[PATCH} Review request : Direct object exception handling for Mips

[Rafael Espíndola]

> Revision 177141 caused a regression in all but
> mips64 little endian. That is because none of the
> other Mips targets had test cases checking the
> contents of the .eh_frame section. This patch fixes
> both the llvm code and the test case to include the
> current 4 flavors.

Are you sure this was a regression? Is mips really the only
architecture using dwarf::DW_EH_PE_absptr? That looks really
suspicious. Is that what the gnu assembler does? If that is the case,
the patch should have a comment on why that is.

Second, this should not use a .ll to .o test. It should use a .s to .o
test. You should probably just update fde-reloc.s to also test mips
(now that I think of it, it probably has to be copied to test/MC/X86
and test/MC/Mips). Note that the test has no assembly instructions,
just cfi directives, so it already works on mips (I just tested that
with "llvm-mc fde-reloc.s -o test.o -filetype=obj -triple
mipsel-unknown-linux").

> Jack

Cheers,
Rafael