wait_for_dump_helpers() calls wake_up/kill_fasync from inside the
wait_event-like loop.  This is not needed and in fact this is not
strictly correct, we can/should do this only once after we change
pipe->writers.  We could even check if it becomes zero.
Change this code to use use wait_event_interruptible(), this can also
help to make this wait freezable.
With this patch we check pipe->readers without pipe_lock(), this is
fine.  Once we see pipe->readers == 1 we know that the handler
decremented the counter, this is all we need.
Signed-off-by: Oleg Nesterov <oleg@redhat.com>
Acked-by: Mandeep Singh Baines <msb@chromium.org>
Cc: Neil Horman <nhorman@redhat.com>
Cc: "Rafael J. Wysocki" <rjw@sisk.pl>
Cc: Tejun Heo <tj@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
        pipe_lock(pipe);
        pipe->readers++;
        pipe->writers--;
+       wake_up_interruptible_sync(&pipe->wait);
+       kill_fasync(&pipe->fasync_readers, SIGIO, POLL_IN);
+       pipe_unlock(pipe);
 
-       while ((pipe->readers > 1) && (!signal_pending(current))) {
-               wake_up_interruptible_sync(&pipe->wait);
-               kill_fasync(&pipe->fasync_readers, SIGIO, POLL_IN);
-               pipe_wait(pipe);
-       }
+       /*
+        * We actually want wait_event_freezable() but then we need
+        * to clear TIF_SIGPENDING and improve dump_interrupted().
+        */
+       wait_event_interruptible(pipe->wait, pipe->readers == 1);
 
+       pipe_lock(pipe);
        pipe->readers--;
        pipe->writers++;
        pipe_unlock(pipe);
-
 }
 
 /*