[PATCH for-next 2/4] io_uring: revise completion_lock locking

[Pavel Begunkov <asml.silence@xxxxxxxxx>]

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@xxxxxxxxx>
---
 io_uring/io_uring.c | 16 ++++++++++++++--
 io_uring/io_uring.h | 11 -----------
 io_uring/timeout.c  |  8 ++++++--
 3 files changed, 20 insertions(+), 15 deletions(-)

diff --git a/io_uring/io_uring.c b/io_uring/io_uring.c
index c30765579a8e..57c1c0da7648 100644
--- a/io_uring/io_uring.c
+++ b/io_uring/io_uring.c
@@ -582,6 +582,18 @@ void __io_commit_cqring_flush(struct io_ring_ctx *ctx)
 		io_eventfd_flush_signal(ctx);
 }
 
+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_inline(struct io_ring_ctx *ctx)
 	__releases(ctx->completion_lock)
@@ -1038,9 +1050,9 @@ 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)
diff --git a/io_uring/io_uring.h b/io_uring/io_uring.h
index 2277c05f52a6..ff84c0cfa2f2 100644
--- a/io_uring/io_uring.h
+++ b/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,
diff --git a/io_uring/timeout.c b/io_uring/timeout.c
index 4c6a5666541c..eae005b2d1d2 100644
--- a/io_uring/timeout.c
+++ b/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;
 }
-- 
2.38.1