block: extend functionality to map bvec iterator

Extend blk_rq_map_user_iov so that it can handle bvec iterator, using
the new blk_rq_map_user_bvec function. It maps the pages from bvec
iterator into a bio and place the bio into request.

This helper will be used by nvme for uring-passthrough path when IO is
done using pre-mapped buffers.

Signed-off-by: Kanchan Joshi <joshi.k@samsung.com>
Signed-off-by: Anuj Gupta <anuj20.g@samsung.com>
Suggested-by: Christoph Hellwig <hch@lst.de>
Link: https://lore.kernel.org/r/20220930062749.152261-11-anuj20.g@samsung.com
Signed-off-by: Jens Axboe <axboe@kernel.dk>

block: factor out blk_rq_map_bio_alloc helper

Move bio allocation logic from bio_map_user_iov to a new helper
blk_rq_map_bio_alloc. It is named so because functionality is opposite
of what is done inside blk_mq_map_bio_put. This is a prep patch.

Signed-off-by: Kanchan Joshi <joshi.k@samsung.com>
Link: https://lore.kernel.org/r/20220930062749.152261-10-anuj20.g@samsung.com
Signed-off-by: Jens Axboe <axboe@kernel.dk>

block: rename bio_map_put to blk_mq_map_bio_put

This patch renames existing bio_map_put function to blk_mq_map_bio_put.

Signed-off-by: Anuj Gupta <anuj20.g@samsung.com>
Suggested-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Link: https://lore.kernel.org/r/20220930062749.152261-9-anuj20.g@samsung.com
Signed-off-by: Jens Axboe <axboe@kernel.dk>

nvme: refactor nvme_alloc_request

nvme_alloc_request expects a large number of parameters.
Split this out into two functions to reduce number of parameters.
First one retains the name nvme_alloc_request, while second one is
named nvme_map_user_request.

Signed-off-by: Kanchan Joshi <joshi.k@samsung.com>
Signed-off-by: Anuj Gupta <anuj20.g@samsung.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Link: https://lore.kernel.org/r/20220930062749.152261-8-anuj20.g@samsung.com
Signed-off-by: Jens Axboe <axboe@kernel.dk>

nvme: refactor nvme_add_user_metadata

Pass struct request rather than bio. It helps to kill a parameter, and
some processing clean-up too.

Signed-off-by: Kanchan Joshi <joshi.k@samsung.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Link: https://lore.kernel.org/r/20220930062749.152261-7-anuj20.g@samsung.com
Signed-off-by: Jens Axboe <axboe@kernel.dk>

nvme: Use blk_rq_map_user_io helper

User blk_rq_map_user_io instead of duplicating the same code at
different places

Signed-off-by: Anuj Gupta <anuj20.g@samsung.com>
Link: https://lore.kernel.org/r/20220930062749.152261-6-anuj20.g@samsung.com
Signed-off-by: Jens Axboe <axboe@kernel.dk>

scsi: Use blk_rq_map_user_io helper

Use the new blk_rq_map_user_io helper instead of duplicating code at
various places. Additionally this also takes advantage of the on-stack
iov fast path.

Signed-off-by: Anuj Gupta <anuj20.g@samsung.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Link: https://lore.kernel.org/r/20220930062749.152261-5-anuj20.g@samsung.com
Signed-off-by: Jens Axboe <axboe@kernel.dk>

block: add blk_rq_map_user_io

Create a helper blk_rq_map_user_io for mapping of vectored as well as
non-vectored requests. This will help in saving dupilcation of code at few
places in scsi and nvme.

Signed-off-by: Anuj Gupta <anuj20.g@samsung.com>
Suggested-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Link: https://lore.kernel.org/r/20220930062749.152261-4-anuj20.g@samsung.com
Signed-off-by: Jens Axboe <axboe@kernel.dk>

io_uring: introduce fixed buffer support for io_uring_cmd

Add IORING_URING_CMD_FIXED flag that is to be used for sending io_uring
command with previously registered buffers. User-space passes the buffer
index in sqe->buf_index, same as done in read/write variants that uses
fixed buffers.

Signed-off-by: Anuj Gupta <anuj20.g@samsung.com>
Signed-off-by: Kanchan Joshi <joshi.k@samsung.com>
Link: https://lore.kernel.org/r/20220930062749.152261-3-anuj20.g@samsung.com
[axboe: shuffle valid flags check before acting on it]
Signed-off-by: Jens Axboe <axboe@kernel.dk>

io_uring: add io_uring_cmd_import_fixed

This is a new helper that callers can use to obtain a bvec iterator for
the previously mapped buffer. This is preparatory work to enable
fixed-buffer support for io_uring_cmd.

Signed-off-by: Anuj Gupta <anuj20.g@samsung.com>
Signed-off-by: Kanchan Joshi <joshi.k@samsung.com>
Link: https://lore.kernel.org/r/20220930062749.152261-2-anuj20.g@samsung.com
Signed-off-by: Jens Axboe <axboe@kernel.dk>

nvme: enable batched completions of passthrough IO

Now that the normal passthrough end_io path doesn't need the request
anymore, we can kill the explicit blk_mq_free_request() and just pass
back RQ_END_IO_FREE instead. This enables the batched completion from
freeing batches of requests at the time.

This brings passthrough IO performance at least on par with bdev based
O_DIRECT with io_uring. With this and batche allocations, peak performance
goes from 110M IOPS to 122M IOPS. For IRQ based, passthrough is now also
about 10% faster than previously, going from ~61M to ~67M IOPS.

Reviewed-by: Anuj Gupta <anuj20.g@samsung.com>
Reviewed-by: Sagi Grimberg <sagi@grimberg.me>
Reviewed-by: Keith Busch <kbusch@kernel.org>
Co-developed-by: Stefan Roesch <shr@fb.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>

nvme: split out metadata vs non metadata end_io uring_cmd completions

By splitting up the metadata and non-metadata end_io handling, we can
remove any request dependencies on the normal non-metadata IO path. This
is in preparation for enabling the normal IO passthrough path to pass
the ownership of the request back to the block layer.

Reviewed-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Anuj Gupta <anuj20.g@samsung.com>
Reviewed-by: Sagi Grimberg <sagi@grimberg.me>
Reviewed-by: Keith Busch <kbusch@kernel.org>
Co-developed-by: Stefan Roesch <shr@fb.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>

block: allow end_io based requests in the completion batch handling

With end_io handlers now being able to potentially pass ownership of
the request upon completion, we can allow requests with end_io handlers
in the batch completion handling.

Reviewed-by: Anuj Gupta <anuj20.g@samsung.com>
Reviewed-by: Keith Busch <kbusch@kernel.org>
Co-developed-by: Stefan Roesch <shr@fb.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>

block: change request end_io handler to pass back a return value

Everything is just converted to returning RQ_END_IO_NONE, and there
should be no functional changes with this patch.

In preparation for allowing the end_io handler to pass ownership back
to the block layer, rather than retain ownership of the request.

Reviewed-by: Keith Busch <kbusch@kernel.org>
Signed-off-by: Jens Axboe <axboe@kernel.dk>

block: enable batched allocation for blk_mq_alloc_request()

The filesystem IO path can take advantage of allocating batches of
requests, if the underlying submitter tells the block layer about it
through the blk_plug. For passthrough IO, the exported API is the
blk_mq_alloc_request() helper, and that one does not allow for
request caching.

Wire up request caching for blk_mq_alloc_request(), which is generally
done without having a bio available upfront.

Tested-by: Anuj Gupta <anuj20.g@samsung.com>
Reviewed-by: Keith Busch <kbusch@kernel.org>
Signed-off-by: Jens Axboe <axboe@kernel.dk>

block: kill deprecated BUG_ON() in the flush handling

We've never had any useful reports from this BUG_ON(), and in fact a
number of the BUG_ON()'s in the flush handling need to be turned into
more graceful handling.

In preparation for allowing batched completions of the end_io handling,
where we can enter the flush completion with queuelist having been reused
for the batch, get rid of this BUG_ON().

Signed-off-by: Jens Axboe <axboe@kernel.dk>

Merge branch 'for-6.1/io_uring' into for-6.1/passthrough

* for-6.1/io_uring: (56 commits)
  io_uring/net: fix notif cqe reordering
  io_uring/net: don't update msg_name if not provided
  io_uring: don't gate task_work run on TIF_NOTIFY_SIGNAL
  io_uring/rw: defer fsnotify calls to task context
  io_uring/net: fix fast_iov assignment in io_setup_async_msg()
  io_uring/net: fix non-zc send with address
  io_uring/net: don't skip notifs for failed requests
  io_uring/rw: don't lose short results on io_setup_async_rw()
  io_uring/rw: fix unexpected link breakage
  io_uring/net: fix cleanup double free free_iov init
  io_uring: fix CQE reordering
  io_uring/net: fix UAF in io_sendrecv_fail()
  selftest/net: adjust io_uring sendzc notif handling
  io_uring: ensure local task_work marks task as running
  io_uring/net: zerocopy sendmsg
  io_uring/net: combine fail handlers
  io_uring/net: rename io_sendzc()
  io_uring/net: support non-zerocopy sendto
  io_uring/net: refactor io_setup_async_addr
  io_uring/net: don't lose partial send_zc on fail
  ...

Merge branch 'for-6.1/block' into for-6.1/passthrough

* for-6.1/block: (162 commits)
  sbitmap: fix lockup while swapping
  block: add rationale for not using blk_mq_plug() when applicable
  block: adapt blk_mq_plug() to not plug for writes that require a zone lock
  s390/dasd: use blk_mq_alloc_disk
  blk-cgroup: don't update the blkg lookup hint in blkg_conf_prep
  nvmet: don't look at the request_queue in nvmet_bdev_set_limits
  nvmet: don't look at the request_queue in nvmet_bdev_zone_mgmt_emulate_all
  blk-mq: use quiesced elevator switch when reinitializing queues
  block: replace blk_queue_nowait with bdev_nowait
  nvme: remove nvme_ctrl_init_connect_q
  nvme-loop: use the tagset alloc/free helpers
  nvme-loop: store the generic nvme_ctrl in set->driver_data
  nvme-loop: initialize sqsize later
  nvme-fc: use the tagset alloc/free helpers
  nvme-fc: store the generic nvme_ctrl in set->driver_data
  nvme-fc: keep ctrl->sqsize in sync with opts->queue_size
  nvme-rdma: use the tagset alloc/free helpers
  nvme-rdma: store the generic nvme_ctrl in set->driver_data
  nvme-tcp: use the tagset alloc/free helpers
  nvme-tcp: store the generic nvme_ctrl in set->driver_data
  ...

Signed-off-by: Jens Axboe <axboe@kernel.dk>

sbitmap: fix lockup while swapping

Commit 4acb83417cad ("sbitmap: fix batched wait_cnt accounting")
is a big improvement: without it, I had to revert to before commit
040b83fcecfb ("sbitmap: fix possible io hung due to lost wakeup")
to avoid the high system time and freezes which that had introduced.

Now okay on the NVME laptop, but 4acb83417cad is a disaster for heavy
swapping (kernel builds in low memory) on another: soon locking up in
sbitmap_queue_wake_up() (into which __sbq_wake_up() is inlined), cycling
around with waitqueue_active() but wait_cnt 0 .  Here is a backtrace,
showing the common pattern of outer sbitmap_queue_wake_up() interrupted
before setting wait_cnt 0 back to wake_batch (in some cases other CPUs
are idle, in other cases they're spinning for a lock in dd_bio_merge()):

sbitmap_queue_wake_up < sbitmap_queue_clear < blk_mq_put_tag <
__blk_mq_free_request < blk_mq_free_request < __blk_mq_end_request <
scsi_end_request < scsi_io_completion < scsi_finish_command <
scsi_complete < blk_complete_reqs < blk_done_softirq < __do_softirq <
__irq_exit_rcu < irq_exit_rcu < common_interrupt < asm_common_interrupt <
_raw_spin_unlock_irqrestore < __wake_up_common_lock < __wake_up <
sbitmap_queue_wake_up < sbitmap_queue_clear < blk_mq_put_tag <
__blk_mq_free_request < blk_mq_free_request < dd_bio_merge <
blk_mq_sched_bio_merge < blk_mq_attempt_bio_merge < blk_mq_submit_bio <
__submit_bio < submit_bio_noacct_nocheck < submit_bio_noacct <
submit_bio < __swap_writepage < swap_writepage < pageout <
shrink_folio_list < evict_folios < lru_gen_shrink_lruvec <
shrink_lruvec < shrink_node < do_try_to_free_pages < try_to_free_pages <
__alloc_pages_slowpath < __alloc_pages < folio_alloc < vma_alloc_folio <
do_anonymous_page < __handle_mm_fault < handle_mm_fault <
do_user_addr_fault < exc_page_fault < asm_exc_page_fault

See how the process-context sbitmap_queue_wake_up() has been interrupted,
after bringing wait_cnt down to 0 (and in this example, after doing its
wakeups), before advancing wake_index and refilling wake_cnt: an
interrupt-context sbitmap_queue_wake_up() of the same sbq gets stuck.

I have almost no grasp of all the possible sbitmap races, and their
consequences: but __sbq_wake_up() can do nothing useful while wait_cnt 0,
so it is better if sbq_wake_ptr() skips on to the next ws in that case:
which fixes the lockup and shows no adverse consequence for me.

The check for wait_cnt being 0 is obviously racy, and ultimately can lead
to lost wakeups: for example, when there is only a single waitqueue with
waiters.  However, lost wakeups are unlikely to matter in these cases,
and a proper fix requires redesign (and benchmarking) of the batched
wakeup code: so let's plug the hole with this bandaid for now.

Signed-off-by: Hugh Dickins <hughd@google.com>
Reviewed-by: Jan Kara <jack@suse.cz>
Reviewed-by: Keith Busch <kbusch@kernel.org>
Link: https://lore.kernel.org/r/9c2038a7-cdc5-5ee-854c-fbc6168bf16@google.com
Signed-off-by: Jens Axboe <axboe@kernel.dk>

io_uring/net: fix notif cqe reordering

send zc is not restricted to !IO_URING_F_UNLOCKED anymore and so
we can't use task-tw ordering trick to order notification cqes
with requests completions. In this case leave it alone and let
io_send_zc_cleanup() flush it.

Cc: stable@vger.kernel.org
Fixes: 53bdc88aac9a2 ("io_uring/notif: order notif vs send CQEs")
Signed-off-by: Pavel Begunkov <asml.silence@gmail.com>
Link: https://lore.kernel.org/r/0031f3a00d492e814a4a0935a2029a46d9c9ba06.1664486545.git.asml.silence@gmail.com
Signed-off-by: Jens Axboe <axboe@kernel.dk>

io_uring/net: don't update msg_name if not provided

io_sendmsg_copy_hdr() may clear msg->msg_name if the userspace didn't
provide it, we should retain NULL in this case.

Cc: stable@vger.kernel.org
Signed-off-by: Pavel Begunkov <asml.silence@gmail.com>
Link: https://lore.kernel.org/r/97d49f61b5ec76d0900df658cfde3aa59ff22121.1664486545.git.asml.silence@gmail.com
Signed-off-by: Jens Axboe <axboe@kernel.dk>

io_uring: don't gate task_work run on TIF_NOTIFY_SIGNAL

This isn't a reliable mechanism to tell if we have task_work pending, we
really should be looking at whether we have any items queued. This is
problematic if forward progress is gated on running said task_work. One
such example is reading from a pipe, where the write side has been closed
right before the read is started. The fput() of the file queues TWA_RESUME
task_work, and we need that task_work to be run before ->release() is
called for the pipe. If ->release() isn't called, then the read will sit
forever waiting on data that will never arise.

Fix this by io_run_task_work() so it checks if we have task_work pending
rather than rely on TIF_NOTIFY_SIGNAL for that. The latter obviously
doesn't work for task_work that is queued without TWA_SIGNAL.

Reported-by: Christiano Haesbaert <haesbaert@haesbaert.org>
Cc: stable@vger.kernel.org
Link: https://github.com/axboe/liburing/issues/665
Signed-off-by: Jens Axboe <axboe@kernel.dk>

io_uring/rw: defer fsnotify calls to task context

We can't call these off the kiocb completion as that might be off
soft/hard irq context. Defer the calls to when we process the
task_work for this request. That avoids valid complaints like:

stack backtrace:
CPU: 1 PID: 0 Comm: swapper/1 Not tainted 6.0.0-rc6-syzkaller-00321-g105a36f3694e #0
Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 08/26/2022
Call Trace:
 <IRQ>
 __dump_stack lib/dump_stack.c:88 [inline]
 dump_stack_lvl+0xcd/0x134 lib/dump_stack.c:106
 print_usage_bug kernel/locking/lockdep.c:3961 [inline]
 valid_state kernel/locking/lockdep.c:3973 [inline]
 mark_lock_irq kernel/locking/lockdep.c:4176 [inline]
 mark_lock.part.0.cold+0x18/0xd8 kernel/locking/lockdep.c:4632
 mark_lock kernel/locking/lockdep.c:4596 [inline]
 mark_usage kernel/locking/lockdep.c:4527 [inline]
 __lock_acquire+0x11d9/0x56d0 kernel/locking/lockdep.c:5007
 lock_acquire kernel/locking/lockdep.c:5666 [inline]
 lock_acquire+0x1ab/0x570 kernel/locking/lockdep.c:5631
 __fs_reclaim_acquire mm/page_alloc.c:4674 [inline]
 fs_reclaim_acquire+0x115/0x160 mm/page_alloc.c:4688
 might_alloc include/linux/sched/mm.h:271 [inline]
 slab_pre_alloc_hook mm/slab.h:700 [inline]
 slab_alloc mm/slab.c:3278 [inline]
 __kmem_cache_alloc_lru mm/slab.c:3471 [inline]
 kmem_cache_alloc+0x39/0x520 mm/slab.c:3491
 fanotify_alloc_fid_event fs/notify/fanotify/fanotify.c:580 [inline]
 fanotify_alloc_event fs/notify/fanotify/fanotify.c:813 [inline]
 fanotify_handle_event+0x1130/0x3f40 fs/notify/fanotify/fanotify.c:948
 send_to_group fs/notify/fsnotify.c:360 [inline]
 fsnotify+0xafb/0x1680 fs/notify/fsnotify.c:570
 __fsnotify_parent+0x62f/0xa60 fs/notify/fsnotify.c:230
 fsnotify_parent include/linux/fsnotify.h:77 [inline]
 fsnotify_file include/linux/fsnotify.h:99 [inline]
 fsnotify_access include/linux/fsnotify.h:309 [inline]
 __io_complete_rw_common+0x485/0x720 io_uring/rw.c:195
 io_complete_rw+0x1a/0x1f0 io_uring/rw.c:228
 iomap_dio_complete_work fs/iomap/direct-io.c:144 [inline]
 iomap_dio_bio_end_io+0x438/0x5e0 fs/iomap/direct-io.c:178
 bio_endio+0x5f9/0x780 block/bio.c:1564
 req_bio_endio block/blk-mq.c:695 [inline]
 blk_update_request+0x3fc/0x1300 block/blk-mq.c:825
 scsi_end_request+0x7a/0x9a0 drivers/scsi/scsi_lib.c:541
 scsi_io_completion+0x173/0x1f70 drivers/scsi/scsi_lib.c:971
 scsi_complete+0x122/0x3b0 drivers/scsi/scsi_lib.c:1438
 blk_complete_reqs+0xad/0xe0 block/blk-mq.c:1022
 __do_softirq+0x1d3/0x9c6 kernel/softirq.c:571
 invoke_softirq kernel/softirq.c:445 [inline]
 __irq_exit_rcu+0x123/0x180 kernel/softirq.c:650
 irq_exit_rcu+0x5/0x20 kernel/softirq.c:662
 common_interrupt+0xa9/0xc0 arch/x86/kernel/irq.c:240

Fixes: f63cf5192fe3 ("io_uring: ensure that fsnotify is always called")
Link: https://lore.kernel.org/all/20220929135627.ykivmdks2w5vzrwg@quack3/
Reported-by: syzbot+dfcc5f4da15868df7d4d@syzkaller.appspotmail.com
Reported-by: Jan Kara <jack@suse.cz>
Signed-off-by: Jens Axboe <axboe@kernel.dk>

block: add rationale for not using blk_mq_plug() when applicable

There are two places in the block layer at the moment where
blk_mq_plug() helper could be used instead of directly accessing the
plug from struct current. In both these cases, directly accessing the plug
should not have any consequences for zoned devices.

Make the intent explicit by adding comments instead of introducing unwanted
checks with blk_mq_plug() helper.[1]

[1] https://lore.kernel.org/linux-block/f6e54907-1035-2b2c-6387-ed178be05ccb@kernel.dk/

Signed-off-by: Pankaj Raghav <p.raghav@samsung.com>
Suggested-by: Jens Axboe <axboe@kernel.dk>
Link: https://lore.kernel.org/r/20220929144141.140077-1-p.raghav@samsung.com
[axboe: fixup multi-line comment style]
Signed-off-by: Jens Axboe <axboe@kernel.dk>

block: adapt blk_mq_plug() to not plug for writes that require a zone lock

The current implementation of blk_mq_plug() disables plugging for all
operations that involves a transfer to the device as we just check if
the last bit in op_is_write() function.

Modify blk_mq_plug() to disable plugging only for REQ_OP_WRITE and
REQ_OP_WRITE_ZEROS as they might require a zone lock.

Suggested-by: Christoph Hellwig <hch@lst.de>
Suggested-by: Damien Le Moal <damien.lemoal@opensource.wdc.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Pankaj Raghav <p.raghav@samsung.com>
Reviewed-by: Damien Le Moal <damien.lemoal@opensource.wdc.com>
Reviewed-by: Johannes Thumshirn <johannes.thumshirn@wdc.com>
Link: https://lore.kernel.org/r/20220929074745.103073-2-p.raghav@samsung.com
Signed-off-by: Jens Axboe <axboe@kernel.dk>

io_uring/net: fix fast_iov assignment in io_setup_async_msg()

I hit a very bad problem during my tests of SENDMSG_ZC.
BUG(); in first_iovec_segment() triggered very easily.
The problem was io_setup_async_msg() in the partial retry case,
which seems to happen more often with _ZC.

iov_iter_iovec_advance() may change i->iov in order to have i->iov_offset
being only relative to the first element.

Which means kmsg->msg.msg_iter.iov is no longer the
same as kmsg->fast_iov.

But this would rewind the copy to be the start of
async_msg->fast_iov, which means the internal
state of sync_msg->msg.msg_iter is inconsitent.

I tested with 5 vectors with length like this 4, 0, 64, 20, 8388608
and got a short writes with:
- ret=2675244 min_ret=8388692 => remaining 5713448 sr->done_io=2675244
- ret=-EAGAIN => io_uring_poll_arm
- ret=4911225 min_ret=5713448 => remaining 802223  sr->done_io=7586469
- ret=-EAGAIN => io_uring_poll_arm
- ret=802223  min_ret=802223  => res=8388692

While this was easily triggered with SENDMSG_ZC (queued for 6.1),
it was a potential problem starting with 7ba89d2af17aa879dda30f5d5d3f152e587fc551
in 5.18 for IORING_OP_RECVMSG.
And also with 4c3c09439c08b03d9503df0ca4c7619c5842892e in 5.19
for IORING_OP_SENDMSG.

However 257e84a5377fbbc336ff563833a8712619acce56 introduced the critical
code into io_setup_async_msg() in 5.11.

Fixes: 7ba89d2af17aa ("io_uring: ensure recv and recvmsg handle MSG_WAITALL correctly")
Fixes: 257e84a5377fb ("io_uring: refactor sendmsg/recvmsg iov managing")
Cc: stable@vger.kernel.org
Signed-off-by: Stefan Metzmacher <metze@samba.org>
Reviewed-by: Pavel Begunkov <asml.silence@gmail.com>
Link: https://lore.kernel.org/r/b2e7be246e2fb173520862b0c7098e55767567a2.1664436949.git.metze@samba.org
Signed-off-by: Jens Axboe <axboe@kernel.dk>

io_uring/net: fix non-zc send with address

We're currently ignoring the dest address with non-zerocopy send because
even though we copy it from the userspace shortly after ->msg_name gets
zeroed. Move msghdr init earlier.

Fixes: 516e82f0e043a ("io_uring/net: support non-zerocopy sendto")
Signed-off-by: Pavel Begunkov <asml.silence@gmail.com>
Link: https://lore.kernel.org/r/176ced5e8568aa5d300ca899b7f05b303ebc49fd.1664409532.git.asml.silence@gmail.com
Signed-off-by: Jens Axboe <axboe@kernel.dk>

Merge tag 'nvme-6.1-2022-09-28' of git://git.infradead.org/nvme into for-6.1/block

Pull NVMe updates from Christoph:

"nvme updates for Linux 6.1

 - handle effects after freeing the request (Keith Busch)
 - copy firmware_rev on each init (Keith Busch)
 - restrict management ioctls to admin (Keith Busch)
 - ensure subsystem reset is single threaded (Keith Busch)
 - report the actual number of tagset maps in nvme-pci (Keith Busch)
 - small fabrics authentication fixups (Christoph Hellwig)
 - add common code for tagset allocation and freeing (Christoph Hellwig)
 - stop using the request_queue in nvmet (Christoph Hellwig)
 - set min_align_mask before calculating max_hw_sectors
   (Rishabh Bhatnagar)
 - send a rediscover uevent when a persistent discovery controller
   reconnects (Sagi Grimberg)
 - misc nvmet-tcp fixes (Varun Prakash, zhenwei pi)"

* tag 'nvme-6.1-2022-09-28' of git://git.infradead.org/nvme: (31 commits)
  nvmet: don't look at the request_queue in nvmet_bdev_set_limits
  nvmet: don't look at the request_queue in nvmet_bdev_zone_mgmt_emulate_all
  nvme: remove nvme_ctrl_init_connect_q
  nvme-loop: use the tagset alloc/free helpers
  nvme-loop: store the generic nvme_ctrl in set->driver_data
  nvme-loop: initialize sqsize later
  nvme-fc: use the tagset alloc/free helpers
  nvme-fc: store the generic nvme_ctrl in set->driver_data
  nvme-fc: keep ctrl->sqsize in sync with opts->queue_size
  nvme-rdma: use the tagset alloc/free helpers
  nvme-rdma: store the generic nvme_ctrl in set->driver_data
  nvme-tcp: use the tagset alloc/free helpers
  nvme-tcp: store the generic nvme_ctrl in set->driver_data
  nvme-tcp: remove the unused queue_size member in nvme_tcp_queue
  nvme: add common helpers to allocate and free tagsets
  nvme-auth: add a MAINTAINERS entry
  nvmet: add helpers to set the result field for connect commands
  nvme: improve the NVME_CONNECT_AUTHREQ* definitions
  nvmet-auth: don't try to cancel a non-initialized work_struct
  nvmet-tcp: remove nvmet_tcp_finish_cmd
  ...

s390/dasd: use blk_mq_alloc_disk

As far as I can tell there is no need for the staged setup in
dasd, so allocate the tagset and the disk with the queue in
dasd_gendisk_alloc.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Stefan Haberland <sth@linux.ibm.com>
Signed-off-by: Stefan Haberland <sth@linux.ibm.com>
Link: https://lore.kernel.org/r/20220928143945.1687114-2-sth@linux.ibm.com
Signed-off-by: Jens Axboe <axboe@kernel.dk>

io_uring/net: don't skip notifs for failed requests

We currently only add a notification CQE when the send succeded, i.e.
cqe.res >= 0. However, it'd be more robust to do buffer notifications
for failed requests as well in case drivers decide do something fanky.

Always return a buffer notification after initial prep, don't hide it.
This behaviour is better aligned with documentation and the patch also
helps the userspace to respect it.

Cc: stable@vger.kernel.org # 6.0
Suggested-by: Stefan Metzmacher <metze@samba.org>
Signed-off-by: Pavel Begunkov <asml.silence@gmail.com>
Link: https://lore.kernel.org/r/9c8bead87b2b980fcec441b8faef52188b4a6588.1664292100.git.asml.silence@gmail.com
Signed-off-by: Jens Axboe <axboe@kernel.dk>

blk-cgroup: don't update the blkg lookup hint in blkg_conf_prep

blkg_conf_prep just creates a new blkg structure, there is no real
need to update the lookup hint which should only be done on a
successful lookup in the I/O path.

Suggested-by: Tejun Heo <tj@kernel.org>
Signed-off-by: Christoph Hellwig <hch@lst.de>
Acked-by: Tejun Heo <tj@kernel.org>
Link: https://lore.kernel.org/r/20220927065425.257876-1-hch@lst.de
Signed-off-by: Jens Axboe <axboe@kernel.dk>

nvmet: don't look at the request_queue in nvmet_bdev_set_limits

nvmet is a consumer of the block layer and should not directly look at
the request_queue.  Use the bdev_ helpers to retrieve the device limits
instead.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Keith Busch <kbusch@kernel.org>

nvmet: don't look at the request_queue in nvmet_bdev_zone_mgmt_emulate_all

nvmet is a consumer of the block layer and should not directly look at
the request_queue.  Just use the NUMA node ID from the gendisk instead of
the request_queue.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Keith Busch <kbusch@kernel.org>

blk-mq: use quiesced elevator switch when reinitializing queues

The hctx's run_work may be racing with the elevator switch when
reinitializing hardware queues. The queue is merely frozen in this
context, but that only prevents requests from allocating and doesn't
stop the hctx work from running. The work may get an elevator pointer
that's being torn down, and can result in use-after-free errors and
kernel panics (example below). Use the quiesced elevator switch instead,
and make the previous one static since it is now only used locally.

  nvme nvme0: resetting controller
  nvme nvme0: 32/0/0 default/read/poll queues
  BUG: kernel NULL pointer dereference, address: 0000000000000008
  #PF: supervisor read access in kernel mode
  #PF: error_code(0x0000) - not-present page
  PGD 80000020c8861067 P4D 80000020c8861067 PUD 250f8c8067 PMD 0
  Oops: 0000 [#1] SMP PTI
  Workqueue: kblockd blk_mq_run_work_fn
  RIP: 0010:kyber_has_work+0x29/0x70

...

  Call Trace:
   __blk_mq_do_dispatch_sched+0x83/0x2b0
   __blk_mq_sched_dispatch_requests+0x12e/0x170
   blk_mq_sched_dispatch_requests+0x30/0x60
   __blk_mq_run_hw_queue+0x2b/0x50
   process_one_work+0x1ef/0x380
   worker_thread+0x2d/0x3e0

Signed-off-by: Keith Busch <kbusch@kernel.org>
Reviewed-by: Ming Lei <ming.lei@redhat.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Link: https://lore.kernel.org/r/20220927155652.3260724-1-kbusch@fb.com
Signed-off-by: Jens Axboe <axboe@kernel.dk>

block: replace blk_queue_nowait with bdev_nowait

Replace blk_queue_nowait with a bdev_nowait helpers that takes the
block_device given that the I/O submission path should not have to
look into the request_queue.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Pankaj Raghav <p.raghav@samsung.com>
Link: https://lore.kernel.org/r/20220927075815.269694-1-hch@lst.de
Signed-off-by: Jens Axboe <axboe@kernel.dk>

nvme: remove nvme_ctrl_init_connect_q

Unused now.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Sagi Grimberg <sagi@grimberg.me>
Reviewed-by: Chaitanya Kulkarni <kch@nvidia.com>

nvme-loop: use the tagset alloc/free helpers

Use the common helpers to allocate and free the tagsets.  To make this
work the generic nvme_ctrl now needs to be stored in the hctx private
data instead of the nvme_loop_ctrl.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Sagi Grimberg <sagi@grimberg.me>
Reviewed-by: Chaitanya Kulkarni <kch@nvidia.com>

nvme-loop: store the generic nvme_ctrl in set->driver_data

Point the private data to the generic controller structure in preparation
of using the common tagset init/exit code.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Sagi Grimberg <sagi@grimberg.me>
Reviewed-by: Chaitanya Kulkarni <kch@nvidia.com>

nvme-loop: initialize sqsize later

Defer initializing the sqsize field from the options until it has been
capped by MAXCMD.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Sagi Grimberg <sagi@grimberg.me>
Reviewed-by: Chaitanya Kulkarni <kch@nvidia.com>

nvme-fc: use the tagset alloc/free helpers

Use the common helpers to allocate and free the tagsets.  To make this
work the generic nvme_ctrl now needs to be stored in the hctx private
data instead of the nvme_fc_ctrl.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Sagi Grimberg <sagi@grimberg.me>
Reviewed-by: James Smart <jsmart2021@gmail.com>

nvme-fc: store the generic nvme_ctrl in set->driver_data

Point the private data to the generic controller structure in preparation
of using the common tagset init/exit code and use the chance the cleanup
the init_hctx methods a bit.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Sagi Grimberg <sagi@grimberg.me>
Reviewed-by: James Smart <jsmart2021@gmail.com>

nvme-fc: keep ctrl->sqsize in sync with opts->queue_size

Also update the sqsize field when capping the queue size, and remove the
check a queue size that is larger than sqsize given that sqsize is only
initialized from opts->queue_size.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Sagi Grimberg <sagi@grimberg.me>
Reviewed-by: James Smart <jsmart2021@gmail.com>

nvme-rdma: use the tagset alloc/free helpers

Use the common helpers to allocate and free the tagsets.  To make this
work the generic nvme_ctrl now needs to be stored in the hctx private
data instead of the nvme_rdma_ctrl.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Sagi Grimberg <sagi@grimberg.me>
Reviewed-by: Chaitanya Kulkarni <kch@nvidia.com>

nvme-rdma: store the generic nvme_ctrl in set->driver_data

Point the private data to the generic controller structure in preparation
of using the common tagset init/exit code.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Sagi Grimberg <sagi@grimberg.me>
Reviewed-by: Chaitanya Kulkarni <kch@nvidia.com>

nvme-tcp: use the tagset alloc/free helpers

Use the common helpers to allocate and free the tagsets.  To make this
work the generic nvme_ctrl now needs to be stored in the hctx private
data instead of the nvme_tcp_ctrl.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Sagi Grimberg <sagi@grimberg.me>
Reviewed-by: Chaitanya Kulkarni <kch@nvidia.com>

nvme-tcp: store the generic nvme_ctrl in set->driver_data

Point the private data to the generic controller structure in preparation
of using the common tagset init/exit code.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Sagi Grimberg <sagi@grimberg.me>
Reviewed-by: Chaitanya Kulkarni <kch@nvidia.com>

nvme-tcp: remove the unused queue_size member in nvme_tcp_queue

->nvme_tcp_queue is not used anywhere, so remove it.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Sagi Grimberg <sagi@grimberg.me>
Reviewed-by: Chaitanya Kulkarni <kch@nvidia.com>

nvme: add common helpers to allocate and free tagsets

Add common helpers to allocate and tear down the admin and I/O tag sets,
including the special queues allocated with them.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Chaitanya Kulkarni <kch@nvidia.com>

nvme-auth: add a MAINTAINERS entry

Add Hannes as the nvme-auth maintainer.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Acked-by: Hannes Reinecke <hare@suse.de>
Reviewed-by: Sagi Grimberg <sagi@grimberg.me>
Reviewed-by: Chaitanya Kulkarni <kch@nvidia.com>

nvmet: add helpers to set the result field for connect commands

The code to set the result field for the admin and I/O connect commands
is not only verbose and duplicated, but also violates the aliasing
rules as it accesses both the u16 and u32 members in the union.

Add a little helper to sort all that out.

Fixes: db1312dd9548 ("nvmet: implement basic In-Band Authentication")
Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Sagi Grimberg <sagi@grimberg.me>
Reviewed-by: Hannes Reinecke <hare@suse.de>

nvme: improve the NVME_CONNECT_AUTHREQ* definitions

Mark them as unsigned so that we don't need extra casts, and define
them relative to cdword0 instead of requiring extra shifts.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Sagi Grimberg <sagi@grimberg.me>
Reviewed-by: Hannes Reinecke <hare@suse.de>

nvmet-auth: don't try to cancel a non-initialized work_struct

Currently blktests nvme/002 trips up debugobjects if CONFIG_NVME_AUTH is
enabled, but authentication is not on a queue.  This is because
nvmet_auth_sq_free cancels sq->auth_expired_work unconditionaly, while
auth_expired_work is only ever initialized if authentication is enabled
for a given controller.

Fix this by calling most of what is nvmet_init_auth unconditionally
when initializing the SQ, and just do the setting of the result
field in the connect command handler.

Fixes: db1312dd9548 ("nvmet: implement basic In-Band Authentication")
Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Sagi Grimberg <sagi@grimberg.me>
Reviewed-by: Hannes Reinecke <hare@suse.de>

nvmet-tcp: remove nvmet_tcp_finish_cmd

There is only a single call-site of nvmet_tcp_finish_cmd(), this
becomes redundant. Remove nvmet_tcp_finish_cmd() and use the original
function body instead.

Suggested-by: Sagi Grimberg <sagi@grimberg.me>
Signed-off-by: zhenwei pi <pizhenwei@bytedance.com>
Reviewed-by: Sagi Grimberg <sagi@grimberg.me>
Signed-off-by: Christoph Hellwig <hch@lst.de>

nvmet-tcp: add bounds check on Transfer Tag

ttag is used as an index to get cmd in nvmet_tcp_handle_h2c_data_pdu(),
add a bounds check to avoid out-of-bounds access.

Signed-off-by: Varun Prakash <varun@chelsio.com>
Reviewed-by: Sagi Grimberg <sagi@grimberg.me>
Signed-off-by: Christoph Hellwig <hch@lst.de>

nvmet-tcp: handle ICReq PDU received in NVMET_TCP_Q_LIVE state

As per NVMe/TCP transport specification ICReq PDU is the first PDU received
by the controller and controller should receive only one ICReq PDU.

If controller receives more than one ICReq PDU then this can be considered
as fatal error.

nvmet-tcp driver does not check for ICReq PDU opcode if queue state is
NVMET_TCP_Q_LIVE. In LIVE state ICReq PDU is treated as CapsuleCmd PDU,
this can result in abnormal behavior.

Add a check for ICReq PDU in nvmet_tcp_done_recv_pdu() to fix this issue.

Signed-off-by: Varun Prakash <varun@chelsio.com>
Reviewed-by: Sagi Grimberg <sagi@grimberg.me>
Signed-off-by: Christoph Hellwig <hch@lst.de>

nvmet-tcp: fix NULL pointer dereference during release

nvmet-tcp frees CMD buffers in nvmet_tcp_uninit_data_in_cmds(),
and waits the inflight IO requests in nvmet_sq_destroy(). During wait
the inflight IO requests, the callback nvmet_tcp_queue_response()
is called from backend after IO complete, this leads a typical
Use-After-Free issue like this:

 BUG: kernel NULL pointer dereference, address: 0000000000000008
 #PF: supervisor read access in kernel mode
 #PF: error_code(0x0000) - not-present page
 PGD 107f80067 P4D 107f80067 PUD 10789e067 PMD 0
 Oops: 0000 [#1] PREEMPT SMP NOPTI
 CPU: 1 PID: 123 Comm: kworker/1:1H Kdump: loaded Tainted: G            E      6.0.0-rc2.bm.1-amd64 #15
 Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS rel-1.16.0-0-gd239552ce722-prebuilt.qemu.org 04/01/2014
 Workqueue: nvmet_tcp_wq nvmet_tcp_io_work [nvmet_tcp]
 RIP: 0010:shash_ahash_digest+0x2b/0x110
 Code: 1f 44 00 00 41 57 41 56 41 55 41 54 55 48 89 fd 53 48 89 f3 48 83 ec 08 44 8b 67 30 45 85 e4 74 1c 48 8b 57 38 b8 00 10 00 00 <44> 8b 7a 08 44 29 f8 39 42 0c 0f 46 42 0c 41 39 c4 76 43 48 8b 03
 RSP: 0018:ffffc9000051bdd8 EFLAGS: 00010206
 RAX: 0000000000001000 RBX: ffff888100ab5470 RCX: 0000000000000000
 RDX: 0000000000000000 RSI: ffff888100ab5470 RDI: ffff888100ab5420
 RBP: ffff888100ab5420 R08: ffff8881024d08c8 R09: ffff888103e1b4b8
 R10: 8080808080808080 R11: 0000000000000000 R12: 0000000000001000
 R13: 0000000000000000 R14: ffff88813412bd4c R15: ffff8881024d0800
 FS:  0000000000000000(0000) GS:ffff88883fa40000(0000) knlGS:0000000000000000
 CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
 CR2: 0000000000000008 CR3: 0000000104b48000 CR4: 0000000000350ee0
 Call Trace:
  <TASK>
  nvmet_tcp_io_work+0xa52/0xb52 [nvmet_tcp]
  ? __switch_to+0x106/0x420
  process_one_work+0x1ae/0x380
  ? process_one_work+0x380/0x380
  worker_thread+0x30/0x360
  ? process_one_work+0x380/0x380
  kthread+0xe6/0x110
  ? kthread_complete_and_exit+0x20/0x20
  ret_from_fork+0x1f/0x30

Separate nvmet_tcp_uninit_data_in_cmds() into two steps:

  uninit data in cmds                  <- new step 1
  nvmet_sq_destroy();
  cancel_work_sync(&queue->io_work);
  free CMD buffers                     <- new step 2

Signed-off-by: zhenwei pi <pizhenwei@bytedance.com>
Reviewed-by: Sagi Grimberg <sagi@grimberg.me>
Signed-off-by: Christoph Hellwig <hch@lst.de>

nvme-pci: report the actual number of tagset maps

We've been reporting 2 maps regardless of whether the module parameter
asked for anything beyond the default queues. A consequence of this
means that blk-mq will reinitialize the all the hardware contexts and io
schedulers on every controller reset when the mapping is exactly the
same as before. This unnecessary overhead is adding several milliseconds
on a reset for environments that don't need it. Report the actual number
of mappings in use.

Signed-off-by: Keith Busch <kbusch@kernel.org>
Signed-off-by: Christoph Hellwig <hch@lst.de>

nvme-pci: set min_align_mask before calculating max_hw_sectors

If swiotlb is force enabled dma_max_mapping_size ends up calling
swiotlb_max_mapping_size which takes into account the min align mask for
the device.  Set the min align mask for nvme driver before calling
dma_max_mapping_size while calculating max hw sectors.

Signed-off-by: Rishabh Bhatnagar <risbhat@amazon.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>

nvme: send a rediscover uevent when a persistent discovery controller reconnects

When a discovery controller is disconnected, no AENs will arrive to
notify the host about discovery log change events.

In order to solve this, send a uevent notification when a
persistent discovery controller reconnects. We add a new ctrl
flag NVME_CTRL_STARTED_ONCE that will be set on the first
start, and consecutive calls will find it set, and send the
event to userspace if the controller is a discovery controller.

Upon the event reception, userspace will re-read the discovery
log page and will act upon changes as it sees fit.

Signed-off-by: Sagi Grimberg <sagi@grimberg.me>
Reviewed-by: Daniel Wagner <dwagner@suse.de>
Reviewed-by: James Smart <jsmart2021@gmail.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>

nvme: enumerate controller flags

We expect to grow a few of these flags for various purposes
so make them a proper enumeration.

Signed-off-by: Sagi Grimberg <sagi@grimberg.me>
Reviewed-by: James Smart <jsmart2021@gmail.com>
Reviewed-by: Daniel Wagner <dwagner@suse.de>
Signed-off-by: Christoph Hellwig <hch@lst.de>

nvme: ensure subsystem reset is single threaded

The subsystem reset writes to a register, so we have to ensure the
device state is capable of handling that otherwise the driver may access
unmapped registers. Use the state machine to ensure the subsystem reset
doesn't try to write registers on a device already undergoing this type
of reset.

Link: https://bugzilla.kernel.org/show_bug.cgi?id=214771
Signed-off-by: Keith Busch <kbusch@kernel.org>
Signed-off-by: Christoph Hellwig <hch@lst.de>

nvme: restrict management ioctls to admin

The passthrough commands already have this restriction, but the other
operations do not. Require the same capabilities for all users as all of
these operations, which include resets and rescans, can be disruptive.

Signed-off-by: Keith Busch <kbusch@kernel.org>
Signed-off-by: Christoph Hellwig <hch@lst.de>

nvme: copy firmware_rev on each init

The firmware revision can change on after a reset so copy the most
recent info each time instead of just the first time, otherwise the
sysfs firmware_rev entry may contain stale data.

Reported-by: Jeff Lien <jeff.lien@wdc.com>
Signed-off-by: Keith Busch <kbusch@kernel.org>
Reviewed-by: Sagi Grimberg <sagi@grimberg.me>
Reviewed-by: Chaitanya Kulkarni <kch@nvidia.com>
Reviewed-by: Chao Leng <lengchao@huawei.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>

nvme: handle effects after freeing the request

If a reset occurs after the scan work attempts to issue a command, the
reset may quisce the admin queue, which blocks the scan work's command
from dispatching. The scan work will not be able to complete while the
queue is quiesced.

Meanwhile, the reset work will cancel all outstanding admin tags and
wait until all requests have transitioned to idle, which includes the
passthrough request. But the passthrough request won't be set to idle
until after the scan_work flushes, so we're deadlocked.

Fix this by handling the end effects after the request has been freed.

Link: https://bugzilla.kernel.org/show_bug.cgi?id=216354
Reported-by: Jonathan Derrick <Jonathan.Derrick@solidigm.com>
Signed-off-by: Keith Busch <kbusch@kernel.org>
Reviewed-by: Sagi Grimberg <sagi@grimberg.me>
Reviewed-by: Chao Leng <lengchao@huawei.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>

blk-cgroup: pass a gendisk to the blkg allocation helpers

Prepare for storing the blkcg information in the gendisk instead of
the request_queue.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Andreas Herrmann <aherrmann@suse.de>
Acked-by: Tejun Heo <tj@kernel.org>
Link: https://lore.kernel.org/r/20220921180501.1539876-18-hch@lst.de
Signed-off-by: Jens Axboe <axboe@kernel.dk>

blk-cgroup: pass a gendisk to blkcg_schedule_throttle

Pass the gendisk to blkcg_schedule_throttle as part of moving the
blk-cgroup infrastructure to be gendisk based.  Remove the unused
!BLK_CGROUP stub while we're at it.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Andreas Herrmann <aherrmann@suse.de>
Acked-by: Tejun Heo <tj@kernel.org>
Link: https://lore.kernel.org/r/20220921180501.1539876-17-hch@lst.de
Signed-off-by: Jens Axboe <axboe@kernel.dk>

blk-cgroup: pass a gendisk to blkg_destroy_all

Pass the gendisk to blkg_destroy_all as part of moving the blk-cgroup
infrastructure to be gendisk based.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Andreas Herrmann <aherrmann@suse.de>
Acked-by: Tejun Heo <tj@kernel.org>
Link: https://lore.kernel.org/r/20220921180501.1539876-16-hch@lst.de
Signed-off-by: Jens Axboe <axboe@kernel.dk>

blk-throttle: pass a gendisk to blk_throtl_cancel_bios

Pass the gendisk to blk_throtl_cancel_bios as part of moving the
blk-cgroup infrastructure to be gendisk based.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Andreas Herrmann <aherrmann@suse.de>
Acked-by: Tejun Heo <tj@kernel.org>
Link: https://lore.kernel.org/r/20220921180501.1539876-15-hch@lst.de
Signed-off-by: Jens Axboe <axboe@kernel.dk>

blk-throttle: pass a gendisk to blk_throtl_register_queue

Pass the gendisk to blk_throtl_register_queue as part of moving the
blk-cgroup infrastructure to be gendisk based.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Andreas Herrmann <aherrmann@suse.de>
Acked-by: Tejun Heo <tj@kernel.org>
Link: https://lore.kernel.org/r/20220921180501.1539876-14-hch@lst.de
Signed-off-by: Jens Axboe <axboe@kernel.dk>

blk-throttle: pass a gendisk to blk_throtl_init and blk_throtl_exit

Pass the gendisk to blk_throtl_init and blk_throtl_exit as part of moving
the blk-cgroup infrastructure to be gendisk based.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Andreas Herrmann <aherrmann@suse.de>
Acked-by: Tejun Heo <tj@kernel.org>
Link: https://lore.kernel.org/r/20220921180501.1539876-13-hch@lst.de
Signed-off-by: Jens Axboe <axboe@kernel.dk>

blk-iocost: cleanup ioc_qos_write

Use a local disk variable instead of retrieving the disk and
request_queue over and over by various means.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Andreas Herrmann <aherrmann@suse.de>
Acked-by: Tejun Heo <tj@kernel.org>
Link: https://lore.kernel.org/r/20220921180501.1539876-12-hch@lst.de
Signed-off-by: Jens Axboe <axboe@kernel.dk>

blk-iocost: pass a gendisk to blk_iocost_init

Pass the gendisk to blk_iocost_init as part of moving the blk-cgroup
infrastructure to be gendisk based.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Andreas Herrmann <aherrmann@suse.de>
Acked-by: Tejun Heo <tj@kernel.org>
Link: https://lore.kernel.org/r/20220921180501.1539876-11-hch@lst.de
Signed-off-by: Jens Axboe <axboe@kernel.dk>

blk-iocost: simplify ioc_name

Just directly dereference the disk name instead of going through multiple
hoops to find the same value.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Andreas Herrmann <aherrmann@suse.de>
Acked-by: Tejun Heo <tj@kernel.org>
Link: https://lore.kernel.org/r/20220921180501.1539876-10-hch@lst.de
Signed-off-by: Jens Axboe <axboe@kernel.dk>

blk-iolatency: pass a gendisk to blk_iolatency_init

Pass the gendisk to blk_iolatency_init as part of moving the blk-cgroup
infrastructure to be gendisk based.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Andreas Herrmann <aherrmann@suse.de>
Acked-by: Tejun Heo <tj@kernel.org>
Link: https://lore.kernel.org/r/20220921180501.1539876-9-hch@lst.de
[axboe: missed inline for blk_iolatency_init() and !CONFIG_BLK_CGROUP_IOLATENCY]
Signed-off-by: Jens Axboe <axboe@kernel.dk>

blk-ioprio: pass a gendisk to blk_ioprio_init and blk_ioprio_exit

Pass the gendisk to blk_ioprio_init and blk_ioprio_exit as part of moving
the blk-cgroup infrastructure to be gendisk based.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Andreas Herrmann <aherrmann@suse.de>
Acked-by: Tejun Heo <tj@kernel.org>
Link: https://lore.kernel.org/r/20220921180501.1539876-8-hch@lst.de
Signed-off-by: Jens Axboe <axboe@kernel.dk>

blk-cgroup: pass a gendisk to blkcg_init_queue and blkcg_exit_queue

Pass the gendisk to blkcg_init_disk and blkcg_exit_disk as part of moving
the blk-cgroup infrastructure to be gendisk based.  Also remove the
rather pointless kerneldoc comments for these internal functions with a
single caller each.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Andreas Herrmann <aherrmann@suse.de>
Acked-by: Tejun Heo <tj@kernel.org>
Link: https://lore.kernel.org/r/20220921180501.1539876-7-hch@lst.de
Signed-off-by: Jens Axboe <axboe@kernel.dk>

blk-cgroup: remove blkg_lookup_check

The combinations of an error check with an ERR_PTR return and a lookup
with a NULL return leads to ugly handling of the return values in the
callers.  Just open coding the check and the lookup is much simpler.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Andreas Herrmann <aherrmann@suse.de>
Acked-by: Tejun Heo <tj@kernel.org>
Link: https://lore.kernel.org/r/20220921180501.1539876-6-hch@lst.de
Signed-off-by: Jens Axboe <axboe@kernel.dk>

blk-cgroup: cleanup the blkg_lookup family of functions

Add a fully inlined blkg_lookup as the extra two checks aren't going
to generated a lot more code vs the call to the slowpath routine, and
open code the hint update in the two callers that care.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Andreas Herrmann <aherrmann@suse.de>
Acked-by: Tejun Heo <tj@kernel.org>
Link: https://lore.kernel.org/r/20220921180501.1539876-5-hch@lst.de
Signed-off-by: Jens Axboe <axboe@kernel.dk>

blk-cgroup: remove open coded blkg_lookup instances

Use blkg_lookup instead of open coding it.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Andreas Herrmann <aherrmann@suse.de>
Acked-by: Tejun Heo <tj@kernel.org>
Link: https://lore.kernel.org/r/20220921180501.1539876-4-hch@lst.de
Signed-off-by: Jens Axboe <axboe@kernel.dk>

blk-cgroup: remove blk_queue_root_blkg

Just open code it in the only caller and drop the unused !BLK_CGROUP
stub.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Andreas Herrmann <aherrmann@suse.de>
Acked-by: Tejun Heo <tj@kernel.org>
Link: https://lore.kernel.org/r/20220921180501.1539876-3-hch@lst.de
Signed-off-by: Jens Axboe <axboe@kernel.dk>

blk-cgroup: fix error unwinding in blkcg_init_queue

When blk_throtl_init fails, we need to call blk_ioprio_exit.  Switch to
proper goto based unwinding to fix this.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Andreas Herrmann <aherrmann@suse.de>
Acked-by: Tejun Heo <tj@kernel.org>
Link: https://lore.kernel.org/r/20220921180501.1539876-2-hch@lst.de
Signed-off-by: Jens Axboe <axboe@kernel.dk>

io_uring/rw: don't lose short results on io_setup_async_rw()

If a retry io_setup_async_rw() fails we lose result from the first
io_iter_do_read(), which is a problem mostly for streams/sockets.

Cc: stable@vger.kernel.org
Signed-off-by: Pavel Begunkov <asml.silence@gmail.com>
Link: https://lore.kernel.org/r/0e8d20cebe5fc9c96ed268463c394237daabc384.1664235732.git.asml.silence@gmail.com
Signed-off-by: Jens Axboe <axboe@kernel.dk>

io_uring/rw: fix unexpected link breakage

req->cqe.res is set in io_read() to the amount of bytes left to be done,
which is used to figure out whether to fail a read or not. However,
io_read() may do another without returning, and we stash the previous
value into ->bytes_done but forget to update cqe.res. Then we ask a read
to do strictly less than cqe.res but expect the return to be exactly
cqe.res.

Fix the bug by updating cqe.res for retries.

Cc: stable@vger.kernel.org
Reported-and-Tested-by: Beld Zhang <beldzhang@gmail.com>
Signed-off-by: Pavel Begunkov <asml.silence@gmail.com>
Link: https://lore.kernel.org/r/3a1088440c7be98e5800267af922a67da0ef9f13.1664235732.git.asml.silence@gmail.com
Signed-off-by: Jens Axboe <axboe@kernel.dk>

io_uring/net: fix cleanup double free free_iov init

Having ->async_data doesn't mean it's initialised and previously we vere
relying on setting F_CLEANUP at the right moment. With zc sendmsg
though, we set F_CLEANUP early in prep when we alloc a notif and so we
may allocate async_data, fail in copy_msg_hdr() leaving
struct io_async_msghdr not initialised correctly but with F_CLEANUP
set, which causes a ->free_iov double free and probably other nastiness.

Always initialise ->free_iov. Also, now it might point to fast_iov when
fails, so avoid freeing it during cleanups.

Reported-by: syzbot+edfd15cd4246a3fc615a@syzkaller.appspotmail.com
Fixes: 493108d95f146 ("io_uring/net: zerocopy sendmsg")
Signed-off-by: Pavel Begunkov <asml.silence@gmail.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>

Linux 6.0-rc7

Merge tag 'ext4_for_linus_stable' of git://git./linux/kernel/git/tytso/ext4

Pull ext4 fixes from Ted Ts'o:
 "Regression and bug fixes:

   - Performance regression fix from 5.18 on a Rasberry Pi

   - Fix extent parsing bug which triggers a BUG_ON when a (corrupted)
     extent tree has has a non-root node when zero entries.

   - Fix a livelock where in the right (wrong) circumstances a large
     number of nfsd threads can try to write to a nearly full file
     system, and retry for hours(!)"

* tag 'ext4_for_linus_stable' of git://git.kernel.org/pub/scm/linux/kernel/git/tytso/ext4:
  ext4: limit the number of retries after discarding preallocations blocks
  ext4: fix bug in extents parsing when eh_entries == 0 and eh_depth > 0
  ext4: use buckets for cr 1 block scan instead of rbtree
  ext4: use locality group preallocation for small closed files
  ext4: make directory inode spreading reflect flexbg size
  ext4: avoid unnecessary spreading of allocations among groups
  ext4: make mballoc try target group first even with mb_optimize_scan

Merge tag 'dax-and-nvdimm-fixes-v6.0-final' of git://git./linux/kernel/git/nvdimm/nvdimm

Pull NVDIMM and DAX fixes from Dan Williams:
 "A recently discovered one-line fix for devdax that further addresses a
  v5.5 regression, and (a bit embarrassing) a small batch of fixes that
  have been sitting in my fixes tree for weeks.

  The older fixes have soaked in linux-next during that time and address
  an fsdax infinite loop and some other minor fixups.

   - Fix a infinite loop bug in fsdax

   - Fix memory-type detection for devdax (EINJ regression)

   - Small cleanups"

* tag 'dax-and-nvdimm-fixes-v6.0-final' of git://git.kernel.org/pub/scm/linux/kernel/git/nvdimm/nvdimm:
  devdax: Fix soft-reservation memory description
  fsdax: Fix infinite loop in dax_iomap_rw()
  nvdimm/namespace: drop nested variable in create_namespace_pmem()
  ndtest: Cleanup all of blk namespace specific code
  pmem: fix a name collision

Merge tag 'i2c-for-6.0-rc7' of git://git./linux/kernel/git/wsa/linux

Pull i2c fixes from Wolfram Sang:
 "I2C driver bugfixes for mlxbf and imx, a few documentation fixes after
  the rework this cycle, and one hardening for the i2c-mux core"

* tag 'i2c-for-6.0-rc7' of git://git.kernel.org/pub/scm/linux/kernel/git/wsa/linux:
  i2c: mux: harden i2c_mux_alloc() against integer overflows
  i2c: mlxbf: Fix frequency calculation
  i2c: mlxbf: prevent stack overflow in mlxbf_i2c_smbus_start_transaction()
  i2c: mlxbf: incorrect base address passed during io write
  Documentation: i2c: fix references to other documents
  MAINTAINERS: remove Nehal Shah from AMD MP2 I2C DRIVER
  i2c: imx: If pm_runtime_get_sync() returned 1 device access is possible

Merge branch 'for-6.0/dax' into libnvdimm-fixes

Pick up another "Soft Reservation" fix for v6.0-final on top of some
straggling nvdimm fixes that missed v5.19.

devdax: Fix soft-reservation memory description

The "hmem" platform-devices that are created to represent the
platform-advertised "Soft Reserved" memory ranges end up inserting a
resource that causes the iomem_resource tree to look like this:

340000000-43fffffff : hmem.0
  340000000-43fffffff : Soft Reserved
    340000000-43fffffff : dax0.0

This is because insert_resource() reparents ranges when they completely
intersect an existing range.

This matters because code that uses region_intersects() to scan for a
given IORES_DESC will only check that top-level 'hmem.0' resource and
not the 'Soft Reserved' descendant.

So, to support EINJ (via einj_error_inject()) to inject errors into
memory hosted by a dax-device, be sure to describe the memory as
IORES_DESC_SOFT_RESERVED. This is a follow-on to:

commit b13a3e5fd40b ("ACPI: APEI: Fix _EINJ vs EFI_MEMORY_SP")

...that fixed EINJ support for "Soft Reserved" ranges in the first
instance.

Fixes: 262b45ae3ab4 ("x86/efi: EFI soft reservation to E820 enumeration")
Reported-by: Ricardo Sandoval Torres <ricardo.sandoval.torres@intel.com>
Tested-by: Ricardo Sandoval Torres <ricardo.sandoval.torres@intel.com>
Cc: <stable@vger.kernel.org>
Cc: Tony Luck <tony.luck@intel.com>
Cc: Omar Avelar <omar.avelar@intel.com>
Cc: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Cc: Mark Gross <markgross@kernel.org>
Link: https://lore.kernel.org/r/166397075670.389916.7435722208896316387.stgit@dwillia2-xfh.jf.intel.com
Signed-off-by: Dan Williams <dan.j.williams@intel.com>

Merge tag 'kbuild-fixes-v6.0-3' of git://git./linux/kernel/git/masahiroy/linux-kbuild

Pull Kbuild fixes from Masahiro Yamada:

 - Fix build error for the combination of SYSTEM_TRUSTED_KEYRING=y and
   X509_CERTIFICATE_PARSER=m

 - Fix DEBUG_INFO_SPLIT to generate debug info for GCC 11+ and Clang 12+

 - Revive debug info for assembly files

 - Remove unused code

* tag 'kbuild-fixes-v6.0-3' of git://git.kernel.org/pub/scm/linux/kernel/git/masahiroy/linux-kbuild:
  Makefile.debug: re-enable debug info for .S files
  Makefile.debug: set -g unconditional on CONFIG_DEBUG_INFO_SPLIT
  certs: make system keyring depend on built-in x509 parser
  Kconfig: remove unused function 'menu_get_root_menu'
  scripts/clang-tools: remove unused module

Merge tag 's390-6.0-5' of git://git./linux/kernel/git/s390/linux

Pull s390 fix from Vasily Gorbik:

 - Fix potential hangs in VFIO AP driver

* tag 's390-6.0-5' of git://git.kernel.org/pub/scm/linux/kernel/git/s390/linux:
  s390/vfio-ap: bypass unnecessary processing of AP resources

Merge tag 'pm-6.0-rc7' of git://git./linux/kernel/git/rafael/linux-pm

Pull power management fixes from Rafael Wysocki:
 "These fix an uninitialized variable usage in the operating performance
  points code and add missing DT bindings for it.

  Specifics:

   - Fix uninitialized variable usage in dev_pm_opp_config_clks_simple()
     (Christophe JAILLET)

   - Add missing OPP DT properties (Rob Herring)"

* tag 'pm-6.0-rc7' of git://git.kernel.org/pub/scm/linux/kernel/git/rafael/linux-pm:
  dt-bindings: opp: Add missing (unevaluated|additional)Properties on child nodes
  OPP: Fix an un-initialized variable usage

Merge tag 'char-misc-6.0-rc7' of git://git./linux/kernel/git/gregkh/char-misc

Pull char/misc driver fixes from Greg KH:
 "Here are three tiny driver fixes for 6.0-rc7.  They include:

   - phy driver reset bugfix

   - fpga memleak bugfix

   - counter irq config bugfix

  The first two have been in linux-next for a while, the last one has
  only been added to my tree in the past few days, but was in linux-next
  under a different commit id. I couldn't pull directly from the counter
  tree due to some gpg key propagation issue, so I took the commit
  directly from email instead"

* tag 'char-misc-6.0-rc7' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/char-misc:
  counter: 104-quad-8: Fix skipped IRQ lines during events configuration
  fpga: m10bmc-sec: Fix possible memory leak of flash_buf
  phy: marvell: phy-mvebu-a3700-comphy: Remove broken reset support

Merge tag 'tty-6.0-rc7' of git://git./linux/kernel/git/gregkh/tty

Pull tty/serial driver fixes from Greg KH:
 "Here are some small, and late, serial driver fixes for 6.0-rc7 to
  resolve some reported problems.

  Included in here are:

   - tegra icount accounting fixes, including a framework function that
     other drivers will be converted over to using in 6.1-rc1.

   - fsl_lpuart reset bugfix

   - 8250 omap 485 bugfix

   - sifive serial clock bugfix

  The last three patches have not shown up in linux-next due to them
  being added to my tree only 2 days ago, but they are tiny and
  self-contained and the developers say they resolve issues that they
  have with 6.0-rc. The other three have been in linux-next for a while
  with no reported issues"

* tag 'tty-6.0-rc7' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/tty:
  serial: sifive: enable clocks for UART when probed
  serial: 8250: omap: Use serial8250_em485_supported
  serial: fsl_lpuart: Reset prior to registration
  serial: tegra-tcu: Use uart_xmit_advance(), fixes icount.tx accounting
  serial: tegra: Use uart_xmit_advance(), fixes icount.tx accounting
  serial: Create uart_xmit_advance()

Merge tag 'cgroup-for-6.0-rc6-fixes' of git://git./linux/kernel/git/tj/cgroup

Pull cgroup fixes from Tejun Heo:

 - Add Waiman Long as a cpuset maintainer

 - cgroup_get_from_id() could be fed a kernfs ID which doesn't point to
   a cgroup directory but a knob file and then crash. Error out if the
   lookup kernfs_node isn't a directory.

* tag 'cgroup-for-6.0-rc6-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/tj/cgroup:
  cgroup: cgroup_get_from_id() must check the looked-up kn is a directory
  cpuset: Add Waiman Long as a cpuset maintainer

Merge tag 'wq-for-6.0-rc6-fixes' of git://git./linux/kernel/git/tj/wq

Pull workqueue fix from Tejun Heo:
 "Just one patch to improve flush lockdep coverage"

* tag 'wq-for-6.0-rc6-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/tj/wq:
  workqueue: don't skip lockdep work dependency in cancel_work_sync()

Merge tag 'io_uring-6.0-2022-09-23' of git://git.kernel.dk/linux

Pull io_uring fix from Jens Axboe:
 "Just a single fix for an issue with un-reaped IOPOLL requests on ring
  exit"

* tag 'io_uring-6.0-2022-09-23' of git://git.kernel.dk/linux:
  io_uring: ensure that cached task references are always put on exit

Merge tag 'block-6.0-2022-09-22' of git://git.kernel.dk/linux

Pull block fixes from Jens Axboe:
 "Fix a regression that's been plaguing us by reverting the offending
  commit, as attempts to both reproduce the issue and fix it in a saner
  fashion have failed.

  Fix for a potential oops condition in the s390 dasd block driver"

* tag 'block-6.0-2022-09-22' of git://git.kernel.dk/linux:
  Revert "block: freeze the queue earlier in del_gendisk"
  s390/dasd: fix Oops in dasd_alias_get_start_dev due to missing pavgroup

blk-mq: don't redirect completion for hctx withs only one ctx mapping

High-performance NVMe devices usually support a large hw queues, which
ensures a 1:1 mapping of hctx and ctx. In this case there will be no
remote request, so we don't need to care about it.

Signed-off-by: Liu Song <liusong@linux.alibaba.com>
Link: https://lore.kernel.org/r/1663731123-81536-1-git-send-email-liusong@linux.alibaba.com
Signed-off-by: Jens Axboe <axboe@kernel.dk>