Problem with glibc test

Dan Nicholson dbn.lists at gmail.com
Wed Oct 8 22:03:28 PDT 2008


On Wed, Oct 8, 2008 at 5:03 PM, Bill Mason <billmason_sp2 at yahoo.com> wrote:
> Thanks Dan,
> I'll check -Wall.
>
> Can you add a bunch of zeros and see if it still works?  I was able to try this on another system, and the problem still existed, but it needed a bit more zeros.

Ah, now I got it. Looking at the log for vsprintf.c, it seems there
were quite a few fixes to do with overflowing on precision.

http://sourceware.org/cgi-bin/cvsweb.cgi/libc/stdio-common/vfprintf.c?cvsroot=glibc

This bug seems relevant, but I think the patch is in 2.5.1.

http://sources.redhat.com/bugzilla/show_bug.cgi?id=4438

In fact, the patch from that bug adds the test that you're failing.

http://sourceware.org/ml/glibc-cvs/2007-q3/msg00113.html

So, uh, I guess that bug wasn't fixed by 2.5.1. If this is your only
error, I'd suggest moving on for now. You can try to keep digging for
the fix from glibc CVS, but I don't know if I'd be applying patches
that I didn't understand to glibc. Next to the kernel, it's pretty
much that component that drives everything.

--
Dan



More information about the lfs-support mailing list