io_uring: fix lost getsockopt completions
authorPavel Begunkov <asml.silence@gmail.com>
Tue, 16 Jul 2024 18:05:46 +0000 (19:05 +0100)
committerJens Axboe <axboe@kernel.dk>
Sat, 20 Jul 2024 17:04:56 +0000 (11:04 -0600)
commit24dce1c538a7ceac43f2f97aae8dfd4bb93ea9b9
treedfcd9428a753654389b77d15e7afaddfa5ac3edd
parentf557af081de6b45a25e27d633b4d8d2dbc2f428e
io_uring: fix lost getsockopt completions

There is a report that iowq executed getsockopt never completes. The
reason being that io_uring_cmd_sock() can return a positive result, and
io_uring_cmd() propagates it back to core io_uring, instead of IOU_OK.
In case of io_wq_submit_work(), the request will be dropped without
completing it.

The offending code was introduced by a hack in
a9c3eda7eada9 ("io_uring: fix submission-failure handling for uring-cmd"),
however it was fine until getsockopt was introduced and started
returning positive results.

The right solution is to always return IOU_OK, since
e0b23d9953b0c ("io_uring: optimise ltimeout for inline execution"),
we should be able to do it without problems, however for the sake of
backporting and minimising side effects, let's keep returning negative
return codes and otherwise do IOU_OK.

Link: https://github.com/axboe/liburing/issues/1181
Cc: stable@vger.kernel.org
Fixes: 8e9fad0e70b7b ("io_uring: Add io_uring command support for sockets")
Signed-off-by: Pavel Begunkov <asml.silence@gmail.com>
Reviewed-by: Breno Leitao <leitao@debian.org>
Link: https://lore.kernel.org/r/ff349cf0654018189b6077e85feed935f0f8839e.1721149870.git.asml.silence@gmail.com
Signed-off-by: Jens Axboe <axboe@kernel.dk>
io_uring/uring_cmd.c