[COMMIT] Infrastructure Updates 4

[Eric W. Biederman]

YhLu <YhLu at tyan.com> writes:

> Eric,
> 
> Without hard_reset, s2885 is ok now.
> 
> In the k8/cpufixup.c need compare mmio_basek and 0x3f0000 and then assign
> that to TOM.

Hmm. What is special about 0x3f0000?

In any event this sounds promising.  It looks like the reset needs to move
a little farther down in the code until after all of the hypertransport
chains have been enumerated.

Eric