Subzero: Improve the representation and handling of the FakeKill instruction.

Subzero: Improve the representation and handling of the FakeKill instruction.

The FakeKill instruction is always used for killing scratch registers, and a fair amount of effort is spent building new copies of the same scratch register list each time (i.e., for each lowered call instruction).  As such, we can create one master list of scratch registers and share it among all FakeKill instructions.

Also, in all situations where an instruction's Srcs[] were considered for liveness, we had to either explicitly ignore an InstFakeKill instruction, or treat it specially.  Now that InstFakeKill lacks any Srcs[] (or Dest), it doesn't need to be specially ignored, and the code is simplified.

In addition, the text asm emitter no longer clutters the output with FakeKill comments (and FakeUse as well).

BUG= none
R=jvoung@chromium.org

Committed: https://gerrit.chromium.org/gerrit/gitweb?p=native_client/pnacl-subzero.git;a=commit;h=bb8b624e7c2a4fcdf9275bad310df779797e91a8

[Jim Stichnoth, 2014-11-04 15:45:09]

stichnot@chromium.org changed reviewers:
+ jvoung@chromium.org

[Jim Stichnoth, 2014-11-04 15:46:37]

This seems to give a marginal performance improvement in translation time, on
the order of 1%.

[jvoung (off chromium), 2014-11-04 16:47:07]

LGTM

https://codereview.chromium.org/691693003/diff/40001/src/IceTargetLoweringX86...
File src/IceTargetLoweringX8632.cpp (right):

https://codereview.chromium.org/691693003/diff/40001/src/IceTargetLoweringX86...
src/IceTargetLoweringX8632.cpp:4544: // allocator.
Maybe this can be an assert that isa<InstFakeKill>(Inst) implies that the
NumSrcs are 0. Then the comment above will be clearer about why InstFakeKill is
brought up here, but no code checks for it anymore.

[Jim Stichnoth, 2014-11-04 17:10:06]

Committed patchset #4 (id:60001) manually as
bb8b624e7c2a4fcdf9275bad310df779797e91a8 (tree was closed).

[Jim Stichnoth, 2014-11-04 17:10:39]

https://codereview.chromium.org/691693003/diff/40001/src/IceTargetLoweringX86...
File src/IceTargetLoweringX8632.cpp (right):

https://codereview.chromium.org/691693003/diff/40001/src/IceTargetLoweringX86...
src/IceTargetLoweringX8632.cpp:4544: // allocator.
On 2014/11/04 16:47:07, jvoung wrote:
> Maybe this can be an assert that isa<InstFakeKill>(Inst) implies that the
> NumSrcs are 0. Then the comment above will be clearer about why InstFakeKill
is
> brought up here, but no code checks for it anymore.

Done.