Switch to use bdev_check_media_changed instead of check_disk_change and
call floppy_revalidate manually. Given that floppy_revalidate only
deals with media change events, the extra call into ->revalidate_disk
from bdev_disk_changed is not required either, so stop wiring up the
method.
Signed-off-by: Christoph Hellwig <hch@lst.de> Reviewed-by: Johannes Thumshirn <johannes.thumshirn@wdc.com> Reviewed-by: Hannes Reinecke <hare@suse.de> Signed-off-by: Jens Axboe <axboe@kernel.dk>
floppy_revalidate mostly duplicates work already done in floppy_open
despite only beeing called from floppy_open. Remove the function and
just clear the ->ejected flag directly under the right condition.
Signed-off-by: Christoph Hellwig <hch@lst.de> Reviewed-by: Johannes Thumshirn <johannes.thumshirn@wdc.com> Reviewed-by: Hannes Reinecke <hare@suse.de> Signed-off-by: Jens Axboe <axboe@kernel.dk>
Switch to use bdev_check_media_change instead of check_disk_change and
call floppy_revalidate manually. Given that floppy_revalidate only
deals with media change events, the extra call into ->revalidate_disk
from bdev_disk_changed is not required either, so stop wiring up the
method.
Signed-off-by: Christoph Hellwig <hch@lst.de> Reviewed-by: Johannes Thumshirn <johannes.thumshirn@wdc.com> Reviewed-by: Hannes Reinecke <hare@suse.de> Signed-off-by: Jens Axboe <axboe@kernel.dk>
Switch to use bdev_check_media_change instead of check_disk_change and
call floppy_revalidate manually. Given that floppy_revalidate only
deals with media change events, the extra call into ->revalidate_disk
from bdev_disk_changed is not required either, so stop wiring up the
method.
Signed-off-by: Christoph Hellwig <hch@lst.de> Reviewed-by: Johannes Thumshirn <johannes.thumshirn@wdc.com> Reviewed-by: Hannes Reinecke <hare@suse.de> Signed-off-by: Jens Axboe <axboe@kernel.dk>
Switch to use bdev_check_media_change instead of check_disk_change and
call floppy_revalidate manually. Given that floppy_revalidate only
deals with media change events, the extra call into ->revalidate_disk
from bdev_disk_changed is not required either, so stop wiring up the
method.
Signed-off-by: Christoph Hellwig <hch@lst.de> Reviewed-by: Johannes Thumshirn <johannes.thumshirn@wdc.com> Reviewed-by: Hannes Reinecke <hare@suse.de> Signed-off-by: Jens Axboe <axboe@kernel.dk>
Jan Kara [Fri, 4 Sep 2020 08:58:52 +0000 (10:58 +0200)]
block: Do not discard buffers under a mounted filesystem
Discarding blocks and buffers under a mounted filesystem is hardly
anything admin wants to do. Usually it will confuse the filesystem and
sometimes the loss of buffer_head state (including b_private field) can
even cause crashes like:
So if we don't have block device open with O_EXCL already, claim the
block device while we truncate buffer cache. This makes sure any
exclusive block device user (such as filesystem) cannot operate on the
device while we are discarding buffer cache.
Reported-by: Ye Bin <yebin10@huawei.com> Signed-off-by: Jan Kara <jack@suse.cz> Reviewed-by: Christoph Hellwig <hch@lst.de>
[axboe: fix !CONFIG_BLOCK error in truncate_bdev_range()] Signed-off-by: Jens Axboe <axboe@kernel.dk>
Jan Kara [Fri, 4 Sep 2020 08:58:51 +0000 (10:58 +0200)]
fs: Don't invalidate page buffers in block_write_full_page()
If block_write_full_page() is called for a page that is beyond current
inode size, it will truncate page buffers for the page and return 0.
This logic has been added in 2.5.62 in commit 81eb69062588 ("fix ext3
BUG due to race with truncate") in history.git tree to fix a problem
with ext3 in data=ordered mode. This particular problem doesn't exist
anymore because ext3 is long gone and ext4 handles ordered data
differently. Also normally buffers are invalidated by truncate code and
there's no need to specially handle this in ->writepage() code.
This invalidation of page buffers in block_write_full_page() is causing
issues to filesystems (e.g. ext4 or ocfs2) when block device is shrunk
under filesystem's hands and metadata buffers get discarded while being
tracked by the journalling layer. Although it is obviously "not
supported" it can cause kernel crashes like:
which is not great. The crash happens because bh->b_private is suddently
NULL although BH_JBD flag is still set (this is because
block_invalidatepage() cleared BH_Mapped flag and subsequent bh lookup
found buffer without BH_Mapped set, called init_page_buffers() which has
rewritten bh->b_private). So just remove the invalidation in
block_write_full_page().
Note that the buffer cache invalidation when block device changes size
is already careful to avoid similar problems by using
invalidate_mapping_pages() which skips busy buffers so it was only this
odd block_write_full_page() behavior that could tear down bdev buffers
under filesystem's hands.
Reported-by: Ye Bin <yebin10@huawei.com> Signed-off-by: Jan Kara <jack@suse.cz> Reviewed-by: Christoph Hellwig <hch@lst.de> CC: stable@vger.kernel.org Signed-off-by: Jens Axboe <axboe@kernel.dk>
Kashyap Desai [Wed, 19 Aug 2020 15:20:28 +0000 (23:20 +0800)]
blk-mq, elevator: Count requests per hctx to improve performance
High CPU utilization on "native_queued_spin_lock_slowpath" due to lock
contention is possible for mq-deadline and bfq IO schedulers
when nr_hw_queues is more than one.
It is because kblockd work queue can submit IO from all online CPUs
(through blk_mq_run_hw_queues()) even though only one hctx has pending
commands.
The elevator callback .has_work for mq-deadline and bfq scheduler considers
pending work if there are any IOs on request queue but it does not account
hctx context.
Add a per-hctx 'elevator_queued' count to the hctx to avoid triggering
the elevator even though there are no requests queued.
[jpg: Relocated atomic_dec() in dd_dispatch_request(), update commit message per Kashyap]
Signed-off-by: Kashyap Desai <kashyap.desai@broadcom.com> Signed-off-by: Hannes Reinecke <hare@suse.de> Signed-off-by: John Garry <john.garry@huawei.com> Tested-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
John Garry [Wed, 19 Aug 2020 15:20:27 +0000 (23:20 +0800)]
blk-mq: Record active_queues_shared_sbitmap per tag_set for when using shared sbitmap
For when using a shared sbitmap, no longer should the number of active
request queues per hctx be relied on for when judging how to share the tag
bitmap.
Instead maintain the number of active request queues per tag_set, and make
the judgement based on that.
Originally-from: Kashyap Desai <kashyap.desai@broadcom.com> Signed-off-by: John Garry <john.garry@huawei.com> Tested-by: Don Brace<don.brace@microsemi.com> #SCSI resv cmds patches used Tested-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
John Garry [Wed, 19 Aug 2020 15:20:26 +0000 (23:20 +0800)]
blk-mq: Record nr_active_requests per queue for when using shared sbitmap
The per-hctx nr_active value can no longer be used to fairly assign a share
of tag depth per request queue for when using a shared sbitmap, as it does
not consider that the tags are shared tags over all hctx's.
For this case, record the nr_active_requests per request_queue, and make
the judgement based on that value.
Co-developed-with: Kashyap Desai <kashyap.desai@broadcom.com> Signed-off-by: John Garry <john.garry@huawei.com> Tested-by: Don Brace<don.brace@microsemi.com> #SCSI resv cmds patches used Tested-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
John Garry [Wed, 19 Aug 2020 15:20:24 +0000 (23:20 +0800)]
blk-mq: Facilitate a shared sbitmap per tagset
Some SCSI HBAs (such as HPSA, megaraid, mpt3sas, hisi_sas_v3 ..) support
multiple reply queues with single hostwide tags.
In addition, these drivers want to use interrupt assignment in
pci_alloc_irq_vectors(PCI_IRQ_AFFINITY). However, as discussed in [0],
CPU hotplug may cause in-flight IO completion to not be serviced when an
interrupt is shutdown. That problem is solved in commit d59cb6e50183
("blk-mq: drain I/O when all CPUs in a hctx are offline").
However, to take advantage of that blk-mq feature, the HBA HW queuess are
required to be mapped to that of the blk-mq hctx's; to do that, the HBA HW
queues need to be exposed to the upper layer.
In making that transition, the per-SCSI command request tags are no
longer unique per Scsi host - they are just unique per hctx. As such, the
HBA LLDD would have to generate this tag internally, which has a certain
performance overhead.
However another problem is that blk-mq assumes the host may accept
(Scsi_host.can_queue * #hw queue) commands. In commit 97bbc2d58b31 ("scsi:
core: avoid host-wide host_busy counter for scsi_mq"), the Scsi host busy
counter was removed, which would stop the LLDD being sent more than
.can_queue commands; however, it should still be ensured that the block
layer does not issue more than .can_queue commands to the Scsi host.
To solve this problem, introduce a shared sbitmap per blk_mq_tag_set,
which may be requested at init time.
New flag BLK_MQ_F_TAG_HCTX_SHARED should be set when requesting the
tagset to indicate whether the shared sbitmap should be used.
Even when BLK_MQ_F_TAG_HCTX_SHARED is set, a full set of tags and requests
are still allocated per hctx; the reason for this is that if tags and
requests were only allocated for a single hctx - like hctx0 - it may break
block drivers which expect a request be associated with a specific hctx,
i.e. not always hctx0. This will introduce extra memory usage.
This change is based on work originally from Ming Lei in [1] and from
Bart's suggestion in [2].
Signed-off-by: John Garry <john.garry@huawei.com> Tested-by: Don Brace<don.brace@microsemi.com> #SCSI resv cmds patches used Tested-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
John Garry [Wed, 19 Aug 2020 15:20:23 +0000 (23:20 +0800)]
blk-mq: Use pointers for blk_mq_tags bitmap tags
Introduce pointers for the blk_mq_tags regular and reserved bitmap tags,
with the goal of later being able to use a common shared tag bitmap across
all HW contexts in a set.
Signed-off-by: John Garry <john.garry@huawei.com> Tested-by: Don Brace<don.brace@microsemi.com> #SCSI resv cmds patches used Tested-by: Douglas Gilbert <dgilbert@interlog.com> Reviewed-by: Hannes Reinecke <hare@suse.de> Signed-off-by: Jens Axboe <axboe@kernel.dk>
Hannes Reinecke [Wed, 19 Aug 2020 15:20:21 +0000 (23:20 +0800)]
blk-mq: Free tags in blk_mq_init_tags() upon error
Since the tags are allocated in blk_mq_init_tags(), it's better practice
to free in that same function upon error, rather than a callee which is to
init the bitmap tags (blk_mq_init_tags()).
[jpg: Split from an earlier patch with a new commit message]
Signed-off-by: Hannes Reinecke <hare@suse.de> Signed-off-by: John Garry <john.garry@huawei.com> Tested-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
Ming Lei [Wed, 19 Aug 2020 15:20:19 +0000 (23:20 +0800)]
blk-mq: Rename BLK_MQ_F_TAG_SHARED as BLK_MQ_F_TAG_QUEUE_SHARED
BLK_MQ_F_TAG_SHARED actually means that tags is shared among request
queues, all of which should belong to LUNs attached to same HBA.
So rename it to make the point explicitly.
[jpg: rebase a few times, add rnbd-clt.c change]
Suggested-by: Bart Van Assche <bvanassche@acm.org> Signed-off-by: Ming Lei <ming.lei@redhat.com> Signed-off-by: John Garry <john.garry@huawei.com> Tested-by: Douglas Gilbert <dgilbert@interlog.com> Reviewed-by: Hannes Reinecke <hare@suse.de> Signed-off-by: Jens Axboe <axboe@kernel.dk>
Signed-off-by: Christoph Hellwig <hch@lst.de> Reviewed-by: Josef Bacik <josef@toxicpanda.com> Reviewed-by: Johannes Thumshirn <johannes.thumshirn@wdc.com> Acked-by: Song Liu <song@kernel.org> Signed-off-by: Jens Axboe <axboe@kernel.dk>
The nvdimm block driver abuse revalidate_disk in a strange way, and
totally unrelated to what other drivers do. Simplify this by just
calling nvdimm_revalidate_disk (which seems rather misnamed) from the
probe routines, as the additional bdev size revalidation is pointless
at this point, and remove the revalidate_disk methods given that
it can only be triggered from add_disk, which is right before the
manual calls.
Signed-off-by: Christoph Hellwig <hch@lst.de> Reviewed-by: Josef Bacik <josef@toxicpanda.com> Reviewed-by: Johannes Thumshirn <johannes.thumshirn@wdc.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
nvme: opencode revalidate_disk in nvme_validate_ns
Keep control in the NVMe driver instead of going through an indirect
call back into ->revalidate_disk. Also reorder the function a bit to be
easier to follow with the additional code.
And now that we have removed all callers of revalidate_disk() in the nvme
code, ->revalidate_disk is only called from the open code when first
opening the device. Which is of course totally pointless as we have
a valid size since the initial scan, and will get an updated view
through the asynchronous notifiation everytime the size changes.
Signed-off-by: Christoph Hellwig <hch@lst.de> Reviewed-by: Josef Bacik <josef@toxicpanda.com> Reviewed-by: Johannes Thumshirn <johannes.thumshirn@wdc.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
block: use revalidate_disk_size in set_capacity_revalidate_and_notify
Only virtio_blk and xen-blkfront set the revalidate argument to true,
and both do not implement the ->revalidate_disk method. So switch
to the helper that just updates the size instead.
Signed-off-by: Christoph Hellwig <hch@lst.de> Reviewed-by: Josef Bacik <josef@toxicpanda.com> Reviewed-by: Johannes Thumshirn <johannes.thumshirn@wdc.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
revalidate_disk is a relative awkward helper for driver use, as it first
calls an optional driver method and then updates the block device size,
while most callers either don't need the method call at all, or want to
keep state between the caller and the called method.
Add a revalidate_disk_size helper that just performs the update of the
block device size from the gendisk one, and switch all drivers that do
not implement ->revalidate_disk to use the new helper instead of
revalidate_disk()
Signed-off-by: Christoph Hellwig <hch@lst.de> Reviewed-by: Josef Bacik <josef@toxicpanda.com> Reviewed-by: Johannes Thumshirn <johannes.thumshirn@wdc.com> Acked-by: Song Liu <song@kernel.org> Signed-off-by: Jens Axboe <axboe@kernel.dk>
block: don't clear bd_invalidated in check_disk_size_change
bd_invalidated is set by check_disk_change or in add_disk to initiate a
partition scan. Move it from check_disk_size_change which is called
from both revalidate_disk() and bdev_disk_changed() to only the latter,
as that is what is called from the block device open code (and nbd) to
deal with the bd_invalidated event. revalidate_disk() on the other hand
is mostly used to propagate a size update from the gendisk to the block
device, which is entirely unrelated.
Signed-off-by: Christoph Hellwig <hch@lst.de> Reviewed-by: Josef Bacik <josef@toxicpanda.com> Reviewed-by: Johannes Thumshirn <johannes.thumshirn@wdc.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
Documentation/filesystems/locking.rst: remove an incorrect sentence
unlock_native_capacity is never called from check_disk_change(), and
while revalidate_disk can be called from it, it can also be called
from two other places at the moment.
Signed-off-by: Christoph Hellwig <hch@lst.de> Reviewed-by: Josef Bacik <josef@toxicpanda.com> Reviewed-by: Johannes Thumshirn <johannes.thumshirn@wdc.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
block: better deal with the delayed not supported case in blk_cloned_rq_check_limits
If WRITE_ZERO/WRITE_SAME operation is not supported by the storage,
blk_cloned_rq_check_limits() will return IO error which will cause
device-mapper to fail the paths.
Instead, if the queue limit is set to 0, return BLK_STS_NOTSUPP.
BLK_STS_NOTSUPP will be ignored by device-mapper and will not fail the
paths.
Suggested-by: Martin K. Petersen <martin.petersen@oracle.com> Signed-off-by: Ritika Srivastava <ritika.srivastava@oracle.com> Reviewed-by: Martin K. Petersen <martin.petersen@oracle.com> Reviewed-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Jens Axboe <axboe@kernel.dk>
When an iocg accumulates too much vtime or gets deactivated, we throw away
some vtime, which lowers the overall device utilization. As the exact amount
which is being thrown away is known, we can compensate by accelerating the
vrate accordingly so that the extra vtime generated in the current period
matches what got lost.
This significantly improves work conservation when involving high weight
cgroups with intermittent and bursty IO patterns.
A low weight iocg can amass a large amount of debt, for example, when
anonymous memory gets reclaimed aggressively. If the system has a lot of
memory paired with a slow IO device, the debt can span multiple seconds or
more. If there are no other subsequent IO issuers, the in-debt iocg may end
up blocked paying its debt while the IO device is idle.
This patch implements a mechanism to protect against such pathological
cases. If the device has been sufficiently idle for a substantial amount of
time, the debts are halved. The criteria are on the conservative side as we
want to resolve the rare extreme cases without impacting regular operation
by forgiving debts too readily.
Curently, iocost syncs the delay duration to the outstanding debt amount,
which seemed enough to protect the system from anon memory hogs. However,
that was mostly because the delay calcuation was using hweight_inuse which
quickly converges towards zero under debt for delay duration calculation,
often pusnishing debtors overly harshly for longer than deserved.
The previous patch fixed the delay calcuation and now the protection against
anonymous memory hogs isn't enough because the effect of delay is indirect
and non-linear and a huge amount of future debt can accumulate abruptly
while unthrottled.
This patch implements delay hysteresis so that delay is decayed
exponentially over time instead of getting cleared immediately as debt is
paid off. While the overall behavior is similar to the blk-cgroup
implementation used by blk-iolatency, a lot of the details are different and
due to the empirical nature of the mechanism, it's challenging to adapt the
mechanism for one controller without negatively impacting the other.
As the delay is gradually decayed now, there's no point in running it from
its own hrtimer. Periodic updates are now performed from ioc_timer_fn() and
the dedicated hrtimer is removed.
* How much inuse a debtor carries wasn't clearly defined. inuse would be
driven down over time from not issuing IOs but it'd be better to clamp it
to minimum immediately once in debt.
* How much can be paid off was determined by hweight_inuse. As inuse was
driven down, the payment amount would fall together regardless of the
debtor's active weight. This means that the debtors were punished harshly.
* ioc_rqos_merge() wasn't calling blkcg_schedule_throttle() after
iocg_kick_delay().
This patch revamps debt handling so that
* Debt handling owns inuse for iocgs in debt and keeps them at zero.
* Payment amount is determined by hweight_active. This is more deterministic
and safer than hweight_inuse but still far from ideal in that it doesn't
factor in possible donations from other iocgs for debt payments. This
likely needs further improvements in the future.
* iocg_rqos_merge() now calls blkcg_schedule_throttle() as necessary.
When the margin drops below the minimum on a donating iocg, donation is
immediately canceled in full. There are a couple shortcomings with the
current behavior.
* It's abrupt. A small temporary budget deficit can lead to a wide swing in
weight allocation and a large surplus.
* It's open coded in the issue path but not implemented for the merge path.
A series of merges at a low inuse can make the iocg incur debts and stall
incorrectly.
This patch reimplements in-period donation snapbacks so that
* inuse adjustment and cost calculations are factored into
adjust_inuse_and_calc_cost() which is called from both the issue and merge
paths.
* Snapbacks are more gradual. It occurs in quarter steps.
* A snapback triggers if the margin goes below the low threshold and is
lower than the budget at the time of the last adjustment.
* For the above, __propagate_weights() stores the margin in
iocg->saved_margin. Move iocg->last_inuse storing together into
__propagate_weights() for consistency.
* Full snapback is guaranteed when there are waiters.
* With precise donation and gradual snapbacks, inuse adjustments are now a
lot more effective and the value of scaling inuse on weight changes isn't
clear. Removed inuse scaling from weight_update().
iocost has various safety nets to combat inuse adjustment calculation
inaccuracies. With Andy's method implemented in transfer_surpluses(), inuse
adjustment calculations are now accurate and we can make donation amount
determinations accurate too.
* Stop keeping track of past usage history and using the maximum. Act on the
immediate usage information.
* Remove donation constraints defined by SURPLUS_* constants. Donate
whatever isn't used.
* Determine the donation amount so that the iocg will end up with
MARGIN_TARGET_PCT budget at the end of the coming period assuming the same
usage as the previous period. TARGET is set at 50% of period, which is the
previous maximum. This provides smooth convergence for most repetitive IO
patterns.
* Apply donation logic early at 20% budget. There's no risk in doing so as
the calculation is based on the delta between the current budget and the
target budget at the end of the coming period.
* Remove preemptive iocg activation for zero cost IOs. As donation can reach
near zero now, the mere activation doesn't provide any protection anymore.
In the unlikely case that this becomes a problem, the right solution is
assigning appropriate costs for such IOs.
This significantly improves the donation determination logic while also
simplifying it. Now all donations are immediate, exact and smooth.
blk-iocost: implement Andy's method for donation weight updates
iocost implements work conservation by reducing iocg->inuse and propagating
the adjustment upwards proportionally. However, while I knew the target
absolute hierarchical proportion - adjusted hweight_inuse, I couldn't figure
out how to determine the iocg->inuse adjustment to achieve that and
approximated the adjustment by scaling iocg->inuse using the proportion of
the needed hweight_inuse changes.
When nested, these scalings aren't accurate even when adjusting a single
node as the donating node also receives the benefit of the donated portion.
When multiple nodes are donating as they often do, they can be wildly wrong.
iocost employed various safety nets to combat the inaccuracies. There are
ample buffers in determining how much to donate, the adjustments are
conservative and gradual. While it can achieve a reasonable level of work
conservation in simple scenarios, the inaccuracies can easily add up leading
to significant loss of total work. This in turn makes it difficult to
closely cap vrate as vrate adjustment is needed to compensate for the loss
of work. The combination of inaccurate donation calculations and vrate
adjustments can lead to wide fluctuations and clunky overall behaviors.
Andy Newell devised a method to calculate the needed ->inuse updates to
achieve the target hweight_inuse's. The method is compatible with the
proportional inuse adjustment propagation which allows all hot path
operations to be local to each iocg.
To roughly summarize, Andy's method divides the tree into donating and
non-donating parts, calculates global donation rate which is used to
determine the target hweight_inuse for each node, and then derives per-level
proportions. There's non-trivial amount of math involved. Please refer to
the following pdfs for detailed descriptions.
This patch implements Andy's method in transfer_surpluses(). This makes the
donation calculations accurate per cycle and enables further improvements in
other parts of the donation logic.
The way the surplus donation logic is structured isn't great. There are two
separate paths for starting/increasing donations and decreasing them making
the logic harder to follow and is prone to unnecessary behavior differences.
In preparation for improved donation handling, this patch restructures the
code so that
* All donors - new, increasing and decreasing - are funneled through the
same code path.
* The target donation calculation is factored into hweight_after_donation()
which is called once from the same spot for all possible donors.
* Actual inuse adjustment is factored into trasnfer_surpluses().
This change introduces a few behavior differences - e.g. donation amount
reduction now uses the max usage of the recent three periods just like new
and increasing donations, and inuse now gets adjusted upwards the same way
it gets downwards. These differences are unlikely to have severely negative
implications and the whole logic will be revamped soon.
This patch also removes two tracepoints. The existing TPs don't quite fit
the new implementation. A later patch will update and reinstate them.
blk-iocost: decouple vrate adjustment from surplus transfers
Budget donations are inaccurate and could take multiple periods to converge.
To prevent triggering vrate adjustments while surplus transfers were
catching up, vrate adjustment was suppressed if donations were increasing,
which was indicated by non-zero nr_surpluses.
This entangling won't be necessary with the scheduled rewrite of donation
mechanism which will make it precise and immediate. Let's decouple the two
in preparation.
blk-iocost: replace iocg->has_surplus with ->surplus_list
Instead of marking iocgs with surplus with a flag and filtering for them
while walking all active iocgs, build a surpluses list. This doesn't make
much difference now but will help implementing improved donation logic which
will iterate iocgs with surplus multiple times.
blk-iocost: calculate iocg->usages[] from iocg->local_stat.usage_us
Currently, iocg->usages[] which are used to guide inuse adjustments are
calculated from vtime deltas. This, however, assumes that the hierarchical
inuse weight at the time of calculation held for the entire period, which
often isn't true and can lead to significant errors.
Now that we have absolute usage information collected, we can derive
iocg->usages[] from iocg->local_stat.usage_us so that inuse adjustment
decisions are made based on actual absolute usage. The calculated usage is
clamped between 1 and WEIGHT_ONE and WEIGHT_ONE is also used to signal
saturation regardless of the current hierarchical inuse weight.
Currently, iocost doesn't collect or expose any statistics punting off all
monitoring duties to drgn based iocost_monitor.py. While it works for some
scenarios, there are some usability and data availability challenges. For
example, accurate per-cgroup usage information can't be tracked by vtime
progression at all and the number available in iocg->usages[] are really
short-term snapshots used for control heuristics with possibly significant
errors.
This patch implements per-cgroup absolute usage stat counter and exposes it
through io.stat along with the current vrate. Usage stat collection and
flushing employ the same method as cgroup rstat on the active iocg's and the
only hot path overhead is preemption toggling and adding to a percpu
counter.
Currently, debt handling requires only iocg->waitq.lock. In the future, we
want to adjust and propagate inuse changes depending on debt status. Let's
grab ioc->lock in debt handling paths in preparation.
* Because ioc->lock nests outside iocg->waitq.lock, the decision to grab
ioc->lock needs to be made before entering the critical sections.
* Add and use iocg_[un]lock() which handles the conditional double locking.
* Add @pay_debt to iocg_kick_waitq() so that debt payment happens only when
the caller grabbed both locks.
This patch is prepatory and the comments contain references to future
changes.
blk-iocost: streamline vtime margin and timer slack handling
The margin handling was pretty inconsistent.
* ioc->margin_us and ioc->inuse_margin_vtime were used as vtime margin
thresholds. However, the two are in different units with the former
requiring conversion to vtime on use.
* iocg_kick_waitq() was using a quarter of WAITQ_TIMER_MARGIN_PCT of
period_us as the timer slack - ~1.2%. While iocg_kick_delay() was using a
quarter of ioc->margin_us - ~12.5%. There aren't strong reasons to use
different values for the two.
This patch cleans up margin and timer slack handling:
* vtime margins are now recorded in ioc->margins.{min, max} on period
duration changes and used consistently.
* Timer slack is now 1% of period_us and recorded in ioc->timer_slack_ns and
used consistently for iocg_kick_waitq() and iocg_kick_delay().
The only functional change is shortening of timer slack. No meaningful
visible change is expected.
blk-iocost: make ioc_now->now and ioc->period_at 64bit
They are in microseconds and wrap in around 1.2 hours with u32. While
unlikely, confusions from wraparounds are still possible. We aren't saving
anything meaningful by keeping these u32. Let's make them u64.
blk-iocost: use WEIGHT_ONE based fixed point number for weights
To improve weight donations, we want to able to scale inuse with a greater
accuracy and down below 1. Let's make non-hierarchical weights to use
WEIGHT_ONE based fixed point numbers too like hierarchical ones.
blk-iocost: make iocg_kick_waitq() call iocg_kick_delay() after paying debt
iocg_kick_waitq() is the function which pays debt and iocg_kick_delay()
updates the actual delay status accordingly. If iocg_kick_delay() is not
called after iocg_kick_delay() updated debt, unnecessarily large delays can
be applied temporarily.
Let's make sure such conditions don't occur by making iocg_kick_waitq()
always call iocg_kick_delay() after paying debt.
blk-iocost: clamp inuse and skip noops in __propagate_weights()
__propagate_weights() currently expects the callers to clamp inuse within
[1, active], which is needlessly fragile. The inuse adjustment logic is
going to be revamped, in preparation, let's make __propagate_weights() clamp
inuse on entry.
Also, make it avoid weight updates altogether if neither active or inuse is
changed.
blk-iocost: rename propagate_active_weights() to propagate_weights()
It already propagates two weights - active and inuse - and there will be
another soon. Let's drop the confusing misnomers. Rename
[__]propagate_active_weights() to [__]propagate_weights() and
commit_active_weights() to commit_weights().
blk-iocost has been reading percpu stat counters from remote cpus which on
some archs can lead to torn reads in really rare occassions. Use local[64]_t
for those counters.
Baolin Wang [Fri, 28 Aug 2020 02:52:56 +0000 (10:52 +0800)]
block: Add a new helper to attempt to merge a bio
There are lots of duplicated code when trying to merge a bio from
plug list and sw queue, we can introduce a new helper to attempt
to merge a bio, which can simplify the blk_bio_list_merge()
and blk_attempt_plug_merge().
Reviewed-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Baolin Wang <baolin.wang@linux.alibaba.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
This comment was added before the multiqueue I/O scheduler framework
was introduced; multiqueue has support for I/O scheduling now, so this
obsolete comment can be removed.
Signed-off-by: Danny Lin <danny@kdrag0n.dev> Signed-off-by: Jens Axboe <axboe@kernel.dk>
Just duplicate a small amount of code in the low-level map into the bio
and copy to the bio routines, leading to much easier to follow and
maintain code, and better shared error handling.
Signed-off-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Jens Axboe <axboe@kernel.dk>
Open code __blk_rq_unmap_user in the two callers. Both never pass a NULL
bio, and one of them can use an existing local variable instead of the bio
flag.
Signed-off-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Jens Axboe <axboe@kernel.dk>
nvme: don't call revalidate_disk from nvme_set_queue_dying
In nvme_set_queue_dying we really just want to ensure the disk and bdev
sizes are set to zero. Going through revalidate_disk leads to a somewhat
arcance and complex callchain relying on special behavior in a few
places. Instead just lift the set_capacity directly to
nvme_set_queue_dying, and rename and move the nvme_mpath_update_disk_size
helper so that we can use it in nvme_set_queue_dying to propagate the
size to the bdev without detours.
Signed-off-by: Christoph Hellwig <hch@lst.de> Reviewed-by: Sagi Grimberg <sagi@grimberg.me> Reviewed-by: Hannes Reinecke <hare@suse.de> Signed-off-by: Jens Axboe <axboe@kernel.dk>
block: fix locking for struct block_device size updates
Two different callers use two different mutexes for updating the
block device size, which obviously doesn't help to actually protect
against concurrent updates from the different callers. In addition
one of the locks, bd_mutex is rather prone to deadlocks with other
parts of the block stack that use it for high level synchronization.
Switch to using a new spinlock protecting just the size updates, as
that is all we need, and make sure everyone does the update through
the proper helper.
This fixes a bug reported with the nvme revalidating disks during a
hot removal operation, which can currently deadlock on bd_mutex.
request_queue.rpm_status is assigned values of the rpm_status enum only,
so reflect that in its type.
Note that including <linux/pm.h> is (currently) a no-op, as it is
already included through <linux/genhd.h> and <linux/device.h>, but it is
better to play it safe.
Signed-off-by: Geert Uytterhoeven <geert+renesas@glider.be> Acked-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
* block-5.9:
blk-stat: make q->stats->lock irqsafe
blk-iocost: ioc_pd_free() shouldn't assume irq disabled
block: fix locking in bdev_del_partition
block: release disk reference in hd_struct_free_work
block: ensure bdi->io_pages is always initialized
nvme-pci: cancel nvme device request before disabling
nvme: only use power of two io boundaries
nvme: fix controller instance leak
nvmet-fc: Fix a missed _irqsave version of spin_lock in 'nvmet_fc_fod_op_done()'
nvme: Fix NULL dereference for pci nvme controllers
nvme-rdma: fix reset hang if controller died in the middle of a reset
nvme-rdma: fix timeout handler
nvme-rdma: serialize controller teardown sequences
nvme-tcp: fix reset hang if controller died in the middle of a reset
nvme-tcp: fix timeout handler
nvme-tcp: serialize controller teardown sequences
nvme: have nvme_wait_freeze_timeout return if it timed out
nvme-fabrics: don't check state NVME_CTRL_NEW for request acceptance
nvmet-tcp: Fix NULL dereference when a connect data comes in h2cdata pdu
blk-iocost calls blk_stat_enable_accounting() while holding an irqsafe lock
which triggers a lockdep splat because q->stats->lock isn't irqsafe. Let's
make it irqsafe.
ioc_pd_free() grabs irq-safe ioc->lock without ensuring that irq is disabled
when it can be called with irq disabled or enabled. This has a small chance
of causing A-A deadlocks and triggers lockdep splats. Use irqsave operations
instead.
We need to hold the whole device bd_mutex to protect against
other thread concurrently deleting out partition before we get
to it, and thus causing a use after free.
Fixes: bc74e398cfe3 ("block: pass a hd_struct to delete_partition") Reported-by: syzbot+6448f3c229bc52b82f69@syzkaller.appspotmail.com Signed-off-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Jens Axboe <axboe@kernel.dk>
Ming Lei [Tue, 1 Sep 2020 10:07:38 +0000 (18:07 +0800)]
block: release disk reference in hd_struct_free_work
Commit 04545fc8c1bf ("block: revert back to synchronous request_queue removal")
stops to release request queue from wq context because that commit
supposed all blk_put_queue() is called in context which is allowed
to sleep. However, this assumption isn't true because we release disk's
reference in partition's percpu_ref's ->release() which doesn't allow
to sleep, because the ->release() is run via call_rcu().
Fixes this issue by moving put disk reference into hd_struct_free_work()
Fixes: 04545fc8c1bf ("block: revert back to synchronous request_queue removal") Reported-by: Ilya Dryomov <idryomov@gmail.com> Signed-off-by: Ming Lei <ming.lei@redhat.com> Tested-by: Ilya Dryomov <idryomov@gmail.com> Reviewed-by: Christoph Hellwig <hch@lst.de> Cc: Luis Chamberlain <mcgrof@kernel.org> Cc: Christoph Hellwig <hch@lst.de> Cc: Bart Van Assche <bvanassche@acm.org> Signed-off-by: Jens Axboe <axboe@kernel.dk>
Jens Axboe [Mon, 31 Aug 2020 17:20:02 +0000 (11:20 -0600)]
block: ensure bdi->io_pages is always initialized
If a driver leaves the limit settings as the defaults, then we don't
initialize bdi->io_pages. This means that file systems may need to
work around bdi->io_pages == 0, which is somewhat messy.
Initialize the default value just like we do for ->ra_pages.
Cc: stable@vger.kernel.org Fixes: 48f03e119c90 ("mm: don't cap request size based on read-ahead setting") Reported-by: OGAWA Hirofumi <hirofumi@mail.parknet.co.jp> Reviewed-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Jens Axboe <axboe@kernel.dk>
Linus Torvalds [Sun, 30 Aug 2020 22:53:44 +0000 (15:53 -0700)]
Merge branch 'linus' of git://git.kernel.org/pub/scm/linux/kernel/git/herbert/crypto-2.6
Pull crypto fixes from Herbert Xu:
- fix regression in af_alg that affects iwd
- restore polling delay in qat
- fix double free in ingenic on error path
- fix potential build failure in sa2ul due to missing Kconfig dependency
* 'linus' of git://git.kernel.org/pub/scm/linux/kernel/git/herbert/crypto-2.6:
crypto: af_alg - Work around empty control messages without MSG_MORE
crypto: sa2ul - add Kconfig selects to fix build error
crypto: ingenic - Drop kfree for memory allocated with devm_kzalloc
crypto: qat - add delay before polling mailbox
Linus Torvalds [Sun, 30 Aug 2020 19:01:23 +0000 (12:01 -0700)]
Merge tag 'x86-urgent-2020-08-30' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip
Pull x86 fixes from Thomas Gleixner:
"Three interrupt related fixes for X86:
- Move disabling of the local APIC after invoking fixup_irqs() to
ensure that interrupts which are incoming are noted in the IRR and
not ignored.
- Unbreak affinity setting.
The rework of the entry code reused the regular exception entry
code for device interrupts. The vector number is pushed into the
errorcode slot on the stack which is then lifted into an argument
and set to -1 because that's regs->orig_ax which is used in quite
some places to check whether the entry came from a syscall.
But it was overlooked that orig_ax is used in the affinity cleanup
code to validate whether the interrupt has arrived on the new
target. It turned out that this vector check is pointless because
interrupts are never moved from one vector to another on the same
CPU. That check is a historical leftover from the time where x86
supported multi-CPU affinities, but not longer needed with the now
strict single CPU affinity. Famous last words ...
- Add a missing check for an empty cpumask into the matrix allocator.
The affinity change added a warning to catch the case where an
interrupt is moved on the same CPU to a different vector. This
triggers because a condition with an empty cpumask returns an
assignment from the allocator as the allocator uses for_each_cpu()
without checking the cpumask for being empty. The historical
inconsistent for_each_cpu() behaviour of ignoring the cpumask and
unconditionally claiming that CPU0 is in the mask struck again.
Sigh.
plus a new entry into the MAINTAINER file for the HPE/UV platform"
* tag 'x86-urgent-2020-08-30' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
genirq/matrix: Deal with the sillyness of for_each_cpu() on UP
x86/irq: Unbreak interrupt affinity setting
x86/hotplug: Silence APIC only after all interrupts are migrated
MAINTAINERS: Add entry for HPE Superdome Flex (UV) maintainers
Linus Torvalds [Sun, 30 Aug 2020 18:56:54 +0000 (11:56 -0700)]
Merge tag 'irq-urgent-2020-08-30' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip
Pull irq fixes from Thomas Gleixner:
"A set of fixes for interrupt chip drivers:
- Revert the platform driver conversion of interrupt chip drivers as
it turned out to create more problems than it solves.
- Fix a trivial typo in the new module helpers which made probing
reliably fail.
- Small fixes in the STM32 and MIPS Ingenic drivers
- The TI firmware rework which had badly managed dependencies and had
to wait post rc1"
* tag 'irq-urgent-2020-08-30' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
irqchip/ingenic: Leave parent IRQ unmasked on suspend
irqchip/stm32-exti: Avoid losing interrupts due to clearing pending bits by mistake
irqchip: Revert modular support for drivers using IRQCHIP_PLATFORM_DRIVER helperse
irqchip: Fix probing deferal when using IRQCHIP_PLATFORM_DRIVER helpers
arm64: dts: k3-am65: Update the RM resource types
arm64: dts: k3-am65: ti-sci-inta/intr: Update to latest bindings
arm64: dts: k3-j721e: ti-sci-inta/intr: Update to latest bindings
irqchip/ti-sci-inta: Add support for INTA directly connecting to GIC
irqchip/ti-sci-inta: Do not store TISCI device id in platform device id field
dt-bindings: irqchip: Convert ti, sci-inta bindings to yaml
dt-bindings: irqchip: ti, sci-inta: Update docs to support different parent.
irqchip/ti-sci-intr: Add support for INTR being a parent to INTR
dt-bindings: irqchip: Convert ti, sci-intr bindings to yaml
dt-bindings: irqchip: ti, sci-intr: Update bindings to drop the usage of gic as parent
firmware: ti_sci: Add support for getting resource with subtype
firmware: ti_sci: Drop unused structure ti_sci_rm_type_map
firmware: ti_sci: Drop the device id to resource type translation
Linus Torvalds [Sun, 30 Aug 2020 18:50:53 +0000 (11:50 -0700)]
Merge tag 'sched-urgent-2020-08-30' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip
Pull scheduler fix from Thomas Gleixner:
"A single fix for the scheduler:
- Make is_idle_task() __always_inline to prevent the compiler from
putting it out of line into the wrong section because it's used
inside noinstr sections"
* tag 'sched-urgent-2020-08-30' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
sched: Use __always_inline on is_idle_task()
Linus Torvalds [Sun, 30 Aug 2020 17:56:12 +0000 (10:56 -0700)]
Merge tag 'powerpc-5.9-4' of git://git.kernel.org/pub/scm/linux/kernel/git/powerpc/linux
Pull powerpc fixes from Michael Ellerman:
- Revert our removal of PROT_SAO, at least one user expressed an
interest in using it on Power9. Instead don't allow it to be used in
guests unless enabled explicitly at compile time.
- A fix for a crash introduced by a recent change to FP handling.
- Revert a change to our idle code that left Power10 with no idle
support.
- One minor fix for the new scv system call path to set PPR.
- Fix a crash in our "generic" PMU if branch stack events were enabled.
- A fix for the IMC PMU, to correctly identify host kernel samples.
- The ADB_PMU powermac code was found to be incompatible with
VMAP_STACK, so make them incompatible in Kconfig until the code can
be fixed.
- A build fix in drivers/video/fbdev/controlfb.c, and a documentation
fix.
Thanks to Alexey Kardashevskiy, Athira Rajeev, Christophe Leroy,
Giuseppe Sacco, Madhavan Srinivasan, Milton Miller, Nicholas Piggin,
Pratik Rajesh Sampat, Randy Dunlap, Shawn Anastasio, Vaidyanathan
Srinivasan.
* tag 'powerpc-5.9-4' of git://git.kernel.org/pub/scm/linux/kernel/git/powerpc/linux:
powerpc/32s: Disable VMAP stack which CONFIG_ADB_PMU
Revert "powerpc/powernv/idle: Replace CPU feature check with PVR check"
powerpc/perf: Fix reading of MSR[HV/PR] bits in trace-imc
powerpc/perf: Fix crashes with generic_compat_pmu & BHRB
powerpc/64s: Fix crash in load_fp_state() due to fpexc_mode
powerpc/64s: scv entry should set PPR
Documentation/powerpc: fix malformed table in syscall64-abi
video: fbdev: controlfb: Fix build for COMPILE_TEST=y && PPC_PMAC=n
selftests/powerpc: Update PROT_SAO test to skip ISA 3.1
powerpc/64s: Disallow PROT_SAO in LPARs by default
Revert "powerpc/64s: Remove PROT_SAO support"
Linus Torvalds [Sun, 30 Aug 2020 17:51:03 +0000 (10:51 -0700)]
Merge tag 'usb-5.9-rc3' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/usb
Pull USB fixes from Greg KH:
"Let's try this again... Here are some USB fixes for 5.9-rc3.
This differs from the previous pull request for this release in that
the usb gadget patch now does not break some systems, and actually
does what it was intended to do. Many thanks to Marek Szyprowski for
quickly noticing and testing the patch from Andy Shevchenko to resolve
this issue.
Additionally, some more new USB quirks have been added to get some new
devices to work properly based on user reports.
Other than that, the patches are all here, and they contain:
- usb gadget driver fixes
- xhci driver fixes
- typec fixes
- new quirks and ids
- fixes for USB patches that went into 5.9-rc1.
All of these have been tested in linux-next with no reported issues"
* tag 'usb-5.9-rc3' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/usb: (33 commits)
usb: storage: Add unusual_uas entry for Sony PSZ drives
USB: Ignore UAS for JMicron JMS567 ATA/ATAPI Bridge
usb: host: ohci-exynos: Fix error handling in exynos_ohci_probe()
USB: gadget: u_f: Unbreak offset calculation in VLAs
USB: quirks: Ignore duplicate endpoint on Sound Devices MixPre-D
usb: typec: tcpm: Fix Fix source hard reset response for TDA 2.3.1.1 and TDA 2.3.1.2 failures
USB: PHY: JZ4770: Fix static checker warning.
USB: gadget: f_ncm: add bounds checks to ncm_unwrap_ntb()
USB: gadget: u_f: add overflow checks to VLA macros
xhci: Always restore EP_SOFT_CLEAR_TOGGLE even if ep reset failed
xhci: Do warm-reset when both CAS and XDEV_RESUME are set
usb: host: xhci: fix ep context print mismatch in debugfs
usb: uas: Add quirk for PNY Pro Elite
tools: usb: move to tools buildsystem
USB: Fix device driver race
USB: Also match device drivers using the ->match vfunc
usb: host: xhci-tegra: fix tegra_xusb_get_phy()
usb: host: xhci-tegra: otg usb2/usb3 port init
usb: hcd: Fix use after free in usb_hcd_pci_remove()
usb: typec: ucsi: Hold con->lock for the entire duration of ucsi_register_port()
...
Linus Torvalds [Sun, 30 Aug 2020 17:47:23 +0000 (10:47 -0700)]
Merge tag 'edac_urgent_for_v5.9_rc3' of git://git.kernel.org/pub/scm/linux/kernel/git/ras/ras
Pull EDAC fix from Borislav Petkov:
"A fix to properly clear ghes_edac driver state on driver remove so
that a subsequent load can probe the system properly (Shiju Jose)"
* tag 'edac_urgent_for_v5.9_rc3' of git://git.kernel.org/pub/scm/linux/kernel/git/ras/ras:
EDAC/ghes: Fix NULL pointer dereference in ghes_edac_register()
projects / kernel.git / log