PATCH: PR ia64/14925: libunwind enabled gcc generates incompatible libgcc_s.so.1

["H. J. Lu" <hjl at lucon dot org>]

On Wed, Aug 25, 2004 at 12:50:38AM -0700, Mark Mitchell wrote:
> David Mosberger wrote:
> 
> >>>>>>On Tue, 24 Aug 2004 23:33:31 -0700, "H. J. Lu" <hjl@lucon.org> said:
> >>>>>>           
> >>>>>>
> >
> > HJ> By default, the current gcc 3.4 generates incompatible
> > HJ> libgcc_s.so.1 on ia64 machines with system libunwind. In that
> > HJ> sense, gcc 3.4 branch has a broken ABI. My patch does change the
> > HJ> ABI by providing the binary compatibility for ia64 machines with
> > HJ> system libunwind. I think it should be considered as a special
> > HJ> case for gcc 3.4 branch.
> >
> >As you probably would expect, I very much agree with HJ that this is a
> >special situation.  IMHO, gcc 3.4.0 and 3.4.1 are horribly broken for
> >ia64 and need to be fixed asap (before they're being widely deployed
> >by distributions).  I believe the ABI-incompatility in 3.4.0 and 3.4.1
> >also affect icc (Intel's compiler), so the compatibility is unlikely
> >to be limited to libunwind.
> > 
> >
> I do not object to the patch, if either Jim Wilson or Richard Henderson 
> approves it.
> 

Here is the version for mainline. The one for gcc 3.4 branch is at

http://gcc.gnu.org/ml/gcc-patches/2004-08/msg01398.html

They have been tested on ia64, ia32 and x86_64. Are they OK for
mainline and 3.4?


H.J.

Attachment: gcc-3.5-libunwind-1.patch
Description: Text document