Re: unordered comparisons for mips

[Eric Christopher <echristo at redhat dot com>]

> Uhm, great this reminds me the old issue we've run across about branch
> reversability.  To get this working reliably, you need update the conditional
> moves, as they currently aborts if someone attempts to create them with
> unordered opcode.
> 

Hrm.

> I have patch somewhere that avoids abort and makes pattern to claim that such a
> conditional move is unsupported, but it is better to get it supported of
> course.  Are you interested in it?
> 

Sure.  I'd love to take a look at it.

-eric


-- 
Look out behind you!