mirror/linux
1
0
Fork 0
mirror of https://github.com/torvalds/linux.git synced 2024-11-22 04:38:03 +00:00
Code Issues Actions Packages Projects Releases Wiki Activity

io_uring: fix io_queue_proc modifying req->flags

Browse Source 
With multiple poll entries __io_queue_proc() might be running in
parallel with poll handlers and possibly task_work, we should not be
carelessly modifying req->flags there. io_poll_double_prepare() handles
a similar case with locking but it's much easier to move it into
__io_arm_poll_handler().

Cc: stable@vger.kernel.org
Fixes: 595e52284d ("io_uring/poll: don't enable lazy wake for POLLEXCLUSIVE")
Signed-off-by: Pavel Begunkov <asml.silence@gmail.com>
Link: https://lore.kernel.org/r/455cc49e38cf32026fa1b49670be8c162c2cb583.1709834755.git.asml.silence@gmail.com
Signed-off-by: Jens Axboe <axboe@kernel.dk>
This commit is contained in:
Pavel Begunkov 2024-03-07 18:06:32 +00:00 committed by Jens Axboe
parent 70581dcd06
commit 1a8ec63b2b
1 changed files with 11 additions and 8 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

19
io_uring/poll.c
View File

@ -540,14 +540,6 @@ static void __io_queue_proc(struct io_poll *poll, struct io_poll_table *pt,
poll->wait.private = (void *) wqe_private; poll->wait.private = (void *) wqe_private;
if (poll->events & EPOLLEXCLUSIVE) { if (poll->events & EPOLLEXCLUSIVE) {
/*
* Exclusive waits may only wake a limited amount of entries
* rather than all of them, this may interfere with lazy
* wake if someone does wait(events > 1). Ensure we don't do
* lazy wake for those, as we need to process each one as they
* come in.
*/
req->flags |= REQ_F_POLL_NO_LAZY;
add_wait_queue_exclusive(head, &poll->wait); add_wait_queue_exclusive(head, &poll->wait);
} else { } else {
add_wait_queue(head, &poll->wait); add_wait_queue(head, &poll->wait);
@ -616,6 +608,17 @@ static int __io_arm_poll_handler(struct io_kiocb *req,
if (issue_flags & IO_URING_F_UNLOCKED) if (issue_flags & IO_URING_F_UNLOCKED)
req->flags &= ~REQ_F_HASH_LOCKED; req->flags &= ~REQ_F_HASH_LOCKED;
/*
* Exclusive waits may only wake a limited amount of entries
* rather than all of them, this may interfere with lazy
* wake if someone does wait(events > 1). Ensure we don't do
* lazy wake for those, as we need to process each one as they
* come in.
*/
if (poll->events & EPOLLEXCLUSIVE)
req->flags |= REQ_F_POLL_NO_LAZY;
mask = vfs_poll(req->file, &ipt->pt) & poll->events; mask = vfs_poll(req->file, &ipt->pt) & poll->events;
if (unlikely(ipt->error || !ipt->nr_entries)) { if (unlikely(ipt->error || !ipt->nr_entries)) {