Don't always bail out of GenerateTypedArrayInitialize

Don't always bail out of GenerateTypedArrayInitialize

In an innocuous cleanup patch, https://code.google.com/p/v8/source/detail?r=21621,
some boolean logic got mixed up and the Hydrogen code for TypedArrayInitialize
was changed to always bail out. This swaps the logic appropriately.

Committed: https://crrev.com/907f0b6ce573333f1cbca00db5cb1df1ead175bb
Cr-Commit-Position: refs/heads/master@{#26643}

[adamk, 2015-02-13 19:52:00]

adamk@chromium.org changed reviewers:
+ dslomov@chromium.org

[adamk, 2015-02-13 19:52:00]

Do we have perf tests that should have caught this?

[Dmitry Lomov (no reviews), 2015-02-13 20:10:12]

Wow. LGTM.

On 2015/02/13 19:52:00, adamk wrote:
> Do we have perf tests that should have caught this?

I am pretty sure some Octane tests took a hit, but nobody noticed in the noise
:(
We should have better direct coverage for typed arrays. It is on my radar for
when I will work on fighting typed array polymorphism and generally doing a
second round of optimizations for them.
We were not in the habit of micro-perftesting at the time this change happened,
but now thanks to much better perf infrastructure we can address this.

[adamk, 2015-02-13 20:11:35]

The CQ bit was checked by adamk@chromium.org

[commit-bot: I haz the power, 2015-02-13 20:12:19]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/922393003/1

[adamk, 2015-02-13 20:13:06]

For this particular problem, we also could have detected that this code path was
never supposed to be hit during "normal" operation (there should be a
CHECK(!fuzzing) inside that bailout).

[Dmitry Lomov (no reviews), 2015-02-13 20:18:28]

On 2015/02/13 20:13:06, adamk wrote:
> For this particular problem, we also could have detected that this code path
was
> never supposed to be hit during "normal" operation (there should be a
> CHECK(!fuzzing) inside that bailout).

Except that there is no "fuzzing" flag. Actually on second thought, I do not
think we actually fuzz builtins in that way anymore, so I think it will actually
be safe to replace that 'if' with a CHECK.

[commit-bot: I haz the power, 2015-02-13 20:50:28]

Committed patchset #1 (id:1)

[commit-bot: I haz the power, 2015-02-13 20:50:41]

Patchset 1 (id:??) landed as
https://crrev.com/907f0b6ce573333f1cbca00db5cb1df1ead175bb
Cr-Commit-Position: refs/heads/master@{#26643}

[Michael Achenbach, 2015-02-15 10:13:55]

A revert of this CL (patchset #1 id:1) has been created in
https://codereview.chromium.org/929773002/ by machenbach@chromium.org.

The reason for reverting is: Speculative revert. This CL seems to break octane..