Hi,
> I made a faster version of AllocSetFreeIndex for x86 architecture.
Neat, I have a version for PowerPC too.
In order to prevent writing multiple copies of AllocSetFreeIndex, I
propose that we add a fls() function ("find last set"); this can be
defined in an architecture-independent manner (ie, shift mask & test in
a loop), and re-defined for arches that have faster ways of doing the
same (ie, cntlz instruction on powerpc).
We can then change AllocSetFreeIndex to use fls().
Patches coming...
Jeremy