eventfd: guard wake_up in eventfd fs calls as well

[ Upstream commit 9f0deaa12d832f488500a5afe9b912e9b3cfc432 ]

Guard wakeups that the user can trigger, and that may end up triggering a
call back into eventfd_signal. This is in addition to the current approach
that only guards in eventfd_signal.

Rename in_eventfd_signal -> in_eventfd at the same time to reflect this.

Without this there would be a deadlock in the following code using libaio:

int main()
{
	struct io_context *ctx = NULL;
	struct iocb iocb;
	struct iocb *iocbs[] = { &iocb };
	int evfd;
        uint64_t val = 1;

	evfd = eventfd(0, EFD_CLOEXEC);
	assert(!io_setup(2, &ctx));
	io_prep_poll(&iocb, evfd, POLLIN);
	io_set_eventfd(&iocb, evfd);
	assert(1 == io_submit(ctx, 1, iocbs));
        write(evfd, &val, 8);
}

Signed-off-by: Dylan Yudaken <dylany@fb.com>
Reviewed-by: Jens Axboe <axboe@kernel.dk>
Link: https://lore.kernel.org/r/20220816135959.1490641-1-dylany@fb.com
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Signed-off-by: Sasha Levin <sashal@kernel.org>

diff --git a/fs/eventfd.c b/fs/eventfd.c
index 3627dd7..c0ffee9 100644 (file)
--- a/fs/eventfd.c
+++ b/fs/eventfd.c
@@ -69,17 +69,17 @@ __u64 eventfd_signal(struct eventfd_ctx *ctx, __u64 n)
         * it returns false, the eventfd_signal() call should be deferred to a
         * safe context.
         */
-       if (WARN_ON_ONCE(current->in_eventfd_signal))
+       if (WARN_ON_ONCE(current->in_eventfd))
                return 0;
 
        spin_lock_irqsave(&ctx->wqh.lock, flags);
-       current->in_eventfd_signal = 1;
+       current->in_eventfd = 1;
        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);
-       current->in_eventfd_signal = 0;
+       current->in_eventfd = 0;
        spin_unlock_irqrestore(&ctx->wqh.lock, flags);
 
        return n;
@@ -253,8 +253,10 @@ static ssize_t eventfd_read(struct kiocb *iocb, struct iov_iter *to)
                __set_current_state(TASK_RUNNING);
        }
        eventfd_ctx_do_read(ctx, &ucnt);
+       current->in_eventfd = 1;
        if (waitqueue_active(&ctx->wqh))
                wake_up_locked_poll(&ctx->wqh, EPOLLOUT);
+       current->in_eventfd = 0;
        spin_unlock_irq(&ctx->wqh.lock);
        if (unlikely(copy_to_iter(&ucnt, sizeof(ucnt), to) != sizeof(ucnt)))
                return -EFAULT;
@@ -301,8 +303,10 @@ static ssize_t eventfd_write(struct file *file, const char __user *buf, size_t c
        }
        if (likely(res > 0)) {
                ctx->count += ucnt;
+               current->in_eventfd = 1;
                if (waitqueue_active(&ctx->wqh))
                        wake_up_locked_poll(&ctx->wqh, EPOLLIN);
+               current->in_eventfd = 0;
        }
        spin_unlock_irq(&ctx->wqh.lock);
 

diff --git a/include/linux/eventfd.h b/include/linux/eventfd.h
index 305d5f1..30eb30d 100644 (file)
--- a/include/linux/eventfd.h
+++ b/include/linux/eventfd.h
@@ -46,7 +46,7 @@ void eventfd_ctx_do_read(struct eventfd_ctx *ctx, __u64 *cnt);
 
 static inline bool eventfd_signal_allowed(void)
 {
-       return !current->in_eventfd_signal;
+       return !current->in_eventfd;
 }
 
 #else /* CONFIG_EVENTFD */

diff --git a/include/linux/sched.h b/include/linux/sched.h
index dcba347..e418935 100644 (file)
--- a/include/linux/sched.h
+++ b/include/linux/sched.h
@@ -933,7 +933,7 @@ struct task_struct {
 #endif
 #ifdef CONFIG_EVENTFD
        /* Recursion prevention for eventfd_signal() */
-       unsigned                        in_eventfd_signal:1;
+       unsigned                        in_eventfd:1;
 #endif
 
        unsigned long                   atomic_flags; /* Flags requiring atomic access. */