io_uring: stop SQPOLL submit on creator's death

[ Upstream commit d9d05217cb6990b9a56e13b56e7a1b71e2551f6c ]

When the creator of SQPOLL io_uring dies (i.e. sqo_task), we don't want
its internals like ->files and ->mm to be poked by the SQPOLL task, it
have never been nice and recently got racy. That can happen when the
owner undergoes destruction and SQPOLL tasks tries to submit new
requests in parallel, and so calls io_sq_thread_acquire*().

That patch halts SQPOLL submissions when sqo_task dies by introducing
sqo_dead flag. Once set, the SQPOLL task must not do any submission,
which is synchronised by uring_lock as well as the new flag.

The tricky part is to make sure that disabling always happens, that
means either the ring is discovered by creator's do_exit() -> cancel,
or if the final close() happens before it's done by the creator. The
last is guaranteed by the fact that for SQPOLL the creator task and only
it holds exactly one file note, so either it pins up to do_exit() or
removed by the creator on the final put in flush. (see comments in
uring_flush() around file->f_count == 2).

One more place that can trigger io_sq_thread_acquire_*() is
__io_req_task_submit(). Shoot off requests on sqo_dead there, even
though actually we don't need to. That's because cancellation of
sqo_task should wait for the request before going any further.

note 1: io_disable_sqo_submit() does io_ring_set_wakeup_flag() so the
caller would enter the ring to get an error, but it still doesn't
guarantee that the flag won't be cleared.

note 2: if final __userspace__ close happens not from the creator
task, the file note will pin the ring until the task dies.

Cc: stable@vger.kernel.org # 5.5+
Fixed: b1b6b5a30dce8 ("kernel/io_uring: cancel io_uring before task works")
Signed-off-by: Pavel Begunkov <asml.silence@gmail.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

1 files changed, 50 insertions, 8 deletions

diff --git a/fs/io_uring.c b/fs/io_uring.c
index 70439e97457d..bcc26fc04d00 100644
--- a/fs/io_uring.c
+++ b/fs/io_uring.c
@@ -260,6 +260,7 @@ struct io_ring_ctx {
 		unsigned int		drain_next: 1;
 		unsigned int		eventfd_async: 1;
 		unsigned int		restricted: 1;
+		unsigned int		sqo_dead: 1;
 
 		/*
 		 * Ring buffer of indices into array of io_uring_sqe, which is
@@ -2083,11 +2084,9 @@ static void io_req_task_cancel(struct callback_head *cb)
 static void __io_req_task_submit(struct io_kiocb *req)
 {
 	struct io_ring_ctx *ctx = req->ctx;
-	bool fail;
 
-	fail = __io_sq_thread_acquire_mm(ctx);
 	mutex_lock(&ctx->uring_lock);
-	if (!fail)
+	if (!ctx->sqo_dead && !__io_sq_thread_acquire_mm(ctx))
 		__io_queue_sqe(req, NULL);
 	else
 		__io_req_task_cancel(req, -EFAULT);
@@ -6796,7 +6795,7 @@ again:
 		to_submit = 8;
 
 	mutex_lock(&ctx->uring_lock);
-	if (likely(!percpu_ref_is_dying(&ctx->refs)))
+	if (likely(!percpu_ref_is_dying(&ctx->refs) && !ctx->sqo_dead))
 		ret = io_submit_sqes(ctx, to_submit);
 	mutex_unlock(&ctx->uring_lock);
 
@@ -8487,6 +8486,10 @@ static void io_ring_ctx_wait_and_kill(struct io_ring_ctx *ctx)
 	mutex_lock(&ctx->uring_lock);
 	percpu_ref_kill(&ctx->refs);
 	/* if force is set, the ring is going away. always drop after that */
+
+	if (WARN_ON_ONCE((ctx->flags & IORING_SETUP_SQPOLL) && !ctx->sqo_dead))
+		ctx->sqo_dead = 1;
+
 	ctx->cq_overflow_flushed = 1;
 	if (ctx->rings)
 		__io_cqring_overflow_flush(ctx, true, NULL, NULL);
@@ -8745,6 +8748,18 @@ static bool __io_uring_cancel_task_requests(struct io_ring_ctx *ctx,
 	return ret;
 }
 
+static void io_disable_sqo_submit(struct io_ring_ctx *ctx)
+{
+	WARN_ON_ONCE(ctx->sqo_task != current);
+
+	mutex_lock(&ctx->uring_lock);
+	ctx->sqo_dead = 1;
+	mutex_unlock(&ctx->uring_lock);
+
+	/* make sure callers enter the ring to get error */
+	io_ring_set_wakeup_flag(ctx);
+}
+
 /*
  * We need to iteratively cancel requests, in case a request has dependent
  * hard links. These persist even for failure of cancelations, hence keep
@@ -8756,6 +8771,8 @@ 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 */
+		io_disable_sqo_submit(ctx);
 		task = ctx->sq_data->thread;
 		atomic_inc(&task->io_uring->in_idle);
 		io_sq_thread_park(ctx->sq_data);
@@ -8927,6 +8944,7 @@ void __io_uring_task_cancel(void)
 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 (!tctx)
 		return 0;
@@ -8942,7 +8960,16 @@ static int io_uring_flush(struct file *file, void *data)
 	if (atomic_long_read(&file->f_count) != 2)
 		return 0;
 
-	io_uring_del_task_file(file);
+	if (ctx->flags & IORING_SETUP_SQPOLL) {
+		/* there is only one file note, which is owned by sqo_task */
+		WARN_ON_ONCE((ctx->sqo_task == current) ==
+			     !xa_load(&tctx->xa, (unsigned long)file));
+
+		io_disable_sqo_submit(ctx);
+	}
+
+	if (!(ctx->flags & IORING_SETUP_SQPOLL) || ctx->sqo_task == current)
+		io_uring_del_task_file(file);
 	return 0;
 }
 
@@ -9016,8 +9043,9 @@ static unsigned long io_uring_nommu_get_unmapped_area(struct file *file,
 
 #endif /* !CONFIG_MMU */
 
-static void io_sqpoll_wait_sq(struct io_ring_ctx *ctx)
+static int io_sqpoll_wait_sq(struct io_ring_ctx *ctx)
 {
+	int ret = 0;
 	DEFINE_WAIT(wait);
 
 	do {
@@ -9026,6 +9054,11 @@ static void io_sqpoll_wait_sq(struct io_ring_ctx *ctx)
 
 		prepare_to_wait(&ctx->sqo_sq_wait, &wait, TASK_INTERRUPTIBLE);
 
+		if (unlikely(ctx->sqo_dead)) {
+			ret = -EOWNERDEAD;
+			goto out;
+		}
+
 		if (!io_sqring_full(ctx))
 			break;
 
@@ -9033,6 +9066,8 @@ static void io_sqpoll_wait_sq(struct io_ring_ctx *ctx)
 	} while (!signal_pending(current));
 
 	finish_wait(&ctx->sqo_sq_wait, &wait);
+out:
+	return ret;
 }
 
 SYSCALL_DEFINE6(io_uring_enter, unsigned int, fd, u32, to_submit,
@@ -9076,10 +9111,16 @@ SYSCALL_DEFINE6(io_uring_enter, unsigned int, fd, u32, to_submit,
 	if (ctx->flags & IORING_SETUP_SQPOLL) {
 		io_cqring_overflow_flush(ctx, false, NULL, NULL);
 
+		ret = -EOWNERDEAD;
+		if (unlikely(ctx->sqo_dead))
+			goto out;
 		if (flags & IORING_ENTER_SQ_WAKEUP)
 			wake_up(&ctx->sq_data->wait);
-		if (flags & IORING_ENTER_SQ_WAIT)
-			io_sqpoll_wait_sq(ctx);
+		if (flags & IORING_ENTER_SQ_WAIT) {
+			ret = io_sqpoll_wait_sq(ctx);
+			if (ret)
+				goto out;
+		}
 		submitted = to_submit;
 	} else if (to_submit) {
 		ret = io_uring_add_task_file(ctx, f.file);
@@ -9498,6 +9539,7 @@ static int io_uring_create(unsigned entries, struct io_uring_params *p,
 	trace_io_uring_create(ret, ctx, p->sq_entries, p->cq_entries, p->flags);
 	return ret;
 err:
+	io_disable_sqo_submit(ctx);
 	io_ring_ctx_wait_and_kill(ctx);
 	return ret;
 }