Bug#443937: openafs-modules-source: Bizzare failures when building on a newer cc/glibc
On Mon, 15 Oct 2007, Chris Hanson wrote: From: Russ Allbery [EMAIL PROTECTED] We think we've tracked this down to not a kernel revision but the upgrade from gcc to 4.2, and specifically its treatment of signed integer overflow as undefined. Could you try this patch and see if it fixes the problem for you? I'm not the original reporter, but I experienced the same symptoms. I can confirm that the patch fixes the bug. Sorry, family health issues kept me offline for a while. However, the problem is indeed solved (and now I can build that .23 kernel as well) - thanks ! -- Rick Nelson Operating Systems Installed: * Debian GNU/Linux 2.1 4 CD Set ($20 from www.chguy.net; price includes taxes, shipping, and a $3 donation to FSF). 2 CDs are binaries, 2 CDs complete source code; * Windows 98 Second Edition Upgrade Version ($136 through Megadepot.com, price does not include taxes/shipping). Surprisingly, no source code is included. -- Bill Stilwell, http://linuxtoday.com/stories/8794.html -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Bug#443937: openafs-modules-source: Bizzare failures when building on a newer cc/glibc
Date: Sun, 07 Oct 2007 12:54:34 -0700 From: Russ Allbery [EMAIL PROTECTED] We think we've tracked this down to not a kernel revision but the upgrade from gcc to 4.2, and specifically its treatment of signed integer overflow as undefined. Could you try this patch and see if it fixes the problem for you? I'm not the original reporter, but I experienced the same symptoms. I can confirm that the patch fixes the bug. Thanks.
Bug#443937: openafs-modules-source: Bizzare failures when building on a newer cc/glibc
Richard A Nelson [EMAIL PROTECTED] writes:
Package: openafs-modules-source
Version: 1.4.4.dfsg1-7
Severity: grave
Justification: renders package unusable
I just built a new x86_64 machine to bridge AFS/NFS/CIFS and after
logging in to an AFS id, I found this:
We think we've tracked this down to not a kernel revision but the upgrade
from gcc to 4.2, and specifically its treatment of signed integer overflow
as undefined.
Could you try this patch and see if it fixes the problem for you?
Index: src/dir/dir.c
===
RCS file: /cvs/openafs/src/dir/dir.c,v
retrieving revision 1.24
diff -u -r1.24 dir.c
--- src/dir/dir.c 13 Oct 2005 15:12:12 - 1.24
+++ src/dir/dir.c 7 Oct 2007 17:10:37 -
@@ -478,8 +478,9 @@
{
/* Hash a string to a number between 0 and NHASHENT. */
register unsigned char tc;
-register int hval;
+unsigned long hval;
register int tval;
+
hval = 0;
while ((tc = (*string++))) {
hval *= 173;
@@ -488,7 +489,7 @@
tval = hval (NHASHENT - 1);
if (tval == 0)
return tval;
-else if (hval 0)
+else if (hval = 131)
tval = NHASHENT - tval;
return tval;
}
--
Russ Allbery ([EMAIL PROTECTED]) http://www.eyrie.org/~eagle/
--
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Bug#443937: openafs-modules-source: Bizzare failures when building on a newer cc/glibc
On Mon, 24 Sep 2007, Russ Allbery wrote: Ah, okay, this, where you see some stuff and not others. Yeah, this was reported on the list as well, so with the additional details you gave, it means something broke in 2.6.22.6 on x86 and possibly in 2.6.22.5 on x86_64. that could be, I didn't have time to do earlier kernel releases before I left this evening - at the time I kinda thought the recent kernel-header issues might've been at fault, but since I build most kernels locally - I guess that doens't really make sense... but it has been a long month so far this week :) I'll follow up to the current mailing list discussion about this and see if we can get to the bottom of it. Cool, I'm relieved I'm not just hallucinating the whole thing Does the current Debian unstable kernel work? Good question ! I'll hopefully be back in the office a few hours tomorrow and will see if I can get that installed. -- Rick Nelson LackOfKan What are 'bots'? ``Erik rsg is a bot, not a human, not a human usable client, just a bot. ``Erik about the same as a quake bot, except irc bots are (usually) built to help, not shoot your ass full of holes -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Bug#443937: openafs-modules-source: Bizzare failures when building on a newer cc/glibc
Package: openafs-modules-source Version: 1.4.4.dfsg1-7 Severity: grave Justification: renders package unusable I just built a new x86_64 machine to bridge AFS/NFS/CIFS and after logging in to an AFS id, I found this: ls -l /u1/cobdev/cobbuild/ total 21 ?- ? ?? ?? /u1/cobdev/cobbuild/bin.tar ?- ? ?? ?? /u1/cobdev/cobbuild/cobol ?- ? ?? ?? /u1/cobdev/cobbuild/cobolw3 ?- ? ?? ?? /u1/cobdev/cobbuild/cobolw4 ?- ? ?? ?? /u1/cobdev/cobbuild/cobolw5 ?- ? ?? ?? /u1/cobdev/cobbuild/cobolwp ?- ? ?? ?? /u1/cobdev/cobbuild/nohup.out ?- ? ?? ?? /u1/cobdev/cobbuild/private ?- ? ?? ?? /u1/cobdev/cobbuild/scheduled ?- ? ?? ?? /u1/cobdev/cobbuild/windows drwxr-xr-x 2 cobbuild cobdev 2048 2007-09-06 14:34 AIX I thought it might be a 64bit, or new kernel issue, as 2.6.22.5 on my x86_32 box worked fine... However, using 2.6.22.5 on the _64 box still showed the error :( So I compiled 2.6.22.7 on the _32 box and see the exact same failure ! This likely coincides with the kernel-headers sharing 32bit and 64bit headers for portions - and I'm guessing is 32bit vs 64bit alignment and/or size issue. -- System Information: Debian Release: lenny/sid APT prefers testing-proposed-updates APT policy: (500, 'testing-proposed-updates'), (500, 'proposed-updates'), (500, 'unstable'), (500, 'testing'), (500, 'stable'), (1, 'experimental') Architecture: i386 (i686) Kernel: Linux 2.6.22.7 (SMP w/1 CPU core; PREEMPT) Locale: LANG=en_US.utf8, LC_CTYPE=en_US.utf8 (charmap=UTF-8) Shell: /bin/sh linked to /bin/bash Versions of packages openafs-modules-source depends on: ii bison 1:2.3.dfsg-5 A parser generator that is compati ii debhelper 5.0.56 helper programs for debian/rules ii flex2.5.33-12A fast lexical analyzer generator. ii kernel-package 11.001 A utility for building Linux kerne ii module-assistant0.10.11 tool to make module package creati openafs-modules-source recommends no packages. -- no debconf information -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Bug#443937: openafs-modules-source: Bizzare failures when building on a newer cc/glibc
Richard A Nelson [EMAIL PROTECTED] writes: Package: openafs-modules-source Version: 1.4.4.dfsg1-7 Severity: grave Justification: renders package unusable I just built a new x86_64 machine to bridge AFS/NFS/CIFS and after logging in to an AFS id, I found this: ls -l /u1/cobdev/cobbuild/ total 21 ?- ? ?? ?? /u1/cobdev/cobbuild/bin.tar ?- ? ?? ?? /u1/cobdev/cobbuild/cobol ?- ? ?? ?? /u1/cobdev/cobbuild/cobolw3 ?- ? ?? ?? /u1/cobdev/cobbuild/cobolw4 ?- ? ?? ?? /u1/cobdev/cobbuild/cobolw5 ?- ? ?? ?? /u1/cobdev/cobbuild/cobolwp ?- ? ?? ?? /u1/cobdev/cobbuild/nohup.out ?- ? ?? ?? /u1/cobdev/cobbuild/private ?- ? ?? ?? /u1/cobdev/cobbuild/scheduled ?- ? ?? ?? /u1/cobdev/cobbuild/windows drwxr-xr-x 2 cobbuild cobdev 2048 2007-09-06 14:34 AIX I thought it might be a 64bit, or new kernel issue, as 2.6.22.5 on my x86_32 box worked fine... However, using 2.6.22.5 on the _64 box still showed the error :( So I compiled 2.6.22.7 on the _32 box and see the exact same failure ! That output looks like you don't have a token. It's the output I'd expect from listing a directory that's listable but not readable. Do you actually have a token? What is the output from the tokens command and what are the ACLs on that directory? This likely coincides with the kernel-headers sharing 32bit and 64bit headers for portions - and I'm guessing is 32bit vs 64bit alignment and/or size issue. This is unlikely given that AFS has worked fine on x86_64 and x86 for years and nothing changed about this in the latest AFS release. I expect it's something else. The x86_64 kernel is rather different from the x86 kernel. -- Russ Allbery ([EMAIL PROTECTED]) http://www.eyrie.org/~eagle/ -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Bug#443937: openafs-modules-source: Bizzare failures when building on a newer cc/glibc
On Mon, 24 Sep 2007, Russ Allbery wrote: That output looks like you don't have a token. It's the output I'd expect from listing a directory that's listable but not readable. Do you actually have a token? What is the output from the tokens command and what are the ACLs on that directory? I thought about tokens initially - but some of the files are corrrect and readable, some (files and directories) completely unavailable. Here is a fresh attempt, to a _32 system: $ tokens Tokens held by the Cache Manager: User's (AFS ID 6976) tokens for [EMAIL PROTECTED] [Expires Sep 26 05:19] --End of list-- $ id uid=6976(cobbuild) gid=210(cobdev) groups=100(users),210(cobdev),666(ssh-user) === this is kerberos 4... thought they're working on moving the server from transarc to openafs === $ fs la Access list for . is Normal rights: system:anyuser rl cowboy rla === that'd be my primary id cobbuild rlidwka === the ID logged onto $ls -la ... -rw--- 1 cobbuild cobdev 2778 Jun 27 16:47 .bash_history ?- ? ?? ?? .bashrc-old -rw--- 1 cobbuild cobdev22 May 1 19:06 .dbxhist ?- ? ?? ?? .envfile -rw-r--r-- 1 cobbuild cobdev20 Jul 14 2005 .forward ?- ? ?? ?? .lesshst -rwxr-xr-x 1 cobbuild cobdev 302 Feb 14 2007 .logout ?- ? ?? ?? .netrc ?- ? ?? ?? .plan ... If I reboot back to the 2.6.22.5 kernel/afs modules, things work just fine: $ls -la ... -rw--- 1 cobbuild cobdev 2778 2007-06-27 16:47 .bash_history -rw-r--r-- 1 cobbuild cobdev 3165 2007-02-20 23:04 .bashrc-old -rw--- 1 cobbuild cobdev22 2007-05-01 19:06 .dbxhist -rwxr-xr-- 1 cobbuild cobdev 2103 2007-02-05 17:49 .envfile -rw-r--r-- 1 cobbuild cobdev20 2005-07-14 10:42 .forward -rw--- 1 cobbuild cobdev77 2007-06-26 23:00 .lesshst -rwxr-xr-x 1 cobbuild cobdev 302 2007-02-14 17:14 .logout lrwxr-xr-x 1 cobbuild cobdev14 2007-07-10 13:21 .netrc - private/.netrc -rwxrwxrwx 1 cobbuild cobdev 1470 2007-09-24 00:40 .plan ... This likely coincides with the kernel-headers sharing 32bit and 64bit headers for portions - and I'm guessing is 32bit vs 64bit alignment and/or size issue. This is unlikely given that AFS has worked fine on x86_64 and x86 for years and nothing changed about this in the latest AFS release. I expect it's something else. The x86_64 kernel is rather different from the x86 kernel. Yes, I run both - and haven't had any problems for years - until now... and istr issues on the lists/irc relating to problems due to using _64 headers on _32 systems... I hadn't actually rebuilt a kernel/module in a while... since 2.6.22.5, so I'm not sure when it exactly started. Anyway, this is fairly new, and totally repeatable, here, on _32 and _64 systems: 2.6.22.5 = kernel = 2.6.22.7, libc6 2.6.1-5, gcc 4.2.1-5 -- Rick Nelson I did this 'cause Linux gives me a woody. It doesn't generate revenue. (Dave '-ddt-` Taylor, announcing DOOM for Linux) -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Bug#443937: openafs-modules-source: Bizzare failures when building on a newer cc/glibc
Richard A Nelson [EMAIL PROTECTED] writes: $ fs la Access list for . is Normal rights: system:anyuser rl cowboy rla === that'd be my primary id cobbuild rlidwka === the ID logged onto $ls -la ... -rw--- 1 cobbuild cobdev 2778 Jun 27 16:47 .bash_history ?- ? ?? ?? .bashrc-old -rw--- 1 cobbuild cobdev22 May 1 19:06 .dbxhist ?- ? ?? ?? .envfile -rw-r--r-- 1 cobbuild cobdev20 Jul 14 2005 .forward ?- ? ?? ?? .lesshst -rwxr-xr-x 1 cobbuild cobdev 302 Feb 14 2007 .logout ?- ? ?? ?? .netrc ?- ? ?? ?? .plan ... Ah, okay, this, where you see some stuff and not others. Yeah, this was reported on the list as well, so with the additional details you gave, it means something broke in 2.6.22.6 on x86 and possibly in 2.6.22.5 on x86_64. I'll follow up to the current mailing list discussion about this and see if we can get to the bottom of it. Does the current Debian unstable kernel work? -- Russ Allbery ([EMAIL PROTECTED]) http://www.eyrie.org/~eagle/ -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
