Subzero: Use cvttss2si and similar instead of cvtss2si for fp->int casts.

Subzero: Use cvttss2si and similar instead of cvtss2si for fp->int casts.

This is the truncating cvt instruction instead of rounding.

A few interesting floating point inputs are added to the cross tests.

Also, the cross test error output is modified to be more clear.

BUG= none
R=jvoung@chromium.org

Committed: https://gerrit.chromium.org/gerrit/gitweb?p=native_client/pnacl-subzero.git;a=commit;h=b63cd88

[Jim Stichnoth, 2014-09-06 21:22:43]

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

[jvoung (off chromium), 2014-09-08 17:27:05]

LGTM

https://codereview.chromium.org/550723002/diff/20001/src/IceTargetLoweringX86...
File src/IceTargetLoweringX8632.cpp (right):

https://codereview.chromium.org/550723002/diff/20001/src/IceTargetLoweringX86...
src/IceTargetLoweringX8632.cpp:2060: // TODO(stichnot): _cvt or _cvtt?
seems like it should be cvtt to match the scalar behavior?

[Jim Stichnoth, 2014-09-08 17:47:06]

https://codereview.chromium.org/550723002/diff/20001/src/IceTargetLoweringX86...
File src/IceTargetLoweringX8632.cpp (right):

https://codereview.chromium.org/550723002/diff/20001/src/IceTargetLoweringX86...
src/IceTargetLoweringX8632.cpp:2060: // TODO(stichnot): _cvt or _cvtt?
On 2014/09/08 17:27:05, jvoung wrote:
> seems like it should be cvtt to match the scalar behavior?

You're probably right.  I wasn't sure since there were no vector cross test
failures after adding fractional values to the tests.  But on closer inspection,
it looks like there are no vector cross tests for cast operations. :(

[Jim Stichnoth, 2014-09-08 17:47:30]

Committed patchset #3 (id:40001) manually as b63cd88 (presubmit successful).