io_uring: file switch work needs to get flushed on exit

We currently flush early, but if we have something in progress and a
new switch is scheduled, we need to ensure to flush after our teardown
as well.

Signed-off-by: Jens Axboe <axboe@kernel.dk>
This commit is contained in:
Jens Axboe 2020-01-17 11:15:34 -07:00
parent b14cca0c84
commit e46a7950d3

View file

@ -5089,11 +5089,14 @@ static int io_sqe_files_unregister(struct io_ring_ctx *ctx)
return -ENXIO;
/* protect against inflight atomic switch, which drops the ref */
flush_work(&data->ref_work);
percpu_ref_get(&data->refs);
/* wait for existing switches */
flush_work(&data->ref_work);
percpu_ref_kill_and_confirm(&data->refs, io_file_ref_kill);
wait_for_completion(&data->done);
percpu_ref_put(&data->refs);
/* flush potential new switch */
flush_work(&data->ref_work);
percpu_ref_exit(&data->refs);
__io_sqe_files_unregister(ctx);