On 10/07/2012 11:48 AM, Blue Swirl wrote: > This is not yet optimal, we could do the same trick as is done for x86 > for cmp/jcc case (target-i386/translate.c:889), but it's still an > improvement so I applied it with others.
It's trickier with sparc, because of the (exceedingly rare) conditional branch in branch delay slot case. I've yet to come up with a plan to handle that which is better than the existing cpu_cond global variable. (Which we now at least initialize with setcond.) r~
