https://codereview.chromium.org/2771563002/ uploaded fine (I had to fix one ubsan test under windows). I'm waiting for help from goma team for uploading the binaries to goma.

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/3b3794c961885d1c714213f751846e4539f04042

commit 3b3794c961885d1c714213f751846e4539f04042
Author: thakis <thakis@chromium.org>
Date: Mon Mar 27 17:00:01 2017

Roll clang 296321:298539.

Ran `tools/clang/scripts/upload_revision.py 298539`.

The roll makes the tsan runtime abort on RTLD_DEEPBIND,
so disable one test under tsan that used that.

BUG= 703833 , 705255 
TBR=hans

Review-Url: https://codereview.chromium.org/2771563002
Cr-Commit-Position: refs/heads/master@{#459802}

[modify] https://crrev.com/3b3794c961885d1c714213f751846e4539f04042/base/native_library_unittest.cc
[modify] https://crrev.com/3b3794c961885d1c714213f751846e4539f04042/tools/clang/scripts/update.py

Yay! Thanks Nico!

Let's hope it sticks, and I will be able to make the first step in getting rid of the Gold plugin: https://codereview.chromium.org/2766333002/#

Anything in this roll that makes former non-throwing "new" throw? We are hitting a few OOMs in V8 that don't properly crash but now bail out in c++. E.g. we now need to fix https://chromium.googlesource.com/v8/v8/+/19f626f076cc/src/allocation.h#46 and explicitly add new(std::nothrow)

On Linux?

Yes, but it doesn't seem so. At least I get a repro of  http://crbug.com/705962  also before this clang roll. Just tried locally. So, maybe, never mind :(

Declaring this a fixed. Objections?