Re: [patch] MIPS -G 0 libgcc vs. (default -G) stdc++ data problem.

["Zack Weinberg" <zackw at stanford dot edu>]

On Sat, Oct 28, 2000 at 08:33:36PM -0700, Chris G. Demetriou wrote:
> Last night i reported an issue with libgcc fighting with libstdc++ in
> a message to gcc-bugs:
> 
> 	http://gcc.gnu.org/ml/gcc-bugs/2000-10/msg00568.html
> 
> In a nutshell, libstdc++ is compiled with default -G value on MIPS,
> but libgcc is compiled with -G 0.  This causes libstdc++'s attempt to
> directly access libgcc's internal variable __terminate_func to fail.

Is __terminate_func used by non-C++ runtimes?  If not, it should
simply be moved over to libstdc++(v3) with all the rest of the C++
specific logic.

> It looks like one other data symbol is exported from libgcc.a on a
> mips-elf-like target: __shtab.  I couldn't find any uses of it in the
> gcc src tree (at all!), though, and it's a 128 byte array, so even if
> it is used it should be safe in my circumstances...

Yeah, what *is* that for?  It's a 32x32 square diagonal bit matrix,
and I can't find any references in gcc, binutils, gdb, or glibc.  No
mention in any change log I have access to, either.

zw