io_uring: fix IOPOLL -EAGAIN retries 5.7-stable
authorJens Axboe <axboe@kernel.dk>
Thu, 27 Aug 2020 22:25:38 +0000 (16:25 -0600)
committerJens Axboe <axboe@kernel.dk>
Fri, 28 Aug 2020 13:46:38 +0000 (07:46 -0600)
commit5eab140f8dfe88a184d78c8f8a5db1e0885ff966
treedf190cd28bc815bf8d8d2428f953731f776295ea
parent84fa6532477f61d8cfc4b8a6f5ea6f254345367b
io_uring: fix IOPOLL -EAGAIN retries

This normally isn't hit, as polling is mostly done on NVMe with deep
queue depths. But if we do run into request starvation, we need to
ensure that retries are properly serialized.

Signed-off-by: Jens Axboe <axboe@kernel.dk>
fs/io_uring.c