Subzero: Fix incorrect address mode inference involving Phi temporaries.

Subzero: Fix incorrect address mode inference involving Phi temporaries.

Also, refactor the key part of the address mode inference into separate functions, since it's getting unwieldy.

The main thing is that we mark phi temporaries as multi-definition, and disallow address mode inference transformations that involve such temporaries, because this is incorrect particular when there are backward branches involved.

BUG= none
R=jvoung@chromium.org

Committed: https://gerrit.chromium.org/gerrit/gitweb?p=native_client/pnacl-subzero.git;a=commit;h=e5ac7db

[Jim Stichnoth, 2014-09-15 12:28:54]

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

[Jim Stichnoth, 2014-09-15 12:29:50]

This fixes the remaining spec2k/O2 components, at least with the training set
input.

[jvoung (off chromium), 2014-09-15 16:37:04]

https://codereview.chromium.org/557953007/diff/20001/src/IceTargetLoweringX86...
File src/IceTargetLoweringX8632.cpp (right):

https://codereview.chromium.org/557953007/diff/20001/src/IceTargetLoweringX86...
src/IceTargetLoweringX8632.cpp:3590: !Var2->getIsMultidef() && true) {
Is there a reason that only Var2 is checked for getIsMultidef()?

[Jim Stichnoth, 2014-09-15 17:26:54]

https://codereview.chromium.org/557953007/diff/20001/src/IceTargetLoweringX86...
File src/IceTargetLoweringX8632.cpp (right):

https://codereview.chromium.org/557953007/diff/20001/src/IceTargetLoweringX86...
src/IceTargetLoweringX8632.cpp:3590: !Var2->getIsMultidef() && true) {
On 2014/09/15 16:37:04, jvoung wrote:
> Is there a reason that only Var2 is checked for getIsMultidef()?

Nice catch, thanks!  Done.

[jvoung (off chromium), 2014-09-15 17:34:01]

lgtm

[Jim Stichnoth, 2014-09-15 17:42:22]

Committed patchset #3 (id:40001) manually as e5ac7db (tree was closed).