GCC 4.9.1 changes.

[Chris Johns]

Hi,

There are a few issues with 4.9.1 which need to be looked at. I have not 
attempted a Windows build yet.

My question is:

  Do we move the architectures that *do* build to 4.9.1 ?

If there are patches or known option required to build the failing 
architectures please let me know.

Is anyone test builind gcc head on a regular basis ?

This is the list of architectures do not build:

bfin:
../../../gcc-4.9.1/libgcc/fp-bit.c: In function '__divsf3':
../../../gcc-4.9.1/libgcc/fp-bit.c:1082:1: internal compiler error: in 
cfg_layout_initialize, at cfgrtl.c:4233

m32c:
configure: error: unable to detect exception model

mips:
xgcc: error: addsf3: No such file or directory

moxie:
We are using binutils-2.24. Is there a later release ?
CC_TLS -DUSE_EMUTLS -o _gcov.o -MT _gcov.o -MD -MP -MF _gcov.dep 
-DL_gcov -c ../../../../gcc-4.9.1/libgcc/libgcov-driver.c
/tmp//cc7Lxv5S.s: Assembler messages:
/tmp//cc7Lxv5S.s:176: Error: unknown opcode sex.b $r0,$r0

   [ AG, a rather interesting name for an instruction ]

powerpc:
../../../../gcc-4.9.1/libgcc/fp-bit.c: In function '_fpadd_parts':
../../../../gcc-4.9.1/libgcc/fp-bit.c:738:1: internal compiler error: in 
dwf_regno, at dwarf2cfi.c:909

Chris