From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from mx2.suse.de ([195.135.220.15]:44146 "EHLO mx2.suse.de" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751050AbdCMPGi (ORCPT ); Mon, 13 Mar 2017 11:06:38 -0400 Date: Mon, 13 Mar 2017 16:05:59 +0100 From: David Sterba To: Holger =?iso-8859-1?Q?Hoffst=E4tte?= Cc: linux-btrfs Subject: Re: Question re. required since btrfs-progs-4.10 Message-ID: <20170313150559.GL14605@twin.jikos.cz> Reply-To: dsterba@suse.cz References: <58C42024.6000401@applied-asynchrony.com> MIME-Version: 1.0 Content-Type: text/plain; charset=iso-8859-1 In-Reply-To: <58C42024.6000401@applied-asynchrony.com> Sender: linux-btrfs-owner@vger.kernel.org List-ID: On Sat, Mar 11, 2017 at 05:04:52PM +0100, Holger Hoffstätte wrote: > I'm on Gentoo and wanted to update Docker to 17.03.0, which failed when it > couldn't build the btrfs driver due to a missing . > This worked fine on another machine the other day, so I dug in and found that > the only difference was an intermediate update to btrfs-progs 4.10. > Sure enough: since 4.10 ctree.h includes , which is nowhere > to be found in my package of linux-headers-4.10 (also not in 4.9). > > Simply copying sizes.h from the 4.9 kernel sources made everything work, > but I'm curious whether this is actually correct - hence my question whether > _should_ actually be installed? Is this a bug with > Gentoo's kernel header package or a new problem with btrfs-progs? That's a bug in btrfs-progs, the build tests do not cover that. There's a local copy of sizes.h that's supposed to be used. The patch was ported from kernel where linus/sizes.h is present but is probably not (yet) part of the public linux-headers. > I don't see any good reason why sizes.h should not be installed, but wanted > to verify since there are probably users of other distributions here as > well. :) The build would fail on any distro. I'll commit a fix and update the tests so this gets caught early. Thanks for the report.