From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: <"evil@g-house.de"@mail.g-house.de> Received: from mail.g-house.de (ns2.g-housing.de [81.169.133.75]) (using TLSv1 with cipher AES256-SHA (256/256 bits)) (Client did not present a certificate) by ozlabs.org (Postfix) with ESMTP id 597B3DDDFF for ; Fri, 13 Jul 2007 09:05:58 +1000 (EST) Date: Fri, 13 Jul 2007 00:37:10 +0200 (CEST) From: Christian Kujau To: linux-kernel@vger.kernel.org Subject: Weird oopses with 2.6.22-rc7 Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; format=flowed; charset=us-ascii Sender: "evil@g-house.de"@mail.g-house.de Cc: linuxppc-dev@ozlabs.org List-Id: Linux on PowerPC Developers Mail List List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Hi there, Since a few days I'm experiencing weird oopses on my iBook/G4 with ubuntu's 2.6.22-7-powerpc and also with a vanilla 2.6.22-rc7-git8. Now for the "weird" part: The oops happens when doing "make install" for a piece of software (xine-ui media player) as *root*. No kidding, it does NOT happen when I compile the software and do "make install" as a normal user. This is 100% reproducible, here're some technical infos before I go on describing whe I did: http://nerdbynature.de/bits/2.6.22-rc7-git8/ I've compiled this software earlier, numerous times, always doing "make" as a user and "sudo make install" afterwards to install to /opt/xine. So when I now tried "make install" as a user I had to chown /opt/xine to the user of course. Even funnier: the dmesg output was captured when I was in runlevel 1 - here "make install" (as root) did produce an oops too, but the system did NOT panic. I typed "make install" a few times - always an oops, but no panic (in runlevel 1). The netconsole.log captured the panics as well, but the output is kinda messy (as always), I'll try to reformat the logfile a bit... When the Oopses started I checked memory (no memtest86 for ppc, but userspace memtest found nothing) and the harddisk (dd'ing from the whole disk) and fsck'ed the ext3 - not a single error. Also, all other applications are running fine (X11, music, video and stuff), the system is stable as usual. I even moved the old target-directory (make install...) out of the way and created a new one (new inode, new luck or so) - but to no avail. "fsck.ext3 -D" to reindex the directories, still the Oopses just won't go away. Searching the net for "Unrecoverable FP Unavailable Exception" did not return much :( Any ideas which magic syscall could cause these oopses? Thanks, Christian. -- BOFH excuse #409: The vulcan-death-grip ping has been applied.