Add an rq private RQF_ELV flag, which tells the block layer that this
request was initialized on a queue that has an IO scheduler attached.
This allows for faster checking in the fast path, rather than having to
deference rq->q later on.

Elevator switching does full quiesce of the queue before detaching an
IO scheduler, so it's safe to cache this in the request itself.
Signed-off-by: NJens Axboe <axboe@kernel.dk>

It's been a while since this was analyzed, move some members around to
better flow with the use case. Initial state up top, and queued state
after that. This improves my peak case by about 1.5%, from 7750K to
7900K IOPS.
Reviewed-by: NChristoph Hellwig <hch@lst.de>
Signed-off-by: NJens Axboe <axboe@kernel.dk>

Replace the blk_poll interface that requires the caller to keep a queue
and cookie from the submissions with polling based on the bio.

Polling for the bio itself leads to a few advantages:

 - the cookie construction can made entirely private in blk-mq.c
 - the caller does not need to remember the request_queue and cookie
   separately and thus sidesteps their lifetime issues
 - keeping the device and the cookie inside the bio allows to trivially
   support polling BIOs remapping by stacking drivers
 - a lot of code to propagate the cookie back up the submission path can
   be removed entirely.
Signed-off-by: NChristoph Hellwig <hch@lst.de>
Tested-by: NMark Wunderlich <mark.wunderlich@intel.com>
Link: https://lore.kernel.org/r/20211012111226.760968-15-hch@lst.deSigned-off-by: NJens Axboe <axboe@kernel.dk>

Fold bio_cur_bytes into the only caller.
Signed-off-by: NChristoph Hellwig <hch@lst.de>
Link: https://lore.kernel.org/r/20211012161804.991559-5-hch@lst.deSigned-off-by: NJens 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: NJens Axboe <axboe@kernel.dk>

Now that shared sbitmap support really means shared tags, rename symbols
to match that.
Signed-off-by: NJohn Garry <john.garry@huawei.com>
Link: https://lore.kernel.org/r/1633429419-228500-15-git-send-email-john.garry@huawei.comSigned-off-by: NJens 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: NJohn Garry <john.garry@huawei.com>
Reviewed-by: NMing Lei <ming.lei@redhat.com>
Link: https://lore.kernel.org/r/1633429419-228500-13-git-send-email-john.garry@huawei.comSigned-off-by: NJens Axboe <axboe@kernel.dk>

It is a bit confusing that there is BLKDEV_MAX_RQ and MAX_SCHED_RQ, as
the name BLKDEV_MAX_RQ would imply the max requests always, which it is
not.

Rename to BLKDEV_MAX_RQ to BLKDEV_DEFAULT_RQ, matching its usage - that being
the default number of requests assigned when allocating a request queue.
Signed-off-by: NJohn Garry <john.garry@huawei.com>
Reviewed-by: NMing Lei <ming.lei@redhat.com>
Reviewed-by: NHannes Reinecke <hare@suse.de>
Link: https://lore.kernel.org/r/1633429419-228500-3-git-send-email-john.garry@huawei.comSigned-off-by: NJens Axboe <axboe@kernel.dk>

struct request is only used by blk-mq drivers, so move it and all
related declarations to blk-mq.h.
Signed-off-by: NChristoph Hellwig <hch@lst.de>
Reviewed-by: NJohannes Thumshirn <johannes.thumshirn@wdc.com>
Link: https://lore.kernel.org/r/20210920123328.1399408-18-hch@lst.deSigned-off-by: NJens Axboe <axboe@kernel.dk>

Pass the lockdep name to the low-level __blk_alloc_disk helper and
hardcode the name for it given that the number of minors or node_id
are not very useful information.  While this passes a pointless
argument for non-lockdep builds that is not really an issue as
disk allocation is a probe time only slow path.
Signed-off-by: NChristoph Hellwig <hch@lst.de>
Link: https://lore.kernel.org/r/20210816131910.615153-5-hch@lst.deSigned-off-by: NJens Axboe <axboe@kernel.dk>

elevator_get_default() uses the following algorithm to select an I/O
scheduler from inside add_disk():
- In case of a single hardware queue or if sharing hardware queues across
  multiple request queues (BLK_MQ_F_TAG_HCTX_SHARED), use mq-deadline.
- Otherwise, use 'none'.

This is a good choice for most but not for all block drivers. Make it
possible to override the selection of mq-deadline with a new flag,
namely BLK_MQ_F_NO_SCHED_BY_DEFAULT.

Cc: Christoph Hellwig <hch@lst.de>
Cc: Ming Lei <ming.lei@redhat.com>
Cc: Tetsuo Handa <penguin-kernel@I-love.SAKURA.ne.jp>
Cc: Martijn Coenen <maco@android.com>
Cc: Jaegeuk Kim <jaegeuk@kernel.org>
Signed-off-by: NBart Van Assche <bvanassche@acm.org>
Link: https://lore.kernel.org/r/20210805174200.3250718-2-bvanassche@acm.orgSigned-off-by: NJens Axboe <axboe@kernel.dk>

All driver uses are gone now.
Signed-off-by: NChristoph Hellwig <hch@lst.de>
Reviewed-by: NBart Van Assche <bvanassche@acm.org>
Link: https://lore.kernel.org/r/20210624081012.256464-1-hch@lst.deSigned-off-by: NJens Axboe <axboe@kernel.dk>

The __blk_mq_alloc_disk() function doesn't return NULLs it returns
error pointers.

Fixes: b461dfc4 ("blk-mq: add the blk_mq_alloc_disk APIs")
Signed-off-by: NDan Carpenter <dan.carpenter@oracle.com>
Reviewed-by: NChristoph Hellwig <hch@lst.de>
Link: https://lore.kernel.org/r/YMyjci35WBqrtqG+@mwandaSigned-off-by: NJens Axboe <axboe@kernel.dk>

All users are gone now.
Signed-off-by: NChristoph Hellwig <hch@lst.de>
Reviewed-by: NChaitanya Kulkarni <chaitanya.kulkarni@wdc.com>
Link: https://lore.kernel.org/r/20210602065345.355274-16-hch@lst.deSigned-off-by: NJens Axboe <axboe@kernel.dk>

Add a new API to allocate a gendisk including the request_queue for use
with blk-mq based drivers.  This is to avoid boilerplate code in drivers.
Signed-off-by: NChristoph Hellwig <hch@lst.de>
Reviewed-by: NChaitanya Kulkarni <chaitanya.kulkarni@wdc.com>
Link: https://lore.kernel.org/r/20210602065345.355274-4-hch@lst.deSigned-off-by: NJens Axboe <axboe@kernel.dk>

Don't return the passed in request_queue but a normal error code, and
drop the elevator_init argument in favor of just calling elevator_init_mq
directly from dm-rq.
Signed-off-by: NChristoph Hellwig <hch@lst.de>
Reviewed-by: NChaitanya Kulkarni <chaitanya.kulkarni@wdc.com>
Link: https://lore.kernel.org/r/20210602065345.355274-3-hch@lst.deSigned-off-by: NJens Axboe <axboe@kernel.dk>

Factour out a helper to initialize a simple single hw queue tag_set from
blk_mq_init_sq_queue.  This will allow to phase out blk_mq_init_sq_queue
in favor of a more symmetric and general API.
Signed-off-by: NChristoph Hellwig <hch@lst.de>
Reviewed-by: NChaitanya Kulkarni <chaitanya.kulkarni@wdc.com>
Link: https://lore.kernel.org/r/20210602065345.355274-2-hch@lst.deSigned-off-by: NJens Axboe <axboe@kernel.dk>

Fixes the following warning when running 'make htmldocs':

  include/linux/blk-mq.h:395: warning: Function parameter or member
  'set_rq_budget_token' not described in 'blk_mq_ops'
  include/linux/blk-mq.h:395: warning: Function parameter or member
  'get_rq_budget_token' not described in 'blk_mq_ops'

[mkp: added warning messages]

Link: https://lore.kernel.org/r/20210421154526.1954174-1-ming.lei@redhat.com
Fixes: d022d18c ("scsi: blk-mq: Add callbacks for storing & retrieving budget token")
Reported-by: NStephen Rothwell <sfr@canb.auug.org.au>
Signed-off-by: NMing Lei <ming.lei@redhat.com>
Signed-off-by: NMartin K. Petersen <martin.petersen@oracle.com>

SCSI uses a global atomic variable to track queue depth for each
LUN/request queue.

This doesn't scale well when there are lots of CPU cores and the disk is
very fast. It has been observed that IOPS is affected a lot by tracking
queue depth via sdev->device_busy in the I/O path.

Return budget token from .get_budget callback. The budget token can be
passed to driver so that we can replace the atomic variable with
sbitmap_queue and alleviate the scaling problems that way.

Link: https://lore.kernel.org/r/20210122023317.687987-9-ming.lei@redhat.com
Cc: Omar Sandoval <osandov@fb.com>
Cc: Kashyap Desai <kashyap.desai@broadcom.com>
Cc: Sumanesh Samanta <sumanesh.samanta@broadcom.com>
Cc: Ewan D. Milne <emilne@redhat.com>
Tested-by: NSumanesh Samanta <sumanesh.samanta@broadcom.com>
Reviewed-by: NHannes Reinecke <hare@suse.de>
Signed-off-by: NMing Lei <ming.lei@redhat.com>
Signed-off-by: NMartin K. Petersen <martin.petersen@oracle.com>

Since SCSI is the only driver which requires dispatch budget move the token
from struct request to struct scsi_cmnd.

Link: https://lore.kernel.org/r/20210122023317.687987-8-ming.lei@redhat.com
Cc: Omar Sandoval <osandov@fb.com>
Cc: Kashyap Desai <kashyap.desai@broadcom.com>
Cc: Sumanesh Samanta <sumanesh.samanta@broadcom.com>
Cc: Ewan D. Milne <emilne@redhat.com>
Cc: Hannes Reinecke <hare@suse.de>
Tested-by: NSumanesh Samanta <sumanesh.samanta@broadcom.com>
Reviewed-by: NHannes Reinecke <hare@suse.de>
Signed-off-by: NMing Lei <ming.lei@redhat.com>
Signed-off-by: NMartin K. Petersen <martin.petersen@oracle.com>

nvme drivers need to set the state of request to MQ_RQ_COMPLETE when
directly complete request in queue_rq.
So add blk_mq_set_request_complete.
Signed-off-by: NChao Leng <lengchao@huawei.com>
Signed-off-by: NChristoph Hellwig <hch@lst.de>

This reverts commit b445547e.

Since both mq-deadline and BFQ completely ignore hctx they are passed to
their dispatch function and dispatch whatever request they deem fit
checking whether any request for a particular hctx is queued is just
pointless since we'll very likely get a request from a different hctx
anyway. In the following commit we'll deal with lock contention in these
IO schedulers in presence of multiple HW queues in a different way.
Signed-off-by: NJan Kara <jack@suse.cz>
Reviewed-by: NMing Lei <ming.lei@redhat.com>
Signed-off-by: NJens Axboe <axboe@kernel.dk>

Replace the gendisk pointer in struct bio with a pointer to the newly
improved struct block device.  From that the gendisk can be trivially
accessed with an extra indirection, but it also allows to directly
look up all information related to partition remapping.
Signed-off-by: NChristoph Hellwig <hch@lst.de>
Acked-by: NTejun Heo <tj@kernel.org>
Signed-off-by: NJens Axboe <axboe@kernel.dk>

Remove flag RQF_PREEMPT and BLK_MQ_REQ_PREEMPT since these are no longer
used by any kernel code.

Link: https://lore.kernel.org/r/20201209052951.16136-8-bvanassche@acm.org
Cc: Can Guo <cang@codeaurora.org>
Cc: Stanley Chu <stanley.chu@mediatek.com>
Cc: Alan Stern <stern@rowland.harvard.edu>
Cc: Ming Lei <ming.lei@redhat.com>
Cc: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Cc: Martin Kepplinger <martin.kepplinger@puri.sm>
Reviewed-by: NChristoph Hellwig <hch@lst.de>
Reviewed-by: NHannes Reinecke <hare@suse.de>
Reviewed-by: NJens Axboe <axboe@kernel.dk>
Reviewed-by: NCan Guo <cang@codeaurora.org>
Signed-off-by: NBart Van Assche <bvanassche@acm.org>
Signed-off-by: NMartin K. Petersen <martin.petersen@oracle.com>

Introduce the BLK_MQ_REQ_PM flag. This flag makes the request allocation
functions set RQF_PM. This is the first step towards removing
BLK_MQ_REQ_PREEMPT.

Link: https://lore.kernel.org/r/20201209052951.16136-3-bvanassche@acm.org
Cc: Alan Stern <stern@rowland.harvard.edu>
Cc: Stanley Chu <stanley.chu@mediatek.com>
Cc: Ming Lei <ming.lei@redhat.com>
Cc: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Cc: Can Guo <cang@codeaurora.org>
Reviewed-by: NChristoph Hellwig <hch@lst.de>
Reviewed-by: NHannes Reinecke <hare@suse.de>
Reviewed-by: NJens Axboe <axboe@kernel.dk>
Reviewed-by: NCan Guo <cang@codeaurora.org>
Signed-off-by: NBart Van Assche <bvanassche@acm.org>
Signed-off-by: NMartin K. Petersen <martin.petersen@oracle.com>

flush_end_io() may be called recursively from some driver, such as
nvme-loop, so lockdep may complain 'possible recursive locking'.
Commit b3c6a599("block: Fix a lockdep complaint triggered by
request queue flushing") tried to address this issue by assigning
dynamically allocated per-flush-queue lock class. This solution
adds synchronize_rcu() for each hctx's release handler, and causes
horrible SCSI MQ probe delay(more than half an hour on megaraid sas).

Add new API of blk_mq_hctx_set_fq_lock_class() for these drivers, so
we just need to use driver specific lock class for avoiding the
lockdep warning of 'possible recursive locking'.
Tested-by: NKashyap Desai <kashyap.desai@broadcom.com>
Reported-by: NQian Cai <cai@redhat.com>
Cc: Sumit Saxena <sumit.saxena@broadcom.com>
Cc: John Garry <john.garry@huawei.com>
Cc: Kashyap Desai <kashyap.desai@broadcom.com>
Cc: Bart Van Assche <bvanassche@acm.org>
Cc: Hannes Reinecke <hare@suse.de>
Signed-off-by: NMing Lei <ming.lei@redhat.com>
Reviewed-by: NHannes Reinecke <hare@suse.de>
Signed-off-by: NJens Axboe <axboe@kernel.dk>

This is a preparation patch to have minimal block layer request bio
append functionality in the context of the NVMeOF Passthru driver which
falls in the fast path and doesn't need calls from blk_rq_append_bio().
Signed-off-by: NChaitanya Kulkarni <chaitanya.kulkarni@wdc.com>
Reviewed-by: NLogan Gunthorpe <logang@deltatee.com>
Signed-off-by: NChristoph Hellwig <hch@lst.de>

As reported by kernel-doc:
	./include/linux/blk-mq.h:267: warning: Function parameter or member 'active_queues_shared_sbitmap' not described in 'blk_mq_tag_set'

There is now a new member for struct blk_mq_tag_set. Add a
description for it, based on the commit that introduced it.

Fixes: f1b49fdc ("blk-mq: Record active_queues_shared_sbitmap per tag_set for when using shared sbitmap")
Signed-off-by: NMauro Carvalho Chehab <mchehab+huawei@kernel.org>
Reviewed-by: NJens Axboe <axboe@kernel.dk>
Reviewed-by: NJohn Garry <john.garry@huawei.com>
Link: https://lore.kernel.org/r/8e513153b83eefc05e358f51f2632b592c3f6772.1603791716.git.mchehab+huawei@kernel.orgSigned-off-by: NJonathan Corbet <corbet@lwn.net>

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: NKashyap Desai <kashyap.desai@broadcom.com>
Signed-off-by: NHannes Reinecke <hare@suse.de>
Signed-off-by: NJohn Garry <john.garry@huawei.com>
Tested-by: NDouglas Gilbert <dgilbert@interlog.com>
Signed-off-by: NJens Axboe <axboe@kernel.dk>

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: NJohn Garry <john.garry@huawei.com>
Tested-by: Don Brace<don.brace@microsemi.com> #SCSI resv cmds patches used
Tested-by: NDouglas Gilbert <dgilbert@interlog.com>
Signed-off-by: NJens Axboe <axboe@kernel.dk>

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 bf0beec0
("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 6eb045e0 ("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].

[0] https://lore.kernel.org/linux-block/alpine.DEB.2.21.1904051331270.1802@nanos.tec.linutronix.de/
[1] https://lore.kernel.org/linux-block/20190531022801.10003-1-ming.lei@redhat.com/
[2] https://lore.kernel.org/linux-block/ff77beff-5fd9-9f05-12b6-826922bace1f@huawei.com/T/#m3db0a602f095cbcbff27e9c884d6b4ae826144beSigned-off-by: NJohn Garry <john.garry@huawei.com>
Tested-by: Don Brace<don.brace@microsemi.com> #SCSI resv cmds patches used
Tested-by: NDouglas Gilbert <dgilbert@interlog.com>
Signed-off-by: NJens Axboe <axboe@kernel.dk>

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: NBart Van Assche <bvanassche@acm.org>
Signed-off-by: NMing Lei <ming.lei@redhat.com>
Signed-off-by: NJohn Garry <john.garry@huawei.com>
Tested-by: NDouglas Gilbert <dgilbert@interlog.com>
Reviewed-by: NHannes Reinecke <hare@suse.de>
Signed-off-by: NJens Axboe <axboe@kernel.dk>

Move the blk_mq_bio_list_merge() into blk-merge.c and
rename it as a generic name.
Reviewed-by: NChristoph Hellwig <hch@lst.de>
Signed-off-by: NBaolin Wang <baolin.wang@linux.alibaba.com>
Signed-off-by: NJens Axboe <axboe@kernel.dk>

No need to define typedefs for the callbacks, because there is not a
single user except blk_mq_ops.
Signed-off-by: NDaniel Wagner <dwagner@suse.de>
Signed-off-by: NJens Axboe <axboe@kernel.dk>

The make_request_fn is a little weird in that it sits directly in
struct request_queue instead of an operation vector.  Replace it with
a block_device_operations method called submit_bio (which describes much
better what it does).  Also remove the request_queue argument to it, as
the queue can be derived pretty trivially from the bio.
Signed-off-by: NChristoph Hellwig <hch@lst.de>
Signed-off-by: NJens Axboe <axboe@kernel.dk>

blk-mq budget is abstract from scsi's device queue depth, and it is
always per-request-queue instead of hctx.

It can be quite absurd to get a budget from one hctx, then dequeue a
request from scheduler queue, and this request may not belong to this
hctx, at least for bfq and deadline.

So fix the mess and always pass request queue to get/put budget
callback.
Signed-off-by: NMing Lei <ming.lei@redhat.com>
Tested-by: NBaolin Wang <baolin.wang7@gmail.com>
Reviewed-by: NJohannes Thumshirn <johannes.thumshirn@wdc.com>
Reviewed-by: NChristoph Hellwig <hch@lst.de>
Reviewed-by: NDouglas Anderson <dianders@chromium.org>
Reviewed-by: NSagi Grimberg <sagi@grimberg.me>
Cc: Sagi Grimberg <sagi@grimberg.me>
Cc: Baolin Wang <baolin.wang7@gmail.com>
Cc: Christoph Hellwig <hch@infradead.org>
Cc: Douglas Anderson <dianders@chromium.org>
Signed-off-by: NJens Axboe <axboe@kernel.dk>

Just check for a non-NULL elevator directly to make the code more clear.
Signed-off-by: NChristoph Hellwig <hch@lst.de>
Signed-off-by: NJens Axboe <axboe@kernel.dk>

This is a variant of blk_mq_complete_request_remote that only completes
the request if it needs to be bounced to another CPU or a softirq.  If
the request can be completed locally the function returns false and lets
the driver complete it without requring and indirect function call.
Reviewed-by: NDaniel Wagner <dwagner@suse.de>
Signed-off-by: NChristoph Hellwig <hch@lst.de>
Signed-off-by: NJens Axboe <axboe@kernel.dk>

Move the call to blk_should_fake_timeout out of blk_mq_complete_request
and into the drivers, skipping call sites that are obvious error
handlers, and remove the now superflous blk_mq_force_complete_rq helper.
This ensures we don't keep injecting errors into completions that just
terminate the Linux request after the hardware has been reset or the
command has been aborted.
Reviewed-by: NDaniel Wagner <dwagner@suse.de>
Signed-off-by: NChristoph Hellwig <hch@lst.de>
Signed-off-by: NJens Axboe <axboe@kernel.dk>

Most of blk-mq drivers depend on managed IRQ's auto-affinity to setup
up queue mapping. Thomas mentioned the following point[1]:

"That was the constraint of managed interrupts from the very beginning:

 The driver/subsystem has to quiesce the interrupt line and the associated
 queue _before_ it gets shutdown in CPU unplug and not fiddle with it
 until it's restarted by the core when the CPU is plugged in again."

However, current blk-mq implementation doesn't quiesce hw queue before
the last CPU in the hctx is shutdown.  Even worse, CPUHP_BLK_MQ_DEAD is a
cpuhp state handled after the CPU is down, so there isn't any chance to
quiesce the hctx before shutting down the CPU.

Add new CPUHP_AP_BLK_MQ_ONLINE state to stop allocating from blk-mq hctxs
where the last CPU goes away, and wait for completion of in-flight
requests.  This guarantees that there is no inflight I/O before shutting
down the managed IRQ.

Add a BLK_MQ_F_STACKING and set it for dm-rq and loop, so we don't need
to wait for completion of in-flight requests from these drivers to avoid
a potential dead-lock. It is safe to do this for stacking drivers as those
do not use interrupts at all and their I/O completions are triggered by
underlying devices I/O completion.

[1] https://lore.kernel.org/linux-block/alpine.DEB.2.21.1904051331270.1802@nanos.tec.linutronix.de/

[hch: different retry mechanism, merged two patches, minor cleanups]
Signed-off-by: NMing Lei <ming.lei@redhat.com>
Signed-off-by: NChristoph Hellwig <hch@lst.de>
Reviewed-by: NHannes Reinecke <hare@suse.de>
Reviewed-by: NDaniel Wagner <dwagner@suse.de>
Signed-off-by: NJens Axboe <axboe@kernel.dk>