Return-Path: Received: from mga11.intel.com (mga11.intel.com [192.55.52.93]) by mx1.pokylinux.org (Postfix) with ESMTP id 371DA4C800A6 for ; Mon, 21 Feb 2011 18:25:27 -0600 (CST) Received: from fmsmga001.fm.intel.com ([10.253.24.23]) by fmsmga102.fm.intel.com with ESMTP; 21 Feb 2011 16:25:26 -0800 X-ExtLoop1: 1 X-IronPort-AV: E=Sophos;i="4.62,203,1297065600"; d="scan'208";a="889857474" Received: from kyu3-dev32.bj.intel.com (HELO localhost) ([10.238.153.102]) by fmsmga001.fm.intel.com with ESMTP; 21 Feb 2011 16:25:26 -0800 Date: Tue, 22 Feb 2011 08:26:18 +0800 From: Yu Ke To: Richard Purdie Message-ID: <20110222002618.GA10710@kyu3-hedt> References: <1298331207.11289.3256.camel@rex> MIME-Version: 1.0 In-Reply-To: <1298331207.11289.3256.camel@rex> User-Agent: Mutt/1.5.20 (2009-06-14) Cc: "poky@yoctoproject.org" Subject: Re: Autobuilder status X-BeenThere: poky@yoctoproject.org X-Mailman-Version: 2.1.13 Precedence: list List-Id: Poky build system developer discussion List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Tue, 22 Feb 2011 00:25:27 -0000 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline On Feb 21, 23:33, Richard Purdie wrote: > Hi, > > There appear to be three issues on the autobuilder at present which we > need to get fixed quickly. > > 1) "zypper search avahi" is causing a segfault on some architectures > with a double free being detected by libc. This seems to be the cause of > all the runtime sanity test failures. > > 2) The tar issues are fixed but there is an underlying problem with > error propagation as it appears bitbake was ignoring the sstate > extraction failure rather than accounting for it in its build task list. > If extraction fails, that task should not be making the "skip" list but > it looks like it does. That bug needs fixing too. > > 3) There are a number of dpkg-deb failures on permissions. This looks to > be pseudo related. Creation of the sstate package and extraction of it > should be happening under pseudo and the error suggests this is not > happening. > > I think PaulE is looking at 2), 3) is new and unassigned and I'm not > sure who is looking at 1) but I'd like to confirm someone is. > > > I'm also aware of some other niggling problems: > > a) do_rootfs depends on the populate_sysroot tasks for all the RDEPENDS. > This appears to be to pull in any tools required for the postinstalls > and dates to changes I made in 2007. This needs revisiting as its crazy > to populate all the sysroots when just the rpms are needed for image > generation. > > Simply put, a "bitbake poky-image-xxx -c rootfs" should just install the > package_write_rpm tasks it needs and the populate_sysroot for any tools > needed. No package data or target populate_sysroot tasks should run. > > b) "all" architecture packages seem to be getting reinstalled a lot by > sstate. Recent changes to the autobuilder should make it clearer when > this is happening but we should find out why. > > > I'll aim to file bugs on these tomorrow but just wanted to write things > down whilst there were fresh in my mind. I suspect there may be a bug > open for 1) above but I'd still appreciate a status update on it. Yes, Bug 722 is for 1), and Qing is looking at it. He will make a status update on this. Regards Ke