This roll will cross the llvm 8 -> 9 version bump, so the lines touched by https://chromium-review.googlesource.com/c/1412828 will need updating when rolling.

Also, any upstream issues discovered while rolling may be relevant for the upstream release too. Please note such issues on https://llvm.org/pr40331

We should start passing -fsplit-lto-unit after this roll. Without this we will start generating slightly worse code for -fwhole-program-vtables once https://reviews.llvm.org/D54815 lands.

r351457 for  bug 829078 .

Kicked off packages at r351476: https://chromium-review.googlesource.com/c/chromium/src/+/1418440

It sounds like the roll will need manual changes to deal with 9.0.0 and -fsplit-lto-unit.

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/d8927fce44de956fae4c310fa13d36be0cfc7378

commit d8927fce44de956fae4c310fa13d36be0cfc7378
Author: Reid Kleckner <rnk@google.com>
Date: Sat Jan 19 00:00:41 2019

Roll clang 350768:351477.

Update VERSION from 8.0.0 to 9.0.0 in update.py and toolchain.gni.

Add -fsplit-lto-unit to account for a change in behavior in upstream
ThinLTO.

Bug:  922384 , 922386
Change-Id: Ic9dbf0a9f4d8c50756b8bdb01c75782ba387bdca
Reviewed-on: https://chromium-review.googlesource.com/c/1418440
Commit-Queue: Nico Weber <thakis@chromium.org>
Reviewed-by: Nico Weber <thakis@chromium.org>
Reviewed-by: Peter Collingbourne <pcc@chromium.org>
Cr-Commit-Position: refs/heads/master@{#624366}
[modify] https://crrev.com/d8927fce44de956fae4c310fa13d36be0cfc7378/build/config/compiler/BUILD.gn
[modify] https://crrev.com/d8927fce44de956fae4c310fa13d36be0cfc7378/build/toolchain/toolchain.gni
[modify] https://crrev.com/d8927fce44de956fae4c310fa13d36be0cfc7378/tools/clang/scripts/update.py

This stuck over the weekend, but it almost got reverted because -fsplit-lto-unit isn't supported by the version of clang that chromeos uses.

Next roll:  https://crbug.com/922384 

I suspect that there are no full chromeos bots in the upstream chromium CQ for good reasons, but it would be good if we could have a bot that builds just Chrome with whatever compiler version chromeos uses. Otherwise, it's too easy to add new cflags to the chromium build that aren't supported by the chromeos toolchain. I guess I'll open a bug for it.

Marking fixed, next roll:  https://crbug.com/922384 

This was the near-miss revert:
https://chromium-review.googlesource.com/c/chromium/src/+/1424477
And the associated bug:  https://crbug.com/923750 

New bug for maybe adding chromeos pfq CQ coverage: https://crbug.com/924192

Minor correction: next roll is https://crbug.com/924185