Merge tag 'io_uring-5.11-2021-01-29' of git://git.kernel.dk/linux-block

Pull io_uring fixes from Jens Axboe:
 "We got the cancelation story sorted now, so for all intents and
  purposes, this should be it for 5.11 outside of any potential little
  fixes that may come in. This contains:

   - task_work task state fixes (Hao, Pavel)

   - Cancelation fixes (me, Pavel)

   - Fix for an inflight req patch in this release (Pavel)

   - Fix for a lock deadlock issue (Pavel)"

* tag 'io_uring-5.11-2021-01-29' of git://git.kernel.dk/linux-block:
  io_uring: reinforce cancel on flush during exit
  io_uring: fix sqo ownership false positive warning
  io_uring: fix list corruption for splice file_get
  io_uring: fix flush cqring overflow list while TASK_INTERRUPTIBLE
  io_uring: fix wqe->lock/completion_lock deadlock
  io_uring: fix cancellation taking mutex while TASK_UNINTERRUPTIBLE
  io_uring: fix __io_uring_files_cancel() with TASK_UNINTERRUPTIBLE
  io_uring: only call io_cqring_ev_posted() if events were posted
  io_uring: if we see flush on exit, cancel related tasks

fs/io_uring.c

@@ -1026,6 +1026,7 @@ static int io_setup_async_rw(struct io_kiocb *req, const struct iovec *iovec,
 			     const struct iovec *fast_iov,
 			     struct iov_iter *iter, bool force);
 static void io_req_drop_files(struct io_kiocb *req);
+static void io_req_task_queue(struct io_kiocb *req);
 
 static struct kmem_cache *req_cachep;
 
@@ -1069,8 +1070,12 @@ static bool io_match_task(struct io_kiocb *head,
 {
 	struct io_kiocb *req;
 
-	if (task && head->task != task)
+	if (task && head->task != task) {
+		/* in terms of cancelation, always match if req task is dead */
+		if (head->task->flags & PF_EXITING)
+			return true;
 		return false;
+	}
 	if (!files)
 		return true;
 
@@ -1630,18 +1635,11 @@ static void __io_queue_deferred(struct io_ring_ctx *ctx)
 	do {
 		struct io_defer_entry *de = list_first_entry(&ctx->defer_list,
 						struct io_defer_entry, list);
-		struct io_kiocb *link;
 
 		if (req_need_defer(de->req, de->seq))
 			break;
 		list_del_init(&de->list);
-		/* punt-init is done before queueing for defer */
-		link = __io_queue_async_work(de->req);
-		if (link) {
-			__io_queue_linked_timeout(link);
-			/* drop submission reference */
-			io_put_req_deferred(link, 1);
-		}
+		io_req_task_queue(de->req);
 		kfree(de);
 	} while (!list_empty(&ctx->defer_list));
 }
@@ -1775,12 +1773,13 @@ static bool __io_cqring_overflow_flush(struct io_ring_ctx *ctx, bool force,
 	struct io_kiocb *req, *tmp;
 	struct io_uring_cqe *cqe;
 	unsigned long flags;
-	bool all_flushed;
+	bool all_flushed, posted;
 	LIST_HEAD(list);
 
 	if (!force && __io_cqring_events(ctx) == rings->cq_ring_entries)
 		return false;
 
+	posted = false;
 	spin_lock_irqsave(&ctx->completion_lock, flags);
 	list_for_each_entry_safe(req, tmp, &ctx->cq_overflow_list, compl.list) {
 		if (!io_match_task(req, tsk, files))
@@ -1800,6 +1799,7 @@ static bool __io_cqring_overflow_flush(struct io_ring_ctx *ctx, bool force,
 			WRITE_ONCE(ctx->rings->cq_overflow,
 				   ctx->cached_cq_overflow);
 		}
+		posted = true;
 	}
 
 	all_flushed = list_empty(&ctx->cq_overflow_list);
@@ -1809,9 +1809,11 @@ static bool __io_cqring_overflow_flush(struct io_ring_ctx *ctx, bool force,
 		ctx->rings->sq_flags &= ~IORING_SQ_CQ_OVERFLOW;
 	}
 
-	io_commit_cqring(ctx);
+	if (posted)
+		io_commit_cqring(ctx);
 	spin_unlock_irqrestore(&ctx->completion_lock, flags);
-	io_cqring_ev_posted(ctx);
+	if (posted)
+		io_cqring_ev_posted(ctx);
 
 	while (!list_empty(&list)) {
 		req = list_first_entry(&list, struct io_kiocb, compl.list);
@@ -6458,7 +6460,8 @@ static struct file *io_file_get(struct io_submit_state *state,
 		file = __io_file_get(state, fd);
 	}
 
-	if (file && file->f_op == &io_uring_fops) {
+	if (file && file->f_op == &io_uring_fops &&
+	    !(req->flags & REQ_F_INFLIGHT)) {
 		io_req_init_async(req);
 		req->flags |= REQ_F_INFLIGHT;
 
@@ -7266,14 +7269,18 @@ static int io_cqring_wait(struct io_ring_ctx *ctx, int min_events,
 						TASK_INTERRUPTIBLE);
 		/* make sure we run task_work before checking for signals */
 		ret = io_run_task_work_sig();
-		if (ret > 0)
+		if (ret > 0) {
+			finish_wait(&ctx->wait, &iowq.wq);
 			continue;
+		}
 		else if (ret < 0)
 			break;
 		if (io_should_wake(&iowq))
 			break;
-		if (test_bit(0, &ctx->cq_check_overflow))
+		if (test_bit(0, &ctx->cq_check_overflow)) {
+			finish_wait(&ctx->wait, &iowq.wq);
 			continue;
+		}
 		if (uts) {
 			timeout = schedule_timeout(timeout);
 			if (timeout == 0) {
@@ -8865,30 +8872,31 @@ static void io_cancel_defer_files(struct io_ring_ctx *ctx,
 	}
 }
 
+static int io_uring_count_inflight(struct io_ring_ctx *ctx,
+				   struct task_struct *task,
+				   struct files_struct *files)
+{
+	struct io_kiocb *req;
+	int cnt = 0;
+
+	spin_lock_irq(&ctx->inflight_lock);
+	list_for_each_entry(req, &ctx->inflight_list, inflight_entry)
+		cnt += io_match_task(req, task, files);
+	spin_unlock_irq(&ctx->inflight_lock);
+	return cnt;
+}
+
 static void io_uring_cancel_files(struct io_ring_ctx *ctx,
 				  struct task_struct *task,
 				  struct files_struct *files)
 {
 	while (!list_empty_careful(&ctx->inflight_list)) {
 		struct io_task_cancel cancel = { .task = task, .files = files };
-		struct io_kiocb *req;
 		DEFINE_WAIT(wait);
-		bool found = false;
-
-		spin_lock_irq(&ctx->inflight_lock);
-		list_for_each_entry(req, &ctx->inflight_list, inflight_entry) {
-			if (!io_match_task(req, task, files))
-				continue;
-			found = true;
-			break;
-		}
-		if (found)
-			prepare_to_wait(&task->io_uring->wait, &wait,
-					TASK_UNINTERRUPTIBLE);
-		spin_unlock_irq(&ctx->inflight_lock);
+		int inflight;
 
-		/* We need to keep going until we don't find a matching req */
-		if (!found)
+		inflight = io_uring_count_inflight(ctx, task, files);
+		if (!inflight)
 			break;
 
 		io_wq_cancel_cb(ctx->io_wq, io_cancel_task_cb, &cancel, true);
@@ -8897,7 +8905,11 @@ static void io_uring_cancel_files(struct io_ring_ctx *ctx,
 		io_cqring_overflow_flush(ctx, true, task, files);
 		/* cancellations _may_ trigger task work */
 		io_run_task_work();
-		schedule();
+
+		prepare_to_wait(&task->io_uring->wait, &wait,
+				TASK_UNINTERRUPTIBLE);
+		if (inflight == io_uring_count_inflight(ctx, task, files))
+			schedule();
 		finish_wait(&task->io_uring->wait, &wait);
 	}
 }
@@ -8955,8 +8967,6 @@ static void io_uring_cancel_task_requests(struct io_ring_ctx *ctx,
 	struct task_struct *task = current;
 
 	if ((ctx->flags & IORING_SETUP_SQPOLL) && ctx->sq_data) {
-		/* for SQPOLL only sqo_task has task notes */
-		WARN_ON_ONCE(ctx->sqo_task != current);
 		io_disable_sqo_submit(ctx);
 		task = ctx->sq_data->thread;
 		atomic_inc(&task->io_uring->in_idle);
@@ -8966,10 +8976,9 @@ static void io_uring_cancel_task_requests(struct io_ring_ctx *ctx,
 	io_cancel_defer_files(ctx, task, files);
 	io_cqring_overflow_flush(ctx, true, task, files);
 
+	io_uring_cancel_files(ctx, task, files);
 	if (!files)
 		__io_uring_cancel_task_requests(ctx, task);
-	else
-		io_uring_cancel_files(ctx, task, files);
 
 	if ((ctx->flags & IORING_SETUP_SQPOLL) && ctx->sq_data) {
 		atomic_dec(&task->io_uring->in_idle);
@@ -9116,16 +9125,15 @@ void __io_uring_task_cancel(void)
 		prepare_to_wait(&tctx->wait, &wait, TASK_UNINTERRUPTIBLE);
 
 		/*
-		 * If we've seen completions, retry. This avoids a race where
-		 * a completion comes in before we did prepare_to_wait().
+		 * If we've seen completions, retry without waiting. This
+		 * avoids a race where a completion comes in before we did
+		 * prepare_to_wait().
 		 */
-		if (inflight != tctx_inflight(tctx))
-			continue;
-		schedule();
+		if (inflight == tctx_inflight(tctx))
+			schedule();
 		finish_wait(&tctx->wait, &wait);
 	} while (1);
 
-	finish_wait(&tctx->wait, &wait);
 	atomic_dec(&tctx->in_idle);
 
 	io_uring_remove_task_files(tctx);
@@ -9136,6 +9144,9 @@ static int io_uring_flush(struct file *file, void *data)
 	struct io_uring_task *tctx = current->io_uring;
 	struct io_ring_ctx *ctx = file->private_data;
 
+	if (fatal_signal_pending(current) || (current->flags & PF_EXITING))
+		io_uring_cancel_task_requests(ctx, NULL);
+
 	if (!tctx)
 		return 0;