We prefer nornal task_works even if it would fail requests inside. Kill a PF_EXITING check in io_req_task_work_add(), task_work_add() handles well dying tasks, i.e. return error when can't enqueue due to late stages of do_exit(). Signed-off-by: Pavel Begunkov <asml.silence@xxxxxxxxx> --- fs/io_uring.c | 4 +--- 1 file changed, 1 insertion(+), 3 deletions(-) diff --git a/fs/io_uring.c b/fs/io_uring.c index 8f18af509afd..ba1df6ae6024 100644 --- a/fs/io_uring.c +++ b/fs/io_uring.c @@ -2003,8 +2003,6 @@ static void io_req_task_work_add(struct io_kiocb *req) if (test_bit(0, &tctx->task_state) || test_and_set_bit(0, &tctx->task_state)) return; - if (unlikely(tsk->flags & PF_EXITING)) - goto fail; /* * SQPOLL kernel thread doesn't need notification, just a wakeup. For @@ -2017,7 +2015,7 @@ static void io_req_task_work_add(struct io_kiocb *req) wake_up_process(tsk); return; } -fail: + clear_bit(0, &tctx->task_state); spin_lock_irqsave(&tctx->task_lock, flags); node = tctx->task_list.first; -- 2.32.0
