Added %_Likely/%_Unlikely intrinsics (special cases of GCC's __builin_expect).

Added %_Likely/%_Unlikely intrinsics (special cases of GCC's __builin_expect).

Currently this only sets branch hints, so we get unlikely code "out of
the way", but in the long run the register allocator needs some love
to treat the unlikely code as, well, unlikely. :-)

Committed: https://crrev.com/677f3d5fd79f2f8804bb8e0ba7f2bbea698c0cd8
Cr-Commit-Position: refs/heads/master@{#27540}

[Sven Panne, 2015-03-31 11:55:22]

svenpanne@chromium.org changed reviewers:
+ mstarzinger@chromium.org

[Michael Starzinger, 2015-03-31 12:49:03]

LGTM (modulo Win failures as discussed offline).

https://codereview.chromium.org/1048063002/diff/1/src/hydrogen.cc
File src/hydrogen.cc (right):

https://codereview.chromium.org/1048063002/diff/1/src/hydrogen.cc#newcode12049
src/hydrogen.cc:12049: CHECK_BAILOUT(Visit(call->arguments()->at(0)));
nit: Not that it matters much, but I don't think we need the CHECK_BAILOUT at
tail positions.

https://codereview.chromium.org/1048063002/diff/1/test/unittests/compiler/js-...
File test/unittests/compiler/js-intrinsic-lowering-unittest.cc (right):

https://codereview.chromium.org/1048063002/diff/1/test/unittests/compiler/js-...
test/unittests/compiler/js-intrinsic-lowering-unittest.cc:263:
ASSERT_TRUE(BranchHint::kNone == BranchHintOf(d.branch->op()));
nit: ASSERT_EQ

https://codereview.chromium.org/1048063002/diff/1/test/unittests/compiler/js-...
test/unittests/compiler/js-intrinsic-lowering-unittest.cc:266:
ASSERT_TRUE(input->op()->Equals(r.replacement()->op()));
nit: EXPECT_THAT(r.replacement(), input)

https://codereview.chromium.org/1048063002/diff/1/test/unittests/compiler/js-...
test/unittests/compiler/js-intrinsic-lowering-unittest.cc:267:
ASSERT_TRUE(BranchHint::kTrue == BranchHintOf(d.branch->op()));
nit: EXPECT_EQ

https://codereview.chromium.org/1048063002/diff/1/test/unittests/compiler/js-...
test/unittests/compiler/js-intrinsic-lowering-unittest.cc:357:
ASSERT_TRUE(BranchHint::kNone == BranchHintOf(d.branch->op()));
nit: Same as the three comments above.

[Sven Panne, 2015-03-31 13:01:35]

Addressed feedback. Changed the runtime function, hopefully the LTO on Windows
is not too "clever" again... :-/

https://codereview.chromium.org/1048063002/diff/1/src/hydrogen.cc
File src/hydrogen.cc (right):

https://codereview.chromium.org/1048063002/diff/1/src/hydrogen.cc#newcode12049
src/hydrogen.cc:12049: CHECK_BAILOUT(Visit(call->arguments()->at(0)));
On 2015/03/31 12:49:03, Michael Starzinger wrote:
> nit: Not that it matters much, but I don't think we need the CHECK_BAILOUT at
> tail positions.

Removed.

https://codereview.chromium.org/1048063002/diff/1/test/unittests/compiler/js-...
File test/unittests/compiler/js-intrinsic-lowering-unittest.cc (right):

https://codereview.chromium.org/1048063002/diff/1/test/unittests/compiler/js-...
test/unittests/compiler/js-intrinsic-lowering-unittest.cc:263:
ASSERT_TRUE(BranchHint::kNone == BranchHintOf(d.branch->op()));
On 2015/03/31 12:49:03, Michael Starzinger wrote:
> nit: ASSERT_EQ

Done.

https://codereview.chromium.org/1048063002/diff/1/test/unittests/compiler/js-...
test/unittests/compiler/js-intrinsic-lowering-unittest.cc:266:
ASSERT_TRUE(input->op()->Equals(r.replacement()->op()));
On 2015/03/31 12:49:03, Michael Starzinger wrote:
> nit: EXPECT_THAT(r.replacement(), input)

Done.

https://codereview.chromium.org/1048063002/diff/1/test/unittests/compiler/js-...
test/unittests/compiler/js-intrinsic-lowering-unittest.cc:267:
ASSERT_TRUE(BranchHint::kTrue == BranchHintOf(d.branch->op()));
On 2015/03/31 12:49:03, Michael Starzinger wrote:
> nit: EXPECT_EQ

Done.

[Sven Panne, 2015-03-31 13:19:20]

The CQ bit was checked by svenpanne@chromium.org

[Sven Panne, 2015-03-31 13:19:21]

The patchset sent to the CQ was uploaded after l-g-t-m from
mstarzinger@chromium.org
Link to the patchset: https://codereview.chromium.org/1048063002/#ps20001
(title: "Addressed feedback. Win fix.")

[commit-bot: I haz the power, 2015-03-31 13:19:37]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1048063002/20001

[commit-bot: I haz the power, 2015-03-31 13:23:12]

Committed patchset #2 (id:20001)

[commit-bot: I haz the power, 2015-03-31 13:23:29]

Patchset 2 (id:??) landed as
https://crrev.com/677f3d5fd79f2f8804bb8e0ba7f2bbea698c0cd8
Cr-Commit-Position: refs/heads/master@{#27540}