Debian Bug report logs - #399724
sparc64: klibc segfaults again

Toggle useless messages

Message #5 received at submit@bugs.debian.org (full text, mbox, reply):

From: Tomas Cernaj <tomas@cernaj.de>

To: submit@bugs.debian.org

Subject: sparc64: klibc segfaults again

Date: Tue, 21 Nov 2006 17:10:48 +0100

Package: libklibc
Version: 1.4.30-1
Severity: important

I tried to install Etch-RC1 on a Sun Ultra 5 (400 MHz UltraSparc IIi,
type sun4u). When it comes to rebooting the system stops in the
initramfs phase because of segfaults in klibc. 
Installing a kernel with yaird works fine (from chroot), so i tried to
run some klibc-utils (such as /usr/lib/klibc/bin/uname -a) as the system
was running, but it segfaults.

I think this is related to bugs #363686 and #347902, as far as I've
found out.

Transscript of a boot session follows (maybe with typos because
transferred by hand...):

<output>
Begin: mounting root file system... ...
Begin: Running /scripts/local-top ...
Done.
Segmentation fault
Begin: Running /scripts/local-premount ...
Done.
Usage: modprobe [-v] [-V] [-C config-file] [-n] [-i] [-q] [-b] [-o
<modname>] <modname> [parameters...]
modprobe -r [-n] [-i] [-v] <modulename> ...
modprobe -l -t <dirname> [ -a <modulename> ...]
mount: Cannot read /etc/fstab: No such file or directory
Begin: Running /scripts/local-bottom ...
Done.
Done.
Begin: Running /scripts/init-bottom ...
Segmentation fault
Done.
mount: Mounting /sys on /root/sys failed: No such file or directory
mount: Mounting /proc on /root/proc failed: No such file or directory
Target filesystem doesn't have /sbin/init


BusyBox v1.1.3 (Debian 1:1.1.3-3) Built-in shell (ash)
Enter 'help' for a list of built-in commands.

/bin/sh: can't access tty; job control turned off
(initramfs) 
</output>

Versions used:
libc6: 2.3.6.ds1-7
linux-image-2.6.17-2-sparc64: 2.6.17-19
libklibc: 1.4.30-1
yaird: 0.0.12-18

Message #10 received at 399724@bugs.debian.org (full text, mbox, reply):

From: maximilian attems <maks@sternwelten.at>

To: Tomas Cernaj <tomas@cernaj.de>, 399724@bugs.debian.org

Cc: jurij@debian.org

Subject: Re: Bug#399724: sparc64: klibc segfaults again

Date: Tue, 21 Nov 2006 18:04:43 +0100

tags 399724 moreinfo
stop

On Tue, 21 Nov 2006, Tomas Cernaj wrote:
> I tried to install Etch-RC1 on a Sun Ultra 5 (400 MHz UltraSparc IIi,
> type sun4u). When it comes to rebooting the system stops in the
> initramfs phase because of segfaults in klibc. 
> Installing a kernel with yaird works fine (from chroot), so i tried to
> run some klibc-utils (such as /usr/lib/klibc/bin/uname -a) as the system
> was running, but it segfaults.
> 
> I think this is related to bugs #363686 and #347902, as far as I've
> found out.
> 
> Transscript of a boot session follows (maybe with typos because
> transferred by hand...):

hmm indeed that happens when fstype goes crazy.
please try by hand (adjust device name to a known partition):
fstype < /dev/sda1

if it segfaults please post
strace fstype < /dev/sda1

looked back in the changelog and found no evident sparc change.
last "official test" according to it was some ago,
cc'ing sparc porter trave11er

best regards

-- 
maks

Message #15 received at 399724@bugs.debian.org (full text, mbox, reply):

From: Tomas Cernaj <tomas@cernaj.de>

To: maximilian attems <maks@sternwelten.at>, 399724@bugs.debian.org

Subject: Re: Bug#399724: sparc64: klibc segfaults again

Date: Tue, 21 Nov 2006 19:11:56 +0100

Am Dienstag, den 21.11.2006, 18:04 +0100 schrieb maximilian attems:
> hmm indeed that happens when fstype goes crazy.
> please try by hand (adjust device name to a known partition):
> fstype < /dev/sda1
> 
> if it segfaults please post
> strace fstype < /dev/sda1
> 

indeed all klibc commands segfault, and that's where the initramfs
scripts fail.

Right now I can only run the test in a chroot environment from Debian
Sarge with custom 2.6.18 kernel, but i hope it helps anyway:

        sun:/# strace /usr/lib/klibc/bin/fstype < /dev/hda4
        execve("/usr/lib/klibc/bin/fstype",
        ["/usr/lib/klibc/bin/fstype"], [/* 18 vars */]) = -1 EFAULT (Bad
        address)
        --- SIGSEGV (Segmentation fault) @ 0 (0) ---
        +++ killed by SIGSEGV +++
        Process 8969 detached
        sun:/#

something else:

        sun:/# strace /usr/lib/klibc/bin/true 
        execve("/usr/lib/klibc/bin/true", ["/usr/lib/klibc/bin/true"],
        [/* 18 vars */]) = -1 EFAULT (Bad address)
        --- SIGSEGV (Segmentation fault) @ 0 (0) ---
        +++ killed by SIGSEGV +++
        Process 8919 detached
        sun:/#

here's my partition table:

        sun:/# fdisk -l /dev/hda
        
        Disk /dev/hda (Sun disk label): 16 heads, 63 sectors, 17660
        cylinders
        Units = cylinders of 1008 * 512 bytes
        
           Device Flag    Start       End    Blocks   Id  System
        /dev/hda1             0      9363   4718952   83  Linux native
        /dev/hda2          9363     10363    504000   82  Linux swap
        /dev/hda3             0     17662   8901648    5  Whole disk
        /dev/hda4         10363     17660   3677688   83  Linux native
        sun:/#

hda1 is Debian Sarge and hda4 is Etch RC1.

Hope this helps...
Tomas

Message #20 received at 399724@bugs.debian.org (full text, mbox, reply):

From: Tomas Cernaj <tomas@cernaj.de>

To: 399724@bugs.debian.org

Subject: some research

Date: Tue, 21 Nov 2006 22:07:44 +0100

Some interesting facts: 

I got the upstream source of klibc-1.4.30 and compiled it under sarge
with kernel 2.6.18. When compiled for 64 bits both the static and shared
versions of the utils work fine, but compiled for 32 bits only the
static ones work. The shared versions all exit with segfault as
described above...

Here some output of 'fstype < [fs]'

(sparc64, shared, /dev/hda4)
        FSTYPE=xfs
        FSSIZE=3765927936
        
(sparc64, shared, /dev/hda1)
        FSTYPE=ext3
        FSSIZE=1207959552
        
(sparc, static, /dev/hda4, BTW: note wrong FSSIZE)
        FSTYPE=xfs
        FSSIZE=16174537324212781056
        
(sparc, static, /dev/hda1)
        FSTYPE=ext3
        FSSIZE=1207959552
        
(sparc, shared, /dev/hda4)
        Segmentation fault

During installation I had indeed formatted /dev/hda4 as XFS first, but
when the installer complained about not being able to boot from XFS I
reformatted it to ext3... 

'fsck.ext3 -f /dev/hda4' checks out all right.

Tomas

Message #25 received at 399724@bugs.debian.org (full text, mbox, reply):

From: Jurij Smakov <jurij@wooyd.org>

To: 399724@bugs.debian.org

Cc: Tomas Cernaj <tomas@cernaj.de>, maximilian attems <maks@sternwelten.at>

Subject: Re: Bug#399724: sparc64: klibc segfaults again

Date: Wed, 22 Nov 2006 21:51:58 -0800

Hi Tomas,

It appears that all the binaries in klibc-utils package are statically 
linked, and I don't see any problem with them on an up-to-date sid 
system (klibc-utils 1.4.30-1):

root@debian:~# file /usr/lib/klibc/bin/fstype
/usr/lib/klibc/bin/fstype: ELF 32-bit MSB executable, SPARC, version 1 
(SYSV), statically linked (uses shared libs), stripped
root@debian:~# ls -al /usr/lib/klibc/bin/fstype 
-rwxr-xr-x 1 root root 3044 2006-11-03 06:56 /usr/lib/klibc/bin/fstype
root@debian:~# sha1sum /usr/lib/klibc/bin/fstype 
cb9a1b8e65a1bd3e8ab58a9affe7b4f32d7fcced  /usr/lib/klibc/bin/fstype
root@debian:~# ldd /usr/lib/klibc/bin/fstype 
        not a dynamic executable
root@debian:~# /usr/lib/klibc/bin/fstype < /dev/sda1
FSTYPE=ext3
FSSIZE=296110080
root@debian:~# /usr/lib/klibc/bin/fstype < /dev/sda2
FSTYPE=ext3
FSSIZE=1949390848
root@debian:~#

Are you sure you don't have some other (non-Debian) version of 
klibc-utils installed, and that got pulled in into the initrd somehow?

Best regards,
-- 
Jurij Smakov                                           jurij@wooyd.org
Key: http://www.wooyd.org/pgpkey/                      KeyID: C99E03CC

Message #30 received at 399724@bugs.debian.org (full text, mbox, reply):

From: Tomas Cernaj <tomas@cernaj.de>

To: Jurij Smakov <jurij@wooyd.org>, 399724@bugs.debian.org

Subject: Re: Bug#399724: sparc64: klibc segfaults again

Date: Thu, 23 Nov 2006 17:40:21 +0100

Hi Jurij,

ok, this is going to be a bit lengthy... :-)

Here's what my system says:

        sun:~# file /usr/lib/klibc/bin/fstype
        /usr/lib/klibc/bin/fstype: ELF 32-bit MSB executable, SPARC,
        version 1 (SYSV), statically linked (uses shared libs), stripped
        sun:~# ls -al /usr/lib/klibc/bin/fstype
        -rwxr-xr-x 1 root root 3044 Nov  3
        15:56 /usr/lib/klibc/bin/fstype
        sun:~# sha1sum /usr/lib/klibc/bin/fstype
        cb9a1b8e65a1bd3e8ab58a9affe7b4f32d7fcced  /usr/lib/klibc/bin/fstype
        sun:~# file /lib/klibc-gZY_j4EiGiPMbozhciEvacptnN8.so
        /lib/klibc-gZY_j4EiGiPMbozhciEvacptnN8.so: ELF 32-bit MSB
        executable, SPARC, version 1 (SYSV), statically linked, stripped
        sun:~# sha1sum /lib/klibc-gZY_j4EiGiPMbozhciEvacptnN8.so
        01392ec3964d79152c4958b8a8095557702458d6  /lib/klibc-gZY_j4EiGiPMbozhciEvacptnN8.so
        sun:~# /usr/lib/klibc/bin/fstype < /dev/hda4
        Segmentation fault
        sun:~#

But what I meant with static vs. shared is this (note "uses shared
libs"):

        root@sun:/usr/src/klibc-1.4.30-sparc/usr/kinit/fstype # file
        shared/fstype
        shared/fstype: ELF 32-bit MSB executable, SPARC, version 1
        (SYSV), statically linked (uses shared libs), stripped
        root@sun:/usr/src/klibc-1.4.30-sparc/usr/kinit/fstype # file
        static/fstype
        static/fstype: ELF 32-bit MSB executable, SPARC, version 1
        (SYSV), statically linked, stripped
        root@sun:/usr/src/klibc-1.4.30-sparc/usr/kinit/fstype #


Another thing, that should probably be filed as another bug is that an
ext3 filesystem that was previously formatted as xfs is still being
recognised as xfs, and there's apparently a problem with FSSIZE:

        root@sun:/usr/src # strace
        klibc-1.4.30-sparc/usr/kinit/fstype/static/fstype < /dev/hda4
        > /dev/null
        execve("klibc-1.4.30-sparc/usr/kinit/fstype/static/fstype",
        ["klibc-1.4.30-sparc/usr/kinit/fstype/static/fstype"], [/* 18
        vars */]) = 0
        pread(0, "XFSB\0\0\20\0\0\0\0\0\0\16\7x\0\0\0\0\0\0\0\0\0\0\0\0
        \0"..., 1024, 0) = 1024
        write(1, "FSTYPE=xfs\nFSSIZE=16174537324212"..., 39) = 39
        exit(0)                                 = ?
        root@sun:/usr/src # 


> Are you sure you don't have some other (non-Debian) version of 
> klibc-utils installed, and that got pulled in into the initrd somehow?

I don't know, how that could have happened (well, that means, I'm sure
it hasn't :-) ). I was doing a fresh install over the net using official
debian mirrors.
But even then this would't explain the segfault when I ran the 32 bit
shared version that I've compiled myself, while the static version
works...

Otherwise the system runs pretty well, in fact I'm writing this mail
right now with Evolution under GNOME from Debian Etch RC1 :-)

Here's the content of /proc/cpuinfo:

        sun:~# cat /proc/cpuinfo
        cpu             : TI UltraSparc IIi (Sabre)
        fpu             : UltraSparc IIi integrated FPU
        prom            : OBP 3.31.0 2001/07/25 20:36
        type            : sun4u
        ncpus probed    : 1
        ncpus active    : 1
        D$ parity tl1   : 0
        I$ parity tl1   : 0
        Cpu0Bogo        : 800.37
        Cpu0ClkTck      : 0000000017d78400
        MMU Type        : Spitfire
        sun:~#


Can I help you with some other specs of my system?

Tomas

Message #35 received at 399724@bugs.debian.org (full text, mbox, reply):

From: Jurij Smakov <jurij@wooyd.org>

To: Tomas Cernaj <tomas@cernaj.de>

Cc: 399724@bugs.debian.org

Subject: Re: Bug#399724: sparc64: klibc segfaults again

Date: Thu, 23 Nov 2006 10:20:14 -0800

On Thu, Nov 23, 2006 at 05:40:21PM +0100, Tomas Cernaj wrote:
> Hi Jurij,
> 
> ok, this is going to be a bit lengthy... :-)

No problem :-). I'm still somewhat confused though. Do I understand 
correctly that if you use the klibc utils and library from the 
archive, it segfaults *only* when you run fstype on your /dev/hda4 
(but runs ok on other partitions)? If that's the case, can you please 
build the klibc with debugging symbols and run gdb on it, to see where 
fstype segfaults in that case? Alternatively, you can provide first 
two kilobytes of /dev/hda4 using something like

dd if=/dev/hda4 of=hda4-header.raw bs=1024 count=2

That should be sufficient to debug the problem.

Best regards,
-- 
Jurij Smakov                                           jurij@wooyd.org
Key: http://www.wooyd.org/pgpkey/                      KeyID: C99E03CC

Message #40 received at 399724@bugs.debian.org (full text, mbox, reply):

From: Tomas Cernaj <tomas@cernaj.de>

To: Jurij Smakov <jurij@wooyd.org>

Cc: 399724@bugs.debian.org

Subject: Re: Bug#399724: sparc64: klibc segfaults again

Date: Fri, 24 Nov 2006 17:15:17 +0100

Am Donnerstag, den 23.11.2006, 10:20 -0800 schrieb Jurij Smakov:

> No problem :-). I'm still somewhat confused though. Do I understand 
> correctly that if you use the klibc utils and library from the 
> archive, it segfaults *only* when you run fstype on your /dev/hda4 
> (but runs ok on other partitions)? 

No, basically the main problem has nothing to do with partitions. It's
just that _none_ of the debian supplied klibc-utils work (not even a
simple one like /usr/lib/klibc/bin/true), because they immediately
segfault. I tried to track down the problem, but due to my limited
knowledge of linux/klibc internals i can only suspect that something
goes wrong the moment /lib/klibc-gZY_j4EiGiPMbozhciEvacptnN8.so is being
loaded and/or initialized. 
Versions used: libklibc-1.4.30-1, klibc-utils-1.4.30-1


What I then tried to do is to compile klibc myself, once for sparc and
then for sparc64.

** First try: sparc (32 bit) **

I unpacked the klibc package, set a link to the linux source tree,
linked linux/include/asm to linux/include/asm-sparc and typed
"KLIBCARCH=sparc make".
Then I tried to run the generated files under usr/utils/shared resp.
usr/utils/static (the shared ones inside a chroot so that they would
access the right klibc-*.so). 

I found out the following:
- the progs in usr/utils/static work fine
- the progs in usr/utils/shared all segfault, just as the ones supplied
by the debian packages


** Second try: sparc64 (64 bit) **

Now I unpacked klibc another time, but this time I linked
linux/include/asm to linux/include/asm-sparc64 and just typed "make".

Result:
- the progs in usr/utils/static work fine
- the progs in usr/utils/shared also work (this time using a 64 bit
klibc-cNP3iUE1BbKETtRZwHq6UJ7ssis.so)


Bottom line: The shared library gets loaded and initialized correctly
when built for sparc64, but not for sparc...


> If that's the case, can you please 
> build the klibc with debugging symbols and run gdb on it, to see where 
> fstype segfaults in that case? Alternatively, you can provide first 
> two kilobytes of /dev/hda4 using something like

Concerning the problem with the unrecognised ext3 partition I will file
a new bug report against klibc-utils.

cu,
Tomas

Message #45 received at 399724@bugs.debian.org (full text, mbox, reply):

From: Jurij Smakov <jurij@wooyd.org>

To: Tomas Cernaj <tomas@cernaj.de>

Cc: 399724@bugs.debian.org

Subject: Re: Bug#399724: sparc64: klibc segfaults again

Date: Sat, 25 Nov 2006 18:12:34 -0800

tag 399724 unreproducible
thanks

On Fri, Nov 24, 2006 at 05:15:17PM +0100, Tomas Cernaj wrote:
 
> No, basically the main problem has nothing to do with partitions. It's
> just that _none_ of the debian supplied klibc-utils work (not even a
> simple one like /usr/lib/klibc/bin/true), because they immediately
> segfault. I tried to track down the problem, but due to my limited
> knowledge of linux/klibc internals i can only suspect that something
> goes wrong the moment /lib/klibc-gZY_j4EiGiPMbozhciEvacptnN8.so is being
> loaded and/or initialized. 
> Versions used: libklibc-1.4.30-1, klibc-utils-1.4.30-1

Sorry, I cannot reproduce it. Today I've installed etch using RC1 
netinst image, ran apt-get update; apt-get upgrade (udev and klibc 
stuff got upgraded, among other things), and rebooted. The system 
booted without problems, and I'm able to run the klibc-provided 
utilities (true, fstype, sleep, cat were the ones I tried) without any 
problem. If you have any ideas about what causes it for you, please 
let me know. In the meantime, I'll tag the bug unreproducible.

Thanks, 
-- 
Jurij Smakov                                           jurij@wooyd.org
Key: http://www.wooyd.org/pgpkey/                      KeyID: C99E03CC

Message #52 received at 399724@bugs.debian.org (full text, mbox, reply):

From: Jurij Smakov <jurij@wooyd.org>

To: Jurij Smakov <jurij@wooyd.org>

Cc: Tomas Cernaj <tomas@cernaj.de>, 399724@bugs.debian.org

Subject: Re: Bug#399724: sparc64: klibc segfaults again

Date: Fri, 5 Jan 2007 23:12:07 -0800

On Sat, Nov 25, 2006 at 06:12:34PM -0800, Jurij Smakov wrote:

> Sorry, I cannot reproduce it. Today I've installed etch using RC1 
> netinst image, ran apt-get update; apt-get upgrade (udev and klibc 
> stuff got upgraded, among other things), and rebooted. The system 
> booted without problems, and I'm able to run the klibc-provided 
> utilities (true, fstype, sleep, cat were the ones I tried) without any 
> problem. If you have any ideas about what causes it for you, please 
> let me know. In the meantime, I'll tag the bug unreproducible.

Hi Tomas,

I didn't see any other reports about klibc problems, and it runs 
fine just fine on my machines. Are you still experiencing any issues, 
or can this bug be closed?

Thanks, 
-- 
Jurij Smakov                                           jurij@wooyd.org
Key: http://www.wooyd.org/pgpkey/                      KeyID: C99E03CC

Message #57 received at 399724-done@bugs.debian.org (full text, mbox, reply):

From: Jurij Smakov <jurij@wooyd.org>

To: 399724-done@bugs.debian.org

Subject: Re: Bug#399724: sparc64: klibc segfaults again

Date: Thu, 15 Feb 2007 20:58:34 -0800

Hi Tomas,

I haven't received any news about this bug for over a month, and I 
don't see any evidence of it on my machines, so I'll close it for 
now. If it persists for you, feel free to reopen it providing more 
information, and I'll be glad to have another look at it.

Best regards,
-- 
Jurij Smakov                                           jurij@wooyd.org
Key: http://www.wooyd.org/pgpkey/                      KeyID: C99E03CC

Message #64 received at 399724@bugs.debian.org (full text, mbox, reply):

From: "Bilderbeek, Manuel" <manuel.bilderbeek@oce.com>

To: <399724@bugs.debian.org>

Subject: Ouch

Date: Wed, 11 Apr 2007 09:05:55 +0200

I got exactly the output (<output>) from the first message in this bug
report on my 
Sun Ultra 10, which I tried to upgrade from Sarge to Etch. (Which is now

unbootable, as for some reason the 2.6.8 kernel doesn't boot either, it
seems it has problems finding the kernel modules...)

What can I do to help and to recover?

--
Kind regards,

Manuel Bilderbeek

This message and attachment(s) are intended solely for use by the addressee and may contain information that is privileged, confidential or otherwise exempt from disclosure under applicable law.

If you are not the intended recipient or agent thereof responsible for delivering this message to the intended recipient, you are hereby notified that any dissemination, distribution or copying of this communication is strictly prohibited.

If you have received this communication in error, please notify the sender immediately by telephone and with a 'reply' message.
  
Thank you for your co-operation.

Message #75 received at 399724@bugs.debian.org (full text, mbox, reply):

From: Bernd Zeimetz <bernd@bzed.de>

To: 399724@bugs.debian.org

Cc: Debian Sparc <debian-sparc@lists.debian.org>

Subject: klibc from unstable segfaults on sparc again

Date: Mon, 29 Oct 2007 12:30:21 +0100

[Message part 1 (text/plain, inline)]

Hi,

while the version which was shipped with Etch (1.4.34-1) boots fine on
the machine here (v880, using US III), the version which is currently in
sid (1.5.7-2) segfaults in the initrd.
If anybody is interested, I managed to save core files from udev and
reboot. They're attached to this mail.

This renders unstable pretty much unusable for me on sparc64.


Cheers,

Bernd


-- 
Bernd Zeimetz
<bernd@bzed.de>                         <http://bzed.de/>

[core.reboot.bz2 (application/x-bzip, inline)]

[core.udev.bz2 (application/x-bzip, inline)]

Message #80 received at 399724@bugs.debian.org (full text, mbox, reply):

From: maximilian attems <max@stro.at>

To: Bernd Zeimetz <bernd@bzed.de>, 399724@bugs.debian.org

Cc: Debian Sparc <debian-sparc@lists.debian.org>

Subject: Re: Bug#399724: klibc from unstable segfaults on sparc again

Date: Tue, 30 Oct 2007 09:20:08 +0100

On Mon, 29 Oct 2007, Bernd Zeimetz wrote:

> while the version which was shipped with Etch (1.4.34-1) boots fine on
> the machine here (v880, using US III), the version which is currently in
> sid (1.5.7-2) segfaults in the initrd.
> If anybody is interested, I managed to save core files from udev and
> reboot. They're attached to this mail.
> 
> This renders unstable pretty much unusable for me on sparc64.

could you please check out the different klibc versions,
as the diff between 1.4.34-1 and 1.5.7-2 is huge,
so i'd really need more information to pinpoint the trouble.
 
you seem to be the first reporter to realibale be able to reproduce that.
so a binary search between the two named versions would be really helpful.

you should be able to find them all at
-> deb http://snapshot.debian.net/archive pool klibc

ah and you'd have to downgrade initramfs-tools most probably.

best regards

-- 
maks

Message #85 received at 399724@bugs.debian.org (full text, mbox, reply):

From: Bernd Zeimetz <bernd@bzed.de>

To: maximilian attems <max@stro.at>

Cc: 399724@bugs.debian.org, Debian Sparc <debian-sparc@lists.debian.org>

Subject: Re: Bug#399724: klibc from unstable segfaults on sparc again

Date: Tue, 30 Oct 2007 12:12:44 +0100

> could you please check out the different klibc versions,
> as the diff between 1.4.34-1 and 1.5.7-2 is huge,
> so i'd really need more information to pinpoint the trouble.

The regression is somewhere between 1.5.5-1 (which doesn't segfault on
the console - didn't boot with it, though) and 1.5.7-2.

titan:~# strace -vfF /usr/lib/klibc/bin/fstype
execve("/usr/lib/klibc/bin/fstype", ["/usr/lib/klibc/bin/fstype"],
["SHELL=/bin/bash", "TERM=xterm", "SSH_CLIENT=[myip] 39403"...,
"SSH_TTY=/dev/pts/0", "USER=root",
"LS_COLORS=no=00:fi=00:di=01;34:l"...,
"PATH=/usr/local/sbin:/usr/local/"..., "MAIL=/var/mail/root",
"PWD=/root", "LANG=en_US.UTF-8", "PS1=\\h:\\w\\$ ", "HOME=/root",
"SHLVL=2", "LS_OPTIONS=--color=auto", "LOGNAME=root",
"SSH_CONNECTION=[myip] 3"..., "_=/usr/bin/strace", "OLDPWD=/"]) = 0
--- SIGSEGV (Segmentation fault) @ 0 (0) ---
+++ killed by SIGSEGV +++
Process 3636 detached


1.5.6 doesn't build on sparc.


> ah and you'd have to downgrade initramfs-tools most probably.

as the segfault also happens by executing /usr/lib/klibc/bin/fstype (for
example) on the normal console - luckily I don't need to boot every time
to figure it out.



-- 
Bernd Zeimetz
<bernd@bzed.de>                         <http://bzed.de/>

Message #90 received at 399724@bugs.debian.org (full text, mbox, reply):

From: Daniel Dehennin <daniel.dehennin@baby-gnu.org>

To: 399724@bugs.debian.org

Subject: Re: Bug#399724: sparc64: klibc segfaults again

Date: Sat, 03 Nov 2007 16:26:53 +0100

[Message part 1 (text/plain, inline)]

Hello,

I have exaclty the same problem on my sparc64 system up-to-date sid
system, all the klibc-utils segfault.

klibc-utils version 1.5.7-2

strace -fF /usr/lib/klibc/bin/fstype /dev/md1
6846  execve("/usr/lib/klibc/bin/fstype", ["/usr/lib/klibc/bin/fstype", "/dev/md1"], [/* 27 vars */]) = 0
6846  --- SIGSEGV (Segmentation fault) @ 0 (0) ---
6846  +++ killed by SIGSEGV +++

file /usr/lib/klibc/bin/fstype
/usr/lib/klibc/bin/fstype: ELF 64-bit MSB executable, SPARC V9,
version 1 (SYSV), statically linked (uses shared libs), stripped

Regards.
-- 
Daniel Dehennin
Récupérer ma clef GPG:
gpg --keyserver pgp.mit.edu --recv-keys 0x6A2540D1

[Message part 2 (application/pgp-signature, inline)]

Message #95 received at 399724@bugs.debian.org (full text, mbox, reply):

From: Bernd Zeimetz <bernd@bzed.de>

To: 399724@bugs.debian.org, Debian Bug Tracking System <control@bugs.debian.org>

Subject: rising severity - should not migrate to testing

Date: Sun, 04 Nov 2007 15:20:09 +0100

severity 399724 grave
thanks

As somebody else confirmed this bug I'm rising the severity as a
migration of klibc-utils in the current state would make sure that every
sparc machine running testing and using initramfs-tools (which is the
default, also on fresh installations) would not be able to boot.

Please let me know if I can help with anything to track this bug down.


Cheers,

Bernd

-- 
Bernd Zeimetz
<bernd@bzed.de>                         <http://bzed.de/>

Message #102 received at 399724@bugs.debian.org (full text, mbox, reply):

From: maximilian attems <max@stro.at>

To: Bernd Zeimetz <bernd@bzed.de>, 399724@bugs.debian.org

Subject: Re: Bug#399724: rising severity - should not migrate to testing

Date: Sun, 4 Nov 2007 17:42:19 +0100

On Sun, 04 Nov 2007, Bernd Zeimetz wrote:

> As somebody else confirmed this bug I'm rising the severity as a
> migration of klibc-utils in the current state would make sure that every
> sparc machine running testing and using initramfs-tools (which is the
> default, also on fresh installations) would not be able to boot.

bah klibc is frozen anyway, so without hint goes nowhere.
anyway leaving ia64 b0rked in testing for quite some time is not nice,
so thanks to your info could you try to build on sparc (yes stupid arch
without accessible porter box), latest klibc with this patch applied
to debian/rules and test the build.

thanks
 
-- 
maks


--- klibc-1.5.7/debian/rules.orig	2007-11-04 17:40:57.000000000 +0100
+++ klibc-1.5.7/debian/rules	2007-11-04 17:41:12.000000000 +0100
@@ -9,9 +9,6 @@ endif
 ifeq ($(DEB_ARCH),ia64)
 DEB_MAKE_ENVVARS := ARCH=ia64
 endif
-ifeq ($(DEB_ARCH),sparc)
-DEB_MAKE_ENVVARS := ARCH=sparc64
-endif
 
 DEB_MAKE_INVOKE := make
 DEB_MAKE_CLEAN_TARGET := clean

Message #107 received at 399724@bugs.debian.org (full text, mbox, reply):

From: Bernd Zeimetz <bernd@bzed.de>

To: maximilian attems <max@stro.at>

Cc: 399724@bugs.debian.org

Subject: Re: Bug#399724: rising severity - should not migrate to testing

Date: Sun, 04 Nov 2007 17:48:03 +0100

maximilian attems wrote:
> On Sun, 04 Nov 2007, Bernd Zeimetz wrote:
> 
>> As somebody else confirmed this bug I'm rising the severity as a
>> migration of klibc-utils in the current state would make sure that every
>> sparc machine running testing and using initramfs-tools (which is the
>> default, also on fresh installations) would not be able to boot.
> 
> bah klibc is frozen anyway, so without hint goes nowhere.

I know, but even such accidents happen. Pretty annoying, I know.

> anyway leaving ia64 b0rked in testing for quite some time is not nice,
> so thanks to your info could you try to build on sparc (yes stupid arch
> without accessible porter box), latest klibc with this patch applied
> to debian/rules and test the build.

I'll build it in a few hours.
If that doesn't fix it - I can organize you access to a sparc machine.

Cheers,

Bernd
-- 
Bernd Zeimetz
<bernd@bzed.de>                         <http://bzed.de/>

Message #112 received at 399724@bugs.debian.org (full text, mbox, reply):

From: Bernd Zeimetz <bernd@bzed.de>

To: maximilian attems <max@stro.at>, 399724@bugs.debian.org

Subject: Re: Bug#399724: rising severity - should not migrate to testing

Date: Mon, 05 Nov 2007 00:38:02 +0100

maximilian attems wrote:
> On Sun, 04 Nov 2007, Bernd Zeimetz wrote:
> 
>> As somebody else confirmed this bug I'm rising the severity as a
>> migration of klibc-utils in the current state would make sure that every
>> sparc machine running testing and using initramfs-tools (which is the
>> default, also on fresh installations) would not be able to boot.
> 
> bah klibc is frozen anyway, so without hint goes nowhere.
> anyway leaving ia64 b0rked in testing for quite some time is not nice,
> so thanks to your info could you try to build on sparc (yes stupid arch
> without accessible porter box), latest klibc with this patch applied
> to debian/rules and test the build.

tested, no difference. Still segfaults at the same place as before.



-- 
Bernd Zeimetz
<bernd@bzed.de>                         <http://bzed.de/>

Message #117 received at 399724@bugs.debian.org (full text, mbox, reply):

From: Alan Atwood <Alan.Atwood@audiocodes.com>

To: <399724@bugs.debian.org>

Subject: sparc64: klibc segfaults again

Date: Thu, 8 Nov 2007 16:54:01 -0600

I'm also seeing this issue on etch (4.0r1).  Occurs while setting up
ramdisk from mini-boot, net boot, and full CD-1 install.

HW is "Netra AX1105-500 (UltraSparc-IIe 500MHz)" built by Tatung.

FWIW Gentoo and NetBSD fail in exactly the same place, while OpenBSD
will work correctly.

Alan

Message #122 received at 399724@bugs.debian.org (full text, mbox, reply):

From: Blars Blarson <blarson@blars.org>

To: 399724@bugs.debian.org

Subject: have problem on ultra 10, not on utra 60

Date: Tue, 13 Nov 2007 16:57:56 -0800

I have what looks like this problem on an ultra-10 I just upgraded
from etch to sid, but not on an ultra-60 running sid.

Maybe it's Ultra IIi only (ultra-60 is Ultra II).

-- 
Blars Blarson			blarson@blars.org
				http://www.blars.org/blars.html
With Microsoft, failure is not an option.  It is a standard feature.

Message #127 received at 399724@bugs.debian.org (full text, mbox, reply):

From: Alan Atwood <aeatwood2@yahoo.com>

To: 399724@bugs.debian.org

Subject: sparc64: klibc segfaults again

Date: Wed, 14 Nov 2007 11:17:03 -0800 (PST)

[Message part 1 (text/plain, inline)]

I think Blars might be on to it here.  My install is on a IIe, and the actual message isn't segfault, it is "Illegal Instruction".
 
 The machines I was trying to use were some old abandoned boxen at work.  If I can buy the 3 of them, I'll try to get a cross compiler working at home and experiment.  I might even be able to make 1 available to the sparc team for testing, but I only have an ISDN line.  That might not be very functional.
 
 Alan 

       
---------------------------------
Be a better pen pal. Text or chat with friends inside Yahoo! Mail. See how.

[Message part 2 (text/html, inline)]

Message #132 received at 399724@bugs.debian.org (full text, mbox, reply):

From: "Bernhard R. Link" <brlink@debian.org>

To: 399724@bugs.debian.org

Cc: debian-sparc@lists.debian.org

Subject: klibc problem on sparc might be caused by 32/64 bit strangeness of gcc

Date: Thu, 22 Nov 2007 17:21:17 +0100

Looking into this problem, to find out why klibc-utils segfault, I found
some things that might be related:

When I replace the ARCH=sparc64 in debian/rules with ARCH=sparc I get
working binaries. Both the static binaries and the dynamic binaries
seem to work. (I only tried to run the static binaries and install the
dynamic ones into a chroot and run /bin/sh.shared in there and issue all
kind of command I could try, which did all work, while the one compiled
with sparc64 always segfault (even the static binaries)). I did not
try anything more sophisticated or creating an initrd with it.
(I personally use yaird, so am not effected by klibc).

Thus making it ARCH=sparc might be a workaround until the issue is
found. (But someone would need to verify that this works)

Something this might be related with, is that it always seem to include
the same libgcc.a file, both in 32 and in 64 bit.
(gcc seems to only have a single one, which I reported as
 http://bugs.debian.org/452402 )

Hochachtungsvoll,
	Bernhard R. Link

Message #137 received at 399724-close@bugs.debian.org (full text, mbox, reply):

From: maximilian attems <maks@debian.org>

To: 399724-close@bugs.debian.org

Subject: Bug#399724: fixed in klibc 1.5.7-3

Date: Thu, 22 Nov 2007 18:17:02 +0000

Source: klibc
Source-Version: 1.5.7-3

We believe that the bug you reported is fixed in the latest version of
klibc, which is due to be installed in the Debian FTP archive:

klibc-utils-floppy-udeb_1.5.7-3_amd64.udeb
  to pool/main/k/klibc/klibc-utils-floppy-udeb_1.5.7-3_amd64.udeb
klibc-utils-udeb_1.5.7-3_amd64.udeb
  to pool/main/k/klibc/klibc-utils-udeb_1.5.7-3_amd64.udeb
klibc-utils_1.5.7-3_amd64.deb
  to pool/main/k/klibc/klibc-utils_1.5.7-3_amd64.deb
klibc_1.5.7-3.diff.gz
  to pool/main/k/klibc/klibc_1.5.7-3.diff.gz
klibc_1.5.7-3.dsc
  to pool/main/k/klibc/klibc_1.5.7-3.dsc
libklibc-dev_1.5.7-3_amd64.deb
  to pool/main/k/klibc/libklibc-dev_1.5.7-3_amd64.deb
libklibc-udeb_1.5.7-3_amd64.udeb
  to pool/main/k/klibc/libklibc-udeb_1.5.7-3_amd64.udeb
libklibc_1.5.7-3_amd64.deb
  to pool/main/k/klibc/libklibc_1.5.7-3_amd64.deb



A summary of the changes between this version and the previous one is
attached.

Thank you for reporting the bug, which will now be closed.  If you
have further comments please address them to 399724@bugs.debian.org,
and the maintainer will reopen the bug report if appropriate.

Debian distribution maintenance software
pp.
maximilian attems <maks@debian.org> (supplier of updated klibc package)

(This message was generated automatically at their request; if you
believe that there is a problem with it please contact the archive
administrators by mailing ftpmaster@debian.org)


-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Format: 1.7
Date: Thu, 22 Nov 2007 18:50:55 +0100
Source: klibc
Binary: libklibc-dev klibc-utils-floppy-udeb libklibc-udeb libklibc klibc-utils klibc-utils-udeb
Architecture: source amd64
Version: 1.5.7-3
Distribution: unstable
Urgency: high
Maintainer: maximilian attems <maks@debian.org>
Changed-By: maximilian attems <maks@debian.org>
Description: 
 klibc-utils - small statically-linked utilities built with klibc
 klibc-utils-floppy-udeb - small utilities built with klibc for the boot floppy (udeb)
 klibc-utils-udeb - small statically-linked utilities built with klibc (udeb)
 libklibc   - minimal libc subset for use with initramfs
 libklibc-dev - kernel headers used during the build of klibc
 libklibc-udeb - minimal libc subset for use with initramfs (udeb)
Closes: 399724
Changes: 
 klibc (1.5.7-3) unstable; urgency=high
 .
   * debian/rules: s/sparc64/sparc/ thanks to the Debian sparc folks.
     (closes: #399724)
Files: 
 b161493ad42faf8f3e288ab67ba6063e 730 libs optional klibc_1.5.7-3.dsc
 6ccd7d30f14276b4bb5cc337be7d0380 10947 libs optional klibc_1.5.7-3.diff.gz
 0e288c58bc88c90ba8f04dc8bb8cf176 832068 libdevel optional libklibc-dev_1.5.7-3_amd64.deb
 b991b5aa667fb7c62847b30523fae1b5 46052 libs optional libklibc_1.5.7-3_amd64.deb
 1325dfb4ff260da326e1d9eff1f3a5cf 183994 libs optional klibc-utils_1.5.7-3_amd64.deb
 99ea6d81d1c809d1fd05d63d99bb20a4 173564 debian-installer optional klibc-utils-udeb_1.5.7-3_amd64.udeb
 87b697fb284d24e4cb933b4a319f4b7b 61810 debian-installer optional klibc-utils-floppy-udeb_1.5.7-3_amd64.udeb
 89338237383759854ebc04b82f90459b 36398 debian-installer optional libklibc-udeb_1.5.7-3_amd64.udeb
Package-Type: udeb

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.6 (GNU/Linux)

iD8DBQFHRcKJeW7Lc5tEHqgRAsFnAJ41teDYCIIHBJrV6shgRkjIlKvgYQCeODIt
Lfw8/ZsUu4X26cQXgXsw27I=
=l+bn
-----END PGP SIGNATURE-----

Send a report that this bug log contains spam.