Fix JSONStringEscapeTest.EscapeBytes on iOS.

Fix JSONStringEscapeTest.EscapeBytes on iOS.

Clang has an apparent bug on arm64 which causes it to replace elements of the
|cases| array with garbage.  Removing \0 from the end of the string constant
works around this bug.  Tracked in https://llvm.org/bugs/show_bug.cgi?id=23437.

BUG=477682
TEST=Arm64 iOS bots go green.

Committed: https://crrev.com/bf21714b4e5432f50f1ca9d7190109e7d322e917
Cr-Commit-Position: refs/heads/master@{#329294}

[rohitrao (ping after 24h), 2015-05-08 11:36:23]

rohitrao@chromium.org changed reviewers:
+ rsesek@chromium.org

[rohitrao (ping after 24h), 2015-05-08 11:36:23]

Robert, I think you wrote this test initially.

The trailing \0 in the string constant is causing clang to output invalid arm64
code.  Removing that allows the test to pass again, but I don't know if that
trailing \0 is important to the test or not.

If it is important, I could add a kTrailingNull constant similar to kEmbedNull
and continue running the test case that way.  What do you think?

Thanks!

[Robert Sesek, 2015-05-11 22:48:49]

I can't remember if/why it is significant to have a trailing \0 in the string,
especially given the embedded \0 test below. LGTM

[rohitrao (ping after 24h), 2015-05-11 22:50:56]

The CQ bit was checked by rohitrao@chromium.org

[commit-bot: I haz the power, 2015-05-11 22:53:29]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1132063004/1

[commit-bot: I haz the power, 2015-05-11 23:03:45]

The CQ bit was unchecked by commit-bot@chromium.org

[commit-bot: I haz the power, 2015-05-11 23:03:46]

Try jobs failed on following builders:
  chromium_presubmit on tryserver.chromium.linux (JOB_FAILED,
http://build.chromium.org/p/tryserver.chromium.linux/builders/chromium_presub...)

[rohitrao (ping after 24h), 2015-05-11 23:07:02]

rohitrao@chromium.org changed reviewers:
+ thakis@chromium.org

[rohitrao (ping after 24h), 2015-05-11 23:07:03]

+thakis for OWNERS because he was helping me look at this last week.  Do you
have any objections to working around the bug while we wait for a clang fix?

[Nico, 2015-05-12 00:08:37]

lgtm

(I kind of expect this to be a problem in other places too :-/)

[rohitrao (ping after 24h), 2015-05-12 00:16:06]

> (I kind of expect this to be a problem in other places too :-/)

Yeah, for example base/strings/string_number_conversions_unittest.cc also has
strings with embedded nuls.  I may take a look to see why those aren't failing.

[rohitrao (ping after 24h), 2015-05-12 00:16:15]

The CQ bit was checked by rohitrao@chromium.org

[commit-bot: I haz the power, 2015-05-12 00:19:28]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1132063004/1

[commit-bot: I haz the power, 2015-05-12 00:26:59]

Committed patchset #1 (id:1)

[commit-bot: I haz the power, 2015-05-12 00:28:35]

Patchset 1 (id:??) landed as
https://crrev.com/bf21714b4e5432f50f1ca9d7190109e7d322e917
Cr-Commit-Position: refs/heads/master@{#329294}