From: Greg Kroah-Hartman gregkh@linuxfoundation.org
From: Jens Axboe axboe@kernel.dk
[ Upstream commit d052d1d685f5125249ab4ff887562c88ba959638 ]
We bypass IOPOLL completion polling (and reaping) for the SQPOLL thread, but if it's the thread itself invoking cancelations, then we still need to perform it or no one will.
Fixes: 9936c7c2bc76 ("io_uring: deduplicate core cancellations sequence") Signed-off-by: Jens Axboe axboe@kernel.dk Signed-off-by: Sasha Levin sashal@kernel.org --- fs/io_uring.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-)
diff --git a/fs/io_uring.c b/fs/io_uring.c index 241313278e5a..00ef0b90d149 100644 --- a/fs/io_uring.c +++ b/fs/io_uring.c @@ -8891,7 +8891,8 @@ static void io_uring_try_cancel_requests(struct io_ring_ctx *ctx, }
/* SQPOLL thread does its own polling */ - if (!(ctx->flags & IORING_SETUP_SQPOLL) && !files) { + if ((!(ctx->flags & IORING_SETUP_SQPOLL) && !files) || + (ctx->sq_data && ctx->sq_data->thread == current)) { while (!list_empty_careful(&ctx->iopoll_list)) { io_iopoll_try_reap_events(ctx); ret = true;