eventfd: track eventfd_signal() recursion depth

eventfd use cases from aio and io_uring can deadlock due to circular
or resursive calling, when eventfd_signal() tries to grab the waitqueue
lock. On top of that, it's also possible to construct notification
chains that are deep enough that we could blow the stack.

Add a percpu counter that tracks the percpu recursion depth, warn if we
exceed it. The counter is also exposed so that users of eventfd_signal()
can do the right thing if it's non-zero in the context where it is
called.

Cc: stable@vger.kernel.org # 4.19+
Signed-off-by: Jens Axboe <axboe@kernel.dk>

1 files changed, 15 insertions, 0 deletions

diff --git a/fs/eventfd.c b/fs/eventfd.c
index 8aa0ea8c55e8..78e41c7c3d05 100644
--- a/fs/eventfd.c
+++ b/fs/eventfd.c
@@ -24,6 +24,8 @@
 #include <linux/seq_file.h>
 #include <linux/idr.h>
 
+DEFINE_PER_CPU(int, eventfd_wake_count);
+
 static DEFINE_IDA(eventfd_ida);
 
 struct eventfd_ctx {
@@ -60,12 +62,25 @@ __u64 eventfd_signal(struct eventfd_ctx *ctx, __u64 n)
 {
 	unsigned long flags;
 
+	/*
+	 * Deadlock or stack overflow issues can happen if we recurse here
+	 * through waitqueue wakeup handlers. If the caller users potentially
+	 * nested waitqueues with custom wakeup handlers, then it should
+	 * check eventfd_signal_count() before calling this function. If
+	 * it returns true, the eventfd_signal() call should be deferred to a
+	 * safe context.
+	 */
+	if (WARN_ON_ONCE(this_cpu_read(eventfd_wake_count)))
+		return 0;
+
 	spin_lock_irqsave(&ctx->wqh.lock, flags);
+	this_cpu_inc(eventfd_wake_count);
 	if (ULLONG_MAX - ctx->count < n)
 		n = ULLONG_MAX - ctx->count;
 	ctx->count += n;
 	if (waitqueue_active(&ctx->wqh))
 		wake_up_locked_poll(&ctx->wqh, EPOLLIN);
+	this_cpu_dec(eventfd_wake_count);
 	spin_unlock_irqrestore(&ctx->wqh.lock, flags);
 
 	return n;