A few tweaks to the CRIS port, nothing to see, move along. (As CRIS is not being a target on which the release depends, I can do this even at this release stage.)
The set was regtested together, but is order-dependent (at least 1 and 2). It was committed separately for possible future bisection needs. Together, the patches did improve coremark results, but by a miniscule factor: speed by 0.002% (from 4887074 to 4886993 cycles) and size by 0.1% (code from 58199 to 58143 bytes) and as you can see, with rounding doing heavy lifting. This is certainly not fallout from rewriting a not-so-new pass, since nobody to my knowledge is working on any pass that's long been dysfunctional, and even if they were, they'd probably also deny everything just as much (how would I know?) and not committing or even alluding to any kind of deadline. brgds, H-P
