Re: Bug#1003201: libc6: Upgrading to libc 2.33-1 causes lots of strange crashes

[Aurelien Jarno]

On 2022-01-06 05:36, Rich wrote:
> On Thu, Jan 6, 2022 at 5:22 AM Aurelien Jarno  wrote:
> 
> > On 2022-01-06 03:36, Rich wrote:
> > > Hi Aurelien,
> > > It's a VM running in qemu on an amd64 Debian bullseye system, no KVM
> > > acceleration to be found here.
> >
> > Ok, that might be a QEMU issue then. Which CPU do you emulate with QEMU?
> >
> 
> I don't explicitly specify a -M or -cpu, so whatever it defaults to, which
> according to -M help seems to be "pseries" mapping to pseries-6.1 here.

Thanks. That allowed me to reproduce the issue locally. I tracked down
it was due to the emulation of a POWER9 CPU, things work fine with a
POWER8 CPU. I found that it has been fixed recently in the upstream 2.33
branch:

https://sourceware.org/git/?p=glibc.git;a=commit;h=c493f6a0e4dcd6fff22da0df9fb2e52ecf41

-- 
Aurelien Jarno  GPG: 4096R/1DDD8C9B
aurel...@aurel32.net http://www.aurel32.net

Re: Bug#1003201: libc6: Upgrading to libc 2.33-1 causes lots of strange crashes

[Rich]

On Thu, Jan 6, 2022 at 5:22 AM Aurelien Jarno  wrote:

> On 2022-01-06 03:36, Rich wrote:
> > Hi Aurelien,
> > It's a VM running in qemu on an amd64 Debian bullseye system, no KVM
> > acceleration to be found here.
>
> Ok, that might be a QEMU issue then. Which CPU do you emulate with QEMU?
>

I don't explicitly specify a -M or -cpu, so whatever it defaults to, which
according to -M help seems to be "pseries" mapping to pseries-6.1 here.

- Rich

- Rich

>
> Regards,
> Aurelien
>
> --
> Aurelien Jarno  GPG: 4096R/1DDD8C9B
> aurel...@aurel32.net http://www.aurel32.net
>

Re: Bug#1003201: libc6: Upgrading to libc 2.33-1 causes lots of strange crashes

[Aurelien Jarno]

On 2022-01-06 03:36, Rich wrote:
> Hi Aurelien,
> It's a VM running in qemu on an amd64 Debian bullseye system, no KVM
> acceleration to be found here.

Ok, that might be a QEMU issue then. Which CPU do you emulate with QEMU?

Regards,
Aurelien

-- 
Aurelien Jarno  GPG: 4096R/1DDD8C9B
aurel...@aurel32.net http://www.aurel32.net

Re: Bug#1003201: libc6: Upgrading to libc 2.33-1 causes lots of strange crashes

[Rich]

https://www.dropbox.com/s/k117zefr83k6b11/ppc64%20bt.png?dl=0 is the
backtrace gdb reports from that core, helpful as it is.

I actually originally had this happen on qemu 5.2, then I upgraded to 6.1
to see if it went away (it does not, and it happily reproduces on fresh
upgrade each time).

- Rich

On Thu, Jan 6, 2022 at 3:36 AM Rich  wrote:

> Hi Aurelien,
> It's a VM running in qemu on an amd64 Debian bullseye system, no KVM
> acceleration to be found here.
>
> dmesg doesn't have any backtraces - the two messages that show up are
> py3compile segfaulting with all the addresses printed as  instead,
> and a couple of programs (like mandb) reporting getting a pointer of
> 0xfff1 or similar and dying in a fire.
>
> The first ones after the upgrade:
> Jan  6 01:30:39 encrepro kernel: [ 6715.078626] mandb[1903]: User access
> of kernel address (8408) - exploit attempt? (uid: 6)
> Jan  6 01:30:39 encrepro kernel: [ 6715.093977] mandb[1903]: segfault (11)
> at 8408 nip 7fffb37f5f28 lr 7fffb37f5f08 code 1 in
> libseccomp.so.2.5.3[7fffb37f+3]
> Jan  6 01:30:39 encrepro kernel: [ 6715.100149] mandb[1903]: code:
> fbe10078 3880 7c7f1b78 4bffddfd e8410028 2c03 41800030 ebe10078
> Jan  6 01:30:39 encrepro kernel: [ 6715.100308] mandb[1903]: code:
> 3860 38210080 6000 e8010010 <906283f8> 7c6307b4 7c0803a6 4e800020
> Jan  6 01:31:31 encrepro kernel: [ 6767.287646] reportbug[1982]: segfault
> (11) at 34c8 nip 34c8 lr 34c8 code 1 in python3.9[1000+5d]
> Jan  6 01:31:31 encrepro kernel: [ 6767.293334] reportbug[1982]: code:
>        
> Jan  6 01:31:31 encrepro kernel: [ 6767.293545] reportbug[1982]: code:
>        
>
> And later:
>
> Jan  6 01:35:30 encrepro systemd[2290]: free(): invalid pointer
>
> and
>
> Jan  6 01:42:53 encrepro systemd[1]: Created slice User Slice of UID 1000.
> Jan  6 01:42:53 encrepro systemd[1]: Starting User Runtime Directory
> /run/user/1000...
> Jan  6 01:42:53 encrepro systemd[1]: Finished User Runtime Directory
> /run/user/1000.
> Jan  6 01:42:53 encrepro systemd[1]: Starting User Manager for UID 1000...
> Jan  6 01:42:53 encrepro systemd[2370]: free(): invalid pointer
> Jan  6 01:42:54 encrepro systemd[1]: user@1000.service: Main process
> exited, code=killed, status=6/ABRT
> Jan  6 01:42:54 encrepro systemd[1]: user@1000.service: Failed with
> result 'signal'.
> Jan  6 01:42:54 encrepro systemd[1]: Failed to start User Manager for UID
> 1000.
>
> I've got a core dump from mandb:
> https://www.dropbox.com/s/4z6bfbuluwub29r/ppc64_libc?dl=0
>
> I don't have a stacktrace from it, though, since I didn't already have gdb
> on the VM, and it wants to upgrade libc to install. (I know I could go find
> an appropriately old section of snapshots.debian.org, but haven't done
> that yet...)
>
> - Rich
>
> On Thu, Jan 6, 2022 at 3:13 AM Aurelien Jarno 
> wrote:
>
>> control: tag -1 + help
>> control: user debian-powerpc@lists.debian.org
>> control: usertag -1 ppc64
>>
>> On 2022-01-06 01:45, Rich Ercolani wrote:
>> > Package: libc6
>> > Version: 2.33-1
>> > Severity: important
>> > X-Debbugs-Cc: rincebr...@gmail.com
>> >
>> > Dear Maintainer,
>> >
>> > (I marked this as serious because it's "just" ppc64, but the system is
>> permaneantly unusable if this upgrade is installed.)
>>
>> I have added the powerpc list in Cc: as the ppc64 porters are the people
>> who can help you there.
>>
>> > I booted my ppc64 VM in qemu 6.1, apt update, apt upgrade, and 20-30
>> packages in, it died horribly
>> > with Python3 packages erroring out with "Cannot get content of
>> [whatever package]".
>>
>> Is it a VM running with KVM or is it using QEMU emulation?
>>
>> > Trying to log into a shell over ssh or at a tty after this happens dies
>> with an error that flashes fast, but
>> > but seems to be "free(): invalid pointer"
>> >
>> > Random applications will now just crash out, in addition to the
>> obvious. (I'm writing this from a session
>> > spawned before the upgrade, which can still spawn children successfully
>> until I log out.)
>> >
>> > If I reboot after upgrading, all services fail to start on boot, and it
>> never spawns a login prompt or rescue
>> > prompt, it just sits forever on a list of failed service starts.
>> >
>> > Anything that would be helpful to debug this? I have a snapshot of the
>> VM before this began, so I can
>> > just roll it back and repeat the exercise.
>>
>> Ideally a backtrace would help, dmesg outputs can also be useful,
>> however given the state of you system, they might be difficult to get.
>>
>> Regards,
>> Aurelien
>>
>> --
>> Aurelien Jarno  GPG: 4096R/1DDD8C9B
>> aurel...@aurel32.net http://www.aurel32.net
>>
>

Re: Bug#1003201: libc6: Upgrading to libc 2.33-1 causes lots of strange crashes

[John Paul Adrian Glaubitz]

Hello!

On 1/6/22 09:13, Aurelien Jarno wrote:
>> (I marked this as serious because it's "just" ppc64, but the system is
>> permaneantly unusable if this upgrade is installed.)
> 
> I have added the powerpc list in Cc: as the ppc64 porters are the people
> who can help you there.
> 
>> I booted my ppc64 VM in qemu 6.1, apt update, apt upgrade, and 20-30 
>> packages in, it died horribly
>> with Python3 packages erroring out with "Cannot get content of [whatever 
>> package]".
> 
> Is it a VM running with KVM or is it using QEMU emulation?

I haven't seen any such issues on the powerpc/ppc64 buildds, but I will check 
whether
I can reproduce this problem on my iBook G4 which has an older processor in 
case this
is a regression that affects older machines only.

Adrian

-- 
 .''`.  John Paul Adrian Glaubitz
: :' :  Debian Developer - glaub...@debian.org
`. `'   Freie Universitaet Berlin - glaub...@physik.fu-berlin.de
  `-GPG: 62FF 8A75 84E0 2956 9546  0006 7426 3B37 F5B5 F913

Re: Bug#1003201: libc6: Upgrading to libc 2.33-1 causes lots of strange crashes

[Rich]

Hi Aurelien,
It's a VM running in qemu on an amd64 Debian bullseye system, no KVM
acceleration to be found here.

dmesg doesn't have any backtraces - the two messages that show up are
py3compile segfaulting with all the addresses printed as  instead,
and a couple of programs (like mandb) reporting getting a pointer of
0xfff1 or similar and dying in a fire.

The first ones after the upgrade:
Jan  6 01:30:39 encrepro kernel: [ 6715.078626] mandb[1903]: User access of
kernel address (8408) - exploit attempt? (uid: 6)
Jan  6 01:30:39 encrepro kernel: [ 6715.093977] mandb[1903]: segfault (11)
at 8408 nip 7fffb37f5f28 lr 7fffb37f5f08 code 1 in
libseccomp.so.2.5.3[7fffb37f+3]
Jan  6 01:30:39 encrepro kernel: [ 6715.100149] mandb[1903]: code: fbe10078
3880 7c7f1b78 4bffddfd e8410028 2c03 41800030 ebe10078
Jan  6 01:30:39 encrepro kernel: [ 6715.100308] mandb[1903]: code: 3860
38210080 6000 e8010010 <906283f8> 7c6307b4 7c0803a6 4e800020
Jan  6 01:31:31 encrepro kernel: [ 6767.287646] reportbug[1982]: segfault
(11) at 34c8 nip 34c8 lr 34c8 code 1 in python3.9[1000+5d]
Jan  6 01:31:31 encrepro kernel: [ 6767.293334] reportbug[1982]: code:
       
Jan  6 01:31:31 encrepro kernel: [ 6767.293545] reportbug[1982]: code:
       

And later:

Jan  6 01:35:30 encrepro systemd[2290]: free(): invalid pointer

and

Jan  6 01:42:53 encrepro systemd[1]: Created slice User Slice of UID 1000.
Jan  6 01:42:53 encrepro systemd[1]: Starting User Runtime Directory
/run/user/1000...
Jan  6 01:42:53 encrepro systemd[1]: Finished User Runtime Directory
/run/user/1000.
Jan  6 01:42:53 encrepro systemd[1]: Starting User Manager for UID 1000...
Jan  6 01:42:53 encrepro systemd[2370]: free(): invalid pointer
Jan  6 01:42:54 encrepro systemd[1]: user@1000.service: Main process
exited, code=killed, status=6/ABRT
Jan  6 01:42:54 encrepro systemd[1]: user@1000.service: Failed with result
'signal'.
Jan  6 01:42:54 encrepro systemd[1]: Failed to start User Manager for UID
1000.

I've got a core dump from mandb:
https://www.dropbox.com/s/4z6bfbuluwub29r/ppc64_libc?dl=0

I don't have a stacktrace from it, though, since I didn't already have gdb
on the VM, and it wants to upgrade libc to install. (I know I could go find
an appropriately old section of snapshots.debian.org, but haven't done that
yet...)

- Rich

On Thu, Jan 6, 2022 at 3:13 AM Aurelien Jarno  wrote:

> control: tag -1 + help
> control: user debian-powerpc@lists.debian.org
> control: usertag -1 ppc64
>
> On 2022-01-06 01:45, Rich Ercolani wrote:
> > Package: libc6
> > Version: 2.33-1
> > Severity: important
> > X-Debbugs-Cc: rincebr...@gmail.com
> >
> > Dear Maintainer,
> >
> > (I marked this as serious because it's "just" ppc64, but the system is
> permaneantly unusable if this upgrade is installed.)
>
> I have added the powerpc list in Cc: as the ppc64 porters are the people
> who can help you there.
>
> > I booted my ppc64 VM in qemu 6.1, apt update, apt upgrade, and 20-30
> packages in, it died horribly
> > with Python3 packages erroring out with "Cannot get content of [whatever
> package]".
>
> Is it a VM running with KVM or is it using QEMU emulation?
>
> > Trying to log into a shell over ssh or at a tty after this happens dies
> with an error that flashes fast, but
> > but seems to be "free(): invalid pointer"
> >
> > Random applications will now just crash out, in addition to the obvious.
> (I'm writing this from a session
> > spawned before the upgrade, which can still spawn children successfully
> until I log out.)
> >
> > If I reboot after upgrading, all services fail to start on boot, and it
> never spawns a login prompt or rescue
> > prompt, it just sits forever on a list of failed service starts.
> >
> > Anything that would be helpful to debug this? I have a snapshot of the
> VM before this began, so I can
> > just roll it back and repeat the exercise.
>
> Ideally a backtrace would help, dmesg outputs can also be useful,
> however given the state of you system, they might be difficult to get.
>
> Regards,
> Aurelien
>
> --
> Aurelien Jarno  GPG: 4096R/1DDD8C9B
> aurel...@aurel32.net http://www.aurel32.net
>

Re: Bug#1003201: libc6: Upgrading to libc 2.33-1 causes lots of strange crashes

[Aurelien Jarno]

control: tag -1 + help
control: user debian-powerpc@lists.debian.org
control: usertag -1 ppc64

On 2022-01-06 01:45, Rich Ercolani wrote:
> Package: libc6
> Version: 2.33-1
> Severity: important
> X-Debbugs-Cc: rincebr...@gmail.com
> 
> Dear Maintainer,
> 
> (I marked this as serious because it's "just" ppc64, but the system is 
> permaneantly unusable if this upgrade is installed.)

I have added the powerpc list in Cc: as the ppc64 porters are the people
who can help you there.

> I booted my ppc64 VM in qemu 6.1, apt update, apt upgrade, and 20-30 packages 
> in, it died horribly
> with Python3 packages erroring out with "Cannot get content of [whatever 
> package]".

Is it a VM running with KVM or is it using QEMU emulation?

> Trying to log into a shell over ssh or at a tty after this happens dies with 
> an error that flashes fast, but
> but seems to be "free(): invalid pointer"
> 
> Random applications will now just crash out, in addition to the obvious. (I'm 
> writing this from a session
> spawned before the upgrade, which can still spawn children successfully until 
> I log out.)
> 
> If I reboot after upgrading, all services fail to start on boot, and it never 
> spawns a login prompt or rescue
> prompt, it just sits forever on a list of failed service starts.
> 
> Anything that would be helpful to debug this? I have a snapshot of the VM 
> before this began, so I can
> just roll it back and repeat the exercise.

Ideally a backtrace would help, dmesg outputs can also be useful,
however given the state of you system, they might be difficult to get.

Regards,
Aurelien

-- 
Aurelien Jarno  GPG: 4096R/1DDD8C9B
aurel...@aurel32.net http://www.aurel32.net

Re: libc6-ppc64 on an Alubook 5,8?

[Johannes Berg]

Hi,

> Interesting. So this means it should be possible to compile some
> software on my 32-bit system for a ppc64 one? 

Yes, the gcc built for ppc by debian is biarch, you can use -m64 to
build a 64-bit binary.

> This is the important point I need to know. Thanks a lot .. :)

:)

johannes


signature.asc
Description: This is a digitally signed message part

Re: libc6-ppc64 on an Alubook 5,8?

[Wolfgang Pfeiffer]

Hi Johannes

On Thu, Jul 27, 2006 at 05:38:44PM +0200, Johannes Berg wrote:
> On Thu, 2006-07-27 at 16:41 +0200, Wolfgang Pfeiffer wrote:
> 
> > I'm relatively sure it was a decision by the package management system
> > to install it, and not my intention. And it looks as if it were
> > installed the first time with previous versions quite some time ago
> > ...
> 
> Well, you probably did a 'apt-get build-dep ...' or similar.

Maybe, yes. But the modification time of the first .deb package I see
in the local apt archives cache seems to be March this year ... I
don't remember what was going on at this time ...

> 
> > Does it make *any* sense to have this package being installed here?
> 
> Only if you want to compile packages that have ppc64 specific binaries,

Interesting. So this means it should be possible to compile some
software on my 32-bit system for a ppc64 one? 

> you'll never be able to actually use those libs.

This is the important point I need to know. Thanks a lot .. :)

Best Regards
Wolfgang

-- 
Wolfgang Pfeiffer: /ICQ: 286585973/ + + +  /AIM: crashinglinux/
http://profiles.yahoo.com/wolfgangpfeiffer

Key ID: E3037113
http://keyserver.mine.nu/pks/lookup?search=0xE3037113&fingerprint=on


-- 
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]

Re: libc6-ppc64 on an Alubook 5,8?

[Johannes Berg]

On Thu, 2006-07-27 at 16:41 +0200, Wolfgang Pfeiffer wrote:

> I'm relatively sure it was a decision by the package management system
> to install it, and not my intention. And it looks as if it were
> installed the first time with previous versions quite some time ago
> ...

Well, you probably did a 'apt-get build-dep ...' or similar.

> Does it make *any* sense to have this package being installed here?

Only if you want to compile packages that have ppc64 specific binaries,
you'll never be able to actually use those libs.

johannes


signature.asc
Description: This is a digitally signed message part

libc6-ppc64 on an Alubook 5,8?

[Wolfgang Pfeiffer]

Hi Everyone

With a recent packages upgrade for unstable I realised i have
libc6-ppc64 installed, on this machine:


$ cat /proc/cpuinfo
processor   : 0
cpu : 7447A, altivec supported
clock   : 1666.666000MHz
revision: 0.5 (pvr 8003 0105)
bogomips: 33.15
timebase: 832
platform: PowerMac
machine : PowerBook5,8
motherboard : PowerBook5,8 MacRISC3 Power Macintosh
detected as : 287 (PowerBook G4 15")
pmac flags  : 0019
L2 cache: 512K unified
pmac-generation : NewWorld


$ dpkg -l libc6-ppc64

 [ ... ]
+++-==-==-
ii  libc6-ppc642.3.6-16   GNU C Library: 64bit Shared libraries for Po


I'm relatively sure it was a decision by the package management system
to install it, and not my intention. And it looks as if it were
installed the first time with previous versions quite some time ago
...


As there's no useful documentation coming with the package to decide
it safely:
Does it make *any* sense to have this package being installed here?

Regards
Wolfgang
-- 
Wolfgang Pfeiffer: /ICQ: 286585973/ + + +  /AIM: crashinglinux/
http://profiles.yahoo.com/wolfgangpfeiffer

Key ID: E3037113
http://keyserver.mine.nu/pks/lookup?search=0xE3037113&fingerprint=on


-- 
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]

libc6

[petereasthope]

Finally I've managed to upgrade the PowerWave 
604|120, Woody -> Sarge -> Etch ... more or less.

The current problem is with libc6.  It appears 
to be stuck at Sarge and is not fully configured.
The report in running dselect is this.

dpkg: error processing libc6 (--configure):

Can anyone suggest how to resolve this or at 
least how I might find why it happens?

Thanks,... Peter Easthope

Desktops.OpenDoc  http://carnot.pathology.ubc.ca/


-- 
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]

Re: libdb1-compat vs. libc6

[Joerg Sommer]

Hello adrian,

adrian crisan <[EMAIL PROTECTED]> wrote:
> hi everybody, i'm trying to install xorg-6.8.99.15 on my beige g3/266
> ... however trying to install libdb1-compat "it depends on libc6" and
> (obvious) trying to install libc6 errors out because "depends on
> libdb1-compat"

You must install them in one transaction: dpkg -i libdb1-compat* libc6*

Regards, Jörg.
-- 
Geld allein macht nicht glücklich, aber es ist besser in einem Taxi zu
weinen, als in der Straßenbahn.
(Marcel Reich-Ranicki)


-- 
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]

libdb1-compat vs. libc6

[adrian crisan]

hi everybody, i'm trying to install  xorg-6.8.99.15 on my beige g3/266 ... 
however trying to install libdb1-compat
"it depends on libc6" and (obvious) trying to install libc6 errors out because 
"depends on libdb1-compat" ... need help please, i'm lost ... i'm in the 
chicken vs. egg dilemma ... thanks for hints/help ... best to all _adrian_


-- 
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]

Re: Bug#326220: libc6 2.3.5 ldconfig segfaults on powerpc oldworld

[Rogério Brito]

On Sep 04 2005, Sven Luther wrote:
> Ok, are you saying that you managed to build miboot floppies using the
> old woody miboot floppies,

Yes, I did.

> (i think woody came with do-it-yourself-miboot stuff, not sure).

I don't know anything about building miboot from scratch.

> This would then be a 2.6 kernel.

Indeed, I'm running:
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
[EMAIL PROTECTED]:~$ uname -a
Linux chagas 2.6.13-rc6-mm1-4.ow #1 Sun Aug 21 07:45:32 BRT 2005 ppc GNU/Linux
[EMAIL PROTECTED]:~$ 
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -

> Could you post me the config file for this kernel ?

Of course. It's attached to this mail.

> Could you also try the same kernel with the miboot from
> http://people.debain.org/~luther/miboot

I will try this later. Please, don't let me forget this.


Hope this helps, Rogério.

-- 
Rogério Brito : [EMAIL PROTECTED] : http://www.ime.usp.br/~rbrito
Homepage of the algorithms package : http://algorithms.berlios.de
Homepage on freshmeat:  http://freshmeat.net/projects/algorithms/


config-2.6.13-rc6-mm1-4.ow.bz2
Description: Binary data

Re: Bug#326220: libc6 2.3.5 ldconfig segfaults on powerpc oldworld

[Sven Luther]

On Wed, Sep 07, 2005 at 03:36:27AM -0300, Rogério Brito wrote:
> On Sep 04 2005, Sven Luther wrote:
> > Ok, are you saying that you managed to build miboot floppies using the
> > old woody miboot floppies,
> 
> Yes, I did.
> 
> > (i think woody came with do-it-yourself-miboot stuff, not sure).
> 
> I don't know anything about building miboot from scratch.
> 
> > This would then be a 2.6 kernel.
> 
> Indeed, I'm running:
> - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
> [EMAIL PROTECTED]:~$ uname -a
> Linux chagas 2.6.13-rc6-mm1-4.ow #1 Sun Aug 21 07:45:32 BRT 2005 ppc GNU/Linux
> [EMAIL PROTECTED]:~$ 
> - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
> 
> > Could you post me the config file for this kernel ?
> 
> Of course. It's attached to this mail.
> 
> > Could you also try the same kernel with the miboot from
> > http://people.debain.org/~luther/miboot
> 
> I will try this later. Please, don't let me forget this.

Ok, i will ping you next WE. if i don't hear from you before and don't forget
myself :)

Friendly,

Sven Luther


-- 
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]

Re: Bug#326220: libc6 2.3.5 ldconfig segfaults on powerpc oldworld

[Sven Luther]

On Sun, Sep 04, 2005 at 06:54:57AM -0300, Rogério Brito wrote:
> I know that this may be off-topic for the bug, but just for archiving
> purposes, I'm also sending it to the BTS.
> 
> On Sep 03 2005, Sven Luther wrote:
> > There has been some discussion about using hfs formatted floppies with
> > the kernel on it without miboot though.
> 
> Are you talking about the discussion we had on debian-powerpc a few days
> ago? If yes, then you perhaps misunderstood what I said.
> 
> I said that I the method I use is to take the HFS floppy from woody and
> replace its kernel with a new generated, bleeding-edge kernel (taken
> directly from Andrew Morton experimental tree).
> 
> I just checked and, yes, the fake Finder and System folders *are*
> miboot.
> 
> For more details, please see the comments I made to myself on
> .

Ok, are you saying that you managed to build miboot floppies using the old
woody miboot floppies, (i think woody came with do-it-yourself-miboot stuff,
not sure). This would then be a 2.6 kernel. Could you post me the config file
for this kernel ?

Could you also try the same kernel with the miboot from
http://people.debain.org/~luther/miboot

Then do :

gzip -9 vmlinux

dd if=/dev/zero of=miboot_floppy bs=1024 count=1440
hformat -l "miboot floppy" miboot_floppy

echo device miboot_floppy > miboot.conf
echo kernel vmlinux.gz root=0200 load_ramdisk=1 prompt_ramdisk=1 
devfs=mount debconf/priority=medium >> miboot.conf
miboot -c miboot.conf

If this works, it would be nice to do the following :

objcopy -O aixcoff-rs6000 -R .stab -R .stabstr -R .comment 
-add-section=image=vmlinux.gz dummy.o vmlinux.gz

just after the gzip step, where dummy.o is taken out of
arch/ppc/boot/openfirmware, you need to do a make zImage to compile it.

mkvmlinuz -a used by d-i currently does this second step, which may be cause
fo this problem.

Friendly,

Sven Luther


-- 
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]

Re: Bug#326220: libc6 2.3.5 ldconfig segfaults on powerpc oldworld

[Rogério Brito]

I know that this may be off-topic for the bug, but just for archiving
purposes, I'm also sending it to the BTS.

On Sep 03 2005, Sven Luther wrote:
> There has been some discussion about using hfs formatted floppies with
> the kernel on it without miboot though.

Are you talking about the discussion we had on debian-powerpc a few days
ago? If yes, then you perhaps misunderstood what I said.

I said that I the method I use is to take the HFS floppy from woody and
replace its kernel with a new generated, bleeding-edge kernel (taken
directly from Andrew Morton experimental tree).

I just checked and, yes, the fake Finder and System folders *are*
miboot.

For more details, please see the comments I made to myself on
.


Hope this clears any confusion, Rogério Brito.

-- 
Rogério Brito : [EMAIL PROTECTED] : http://www.ime.usp.br/~rbrito
Homepage of the algorithms package : http://algorithms.berlios.de
Homepage on freshmeat:  http://freshmeat.net/projects/algorithms/


-- 
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]

Re: Bug#326220: libc6 2.3.5 ldconfig segfaults on powerpc oldworld

[Michael Schmitz]

> > So please reassign the bug to whatever seems appropriate. Not being able
> > to upgrade from woody to current testing/unstable is what I'd consider a
> > bug in its own right.
>
> It's seems you are not aware that Sarge is out and is now the stable

I'm well aware of that, thank you very much.

> version. Debian does not support the upgrade from stable (Woody) to
> stable + 2 (the future Etch). You have to upgrade to stable + 1 (Sarge)
> to first.

That's what I've been missing. Please note that this used to be different
in earlier releases, so I had a reasonable expectation this would work.
And there was no warning during the attempted upgrade to make me
reconsider.

> So please upgrade to Sarge first.

What, now that I've already shoehorned testing onto the box? You gotta be
kidding. What would this solve?

> > Besides, what's the proposed fix? etch ldconfig just doesn't work, so what
> > piece am I missing?
>
> Update to Sarge first, and also to a Sarge kernel.

Please add a check to libc6 preinst that makes sure the running kernel (as
determined by uname -r, just like is already done for i386, sparc sun4m,
hppa, parisc64, mips and amd64) is recent enough for powerpc ldconfig from
libc6 2.3.5 to work. You have this check for other architectures to
prevent bad things from happening, so apparently permitting ldconfig
breakage on powerpc is a bug, and a serious one at that.

To point out the not-so-obvious: I recovered my system just fine. I ask
for such a check to be added for the next person to upgrade from woody.

I'm willing to help testing just what kernel feature is missing in my
kernel, or what exact kernel version would be OK to use. I indicated as
much in my initial bug report. If you already know the appropriate kernel
versions, why not have that check? If you don't know them, why do you give
me that 'dont do that, stupid' crap instead of taking me up on my offer to
help debug this?

To put this into perspective: I need this old G3 as a fallback developer
machine to debug a problem with nvsetenv on OldWorld, as long as my
PowerBook is being repaired. While updating that G3, I trip over a severe
libc6 bug, and report it. Color me clueless for assuming these reports are
actually welcomed.

I'm getting seriously fed up with this overgrown project. Any other
distributions out there with decent powerpc support?

Michael


-- 
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]

Re: Bug#326220: libc6 2.3.5 ldconfig segfaults on powerpc oldworld

[Aurelien Jarno]

Michael Schmitz a écrit :

reassign 326220 apt



Your system looks older than a testing from April 2005. libc6 version
2.2.5-11.8 is a libc6 from Woody, an Debian does not support direct
Woody -> Etch upgrade.



So please reassign the bug to whatever seems appropriate. Not being able
to upgrade from woody to current testing/unstable is what I'd consider a
bug in its own right.


It's seems you are not aware that Sarge is out and is now the stable 
version. Debian does not support the upgrade from stable (Woody) to 
stable + 2 (the future Etch). You have to upgrade to stable + 1 (Sarge) 
to first.


So please upgrade to Sarge first.



Besides, what's the proposed fix? etch ldconfig just doesn't work, so what
piece am I missing?


Update to Sarge first, and also to a Sarge kernel.


--
  .''`.  Aurelien Jarno | GPG: 1024D/F1BCDB73
 : :' :  Debian GNU/Linux developer | Electrical Engineer
 `. `'   [EMAIL PROTECTED] | [EMAIL PROTECTED]
   `-people.debian.org/~aurel32 | www.aurel32.net


--
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]

Re: Bug#326220: libc6 2.3.5 ldconfig segfaults on powerpc oldworld

[Sven Luther]

On Sat, Sep 03, 2005 at 11:50:09AM +0200, Michael Schmitz wrote:
> > > Well, apt or dpkg should have figured that out, and warned me off to not
> > > attempt the upgrade, right? Yet another bug.
> > >
> > > The system is booted using a floppy, I'm not sure I can even fit a 2.4 or
> > > 2.6 kernel on there. Either way, 'you should not try that' is not an
> >
> > The 2.6.8 sarge kernel did fit on a miboot floppy just fine, thank you, 
> > miboot
> 
> Thanks for pointing this out. Now what in libc6 2.3.5 doesn't work anymore
> with 2.2 kernels, and why doesn't libc6 test for the running kernel
> version before attempting an upgrade?
> 
> Should this bug be assigned to apt, or libc6, or what?
> 
> I'll try your 2.6.8 sometime; is the stuff required to build miboot

Well, we only really tried miboot in debian-installer setup, and i never
really had success with it with 2.6 floppies, except two mysterious times in
oldenbourg one year ago. We use 2.4 miboot floppies there. There has been some
discussion about using hfs formatted floppies with the kernel on it without
miboot though.

> floppies in the powerpc kernel source, and would building a minimal 2.6.13
> kernel fit on floppy?

Yes, with 2.6.8 though the box died when switching the framebuffer on or
something.

you need the miboot package from p.d.o/~luther/miboot, and can look at the
instructions on how to use it in the d-i build stuff.

Friendly,

Sven LUther


-- 
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]

Re: Bug#326220: libc6 2.3.5 ldconfig segfaults on powerpc oldworld

[Michael Schmitz]

> > Well, apt or dpkg should have figured that out, and warned me off to not
> > attempt the upgrade, right? Yet another bug.
> >
> > The system is booted using a floppy, I'm not sure I can even fit a 2.4 or
> > 2.6 kernel on there. Either way, 'you should not try that' is not an
>
> The 2.6.8 sarge kernel did fit on a miboot floppy just fine, thank you, miboot

Thanks for pointing this out. Now what in libc6 2.3.5 doesn't work anymore
with 2.2 kernels, and why doesn't libc6 test for the running kernel
version before attempting an upgrade?

Should this bug be assigned to apt, or libc6, or what?

I'll try your 2.6.8 sometime; is the stuff required to build miboot
floppies in the powerpc kernel source, and would building a minimal 2.6.13
kernel fit on floppy?

Michael



-- 
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]

Re: Bug#326220: libc6 2.3.5 ldconfig segfaults on powerpc oldworld

[Sven Luther]

On Sat, Sep 03, 2005 at 10:30:22AM +0200, Michael Schmitz wrote:
> reassign 326220 apt
> 
> > Your system looks older than a testing from April 2005. libc6 version
> > 2.2.5-11.8 is a libc6 from Woody, an Debian does not support direct
> > Woody -> Etch upgrade.
> 
> So please reassign the bug to whatever seems appropriate. Not being able
> to upgrade from woody to current testing/unstable is what I'd consider a
> bug in its own right.
> 
> Besides, what's the proposed fix? etch ldconfig just doesn't work, so what
> piece am I missing?
> 
> IIRC woody was still stable at that time, so it isn't all that outdated.
> 
> > > kernel version: 2.2.20-pmac (Debian package, unknown version)
> >
> > This is a rather old version, not even available in Debian Sarge.
> >
> > I suggest you to upgrade to Debian Sarge first (and also your kernel to
> > a 2.4 or 2.6 version), before trying to upgrade to testing or unstable.
> 
> Well, apt or dpkg should have figured that out, and warned me off to not
> attempt the upgrade, right? Yet another bug.
> 
> The system is booted using a floppy, I'm not sure I can even fit a 2.4 or
> 2.6 kernel on there. Either way, 'you should not try that' is not an

The 2.6.8 sarge kernel did fit on a miboot floppy just fine, thank you, miboot
is not in main though because of the apple-owned boot sector we have no
licence for, but you can find it on :

  http://people.debian.org/~luther/miboot

The current 2.6.12 sid kernel is too big by around 200K, but we will work it
to fit on a miboot size again before the etch release, hopefully, and by then
we may even have quik-floppy support or whatever, so ...

Friendly,

Sven Luther


-- 
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]

Re: Bug#326220: libc6 2.3.5 ldconfig segfaults on powerpc oldworld

[Michael Schmitz]

reassign 326220 apt

> Your system looks older than a testing from April 2005. libc6 version
> 2.2.5-11.8 is a libc6 from Woody, an Debian does not support direct
> Woody -> Etch upgrade.

So please reassign the bug to whatever seems appropriate. Not being able
to upgrade from woody to current testing/unstable is what I'd consider a
bug in its own right.

Besides, what's the proposed fix? etch ldconfig just doesn't work, so what
piece am I missing?

IIRC woody was still stable at that time, so it isn't all that outdated.

> > kernel version: 2.2.20-pmac (Debian package, unknown version)
>
> This is a rather old version, not even available in Debian Sarge.
>
> I suggest you to upgrade to Debian Sarge first (and also your kernel to
> a 2.4 or 2.6 version), before trying to upgrade to testing or unstable.

Well, apt or dpkg should have figured that out, and warned me off to not
attempt the upgrade, right? Yet another bug.

The system is booted using a floppy, I'm not sure I can even fit a 2.4 or
2.6 kernel on there. Either way, 'you should not try that' is not an
option if there's no warning during the upgrade. Once ldconfig has been
installed, the upgrade is shot. Nice way to tell your average user 'just
wipe the system and install ubuntu'. Debian used to pride itself on being
able to upgrade from oldstable to unstable without a hitch, and I've done
that before, even remotely.

Reassigning to apt, CC to debian-powerpc as a warning to others. Repeat
after me: 'it's not libc's fault if ldconfig breaks'.

Michael


-- 
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]

Re: WARNING: libc6 in unstable

[Mich Lanners]

On  12 Jul, this message from Michel Dänzer echoed through cyberspace:
>> Yeah; not a lot you can do... I was able to mke it boot by specifying
>> init=/bin/sh. However, I found no way to downgrade libc6 before I
>> shot myself in the foot and was forced to boot from another disk
> 
> Wasn't the old libc6 still in /var/cache/apt/archives for you to dpkg
> -i
> ?

Probably yes... But for some reason I didn't see that as a viable
option. Maybe because the bug reports talked about segfaults during
libc6 upgrade...

Cheers

Michel

-
Michel Lanners |  " Read Philosophy.  Study Art.
23, Rue Paul Henkes|Ask Questions.  Make Mistakes.
L-1710 Luxembourg  |
email   [EMAIL PROTECTED]|
http://www.cpu.lu/~mlan| Learn Always. "

Re: WARNING: libc6 in unstable

[Chris Tillman]

On Sat, Jul 12, 2003 at 10:13:46PM -0400, David M. Cooke wrote:
> On Sat, Jul 12, 2003 at 11:03:32AM -0700, Chris Tillman wrote:
> [...]
> > Since re-installing the base system overwrites the status file, my
> > machine had no idea what software had been installed. I wonder if a
> > utility for apt would be useful to reconstruct the status file from
> > what's on disk.
> 
> Does it overwrite the backup of dpkg.status in /var/backups? (There's a
> cron job that backs up key config files every day).

I don't know when it happened, but by the time I got there status.old
was the same as status. It would have been great to know about
/var/backups!! I'll submit a change to the debian-reference.

-- 
Debian GNU/Linux Operating System
  By the People, For the People
Chris Tillman (a people instance)
   toff one at cox dot net

Re: WARNING: libc6 in unstable

[Menaka Lashitha Bandara]

On Sat, 2003-07-12 at 17:33, Mich Lanners wrote:
> To those that are tracking unstable and haven't noticed (or haven't been
> caught :-) yet:

I've been caught allright. And recovery was a 3 hour process!

If anyone needs the libc6 2.3.1-16 packages, they can get it from
ftp.debian.org. If not, mail me, and I'll upload it to my web space.

If you can't downgrade, then the best thing to do is to get
libc6-2.3.1-16, dpkg -X it  (ie extract it to somewhere safe, ie /root)
remove libc6-2.3.1-17 (this will make your system break!), and then boot
using a cd  (like debian install), mount your disk, and do a cp -dpRvf
of the extracted to your target disk. Then reboot, and dpkg --install
your libc6-2.3.1-16 just to sat dependencies and make things clean.

When this problem occured for me, I could not install libc6-2.3.1-16
because I couldn't even chroot. So extracting by hand and then
reinstalling was the only option really.

\LaShI

-- 
Debian GNU/Linux 3.0 (Sid) (ibook2,powerpc), kernel 2.4.21-ben2-xfs

Grub first, then ideals.
-- Bertolt Brecht

<[EMAIL PROTECTED]>
ICQ: 7181075

pub  1024D/CC1F67BC 2002-06-03 Menaka Lashitha Bandara (Freedom!)
<[EMAIL PROTECTED]>
 Key fingerprint = 1874 5597 C754 5209 0A7D  4DA6 9112 5D0E CC1F
67BC


signature.asc
Description: This is a digitally signed message part

Re: WARNING: libc6 in unstable

[Leandro Guimarães Faria Corsetti Dutra]

Em Sat, 12 Jul 2003 23:48:56 -0500, Rick Cook escreveu:

> I upgraded to that 17.0.2 (and rebooted) about 7 hours ago. So far, no 
> problems.

Thanks, same here from half an hour.


-- 
 _
/ \  Leandro Guimarães Faria Corsetti Dutra +41 (21) 648 11 34
\ /  http://br.geocities.com./lgcdutra/ +41 (78) 778 11 34
/ \  Responda à lista, não a mim diretamente!   +55 (11) 5686 2219
Dê-me nota se te ajudei:   http://svcs.affero.net/rm.php?r=leandro

Re: WARNING: libc6 in unstable

[Rick Cook]

On Saturday 12 July 2003 21:39, Leandro Guimarães Faria Corsetti Dutra 
wrote:
> Em Sat, 12 Jul 2003 09:33:22 +0200, Mich Lanners escreveu:
> > DON'T upgrade to libc6-2.3.1-17.0.1 in unstable. It breaks your system!
>
>   There is now libc6-2.3.1-17.0.2... anyone knows if that's safe?
>
>

I upgraded to that 17.0.2 (and rebooted) about 7 hours ago. So far, no 
problems.


Rick Cook

Re: WARNING: libc6 in unstable

[Leandro Guimarães Faria Corsetti Dutra]

Em Sat, 12 Jul 2003 09:33:22 +0200, Mich Lanners escreveu:

> DON'T upgrade to libc6-2.3.1-17.0.1 in unstable. It breaks your system!

There is now libc6-2.3.1-17.0.2... anyone knows if that's safe?


-- 
 _
/ \  Leandro Guimarães Faria Corsetti Dutra +41 (21) 648 11 34
\ /  http://br.geocities.com./lgcdutra/ +41 (78) 778 11 34
/ \  Responda à lista, não a mim diretamente!   +55 (11) 5686 2219
Dê-me nota se te ajudei:   http://svcs.affero.net/rm.php?r=leandro

Re: WARNING: libc6 in unstable

[David M. Cooke]

On Sat, Jul 12, 2003 at 11:03:32AM -0700, Chris Tillman wrote:
[...]
> Since re-installing the base system overwrites the status file, my
> machine had no idea what software had been installed. I wonder if a
> utility for apt would be useful to reconstruct the status file from
> what's on disk.

Does it overwrite the backup of dpkg.status in /var/backups? (There's a
cron job that backs up key config files every day).

-- 
|>|\/|<
/--\
|David M. Cooke  http://arbutus.physics.mcmaster.ca/dmc/
|[EMAIL PROTECTED]

Re: WARNING: libc6 in unstable

[Leandro Guimarães Faria Corsetti Dutra]

Em Sat, 12 Jul 2003 22:09:00 +0200, Mich Lanners escreveu:

> On  12 Jul, this message from Leandro Guimarães Faria Corcete Dutra
> echoed through cyberspace:
>>
>>  Is there a FTP there, or does the install CD has some http utility?
> 
> No FTP, sorry. But doesn't the installer have wget?

Yes, it does but not quite... it's in busybox, and I had
forgotten about busybox; since there is no command line completion, I
felt left in the cold...

One note for whomever else is caught unawares: you have to
activate the network, obviously... it's still not activated when you
mount the / and switch to the alternate console.


-- 
 _
/ \  Leandro Guimarães Faria Corsetti Dutra +41 (21) 648 11 34
\ /  http://br.geocities.com./lgcdutra/ +41 (78) 778 11 34
/ \  Responda à lista, não a mim diretamente!   +55 (11) 5686 2219
Dê-me nota se te ajudei:   http://svcs.affero.net/rm.php?r=leandro

Re: WARNING: libc6 in unstable

[Chris Tillman]

On Sat, Jul 12, 2003 at 10:09:00PM +0200, Mich Lanners wrote:
> On  12 Jul, this message from Leandro Guimarães Faria Corcete Dutra
> echoed through cyberspace:
> > Em Sat, 12 Jul 2003 11:03:59 +0200, Mich Lanners escreveu:
> > 
> >> http://www.cpu.lu/~mlan/ftp/debian/libc6/
> >
> > Is there a FTP there, or does the install CD has some http utility?
> 
> No FTP, sorry. But doesn't the installer have wget?

yes, it does.

-- 
Debian GNU/Linux Operating System
  By the People, For the People
Chris Tillman (a people instance)
   toff one at cox dot net

Re: WARNING: libc6 in unstable

[Mich Lanners]

On  12 Jul, this message from Leandro Guimarães Faria Corcete Dutra
echoed through cyberspace:
> Em Sat, 12 Jul 2003 11:03:59 +0200, Mich Lanners escreveu:
> 
>> http://www.cpu.lu/~mlan/ftp/debian/libc6/
>
>   Is there a FTP there, or does the install CD has some http utility?

No FTP, sorry. But doesn't the installer have wget?

Cheers

Michel

-
Michel Lanners |  " Read Philosophy.  Study Art.
23, Rue Paul Henkes|Ask Questions.  Make Mistakes.
L-1710 Luxembourg  |
email   [EMAIL PROTECTED]|
http://www.cpu.lu/~mlan| Learn Always. "

Re: WARNING: libc6 in unstable

[Leandro Guimarães Faria Corcete Dutra]

Em Sat, 12 Jul 2003 11:03:59 +0200, Mich Lanners escreveu:

> http://www.cpu.lu/~mlan/ftp/debian/libc6/

Is there a FTP there, or does the install CD has some http utility?


-- 
Leandro Dutra

Re: WARNING: libc6 in unstable

[Chris Tillman]

On Sat, Jul 12, 2003 at 01:49:20PM -0400, Pan wrote:
> I'm stuck on the same place.
> 
> The problem right now is I can't boot from the cdrom.
> 
> I have a blue&white g3 (450) and I can't get it to boot of the cd. I
> have 'boot of cd' and 'boot of of' on the yaboot menu, but apparently
> yaboot can't reach my cd?
> 
> I tried booting directly to OF (cmd+apl+O+F) and typing from there all I
> could imagine:
> 
> boot cd:\\yaboot
> boot cd:,\\yaboot
> boot cd:2,\\yaboot
> boot cd:install\powermac\yaboot
> boot cd:install\powermac\vmlinux
> boot cd:,install\powermac\yaboot
> boot cd:,install\powermac\vmlinux
> boot cd:2,install\powermac\yaboot
> boot cd:2,install\powermac\vmlinux

Try boot cd:,\\install\\powermac\\yaboot

> 
> and none of it worked...
> 
> Usually I get sent back to the OF screen, and the 3 first ones land me
> on the yaboot menu of my broken debian install.

>From boot:, you could also try the OF path to the CD linux.bin file.

> 
> Any idea what I'm doing wrong?
> 
> Of course I also tried booting of the cd while pressing 'c', but it
> doesn't work

Another key combination to try is Ctrl-Option-Shift-Delete to force
CD booting.

Beware, don't re-install base (see my other post).

-- 
Debian GNU/Linux Operating System
  By the People, For the People
Chris Tillman (a people instance)
   toff one at cox dot net

Re: WARNING: libc6 in unstable

[Chris Tillman]

On Sat, Jul 12, 2003 at 09:33:22AM +0200, Mich Lanners wrote:
> Hi all,
> 
> To those that are tracking unstable and haven't noticed (or haven't been
> caught :-) yet:
> 
> DON'T upgrade to libc6-2.3.1-17.0.1 in unstable. It breaks your system!
> See for instance:
> http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=200887
> http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=200833
> http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=200827
> 
> If you have already upgraded, try to downgrade _before_ rebooting the
> system.
> 
> Although I had no problem when installing this version, my system
> wouldn't boot anymore afterwards: init segfaulted during startup.
> 
> Had to burn a root CD from another PPC system and boot that one to put
> back the old glibc...

I got burnt with this one too. I had also upgraded a few other things,
so I wasn't sure which was the problem. I ended up re-installing the
base system, which is NOT advised, believe me. I had to reinstall 
everything.

Since re-installing the base system overwrites the status file, my
machine had no idea what software had been installed. I wonder if a
utility for apt would be useful to reconstruct the status file from
what's on disk.

>From the debian-reference, I got the clue to use /usr/share/doc as a
package list, and combined that with a list gleaned from
/var/lib/dpkg/info using sed. Then I did dpkg --set-selections to set
their status to 'install'. The only problem was then I still had to
download 1G of packages which in reality were already here.

Does anyone know of a better solution for lusers like me who make
this mistake?

-- 
Debian GNU/Linux Operating System
  By the People, For the People
Chris Tillman (a people instance)
   toff one at cox dot net

Re: WARNING: libc6 in unstable

[Pan]

I'm stuck on the same place.

The problem right now is I can't boot from the cdrom.

I have a blue&white g3 (450) and I can't get it to boot of the cd. I
have 'boot of cd' and 'boot of of' on the yaboot menu, but apparently
yaboot can't reach my cd?

I tried booting directly to OF (cmd+apl+O+F) and typing from there all I
could imagine:

boot cd:\\yaboot
boot cd:,\\yaboot
boot cd:2,\\yaboot
boot cd:install\powermac\yaboot
boot cd:install\powermac\vmlinux
boot cd:,install\powermac\yaboot
boot cd:,install\powermac\vmlinux
boot cd:2,install\powermac\yaboot
boot cd:2,install\powermac\vmlinux

and none of it worked...

Usually I get sent back to the OF screen, and the 3 first ones land me
on the yaboot menu of my broken debian install.

Any idea what I'm doing wrong?

Of course I also tried booting of the cd while pressing 'c', but it
doesn't work

Benja

On Sat Jul 12, 2003 at 04:54:35PM +0100 or thereabouts, Joss Winn wrote:
> Thanks. That worked and I'm back into Debian.  I downloaded the
> tar.gz file and opened it in OSX.  Copied the files to my HFS partition and
> rebooted using the Debian install CD 1.  Mounted the / partition and
> the HFS partition and then copied the libc-2.3.1.so file over the
> broken one.
> 
> rebooted and all is fine.
> 
> Thanks
> 
> Joss
> 
> On Sat, Jul 12, 2003 at 11:03:59AM +0200, Mich Lanners wrote:
> > On  12 Jul, this message from Joss Winn echoed through cyberspace:
> > > Yep, this caught me out at midnight last night and I've yet to
> > > tackle it.
> > > I can only boot into runlevel two (where booting now stops).  It
> > > won't let me login though.
> > 
> > Yeah; not a lot you can do... I was able to mke it boot by specifying
> > init=/bin/sh. However, I found no way to downgrade libc6 before I shot
> > myself in the foot and was forced to boot from another disk
> > 
> > > My plan was to use the debian install cd 1 to boot and mount the
> > > linux partition.  I should be able to downgrade from there, right?
> > 
> > That should be possible... hmmm... I have put up a .tgz archive on my
> > site containing just libc and its associated symbolic link. When booting
> > from the install CD, you can untar that over the bad libc (from within
> > /target/lib).
> > 
> > It's available here:
> > 
> > http://www.cpu.lu/~mlan/ftp/debian/libc6/
> > 
> > > What is the downgraded command I should be using?
> > 
> > I usd dpkg; I'm sure some apt-get can also do it, but I was too lazy to
> > look it up. On my site I have put the relevant .debs; get them and
> > insall them thus:
> > 
> > dpkg -i libc6*deb locale*deb
> > 
> > Cheers
> > 
> > Michel
> > 
> > -
> > Michel Lanners |  " Read Philosophy.  Study Art.
> > 23, Rue Paul Henkes|Ask Questions.  Make Mistakes.
> > L-1710 Luxembourg  |
> > email   [EMAIL PROTECTED]|
> > http://www.cpu.lu/~mlan| Learn Always. "
> 
> -- 
> 
> 
> -- 
> To UNSUBSCRIBE, email to [EMAIL PROTECTED]
> with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]

Re: WARNING: libc6 in unstable

[Joss Winn]

Thanks. That worked and I'm back into Debian.  I downloaded the
tar.gz file and opened it in OSX.  Copied the files to my HFS partition and
rebooted using the Debian install CD 1.  Mounted the / partition and
the HFS partition and then copied the libc-2.3.1.so file over the
broken one.

rebooted and all is fine.

Thanks

Joss

On Sat, Jul 12, 2003 at 11:03:59AM +0200, Mich Lanners wrote:
> On  12 Jul, this message from Joss Winn echoed through cyberspace:
> > Yep, this caught me out at midnight last night and I've yet to
> > tackle it.
> > I can only boot into runlevel two (where booting now stops).  It
> > won't let me login though.
> 
> Yeah; not a lot you can do... I was able to mke it boot by specifying
> init=/bin/sh. However, I found no way to downgrade libc6 before I shot
> myself in the foot and was forced to boot from another disk
> 
> > My plan was to use the debian install cd 1 to boot and mount the
> > linux partition.  I should be able to downgrade from there, right?
> 
> That should be possible... hmmm... I have put up a .tgz archive on my
> site containing just libc and its associated symbolic link. When booting
> from the install CD, you can untar that over the bad libc (from within
> /target/lib).
> 
> It's available here:
> 
> http://www.cpu.lu/~mlan/ftp/debian/libc6/
> 
> > What is the downgraded command I should be using?
> 
> I usd dpkg; I'm sure some apt-get can also do it, but I was too lazy to
> look it up. On my site I have put the relevant .debs; get them and
> insall them thus:
> 
> dpkg -i libc6*deb locale*deb
> 
> Cheers
> 
> Michel
> 
> -
> Michel Lanners |  " Read Philosophy.  Study Art.
> 23, Rue Paul Henkes|Ask Questions.  Make Mistakes.
> L-1710 Luxembourg  |
> email   [EMAIL PROTECTED]|
> http://www.cpu.lu/~mlan| Learn Always. "

-- 

Re: WARNING: libc6 in unstable

[Michel Dänzer]

On Sat, 2003-07-12 at 11:03, Mich Lanners wrote:
> On  12 Jul, this message from Joss Winn echoed through cyberspace:
> > Yep, this caught me out at midnight last night and I've yet to
> > tackle it.
> > I can only boot into runlevel two (where booting now stops).  It
> > won't let me login though.
> 
> Yeah; not a lot you can do... I was able to mke it boot by specifying
> init=/bin/sh. However, I found no way to downgrade libc6 before I shot
> myself in the foot and was forced to boot from another disk

Wasn't the old libc6 still in /var/cache/apt/archives for you to dpkg -i
?


-- 
Earthling Michel Dänzer   \  Debian (powerpc), XFree86 and DRI developer
Software libre enthusiast  \ http://svcs.affero.net/rm.php?r=daenzer

Re: WARNING: libc6 in unstable

[[EMAIL PROTECTED]]

Hi,

Hu...I see I was not alone not sleeping last night ;-)

I've burned a CD too..and all is OK for now on my ibook  (Ouf !)


Best regards


--
eric b

Re: WARNING: libc6 in unstable

[Mich Lanners]

On  12 Jul, this message from Joss Winn echoed through cyberspace:
> Yep, this caught me out at midnight last night and I've yet to
> tackle it.
> I can only boot into runlevel two (where booting now stops).  It
> won't let me login though.

Yeah; not a lot you can do... I was able to mke it boot by specifying
init=/bin/sh. However, I found no way to downgrade libc6 before I shot
myself in the foot and was forced to boot from another disk

> My plan was to use the debian install cd 1 to boot and mount the
> linux partition.  I should be able to downgrade from there, right?

That should be possible... hmmm... I have put up a .tgz archive on my
site containing just libc and its associated symbolic link. When booting
from the install CD, you can untar that over the bad libc (from within
/target/lib).

It's available here:

http://www.cpu.lu/~mlan/ftp/debian/libc6/

> What is the downgraded command I should be using?

I usd dpkg; I'm sure some apt-get can also do it, but I was too lazy to
look it up. On my site I have put the relevant .debs; get them and
insall them thus:

dpkg -i libc6*deb locale*deb

Cheers

Michel

-
Michel Lanners |  " Read Philosophy.  Study Art.
23, Rue Paul Henkes|Ask Questions.  Make Mistakes.
L-1710 Luxembourg  |
email   [EMAIL PROTECTED]|
http://www.cpu.lu/~mlan| Learn Always. "

Re: WARNING: libc6 in unstable

[Joss Winn]

Yep, this caught me out at midnight last night and I've yet to
tackle it.
I can only boot into runlevel two (where booting now stops).  It
won't let me login though.

My plan was to use the debian install cd 1 to boot and mount the
linux partition.  I should be able to downgrade from there, right?

What is the downgraded command I should be using?

thanks,

Joss

On Sat, Jul 12, 2003 at 09:33:22AM +0200, Mich Lanners wrote:
> Hi all,
> 
> To those that are tracking unstable and haven't noticed (or haven't been
> caught :-) yet:
> 
> DON'T upgrade to libc6-2.3.1-17.0.1 in unstable. It breaks your system!
> See for instance:
> http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=200887
> http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=200833
> http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=200827
> 
> If you have already upgraded, try to downgrade _before_ rebooting the
> system.
> 
> Although I had no problem when installing this version, my system
> wouldn't boot anymore afterwards: init segfaulted during startup.
> 
> Had to burn a root CD from another PPC system and boot that one to put
> back the old glibc...
> 
> Cheers
> 
> Michel
> 
> -
> Michel Lanners |  " Read Philosophy.  Study Art.
> 23, Rue Paul Henkes|Ask Questions.  Make Mistakes.
> L-1710 Luxembourg  |
> email   [EMAIL PROTECTED]|
> http://www.cpu.lu/~mlan| Learn Always. "
> 
> 
> -- 
> To UNSUBSCRIBE, email to [EMAIL PROTECTED]
> with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
> 

-- 
Sent via SDF Public Access UNIX System - http://sdf.lonestar.org

WARNING: libc6 in unstable

[Mich Lanners]

Hi all,

To those that are tracking unstable and haven't noticed (or haven't been
caught :-) yet:

DON'T upgrade to libc6-2.3.1-17.0.1 in unstable. It breaks your system!
See for instance:
http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=200887
http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=200833
http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=200827

If you have already upgraded, try to downgrade _before_ rebooting the
system.

Although I had no problem when installing this version, my system
wouldn't boot anymore afterwards: init segfaulted during startup.

Had to burn a root CD from another PPC system and boot that one to put
back the old glibc...

Cheers

Michel

-
Michel Lanners |  " Read Philosophy.  Study Art.
23, Rue Paul Henkes|Ask Questions.  Make Mistakes.
L-1710 Luxembourg  |
email   [EMAIL PROTECTED]|
http://www.cpu.lu/~mlan| Learn Always. "

Re: better fix libbb than libc6-dev

[Gaudenz Steinlin]

tag 194631 patch
thanks

Here is a patch to fix the build problem of busybox-cvs on the
busybox-side. It includes the same code for powerpc, that was already
there for ia64 and some new ifdefs for powerpc. Not sure about this, but
perhaps other archs will need this too.

Plase can somebody with more c system programmming knowledge look at the
patch and eventually send it to the upstream maintainer.
--- libbb/syscalls.c.orig   2003-05-27 13:49:08.0 +0200
+++ libbb/syscalls.c2003-05-27 12:39:10.0 +0200
@@ -34,7 +34,7 @@
 #endif
 #include "libbb.h"
 
-#if defined(__ia64__)
+#if (defined(__ia64__) || defined(__powerpc__))
 int sysfs( int option, unsigned int fs_index, char * buf)
 {
return(syscall(__NR_sysfs, option, fs_index, buf));
@@ -59,7 +59,7 @@
return -1;
 }
 #else
-#  if defined(__ia64__)
+#  if (defined(__ia64__) || defined(__powerpc__))
int pivot_root(const char * new_root,const char * put_old)
{
return(syscall(__NR_pivot_root, new_root, put_old));
--- util-linux/fdisk.c.orig 2003-05-27 13:47:41.0 +0200
+++ util-linux/fdisk.c  2003-05-27 13:04:11.0 +0200
@@ -865,10 +865,21 @@
 
 #ifdef __NR__llseek
 
+#if defined(__powerpc__)
+static int _llseek (unsigned int f_d, unsigned long offset_high,
+  unsigned long offset_low, ext2_loff_t * result, 
+  unsigned int origin) {
+
+   return(syscall(__NR__llseek, f_d, offset_high, offset_low, result, 
origin));
+}
+
+#else
 static _syscall5(int,_llseek,unsigned int,f_d,unsigned long,offset_high,
 unsigned long, offset_low,ext2_loff_t *,result,
 unsigned int, origin)
 
+#endif
+
 #else
 
 /* no __NR__llseek on compilation machine - might give it explicitly */
--- libpwdgrp/setgroups.c.orig  2003-05-27 13:48:10.0 +0200
+++ libpwdgrp/setgroups.c   2003-05-27 13:13:29.0 +0200
@@ -38,5 +38,10 @@
 #include "grp_.h"
 
 //#define __NR_setgroups81
+#if defined(__powerpc__)
+int setgroups(size_t size, const gid_t * list) {
+   return(syscall(__NR_setgroups, size, list));
+}
+#else
 _syscall2(int, setgroups, size_t, size, const gid_t *, list);
-
+#endif

Re: libc6 (>= 2.2.5-13) & liba52-0.7.4

[erik]

On Fri, Dec 20, 2002 at 05:28:42PM -0500, Nirmal Govind wrote:
> > OTOH, the stock woody version doesn't play encrypted dvd's.  If your X
> > isn't as squirrelly as mine those lines in your sources.list should
> > let you install and compile from the source package and get around
> > your versioning problems.  (I'm using liba52-0.7.3 on x86 - my
> > laptop's at home and running OS X at the moment).
> > 
> 
> thanks Erik.. I managed (after several installs and reinstalls mainly
> due to the unorderly fashion in which I set it up) to get vlc
> working.. the volume however is very low.. I went to max vol in the
> gmix but doesn't help much.. when xmms plays mp3's at the same
> vol. level, it's very audible though... and I couldn't find any
> setting in Vlc for an audio output stream etc.. another thing is that
> as the vol is increased the quality goes down.. again, doesn't happen
> while using xmms to listen to music... I haven't installed alsa
> yet.. would that be the solution?


start up vlc

make sure you're not playing anything.

go to Settings->Preferences-> Audio tab.

Set "audio output volume" to 1024 (the max).  hit "OK".  Now when you
start playing it'll be MUCH louder.  AFAICS, vlc doesn't adjust the
system volume, instead it manipulates the volume of the audio it sends
to the audio output plugin.  Oh, and at least on the versions I've
got, it'll forget this setting each time you close the app - "save"
and "apply" don't seem to do a damned thing.

As for system volume, that's incredibly dependent on sound card,
speakers, etc.  If using OSS, make sure both the pcm (aumix -w) volume
is cranked and the main volume (aumix -v).  I've got a Turtle Beach
card that plays very softly and a I find that setting pcm to 90 then
adjusting volume with the main control gets the job done.  YMMV, of
course.

I've not played around with trying for the best sound quality from
vlc, but I'm betting there's some magical balance between "audio
output volume" and the system volume settings.  Personally I've never
found a card where setting the -v very high gave good quality - I
always want to get the volume up before or after that (e.g. volume on
speakers, tuner, input gain, etc).  Using "aumix -w90 -v80" on my
desktop is a desperation move - and I don't expect perfect sound to
come from it.  I'm impressed with how well this card handles it,
though.

Hope that helps.  I spent waaay too much time fiddling with things
trying to get vlc to work in all the ways I wanted it to.


Erik

Re: libc6 (>= 2.2.5-13) & liba52-0.7.4

[Nirmal Govind]

> OTOH, the stock woody version doesn't play encrypted dvd's.  If your X
> isn't as squirrelly as mine those lines in your sources.list should
> let you install and compile from the source package and get around
> your versioning problems.  (I'm using liba52-0.7.3 on x86 - my
> laptop's at home and running OS X at the moment).
> 

thanks Erik.. I managed (after several installs and reinstalls mainly due to 
the unorderly fashion in which I set it up) to get vlc working.. the volume 
however is very low.. I went to max vol in the gmix but doesn't help much.. 
when xmms plays mp3's at the same vol. level, it's very audible though... and I 
couldn't find any setting in Vlc for an audio output stream etc.. another thing 
is that as the vol is increased the quality goes down.. again, doesn't happen 
while using xmms to listen to music... I haven't installed alsa yet.. would 
that be the solution?

Thanks,
nirmal

Re: libc6 (>= 2.2.5-13) & liba52-0.7.4

[erik]

On Fri, Dec 20, 2002 at 01:27:51PM -0500, Nirmal Govind wrote:
>  Hi.. where do I get debs for liba52-0.7.4 and libc6 (>= 2.2.5-13)? I'm 
> trying to install videolan for playing dvd's and it needs these libs.. I have 
> libc6 but it's not >13.. I searched around but couldn't find >13 debs.. or do 
> I need to compile it from source?

There's a version in normal woody, which works.  The most recent I can
get from videolan.org debian repo won't play properly on my ibook
(looks like it doesn't work properly w/ x11 rendering).

deb http://www.videolan.org/pub/videolan/debian $(ARCH)/
deb-src http://www.videolan.org/pub/videolan/debian sources/

OTOH, the stock woody version doesn't play encrypted dvd's.  If your X
isn't as squirrelly as mine those lines in your sources.list should
let you install and compile from the source package and get around
your versioning problems.  (I'm using liba52-0.7.3 on x86 - my
laptop's at home and running OS X at the moment).


Erik


-- 
If I didn't ruin the mood by saying something stupid, 
   I wouldn't be me, would I?
 - Davan (www.somethingpositive.net)

libc6 (>= 2.2.5-13) & liba52-0.7.4

[Nirmal Govind]

 Hi.. where do I get debs for liba52-0.7.4 and libc6 (>= 2.2.5-13)? I'm trying 
to install videolan for playing dvd's and it needs these libs.. I have libc6 
but it's not >13.. I searched around but couldn't find >13 debs.. or do I need 
to compile it from source?

Thanks,
nirmal

Re: pb with apt-get and libc6

[Sven Luther]

On Wed, Nov 20, 2002 at 02:06:14PM +0100, jan Govaere wrote:
> I've got a problem : yesterday I change my woody -> sid(unstable)
> It seems to be a problem with the libc6 ...
> 
> #dpkg --configure libc6
> Package 'ssh-nonfree' isq not installed and no info is available
> sed: can't read exim: No such file or directory
> sed: can't read ssh: No such file or directory
> sed: can't read netbase: No such file or directory
> sed: can't read cron: No such file or directory
> sed: can't read lpr: No such file or directory
> dpkg: error processing libc6 (--configure)
> subprocess post-installation script returned error exit status 2
> Errors were encountered while processing libc6

I suppose you now have 2.3.1-4 installed, or maybe 2.3.1-5 ?

Anyway, there are already 3 bugs about this in the BTS.

Mmm, please try 2.3.1-5, which seem to have fixed some of them.

Friendly,

Sven Luther

pb with apt-get and libc6

[jan Govaere]

I've got a problem : yesterday I change my woody -> sid(unstable)
It seems to be a problem with the libc6 ...

#dpkg --configure libc6
Package 'ssh-nonfree' isq not installed and no info is available
sed: can't read exim: No such file or directory
sed: can't read ssh: No such file or directory
sed: can't read netbase: No such file or directory
sed: can't read cron: No such file or directory
sed: can't read lpr: No such file or directory
dpkg: error processing libc6 (--configure)
subprocess post-installation script returned error exit status 2
Errors were encountered while processing libc6

Any idea ? I don't know what to do ...

Thanks ...

--
JAn

Re: what is libc6 2.2.4-6.0.1 in ppc?

[Daniel Jacobowitz]

On Wed, Dec 05, 2001 at 11:02:42AM +0900, Atsuhito Kohda wrote:
> Recently when I was upgrading sid system (ppc), dpkg complained
> that locales 2.2.4-6 depended on glibc 2.2.4-6 but there was
> no such file.  In fact there is only libc6 2.2.4-6.0.1 now.
> 
> What is this libc6 2.2.4-6.0.1 and why there is no associating
> locales?
> 
> If this is already known problem, I'm very sorry.

I uploaded it?

Ack!  I uploaded it!  I'm sorry!

I'll try to get this corrected soon.

-- 
Daniel Jacobowitz   Carnegie Mellon University
MontaVista Software Debian GNU/Linux Developer

Re: what is libc6 2.2.4-6.0.1 in ppc?

[Atsuhito Kohda]

> What is this libc6 2.2.4-6.0.1 and why there is no associating
> locales?

I was taught that this was because of voltaire corruption.

Sorry for my noise.

Best regards,   2001.12.5

-- 
 Debian Developer & Debian JP Developer - much more I18N of Debian
 Atsuhito Kohda <[EMAIL PROTECTED]>
 Department of Math., Tokushima Univ.

what is libc6 2.2.4-6.0.1 in ppc?

[Atsuhito Kohda]

Recently when I was upgrading sid system (ppc), dpkg complained
that locales 2.2.4-6 depended on glibc 2.2.4-6 but there was
no such file.  In fact there is only libc6 2.2.4-6.0.1 now.

What is this libc6 2.2.4-6.0.1 and why there is no associating
locales?

If this is already known problem, I'm very sorry.

Best regards,2001.12.5

-- 
 Debian Developer & Debian JP Developer - much more I18N of Debian
 Atsuhito Kohda <[EMAIL PROTECTED]>
 Department of Math., Tokushima Univ.

Re: Upgrading libc6 and binutils from Potato to Woody

[Michael Schmitz]

> How could I get libc6 and binutils from Woody? I don't see them
> while looking at the binary-powerpc directories on the FTP site.

Look in 'pool'.

Michael

Upgrading libc6 and binutils from Potato to Woody

[Jose A. Accino]

How could I get libc6 and binutils from Woody? I don't see them
while looking at the binary-powerpc directories on the FTP site.
Also, a search in packages.debian.org always offers returns the
i386 package to download, but no powerpc version...
What am I missing?

TIA,
JA.

libc6 2.1.3-19 causes segfaults?

[Jens Kutilek]

Has anyone upgraded successfully to libc 2.1.3-19 on potato (proposed-updates)?
Shortly after I upgraded, kswapd segfaulted and things went kinda downhill from 
then
on ... even shutdown, sync and init crashed (that reads: inevitable fs damage).
I managed to downgrade libc6 to 2.1.3-18, had a kind person at my provider press
the reset button, repaired some fs and now it seems to run fine again. Don't 
want to
try again, though.
Here's what kern.log said on the first occurrence:

Sep 11 12:27:00 www kernel: Oops: kernel access of bad area, sig: 11
Sep 11 12:27:00 www kernel: NIP: C0033D7C XER:  LR: C0033CBC SP: 
C04D7F90 REGS: c04d7ee0 TRAP:0300
Sep 11 12:27:00 www kernel: MSR: 9032 EE: 1 PR: 0 FP: 0 ME: 1 IR/DR: 11
Sep 11 12:27:00 www kernel: DAR: , DSISR: 4200
Sep 11 12:27:00 www kernel: TASK = c04d6000[4] 'kswapd' Last syscall: -1
Sep 11 12:27:00 www kernel: last math cc41 last altivec 
Sep 11 12:27:00 www kernel: GPR00: B4D5 C04D7F90 C04D6000 0001 0006 
 DFED84E0 
Sep 11 12:27:00 www kernel: GPR08: C01427A0  C0A81F54 C01A4334 DFAC4780 
10075648  
Sep 11 12:27:00 www kernel: GPR16:     003FF000 
 C01A C01A
Sep 11 12:27:00 www kernel: GPR24:  C01A4334   080E 
C0A7FFE0 C0A7FFFC C0A7FFF8
Sep 11 12:27:00 www kernel: Call backtrace:
Sep 11 12:27:00 www kernel: C0014268 C00342D4 C000640C

The machine is an IBM RS/6000 B50.

bye,
Jens

-- 
Jens Kutilek
Web-Design

ISITRAIN Schulung und Systemlösungen GmbH
Telefon: (05303) 941014, Telefax: (05303) 941016
E-Mail: {HYPERLINK "mailto:[EMAIL PROTECTED]"[EMAIL PROTECTED], Internet: 
{HYPERLINK "http://www.isitrain.de"}www.isitrain.de

Re: libc6 is broken (still)

[Michael Schmitz]

> >> From what I understood, the current ld.so breakage is not worth a new
> >>upload in Ben Collins' opinion. And this is driving me crazy. This bug
> >>has been present for over a month now.

Which ld.so breakage (which bug number, ...)?

Michael

Re: libc6 is broken (still)

[Bastien Nocera]

Daniel Jacobowitz wrote:


On Tue, Jul 10, 2001 at 12:55:06AM +0100, Bastien Nocera wrote:


Hi,

From what I understood, the current ld.so breakage is not worth a new 
upload in Ben Collins' opinion. And this is driving me crazy. This bug 
has been present for over a month now.


Would somebody care making a new version of the libc6 with the patch here:
http://sources.redhat.com/ml/libc-alpha/2001-06/msg00238.html

I have tried to do the compilation myself, but it fails some tests, and 
craps out (I'm probably using the wrong compiler version or something). 
I attached the patch I added to debian/patches (I also modified the 
0list file in the same directory).


I hope someone can help, and I apologise in advance if mozilla is to 
send this mail in HTML




I'll try to talk Ben into a new upload.  In fact, he seems to have
already been talked into it.



He might have seen me yelling and kicking yesterday evening on 
#debian-devel. glibc is definitely not something you want broken...


Well, I reckon you don't need the .dpatch file that I forgot to attach 
to my previous email then ;)


BTW, could you explain to us what this problem is exactly about, and 
what needs to be done to other software (like Michael asked about X) ?


Thanks a lot for taking care of that.

---

/Bastien Nocera

http://hadess.net

Re: libc6 is broken (still)

[Daniel Jacobowitz]

On Tue, Jul 10, 2001 at 09:35:31AM +0200, Michel Dänzer wrote:
> Does this mean potentially broken software like XFree86 doesn't have to care
> about this? We've had quite some discussion about a solution...

No!  It needs to be fixed.  The glibc change is not only a workaround
but a significant performance improvement as well.

-- 
Daniel Jacobowitz   Carnegie Mellon University
MontaVista Software Debian GNU/Linux Developer

Re: libc6 is broken (still)

[Michel Dänzer]

Daniel Jacobowitz wrote:
> 
> On Tue, Jul 10, 2001 at 12:55:06AM +0100, Bastien Nocera wrote:
> > Hi,
> >
> >  From what I understood, the current ld.so breakage is not worth a new
> > upload in Ben Collins' opinion. And this is driving me crazy. This bug
> > has been present for over a month now.
> >
> > Would somebody care making a new version of the libc6 with the patch here:
> > http://sources.redhat.com/ml/libc-alpha/2001-06/msg00238.html
> >
> > I have tried to do the compilation myself, but it fails some tests, and
> > craps out (I'm probably using the wrong compiler version or something).
> > I attached the patch I added to debian/patches (I also modified the
> > 0list file in the same directory).
> >
> > I hope someone can help, and I apologise in advance if mozilla is to
> > send this mail in HTML
> 
> I'll try to talk Ben into a new upload.  In fact, he seems to have
> already been talked into it.

Does this mean potentially broken software like XFree86 doesn't have to care
about this? We've had quite some discussion about a solution...


-- 
Earthling Michel Dänzer (MrCooper)\   Debian GNU/Linux (powerpc) developer
CS student, Free Software enthusiast   \XFree86 and DRI project member

Re: libc6 is broken (still)

[Daniel Jacobowitz]

On Tue, Jul 10, 2001 at 12:55:06AM +0100, Bastien Nocera wrote:
> Hi,
> 
>  From what I understood, the current ld.so breakage is not worth a new 
> upload in Ben Collins' opinion. And this is driving me crazy. This bug 
> has been present for over a month now.
> 
> Would somebody care making a new version of the libc6 with the patch here:
> http://sources.redhat.com/ml/libc-alpha/2001-06/msg00238.html
> 
> I have tried to do the compilation myself, but it fails some tests, and 
> craps out (I'm probably using the wrong compiler version or something). 
> I attached the patch I added to debian/patches (I also modified the 
> 0list file in the same directory).
> 
> I hope someone can help, and I apologise in advance if mozilla is to 
> send this mail in HTML

I'll try to talk Ben into a new upload.  In fact, he seems to have
already been talked into it.

-- 
Daniel Jacobowitz   Carnegie Mellon University
MontaVista Software Debian GNU/Linux Developer

libc6 is broken (still)

[Bastien Nocera]

Hi,

From what I understood, the current ld.so breakage is not worth a new 
upload in Ben Collins' opinion. And this is driving me crazy. This bug 
has been present for over a month now.


Would somebody care making a new version of the libc6 with the patch here:
http://sources.redhat.com/ml/libc-alpha/2001-06/msg00238.html

I have tried to do the compilation myself, but it fails some tests, and 
craps out (I'm probably using the wrong compiler version or something). 
I attached the patch I added to debian/patches (I also modified the 
0list file in the same directory).


I hope someone can help, and I apologise in advance if mozilla is to 
send this mail in HTML


---
/Bastien Nocera
http://hadess.net

Re: broken libc6, looking for a working dpkg

[Tuomas Kuosmanen]

On Fri, Jun 15, 2001 at 02:18:51PM +0100, thus said Bastien Nocera:

> > or is it something different? I am wondering why this broke, is it a
> > broken glibc package or my own fault, and how to fix it the right way?
> 
> Much like it. I am _not_ restarting evolution until this is fixed ;) so
> here is what I get when I launch xmms, some plugins fail to load:
> 
> /usr/local/lib/libSDL-1.2.so.0: R_PPC_REL24 relocation at 0x6f6596d4 for
> symbol `' out of range
> /home/hadess/.xmms/Plugins/libkjofol.so: R_PPC_REL24 relocation at
> 0x6f46aeb8 for symbol `' out of range

% xmms 
usr/lib/libSDL-1.2.so.0: R_PPC_REL24 relocation at 0x6ec51f38 for
/symbol `' out of range

So same thing. The weird thing is most apps seem to work, for example
Nautilus is working fine, no problem on it. Was there something changing in
the glibc that broke things for us between these versions? It seems to be
related to linking or library loading, but I dont have much clue about those
things.. :(

Tuomas

-- 

.--->  [EMAIL PROTECTED].|\,/|  [EMAIL PROTECTED]  <-.
+>  www.ximian.com -  ()-@@  ,   tigert.gimp.org  <+
`->  art director   ,  `--')/   a gimp artist  <---'

Re: broken libc6, looking for a working dpkg

[Bastien Nocera]

On 15 Jun 2001 15:38:59 +0300, Tuomas Kuosmanen wrote:
> On Fri, Jun 15, 2001 at 12:19:46AM +0100, thus said Bastien Nocera:
> > Hi,
> > 
> > After downgrading my libc6 from 2.2.3-1 to 2.2.2-1, because the upgrade
> > broke most of xmms' plugins and evolution (that even a recompile
> > couldn't fix), I'm stuck with a version of dpkg that requires libc6
> > 2.2.3.
> > 
> > Does anybody have a binary of dpkg that I could use on a glibc 2.2.2
> > based system ?

I managed to get it back to a working state, thanks Ethan.

> > Did anybody else have these plugins problems ? I reckon it could be a
> > problem with glib's dlopen() code (not glibc, glib), but didn't have the
> > time to investigate.
> 
> Ah! At least someone in the same boat I guess.. My Evolution broke as well.
> I noticed gdialog segfaulted after the glibc upgrade, but recompiling the
> gnome-utils package of which gdialog is part of, fixed it. Did you get
> something like this with Evo?
> 
>   http://tigert.gimp.org/things/aieee.png
> 
> or is it something different? I am wondering why this broke, is it a
> broken glibc package or my own fault, and how to fix it the right way?

Much like it. I am _not_ restarting evolution until this is fixed ;) so
here is what I get when I launch xmms, some plugins fail to load:

/usr/local/lib/libSDL-1.2.so.0: R_PPC_REL24 relocation at 0x6f6596d4 for
symbol `' out of range
/home/hadess/.xmms/Plugins/libkjofol.so: R_PPC_REL24 relocation at
0x6f46aeb8 for symbol `' out of range

I tried recompiling glib with the new glibc, and not more success. Does
anybody have an idea about this ?

Cheers

-- 
/Bastien Nocera
http://hadess.net

Re: broken libc6, looking for a working dpkg

[Tuomas Kuosmanen]

On Fri, Jun 15, 2001 at 12:19:46AM +0100, thus said Bastien Nocera:
> Hi,
> 
> After downgrading my libc6 from 2.2.3-1 to 2.2.2-1, because the upgrade
> broke most of xmms' plugins and evolution (that even a recompile
> couldn't fix), I'm stuck with a version of dpkg that requires libc6
> 2.2.3.
> 
> Does anybody have a binary of dpkg that I could use on a glibc 2.2.2
> based system ?
> 
> Did anybody else have these plugins problems ? I reckon it could be a
> problem with glib's dlopen() code (not glibc, glib), but didn't have the
> time to investigate.

Ah! At least someone in the same boat I guess.. My Evolution broke as well.
I noticed gdialog segfaulted after the glibc upgrade, but recompiling the
gnome-utils package of which gdialog is part of, fixed it. Did you get
something like this with Evo?

http://tigert.gimp.org/things/aieee.png

or is it something different? I am wondering why this broke, is it a
broken glibc package or my own fault, and how to fix it the right way?

Tuomas

-- 

.--->  [EMAIL PROTECTED].|\,/|  [EMAIL PROTECTED]  <-.
+>  www.ximian.com -  ()-@@  ,   tigert.gimp.org  <+
`->  art director   ,  `--')/   a gimp artist  <---'

broken libc6, looking for a working dpkg

[Bastien Nocera]

Hi,

After downgrading my libc6 from 2.2.3-1 to 2.2.2-1, because the upgrade
broke most of xmms' plugins and evolution (that even a recompile
couldn't fix), I'm stuck with a version of dpkg that requires libc6
2.2.3.

Does anybody have a binary of dpkg that I could use on a glibc 2.2.2
based system ?

Did anybody else have these plugins problems ? I reckon it could be a
problem with glib's dlopen() code (not glibc, glib), but didn't have the
time to investigate.

Thanks

-- 
/Bastien Nocera
http://hadess.net

Re: Upgrading libc6 problem

[Andrew Sharp]

Michel Dänzer wrote:
> 
> Vinod Kurup wrote:
> 
> > I just did an apt-get upgrade on my unstable powerpc machine and it
> > upgraded libc6.
> >
> > shortly thereafter, I started getting this message in my syslog:
> >
> > May 28 22:10:05 localhost inetd[2673]: getpwnam: mail: No such user
> >
> > And my mail stopped being delivered. Thanks to the magic of google, I found
> > this message from Colin Watson:
> >
> > <http://lists.debian.org/debian-devel-0009/msg01990.html>
> >
> > Quoting:
> >
> > > Going down to single-user mode and then back up to runlevel 2 seems to
> > > have fixed most such problems. Sending a SIGHUP to inetd, as libc6's
> > > postinst does via '/etc/init.d/inetd restart', is not enough - it needs
> > > to be stopped and started again.
> >
> > So i went to runlevel 1 and then back to 2 and everything was back to
> > normal. Just thought I'd mention it in case it happens to anyone
> > else.
> 
> /etc/init.d/inetd restart
> 
> wouldn't do?

At the very least /etc/init.d/inetd stop ; /etc/init.d/inetd start
should do ~:^)

a

Re: Upgrading libc6 problem

[Michel Dänzer]

Vinod Kurup wrote:
> 
> On Tue, May 29, 2001 at 05:22:00PM +0200, Michel D?nzer wrote:
> > Vinod Kurup wrote:
> >
> > > I just did an apt-get upgrade on my unstable powerpc machine and it
> > > upgraded libc6.
> > >
> > > shortly thereafter, I started getting this message in my syslog:
> > >
> > > May 28 22:10:05 localhost inetd[2673]: getpwnam: mail: No such user
> > >
> > > And my mail stopped being delivered. Thanks to the magic of google, I
> > > found this message from Colin Watson:
> > >
> > > <http://lists.debian.org/debian-devel-0009/msg01990.html>
> > >
> > > Quoting:
> > >
> > > > Going down to single-user mode and then back up to runlevel 2 seems to
> > > > have fixed most such problems. Sending a SIGHUP to inetd, as libc6's
> > > > postinst does via '/etc/init.d/inetd restart', is not enough - it
> > > > needs to be stopped and started again.
> > >
> > > So i went to runlevel 1 and then back to 2 and everything was back to
> > > normal. Just thought I'd mention it in case it happens to anyone
> > > else.
> >
> > /etc/init.d/inetd restart
> > 
> > wouldn't do?
> 
> Actually, you're right - it should do. I should have tried that first, but
> I just went for the big hammer.
> 
> The problem (I think) is that the libc6 postinst script calls
> /etc/init.d/inetd restart before some other important piece is upgraded. At
> least, that's what I gather from reading the rest of the thread I
> referenced above.

In fact, the post the link points to is a bit contradictory. It says
/etc/init.d/inetd restart just sends SIGHUP to inetd (which of course can be
not enough), but at least on this box it stops and starts it. Maybe
libc6.postinst really uses /etc/init.d/inetd reload or force-reload? That
would be a bug in either libc6 or inetd then.


-- 
Earthling Michel Dänzer (MrCooper)\   Debian GNU/Linux (powerpc) developer
CS student, Free Software enthusiast   \XFree86 and DRI project member

Re: Upgrading libc6 problem

[Vinod Kurup]

On Tue, May 29, 2001 at 05:22:00PM +0200, Michel D?nzer wrote:
> Vinod Kurup wrote:
> 
> > I just did an apt-get upgrade on my unstable powerpc machine and it
> > upgraded libc6.
> > 
> > shortly thereafter, I started getting this message in my syslog:
> > 
> > May 28 22:10:05 localhost inetd[2673]: getpwnam: mail: No such user
> > 
> > And my mail stopped being delivered. Thanks to the magic of google, I found
> > this message from Colin Watson:
> > 
> > <http://lists.debian.org/debian-devel-0009/msg01990.html>
> > 
> > Quoting:
> > 
> > > Going down to single-user mode and then back up to runlevel 2 seems to
> > > have fixed most such problems. Sending a SIGHUP to inetd, as libc6's
> > > postinst does via '/etc/init.d/inetd restart', is not enough - it needs
> > > to be stopped and started again.
> > 
> > So i went to runlevel 1 and then back to 2 and everything was back to
> > normal. Just thought I'd mention it in case it happens to anyone
> > else.
> 
> /etc/init.d/inetd restart
> 
> wouldn't do?

Actually, you're right - it should do. I should have tried that first, but
I just went for the big hammer.

The problem (I think) is that the libc6 postinst script calls
/etc/init.d/inetd restart before some other important piece is upgraded. At
least, that's what I gather from reading the rest of the thread I
referenced above.

Vinod

-- 
_
Vinod Kurup, MD
email: [EMAIL PROTECTED]
phone: 617.277.2012
cell:  617.359.5990
http://www.kurup.com
aim: vvkurup

Re: Upgrading libc6 problem

[Michel Dänzer]

Vinod Kurup wrote:

> I just did an apt-get upgrade on my unstable powerpc machine and it
> upgraded libc6.
> 
> shortly thereafter, I started getting this message in my syslog:
> 
> May 28 22:10:05 localhost inetd[2673]: getpwnam: mail: No such user
> 
> And my mail stopped being delivered. Thanks to the magic of google, I found
> this message from Colin Watson:
> 
> <http://lists.debian.org/debian-devel-0009/msg01990.html>
> 
> Quoting:
> 
> > Going down to single-user mode and then back up to runlevel 2 seems to
> > have fixed most such problems. Sending a SIGHUP to inetd, as libc6's
> > postinst does via '/etc/init.d/inetd restart', is not enough - it needs
> > to be stopped and started again.
> 
> So i went to runlevel 1 and then back to 2 and everything was back to
> normal. Just thought I'd mention it in case it happens to anyone
> else.

/etc/init.d/inetd restart

wouldn't do?


-- 
Michel Dänzer   |  Relog-Reagency AG  |http://relog.ch 
Weinbergstrasse 108 | 8006 Zürich | +41 (0)1 368 55 55

Upgrading libc6 problem

[Vinod Kurup]

Hi all,

I just did an apt-get upgrade on my unstable powerpc machine and it
upgraded libc6.

shortly thereafter, I started getting this message in my syslog:

May 28 22:10:05 localhost inetd[2673]: getpwnam: mail: No such user

And my mail stopped being delivered. Thanks to the magic of google, I found
this message from Colin Watson:

<http://lists.debian.org/debian-devel-0009/msg01990.html>

Quoting:

> Going down to single-user mode and then back up to runlevel 2 seems to
> have fixed most such problems. Sending a SIGHUP to inetd, as libc6's
> postinst does via '/etc/init.d/inetd restart', is not enough - it needs
> to be stopped and started again.

So i went to runlevel 1 and then back to 2 and everything was back to
normal. Just thought I'd mention it in case it happens to anyone
else.

Vinod

-- 
_
Vinod Kurup, MD
email: [EMAIL PROTECTED]
phone: 617.277.2012
cell:  617.359.5990
http://www.kurup.com
aim: vvkurup

Re: Is libc6 OK on powerpc?

[Drew Parsons]

On Thu, Mar 22, 2001 at 08:57:56AM -0800, David J. Roundy wrote:
> On Wed, Mar 21, 2001 at 02:21:09PM -0500, Daniel Jacobowitz wrote:
> > 
> > Somewhere, you left out one of the va_arg calls.  It'll be using a char
> > or short argument.  This is illegal C; only promoted types (int,
> > pointers, etc) may be given to va_arg.
> 
> This was indeed the problem.  The bug was with a call of va_arg(ap,boolean)
> which should have been an int.
> 
> With the patch below it compiles and runs just fine.
> 

Thanks!  I'll put the patch in soon :)

Thanks for the pointers, all!

Drew

-- 
PGP public key available at http://dparsons.webjump.com/drewskey.txt
Fingerprint: A110 EAE1 D7D2 8076 5FE0  EC0A B6CE 7041 6412 4E4A

Re: Is libc6 OK on powerpc?

[Daniel Jacobowitz]

On Thu, Mar 22, 2001 at 08:57:56AM -0800, David J. Roundy wrote:
> "__va_arg_type_violation" is a function which doesn't exist, which gcc
> intentionally makes a call to if you make a bad va_arg call.  See
>  (which is where I read about it).
> 
> I'm not sure why it does this rather than giving a compile time error,
> which would be much more user-friendly, since it would give the line of
> code that is causing the problem.  Also, it doesn't do it when you turn the
> optimization off, which I also don't understand.

It can't give a compile-time error; the reason is briefly alluded to in
va-ppc.h, IIRC.  GCC 3.0 -will- be able to, on the other hand.  I
think.

-- 
Daniel Jacobowitz   Debian GNU/Linux Developer
Monta Vista Software  Debian Security Team
 "I am croutons!"

Re: Is libc6 OK on powerpc?

[David J. Roundy]

On Wed, Mar 21, 2001 at 02:21:09PM -0500, Daniel Jacobowitz wrote:
>
> On Thu, Mar 22, 2001 at 12:51:02AM +1100, Drew Parsons wrote:
> > gcc  main.o init.o events.o tools.o screens.o game.o editor.o files.o
> > cartoons.o libgame/libgame.a -lSDL_image -lSDL_mixer -L/usr/lib -lSDL
> > -lpthread  -lm -o ../mirrormagic
> > libgame/libgame.a(gadgets.o): In function HandleGadgetTags':
> > gadgets.o(.text+0x9f8): undefined reference to __va_arg_type_violation'
> > gadgets.o(.text+0x9f8): relocation truncated to fit: R_PPC_REL24
> > __va_arg_type_violation
> > collect2: ld returned 1 exit status
> > make[2]: *** [../mirrormagic] Error 1
> 
> 
> > The code in question is declared as:
> > 
> > static void HandleGadgetTags(struct GadgetInfo *gi, int first_tag, va_list 
> > ap)
> > 
> > and the va_arg argument is used in several places inside, e.g.
> > gi->custom_id = va_arg(ap, int); 
> > strncpy(gi->info_text, va_arg(ap, char *), max_textsize); 
> > gi->type = va_arg(ap, unsigned long); 
> > gi->deco.width = va_arg(ap, int); 
> > gi->deco.height = va_arg(ap, int); 
> > gi->design[GD_BUTTON_PRESSED].bitmap = va_arg(ap, Bitmap *); 
> > gi->event_mask = va_arg(ap, unsigned long); 
> > gi->callback_info = va_arg(ap, gadget_function); 
> 
> Somewhere, you left out one of the va_arg calls.  It'll be using a char
> or short argument.  This is illegal C; only promoted types (int,
> pointers, etc) may be given to va_arg.

This was indeed the problem.  The bug was with a call of va_arg(ap,boolean)
which should have been an int.

With the patch below it compiles and runs just fine.

> > So this "undefined reference to __va_arg_type_violation", is it a known
> > problem on the powerpc?  Have others met it?  Or does it mean the code has
> > been written poorly?  There's nothing obviously bad in the code that I can
> > see, but I've never had to deal with powerpc or endian issues before (though
> > I can't see any issues here). It works fine on x86.
> 
> It works on x86 because x86 is a mostly unaligned architecture, that's
> all :)

"__va_arg_type_violation" is a function which doesn't exist, which gcc
intentionally makes a call to if you make a bad va_arg call.  See
 (which is where I read about it).

I'm not sure why it does this rather than giving a compile time error,
which would be much more user-friendly, since it would give the line of
code that is causing the problem.  Also, it doesn't do it when you turn the
optimization off, which I also don't understand.

But at least I now understand what this error is all about.
-- 
David Roundy
http://civet.berkeley.edu/droundy/


--- mirrormagic-2.0.0/src/libgame/gadgets.c Mon Dec  4 14:44:04 2000
+++ mirrormagic-2.0.0.new/src/libgame/gadgets.c Thu Mar 22 08:44:23 2001
@@ -309,7 +309,7 @@
break;

   case GDI_CHECKED:
-   gi->checked = va_arg(ap, boolean);
+   gi->checked = (boolean) va_arg(ap, int);
break;

   case GDI_RADIO_NR:

Re: Is libc6 OK on powerpc?

[Christophe Merlet \(RedFox\)]

Daniel Jacobowitz wrote:
> 
> On Wed, Mar 21, 2001 at 06:52:06PM +0100, Christophe Merlet (RedFox) wrote:
> > See the file ./gcc/ginclude/va-ppc.h in the source of gcc for
> > explanation about this undefined reference.
> >
> > Compile your code, or just the file gadgets.c, without optimisation.
> 
> No!  Gods, no!  Don't do that!

Oupsss, sorry

> It's an -error-, not a warning.  With optimization off perhaps we fail
> to detect it.  In GCC 3 it will most likely be an error on all
> architectures.
> 
> See my other message for (the FAQ) of the problem.

Thanks for the explanation. The next time I see this -error- in a
software, I'll send your mail to the mainteners.

--
Christophe merlet (RedFox)

Re: Is libc6 OK on powerpc?

[Michael Schmitz]

> > and the va_arg argument is used in several places inside, e.g.
> > gi->custom_id = va_arg(ap, int);
> > strncpy(gi->info_text, va_arg(ap, char *), max_textsize);
> > gi->type = va_arg(ap, unsigned long);
> > gi->deco.width = va_arg(ap, int);
> > gi->deco.height = va_arg(ap, int);
> > gi->design[GD_BUTTON_PRESSED].bitmap = va_arg(ap, Bitmap *);
> > gi->event_mask = va_arg(ap, unsigned long);
> > gi->callback_info = va_arg(ap, gadget_function);
>
> Somewhere, you left out one of the va_arg calls.  It'll be using a char
> or short argument.  This is illegal C; only promoted types (int,
> pointers, etc) may be given to va_arg.

The Bitmap* or gadget_function argument should be cast to void* I think.

> > So this "undefined reference to __va_arg_type_violation", is it a known
> > problem on the powerpc?  Have others met it?  Or does it mean the code has
> > been written poorly?  There's nothing obviously bad in the code that I can

You must be using illegal types as arguments to va_arg (char, short, float,
perhaps the pointers), that's what's bad.

> > see, but I've never had to deal with powerpc or endian issues before (though
> > I can't see any issues here). It works fine on x86.
>
> It works on x86 because x86 is a mostly unaligned architecture, that's
> all :)

Prepare for more breakage (run time, horrible stack corruption) after it
compiles. Passing pointers to va_lists, using va_list_save = va_list
instead of va_copy, all this makes for a fun debugging project.

Lots of ix86 code using varargs isn't written portably, or even to the
spec.

Michael

Re: Is libc6 OK on powerpc?

[Daniel Jacobowitz]

On Thu, Mar 22, 2001 at 12:51:02AM +1100, Drew Parsons wrote:
> gcc  main.o init.o events.o tools.o screens.o game.o editor.o files.o
> cartoons.o libgame/libgame.a -lSDL_image -lSDL_mixer -L/usr/lib -lSDL
> -lpthread  -lm -o ../mirrormagic
> libgame/libgame.a(gadgets.o): In function HandleGadgetTags':
> gadgets.o(.text+0x9f8): undefined reference to __va_arg_type_violation'
> gadgets.o(.text+0x9f8): relocation truncated to fit: R_PPC_REL24
> __va_arg_type_violation
> collect2: ld returned 1 exit status
> make[2]: *** [../mirrormagic] Error 1


> The code in question is declared as:
> 
> static void HandleGadgetTags(struct GadgetInfo *gi, int first_tag, va_list ap)
> 
> and the va_arg argument is used in several places inside, e.g.
> gi->custom_id = va_arg(ap, int); 
> strncpy(gi->info_text, va_arg(ap, char *), max_textsize); 
> gi->type = va_arg(ap, unsigned long); 
> gi->deco.width = va_arg(ap, int); 
> gi->deco.height = va_arg(ap, int); 
> gi->design[GD_BUTTON_PRESSED].bitmap = va_arg(ap, Bitmap *); 
> gi->event_mask = va_arg(ap, unsigned long); 
> gi->callback_info = va_arg(ap, gadget_function); 

Somewhere, you left out one of the va_arg calls.  It'll be using a char
or short argument.  This is illegal C; only promoted types (int,
pointers, etc) may be given to va_arg.


> So this "undefined reference to __va_arg_type_violation", is it a known
> problem on the powerpc?  Have others met it?  Or does it mean the code has
> been written poorly?  There's nothing obviously bad in the code that I can
> see, but I've never had to deal with powerpc or endian issues before (though
> I can't see any issues here). It works fine on x86.

It works on x86 because x86 is a mostly unaligned architecture, that's
all :)

-- 
Daniel Jacobowitz   Debian GNU/Linux Developer
Monta Vista Software  Debian Security Team
 "I am croutons!"

Re: Is libc6 OK on powerpc?

[Daniel Jacobowitz]

On Wed, Mar 21, 2001 at 06:52:06PM +0100, Christophe Merlet (RedFox) wrote:
> See the file ./gcc/ginclude/va-ppc.h in the source of gcc for
> explanation about this undefined reference.
> 
> Compile your code, or just the file gadgets.c, without optimisation.

No!  Gods, no!  Don't do that!

It's an -error-, not a warning.  With optimization off perhaps we fail
to detect it.  In GCC 3 it will most likely be an error on all
architectures.

See my other message for (the FAQ) of the problem.

-- 
Daniel Jacobowitz   Debian GNU/Linux Developer
Monta Vista Software  Debian Security Team
 "I am croutons!"

Re: Is libc6 OK on powerpc?

[Christophe Merlet \(RedFox\)]

Drew Parsons wrote:
> 
> Hello there,
> 
> I'm the maintainer for mirrormagic.  I've got an x86 system, hope you can
> forgive my little inadequacies.
> 
> update_excuses.html says mirrormagic has issues getting into testing because
> of the following bug on powerpc (well, I gather this excuse has been
> overridden, and mirrormagic 2.0.0 isn't in testing because libsdl1.1 isn't,
> but anyway...)
> 
> According to http://voltaire.debian.org/buildd/buildlogs/mirrormagic/latest,
> we have this compile-time error:
> 
> gcc  main.o init.o events.o tools.o screens.o game.o editor.o files.o
> cartoons.o libgame/libgame.a -lSDL_image -lSDL_mixer -L/usr/lib -lSDL
> -lpthread  -lm -o ../mirrormagic
> libgame/libgame.a(gadgets.o): In function HandleGadgetTags':
> gadgets.o(.text+0x9f8): undefined reference to __va_arg_type_violation'
> gadgets.o(.text+0x9f8): relocation truncated to fit: R_PPC_REL24
> __va_arg_type_violation
> collect2: ld returned 1 exit status
> make[2]: *** [../mirrormagic] Error 1

See the file ./gcc/ginclude/va-ppc.h in the source of gcc for
explanation about this undefined reference.

Compile your code, or just the file gadgets.c, without optimisation.

--
Christophe Merlet (RedFox)

Re: Is libc6 OK on powerpc?

[David J. Roundy]

On Thu, Mar 22, 2001 at 12:51:02AM +1100, Drew Parsons wrote:
> 
> gcc  main.o init.o events.o tools.o screens.o game.o editor.o files.o
> cartoons.o libgame/libgame.a -lSDL_image -lSDL_mixer -L/usr/lib -lSDL
> -lpthread  -lm -o ../mirrormagic
> libgame/libgame.a(gadgets.o): In function HandleGadgetTags':
> gadgets.o(.text+0x9f8): undefined reference to __va_arg_type_violation'
> gadgets.o(.text+0x9f8): relocation truncated to fit: R_PPC_REL24
> __va_arg_type_violation
> collect2: ld returned 1 exit status
> make[2]: *** [../mirrormagic] Error 1

I've seen similar errors in the past while compiling KDE, but in that case
the libraries were dynamically linked, so it didn't show up until runtime.
Alas, I never did figure out what caused it.

I would be very interested in finding out what is causing this error.  I
just reproduced the same error, but am at a loss as to how to fix it.
Sorry I can't be any more helpful.
-- 
David Roundy
http://civet.berkeley.edu/droundy/

Is libc6 OK on powerpc?

[Drew Parsons]

Hello there,

I'm the maintainer for mirrormagic.  I've got an x86 system, hope you can
forgive my little inadequacies.

update_excuses.html says mirrormagic has issues getting into testing because
of the following bug on powerpc (well, I gather this excuse has been
overridden, and mirrormagic 2.0.0 isn't in testing because libsdl1.1 isn't,
but anyway...)

According to http://voltaire.debian.org/buildd/buildlogs/mirrormagic/latest,
we have this compile-time error:

gcc  main.o init.o events.o tools.o screens.o game.o editor.o files.o
cartoons.o libgame/libgame.a -lSDL_image -lSDL_mixer -L/usr/lib -lSDL
-lpthread  -lm -o ../mirrormagic
libgame/libgame.a(gadgets.o): In function HandleGadgetTags':
gadgets.o(.text+0x9f8): undefined reference to __va_arg_type_violation'
gadgets.o(.text+0x9f8): relocation truncated to fit: R_PPC_REL24
__va_arg_type_violation
collect2: ld returned 1 exit status
make[2]: *** [../mirrormagic] Error 1



The code in question is declared as:

static void HandleGadgetTags(struct GadgetInfo *gi, int first_tag, va_list ap)

and the va_arg argument is used in several places inside, e.g.
gi->custom_id = va_arg(ap, int); 
strncpy(gi->info_text, va_arg(ap, char *), max_textsize); 
gi->type = va_arg(ap, unsigned long); 
gi->deco.width = va_arg(ap, int); 
gi->deco.height = va_arg(ap, int); 
gi->design[GD_BUTTON_PRESSED].bitmap = va_arg(ap, Bitmap *); 
gi->event_mask = va_arg(ap, unsigned long); 
gi->callback_info = va_arg(ap, gadget_function); 
etc.

gi (GadgetInfo) looks like:

struct GadgetInfo
{
  int id;   /* internal gadget identifier */
  int custom_id;/* custom gadget identifier */
  int custom_type_id;   /* custom gadget type identifier */
  char info_text[MAX_INFO_TEXTSIZE];/* short popup info text */
  int x, y; /* gadget position */
  int width, height;/* gadget size */
  unsigned long type;   /* type (button, text input, ...) */
  unsigned long state;  /* state (pressed, released, ...) */
  boolean checked;  /* check/radio button state */
  int radio_nr; /* number of radio button series */
  boolean mapped;   /* gadget is active */
  struct GadgetBorder border;   /* gadget border design */
  struct GadgetDesign design[2];/* 0: normal; 1: pressed */
  struct GadgetDesign alt_design[2];/* alternative design */
  struct GadgetDecoration deco; /* decoration on top of gadget */
  unsigned long event_mask; /* possible events for this gadget */
  struct GadgetEvent event; /* actual gadget event */
  gadget_function callback_info;/* function for pop-up info text */
  gadget_function callback_action;  /* function for gadget action */
  struct GadgetDrawingArea drawing; /* fields for drawing area gadget */
  struct GadgetTextInput text;  /* fields for text input gadget */
  struct GadgetScrollbar scrollbar; /* fields for scrollbar gadget */
  struct GadgetInfo *next;  /* next list entry */
};



So this "undefined reference to __va_arg_type_violation", is it a known
problem on the powerpc?  Have others met it?  Or does it mean the code has
been written poorly?  There's nothing obviously bad in the code that I can
see, but I've never had to deal with powerpc or endian issues before (though
I can't see any issues here). It works fine on x86.

Thanks for any clues,

Drew Parsons

-- 
PGP public key available at http://dparsons.webjump.com/drewskey.txt
Fingerprint: A110 EAE1 D7D2 8076 5FE0  EC0A B6CE 7041 6412 4E4A

Re: 2.2.18 source patch; libc6-dev internal conflict?

[Adam C Powell IV]

Adam C Powell IV wrote:

> Unfortunately, openafs doesn't support 2.2 on PPC (yet?), so I patched
> it and tried to build it for 2.4.0-test12, details including the patch
> can be found at bug #80835.  Patched as such, it failed on a curious
> error, which can be found in that bug, but I'll repeat it here for
> discussion, as it appears to indicate an internal conflict within
> libc6-dev:
>
> cc -Dppc_linux24 -O2 -I.
> -I/usr/src/modules/openafs/ppc_linux24/dest/include  -O2 -g-c -o
> partition.o partition.c

Turns out partition.c needs a patch.  I'll build and test as soon as I can, and
report back to the list.

The mac_hid kernel problem is still there in kernel-source-2.2.18.  I just filed
bug #80959.

Zeen,

-Adam P.

   Welcome to the best software in the world today cafe!

2.2.18 source patch; libc6-dev internal conflict?

[Adam C Powell IV]

Greetings,

More "new input layer" drama...

I got and unpacked kernel-source-2.2.18, copied /boot/config-2.2.18pre21
to .config, did make oldconfig (which changed nothing) and make-kpkg
kernel_image.  A pretty straightforward kernel build, which I'm doing to
try the openafs modules.

But it died in drivers/macintosh/mac_hid.c with:

mac_hid.c: In function `mac_hid_sysctl_keycodes':
mac_hid.c:296: structure has no member named `ppc_kbd_sysrq_xlate'
mac_hid.c:303: structure has no member named `ppc_kbd_sysrq_xlate'

The attached patch lets it build, but I haven't yet tried booting it.

Unfortunately, openafs doesn't support 2.2 on PPC (yet?), so I patched
it and tried to build it for 2.4.0-test12, details including the patch
can be found at bug #80835.  Patched as such, it failed on a curious
error, which can be found in that bug, but I'll repeat it here for
discussion, as it appears to indicate an internal conflict within
libc6-dev:

cc -Dppc_linux24 -O2 -I.
-I/usr/src/modules/openafs/ppc_linux24/dest/include  -O2 -g-c -o
partition.o partition.c
In file included from
/usr/src/modules/openafs/ppc_linux24/dest/include/rx/rx.h:58,
 from
/usr/src/modules/openafs/ppc_linux24/dest/include/afs/afsint.h:62,
 from partition.c:93:
/usr/include/netinet/in.h:256: warning: `ntohl' redefined
/usr/include/linux/byteorder/generic.h:169: warning: this is the
location of the previous definition
/usr/include/netinet/in.h:257: warning: `ntohs' redefined
/usr/include/linux/byteorder/generic.h:175: warning: this is the
location of the previous definition
/usr/include/netinet/in.h:258: warning: `htonl' redefined
/usr/include/linux/byteorder/generic.h:168: warning: this is the
location of the previous definition
/usr/include/netinet/in.h:259: warning: `htons' redefined
/usr/include/linux/byteorder/generic.h:174: warning: this is the
location of the previous definition
In file included from
/usr/src/modules/openafs/ppc_linux24/dest/include/rx/rx.h:58,
 from
/usr/src/modules/openafs/ppc_linux24/dest/include/afs/afsint.h:62,
 from partition.c:93:
/usr/include/netinet/in.h:236: `__u32' redeclared as different kind of
symbol
/usr/include/asm/types.h:18: previous declaration of `__u32'
/usr/include/netinet/in.h:238: `__u16' redeclared as different kind of
symbol
/usr/include/asm/types.h:15: previous declaration of `__u16'
make[3]: *** [partition.o] Error 1
make[3]: Leaving directory
`/usr/src/modules/openafs/ppc_linux24/obj/vol'

The first bunch I think I understand, but the __u32 and __u16 parts are
beyond me...

Thanks,

-Adam P.

  Welcome to the best software in the world today cafe!
--- drivers/macintosh/mac_hid.c.bak Thu Dec 28 23:21:37 2000
+++ drivers/macintosh/mac_hid.c Thu Dec 28 23:23:19 2000
@@ -293,14 +293,14 @@
&& keyboard_sends_linux_keycodes != val) {
if (!keyboard_sends_linux_keycodes) {
 #ifdef CONFIG_MAGIC_SYSRQ
-   ppc_md.ppc_kbd_sysrq_xlate   = mac_hid_kbd_sysrq_xlate;
+   ppc_md.sysrq_xlate   = mac_hid_kbd_sysrq_xlate;
SYSRQ_KEY= 0x69;
 #endif
memcpy(pc_key_maps_save, key_maps, sizeof(key_maps));
memcpy(key_maps, mac_key_maps_save, sizeof(key_maps));
} else {
 #ifdef CONFIG_MAGIC_SYSRQ
-   ppc_md.ppc_kbd_sysrq_xlate   = pckbd_sysrq_xlate;
+   ppc_md.sysrq_xlate   = pckbd_sysrq_xlate;
SYSRQ_KEY= 0x54;
 #endif
memcpy(mac_key_maps_save, key_maps, sizeof(key_maps));

Re: Major libc6 problems (with libc6 2.1.97-1 and friends)

[Ben Collins]

On Thu, Nov 23, 2000 at 01:41:30AM -0800, C.M. Connelly wrote:
> 
> "BC" => Ben Collins <[EMAIL PROTECTED]>
> 
> Me> [*] One extremely annoying discovery was that the libc6
> Me> packages *must be built as root*.  Leaving aside the
> Me> question of *why* the packages can't be built with
> Me> fakeroot (or as an ordinary user, even) -- the Red Hat
> Me> RPMs do not require you to be root -- it would have been
> Me> really nice if the package had checked for root *first*,
> Me> rather than running for 4.5 hours and only then realizing
> Me> it couldn't finish.
> 
> BC> Problem is, fakeroot preloads a library to "fake"
> BC> things. This is bad when working with building a system
> BC> lib such as glibc. It has caused many bugs before, which
> BC> is why I explicitly forced the glibc to make sure it was
> BC> using real root, and not fakeroot.
> 
> If fakeroot causes problems, then perhaps we should either fix it
> so that it causes fewer problems, or figure out a way to avoid
> using it.  Other packaging systems, such as RPM, allow people to
> build packages as ordinary users, and that includes glibc -- my
> partner has built RPMs of several versions of glibc over the past
> few days on a faster machine where root access wasn't an option.
> Until dpkg includes that functionality, however, I'll believe you
> when you say that fakeroot causes some problems and accept that
> the safest workaround is to build the packages as root.

That is because RPM uses neither root nor a type of fakeroot. This is part
of the package program (rpm). The rpm program uses an internal archiver
(cpio derivative) that sets the uid/gid itself. IOW, it does everything
for you. If you do this for Debian, then I can overcome this problem.
Fakeroot cannot be fixed in this case, because of the fact that it is based
on preloading a library, which always preloads a libc.so.6 that isn't the
one we are using. This library gets preloaded with the new libc.so.6 that
we are building. So the problem is unavoidable.

> BC> This wont change, and the "4 hours later" problem is a
> BC> result of the Debian build process, not something specific
> BC> to glibc.
> 
> Except that the test comes right at the end and is both defined in
> and performed by code in the debian/rules file:
> 
>define checkroot
>   @test root = "`whoami`" || (echo Need root privileges; exit 1)
>   @test -z "$$FAKEROOTKEY" || (echo Need to compile as root, not 
> fakeroot; exit 1)
>endef
> 
> This code is called (as ``$(checkroot)'') at the start of the
> ``install'' target, which is one of the last targets to run during
> the build process.  If you can run this test just before building
> packages, you can run it earlier in the process -- preferably
> right at the very beginning (in the ``configure'' target), before
> the builder has walked away, expecting to come back to a set of
> freshly built glibc packages.

That's because most people do "dpkg-buildpackage -rfakeroot", and it only
uses "fakeroot" at this point in the build, not for the entire build. This
must be where you are messing up. You need to run "dpkg-buildpackage
-rsudo" and not "sudo dpkg-buildpackage".

> You could even perform the test for *each* target, which would
> help if someone was running an intermediate target rather than
> starting the build process from scratch.

I only run it on the targets that the package policy defines as needing
root. Any other way is wrong (and all packages do this the right way, so
don't point fingers at glibc).

-- 
 ---===-=-==-=---==-=--
/  Ben Collins  --  ...on that fantastic voyage...  --  Debian GNU/Linux   \
`  [EMAIL PROTECTED]  --  [EMAIL PROTECTED]  --  [EMAIL PROTECTED]  '
 `---=--===-=-=-=-===-==---=--=---'

Re: Major libc6 problems (with libc6 2.1.97-1 and friends)

[David Brown]

On Thu, Nov 23, 2000 at 04:28:52AM -0900, Ethan Benson wrote:
> On Thu, Nov 23, 2000 at 01:41:30AM -0800, C.M. Connelly wrote:
> > 
> > If fakeroot causes problems, then perhaps we should either fix it
> > so that it causes fewer problems, or figure out a way to avoid
> > using it.  Other packaging systems, such as RPM, allow people to
> > build packages as ordinary users, and that includes glibc -- my
> > partner has built RPMs of several versions of glibc over the past
> > few days on a faster machine where root access wasn't an option.
> > Until dpkg includes that functionality, however, I'll believe you
> > when you say that fakeroot causes some problems and accept that
> > the safest workaround is to build the packages as root.
> 
> AFAIK the only part of the .deb build process that usually needs root
> privileges is where all the binaries are actually packed into the
> .deb, root is needed to do chown and whatnot.  

Or have dpkg-deb use something like tardy
 to set the ownership and
permissions correctly in the control and data files.  The package build
scripts already know who should own what, and this information could be set
another way.

Dave Brown


pgpBaii810Nto.pgp
Description: PGP signature

Re: Major libc6 problems (with libc6 2.1.97-1 and friends)

[Ethan Benson]

On Thu, Nov 23, 2000 at 01:41:30AM -0800, C.M. Connelly wrote:
> 
> If fakeroot causes problems, then perhaps we should either fix it
> so that it causes fewer problems, or figure out a way to avoid
> using it.  Other packaging systems, such as RPM, allow people to
> build packages as ordinary users, and that includes glibc -- my
> partner has built RPMs of several versions of glibc over the past
> few days on a faster machine where root access wasn't an option.
> Until dpkg includes that functionality, however, I'll believe you
> when you say that fakeroot causes some problems and accept that
> the safest workaround is to build the packages as root.

AFAIK the only part of the .deb build process that usually needs root
privileges is where all the binaries are actually packed into the
.deb, root is needed to do chown and whatnot.  

i would imagine if the deb build process were altered so fakeroot was
only called when its actually needed: the packing time, where all the
chown, chmod stuff is happening.  this should fix this type of
problem since the compilers and other build programs are not running
in a fakeroot environment anymore. 

just a thought.

-- 
Ethan Benson
http://www.alaska.net/~erbenson/


pgp2v3fvQ8mkm.pgp
Description: PGP signature

Re: Major libc6 problems (with libc6 2.1.97-1 and friends)

[C.M. Connelly]

"BC" => Ben Collins <[EMAIL PROTECTED]>

Me> [*] One extremely annoying discovery was that the libc6
Me> packages *must be built as root*.  Leaving aside the
Me> question of *why* the packages can't be built with
Me> fakeroot (or as an ordinary user, even) -- the Red Hat
Me> RPMs do not require you to be root -- it would have been
Me> really nice if the package had checked for root *first*,
Me> rather than running for 4.5 hours and only then realizing
Me> it couldn't finish.

BC> Problem is, fakeroot preloads a library to "fake"
BC> things. This is bad when working with building a system
BC> lib such as glibc. It has caused many bugs before, which
BC> is why I explicitly forced the glibc to make sure it was
BC> using real root, and not fakeroot.

If fakeroot causes problems, then perhaps we should either fix it
so that it causes fewer problems, or figure out a way to avoid
using it.  Other packaging systems, such as RPM, allow people to
build packages as ordinary users, and that includes glibc -- my
partner has built RPMs of several versions of glibc over the past
few days on a faster machine where root access wasn't an option.
Until dpkg includes that functionality, however, I'll believe you
when you say that fakeroot causes some problems and accept that
the safest workaround is to build the packages as root.


BC> This wont change, and the "4 hours later" problem is a
BC> result of the Debian build process, not something specific
BC> to glibc.

Except that the test comes right at the end and is both defined in
and performed by code in the debian/rules file:

   define checkroot
  @test root = "`whoami`" || (echo Need root privileges; exit 1)
  @test -z "$$FAKEROOTKEY" || (echo Need to compile as root, not 
fakeroot; exit 1)
   endef

This code is called (as ``$(checkroot)'') at the start of the
``install'' target, which is one of the last targets to run during
the build process.  If you can run this test just before building
packages, you can run it earlier in the process -- preferably
right at the very beginning (in the ``configure'' target), before
the builder has walked away, expecting to come back to a set of
freshly built glibc packages.

You could even perform the test for *each* target, which would
help if someone was running an intermediate target rather than
starting the build process from scratch.

Even better -- you could perform the test at the start of the
build, and if the user wasn't root, you could print a message to
the effect that the libraries could be built as an ordinary user,
but that the user would need root privileges to build the
packages.  The test at the ``install'' target could then print
some useful information about resuming the build, if that's
possible.  I opted to start over because I wasn't sure whether the
problem was related to building the packages or building the
libraries (and I wasn't planning to use my machine for anything
else), but if I had known that it was safe to just su to root and
type ``debian/rules binary'' (or whatever), I would have done so.

At the very least, you could have the build script print an
obnoxious message on the terminal that warns the user that they
need to be root.  Maybe something like

   **
   WARNING!  WARNING!  WARNING!  WARNING!  WARNING!  WARNING!
   **
     
   This package cannot be built with fakeroot!   
     
   Please do at least one of the following:  
     * Build as root 
     * Devise workarounds for fakeroot's 
   deficiencies   
     * Rewrite fakeroot so it works  
     * Rewrite dpkg so you don't need to be  
   or pretend to be root 
     
   ('su'ing is the easiest one, I think.)
     
   **
   WARNING!  WARNING!  WARNING!  WARNING!  WARNING!  WARNING!
   **

would do the trick

Thanks,

   CMC

+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+
 Behind the counter a boy with a shaven head stared vacantly into space, 
 a dozen spikes of microsoft protruding from the socket behind his ear.
+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+
   C.M. Connelly   [EMAIL PROTECTED]   SHC, DS
+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+ 

Re: Major libc6 problems (with libc6 2.1.97-1 and friends)

[Ben Collins]

> [*] One extremely annoying discovery was that the libc6 packages
> *must be built as root*.  Leaving aside the question of *why*
> the packages can't be built with fakeroot (or as an ordinary
> user, even) -- the Red Hat RPMs do not require you to be root -- 
> it would have been really nice if the package had checked
> for root *first*, rather than running for 4.5 hours and only
> then realizing it couldn't finish.

No, it does not need root to be *built*. Notice that debian packages
require root in order to run the binary target of the build process (the
part where it installs things and builds the .deb's). This is normal, and
every package requires it.

However, we have fakeroot, which allows us to not have a real root access
in order to build packages. Problem is, fakeroot preloads a library to
"fake" things. This is bad when working with building a system lib such as
glibc. It has caused many bugs before, which is why I explicitly forced
the glibc to make sure it was using real root, and not fakeroot.

This wont change, and the "4 hours later" problem is a result of the
Debian build process, not something specific to glibc.

-- 
 ---===-=-==-=---==-=--
/  Ben Collins  --  ...on that fantastic voyage...  --  Debian GNU/Linux   \
`  [EMAIL PROTECTED]  --  [EMAIL PROTECTED]  --  [EMAIL PROTECTED]  '
 `---=--===-=-=-=-===-==---=--=---'

Re: Major libc6 problems (with libc6 2.1.97-1 and friends)

[C.M. Connelly]

Back on Saturday, November 11, 2000, I wrote

CMC> I just had hours of ``fun'' after typing the fateful
CMC> command ``apt-get upgrade'' and having my glibc packages
CMC> upgraded from 2.1.3-13 to 2.1.97-1.

This weekend, we had another go, and managed to figure out what
the problem was (it wasn't the libc6 packages at all) and get
things working.

As you may recall, the main symptom was that some binaries --
notably /bin/login -- caused Illegal instruction errors (meaning
that it is impossible to log in to the machine after an upgrade).
More specifically, the error was

   Trace/BPT trap

With gdb, the error was slightly more verbose:

   Program received signal SIGTRAP, Trace/Breakpoint trap.
   0x7fffd188 in ??? ()
   (gdb) bt
   #0 0x7fffd188 in ??? ()

Weirdly, though, if the debugging versions of the libraries were
used (by setting the environment variable LD_LIBRARY_PATH to
/usr/lib/debug), the program would execute normally, indicating
that the problem was related to the main libraries.

Because there were no debugging symbols in the main libraries, we
decided to build a set of libraries that contained those
symbols.[*]

When I downloaded the source (with ``apt-get source libc6''), I
ended up with libc6 2.2-1, and after compiling and installing, the
2.2-1 libraries gave the same error with /bin/login.  To figure
out exactly where the problem was, we ended up building unstripped
versions of several additional packages (login, libpam) before
finally discovering that the problem occurred when loading
libdl.so.2.  When I checked the file, I found that /lib/libdl.so.2
was a symbolic link to the wrong file -- /lib/libdl.so rather than
/lib/libdl-2.2.so:

-rw-r--r--1 root root56974 Nov 20 12:18 /lib/libdl-2.2.so
-rw-r--r--1 root root10640 Mar 12  2000 /lib/libdl.so
lrwxrwxrwx1 root root8 Nov 20 21:29 /lib/libdl.so.2 -> 
libdl.so

When I replaced the link with one to libdl-2.2.so, the errors
stopped.  As long as a /lib/libdl.so file was present, however,
running ldconfig would recreate the link to the old, incorrect
file!

That leaves me with the mystery of where the /lib/libdl.so file
came from, and why it was still present on my system.  When I did
a contents search, the only ``libdl.so'' found was a symbolic link
in /usr/lib, and is part of libc6-dev.  Is it possible that the
file in /lib came from some other package that existed in March of
2000?  Any other ideas?  (And, no, I've never built C libraries
outside of the Debian packages.)

At any rate, everything seems to be fine now.  Thanks for the
helpful followups to my original report -- knowing that other
people hadn't had the problem helped localize things a bit -- and
apologies for rasing a false alarm.

   CMC

+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+
 Behind the counter a boy with a shaven head stared vacantly into space, 
 a dozen spikes of microsoft protruding from the socket behind his ear.
+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+
   C.M. Connelly   [EMAIL PROTECTED]   SHC, DS
+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+ 

[*] One extremely annoying discovery was that the libc6 packages
*must be built as root*.  Leaving aside the question of *why*
the packages can't be built with fakeroot (or as an ordinary
user, even) -- the Red Hat RPMs do not require you to be root -- 
it would have been really nice if the package had checked
for root *first*, rather than running for 4.5 hours and only
then realizing it couldn't finish.

Re: Major libc6 problems (with libc6 2.1.97-1 and friends)

[Adam C Powell IV]

"C.M. Connelly" wrote:

> I just had hours of ``fun'' after typing the fateful command
> ``apt-get upgrade'' and having my glibc packages upgraded from
> 2.1.3-13 to 2.1.97-1.

I'm sorry, the only contribution I can make is that I did the same upgrade about
a week ago and "it worked for me".  I have upgraded to woody in steps though,
starting with GNOME and E and only recently doing libc6, so I'm down to about 80
packages held now...

Did you get the new libnss-db?  And everything that goes with libc6, e.g.
locales, ncsd, i18n-data...?  I did it via dselect rather than apt-get, that's
the only difference I can think of...

-Adam P.

  Welcome to the best software in the world today cafe!

Re: Major libc6 problems (with libc6 2.1.97-1 and friends)

[Peter Abrahamsen]

Strange. I compiled 2.1.96 myself, and aside from a big loop of compilation
dependencies (partially caused by my confusion) involving db/lidb, the debs
work fine. I can send them to someone if they like.

Just so you don't have to look at everything that happened between 2.1.3 and
2.1.97 to see what might have gone wrong...

Peter


On Sat, Nov 11, 2000 at 12:45:43AM -0800, C.M. Connelly wrote:
> I just had hours of ``fun'' after typing the fateful command
> ``apt-get upgrade'' and having my glibc packages upgraded from
> 2.1.3-13 to 2.1.97-1.
> 
> Once the process was complete, trying to start any new programs
> (including stuff like `ls') failed, after rebooting, I had lots of
> ``Illegal instruction'' errors affecting programs such as `login',
> and ended up having to jump through an enormous number of hoops to
> downgrade the libraries.  Luckily, I still had the older packages
> in my package cache (because they are no longer available from the
> Debian archives!).
> 
> Among other things, it was impossible to log into the machine
> (outside of the single-user mode root login, which doesn't use
> /bin/login).  Running /bin/login resulted in Illegal instruction
> errors, with segfaults if you typed C-d to terminate the program.
> Using the debugging libraries (with LD_LIBRARY_PATH) allowed you
> to get a login: prompt, but the program still failed.
> 
> Downgrading was especially fun, because the set of packages that
> constitute the C libraries has changed, meaning that some of the
> files in the newer packages are overwritten by the older packages,
> and the system won't allow you to do that without a lot of
> coaxing.  The libnss-db package, which is new, and not needed with
> the older libc6 packages, is marked essential, and can only be
> removed with dpkg and the --force-remove-essential flag.
> /sbin/ldconfig is now included in the libc6 package, which means
> that attempting to downgrade the libc6 package results in ldconfig
> vanishing.  That leaves dpkg in a chicken-and-egg situation where
> it refuses to install ldso, which contains /sbin/ldconfig, because
> it can't find ldconfig.
> 
> I don't know exactly what the problem here is, but you might want
> to consider putting your libc6 packages on hold, or avoiding
> upgrades until you hear that everything's fine -- unless, of
> course, you have lots of time and patience, a complete set of
> older libc6 packages, and maybe an extra bootable partition
> somewhere.
> 
> (Although it's unlikely that this information is important, my
> system is a PowerCenter 132 -- an ``old world'' Mac clone.)
> 
>   CMC
> 
> +=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+
>  Behind the counter a boy with a shaven head stared vacantly into space, 
>  a dozen spikes of microsoft protruding from the socket behind his ear.
> +=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+
>C.M. Connelly   [EMAIL PROTECTED]   SHC, DS
> +=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+ 
> 
> 
> --  
> To UNSUBSCRIBE, email to [EMAIL PROTECTED]
> with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
> 
> 

Major libc6 problems (with libc6 2.1.97-1 and friends)

[C.M. Connelly]

I just had hours of ``fun'' after typing the fateful command
``apt-get upgrade'' and having my glibc packages upgraded from
2.1.3-13 to 2.1.97-1.

Once the process was complete, trying to start any new programs
(including stuff like `ls') failed, after rebooting, I had lots of
``Illegal instruction'' errors affecting programs such as `login',
and ended up having to jump through an enormous number of hoops to
downgrade the libraries.  Luckily, I still had the older packages
in my package cache (because they are no longer available from the
Debian archives!).

Among other things, it was impossible to log into the machine
(outside of the single-user mode root login, which doesn't use
/bin/login).  Running /bin/login resulted in Illegal instruction
errors, with segfaults if you typed C-d to terminate the program.
Using the debugging libraries (with LD_LIBRARY_PATH) allowed you
to get a login: prompt, but the program still failed.

Downgrading was especially fun, because the set of packages that
constitute the C libraries has changed, meaning that some of the
files in the newer packages are overwritten by the older packages,
and the system won't allow you to do that without a lot of
coaxing.  The libnss-db package, which is new, and not needed with
the older libc6 packages, is marked essential, and can only be
removed with dpkg and the --force-remove-essential flag.
/sbin/ldconfig is now included in the libc6 package, which means
that attempting to downgrade the libc6 package results in ldconfig
vanishing.  That leaves dpkg in a chicken-and-egg situation where
it refuses to install ldso, which contains /sbin/ldconfig, because
it can't find ldconfig.

I don't know exactly what the problem here is, but you might want
to consider putting your libc6 packages on hold, or avoiding
upgrades until you hear that everything's fine -- unless, of
course, you have lots of time and patience, a complete set of
older libc6 packages, and maybe an extra bootable partition
somewhere.

(Although it's unlikely that this information is important, my
system is a PowerCenter 132 -- an ``old world'' Mac clone.)

  CMC

+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+
 Behind the counter a boy with a shaven head stared vacantly into space, 
 a dozen spikes of microsoft protruding from the socket behind his ear.
+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+
   C.M. Connelly   [EMAIL PROTECTED]   SHC, DS
+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+ 

Re: apt-get upgrade safe / libc6-2.1.94?

[Daniel Jacobowitz]

On Tue, Oct 17, 2000 at 05:20:58PM +0200, Jan Nieuwenhuizen wrote:
> On Tuesday, 17 October 2000, Robert Ramiega writes:
> > On Tue, Oct 17, 2000 at 11:49:00AM +0200, Jan Nieuwenhuizen wrote:
> > > But it seems that some people think debian=i386 and
> > > we might still be in trouble?
> >  THere is no libc6 package higher than 2.1.3-13 yet on powerpc (that latter
> > is on security.d.o)
> 
> That's what I meant: the version that was reported to be safe is
> not yet available for us.  Do we need to wait (for 2.1.94)?

No.  It was 2.1.93 and 2.1.94-{1,2} that were unhappy - 2.1.3 is
perfectly fine.  I'm not pushing powerpc to 2.1.95 until I'm convinced
it's safe, which should be later this week.

Do be alert for it... it will break things!  But not hideously, and it
may fix exim.  SSH without ipv6 enabled is a little prone to
segfaulting...

Dan

/\  /\
|   Daniel Jacobowitz|__|SCS Class of 2002   |
|   Debian GNU/Linux Developer__Carnegie Mellon University   |
| [EMAIL PROTECTED] |  |   [EMAIL PROTECTED]  |
\/  \/

Re: apt-get upgrade safe / libc6-2.1.94?

[Tom Rini]

On Tue, Oct 17, 2000 at 05:20:58PM +0200, Jan Nieuwenhuizen wrote:
> On Tuesday, 17 October 2000, Robert Ramiega writes:
> > On Tue, Oct 17, 2000 at 11:49:00AM +0200, Jan Nieuwenhuizen wrote:
> > > But it seems that some people think debian=i386 and
> > > we might still be in trouble?
> >  THere is no libc6 package higher than 2.1.3-13 yet on powerpc (that latter
> > is on security.d.o)
> 
> That's what I meant: the version that was reported to be safe is
> not yet available for us.  Do we need to wait (for 2.1.94)?

glibc 2.1.95+current binutils test release is happy on PPC.  Anything before
then is very, very unhappy.  But Dan knows about this so fear not. :)

-- 
Tom Rini (TR1265)
http://gate.crashing.org/~trini/

Re: apt-get upgrade safe / libc6-2.1.94?

[Jan Nieuwenhuizen]

On Tuesday, 17 October 2000, Robert Ramiega writes:
> On Tue, Oct 17, 2000 at 11:49:00AM +0200, Jan Nieuwenhuizen wrote:
> > But it seems that some people think debian=i386 and
> > we might still be in trouble?
>  THere is no libc6 package higher than 2.1.3-13 yet on powerpc (that latter
> is on security.d.o)

That's what I meant: the version that was reported to be safe is
not yet available for us.  Do we need to wait (for 2.1.94)?

> > I suffered problems on x86 at work, so I'm holding back my
> > apt-gets.  Maybe powerpc wasn't affected at all, can someone
> > clarify?
>  You don't need to hold'em (unless it's fully automatic which is not good
> thing in unstable i think). Just inspect what dselect says and eventually
> hold libc6 package (and all that depends). I've done it on one of i386 boxes
> i have.

Dselect?  I only switched to Debian when they fixed their interactive
install/upgrade.

Greetings,
Jan.

-- 
Jan Nieuwenhuizen <[EMAIL PROTECTED]> | GNU LilyPond - The music typesetter
http://www.xs4all.nl/~jantien   | http://www.lilypond.org

Re: apt-get upgrade safe / libc6-2.1.94?

[Robert Ramiega]

On Tue, Oct 17, 2000 at 11:49:00AM +0200, Jan Nieuwenhuizen wrote:
> But it seems that some people think debian=i386 and
> we might still be in trouble?
 THere is no libc6 package higher than 2.1.3-13 yet on powerpc (that latter
is on security.d.o)

> I suffered problems on x86 at work, so I'm holding back my
> apt-gets.  Maybe powerpc wasn't affected at all, can someone
> clarify?
 You don't need to hold'em (unless it's fully automatic which is not good
thing in unstable i think). Just inspect what dselect says and eventually
hold libc6 package (and all that depends). I've done it on one of i386 boxes
i have.

-- 
 Robert Ramiega  | [EMAIL PROTECTED]  IRC: _Jedi_ | Don't underestimate 
 UIN: 13201047   | http://www.plukwa.net/ | the power of Source

apt-get upgrade safe / libc6-2.1.94?

[Jan Nieuwenhuizen]

Hi list,

Dwn of October 3rd 

http://www.debian.org/News/weekly/2000/32/

mentions unstable is safe again, with the upload of:

Installed glibc 2.1.94-3 (all source i386)

But it seems that some people think debian=i386 and
we might still be in trouble?

11:39:40 appel ~ grep -A6 Package:\ libc6$ 
/var/state/apt/lists/ftp.nl.debian.org_pub_linux_debian_dists_unstable_main_binary-powerpc_Packages
Package: libc6
Priority: required
Section: base
Installed-Size: 6220
Maintainer: Joel Klecker 
Source: glibc
Version: 2.1.3-10

I suffered problems on x86 at work, so I'm holding back my
apt-gets.  Maybe powerpc wasn't affected at all, can someone
clarify?

Greetings,
Jan.

-- 
Jan Nieuwenhuizen <[EMAIL PROTECTED]> | GNU LilyPond - The music typesetter
http://www.xs4all.nl/~jantien   | http://www.lilypond.org

Re: exim 3.16-3.1 and 3.16-4 (and libc6 2.1.3-13)

[Robert Ramiega]

  Hi!

On Tue, Oct 03, 2000 at 03:54:33PM +0100, Sergio Brandano wrote:
>  I am using libc6 2.1.3-10, that also seems the only one available
>  for woody, together with exim 3.16-3.1 and I am reporting no
>  problems so far. Where did you get libc6 2.1.94 from? The c libs
 I guess my wording is not to good (sorry English is not my native language).
Anyway i *don't* have libc6 2.1.94 installed (two days ago it was not
available) all i did was upgraded exim to 3.16-3.1 (there were some other
packages upgraded; unfortunately i don't recall which ones). I saw thread on
debian-devel on libc6 upgrades and on Intel box i simply put those packages
on hold. On my powerpc box i let dselect upgrade exim as i didn't suspect any
problems. 
 When i tried to run exim in daemon mode it simply segfaulted. When i put it
in inetd it started to receive messages but was crashing while processing the
queue.
 All this was gone when i switched to exim 3.16-3.
>  are the most delicate thing to compile by yourself, and I would
>  recommend to stick with 2.1.3-10.
 2.1.3-13 is in security.d.o  and generally You are correct (this time i'll
wait till some of You install latest glibc and what came out of it =o))

-- 
 Robert Ramiega  | [EMAIL PROTECTED]  IRC: _Jedi_ | Don't underestimate 
 UIN: 13201047   | http://www.plukwa.net/ | the power of Source

Re: exim 3.16-3.1 and 3.16-4 (and libc6 2.1.3-13)

[Sergio Brandano]

 Hi Robert,

 I am using libc6 2.1.3-10, that also seems the only one available
 for woody, together with exim 3.16-3.1 and I am reporting no
 problems so far. Where did you get libc6 2.1.94 from? The c libs
 are the most delicate thing to compile by yourself, and I would
 recommend to stick with 2.1.3-10.

 Cheers,
 Sergio


Robert Ramiega wrote:

> Just a quick note:
> Yestradys night i've upgraded my system. I saw that there is exim-3.16_3.1
>(and thou i knew it was there to fix incompatibilites with libc 2.1.94 i
>didn't expect it to break). This was a BIG mistake. I ended up without MTA
>for more that 15 hours (exim seg faulted when trying to enter daemon mode). I
>recompiled both 3.1 and 4 package fro sources to no avail. After some
>twidling i got my ystem to accept messages (exim was running from inetd) but
>couldn't sent messages out (exim -q crashed with signal 11).
>
> Thankfully i found some lagged debian mirror that stil had exim-3.16_3.deb 
>
> I  just wonder how did exim-3.16_3.1.deb slip out? Was it compiled against
>libc6 2.1.94? Or libc6 2.1.3? (to say it loud and clear: i don;t blame anyone
>for my problems! I should hold exim since i knew it was meant for libc6
>package that is not available for ppc at that moment; i'm just curious =o))
>
>(is the build daemon running again?)