patches to reorg.c for c4x

Michael Hayes m.hayes@elec.canterbury.ac.nz
Wed Sep 30 23:08:00 GMT 1998


Jeffrey A Law writes:

 > My first question is how did we end up with an insn which sets CC0
 > in a delay slot?  I think that is the real root of the problem -- I
 > do not think that should be happening.

 > I suspect there's something funny going on in the c4x backend which
 > is leading to this problem.  The conditional branch should
 > (effectively) use the cc0 reg, which should in turn prevent an insn 
 >  which sets the cc0 reg from being placed in its delay slot.  Right?

The c4x doesn't use cc0, it uses a hard register for CC. 

I think the problem this patch fixes is not exposed by other targets
which only have a single delay slot.

Michael.






More information about the Gcc-patches mailing list