Re: ed SEGV
On Tue, Feb 05, 2002 at 11:40:01PM -0500, utsl@quic.net wrote:
> On Tue, Feb 05, 2002 at 09:20:25PM -0700, Joel Baker wrote:
> > On Wed, Feb 06, 2002 at 03:18:49AM +0000, Matthew Vernon wrote:
> > > "Joel Baker" <lucifer@lightbearer.com> writes:
> > >
> > >
> > > > It's apparently being caused by stack corruption... lord knows how THAT
> > > > is happening. Time to dig out the toolkit. Or rather, try to compile the
> > > > toolkit... (GDB should be up, now, for the record).
> > >
> > > I suspect Crap Code(TM). Or someone relying on a behaviour of glibc
> > > that isn't standard.
> >
> > Except that the source is taken from FreeBSD's /usr/src tree...
> >
> > However, GCC 2.95.4 compiled; I'm now going to back up my current tarball,
> > and do some test compiling with the new GCC to ensure that it works sanely,
> > then try ed under 2.95.4 to try to weed out compiler issues.
>
> Which sources are you using for ed? I used the ed package from Debian, and all
> I had to do was put libed.a back into the ed target in Makefile.in. make check
> passes just fine.
Forcing it to link with libed.a (which, I will note, provides it's own
regex engine) appears to avoid the problems.
I strongly suspect the chroot libc has some issues in the regex routes. I
intend to try to do a native-in-the-chroot compile of a kernel and libc
at some point in the not-too-distant future.
GCC 2.95.4 compiles and packages; I'm going to try to get it to use dejagnu
and pass it's testsuite, however, before I put it up publically.
--
***************************************************************************
Joel Baker System Administrator - lightbearer.com
lucifer@lightbearer.com http://users.lightbearer.com/lucifer/
Reply to: