bug-coreutils
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

glibc-2.4 fchownat bug [Re: 2 test failures


From: Jim Meyering
Subject: glibc-2.4 fchownat bug [Re: 2 test failures
Date: Mon, 23 Oct 2006 19:50:01 +0200

Greg Schafer <address@hidden> wrote:
> This is on i686-pc-linux-gnu against glibc-2.4 and applies to both coreutils
> 6.3 and 6.4.

Thanks for reporting that.
Does your glibc-2.4 have any patches?
What distribution?

> IMPORTANT INFO - both failing tests pass when built aginst glibc-2.5 or if
> coreutils is configured like this:
>
> ./configure ac_cv_func_openat=no
>
> This suggests both failures are caused by bugs in glibc-2.4 so maybe this is
> a useless bug report for the coreutils team. But then again, maybe coreutils
> needs to work around glibc bugs?

This bug was apparently fixed before 2.5, but I haven't tried to find
out when.  I reported it against Ubuntu, because at the time it failed
with their glibc-2.4, but not the one from upstream's rawhide, which
was still glibc-2.4:

    https://launchpad.net/distros/ubuntu/+source/glibc/+bug/59083

> The first test to fail (running as root) is:
>
> tests/chown/basic (verbose log attached - log1.txt)
>
> *** incidentally, the "check-root" target is a bit buggy because it doesn't
> continue on in the event of failures even if you run "make -k check" as per
> the docs ***

Thanks for running those tests as root.
Distributions that track upstream glibc-2.4 should definitely
look into fixing this.  It's a bug that can have serious consequences.

So far, I've encountered this bug only in Ubuntu's unstable.
If you know of another distribution that is affected, please
let us know.  If enough people are affected, and if it looks
like the distro-glibc maintainers are not interested in fixing
the bug, then I'll consider working around it in the coreutils.

> The second test to fail is:
>
> tests/chgrp/basic (verbose log attached - log2.txt)

I think it's the same underlying problem.

> As an aside, the Debian utility "fakeroot" no longer works properly when
> latest coreutils are built against glibc 2.4 or 2.5. It seems some of the
> new *at() calls will need to be wrapped. Again, use the `./configure
> ac_cv_func_openat=no' hack as a workaround.




reply via email to

[Prev in Thread] Current Thread [Next in Thread]