From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from userp2130.oracle.com ([156.151.31.86]:40688 "EHLO userp2130.oracle.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1727650AbeLQR7e (ORCPT ); Mon, 17 Dec 2018 12:59:34 -0500 Date: Mon, 17 Dec 2018 09:59:22 -0800 From: "Darrick J. Wong" Subject: Re: COW improvements and always_cow support V3 Message-ID: <20181217175922.GI24487@magnolia> References: <20181203222503.30649-1-hch@lst.de> <20181206010550.GN24487@magnolia> <20181206200930.GA31797@lst.de> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <20181206200930.GA31797@lst.de> Sender: linux-xfs-owner@vger.kernel.org List-ID: List-Id: xfs To: Christoph Hellwig Cc: linux-xfs@vger.kernel.org On Thu, Dec 06, 2018 at 09:09:30PM +0100, Christoph Hellwig wrote: > I see the 050 one, and it looks odd. > > > --- /tmp/xfstests/tests/generic/075.out 2017-02-28 09:23:56.049065953 -0800 > > +++ /var/tmp/xfstests//generic/075.out.bad 2018-12-03 22:11:12.392566688 -0800 > > This and a few other fsx tests assume you can always fallocate > on XFS. I sent a series for this: > > https://www.spinics.net/lists/linux-xfs/msg23433.html > > But I need to rework some of the patches a little more based on the > review feedback. "the patches"... as in the fstests patches, or the always_cow series? > Similarly the badly fragmented warnings are simply expected, that > is the nature of an always out of place write fs. But the quota > run turned up a few more failures than I expected for the always_cow > mode, I can look into them. But so far 95% of the always_cow bugs > were either issues with the text, or genuine XFS COW path bugs not > otherwise uncovered.. generic/311 and generic/476 regularly trigger assertion warnings about negative quota counts, which I have yet to diagnose. FWIW fstests' post-run quota checking never complains about problems, which means we're probably just moving quota accounting in the wrong order or something. --D