io_uring: don't gate task_work run on TIF_NOTIFY_SIGNAL
authorJens Axboe <axboe@kernel.dk>
Sat, 21 Jan 2023 03:50:24 +0000 (20:50 -0700)
committerGreg Kroah-Hartman <gregkh@linuxfoundation.org>
Tue, 24 Jan 2023 06:22:43 +0000 (07:22 +0100)
commita9aa4aa7a5b293b8dae8add6abd4dcf37b1789eb
tree0ec8dec89557f0ff398e5aee5562582403df0ee6
parentbd9a23a4bb8a320ffa6e45cf09a068ec0a335350
io_uring: don't gate task_work run on TIF_NOTIFY_SIGNAL

commit 46a525e199e4037516f7e498c18f065b09df32ac upstream.

This isn't a reliable mechanism to tell if we have task_work pending, we
really should be looking at whether we have any items queued. This is
problematic if forward progress is gated on running said task_work. One
such example is reading from a pipe, where the write side has been closed
right before the read is started. The fput() of the file queues TWA_RESUME
task_work, and we need that task_work to be run before ->release() is
called for the pipe. If ->release() isn't called, then the read will sit
forever waiting on data that will never arise.

Fix this by io_run_task_work() so it checks if we have task_work pending
rather than rely on TIF_NOTIFY_SIGNAL for that. The latter obviously
doesn't work for task_work that is queued without TWA_SIGNAL.

Reported-by: Christiano Haesbaert <haesbaert@haesbaert.org>
Cc: stable@vger.kernel.org
Link: https://github.com/axboe/liburing/issues/665
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Signed-off-by: Sasha Levin <sashal@kernel.org>
io_uring/io-wq.c