when doing normal scsi-mq test

[69832.239032] ==================================================================
[69832.241810] BUG: KASAN: use-after-free in bfq_dispatch_request+0x1045/0x44b0
[69832.243267] Read of size 8 at addr ffff88802622ba88 by task kworker/3:1H/155
[69832.244656]
[69832.245007] CPU: 3 PID: 155 Comm: kworker/3:1H Not tainted 5.10.0-10295-g576c6382529e #8
[69832.246626] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS rel-1.14.0-0-g155821a1990b-prebuilt.qemu.org 04/01/2014
[69832.249069] Workqueue: kblockd blk_mq_run_work_fn
[69832.250022] Call Trace:
[69832.250541]  dump_stack+0x9b/0xce
[69832.251232]  ? bfq_dispatch_request+0x1045/0x44b0
[69832.252243]  print_address_description.constprop.6+0x3e/0x60
[69832.253381]  ? __cpuidle_text_end+0x5/0x5
[69832.254211]  ? vprintk_func+0x6b/0x120
[69832.254994]  ? bfq_dispatch_request+0x1045/0x44b0
[69832.255952]  ? bfq_dispatch_request+0x1045/0x44b0
[69832.256914]  kasan_report.cold.9+0x22/0x3a
[69832.257753]  ? bfq_dispatch_request+0x1045/0x44b0
[69832.258755]  check_memory_region+0x1c1/0x1e0
[69832.260248]  bfq_dispatch_request+0x1045/0x44b0
[69832.261181]  ? bfq_bfqq_expire+0x2440/0x2440
[69832.262032]  ? blk_mq_delay_run_hw_queues+0xf9/0x170
[69832.263022]  __blk_mq_do_dispatch_sched+0x52f/0x830
[69832.264011]  ? blk_mq_sched_request_inserted+0x100/0x100
[69832.265101]  __blk_mq_sched_dispatch_requests+0x398/0x4f0
[69832.266206]  ? blk_mq_do_dispatch_ctx+0x570/0x570
[69832.267147]  ? __switch_to+0x5f4/0xee0
[69832.267898]  blk_mq_sched_dispatch_requests+0xdf/0x140
[69832.268946]  __blk_mq_run_hw_queue+0xc0/0x270
[69832.269840]  blk_mq_run_work_fn+0x51/0x60
[69832.278170]  process_one_work+0x6d4/0xfe0
[69832.278984]  worker_thread+0x91/0xc80
[69832.279726]  ? __kthread_parkme+0xb0/0x110
[69832.280554]  ? process_one_work+0xfe0/0xfe0
[69832.281414]  kthread+0x32d/0x3f0
[69832.282082]  ? kthread_park+0x170/0x170
[69832.282849]  ret_from_fork+0x1f/0x30
[69832.283573]
[69832.283886] Allocated by task 7725:
[69832.284599]  kasan_save_stack+0x19/0x40
[69832.285385]  __kasan_kmalloc.constprop.2+0xc1/0xd0
[69832.286350]  kmem_cache_alloc_node+0x13f/0x460
[69832.287237]  bfq_get_queue+0x3d4/0x1140
[69832.287993]  bfq_get_bfqq_handle_split+0x103/0x510
[69832.289015]  bfq_init_rq+0x337/0x2d50
[69832.289749]  bfq_insert_requests+0x304/0x4e10
[69832.290634]  blk_mq_sched_insert_requests+0x13e/0x390
[69832.291629]  blk_mq_flush_plug_list+0x4b4/0x760
[69832.292538]  blk_flush_plug_list+0x2c5/0x480
[69832.293392]  io_schedule_prepare+0xb2/0xd0
[69832.294209]  io_schedule_timeout+0x13/0x80
[69832.295014]  wait_for_common_io.constprop.1+0x13c/0x270
[69832.296137]  submit_bio_wait+0x103/0x1a0
[69832.296932]  blkdev_issue_discard+0xe6/0x160
[69832.297794]  blk_ioctl_discard+0x219/0x290
[69832.298614]  blkdev_common_ioctl+0x50a/0x1750
[69832.304715]  blkdev_ioctl+0x470/0x600
[69832.305474]  block_ioctl+0xde/0x120
[69832.306232]  vfs_ioctl+0x6c/0xc0
[69832.306877]  __se_sys_ioctl+0x90/0xa0
[69832.307629]  do_syscall_64+0x2d/0x40
[69832.308362]  entry_SYSCALL_64_after_hwframe+0x44/0xa9
[69832.309382]
[69832.309701] Freed by task 155:
[69832.310328]  kasan_save_stack+0x19/0x40
[69832.311121]  kasan_set_track+0x1c/0x30
[69832.311868]  kasan_set_free_info+0x1b/0x30
[69832.312699]  __kasan_slab_free+0x111/0x160
[69832.313524]  kmem_cache_free+0x94/0x460
[69832.314367]  bfq_put_queue+0x582/0x940
[69832.315112]  __bfq_bfqd_reset_in_service+0x166/0x1d0
[69832.317275]  bfq_bfqq_expire+0xb27/0x2440
[69832.318084]  bfq_dispatch_request+0x697/0x44b0
[69832.318991]  __blk_mq_do_dispatch_sched+0x52f/0x830
[69832.319984]  __blk_mq_sched_dispatch_requests+0x398/0x4f0
[69832.321087]  blk_mq_sched_dispatch_requests+0xdf/0x140
[69832.322225]  __blk_mq_run_hw_queue+0xc0/0x270
[69832.323114]  blk_mq_run_work_fn+0x51/0x60
[69832.323942]  process_one_work+0x6d4/0xfe0
[69832.324772]  worker_thread+0x91/0xc80
[69832.325518]  kthread+0x32d/0x3f0
[69832.326205]  ret_from_fork+0x1f/0x30
[69832.326932]
[69832.338297] The buggy address belongs to the object at ffff88802622b968
[69832.338297]  which belongs to the cache bfq_queue of size 512
[69832.340766] The buggy address is located 288 bytes inside of
[69832.340766]  512-byte region [ffff88802622b968, ffff88802622bb68)
[69832.343091] The buggy address belongs to the page:
[69832.344097] page:ffffea0000988a00 refcount:1 mapcount:0 mapping:0000000000000000 index:0xffff88802622a528 pfn:0x26228
[69832.346214] head:ffffea0000988a00 order:2 compound_mapcount:0 compound_pincount:0
[69832.347719] flags: 0x1fffff80010200(slab|head)
[69832.348625] raw: 001fffff80010200 ffffea0000dbac08 ffff888017a57650 ffff8880179fe840
[69832.354972] raw: ffff88802622a528 0000000000120008 00000001ffffffff 0000000000000000
[69832.356547] page dumped because: kasan: bad access detected
[69832.357652]
[69832.357970] Memory state around the buggy address:
[69832.358926]  ffff88802622b980: fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb
[69832.360358]  ffff88802622ba00: fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb
[69832.361810] >ffff88802622ba80: fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb
[69832.363273]                       ^
[69832.363975]  ffff88802622bb00: fb fb fb fb fb fb fb fb fb fb fb fb fb fc fc fc
[69832.375960]  ffff88802622bb80: fc fc fc fc fc fc fc fc fc fc fc fc fc fc fc fc
[69832.377405] ==================================================================

In bfq_dispatch_requestfunction, it may have function call:

bfq_dispatch_request
__bfq_dispatch_request
bfq_select_queue
bfq_bfqq_expire
__bfq_bfqd_reset_in_service
bfq_put_queue
kmem_cache_free
In this function call, in_serv_queue has beed expired and meet the
conditions to free. In the function bfq_dispatch_request, the address
of in_serv_queue pointing to has been released. For getting the value
of idle_timer_disabled, it will get flags value from the address which
in_serv_queue pointing to, then the problem of use-after-free happens;

Fix the problem by check in_serv_queue == bfqd->in_service_queue, to
get the value of idle_timer_disabled if in_serve_queue is equel to
bfqd->in_service_queue. If the space of in_serv_queue pointing has
been released, this judge will aviod use-after-free problem.
And if in_serv_queue may be expired or finished, the idle_timer_disabled
will be false which would not give effects to bfq_update_dispatch_stats.

Reported-by: Hulk Robot <hulkci@huawei.com>
Signed-off-by: Zhang Wensheng <zhangwensheng5@huawei.com>
Link: https://lore.kernel.org/r/20220303070334.3020168-1-zhangwensheng5@huawei.com
Signed-off-by: Jens Axboe <axboe@kernel.dk>

blk-crypto: show crypto capabilities in sysfs

Add sysfs files that expose the inline encryption capabilities of
request queues:

/sys/block/$disk/queue/crypto/max_dun_bits
/sys/block/$disk/queue/crypto/modes/$mode
/sys/block/$disk/queue/crypto/num_keyslots

Userspace can use these new files to decide what encryption settings to
use, or whether to use inline encryption at all.  This also brings the
crypto capabilities in line with the other queue properties, which are
already discoverable via the queue directory in sysfs.

Design notes:

  - Place the new files in a new subdirectory "crypto" to group them
    together and to avoid complicating the main "queue" directory.  This
    also makes it possible to replace "crypto" with a symlink later if
    we ever make the blk_crypto_profiles into real kobjects (see below).

  - It was necessary to define a new kobject that corresponds to the
    crypto subdirectory.  For now, this kobject just contains a pointer
    to the blk_crypto_profile.  Note that multiple queues (and hence
    multiple such kobjects) may refer to the same blk_crypto_profile.

    An alternative design would more closely match the current kernel
    data structures: the blk_crypto_profile could be a kobject itself,
    located directly under the host controller device's kobject, while
    /sys/block/$disk/queue/crypto would be a symlink to it.

    I decided not to do that for now because it would require a lot more
    changes, such as no longer embedding blk_crypto_profile in other
    structures, and also because I'm not sure we can rule out moving the
    crypto capabilities into 'struct queue_limits' in the future.  (Even
    if multiple queues share the same crypto engine, maybe the supported
    data unit sizes could differ due to other queue properties.)  It
    would also still be possible to switch to that design later without
    breaking userspace, by replacing the directory with a symlink.

  - Use "max_dun_bits" instead of "max_dun_bytes".  Currently, the
    kernel internally stores this value in bytes, but that's an
    implementation detail.  It probably makes more sense to talk about
    this value in bits, and choosing bits is more future-proof.

  - "modes" is a sub-subdirectory, since there may be multiple supported
    crypto modes, sysfs is supposed to have one value per file, and it
    makes sense to group all the mode files together.

  - Each mode had to be named.  The crypto API names like "xts(aes)" are
    not appropriate because they don't specify the key size.  Therefore,
    I assigned new names.  The exact names chosen are arbitrary, but
    they happen to match the names used in log messages in fs/crypto/.

  - The "num_keyslots" file is a bit different from the others in that
    it is only useful to know for performance reasons.  However, it's
    included as it can still be useful.  For example, a user might not
    want to use inline encryption if there aren't very many keyslots.

Reviewed-by: Hannes Reinecke <hare@suse.de>
Signed-off-by: Eric Biggers <ebiggers@google.com>
Link: https://lore.kernel.org/r/20220124215938.2769-4-ebiggers@kernel.org
Signed-off-by: Jens Axboe <axboe@kernel.dk>

block: don't delete queue kobject before its children

kobjects aren't supposed to be deleted before their child kobjects are
deleted.  Apparently this is usually benign; however, a WARN will be
triggered if one of the child kobjects has a named attribute group:

    sysfs group 'modes' not found for kobject 'crypto'
    WARNING: CPU: 0 PID: 1 at fs/sysfs/group.c:278 sysfs_remove_group+0x72/0x80
    ...
    Call Trace:
      sysfs_remove_groups+0x29/0x40 fs/sysfs/group.c:312
      __kobject_del+0x20/0x80 lib/kobject.c:611
      kobject_cleanup+0xa4/0x140 lib/kobject.c:696
      kobject_release lib/kobject.c:736 [inline]
      kref_put include/linux/kref.h:65 [inline]
      kobject_put+0x53/0x70 lib/kobject.c:753
      blk_crypto_sysfs_unregister+0x10/0x20 block/blk-crypto-sysfs.c:159
      blk_unregister_queue+0xb0/0x110 block/blk-sysfs.c:962
      del_gendisk+0x117/0x250 block/genhd.c:610

Fix this by moving the kobject_del() and the corresponding
kobject_uevent() to the correct place.

Fixes: 2c2086afc2b8 ("block: Protect less code with sysfs_lock in blk_{un,}register_queue()")
Reviewed-by: Hannes Reinecke <hare@suse.de>
Reviewed-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Reviewed-by: Bart Van Assche <bvanassche@acm.org>
Signed-off-by: Eric Biggers <ebiggers@google.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Link: https://lore.kernel.org/r/20220124215938.2769-3-ebiggers@kernel.org
Signed-off-by: Jens Axboe <axboe@kernel.dk>

block: simplify calling convention of elv_unregister_queue()

Make elv_unregister_queue() a no-op if q->elevator is NULL or is not
registered.

This simplifies the existing callers, as well as the future caller in
the error path of blk_register_queue().

Also don't bother checking whether q is NULL, since it never is.

Reviewed-by: Hannes Reinecke <hare@suse.de>
Reviewed-by: Bart Van Assche <bvanassche@acm.org>
Signed-off-by: Eric Biggers <ebiggers@google.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Link: https://lore.kernel.org/r/20220124215938.2769-2-ebiggers@kernel.org
Signed-off-by: Jens Axboe <axboe@kernel.dk>

block: remove redundant semicolon

Remove redundant semicolon from block/bdev.c

Signed-off-by: Nian Yanchuan <yanchuan@nfschina.com>
Link: https://lore.kernel.org/r/20220227170124.GA14658@localhost.localdomain
Signed-off-by: Jens Axboe <axboe@kernel.dk>

block: default BLOCK_LEGACY_AUTOLOAD to y

As Luis reported, losetup currently doesn't properly create the loop
device without this if the device node already exists because old
scripts created it manually.  So default to y for now and remove the
aggressive removal schedule.

Reported-by: Luis Chamberlain <mcgrof@kernel.org>
Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Chaitanya Kulkarni <kch@nvidia.com>
Link: https://lore.kernel.org/r/20220225181440.1351591-1-hch@lst.de
Signed-off-by: Jens Axboe <axboe@kernel.dk>

block: update io_ticks when io hang

When the inflight IOs are slow and no new IOs are issued, we expect
iostat could manifest the IO hang problem. However after
commit 5b18b5a73760 ("block: delete part_round_stats and switch to less
precise counting"), io_tick and time_in_queue will not be updated until
the end of IO, and the avgqu-sz and %util columns of iostat will be zero.

Because it has using stat.nsecs accumulation to express time_in_queue
which is not suitable to change, and may %util will express the status
better when io hang occur. To fix io_ticks, we use update_io_ticks and
inflight to update io_ticks when diskstats_show and part_stat_show
been called.

Fixes: 5b18b5a73760 ("block: delete part_round_stats and switch to less precise counting")
Signed-off-by: Zhang Wensheng <zhangwensheng5@huawei.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Link: https://lore.kernel.org/r/20220217064247.4041435-1-zhangwensheng5@huawei.com
Signed-off-by: Jens Axboe <axboe@kernel.dk>

block, bfq: don't move oom_bfqq

Our test report a UAF:

[ 2073.019181] ==================================================================
[ 2073.019188] BUG: KASAN: use-after-free in __bfq_put_async_bfqq+0xa0/0x168
[ 2073.019191] Write of size 8 at addr ffff8000ccf64128 by task rmmod/72584
[ 2073.019192]
[ 2073.019196] CPU: 0 PID: 72584 Comm: rmmod Kdump: loaded Not tainted 4.19.90-yk #5
[ 2073.019198] Hardware name: QEMU KVM Virtual Machine, BIOS 0.0.0 02/06/2015
[ 2073.019200] Call trace:
[ 2073.019203]  dump_backtrace+0x0/0x310
[ 2073.019206]  show_stack+0x28/0x38
[ 2073.019210]  dump_stack+0xec/0x15c
[ 2073.019216]  print_address_description+0x68/0x2d0
[ 2073.019220]  kasan_report+0x238/0x2f0
[ 2073.019224]  __asan_store8+0x88/0xb0
[ 2073.019229]  __bfq_put_async_bfqq+0xa0/0x168
[ 2073.019233]  bfq_put_async_queues+0xbc/0x208
[ 2073.019236]  bfq_pd_offline+0x178/0x238
[ 2073.019240]  blkcg_deactivate_policy+0x1f0/0x420
[ 2073.019244]  bfq_exit_queue+0x128/0x178
[ 2073.019249]  blk_mq_exit_sched+0x12c/0x160
[ 2073.019252]  elevator_exit+0xc8/0xd0
[ 2073.019256]  blk_exit_queue+0x50/0x88
[ 2073.019259]  blk_cleanup_queue+0x228/0x3d8
[ 2073.019267]  null_del_dev+0xfc/0x1e0 [null_blk]
[ 2073.019274]  null_exit+0x90/0x114 [null_blk]
[ 2073.019278]  __arm64_sys_delete_module+0x358/0x5a0
[ 2073.019282]  el0_svc_common+0xc8/0x320
[ 2073.019287]  el0_svc_handler+0xf8/0x160
[ 2073.019290]  el0_svc+0x10/0x218
[ 2073.019291]
[ 2073.019294] Allocated by task 14163:
[ 2073.019301]  kasan_kmalloc+0xe0/0x190
[ 2073.019305]  kmem_cache_alloc_node_trace+0x1cc/0x418
[ 2073.019308]  bfq_pd_alloc+0x54/0x118
[ 2073.019313]  blkcg_activate_policy+0x250/0x460
[ 2073.019317]  bfq_create_group_hierarchy+0x38/0x110
[ 2073.019321]  bfq_init_queue+0x6d0/0x948
[ 2073.019325]  blk_mq_init_sched+0x1d8/0x390
[ 2073.019330]  elevator_switch_mq+0x88/0x170
[ 2073.019334]  elevator_switch+0x140/0x270
[ 2073.019338]  elv_iosched_store+0x1a4/0x2a0
[ 2073.019342]  queue_attr_store+0x90/0xe0
[ 2073.019348]  sysfs_kf_write+0xa8/0xe8
[ 2073.019351]  kernfs_fop_write+0x1f8/0x378
[ 2073.019359]  __vfs_write+0xe0/0x360
[ 2073.019363]  vfs_write+0xf0/0x270
[ 2073.019367]  ksys_write+0xdc/0x1b8
[ 2073.019371]  __arm64_sys_write+0x50/0x60
[ 2073.019375]  el0_svc_common+0xc8/0x320
[ 2073.019380]  el0_svc_handler+0xf8/0x160
[ 2073.019383]  el0_svc+0x10/0x218
[ 2073.019385]
[ 2073.019387] Freed by task 72584:
[ 2073.019391]  __kasan_slab_free+0x120/0x228
[ 2073.019394]  kasan_slab_free+0x10/0x18
[ 2073.019397]  kfree+0x94/0x368
[ 2073.019400]  bfqg_put+0x64/0xb0
[ 2073.019404]  bfqg_and_blkg_put+0x90/0xb0
[ 2073.019408]  bfq_put_queue+0x220/0x228
[ 2073.019413]  __bfq_put_async_bfqq+0x98/0x168
[ 2073.019416]  bfq_put_async_queues+0xbc/0x208
[ 2073.019420]  bfq_pd_offline+0x178/0x238
[ 2073.019424]  blkcg_deactivate_policy+0x1f0/0x420
[ 2073.019429]  bfq_exit_queue+0x128/0x178
[ 2073.019433]  blk_mq_exit_sched+0x12c/0x160
[ 2073.019437]  elevator_exit+0xc8/0xd0
[ 2073.019440]  blk_exit_queue+0x50/0x88
[ 2073.019443]  blk_cleanup_queue+0x228/0x3d8
[ 2073.019451]  null_del_dev+0xfc/0x1e0 [null_blk]
[ 2073.019459]  null_exit+0x90/0x114 [null_blk]
[ 2073.019462]  __arm64_sys_delete_module+0x358/0x5a0
[ 2073.019467]  el0_svc_common+0xc8/0x320
[ 2073.019471]  el0_svc_handler+0xf8/0x160
[ 2073.019474]  el0_svc+0x10/0x218
[ 2073.019475]
[ 2073.019479] The buggy address belongs to the object at ffff8000ccf63f00
 which belongs to the cache kmalloc-1024 of size 1024
[ 2073.019484] The buggy address is located 552 bytes inside of
 1024-byte region [ffff8000ccf63f00, ffff8000ccf64300)
[ 2073.019486] The buggy address belongs to the page:
[ 2073.019492] page:ffff7e000333d800 count:1 mapcount:0 mapping:ffff8000c0003a00 index:0x0 compound_mapcount: 0
[ 2073.020123] flags: 0x7ffff0000008100(slab|head)
[ 2073.020403] raw: 07ffff0000008100 ffff7e0003334c08 ffff7e00001f5a08 ffff8000c0003a00
[ 2073.020409] raw: 0000000000000000 00000000001c001c 00000001ffffffff 0000000000000000
[ 2073.020411] page dumped because: kasan: bad access detected
[ 2073.020412]
[ 2073.020414] Memory state around the buggy address:
[ 2073.020420]  ffff8000ccf64000: fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb
[ 2073.020424]  ffff8000ccf64080: fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb
[ 2073.020428] >ffff8000ccf64100: fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb
[ 2073.020430]                                   ^
[ 2073.020434]  ffff8000ccf64180: fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb
[ 2073.020438]  ffff8000ccf64200: fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb
[ 2073.020439] ==================================================================

The same problem exist in mainline as well.

This is because oom_bfqq is moved to a non-root group, thus root_group
is freed earlier.

Thus fix the problem by don't move oom_bfqq.

Signed-off-by: Yu Kuai <yukuai3@huawei.com>
Reviewed-by: Jan Kara <jack@suse.cz>
Acked-by: Paolo Valente <paolo.valente@linaro.org>
Link: https://lore.kernel.org/r/20220129015924.3958918-4-yukuai3@huawei.com
Signed-off-by: Jens Axboe <axboe@kernel.dk>

block, bfq: avoid moving bfqq to it's parent bfqg

Moving bfqq to it's parent bfqg is pointless.

Signed-off-by: Yu Kuai <yukuai3@huawei.com>
Link: https://lore.kernel.org/r/20220129015924.3958918-3-yukuai3@huawei.com
Signed-off-by: Jens Axboe <axboe@kernel.dk>

block, bfq: cleanup bfq_bfqq_to_bfqg()

Use bfq_group() instead, which do the same thing.

Signed-off-by: Yu Kuai <yukuai3@huawei.com>
Reviewed-by: Jan Kara <jack@suse.cz>
Acked-by: Paolo Valente <paolo.valente@linaro.org>
Link: https://lore.kernel.org/r/20220129015924.3958918-2-yukuai3@huawei.com
Signed-off-by: Jens Axboe <axboe@kernel.dk>

block/bfq_wf2q: correct weight to ioprio

The return value is ioprio * BFQ_WEIGHT_CONVERSION_COEFF or 0.
What we want is ioprio or 0.
Correct this by changing the calculation.

Signed-off-by: Yahu Gao <gaoyahu19@gmail.com>
Acked-by: Paolo Valente <paolo.valente@linaro.org>
Link: https://lore.kernel.org/r/20220107065859.25689-1-gaoyahu19@gmail.com
Signed-off-by: Jens Axboe <axboe@kernel.dk>

blk-mq: avoid extending delays of active hctx from blk_mq_delay_run_hw_queues

When blk_mq_delay_run_hw_queues sets an hctx to run in the future, it can
reset the delay length for an already pending delayed work run_work. This
creates a scenario where multiple hctx may have their queues set to run,
but if one runs first and finds nothing to do, it can reset the delay of
another hctx and stall the other hctx's ability to run requests.

To avoid this I/O stall when an hctx's run_work is already pending,
leave it untouched to run at its current designated time rather than
extending its delay. The work will still run which keeps closed the race
calling blk_mq_delay_run_hw_queues is needed for while also avoiding the
I/O stall.

Signed-off-by: David Jeffery <djeffery@redhat.com>
Reviewed-by: Ming Lei <ming.lei@redhat.com>
Link: https://lore.kernel.org/r/20220131203337.GA17666@redhat
Signed-off-by: Jens Axboe <axboe@kernel.dk>

virtio_blk: simplify refcounting

Implement the ->free_disk method to free the virtio_blk structure only
once the last gendisk reference goes away instead of keeping a local
refcount.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Acked-by: Michael S. Tsirkin <mst@redhat.com>
Link: https://lore.kernel.org/r/20220215094514.3828912-6-hch@lst.de
Signed-off-by: Jens Axboe <axboe@kernel.dk>

memstick/mspro_block: simplify refcounting

Implement the ->free_disk method to free the msb_data structure only once
the last gendisk reference goes away instead of keeping a local
refcount.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Link: https://lore.kernel.org/r/20220215094514.3828912-5-hch@lst.de
Signed-off-by: Jens Axboe <axboe@kernel.dk>

memstick/mspro_block: fix handling of read-only devices

Use set_disk_ro to propagate the read-only state to the block layer
instead of checking for it in ->open and leaking a reference in case
of a read-only device.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Link: https://lore.kernel.org/r/20220215094514.3828912-4-hch@lst.de
Signed-off-by: Jens Axboe <axboe@kernel.dk>

memstick/ms_block: simplify refcounting

Implement the ->free_disk method to free the msb_data structure only once
the last gendisk reference goes away instead of keeping a local refcount.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Link: https://lore.kernel.org/r/20220215094514.3828912-3-hch@lst.de
Signed-off-by: Jens Axboe <axboe@kernel.dk>

block: add a ->free_disk method

Add a method to notify the driver that the gendisk is about to be freed.
This allows drivers to tie the lifetime of their private data to that of
the gendisk and thus deal with device removal races without expensive
synchronization and boilerplate code.

A new flag is added so that ->free_disk is only called after a successful
call to add_disk, which significantly simplifies the error handling path
during probing.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Link: https://lore.kernel.org/r/20220215094514.3828912-2-hch@lst.de
Signed-off-by: Jens Axboe <axboe@kernel.dk>

block: revert 4f1e9630afe6 ("blk-throtl: optimize IOPS throttle for large IO scenarios")

Revert commit 4f1e9630afe6 ("blk-throtl: optimize IOPS throttle for large
IO scenarios") since we have another easier way to address this issue and
get better iops throttling result.

Acked-by: Tejun Heo <tj@kernel.org>
Signed-off-by: Ming Lei <ming.lei@redhat.com>
Link: https://lore.kernel.org/r/20220216044514.2903784-9-ming.lei@redhat.com
Signed-off-by: Jens Axboe <axboe@kernel.dk>

block: don't try to throttle split bio if iops limit isn't set

We need to throttle split bio in case of IOPS limit even though the
split bio has been marked as BIO_THROTTLED since block layer
accounts split bio actually.

If only throughput throttle is setup, no need to throttle any more
if BIO_THROTTLED is set since we have accounted & considered the
whole bio bytes already.

Add one flag of THROTL_TG_HAS_IOPS_LIMIT for serving this purpose.

Acked-by: Tejun Heo <tj@kernel.org>
Signed-off-by: Ming Lei <ming.lei@redhat.com>
Link: https://lore.kernel.org/r/20220216044514.2903784-8-ming.lei@redhat.com
Signed-off-by: Jens Axboe <axboe@kernel.dk>

block: throttle split bio in case of iops limit

Commit 111be8839817 ("block-throttle: avoid double charge") marks bio as
BIO_THROTTLED unconditionally if __blk_throtl_bio() is called on this bio,
then this bio won't be called into __blk_throtl_bio() any more. This way
is to avoid double charge in case of bio splitting. It is reasonable for
read/write throughput limit, but not reasonable for IOPS limit because
block layer provides io accounting against split bio.

Chunguang Xu has already observed this issue and fixed it in commit
4f1e9630afe6 ("blk-throtl: optimize IOPS throttle for large IO scenarios").
However, that patch only covers bio splitting in __blk_queue_split(), and
we have other kind of bio splitting, such as bio_split() &
submit_bio_noacct() and other ways.

This patch tries to fix the issue in one generic way by always charging
the bio for iops limit in blk_throtl_bio(). This way is reasonable:
re-submission & fast-cloned bio is charged if it is submitted to same
disk/queue, and BIO_THROTTLED will be cleared if bio->bi_bdev is changed.

This new approach can get much more smooth/stable iops limit compared with
commit 4f1e9630afe6 ("blk-throtl: optimize IOPS throttle for large IO
scenarios") since that commit can't throttle current split bios actually.

Also this way won't cause new double bio iops charge in
blk_throtl_dispatch_work_fn() in which blk_throtl_bio() won't be called
any more.

Reported-by: Ning Li <lining2020x@163.com>
Acked-by: Tejun Heo <tj@kernel.org>
Cc: Chunguang Xu <brookxu@tencent.com>
Signed-off-by: Ming Lei <ming.lei@redhat.com>
Link: https://lore.kernel.org/r/20220216044514.2903784-7-ming.lei@redhat.com
Signed-off-by: Jens Axboe <axboe@kernel.dk>

block: merge submit_bio_checks() into submit_bio_noacct

Now submit_bio_checks() is only called by submit_bio_noacct(), so merge
it into submit_bio_noacct().

Suggested-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Ming Lei <ming.lei@redhat.com>
Reviewed-by: Chaitanya Kulkarni <kch@nvidia.com>
Link: https://lore.kernel.org/r/20220216044514.2903784-6-ming.lei@redhat.com
Signed-off-by: Jens Axboe <axboe@kernel.dk>

block: don't check bio in blk_throtl_dispatch_work_fn

The bio has been checked already before throttling, so no need to check
it again before dispatching it from throttle queue.

Add a helper of submit_bio_noacct_nocheck() for this purpose.

Signed-off-by: Ming Lei <ming.lei@redhat.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Link: https://lore.kernel.org/r/20220216044514.2903784-5-ming.lei@redhat.com
Signed-off-by: Jens Axboe <axboe@kernel.dk>

block: don't declare submit_bio_checks in local header

submit_bio_checks() won't be called outside of block/blk-core.c any more
since commit 9d497e2941c3 ("block: don't protect submit_bio_checks by
q_usage_counter"), so mark it as one local helper.

Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Ming Lei <ming.lei@redhat.com>
Reviewed-by: Chaitanya Kulkarni <kch@nvidia.com>
Link: https://lore.kernel.org/r/20220216044514.2903784-4-ming.lei@redhat.com
Signed-off-by: Jens Axboe <axboe@kernel.dk>

block: move blk_crypto_bio_prep() out of blk-mq.c

blk_crypto_bio_prep() is called for both bio based and blk-mq drivers,
so move it out of blk-mq.c, then we can unify this kind of handling.

Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Ming Lei <ming.lei@redhat.com>
Reviewed-by: Chaitanya Kulkarni <kch@nvidia.com>
Link: https://lore.kernel.org/r/20220216044514.2903784-3-ming.lei@redhat.com
Signed-off-by: Jens Axboe <axboe@kernel.dk>

block: move submit_bio_checks() into submit_bio_noacct

It is more clean & readable to check bio when starting to submit it,
instead of just before calling ->submit_bio() or blk_mq_submit_bio().

Also it provides us chance to optimize bio submission without checking
bio.

Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Ming Lei <ming.lei@redhat.com>
Reviewed-by: Chaitanya Kulkarni <kch@nvidia.com>
Link: https://lore.kernel.org/r/20220216044514.2903784-2-ming.lei@redhat.com
Signed-off-by: Jens Axboe <axboe@kernel.dk>

dm: remove dm_dispatch_clone_request

Fold dm_dispatch_clone_request into it's only caller, and use a switch
statement to single dispatch for the handling of the different return
values from blk_insert_cloned_request.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Mike Snitzer <snitzer@redhat.com>
Link: https://lore.kernel.org/r/20220215100540.3892965-6-hch@lst.de
Signed-off-by: Jens Axboe <axboe@kernel.dk>

dm: remove useless code from dm_dispatch_clone_request

Both ->start_time_ns and the RQF_IO_STAT are set when the request is
allocated using blk_mq_alloc_request by dm-mpath in blk_mq_rq_ctx_init.
The block layer also ensures ->start_time_ns is only set when actually
needed.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Mike Snitzer <snitzer@redhat.com>
Link: https://lore.kernel.org/r/20220215100540.3892965-5-hch@lst.de
Signed-off-by: Jens Axboe <axboe@kernel.dk>

blk-mq: remove the request_queue argument to blk_insert_cloned_request

The request must be submitted to the queue it was allocated for, so
remove the extra request_queue argument.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Mike Snitzer <snitzer@redhat.com>
Link: https://lore.kernel.org/r/20220215100540.3892965-4-hch@lst.de
Signed-off-by: Jens Axboe <axboe@kernel.dk>

blk-mq: fold blk_cloned_rq_check_limits into blk_insert_cloned_request

Fold blk_cloned_rq_check_limits into its only caller.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Mike Snitzer <snitzer@redhat.com>
Link: https://lore.kernel.org/r/20220215100540.3892965-3-hch@lst.de
Signed-off-by: Jens Axboe <axboe@kernel.dk>

blk-mq: make the blk-mq stacking code optional

The code to stack blk-mq drivers is only used by dm-multipath, and
will preferably stay that way.  Make it optional and only selected
by device mapper, so that the buildbots more easily catch abuses
like the one that slipped in in the ufs driver in the last merged
window.  Another positive side effects is that kernel builds without
device mapper shrink a little bit as well.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Mike Snitzer <snitzer@redhat.com>
Link: https://lore.kernel.org/r/20220215100540.3892965-2-hch@lst.de
Signed-off-by: Jens Axboe <axboe@kernel.dk>

blk-cgroup: set blkg iostat after percpu stat aggregation

Don't need to do blkg_iostat_set for top blkg iostat on each CPU,
so move it after percpu stat aggregation.

Fixes: ef45fe470e1e ("blk-cgroup: show global disk stats in root cgroup io.stat")
Signed-off-by: Chengming Zhou <zhouchengming@bytedance.com>
Acked-by: Tejun Heo <tj@kernel.org>
Link: https://lore.kernel.org/r/20220213085902.88884-1-zhouchengming@bytedance.com
Signed-off-by: Jens Axboe <axboe@kernel.dk>

blk-lib: don't check bdev_get_queue() NULL check

Based on the comment present in the bdev_get_queue()
bdev->bd_queue can never be NULL. Remove the NULL check for the local
variable q that is set from bdev_get_queue() for discard, write_same,
and write_zeroes.

Signed-off-by: Chaitanya Kulkarni <kch@nvidia.com>
Reviewed-by: Ming Lei <ming.lei@redhat.com>
Link: https://lore.kernel.org/r/20220215115247.11717-2-kch@nvidia.com
Signed-off-by: Jens Axboe <axboe@kernel.dk>

block: remove biodoc.rst

This document is completely out of date and extremely misleading. In
general the existing kerneldoc comment serve as a much better
documentation of the still existing functionality, while the history
blurbs are pretty much irrelevant today.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Johannes Thumshirn <johannes.thumshirn@wdc.com>
Link: https://lore.kernel.org/r/20220215081047.3693582-1-hch@lst.de
Signed-off-by: Jens Axboe <axboe@kernel.dk>

docs: block: biodoc.rst: Drop the obsolete and incorrect content

Since commit 7eaceaccab5f ("block: remove per-queue plugging"), kernel
has removed blk_run_address_space(), blk_unplug() and sync_buffer(),
and moved to on-stack plugging. The document has been obsolete for
years.
Given that there is no obvious counterparts in the new mechinism to
replace old APIs, this patch drops the content directly.

Signed-off-by: Barry Song <song.bao.hua@hisilicon.com>
Link: https://lore.kernel.org/r/20220207074931.20067-1-song.bao.hua@hisilicon.com
Signed-off-by: Jens Axboe <axboe@kernel.dk>

block: partition include/linux/blk-cgroup.h

Partition include/linux/blk-cgroup.h into two parts: one is public part,
the other is block layer private part.

Suggested by Christoph Hellwig.

Signed-off-by: Ming Lei <ming.lei@redhat.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Link: https://lore.kernel.org/r/20220211101149.2368042-4-ming.lei@redhat.com
Signed-off-by: Jens Axboe <axboe@kernel.dk>

block: move initialization of q->blkg_list into blkcg_init_queue

q->blkg_list is only used by blkcg code, so move it into
blkcg_init_queue.

Reviewed-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Bart Van Assche <bvanassche@acm.org>
Signed-off-by: Ming Lei <ming.lei@redhat.com>
Link: https://lore.kernel.org/r/20220211101149.2368042-3-ming.lei@redhat.com
Signed-off-by: Jens Axboe <axboe@kernel.dk>

block: remove THROTL_IOPS_MAX

No one uses THROTL_IOPS_MAX any more, so remove it.

Signed-off-by: Ming Lei <ming.lei@redhat.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Link: https://lore.kernel.org/r/20220211101149.2368042-2-ming.lei@redhat.com
Signed-off-by: Jens Axboe <axboe@kernel.dk>

block: introduce block_rq_error tracepoint

Currently, rasdaemon uses the existing tracepoint block_rq_complete
and filters out non-error cases in order to capture block disk errors.

But there are a few problems with this approach:

1. Even kernel trace filter could do the filtering work, there is
   still some overhead after we enable this tracepoint.

2. The filter is merely based on errno, which does not align with kernel
   logic to check the errors for print_req_error().

3. block_rq_complete only provides dev major and minor to identify
   the block device, it is not convenient to use in user-space.

So introduce a new tracepoint block_rq_error just for the error case.
With this patch, rasdaemon could switch to block_rq_error.

Since the new tracepoint has the similar implementation with
block_rq_complete, so move the existing code from TRACE_EVENT
block_rq_complete() into new event class block_rq_completion(). Then add
event for block_rq_complete and block_rq_err respectively from the newly
created event class per the suggestion from Chaitanya Kulkarni.

Cc: Jens Axboe <axboe@kernel.dk>
Cc: Christoph Hellwig <hch@infradead.org>
Reviewed-by: Steven Rostedt <rostedt@goodmis.org>
Signed-off-by: Cong Wang <xiyou.wangcong@gmail.com>
Signed-off-by: Chaitanya Kulkarni <kch@nvidia.com>
Signed-off-by: Yang Shi <shy828301@gmail.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Link: https://lore.kernel.org/r/20220210225222.260069-1-shy828301@gmail.com
Signed-off-by: Jens Axboe <axboe@kernel.dk>

sbitmap: Delete old sbitmap_queue_get_shallow()

Since __sbitmap_queue_get_shallow() was introduced in commit c05e66733788
("sbitmap: add sbitmap_get_shallow() operation"), it has not been used.

Delete __sbitmap_queue_get_shallow() and rename public
__sbitmap_queue_get_shallow() -> sbitmap_queue_get_shallow() as it is odd
to have public __foo but no foo at all.

Signed-off-by: John Garry <john.garry@huawei.com>
Link: https://lore.kernel.org/r/1644322024-105340-1-git-send-email-john.garry@huawei.com
Signed-off-by: Jens Axboe <axboe@kernel.dk>

lib/sbitmap: kill 'depth' from sbitmap_word

Only the last sbitmap_word can have different depth, and all the others
must have same depth of 1U << sb->shift, so not necessary to store it in
sbitmap_word, and it can be retrieved easily and efficiently by adding
one internal helper of __map_depth(sb, index).

Remove 'depth' field from sbitmap_word, then the annotation of
____cacheline_aligned_in_smp for 'word' isn't needed any more.

Not see performance effect when running high parallel IOPS test on
null_blk.

This way saves us one cacheline(usually 64 words) per each sbitmap_word.

Cc: Martin Wilck <martin.wilck@suse.com>
Signed-off-by: Ming Lei <ming.lei@redhat.com>
Reviewed-by: Martin Wilck <mwilck@suse.com>
Reviewed-by: John Garry <john.garry@huawei.com>
Link: https://lore.kernel.org/r/20220110072945.347535-1-ming.lei@redhat.com
Signed-off-by: Jens Axboe <axboe@kernel.dk>

block: pass a block_device to bio_clone_fast

Pass a block_device to bio_clone_fast and __bio_clone_fast and give
the functions more suitable names.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Mike Snitzer <snitzer@redhat.com>
Link: https://lore.kernel.org/r/20220202160109.108149-14-hch@lst.de
Signed-off-by: Jens Axboe <axboe@kernel.dk>

block: initialize the target bio in __bio_clone_fast

All callers of __bio_clone_fast initialize the bio first.  Move that
initialization into __bio_clone_fast instead.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Mike Snitzer <snitzer@redhat.com>
Link: https://lore.kernel.org/r/20220202160109.108149-13-hch@lst.de
Signed-off-by: Jens Axboe <axboe@kernel.dk>

dm: use bio_clone_fast in alloc_io/alloc_tio

Replace open coded bio_clone_fast implementations with the actual helper.
Note that the bio allocated as part of the dm_io structure in alloc_io
will only actually be used later in alloc_tio, making this earlier
cloning of the information safe.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Mike Snitzer <snitzer@redhat.com>
Link: https://lore.kernel.org/r/20220202160109.108149-12-hch@lst.de
Signed-off-by: Jens Axboe <axboe@kernel.dk>

block: clone crypto and integrity data in __bio_clone_fast

__bio_clone_fast should also clone integrity and crypto data, as a clone
without those is incomplete.  Right now the only caller that can actually
support crypto and integrity data (dm) does it manually for the one
callchain that supports these, but we better do it properly in the core.

Note that all callers except for the above mentioned one also don't need
to handle failure at all, given that the integrity and crypto clones are
based on mempool allocations that won't fail for sleeping allocations.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Mike Snitzer <snitzer@redhat.com>
Link: https://lore.kernel.org/r/20220202160109.108149-11-hch@lst.de
Signed-off-by: Jens Axboe <axboe@kernel.dk>

dm-cache: remove __remap_to_origin_clear_discard

Fold __remap_to_origin_clear_discard into the two callers to prepare
for bio cloning refactoring.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Mike Snitzer <snitzer@redhat.com>
Link: https://lore.kernel.org/r/20220202160109.108149-10-hch@lst.de
Signed-off-by: Jens Axboe <axboe@kernel.dk>

dm: simplify the single bio fast path in __send_duplicate_bios

Most targets just need a single flush bio.  Open code that case in
__send_duplicate_bios without the need to add the bio to a list.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Mike Snitzer <snitzer@redhat.com>
Link: https://lore.kernel.org/r/20220202160109.108149-9-hch@lst.de
Signed-off-by: Jens Axboe <axboe@kernel.dk>

dm: retun the clone bio from alloc_tio

Return the clone bio embedded into the tio as that is what the callers
actually want.  Similar for the free side.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Mike Snitzer <snitzer@redhat.com>
Link: https://lore.kernel.org/r/20220202160109.108149-8-hch@lst.de
Signed-off-by: Jens Axboe <axboe@kernel.dk>

dm: pass the bio instead of tio to __map_bio

This simplifies the callers a bit.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Mike Snitzer <snitzer@redhat.com>
Link: https://lore.kernel.org/r/20220202160109.108149-7-hch@lst.de
Signed-off-by: Jens Axboe <axboe@kernel.dk>

dm: move cloning the bio into alloc_tio

Move the call to __bio_clone_fast and the assignment of ->len_ptr from
the callers into alloc_tio to prepare for changes to the bio clone API.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Mike Snitzer <snitzer@redhat.com>
Link: https://lore.kernel.org/r/20220202160109.108149-6-hch@lst.de
Signed-off-by: Jens Axboe <axboe@kernel.dk>

dm: fold __send_duplicate_bios into __clone_and_map_simple_bio

Fold __send_duplicate_bios into its only caller to prepare for
refactoring.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Mike Snitzer <snitzer@redhat.com>
Link: https://lore.kernel.org/r/20220202160109.108149-5-hch@lst.de
Signed-off-by: Jens Axboe <axboe@kernel.dk>

dm: fold clone_bio into __clone_and_map_data_bio

Fold clone_bio into its only caller to prepare for refactoring.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Mike Snitzer <snitzer@redhat.com>
Link: https://lore.kernel.org/r/20220202160109.108149-4-hch@lst.de
Signed-off-by: Jens Axboe <axboe@kernel.dk>

dm: add a clone_to_tio helper

Add a helper to stop open coding the container_of operations to get
from the clone bio to the tio structure.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Mike Snitzer <snitzer@redhat.com>
Link: https://lore.kernel.org/r/20220202160109.108149-3-hch@lst.de
Signed-off-by: Jens Axboe <axboe@kernel.dk>

drbd: set ->bi_bdev in drbd_req_new

Make sure the newly allocated bio has the correct bi_bdev set from the
start.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Mike Snitzer <snitzer@redhat.com>
Link: https://lore.kernel.org/r/20220202160109.108149-2-hch@lst.de
Signed-off-by: Jens Axboe <axboe@kernel.dk>

block: call bio_associate_blkg from bio_reset

Call bio_associate_blkg just like bio_set_dev did in the callers before
the conversion to set the block device in bio_reset.

Fixes: a7c50c940477 ("block: pass a block_device and opf to bio_reset")
Reported-by: syzbot+2b3f18414c37b42dcc94@syzkaller.appspotmail.com
Tested-by: syzbot+2b3f18414c37b42dcc94@syzkaller.appspotmail.com
Signed-off-by: Christoph Hellwig <hch@lst.de>
Tested-by: Chaitanya Kulkarni <kch@nvidia.com>
Reviewed-by: Chaitanya Kulkarni <kch@nvidia.com>
Link: https://lore.kernel.org/r/20220204071934.168469-1-hch@lst.de
Signed-off-by: Jens Axboe <axboe@kernel.dk>

scsi: use BLK_STS_OFFLINE for not fully online devices

The new error message for such case looks like

[  172.809565] device offline error, dev sda, sector 3138208 ...

which will not be confused with regular I/O error (BLK_STS_IOERR).

Reviewed-by: Hannes Reinecke <hare@suse.de>
Signed-off-by: Song Liu <song@kernel.org>
Reviewed-by: Martin K. Petersen <martin.petersen@oracle.com>
Link: https://lore.kernel.org/r/20220203192827.1370270-4-song@kernel.org
Signed-off-by: Jens Axboe <axboe@kernel.dk>

block: return -ENODEV for BLK_STS_OFFLINE

Change the user visible return value for BLK_STS_OFFLINE to -ENODEV, which
is more descriptive than existing -EIO.

Signed-off-by: Song Liu <song@kernel.org>
Reviewed-by: Martin K. Petersen <martin.petersen@oracle.com>
Link: https://lore.kernel.org/r/20220203192827.1370270-3-song@kernel.org
Signed-off-by: Jens Axboe <axboe@kernel.dk>

block: introduce BLK_STS_OFFLINE

Currently, drivers reports BLK_STS_IOERR for devices that are not full
online or being removed. This behavior could cause confusion for users,
as they are not really I/O errors from the device.

Solve this issue with a new state BLK_STS_OFFLINE, which reports "device
offline error" in dmesg instead of "I/O error".

EIO is intentionally kept to not change user visible return value.

Signed-off-by: Song Liu <song@kernel.org>
Reviewed-by: Martin K. Petersen <martin.petersen@oracle.com>
Link: https://lore.kernel.org/r/20220203192827.1370270-2-song@kernel.org
Signed-off-by: Jens Axboe <axboe@kernel.dk>

fs/ntfs3: remove unnecessary NULL check

This code triggers a Smatch warning:

    fs/ntfs3/fsntfs.c:1606 ntfs_bio_fill_1()
    warn: variable dereferenced before check 'bio' (see line 1591)

The "bio" pointer cannot be NULL so there is no need to check.
Originally there was more extensive NULL checking but it was removed
because bio_alloc() will never fail if it is allowed to sleep.

Remove this check as well.

Fixes: 39146b6f66ba ("ntfs3: remove ntfs_alloc_bio")
Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Link: https://lore.kernel.org/r/20220128140922.GA29766@kili
Signed-off-by: Jens Axboe <axboe@kernel.dk>

block: fix boolreturn.cocci warning

Return statements in functions returning bool should use true/false
instead of 1/0.

./block/bio.c:1081:9-10: WARNING: return of 0/1 in function
'bio_add_folio' with return type bool.

Reported-by: Abaci Robot <abaci@linux.alibaba.com>
Signed-off-by: Jiapeng Chong <jiapeng.chong@linux.alibaba.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Chaitanya Kulkarni <kch@nvidia.com>
Link: https://lore.kernel.org/r/20220128043454.68927-1-jiapeng.chong@linux.alibaba.com
Signed-off-by: Jens Axboe <axboe@kernel.dk>

MAINTAINERS: add bio.h to the block section

bio.h is part of the block layer, so list it in the MAINTAINERS file
as such.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Link: https://lore.kernel.org/r/20220127064221.1314477-1-hch@lst.de
Signed-off-by: Jens Axboe <axboe@kernel.dk>