io_uring: fix regression with always ignoring signals in io_cqring_wait()

When switching to TWA_SIGNAL for task_work notifications, we also made
any signal based condition in io_cqring_wait() return -ERESTARTSYS.
This breaks applications that rely on using signals to abort someone
waiting for events.

Check if we have a signal pending because of queued task_work, and
repeat the signal check once we've run the task_work. This provides a
reliable way of telling the two apart.

Additionally, only use TWA_SIGNAL if we are using an eventfd. If not,
we don't have the dependency situation described in the original commit,
and we can get by with just using TWA_RESUME like we previously did.

Fixes: ce593a6c48 ("io_uring: use signal based task_work running")
Cc: stable@vger.kernel.org # v5.7
Reported-by: Andres Freund <andres@anarazel.de>
Tested-by: Andres Freund <andres@anarazel.de>
Signed-off-by: Jens Axboe <axboe@kernel.dk>

fs/io_uring.c

@@ -4072,14 +4072,22 @@ struct io_poll_table {
 	int error;
 };
 
-static int io_req_task_work_add(struct io_kiocb *req, struct callback_head *cb,
-				int notify)
+static int io_req_task_work_add(struct io_kiocb *req, struct callback_head *cb)
 {
 	struct task_struct *tsk = req->task;
-	int ret;
+	struct io_ring_ctx *ctx = req->ctx;
+	int ret, notify = TWA_RESUME;
 
-	if (req->ctx->flags & IORING_SETUP_SQPOLL)
+	/*
+	 * SQPOLL kernel thread doesn't need notification, just a wakeup.
+	 * If we're not using an eventfd, then TWA_RESUME is always fine,
+	 * as we won't have dependencies between request completions for
+	 * other kernel wait conditions.
+	 */
+	if (ctx->flags & IORING_SETUP_SQPOLL)
 		notify = 0;
+	else if (ctx->cq_ev_fd)
+		notify = TWA_SIGNAL;
 
 	ret = task_work_add(tsk, cb, notify);
 	if (!ret)
@@ -4110,7 +4118,7 @@ static int __io_async_wake(struct io_kiocb *req, struct io_poll_iocb *poll,
 	 * of executing it. We can't safely execute it anyway, as we may not
 	 * have the needed state needed for it anyway.
 	 */
-	ret = io_req_task_work_add(req, &req->task_work, TWA_SIGNAL);
+	ret = io_req_task_work_add(req, &req->task_work);
 	if (unlikely(ret)) {
 		WRITE_ONCE(poll->canceled, true);
 		tsk = io_wq_get_task(req->ctx->io_wq);
@@ -6201,7 +6209,14 @@ static int io_cqring_wait(struct io_ring_ctx *ctx, int min_events,
 		if (current->task_works)
 			task_work_run();
 		if (signal_pending(current)) {
-			ret = -ERESTARTSYS;
+			if (current->jobctl & JOBCTL_TASK_WORK) {
+				spin_lock_irq(&current->sighand->siglock);
+				current->jobctl &= ~JOBCTL_TASK_WORK;
+				recalc_sigpending();
+				spin_unlock_irq(&current->sighand->siglock);
+				continue;
+			}
+			ret = -EINTR;
 			break;
 		}
 		if (io_should_wake(&iowq, false))
@@ -6210,7 +6225,7 @@ static int io_cqring_wait(struct io_ring_ctx *ctx, int min_events,
 	} while (1);
 	finish_wait(&ctx->wait, &iowq.wq);
 
-	restore_saved_sigmask_unless(ret == -ERESTARTSYS);
+	restore_saved_sigmask_unless(ret == -EINTR);
 
 	return READ_ONCE(rings->cq.head) == READ_ONCE(rings->cq.tail) ? ret : 0;
 }