On Tue, Oct 21, 2003 at 11:26:32AM +1000, Duraid Madina wrote: > Daniel Jacobowitz wrote: > > >You didn't say which version of the Intel compiler broke. I'm told > >that this compiler has known issues with recent versions of glibc so > >I'm inclined to suspect that the compiler is at fault. > > (%:~)- ecc -V > Intel(R) C++ Itanium(R) Compiler for Itanium(R)-based applications > Version 7.1, Build 20030924 > Copyright (C) 1985-2003 Intel Corporation. All rights reserved. > FOR NON-COMMERCIAL USE ONLY > > GNU ld version 2.14.90.0.6 20030820 Debian GNU/Linux > Supported emulations: > elf64_ia64 > > Intel claims this version supports glibc 2.3.2, but who knows.
If earlier versions were broken on glibc 2.3.2 it isn't surprising to me that current versions are broken with later glibc. > >JVMs are famous for abusing glibc interfaces, also. > > The thing is, in both these cases (compiler+jvm) they're not actually > _doing_ much: they segfault immediately, or very close to it. That doesn't say anything, really. There's a dozen things it could be misusing in startup. > >X is worrisome. Can you reproduce this without using a beta CVS > >version of XFree86? > > No, because there's only so far back I can go with XF86 before it just > doesn't work on my platform. However, I can say that the bug also exists > in a CVS version from about 3 months ago. > > >Or any less monolithic application? You haven't given us much to work > >with. > > To reproduce this, all you should have to do is: > > 1) download > http://download2.bea.com/pub/jrockit/81/jrockit-8.1sp1-j2se1.4.1-linux64.bin > 2) run (to install) > 3) type 'java' > > Hopefully you can then tell if it's the JVM at fault, or a bug in > glibc. > > Sorry I'm not being very helpful! I can't test ia64. Someone else will have to. -- Daniel Jacobowitz MontaVista Software Debian GNU/Linux Developer
