Randy McMurchy wrote:
Hi all,
I would like to propose a consideration for LFS to move towards the
GCC-4 branch as the default LFS build. There are issues, but none that
are really show-stoppers.
Well, there's currently three issues I'm wanting to see fixed/addressed
before considering a GCC-4 based release.
1) Incorrect password given to 'su' causes it to segfault if compiled
against PAM. We need to ascertain whether this is because of Shadow or
because of PAM. I'm guessing, because it doesn't trigger without PAM,
then it must be a PAM issue.
2) Inetutils doesn't compile without a patch, but the patch we have
causes 'ftp' to segfault. (Bug 1602)
3) 'cfdisk -P t' segfaults on particular partition layouts (at least I'm
guessing it's dependent on partition layouts, as reproducing it appears
to be 'tricky' at best). I've tried working with Chris Beaker on this,
but we were unable to reliably trigger/fix the bug. Changing the
optimisation from -O2 to -O makes the problem go away. Compiling with
'-O -funit-at-a-time' triggers it again (and it's the only one of the
'-O2' level optimisation passes that does so). Compiling with '-O2
-fno-unit-at-a-time' still segfaults though. (1604).
Cheers,
Matt.
--
http://linuxfromscratch.org/mailman/listinfo/lfs-dev
FAQ: http://www.linuxfromscratch.org/faq/
Unsubscribe: See the above information page
