This is the mail archive of the gcc@gcc.gnu.org mailing list for the GCC project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: IA64 Crashes in stack unwinding


On Thursday 18 March 2004 08:59, Jim Wilson wrote:
> Otto Bruggeman wrote:
> > I'm seeing crashes with a 3.3.2 gcc and what will be 2.3.3 glibc on a quad 
> > Itanium 2 system during backtrace() calls. Can anyone confirm this ?
> 
> You should look at the optional libunwind package.  This is a much 
> better unwinder than the one built into gcc.  In particular, it should 
> work much better for unusual cases such as unwinding from signal handlers.
> 
> If you build and install libunwind, and then build and install gcc, then 
> gcc will automatically use libunwind.
> 
> See
> 	http://www.hpl.hp.com/research/linux/libunwind/

Ok, from which version is this supported ? gcc 3.3.something or later like 
3.4/5 ?  Or does 3.2.2 already support this in its configure, if not can you 
give me an estimate on when and where it was added to CVS so i can search the 
commit logs for the source changes ?

I've started testing this because i'll have to backport the backtrace 
functionality from gcc 3.3.x to 3.2.2 and glibc 2.2.5 for IA64  for SuSE SLES 
8. We unfortunately cannot use libunwind directly by linking to it (some 
legal issues i guess that i don't (want/need to) know about), but we can use 
it if the compiler and base libraries have built-in support for it. So i have 
to port this functionality to SLES8 for IA64 because there is a need for it 
on this platform and waiting for SLES9 is not an option.

Thanks for the help,
Otto Bruggeman


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]