io_uring: fix __io_iopoll_check deadlock in io_sq_thread
PATCH 1/1 io_uring: fix __io_iopoll_check deadlock in io_sq_thread
Since commit a3a0e43fd770 ("io_uring: don't enter poll loop if we have CQEs pending"), if we already events pending, we won't enter poll loop. In case SETUP_IOPOLL and SETUP_SQPOLL are both enabled, if app has been terminated and don't reap pengding events which are already in cq ring, and there are some reqs in poll_list, io_sq_thread will enter __io_iopoll_check(), and find pending events, then return, this loop will never have a chance to exit.
I have seem this issue in fio stress tests, to fix this issue, let io_sq_thread call io_iopoll_getevents() with argument 'min' being zero, and remove __io_iopoll_check().
--- fs/io_uring.c | 27 +++++++++------------------ 1 file changed, 9 insertions(+), 18 deletions(-)
fs/io_uring.c +9 -18
@@ -1635,11 +1635,17 @@ static void io_iopoll_reap_events(struct io_ring_ctx *ctx)mutex_unlock (&ctx->uring_lock); } -static int __io_iopoll_check (struct io_ring_ctx *ctx,unsigned *nr_events, -long min) +static int io_iopoll_check (struct io_ring_ctx *ctx,unsigned *nr_events, +long min) {int iters =0 , ret =0 ; +/* +* We disallow the app entering submit/complete with polling, but we +* still need to lock the ring to prevent racing with polled issue +* that got punted to a workqueue. +*/ +mutex_lock (&ctx->uring_lock); do { int tmin = 0;
@@ -1675,21 +1681,6 @@ static int __io_iopoll_check(struct io_ring_ctx *ctx, unsigned *nr_events, ret =0 ; }while (min && !*nr_events && !need_resched ()); -return ret; -} - -static int io_iopoll_check (struct io_ring_ctx *ctx,unsigned *nr_events, -long min) -{ -int ret; - -/* -* We disallow the app entering submit/complete with polling, but we -* still need to lock the ring to prevent racing with polled issue -* that got punted to a workqueue. -*/ -mutex_lock (&ctx->uring_lock); - ret =__io_iopoll_check (ctx, nr_events, min);mutex_unlock (&ctx->uring_lock);return ret; }
@@ -5039,7 +5030,7 @@ static int io_sq_thread(void *data)*/ mutex_lock (&ctx->uring_lock);if (!list_empty (&ctx->poll_list)) -__io_iopoll_check (ctx, &nr_events,0 ); +io_iopoll_getevents (ctx, &nr_events,0 );else inflight =0 ;mutex_unlock (&ctx->uring_lock);
Signature · 1 line
2.17.2
Jens Axboe
maintainerQuoted 11 lines · Xiaoguang Wang
Since commit
a3a0e43fd770("io_uring: don't enter poll loop if we have CQEs pending"), if we already events pending, we won't enter poll loop. In caseSETUP_IOPOLLandSETUP_SQPOLLare both enabled, if app has been terminated and don't reap pengding events which are already in cq ring, and there are some reqs inpoll_list,io_sq_threadwill enter__io_iopoll_check(), and find pending events, then return, this loop will never have a chance to exit.I have seem this issue in fio stress tests, to fix this issue, let
io_sq_threadcallio_iopoll_getevents()with argument 'min' being zero, and remove__io_iopoll_check().
Thanks, nice and clean fix. Applied for 5.6.
Signature · 1 line
Jens Axboe