PATCH: First part of 64-bit Darwin support, revised

David Edelsohn dje@watson.ibm.com
Tue Aug 17 18:38:00 GMT 2004


>>>>> Stan Shebs writes:

>> And the match_parallel patterns appear to be generating RTL that
>> is unrecognized by lmw patterns in rs6000.md as seen by ppclinux32 and
>> aix32.

Stan> That's pretty weird, because they're all 64-bit and three are
Stan> altivec-only. 

	Zack warned about "any_operand" (and now "any_parallel_operand")
predicate after his genpred changes.  AIX experienced the problem with
-mcpu=power.  Andreas experienced it with -Os, which implicitly enabled
MASK_MULTIPLE and MASK_STRING in FSF (I think that was changed in
apple-ppc branch).  Maybe something about your patch causes confusion with
all of the match_parallel patterns related to any_operand, even though
your new patterns are 64-bit specific.

David



More information about the Gcc-patches mailing list