From: Pavel Begunkov asml.silence@gmail.com
commit 9ffa13ff78a0a55df968a72d6f0ebffccee5c9f4 upstream.
Unlike normal tw, nothing prevents deferred tw to be executed right after an tw item added to ->work_llist in io_req_local_work_add(). For instance, the waiting task may get waken up by CQ posting or a normal tw. Thus we need to pin the ring for the rest of io_req_local_work_add()
Cc: stable@vger.kernel.org Fixes: c0e0d6ba25f18 ("io_uring: add IORING_SETUP_DEFER_TASKRUN") Signed-off-by: Pavel Begunkov asml.silence@gmail.com Link: https://lore.kernel.org/r/1a79362b9c10b8523ef70b061d96523650a23344.167279599... Signed-off-by: Jens Axboe axboe@kernel.dk Signed-off-by: Greg Kroah-Hartman gregkh@linuxfoundation.org --- io_uring/io_uring.c | 8 +++++++- 1 file changed, 7 insertions(+), 1 deletion(-)
--- a/io_uring/io_uring.c +++ b/io_uring/io_uring.c @@ -1109,13 +1109,18 @@ static void io_req_local_work_add(struct { struct io_ring_ctx *ctx = req->ctx;
- if (!llist_add(&req->io_task_work.node, &ctx->work_llist)) + percpu_ref_get(&ctx->refs); + + if (!llist_add(&req->io_task_work.node, &ctx->work_llist)) { + percpu_ref_put(&ctx->refs); return; + } /* need it for the following io_cqring_wake() */ smp_mb__after_atomic();
if (unlikely(atomic_read(&req->task->io_uring->in_idle))) { io_move_task_work_from_local(ctx); + percpu_ref_put(&ctx->refs); return; }
@@ -1125,6 +1130,7 @@ static void io_req_local_work_add(struct if (ctx->has_evfd) io_eventfd_signal(ctx); __io_cqring_wake(ctx); + percpu_ref_put(&ctx->refs); }
static inline void __io_req_task_work_add(struct io_kiocb *req, bool allow_local)