The fio's option "fsync" can make rbd engine issue sync request. In practice, we would like to use aio_rbd_flush to make it non-block. So it will be more reliable to use fio to evaluate actual io scene like qemu.
Signed-off-by: Haomai Wang <[email protected]> --- engines/rbd.c | 30 ++++++++++++++++++++++++++++-- 1 file changed, 28 insertions(+), 2 deletions(-) diff --git a/engines/rbd.c b/engines/rbd.c index d006123..32ce60c 100644 --- a/engines/rbd.c +++ b/engines/rbd.c @@ -195,6 +195,23 @@ static void _fio_rbd_finish_read_aiocb(rbd_completion_t comp, void *data) return; } +static void _fio_rbd_finish_sync_aiocb(rbd_completion_t comp, void *data) +{ + struct io_u *io_u = (struct io_u *)data; + struct fio_rbd_iou *fio_rbd_iou = + (struct fio_rbd_iou *)io_u->engine_data; + + fio_rbd_iou->io_complete = 1; + + /* if sync needs to be verified - we should not release comp here + without fetching the result */ + rbd_aio_release(comp); + + /* TODO handle error */ + + return; +} + static struct io_u *fio_rbd_event(struct thread_data *td, int event) { struct rbd_data *rbd_data = td->io_ops->data; @@ -281,13 +298,22 @@ static int fio_rbd_queue(struct thread_data *td, struct io_u *io_u) } } else if (io_u->ddir == DDIR_SYNC) { - r = rbd_flush(rbd_data->image); + r = rbd_aio_create_completion(io_u, + (rbd_callback_t) + _fio_rbd_finish_sync_aiocb, + &comp); + if (r < 0) { + log_err + ("rbd_aio_create_completion for DDIR_SYNC failed.\n"); + goto failed; + } + + r = rbd_aio_flush(rbd_data->image, comp); if (r < 0) { log_err("rbd_flush failed.\n"); goto failed; } - return FIO_Q_COMPLETED; } else { dprint(FD_IO, "%s: Warning: unhandled ddir: %d\n", __func__, io_u->ddir); -- 1.8.5.2 (Apple Git-48) -- To unsubscribe from this list: send the line "unsubscribe fio" in the body of a message to [email protected] More majordomo info at http://vger.kernel.org/majordomo-info.html
