* FAILED: patch "[PATCH] io_uring: lock overflowing for IOPOLL" failed to apply to 5.15-stable tree
@ 2023-01-14 9:51 gregkh
2023-01-14 16:15 ` Jens Axboe
0 siblings, 1 reply; 5+ messages in thread
From: gregkh @ 2023-01-14 9:51 UTC (permalink / raw)
To: asml.silence, axboe; +Cc: stable
The patch below does not apply to the 5.15-stable tree.
If someone wants it applied there, or to any other stable or longterm
tree, then please email the backport, including the original git commit
id to <stable@vger.kernel.org>.
Possible dependencies:
544d163d659d ("io_uring: lock overflowing for IOPOLL")
a8cf95f93610 ("io_uring: fix overflow handling regression")
fa18fa2272c7 ("io_uring: inline __io_req_complete_put()")
f9d567c75ec2 ("io_uring: inline __io_req_complete_post()")
52120f0fadcb ("io_uring: add allow_overflow to io_post_aux_cqe")
e6130eba8a84 ("io_uring: add support for passing fixed file descriptors")
253993210bd8 ("io_uring: introduce locking helpers for CQE posting")
305bef988708 ("io_uring: hide eventfd assumptions in eventfd paths")
d9dee4302a7c ("io_uring: remove ->flush_cqes optimisation")
a830ffd28780 ("io_uring: move io_eventfd_signal()")
9046c6415be6 ("io_uring: reshuffle io_uring/io_uring.h")
d142c3ec8d16 ("io_uring: remove extra io_commit_cqring()")
68494a65d0e2 ("io_uring: introduce io_req_cqe_overflow()")
faf88dde060f ("io_uring: don't inline __io_get_cqe()")
d245bca6375b ("io_uring: don't expose io_fill_cqe_aux()")
aa1e90f64ee5 ("io_uring: move small helpers to headers")
aa1e90f64ee5 ("io_uring: move small helpers to headers")
aa1e90f64ee5 ("io_uring: move small helpers to headers")
aa1e90f64ee5 ("io_uring: move small helpers to headers")
aa1e90f64ee5 ("io_uring: move small helpers to headers")
thanks,
greg k-h
------------------ original commit in Linus's tree ------------------
From 544d163d659d45a206d8929370d5a2984e546cb7 Mon Sep 17 00:00:00 2001
From: Pavel Begunkov <asml.silence@gmail.com>
Date: Thu, 12 Jan 2023 13:08:56 +0000
Subject: [PATCH] io_uring: lock overflowing for IOPOLL
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit
syzbot reports an issue with overflow filling for IOPOLL:
WARNING: CPU: 0 PID: 28 at io_uring/io_uring.c:734 io_cqring_event_overflow+0x1c0/0x230 io_uring/io_uring.c:734
CPU: 0 PID: 28 Comm: kworker/u4:1 Not tainted 6.2.0-rc3-syzkaller-16369-g358a161a6a9e #0
Workqueue: events_unbound io_ring_exit_work
Call trace:
io_cqring_event_overflow+0x1c0/0x230 io_uring/io_uring.c:734
io_req_cqe_overflow+0x5c/0x70 io_uring/io_uring.c:773
io_fill_cqe_req io_uring/io_uring.h:168 [inline]
io_do_iopoll+0x474/0x62c io_uring/rw.c:1065
io_iopoll_try_reap_events+0x6c/0x108 io_uring/io_uring.c:1513
io_uring_try_cancel_requests+0x13c/0x258 io_uring/io_uring.c:3056
io_ring_exit_work+0xec/0x390 io_uring/io_uring.c:2869
process_one_work+0x2d8/0x504 kernel/workqueue.c:2289
worker_thread+0x340/0x610 kernel/workqueue.c:2436
kthread+0x12c/0x158 kernel/kthread.c:376
ret_from_fork+0x10/0x20 arch/arm64/kernel/entry.S:863
There is no real problem for normal IOPOLL as flush is also called with
uring_lock taken, but it's getting more complicated for IOPOLL|SQPOLL,
for which __io_cqring_overflow_flush() happens from the CQ waiting path.
Reported-and-tested-by: syzbot+6805087452d72929404e@syzkaller.appspotmail.com
Cc: stable@vger.kernel.org # 5.10+
Signed-off-by: Pavel Begunkov <asml.silence@gmail.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
diff --git a/io_uring/rw.c b/io_uring/rw.c
index 8227af2e1c0f..9c3ddd46a1ad 100644
--- a/io_uring/rw.c
+++ b/io_uring/rw.c
@@ -1062,7 +1062,11 @@ int io_do_iopoll(struct io_ring_ctx *ctx, bool force_nonspin)
continue;
req->cqe.flags = io_put_kbuf(req, 0);
- io_fill_cqe_req(req->ctx, req);
+ if (unlikely(!__io_fill_cqe_req(ctx, req))) {
+ spin_lock(&ctx->completion_lock);
+ io_req_cqe_overflow(req);
+ spin_unlock(&ctx->completion_lock);
+ }
}
if (unlikely(!nr_events))
^ permalink raw reply related [flat|nested] 5+ messages in thread* Re: FAILED: patch "[PATCH] io_uring: lock overflowing for IOPOLL" failed to apply to 5.15-stable tree 2023-01-14 9:51 FAILED: patch "[PATCH] io_uring: lock overflowing for IOPOLL" failed to apply to 5.15-stable tree gregkh @ 2023-01-14 16:15 ` Jens Axboe 2023-01-14 16:21 ` Jens Axboe 0 siblings, 1 reply; 5+ messages in thread From: Jens Axboe @ 2023-01-14 16:15 UTC (permalink / raw) To: gregkh, asml.silence; +Cc: stable [-- Attachment #1: Type: text/plain, Size: 479 bytes --] On 1/14/23 2:51 AM, gregkh@linuxfoundation.org wrote: > > The patch below does not apply to the 5.15-stable tree. > If someone wants it applied there, or to any other stable or longterm > tree, then please email the backport, including the original git commit > id to <stable@vger.kernel.org>. This has to be done a bit differently, but this one should work. I tested it on 5.10-stable, but should apply to 5.15-stable as well as they are the same base now. -- Jens Axboe [-- Attachment #2: 0001-io_uring-lock-overflowing-for-IOPOLL.patch --] [-- Type: text/x-patch, Size: 2784 bytes --] From 313cc3d4c3f6a1c6d0df525f0368a2c5a7ce97c8 Mon Sep 17 00:00:00 2001 From: Pavel Begunkov <asml.silence@gmail.com> Date: Sat, 14 Jan 2023 09:14:03 -0700 Subject: [PATCH] io_uring: lock overflowing for IOPOLL MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit commit 544d163d659d45a206d8929370d5a2984e546cb7 upstream. syzbot reports an issue with overflow filling for IOPOLL: WARNING: CPU: 0 PID: 28 at io_uring/io_uring.c:734 io_cqring_event_overflow+0x1c0/0x230 io_uring/io_uring.c:734 CPU: 0 PID: 28 Comm: kworker/u4:1 Not tainted 6.2.0-rc3-syzkaller-16369-g358a161a6a9e #0 Workqueue: events_unbound io_ring_exit_work Call trace: io_cqring_event_overflow+0x1c0/0x230 io_uring/io_uring.c:734 io_req_cqe_overflow+0x5c/0x70 io_uring/io_uring.c:773 io_fill_cqe_req io_uring/io_uring.h:168 [inline] io_do_iopoll+0x474/0x62c io_uring/rw.c:1065 io_iopoll_try_reap_events+0x6c/0x108 io_uring/io_uring.c:1513 io_uring_try_cancel_requests+0x13c/0x258 io_uring/io_uring.c:3056 io_ring_exit_work+0xec/0x390 io_uring/io_uring.c:2869 process_one_work+0x2d8/0x504 kernel/workqueue.c:2289 worker_thread+0x340/0x610 kernel/workqueue.c:2436 kthread+0x12c/0x158 kernel/kthread.c:376 ret_from_fork+0x10/0x20 arch/arm64/kernel/entry.S:863 There is no real problem for normal IOPOLL as flush is also called with uring_lock taken, but it's getting more complicated for IOPOLL|SQPOLL, for which __io_cqring_overflow_flush() happens from the CQ waiting path. Reported-and-tested-by: syzbot+6805087452d72929404e@syzkaller.appspotmail.com Cc: stable@vger.kernel.org # 5.10+ Signed-off-by: Pavel Begunkov <asml.silence@gmail.com> Signed-off-by: Jens Axboe <axboe@kernel.dk> --- io_uring/io_uring.c | 16 +++++++++++++++- 1 file changed, 15 insertions(+), 1 deletion(-) diff --git a/io_uring/io_uring.c b/io_uring/io_uring.c index 0c4d16afb9ef..1a3f360af373 100644 --- a/io_uring/io_uring.c +++ b/io_uring/io_uring.c @@ -2478,10 +2478,24 @@ static void io_iopoll_complete(struct io_ring_ctx *ctx, unsigned int *nr_events, io_init_req_batch(&rb); while (!list_empty(done)) { + struct io_uring_cqe *cqe; + unsigned cflags; + req = list_first_entry(done, struct io_kiocb, inflight_entry); list_del(&req->inflight_entry); + cflags = io_put_rw_kbuf(req); - io_fill_cqe_req(req, req->result, io_put_rw_kbuf(req)); + cqe = io_get_cqe(ctx); + if (unlikely(!cqe)) { + spin_lock(&ctx->completion_lock); + io_cqring_event_overflow(ctx, req->user_data, + req->result, cflags); + spin_unlock(&ctx->completion_lock); + continue; + } + WRITE_ONCE(cqe->user_data, req->user_data); + WRITE_ONCE(cqe->res, req->result); + WRITE_ONCE(cqe->flags, cflags); (*nr_events)++; if (req_ref_put_and_test(req)) -- 2.39.0 ^ permalink raw reply related [flat|nested] 5+ messages in thread
* Re: FAILED: patch "[PATCH] io_uring: lock overflowing for IOPOLL" failed to apply to 5.15-stable tree 2023-01-14 16:15 ` Jens Axboe @ 2023-01-14 16:21 ` Jens Axboe 2023-01-14 16:48 ` Jens Axboe 0 siblings, 1 reply; 5+ messages in thread From: Jens Axboe @ 2023-01-14 16:21 UTC (permalink / raw) To: gregkh, asml.silence; +Cc: stable [-- Attachment #1: Type: text/plain, Size: 619 bytes --] On 1/14/23 9:15 AM, Jens Axboe wrote: > On 1/14/23 2:51 AM, gregkh@linuxfoundation.org wrote: >> >> The patch below does not apply to the 5.15-stable tree. >> If someone wants it applied there, or to any other stable or longterm >> tree, then please email the backport, including the original git commit >> id to <stable@vger.kernel.org>. > > This has to be done a bit differently, but this one should work. I tested > it on 5.10-stable, but should apply to 5.15-stable as well as they are > the same base now. Oops, missed accounting for overflow. Please use this one instead! Sorry about that. -- Jens Axboe [-- Attachment #2: 0001-io_uring-lock-overflowing-for-IOPOLL.patch --] [-- Type: text/x-patch, Size: 2846 bytes --] From 929e5c54ff2ad116e9dcee967fc5df1dbb6b1d90 Mon Sep 17 00:00:00 2001 From: Pavel Begunkov <asml.silence@gmail.com> Date: Sat, 14 Jan 2023 09:14:03 -0700 Subject: [PATCH] io_uring: lock overflowing for IOPOLL MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit commit 544d163d659d45a206d8929370d5a2984e546cb7 upstream. syzbot reports an issue with overflow filling for IOPOLL: WARNING: CPU: 0 PID: 28 at io_uring/io_uring.c:734 io_cqring_event_overflow+0x1c0/0x230 io_uring/io_uring.c:734 CPU: 0 PID: 28 Comm: kworker/u4:1 Not tainted 6.2.0-rc3-syzkaller-16369-g358a161a6a9e #0 Workqueue: events_unbound io_ring_exit_work Call trace: io_cqring_event_overflow+0x1c0/0x230 io_uring/io_uring.c:734 io_req_cqe_overflow+0x5c/0x70 io_uring/io_uring.c:773 io_fill_cqe_req io_uring/io_uring.h:168 [inline] io_do_iopoll+0x474/0x62c io_uring/rw.c:1065 io_iopoll_try_reap_events+0x6c/0x108 io_uring/io_uring.c:1513 io_uring_try_cancel_requests+0x13c/0x258 io_uring/io_uring.c:3056 io_ring_exit_work+0xec/0x390 io_uring/io_uring.c:2869 process_one_work+0x2d8/0x504 kernel/workqueue.c:2289 worker_thread+0x340/0x610 kernel/workqueue.c:2436 kthread+0x12c/0x158 kernel/kthread.c:376 ret_from_fork+0x10/0x20 arch/arm64/kernel/entry.S:863 There is no real problem for normal IOPOLL as flush is also called with uring_lock taken, but it's getting more complicated for IOPOLL|SQPOLL, for which __io_cqring_overflow_flush() happens from the CQ waiting path. Reported-and-tested-by: syzbot+6805087452d72929404e@syzkaller.appspotmail.com Cc: stable@vger.kernel.org # 5.10+ Signed-off-by: Pavel Begunkov <asml.silence@gmail.com> Signed-off-by: Jens Axboe <axboe@kernel.dk> --- io_uring/io_uring.c | 18 ++++++++++++++++-- 1 file changed, 16 insertions(+), 2 deletions(-) diff --git a/io_uring/io_uring.c b/io_uring/io_uring.c index 0c4d16afb9ef..c3838a2c5c6f 100644 --- a/io_uring/io_uring.c +++ b/io_uring/io_uring.c @@ -2478,12 +2478,26 @@ static void io_iopoll_complete(struct io_ring_ctx *ctx, unsigned int *nr_events, io_init_req_batch(&rb); while (!list_empty(done)) { + struct io_uring_cqe *cqe; + unsigned cflags; + req = list_first_entry(done, struct io_kiocb, inflight_entry); list_del(&req->inflight_entry); - - io_fill_cqe_req(req, req->result, io_put_rw_kbuf(req)); + cflags = io_put_rw_kbuf(req); (*nr_events)++; + cqe = io_get_cqe(ctx); + if (unlikely(!cqe)) { + spin_lock(&ctx->completion_lock); + io_cqring_event_overflow(ctx, req->user_data, + req->result, cflags); + spin_unlock(&ctx->completion_lock); + continue; + } + WRITE_ONCE(cqe->user_data, req->user_data); + WRITE_ONCE(cqe->res, req->result); + WRITE_ONCE(cqe->flags, cflags); + if (req_ref_put_and_test(req)) io_req_free_batch(&rb, req, &ctx->submit_state); } -- 2.39.0 ^ permalink raw reply related [flat|nested] 5+ messages in thread
* Re: FAILED: patch "[PATCH] io_uring: lock overflowing for IOPOLL" failed to apply to 5.15-stable tree 2023-01-14 16:21 ` Jens Axboe @ 2023-01-14 16:48 ` Jens Axboe 2023-01-16 14:47 ` Greg KH 0 siblings, 1 reply; 5+ messages in thread From: Jens Axboe @ 2023-01-14 16:48 UTC (permalink / raw) To: gregkh, asml.silence; +Cc: stable [-- Attachment #1: Type: text/plain, Size: 795 bytes --] On 1/14/23 9:21?AM, Jens Axboe wrote: > On 1/14/23 9:15?AM, Jens Axboe wrote: >> On 1/14/23 2:51?AM, gregkh@linuxfoundation.org wrote: >>> >>> The patch below does not apply to the 5.15-stable tree. >>> If someone wants it applied there, or to any other stable or longterm >>> tree, then please email the backport, including the original git commit >>> id to <stable@vger.kernel.org>. >> >> This has to be done a bit differently, but this one should work. I tested >> it on 5.10-stable, but should apply to 5.15-stable as well as they are >> the same base now. > > Oops, missed accounting for overflow. Please use this one instead! Sorry > about that. Not batting 1000 on this one today. Wrote a test case for this specifically now and verified it, this one is good. For real. -- Jens Axboe [-- Attachment #2: 0001-io_uring-lock-overflowing-for-IOPOLL.patch --] [-- Type: text/x-patch, Size: 2836 bytes --] From 31f897396bbbd37010260119293e84007736dbc7 Mon Sep 17 00:00:00 2001 From: Pavel Begunkov <asml.silence@gmail.com> Date: Sat, 14 Jan 2023 09:14:03 -0700 Subject: [PATCH] io_uring: lock overflowing for IOPOLL MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit commit 544d163d659d45a206d8929370d5a2984e546cb7 upstream. syzbot reports an issue with overflow filling for IOPOLL: WARNING: CPU: 0 PID: 28 at io_uring/io_uring.c:734 io_cqring_event_overflow+0x1c0/0x230 io_uring/io_uring.c:734 CPU: 0 PID: 28 Comm: kworker/u4:1 Not tainted 6.2.0-rc3-syzkaller-16369-g358a161a6a9e #0 Workqueue: events_unbound io_ring_exit_work Call trace: io_cqring_event_overflow+0x1c0/0x230 io_uring/io_uring.c:734 io_req_cqe_overflow+0x5c/0x70 io_uring/io_uring.c:773 io_fill_cqe_req io_uring/io_uring.h:168 [inline] io_do_iopoll+0x474/0x62c io_uring/rw.c:1065 io_iopoll_try_reap_events+0x6c/0x108 io_uring/io_uring.c:1513 io_uring_try_cancel_requests+0x13c/0x258 io_uring/io_uring.c:3056 io_ring_exit_work+0xec/0x390 io_uring/io_uring.c:2869 process_one_work+0x2d8/0x504 kernel/workqueue.c:2289 worker_thread+0x340/0x610 kernel/workqueue.c:2436 kthread+0x12c/0x158 kernel/kthread.c:376 ret_from_fork+0x10/0x20 arch/arm64/kernel/entry.S:863 There is no real problem for normal IOPOLL as flush is also called with uring_lock taken, but it's getting more complicated for IOPOLL|SQPOLL, for which __io_cqring_overflow_flush() happens from the CQ waiting path. Reported-and-tested-by: syzbot+6805087452d72929404e@syzkaller.appspotmail.com Cc: stable@vger.kernel.org # 5.10+ Signed-off-by: Pavel Begunkov <asml.silence@gmail.com> Signed-off-by: Jens Axboe <axboe@kernel.dk> --- io_uring/io_uring.c | 18 ++++++++++++++++-- 1 file changed, 16 insertions(+), 2 deletions(-) diff --git a/io_uring/io_uring.c b/io_uring/io_uring.c index 0c4d16afb9ef..cb83ba50b017 100644 --- a/io_uring/io_uring.c +++ b/io_uring/io_uring.c @@ -2478,12 +2478,26 @@ static void io_iopoll_complete(struct io_ring_ctx *ctx, unsigned int *nr_events, io_init_req_batch(&rb); while (!list_empty(done)) { + struct io_uring_cqe *cqe; + unsigned cflags; + req = list_first_entry(done, struct io_kiocb, inflight_entry); list_del(&req->inflight_entry); - - io_fill_cqe_req(req, req->result, io_put_rw_kbuf(req)); + cflags = io_put_rw_kbuf(req); (*nr_events)++; + cqe = io_get_cqe(ctx); + if (cqe) { + WRITE_ONCE(cqe->user_data, req->user_data); + WRITE_ONCE(cqe->res, req->result); + WRITE_ONCE(cqe->flags, cflags); + } else { + spin_lock(&ctx->completion_lock); + io_cqring_event_overflow(ctx, req->user_data, + req->result, cflags); + spin_unlock(&ctx->completion_lock); + } + if (req_ref_put_and_test(req)) io_req_free_batch(&rb, req, &ctx->submit_state); } -- 2.39.0 ^ permalink raw reply related [flat|nested] 5+ messages in thread
* Re: FAILED: patch "[PATCH] io_uring: lock overflowing for IOPOLL" failed to apply to 5.15-stable tree 2023-01-14 16:48 ` Jens Axboe @ 2023-01-16 14:47 ` Greg KH 0 siblings, 0 replies; 5+ messages in thread From: Greg KH @ 2023-01-16 14:47 UTC (permalink / raw) To: Jens Axboe; +Cc: asml.silence, stable On Sat, Jan 14, 2023 at 09:48:34AM -0700, Jens Axboe wrote: > On 1/14/23 9:21?AM, Jens Axboe wrote: > > On 1/14/23 9:15?AM, Jens Axboe wrote: > >> On 1/14/23 2:51?AM, gregkh@linuxfoundation.org wrote: > >>> > >>> The patch below does not apply to the 5.15-stable tree. > >>> If someone wants it applied there, or to any other stable or longterm > >>> tree, then please email the backport, including the original git commit > >>> id to <stable@vger.kernel.org>. > >> > >> This has to be done a bit differently, but this one should work. I tested > >> it on 5.10-stable, but should apply to 5.15-stable as well as they are > >> the same base now. > > > > Oops, missed accounting for overflow. Please use this one instead! Sorry > > about that. > > Not batting 1000 on this one today. Wrote a test case for this > specifically now and verified it, this one is good. For real. Looks like Sasha grabbed all of these already, thanks for the backports! greg k-h ^ permalink raw reply [flat|nested] 5+ messages in thread
end of thread, other threads:[~2023-01-16 14:58 UTC | newest] Thread overview: 5+ messages (download: mbox.gz follow: Atom feed -- links below jump to the message on this page -- 2023-01-14 9:51 FAILED: patch "[PATCH] io_uring: lock overflowing for IOPOLL" failed to apply to 5.15-stable tree gregkh 2023-01-14 16:15 ` Jens Axboe 2023-01-14 16:21 ` Jens Axboe 2023-01-14 16:48 ` Jens Axboe 2023-01-16 14:47 ` Greg KH
This is an external index of several public inboxes, see mirroring instructions on how to clone and mirror all data and code used by this external index.