On Friday 04 September 2009 20:59:49 DJ Delorie wrote:
> > m32c-elf/log0:FAIL: comment.s: comments in listings
> > m32c-elf/log0:FAIL: align
> > m32c-elf/log0:FAIL: incbin
> > m32c-elf/log0:FAIL: semi
>
> These are all caused by a fluke in the way m32c works.  The CGEN
> engine always reads enough bytes for the longest opcode it might
> encounter.  Thus, gas always adds extra bytes to the end of the last
> section in each object, which is messing up the tests.  Without this,
> many real operations fail due to bfd trying to read past the end of
> the section.  So, the failures aren't "real" failures, just annoying
> ones.
>
> The only real way to solve this one is to replace the CGEN engine with
> all-new code.  Not gonna happen for 2.20 ;-)

we can tag them as XFAIL in the branch though ...
-mike