libc6 update breaks dchroot, recompile fixes

Hi. Are you running feisty?

Is this still a issue for you?

I'm not running feisty, and I imagine it's not an issue in feisty as the package has likely been recompiled for a different version of libc6 anyway. I am still having this problem in edgy last time I checked.

pasted from #ubuntu-devel:

13:11:17 < siretart> pitti: could you please have a look at #81893? if the submitter is right, there might be a lot more packages affected as well...
13:13:21 < pitti> siretart: right, this looks like an adequate SRU
13:13:54 < siretart> pitti: adequate or not, why does the libc6 update break dchroot in the first place, and how can we sure that dchroot is the only affected binary?
13:13:57 < pitti> siretart: still, this is worrying me -- was this an ABI change or so?
13:14:42 < Fujitsu> I'd say it is rather worrying, personally.
13:14:54 < pitti> siretart: can't say yet, I'd need to look at a stack trace and check if downgrading indeed fixes it
13:15:30 < pitti> siretart: also, the architecture would be interesting; there had been some issues on amd64, and rebuilding libc with a new compiler in edgy might have caused that

does this happen on all architectures or just on some architectures?

oh, and could you please attach an crash file?

I've only seen this on my AMD64 box (as that's the only place I have a chroot environment ;) ), so I can't verify whether other archs are affected.

As for crash file, how should I generate one for you?

Just went to try and generate a core dump with the packages in the repository instead of the ones I recompiled, and I can't reproduce the problem.

I had a thought about the cause though... I've been using prelink on my systems under the assumption that it speeds load time. Is it possible that the prelinking of this binary broke compatibility when libc was updated, and that a recompile of that package was a coincidental fix (as it replaced the prelinked binary with a nonprelinked one), and not directly related to the problem? Or is that not a possible side affect of prelinking.

OTOH, is it possible the package in the repo been recomplied since I filed this bug? I have a hard time telling that from the repository information.

Hi Graeme,

Graeme Humphries [2007-03-21 16:47 -0000]:
> I had a thought about the cause though... I've been using prelink on my
> systems under the assumption that it speeds load time. Is it possible
> that the prelinking of this binary broke compatibility when libc was
> updated

I am not entirely familiar with prelinking, but it does sound like a
good culprit candidate.

> OTOH, is it possible the package in the repo been recomplied since I
> filed this bug? I have a hard time telling that from the repository
> information.

https://launchpad.net/distros/ubuntu/+source/dchroot shows that the
package has never been rebuilt for dapper after dapper's release.

Ok, well then it was likely a prelinking problem, which is local to me. Sorry for filing an erroneous bug. But at least you know it's not a libc ABI problem. ;)

I'll close this now.