io_uring: revise completion_lock locking

io_kill_timeouts() doesn't post any events but queues everything to
task_work. Locking there is needed for protecting linked requests
traversing, we should grab completion_lock directly instead of using
io_cq_[un]lock helpers. Same goes for __io_req_find_next_prep().
Signed-off-by: Pavel Begunkov <asml.silence@gmail.com>
Link: https://lore.kernel.org/r/88e75d481a65dc295cb59722bb1cf76402d1c06b.1670002973.git.asml.silence@gmail.comSigned-off-by: Jens Axboe <axboe@kernel.dk>

io_uring/io_uring.c

@@ -597,6 +597,18 @@ static inline void __io_cq_unlock(struct io_ring_ctx *ctx)
 		spin_unlock(&ctx->completion_lock);
 }
 
+static inline void io_cq_lock(struct io_ring_ctx *ctx)
+	__acquires(ctx->completion_lock)
+{
+	spin_lock(&ctx->completion_lock);
+}
+
+static inline void io_cq_unlock(struct io_ring_ctx *ctx)
+	__releases(ctx->completion_lock)
+{
+	spin_unlock(&ctx->completion_lock);
+}
+
 /* keep it inlined for io_submit_flush_completions() */
 static inline void __io_cq_unlock_post(struct io_ring_ctx *ctx)
 	__releases(ctx->completion_lock)
@@ -1074,9 +1086,9 @@ static void __io_req_find_next_prep(struct io_kiocb *req)
 {
 	struct io_ring_ctx *ctx = req->ctx;
 
-	io_cq_lock(ctx);
+	spin_lock(&ctx->completion_lock);
 	io_disarm_next(req);
-	io_cq_unlock_post(ctx);
+	spin_unlock(&ctx->completion_lock);
 }
 
 static inline struct io_kiocb *io_req_find_next(struct io_kiocb *req)

io_uring/io_uring.h

@@ -87,17 +87,6 @@ static inline void io_req_task_work_add(struct io_kiocb *req)
 #define io_for_each_link(pos, head) \
 	for (pos = (head); pos; pos = pos->link)
 
-static inline void io_cq_lock(struct io_ring_ctx *ctx)
-	__acquires(ctx->completion_lock)
-{
-	spin_lock(&ctx->completion_lock);
-}
-
-static inline void io_cq_unlock(struct io_ring_ctx *ctx)
-{
-	spin_unlock(&ctx->completion_lock);
-}
-
 void io_cq_unlock_post(struct io_ring_ctx *ctx);
 
 static inline struct io_uring_cqe *io_get_cqe_overflow(struct io_ring_ctx *ctx,

io_uring/timeout.c

@@ -624,7 +624,11 @@ __cold bool io_kill_timeouts(struct io_ring_ctx *ctx, struct task_struct *tsk,
 	struct io_timeout *timeout, *tmp;
 	int canceled = 0;
 
-	io_cq_lock(ctx);
+	/*
+	 * completion_lock is needed for io_match_task(). Take it before
+	 * timeout_lockfirst to keep locking ordering.
+	 */
+	spin_lock(&ctx->completion_lock);
 	spin_lock_irq(&ctx->timeout_lock);
 	list_for_each_entry_safe(timeout, tmp, &ctx->timeout_list, list) {
 		struct io_kiocb *req = cmd_to_io_kiocb(timeout);
@@ -634,6 +638,6 @@ __cold bool io_kill_timeouts(struct io_ring_ctx *ctx, struct task_struct *tsk,
 			canceled++;
 	}
 	spin_unlock_irq(&ctx->timeout_lock);
-	io_cq_unlock_post(ctx);
+	spin_unlock(&ctx->completion_lock);
 	return canceled != 0;
 }