Revert of Roll Clang 216630:217949

Revert of Roll Clang 216630:217949 (patchset #5 id:80001 of https://codereview.chromium.org/587433002/)

Reason for revert:
Reverting due to http://crbug.com/418234. Clang was branching on undefined values.

Original issue's description:
> Roll Clang 216630:217949
>
> Also make the script revert local changes before 'svn co'
> rather than afterwards to avoid merge conflicts if a previously
> patched file changes upstream.
>
> BUG=410810
>
> Committed: https://crrev.com/ba4cf8bb6dc77ea033fe13989eb604d6069daa97
> Cr-Commit-Position: refs/heads/master@{#296870}

TBR=thakis@chromium.org,brettw@chromium.org
NOTREECHECKS=true
NOTRY=true
BUG=410810
Committed: https://crrev.com/ad66334103b9b9a3129e98fc9decbc0c94c9aa11
Cr-Commit-Position: refs/heads/master@{#297073}

[hans, 2014-09-26 23:45:06]

Created Revert of Roll Clang 216630:217949

[commit-bot: I haz the power, 2014-09-26 23:46:18]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/614453002/1

[commit-bot: I haz the power, 2014-09-26 23:46:55]

Committed patchset #1 (id:1) as 3c80ac9fd48da3d11852eecd6999053d6d84c56c

[commit-bot: I haz the power, 2014-09-26 23:48:01]

Patchset 1 (id:??) landed as
https://crrev.com/ad66334103b9b9a3129e98fc9decbc0c94c9aa11
Cr-Commit-Position: refs/heads/master@{#297073}

[Nico, 2014-09-27 05:39:26]

(This looks superficially similar to
https://code.google.com/p/chromium/issues/detail?id=399852#c6 and #c8. I
think this is working as expected.)

On Sat, Sep 27, 2014 at 1:48 AM, <commit-bot@chromium.org> wrote:

> Patchset 1 (id:??) landed as
> https://crrev.com/ad66334103b9b9a3129e98fc9decbc0c94c9aa11
> Cr-Commit-Position: refs/heads/master@{#297073}
>
> https://codereview.chromium.org/614453002/
>

To unsubscribe from this group and stop receiving emails from it, send an email
to chromium-reviews+unsubscribe@chromium.org.

[hans, 2014-09-27 18:27:39]

On 2014/09/27 05:39:26, Nico (away until Oct 27) wrote:
> (This looks superficially similar to
> https://code.google.com/p/chromium/issues/detail?id=399852#c6 and #c8. I
> think this is working as expected.)

Thanks for the link! I hadn't seen this before.

After discussing this with LLVM folks we concluded that the generated code is
fine and we should just add Valgrind suppressions. Since LLVM has been doing
this for a very long time and it's only raised Valgrind reports a couple of
times it doesn't seem like a big problem.

I plan to land the roll again on Monday when I have time to watch it.