[Bug target/20126] [3.4 Regression] Inlined memcmp makes one argument null on entry

[uweigand at gcc dot gnu dot org]

------- Additional Comments From uweigand at gcc dot gnu dot org  2005-07-08 15:48 -------
(In reply to comment #51)
> I don't know whether I just forgot about it, or figured we'd be better off
> leaving it as it was for a bit longer, so as to expose more cases we could
> handle especially.

Well, in my case the problem is that the pattern has two memory operands 
whose addresses need to agree, and loop is trying to change one of them
--> the insn predicate rejects.

I don't see how this can be fixed easily, and don't think much effort
should be put into the old loop code -- but we need a safe fall-back
to avoid the ICE.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=20126