On 6/30/22 10:39 PM, Al Viro wrote: > On Wed, Jun 01, 2022 at 02:01:41PM -0700, Stefan Roesch wrote: >> This adds the async buffered write support to XFS. For async buffered >> write requests, the request will return -EAGAIN if the ilock cannot be >> obtained immediately. > > breaks generic/471... That test case is odd, because it makes some weird assumptions about what RWF_NOWAIT means. Most notably that it makes it mean if we should instantiate blocks or not. Where did those assumed semantics come from? On the read side, we have clearly documented that it should "not wait for data which is not immediately available". Now it is possible that we're returning a spurious -EAGAIN here when we should not be. And that would be a bug imho. I'll dig in and see what's going on. -- Jens Axboe
