bug report - g++ + linuxthreads == seg fault

[llewelly@dbritsch.dsl.xmission.com]

On Tue, 9 May 2000, Brent Verner wrote:

> on Tue, May 09, 2000 at 08:57:29PM -0600, llewelly@dbritsch.dsl.xmission.com typed aloud:
> | On Tue, 9 May 2000, Brent Verner wrote:
> | 
> | I compiled and ran this source on my i686-pc-linux-gnu (redhat 6.1 +
> |   mods), and I got:
> | 
> | {~/cc_exer}g++ -g -Wall -O2 -lpthread c++_threads.cc
> | {~/cc_exer}./a.out
> | calling thread
> | running me
> | thread joined
> | {~/cc_exer}gcc --version
> | 2.95.2
> | 
> | binutils: 2.95.0.41
> | glibc: 2.1.2, with matching linux threads
> 
> I have:
>     glibc 2.1.3 + matching linuxthreads
>     gcc version 2.96 20000508 (experimental) [cvs co about 2pm]
>     binutils 2.95.0.41
>  
> note, that I've seen this problem since the 'cvs co egcs' on 20000428.

I did not realize you were using cvs gcc.

> 
> | So upgrading each of the above may help.
> | 
> |   gcc is easiest to upgrade, so try upgrading it first (if you do not have
> |     2.95.2 already) (do not forget to pass --enable-threads to configure.)
> | 
> |   glibc + linuxthreads is probably the most difficult upgrade, so try that 
> |     last.
> | 
> |   If you already have the versions above (or newer), you may have an
> |     installation problem.
> 
> what can I do to fix any installation problem if that's what it is?

I was thinking of the --enable-threads issue, but that usually shows up as
  a segfault when throwing an exception in a threaded program.

On second thought, I suspect a bug in cvs gcc. 

I am not an expert or a gcc developer, so I do not know where to look
  next.

> 
> | > 
> | > ==========================
> | > the source: 
> | > ==========================
> | >     (g++ -v --save-temps -O2 -lpthreads breaks.cc) [attached]
> | 
> | hm. I did not get any attachements.
> 
> sorry. ddd hung my box, so I quickly sent (w/o attachment) email
> so I could kill X... 
> 
> it is now attached.
> 
> brent
>