diff options
| author | jchaffraix <jchaffraix@chromium.org> | 2016-01-12 08:55:53 -0800 |
|---|---|---|
| committer | Commit bot <commit-bot@chromium.org> | 2016-01-12 16:56:34 +0000 |
| commit | a622d4a967550c211ef7ff63008dd9f83c31acad (patch) | |
| tree | 9520bd79242ed551f7ca97efbd5d3a489b0e05ca /tools/clang | |
| parent | be2698cbeb075f3465a02f8ddb91367ef0ccd92f (diff) | |
| download | chromium_src-a622d4a967550c211ef7ff63008dd9f83c31acad.zip chromium_src-a622d4a967550c211ef7ff63008dd9f83c31acad.tar.gz chromium_src-a622d4a967550c211ef7ff63008dd9f83c31acad.tar.bz2 | |
Revert of roll clang 255169:256316 (patchset #10 id:180001 of https://codereview.chromium.org/1539373005/ )
Reason for revert:
The Mac OS 10.7 Debug Blink bot is hitting the regression mentioned in the CL while running several Blink unittests:
dyld: lazy symbol binding failed: Symbol not found: __ZNSt3__114basic_iostreamIcNS_11char_traitsIcEEEC2EPNS_15basic_streambufIcS2_EE
Referenced from: /Volumes/data/b/swarm_slave/work/isolated/runImpjEb/third_party/WebKit/Source/wtf/../../../../out/Debug/wtf_unittests
Expected in: /usr/lib/libc++.1.dylib
dyld: Symbol not found: __ZNSt3__114basic_iostreamIcNS_11char_traitsIcEEEC2EPNS_15basic_streambufIcS2_EE
Referenced from: /Volumes/data/b/swarm_slave/work/isolated/runImpjEb/third_party/WebKit/Source/wtf/../../../../out/Debug/wtf_unittests
Expected in: /usr/lib/libc++.1.dylib
Full output:
https://build.chromium.org/p/chromium.webkit/builders/WebKit%20Mac10.7%20%28dbg%29/builds/26318/steps/steps/logs/stdio
Original issue's description:
> roll clang 255169:256316
>
> New:
> * /W4 in clang-cl enables -Wextra warnings
> * fix for link.exe /incremental build flakes
> * /Brepro suppresses timestamps for deterministic builds
> (doesn't work with link.exe /incremental)
>
> Note that this brings in a libc++ regression on Mac [1]: some of the
> symbols that are hidden symbols in the system libc++ in OS X 10.8
> are marked as "inline" in the headers shipped with this compiler --
> this means that binaries produced by this compiler might not run
> on OS X 10.8 if these calls don't get inlined (or, if the 10.8 SDK is
> used, they won't link). That's ok for us though because production
> builds on OS X statically link a newer libc++ with these symbols
> exported, the iOS bots run newer OS X versions than 10.8, and
> so do the Mac bots that do component builds (which use the
> system libc++) or asan builds. And soon we'll roll in r257193 which
> fixes this again.
>
> 1: http://lists.llvm.org/pipermail/cfe-commits/Week-of-Mon-20160104/146428.html
>
> BUG=571691
> TBR=hans
>
> Committed: https://crrev.com/483df9304bb678dd50b435a41c063a3aed96d10f
> Cr-Commit-Position: refs/heads/master@{#368796}
TBR=hans@chromium.org,thakis@chromium.org
# Skipping CQ checks because original CL landed less than 1 days ago.
NOPRESUBMIT=true
NOTREECHECKS=true
NOTRY=true
BUG=571691
Review URL: https://codereview.chromium.org/1575393003
Cr-Commit-Position: refs/heads/master@{#368900}
Diffstat (limited to 'tools/clang')
| -rwxr-xr-x | tools/clang/scripts/update.py | 2 |
1 files changed, 1 insertions, 1 deletions
diff --git a/tools/clang/scripts/update.py b/tools/clang/scripts/update.py index 75b6d29..f784c62 100755 --- a/tools/clang/scripts/update.py +++ b/tools/clang/scripts/update.py @@ -26,7 +26,7 @@ import zipfile # Do NOT CHANGE this if you don't know what you're doing -- see # https://code.google.com/p/chromium/wiki/UpdatingClang # Reverting problematic clang rolls is safe, though. -CLANG_REVISION = '256316' +CLANG_REVISION = '255169' use_head_revision = 'LLVM_FORCE_HEAD_REVISION' in os.environ if use_head_revision: |