Re: libc resolver problem solved (critical bug)
On Wed, Nov 18, 1998 at 05:04:22PM +0100, Gergely Madarasz wrote:
> On Wed, 18 Nov 1998, Oleg Krivosheev wrote:
> > Hello!
> > I just got an idea from a recent thread on debian-devel... libc 2.0.7u was
> > compiled with egcs, so that might have caused the problems in the resolver
> > (see critical bug #27334). And it proved to be correct. I just compiled
> > 2.0.7u-5 with gcc on hamm, and it works nicely. The test environment:
> > a list server with 35k deliveries a day. I installed the official
> > 2.0.7u-5, and got 50 "host not found" messages from sendmail in a minute.
> > After that I installed the version compiled with gcc... it works
> > flawlessly for 20 minutes now. So please recompile 2.0.7u with gcc
> > (current 2.0.7u-5: Compiled by GNU CC version egcs-2.90.27 980315
> > (egcs-1.0.2 release).)
> > hmm...
> > there are bugs (two, i think) in egcs-1.0.2. If you'd like, i can
> > provide you with 2.0.7u-5 compiled by egcs-1.1. Just tell me where
> > to put all this stuff
> Actually this might be the easiest way of solving the problem... if it
> works... especially considering __register_frame_info ... but from then on
> we'll always have to compile libc with egcs, and the versioned dependency
> (>= 2.0.7u) will have to stay. actually i'd suggest a >=
> 2.0.7u-(first_fixed_revision). Anyway, I'll compile it with egcs 1.1 now
> and test it, thanks for the idea.
If I'm not completely insane, Zack said something about having sent a
patch to Ulrich and Dale about __register_frame_info, and it being an
actual glibc source bug.