https://bugzilla.kernel.org/show_bug.cgi?id=209243 Zorro Lang (zlang@xxxxxxxxxx) changed: What |Removed |Added ---------------------------------------------------------------------------- CC| |axboe@xxxxxxxxx Summary|fsx IO_URING reading get |[regression] fsx IO_URING |BAD DATA |reading get BAD DATA --- Comment #5 from Zorro Lang (zlang@xxxxxxxxxx) --- Finally, I find the first commit which can reproduce this failure: commit c3cf992c25c7ff04bfc4dec5c916705a5332320e Author: Jens Axboe <axboe@xxxxxxxxx> Date: Fri May 22 09:24:42 2020 -0600 io_uring: support true async buffered reads, if file provides it If the file is flagged with FMODE_BUF_RASYNC, then we don't have to punt the buffered read to an io-wq worker. Instead we can rely on page unlocking callbacks to support retry based async IO. This is a lot more efficient than doing async thread offload. The retry is done similarly to how we handle poll based retry. From the unlock callback, we simply queue the retry to a task_work based handler. Signed-off-by: Jens Axboe <axboe@xxxxxxxxx> So CC the author of this patch to get more review. -- You are receiving this mail because: You are watching the assignee of the bug.
