On Sat, Apr 16, 2022 at 05:03:35PM +0800, Ming Lei wrote: > > Yes. But not doing this automatically also means you keep easily > > forgetting callsites. For example iomap still does not flush the plug > > in your patch. > > It is reasonable for flush user(usually submission) to be responsible > for finishing/flushing plug. Well, I very much disagree here. blk_flush_plug is not a publіc, exported API, and that is for a reason. A bio submission interface that requires flushing the plug to be useful is rather broken. > iomap is one good example to show this point, since it does flush the plug > before call bio_poll(), see __iomap_dio_rw(). iomap does not do a manual plug flush anywhere. iomap does finish the plug before polling, which makes sense. Now of course __blkdev_direct_IO_simple doesn't even use a plug to start with, so I'm wondering what plug this patch even tries to flush?
