Add 64-bit support for PowerPC Book-E

[jhibbits]

Summary

Extend the Book-E support into 64-bits. Fix drivers that make 32-bit
assumptions.

Test Plan

Continuously testing on a P5020 board (Amiga X5000/20)

[jhibbits]

I'll note this is not complete, but I'll take feedback as I progress.

[jhibbits]

Remove driver updates, to be done separately. Add powerpc 32-bit binary support.

Tested with running /sbin/init (64-bit and 32-bit) from an embedded md.

[stevek]

sys/powerpc/booke/locore.S

585–589

Shouldn't that just be the following (since you have the #defines for it):
LOAD %r3, 0(%r3)

[jhibbits]

sys/powerpc/booke/locore.S

585–589

Yes it should.

[stevek]

sys/powerpc/booke/locore.S

982–992

Given this and other usage elsewhere. Would it make more sense to create a macro for the use of .llong or .long based on __powerpc64__ defined or not?

sys/powerpc/booke/pmap.c

611

Typo in comment. Should be...

/* Free corresponding ptbl buf. */

769

Comment says "this will sleep", yet you have provisions for nosleep flag.

1352

Why is this needed? It doesn't seem to be used in these additions.

1460

Line seems a bit long, candidate for wrapping.

4039

Shouldn't the wasted space part be using PRI0ptrX also?

4071

It would be better to have #defines for the VA and PA in this tlb1_set_entry, instead of bare values.

[jhibbits]

sys/powerpc/booke/locore.S

982–992

Good idea.

sys/powerpc/booke/pmap.c

611

Oops, typo copy.

769

This was copied from Juniper code. I forgot to clean up the comments when adapting it to head.

1352

Oops, this was for debug purposes. Need to remove this and one other.

1460

Probably a better candidate for macroizing.

4039

Yes it should. Good catch.

4071

This was also debug. It'll be committed separately with EARLY_PRINTF changes.

[jhibbits]

Address Steve's comments

[jhibbits]

Fix 32-bit builds.