Add multiplication and division to x64 assembler. Add emit_modrm() function.

Add multiplication and division to x64 assembler.  Add emit_modrm() function.

Committed: http://code.google.com/p/v8/source/detail?r=2098

[Kevin Millikin (Chromium), 2009-06-03 13:13:25]

LGTM.

http://codereview.chromium.org/119078/diff/1/2
File src/x64/assembler-x64.cc (right):

http://codereview.chromium.org/119078/diff/1/2#newcode482
Line 482: if ((adr.code() & 0x8) == 0x8) {
(adr.code() > 7), or
(adr.code() >= 8).

http://codereview.chromium.org/119078/diff/1/2#newcode672
Line 672: if ((target.code() & 0x8) == 0x8) {
Same.

[Lasse Reichstein, 2009-06-03 16:49:15]

LGTM.

http://codereview.chromium.org/119078/diff/1/2
File src/x64/assembler-x64.cc (right):

http://codereview.chromium.org/119078/diff/1/2#newcode482
Line 482: if ((adr.code() & 0x8) == 0x8) {
Or 
(adr.code() & 0x8) != 0
We use this test so often, Register should have a method for it. Perhaps
Register::IsExtendedRegister().

http://codereview.chromium.org/119078/diff/1/2#newcode790
Line 790: emit(0xB8 | (dst.code() & 0x7));  // Not a ModR/M byte.
We don't write that comment in other places with the same type of opcode.
To match the opcode specification, "B8+ rd", better, you could write + instead
of |, i.e.:
 0xB8 + (dst.code() & 0x7)
Perhaps introduce a function on Register returning the low-bits, so it's 
 0xB8 + dst.low_bits()
or something similar.

http://codereview.chromium.org/119078/diff/1/2#newcode1142
Line 1142: 
No testq(Register,Immediate)?

[William Hesse, 2009-06-12 11:55:58]

The comment about using register.code() & 0x7 and
register.code() & 0x8 so much is good.  If we add
register.low_bits() and register.high_bit(), it should
clean up the code significantly.