io_uring: fix multishot poll on overflow
authorDylan Yudaken <dylany@fb.com>
Thu, 30 Jun 2022 09:12:27 +0000 (02:12 -0700)
committerJens Axboe <axboe@kernel.dk>
Thu, 21 Jul 2022 14:46:58 +0000 (08:46 -0600)
commit8c71a9efa1e23223ba2d1336398b23742537d86d
tree73de760e2fc244431aab6048ce3a71a673fcbfcb
parent61af7ee3f52d80e1def5260d9a6744bf96a704cc
io_uring: fix multishot poll on overflow

On overflow, multishot poll can still complete with the IORING_CQE_F_MORE
flag set.
If in the meantime the user clears a CQE and a the poll was cancelled then
the poll will post a CQE without the IORING_CQE_F_MORE (and likely result
-ECANCELED).

However when processing the application will encounter the non-overflow
CQE which indicates that there will be no more events posted. Typical
userspace applications would free memory associated with the poll in this
case.
It will then subsequently receive the earlier CQE which has overflowed,
which breaks the contract given by the IORING_CQE_F_MORE flag.

Signed-off-by: Dylan Yudaken <dylany@fb.com>
Link: https://lore.kernel.org/r/20220630091231.1456789-9-dylany@fb.com
Signed-off-by: Jens Axboe <axboe@kernel.dk>
io_uring/poll.c