Posted (Feb. 26, 2011, 8:55 a.m.)
I see what you're doing here, I think. For the strategy your taking your implementation seems ok (other than some weird whitespace), but I'm not sure this is the way to go. The itstate should be part of the pc and then wouldn't depend on anything in the previous instruction. The way you've implemented this forced itstate stuff may be the real culprit. I don't remember all the details of our previous conversation about it, but I do remember suggesting a way to implement that that fit better with the rest of the pc state information and I expect would make this change unnecessary.
-
src/arch/arm/predecoder.hh (Diff revision 1) -
Why all the extra blank lines?
