The newly added loop for the cached requests in __blk_mq_alloc_request
is a little too convoluted for my taste, so unwind it a bit. Also
rename the function to __blk_mq_alloc_requests now that it can allocate
more than a single request.
Signed-off-by: Christoph Hellwig <hch@lst.de>
Link: https://lore.kernel.org/r/20211012104045.658051-2-hch@lst.de
Signed-off-by: Jens Axboe <axboe@kernel.dk>
The caller typically has a good (or even exact) idea of how many requests
it needs to submit. We can make the request/tag allocation a lot more
efficient if we just allocate N requests/tags upfront when we queue the
first bio from the batch.
Provide a new plug start helper that allows the caller to specify how many
IOs are expected. This sets plug->nr_ios, and we can use that for smarter
request allocation. The plug provides a holding spot for requests, and
request allocation will check it before calling into the normal request
allocation path.
The blk_finish_plug() is called, check if there are unused requests and
free them. This should not happen in normal operations. The exception is
if we get merging, then we may be left with requests that need freeing
when done.
This raises the per-core performance on my setup from ~5.8M to ~6.1M
IOPS.
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Particularly for NVMe with efficient deferred submission for many
requests, there are nice benefits to be seen by bumping the default max
plug count from 16 to 32. This is especially true for virtualized setups,
where the submit part is more expensive. But can be noticed even on
native hardware.
Reduce the multiple queue factor from 4 to 2, since we're changing the
default size.
While changing it, move the defines into the block layer private header.
These aren't values that anyone outside of the block layer uses, or
should use.
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Doing high IOPS testing with blk-cgroups enabled spends ~15-20% of the
time just doing ktime_get_ns() -> readtsc. We essentially read and
set the start time twice, one for the bio and then again when that bio
is mapped to a request.
Given that the time between the two is very short, inherit the bio
start time instead of reading it again. This cuts 1/3rd of the overhead
of the time keeping.
Acked-by: Tejun Heo <tj@kernel.org>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Even if no policies are defined, we spend ~2% of the total IO time
checking. Move the fast path inline.
Acked-by: Tejun Heo <tj@kernel.org>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Currently we use separate sbitmap pairs and active_queues atomic_t for
shared sbitmap support.
However a full sets of static requests are used per HW queue, which is
quite wasteful, considering that the total number of requests usable at
any given time across all HW queues is limited by the shared sbitmap depth.
As such, it is considerably more memory efficient in the case of shared
sbitmap to allocate a set of static rqs per tag set or request queue, and
not per HW queue.
So replace the sbitmap pairs and active_queues atomic_t with a shared
tags per tagset and request queue, which will hold a set of shared static
rqs.
Since there is now no valid HW queue index to be passed to the blk_mq_ops
.init and .exit_request callbacks, pass an invalid index token. This
changes the semantics of the APIs, such that the callback would need to
validate the HW queue index before using it. Currently no user of shared
sbitmap actually uses the HW queue index (as would be expected).
Signed-off-by: John Garry <john.garry@huawei.com>
Reviewed-by: Ming Lei <ming.lei@redhat.com>
Link: https://lore.kernel.org/r/1633429419-228500-13-git-send-email-john.garry@huawei.com
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Add a function to combine allocating tags and the associated requests,
and factor out common patterns to use this new function.
Some function only call blk_mq_alloc_map_and_rqs() now, but more
functionality will be added later.
Also make blk_mq_alloc_rq_map() and blk_mq_alloc_rqs() static since they
are only used in blk-mq.c, and finally rename some functions for
conciseness and consistency with other function names:
- __blk_mq_alloc_map_and_{request -> rqs}()
- blk_mq_alloc_{map_and_requests -> set_map_and_rqs}()
Suggested-by: Ming Lei <ming.lei@redhat.com>
Signed-off-by: John Garry <john.garry@huawei.com>
Reviewed-by: Hannes Reinecke <hare@suse.de>
Reviewed-by: Ming Lei <ming.lei@redhat.com>
Link: https://lore.kernel.org/r/1633429419-228500-11-git-send-email-john.garry@huawei.com
Signed-off-by: Jens Axboe <axboe@kernel.dk>
For shared sbitmap, if the call to blk_mq_tag_update_depth() was
successful for any hctx when hctx->sched_tags is not set, then it would be
successful for all (due to nature in which blk_mq_tag_update_depth()
fails).
As such, there is no need to call blk_mq_tag_resize_shared_sbitmap() for
each hctx. So relocate the call until after the hctx iteration under the
!q->elevator check, which is equivalent (to !hctx->sched_tags).
Signed-off-by: John Garry <john.garry@huawei.com>
Reviewed-by: Ming Lei <ming.lei@redhat.com>
Reviewed-by: Hannes Reinecke <hare@suse.de>
Link: https://lore.kernel.org/r/1633429419-228500-4-git-send-email-john.garry@huawei.com
Signed-off-by: Jens Axboe <axboe@kernel.dk>
In addition to reverting commit 7b05bf7710 ("Revert "block/mq-deadline:
Prioritize high-priority requests""), this patch uses 'jiffies' instead
of ktime_get() in the code for aging lower priority requests.
This patch has been tested as follows:
Measured QD=1/jobs=1 IOPS for nullb with the mq-deadline scheduler.
Result without and with this patch: 555 K IOPS.
Measured QD=1/jobs=8 IOPS for nullb with the mq-deadline scheduler.
Result without and with this patch: about 380 K IOPS.
Ran the following script:
set -e
scriptdir=$(dirname "$0")
if [ -e /sys/module/scsi_debug ]; then modprobe -r scsi_debug; fi
modprobe scsi_debug ndelay=1000000 max_queue=16
sd=''
while [ -z "$sd" ]; do
sd=$(basename /sys/bus/pseudo/drivers/scsi_debug/adapter*/host*/target*/*/block/*)
done
echo $((100*1000)) > "/sys/block/$sd/queue/iosched/prio_aging_expire"
if [ -e /sys/fs/cgroup/io.prio.class ]; then
cd /sys/fs/cgroup
echo restrict-to-be >io.prio.class
echo +io > cgroup.subtree_control
else
cd /sys/fs/cgroup/blkio/
echo restrict-to-be >blkio.prio.class
fi
echo $$ >cgroup.procs
mkdir -p hipri
cd hipri
if [ -e io.prio.class ]; then
echo none-to-rt >io.prio.class
else
echo none-to-rt >blkio.prio.class
fi
{ "${scriptdir}/max-iops" -a1 -d32 -j1 -e mq-deadline "/dev/$sd" >& ~/low-pri.txt & }
echo $$ >cgroup.procs
"${scriptdir}/max-iops" -a1 -d32 -j1 -e mq-deadline "/dev/$sd" >& ~/hi-pri.txt
Result:
* 11000 IOPS for the high-priority job
* 40 IOPS for the low-priority job
If the prio aging expiry time is changed from 100s into 0, the IOPS results
change into 6712 and 6796 IOPS.
The max-iops script is a script that runs fio with the following arguments:
--bs=4K --gtod_reduce=1 --ioengine=libaio --ioscheduler=${arg_e} --runtime=60
--norandommap --rw=read --thread --buffered=0 --numjobs=${arg_j}
--iodepth=${arg_d} --iodepth_batch_submit=${arg_a}
--iodepth_batch_complete=$((arg_d / 2)) --name=${positional_argument_1}
--filename=${positional_argument_1}
Cc: Damien Le Moal <damien.lemoal@wdc.com>
Cc: Niklas Cassel <Niklas.Cassel@wdc.com>
Cc: Hannes Reinecke <hare@suse.de>
Signed-off-by: Bart Van Assche <bvanassche@acm.org>
Reviewed-by: Damien Le Moal <damien.lemoal@opensource.wdc.com>
Link: https://lore.kernel.org/r/20210927220328.1410161-5-bvanassche@acm.org
[axboe: @latest -> @latest_start]
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Calculating the sum over all CPUs of per-CPU counters frequently is
inefficient. Hence switch from per-CPU to individual counters. Three
counters are protected by the mq-deadline spinlock since these are
only accessed from contexts that already hold that spinlock. The fourth
counter is atomic because protecting it with the mq-deadline spinlock
would trigger lock contention.
Reviewed-by: Damien Le Moal <damien.lemoal@wdc.com>
Reviewed-by: Niklas Cassel <Niklas.Cassel@wdc.com>
Cc: Hannes Reinecke <hare@suse.de>
Signed-off-by: Bart Van Assche <bvanassche@acm.org>
Link: https://lore.kernel.org/r/20210927220328.1410161-4-bvanassche@acm.org
Signed-off-by: Jens Axboe <axboe@kernel.dk>
The scheduler .insert_requests() callback is called when a request is
queued for the first time and also when it is requeued. Only count a
request the first time it is queued. Additionally, since the mq-deadline
scheduler only performs zone locking for requests that have been
inserted, skip the zone unlock code for requests that have not been
inserted into the mq-deadline scheduler.
Fixes: 38ba64d12d ("block/mq-deadline: Track I/O statistics")
Reviewed-by: Damien Le Moal <damien.lemoal@wdc.com>
Reviewed-by: Niklas Cassel <Niklas.Cassel@wdc.com>
Cc: Hannes Reinecke <hare@suse.de>
Signed-off-by: Bart Van Assche <bvanassche@acm.org>
Link: https://lore.kernel.org/r/20210927220328.1410161-2-bvanassche@acm.org
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Except for the features passed to blk_queue_required_elevator_features,
elevator.h is only needed internally to the block layer. Move the
ELEVATOR_F_* definitions to blkdev.h, and the move elevator.h to
block/, dropping all the spurious includes outside of that.
Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Johannes Thumshirn <johannes.thumshirn@wdc.com>
Link: https://lore.kernel.org/r/20210920123328.1399408-13-hch@lst.de
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Pull block fixes from Jens Axboe:
"Bigger than usual for this point in time, the majority is fixing some
issues around BDI lifetimes with the move from the request_queue to
the disk in this release. In detail:
- Series on draining fs IO for del_gendisk() (Christoph)
- NVMe pull request via Christoph:
- fix the abort command id (Keith Busch)
- nvme: fix per-namespace chardev deletion (Adam Manzanares)
- brd locking scope fix (Tetsuo)
- BFQ fix (Paolo)"
* tag 'block-5.15-2021-10-17' of git://git.kernel.dk/linux-block:
block, bfq: reset last_bfqq_created on group change
block: warn when putting the final reference on a registered disk
brd: reduce the brd_devices_mutex scope
kyber: avoid q->disk dereferences in trace points
block: keep q_usage_counter in atomic mode after del_gendisk
block: drain file system I/O on del_gendisk
block: split bio_queue_enter from blk_queue_enter
block: factor out a blk_try_enter_queue helper
block: call submit_bio_checks under q_usage_counter
nvme: fix per-namespace chardev deletion
block/rnbd-clt-sysfs: fix a couple uninitialized variable bugs
nvme-pci: Fix abort command id
Since commit 430a67f9d6 ("block, bfq: merge bursts of newly-created
queues"), BFQ maintains a per-group pointer to the last bfq_queue
created. If such a queue, say bfqq, happens to move to a different
group, then bfqq is no more a valid last bfq_queue created for its
previous group. That pointer must then be cleared. Not resetting such
a pointer may also cause UAF, if bfqq happens to also be freed after
being moved to a different group. This commit performs this missing
reset. As such it fixes commit 430a67f9d6 ("block, bfq: merge bursts
of newly-created queues").
Such a missing reset is most likely the cause of the crash reported in [1].
With some analysis, we found that this crash was due to the
above UAF. And such UAF did go away with this commit applied [1].
Anyway, before this commit, that crash happened to be triggered in
conjunction with commit 2d52c58b9c ("block, bfq: honor already-setup
queue merges"). The latter was then reverted by commit ebc69e897e
("Revert "block, bfq: honor already-setup queue merges""). Yet commit
2d52c58b9c ("block, bfq: honor already-setup queue merges") contains
no error related with the above UAF, and can then be restored.
[1] https://bugzilla.kernel.org/show_bug.cgi?id=214503
Fixes: 430a67f9d6 ("block, bfq: merge bursts of newly-created queues")
Tested-by: Grzegorz Kowal <custos.mentis@gmail.com>
Signed-off-by: Paolo Valente <paolo.valente@linaro.org>
Link: https://lore.kernel.org/r/20211015144336.45894-2-paolo.valente@linaro.org
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Warn when the last reference on a live disk is put without calling
del_gendisk first. There are some BDI related bug reports that look
like a case of this, so make sure we have the proper instrumentation
to catch it.
Signed-off-by: Christoph Hellwig <hch@lst.de>
Link: https://lore.kernel.org/r/20211014130231.1468538-1-hch@lst.de
Signed-off-by: Jens Axboe <axboe@kernel.dk>
q->disk becomes invalid after the gendisk is removed. Work around this
by caching the dev_t for the tracepoints. The real fix would be to
properly tear down the I/O schedulers with the gendisk, but that is
a much more invasive change.
Signed-off-by: Christoph Hellwig <hch@lst.de>
Link: https://lore.kernel.org/r/20211012093301.GA27795@lst.de
Tested-by: Yi Zhang <yi.zhang@redhat.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Instead of delaying draining of file system I/O related items like the
blk-qos queues, the integrity read workqueue and timeouts only when the
request_queue is removed, do that when del_gendisk is called. This is
important for SCSI where the upper level drivers that control the gendisk
are separate entities, and the disk can be freed much earlier than the
request_queue, or can even be unbound without tearing down the queue.
Fixes: edb0872f44 ("block: move the bdi from the request_queue to the gendisk")
Reported-by: Ming Lei <ming.lei@redhat.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>
Tested-by: Darrick J. Wong <djwong@kernel.org>
Link: https://lore.kernel.org/r/20210929071241.934472-5-hch@lst.de
Tested-by: Yi Zhang <yi.zhang@redhat.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Pull block fixes from Jens Axboe:
"Two small fixes for this release:
- Add missing QUEUE_FLAG_HCTX_ACTIVE in the debugfs handling
(Johannes)
- Fix double free / UAF issue in __alloc_disk_node (Tetsuo)"
* tag 'block-5.15-2021-10-09' of git://git.kernel.dk/linux-block:
block: decode QUEUE_FLAG_HCTX_ACTIVE in debugfs output
block: genhd: fix double kfree() in __alloc_disk_node()
Pull block fixes from Jens Axboe:
"A few block fixes for this release:
- Revert a BFQ commit that causes breakage for people. Unfortunately
it was auto-selected for stable as well, so now 5.14.7 suffers from
it too. Hopefully stable will pick up this revert quickly too, so
we can remove the issue on that end as well.
- Add a quirk for Apple NVMe controllers, which due to their
non-compliance broke due to the introduction of command sequences
(Keith)
- Use shifts in nbd, fixing a __divdi3 issue (Nick)"
* tag 'block-5.15-2021-10-01' of git://git.kernel.dk/linux-block:
nbd: use shifts rather than multiplies
Revert "block, bfq: honor already-setup queue merges"
nvme: add command id quirk for apple controllers
This reverts commit 2d52c58b9c.
We have had several folks complain that this causes hangs for them, which
is especially problematic as the commit has also hit stable already.
As no resolution seems to be forthcoming right now, revert the patch.
Link: https://bugzilla.kernel.org/show_bug.cgi?id=214503
Fixes: 2d52c58b9c ("block, bfq: honor already-setup queue merges")
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Pull SCSI fixes from James Bottomley:
"Thirty-three fixes, I'm afraid.
Essentially the build up from the last couple of weeks while I've been
dealling with Linux Plumbers conference infrastructure issues. It's
mostly the usual assortment of spelling fixes and minor corrections.
The only core relevant changes are to the sd driver to reduce the spin
up message spew and fix a small memory leak on the freeing path"
* tag 'scsi-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/jejb/scsi: (33 commits)
scsi: ses: Retry failed Send/Receive Diagnostic commands
scsi: target: Fix spelling mistake "CONFLIFT" -> "CONFLICT"
scsi: lpfc: Fix gcc -Wstringop-overread warning, again
scsi: lpfc: Use correct scnprintf() limit
scsi: lpfc: Fix sprintf() overflow in lpfc_display_fpin_wwpn()
scsi: core: Remove 'current_tag'
scsi: acornscsi: Remove tagged queuing vestiges
scsi: fas216: Kill scmd->tag
scsi: qla2xxx: Restore initiator in dual mode
scsi: ufs: core: Unbreak the reset handler
scsi: sd_zbc: Support disks with more than 2**32 logical blocks
scsi: ufs: core: Revert "scsi: ufs: Synchronize SCSI and UFS error handling"
scsi: bsg: Fix device unregistration
scsi: sd: Make sd_spinup_disk() less noisy
scsi: ufs: ufs-pci: Fix Intel LKF link stability
scsi: mpt3sas: Clean up some inconsistent indenting
scsi: megaraid: Clean up some inconsistent indenting
scsi: sr: Fix spelling mistake "does'nt" -> "doesn't"
scsi: Remove SCSI CDROM MAINTAINERS entry
scsi: megaraid: Fix Coccinelle warning
...
When running ->fallocate(), blkdev_fallocate() should hold
mapping->invalidate_lock to prevent page cache from being accessed,
otherwise stale data may be read in page cache.
Without this patch, blktests block/009 fails sometimes. With this patch,
block/009 can pass always.
Also as Jan pointed out, no pages can be created in the discarded area
while you are holding the invalidate_lock, so remove the 2nd
truncate_bdev_range().
Cc: Jan Kara <jack@suse.cz>
Signed-off-by: Ming Lei <ming.lei@redhat.com>
Reviewed-by: Jan Kara <jack@suse.cz>
Link: https://lore.kernel.org/r/20210923023751.1441091-1-ming.lei@redhat.com
Signed-off-by: Jens Axboe <axboe@kernel.dk>
rq_qos framework is only applied on request based driver, so:
1) rq_qos_done_bio() needn't to be called for bio based driver
2) rq_qos_done_bio() needn't to be called for bio which isn't tracked,
such as bios ended from error handling code.
Especially in bio_endio():
1) request queue is referred via bio->bi_bdev->bd_disk->queue, which
may be gone since request queue refcount may not be held in above two
cases
2) q->rq_qos may be freed in blk_cleanup_queue() when calling into
__rq_qos_done_bio()
Fix the potential kernel panic by not calling rq_qos_ops->done_bio if
the bio isn't tracked. This way is safe because both ioc_rqos_done_bio()
and blkcg_iolatency_done_bio() are nop if the bio isn't tracked.
Reported-by: Yu Kuai <yukuai3@huawei.com>
Cc: tj@kernel.org
Signed-off-by: Ming Lei <ming.lei@redhat.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Acked-by: Tejun Heo <tj@kernel.org>
Link: https://lore.kernel.org/r/20210924110704.1541818-1-ming.lei@redhat.com
Signed-off-by: Jens Axboe <axboe@kernel.dk>
