From mboxrd@z Thu Jan 1 00:00:00 1970 From: Thomas Petazzoni Date: Fri, 28 Feb 2014 20:54:51 +0100 Subject: [Buildroot] Significant problem with the parallel build patch series In-Reply-To: <20140228201620.1f4b13ba@skate> References: <1393605956-18133-1-git-send-email-yann.morin.1998@free.fr> <20140228201620.1f4b13ba@skate> Message-ID: <20140228205451.76e9996f@skate> List-Id: MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit To: buildroot@busybox.net Hello, On Fri, 28 Feb 2014 20:16:20 +0100, Thomas Petazzoni wrote: > While I agree with the patch you're proposing, I was wondering which > patch was introducing the regression. And I didn't find where the > regression could be coming from. Looking more closely at one of the > build results, you see that the tarballs are only extracted at the > make legal-info invocation, which doesn't make any sense: "make" is > executed before "make legal-info" in the autobuilders. So apparently, > nothing gets built at all during "make", and it's only at the "make > legal-info" step that packages are extracted. > > So indeed your patch looks good, but it will not solve the autobuilders > problems I believe. Ok, after discussing with Gustavo, he told me that Vicente had already investigated the problem: it is caused by the parallel build patch series from Fabio. The problem is that the autobuilders disable the construction of any root filesystem image (even the tarball one). And when you disable all images, then "make all" doesn't do anything (in fact, it runs the post-image scripts if you have defined any). That's because of the reorganization of the target dependencies for the parallel build patch series. Fabio, which solution would you propose to ensure that packages get built even if no filesystem image format has been selected? Thanks! Thomas -- Thomas Petazzoni, CTO, Free Electrons Embedded Linux, Kernel and Android engineering http://free-electrons.com