uClibc gcc-svn support

[Peter S. Mazinger]

On Tue, 7 Mar 2006, Joseph S. Myers wrote:

> On Tue, 7 Mar 2006, Peter S. Mazinger wrote:
> 
> > it seems that unwind-dw2-fde-glibc.c has a check for DT_CONFIG, that would 
> > make it use unwind-dw2-fde.c on uClibc I think, so the only offending line 
> > to get omitted for uClibc is
> > SHLIB_MAPFILES += $(srcdir)/config/libgcc-glibc.ver
> > 
> > what check could be used in this file to see if we are building for 
> > uClibc?
> 
> When toplevel libgcc is merged to mainline GCC, it will be straightforward 
> to put in configure-time or make-time tests that get run for each 
> multilib, test whether that multilib is uClibc and adjust such things 
> accordingly.  Until then it's difficult to do such things correctly, but I 
> don't think the different symbol versions do much harm.
> 
> I.e., for buildroot you might continue to condition on the target triplet 
> - that is the status quo for buildroot.  For mainline GCC, once toplevel 
> libgcc is in a proper fix could go in mainline GCC, and the target 
> triplet conditional could be removed from buildroot for 4.2.0 and later.

Ok, I do not quite see how to do it without copying t-linux.
Would you commit to gcc-svn the remaining bits (patch I attached 
earlier)? I redo it so that it applies correctly if you would, 
omitting the sh* part. Then buildroot and gcc-svn would be in sync 
(only the clocale=uclibc support being kept in buildroot)

Thanks, Peter

-- 
Peter S. Mazinger <ps dot m at gmx dot net>           ID: 0xA5F059F2
Key fingerprint = 92A4 31E1 56BC 3D5A 2D08  BB6E C389 975E A5F0 59F2