io_uring: avoid normal tw intermediate fallback

When a DEFER_TASKRUN io_uring is terminating it requeues deferred task
work items as normal tw, which can further fallback to kthread
execution. Avoid this extra step and always push them to the fallback
kthread.

Signed-off-by: Pavel Begunkov <asml.silence@gmail.com>
Link: https://lore.kernel.org/r/d1cd472cec2230c66bd1c8d412a5833f0af75384.1730772720.git.asml.silence@gmail.com
Signed-off-by: Jens Axboe <axboe@kernel.dk>

diff --git a/io_uring/io_uring.c b/io_uring/io_uring.c
index 751b9e19da6eb53671789b2a22c6159870edb51b..042a65d38d0c50b3eda2f6be9d43b13d1707f581 100644 (file)
--- a/io_uring/io_uring.c
+++ b/io_uring/io_uring.c
@@ -1066,9 +1066,8 @@ struct llist_node *io_handle_tw_list(struct llist_node *node,
        return node;
 }
 
-static __cold void io_fallback_tw(struct io_uring_task *tctx, bool sync)
+static __cold void __io_fallback_tw(struct llist_node *node, bool sync)
 {
-       struct llist_node *node = llist_del_all(&tctx->task_list);
        struct io_ring_ctx *last_ctx = NULL;
        struct io_kiocb *req;
 
@@ -1094,6 +1093,13 @@ static __cold void io_fallback_tw(struct io_uring_task *tctx, bool sync)
        }
 }
 
+static void io_fallback_tw(struct io_uring_task *tctx, bool sync)
+{
+       struct llist_node *node = llist_del_all(&tctx->task_list);
+
+       __io_fallback_tw(node, sync);
+}
+
 struct llist_node *tctx_task_work_run(struct io_uring_task *tctx,
                                      unsigned int max_entries,
                                      unsigned int *count)
@@ -1247,16 +1253,9 @@ void io_req_task_work_add_remote(struct io_kiocb *req, struct io_ring_ctx *ctx,
 
 static void __cold io_move_task_work_from_local(struct io_ring_ctx *ctx)
 {
-       struct llist_node *node;
+       struct llist_node *node = llist_del_all(&ctx->work_llist);
 
-       node = llist_del_all(&ctx->work_llist);
-       while (node) {
-               struct io_kiocb *req = container_of(node, struct io_kiocb,
-                                                   io_task_work.node);
-
-               node = node->next;
-               io_req_normal_work_add(req);
-       }
+       __io_fallback_tw(node, false);
 }
 
 static bool io_run_local_work_continue(struct io_ring_ctx *ctx, int events,

diff --git a/io_uring/io_uring.h b/io_uring/io_uring.h
index 702c8e987430de1728023720caa7af1f68866300..4070d4c8ef9713ed1baa6b4bc28e63dc10255cf2 100644 (file)
--- a/io_uring/io_uring.h
+++ b/io_uring/io_uring.h
@@ -136,7 +136,7 @@ static inline void io_lockdep_assert_cq_locked(struct io_ring_ctx *ctx)
                 * Not from an SQE, as those cannot be submitted, but via
                 * updating tagged resources.
                 */
-               if (ctx->submitter_task->flags & PF_EXITING)
+               if (percpu_ref_is_dying(&ctx->refs))
                        lockdep_assert(current_work());
                else
                        lockdep_assert(current == ctx->submitter_task);