gmp 4.3.2 regressions when compiled with llvm-gcc
howarth at bromo.med.uc.edu
Thu Mar 31 19:32:52 CEST 2011
On Thu, Mar 31, 2011 at 07:17:45PM +0200, Torbjorn Granlund wrote:
> I assume your message is a "FYI", and that it is not a request for
> workarounds in GMP.
> We are quite restrictive with adding workarounds for compiler bugs in
> the GMP sources, even when the problem compiler is GNU's own (i.e.,
> Mac OS X has been the problem platform for GMP for all Mac OS X's 10
> years of existance. The linker, aseembler, and Apple's hacked GCC all
> have substandard quality. There are signs that it will go from bad to
> worse with LLVM...
If you have access to a x86_64 linux box which defaults gmp to core2,
you could test this problem locally using the prebuilt linux x86_64 binaries
for llvm-gcc4.2 from http://llvm.org/releases/download.html#2.8. I suspect
this issue may not be darwin specific but a general llvm-gcc/-mtune=core2
Note I also tried a build under clang 2.9 but ran into a darwin linker
bug. Once that is fixed upstream, I'll let you know if clang exhibits
the same issues.
ps Aren't the other BSDs moving to llvm/clang? If so, I think these issues
may have a wider audience than just darwin.
More information about the gmp-bugs