> r311217 for  bug 769759 
That's not the one.

What we need is a fix for https://bugs.llvm.org/show_bug.cgi?id=34769

Looks like the error goes away with r313905.

want r314114 for  bug 753736 

r314588 is nice for gold perf

We'll need a fix for https://bugs.llvm.org/show_bug.cgi?id=34811 in order to unbreak CFI.

need r314807 for  issue 771196 

r315025 for  Bug 771710  (but next roll would be fine too)

r315119 for  issue 772559 .

r315330 for  Bug 771710  (aim for r315332+ due to build breakages)

r315374 for a failing ubsan test on windows

Trying r315377: https://chromium-review.googlesource.com/c/chromium/src/+/710218

goma is having trouble with SetFileInformationByHandle which llvm started using in r315079 (b/67663665)

r315513 (really r315515) for rnk's FPO patch
r315520 for the goma/wine issue

+gbiv

Trying r315520: https://chromium-review.googlesource.com/c/chromium/src/+/714337

rnk said he broke something with his fpo patch and we need at least r315542
i'll kick off a build, but upstream is pretty red right now, so i'm not sure it'll work

also, i'll take r315520 to goma anyway to see if the wine issue is resolved

r315520 still shows the mysterious "cannot open input file" error. Not sure what's going on there.

And the win dbg build is broken by rnk's change, e.g. https://build.chromium.org/p/chromium.clang/builders/ToTWin%28dbg%29/builds/12

I'll keep trying tomorrow.

vtsyrklevich: Is there a particular revision you're waiting for?

The goma problem is resolved. rnk landed an FPO fix this morning, and I'm giving 315613 a try: https://chromium-review.googlesource.com/c/chromium/src/+/714337/6

r314171 for  bug 772906 

> I'm giving 315613 a try: https://chromium-review.googlesource.com/c/chromium/src/+/714337/6

Pushed to goma and running tryjobs now. Fingers crossed.

The roll is looking promising. I'm just trying to figure out exactly what's happening with the android_n5x_swarming_rel trybot.

The android_n5x_swarming_rel static initializer count difference reproduces locally. With the roll:

$ python ../../build/android/gyp/assert_static_initializers.py --expected-count=6 --tool-prefix ../../third_party/android_tools/ndk/toolchains/aarch64-linux-android-4.9/prebuilt/linux-x86_64/bin/aarch64-linux-android- --touch gen/chrome/android/monochrome_static_initializers.stamp apks/MonochromePublic.apk
Expected 6 static initializers, but found 4.
Dumping static initializers via dump-static-initializers.py:
# SkCurveMeasure.cpp guard variable for GrGLImprovedPerlinNoise::emitCode(GrGLSLFragmentProcessor::EmitArgs&)::lerpArgs
# SkCurveMeasure.cpp content::protocol::(anonymous namespace)::kDefaultScreenshotQuality+0x42c
# aw_contents.cc blink::g_languages+0xcc0
# aw_contents.cc std::__ndk1::basic_string<char, std::__ndk1::char_traits<char>, pool_allocator<char> >::basic_string(pool_allocator<char> const&)
# aw_contents.cc std::__ndk1::ostreambuf_iterator<char, std::__ndk1::char_traits<char> > std::__ndk1::__pad_and_output<char, std::__ndk1::char_traits<char> >(std::__ndk1::ostreambuf_iterator<char, std::__ndk1::char_traits<char> >, char const*, char const*, char const*, std::__ndk1::ios_base&, char)+0x1c
# aw_contents.cc __cxa_atexit@plt [registers a dtor to run at exit]
# iostream.cpp _GLOBAL_OFFSET_TABLE_+0x11ea8
# iostream.cpp _GLOBAL_OFFSET_TABLE_+0x12ea8
# iostream.cpp std::__ndk1::ios_base::Init::Init()
# iostream.cpp __cxa_atexit@plt [registers a dtor to run at exit]


This is what it looks like without the roll:


Expected 6 static initializers, but found 6.
Dumping static initializers via dump-static-initializers.py:
# SkCurveMeasure.cpp GrGLImprovedPerlinNoise::emitCode(GrGLSLFragmentProcessor::EmitArgs&)::gradArgs
# SkCurveMeasure.cpp trstandard+0x90
# aw_contents.cc blink::g_languages+0xbc0
# aw_contents.cc std::__ndk1::basic_string<char, std::__ndk1::char_traits<char>, pool_allocator<char> >::basic_string(pool_allocator<char> const&)
# aw_contents.cc std::__ndk1::ostreambuf_iterator<char, std::__ndk1::char_traits<char> > std::__ndk1::__pad_and_output<char, std::__ndk1::char_traits<char> >(std::__ndk1::ostreambuf_iterator<char, std::__ndk1::char_traits<char> >, char const*, char const*, char const*, std::__ndk1::ios_base&, char)+0x9c
# aw_contents.cc __cxa_atexit@plt [registers a dtor to run at exit]
# bbr_sender.cc icu_59::gPower10+0x2b0
# bbr_sender.cc net::HostCache::Set(net::HostCache::Key const&, net::HostCache::Entry const&, base::TimeTicks, base::TimeDelta)::trace_event_unique_atomic210
# iostream.cpp _GLOBAL_OFFSET_TABLE_+0x11ea8
# iostream.cpp _GLOBAL_OFFSET_TABLE_+0x12ea8
# iostream.cpp std::__ndk1::ios_base::Init::Init()
# iostream.cpp __cxa_atexit@plt [registers a dtor to run at exit]
# pacing_sender.cc icu_59::gPower10+0x2b0
# pacing_sender.cc net::HostCache::Set(net::HostCache::Key const&, net::HostCache::Entry const&, base::TimeTicks, base::TimeDelta)::trace_event_unique_atomic210


Maybe some inlining decision changed?

So it reduces the number of static initializers by 2? Sounds great, ship it :-)

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/8316001e3f423f6ca2e7a9a7258c22944d16ae13

commit 8316001e3f423f6ca2e7a9a7258c22944d16ae13
Author: Hans Wennborg <hans@chromium.org>
Date: Fri Oct 13 20:32:51 2017

Roll Clang 313786:315613

Bug:  769761 
Change-Id: I3da32861afe5912be8478ade1bf56fe207aa603f
Reviewed-on: https://chromium-review.googlesource.com/714337
Reviewed-by: Nico Weber <thakis@chromium.org>
Commit-Queue: Hans Wennborg <hans@chromium.org>
Cr-Commit-Position: refs/heads/master@{#508804}
[modify] https://crrev.com/8316001e3f423f6ca2e7a9a7258c22944d16ae13/chrome/android/static_initializers.gni
[modify] https://crrev.com/8316001e3f423f6ca2e7a9a7258c22944d16ae13/tools/clang/scripts/update.py

The roll is in, but there's some fall-out:

b/67839378 (not a problem)
 Bug 774973  (probably not revert-worthy)
Bug 774742 (pretty bad)

Landed https://chromium-review.googlesource.com/c/chromium/src/+/721262 for Bug 774742, so the roll is still sticking.

Next roll is tracked by  Bug 775171 

The roll appears to be sticking.