[compiler-rt] r231367 - tsan: fix signal handling during stop-the-world

Galina Kistanova gkistanova at accesssoftek.com
Tue Mar 24 15:26:08 PDT 2015


I reduced number of threads for build to 32 and rebooted the box. Hope this will help.

Thanks

Galina


________________________________
From: Kostya Serebryany [mailto:kcc at google.com]
Sent: Tuesday, March 24, 2015 1:00 PM
To: Galina Kistanova
Cc: Hans Wennborg; David Blaikie; llvm-commits at cs.uiuc.edu
Subject: Re: [compiler-rt] r231367 - tsan: fix signal handling during stop-the-world

The bot is very heavily loaded, it has 40+ processes running since March 1:

41500 buildsl+  20   0 1208380 1.040g   1044 R 100.0  2.2  17567:12 clang-3.7
41534 buildsl+  20   0  347420 246492    976 R 100.0  0.5  18093:57 clang-3.7
41438 buildsl+  20   0  535216 427720   1020 R 100.0  0.9  17862:52 clang-3.7
41364 buildsl+  20   0  538920 427928   1132 R  99.6  0.9  17835:58 clang-3.7
41400 buildsl+  20   0  481600 374952   1044 R  99.0  0.8  17961:05 clang-3.7

buildsl+ 41534 55.0  0.4 347420 246500 ?       R    Mar01 18094:29 /home/buildslave/slave_as-bldslv8/clang-x86_64-linux-selfhost-abi-test/stage1.install/bin/clang-3.7 -cc1 -triple x86_64-unknown-linux-gnu -emit-obj -disable-free -main-file-name Errno.cpp -mrelocation-model pic -pic-level 2 -mthread-model posix -fmath-errno -masm-verbose -mconstructor-aliases -munwind-tables -fuse-init-array -target-cpu x86-64 -momit-leaf-frame-pointer -dwarf-column-info -ffunction-sections -fdata-sections -coverage-file /home/buildslave/slave_as-bldslv8/clang-x86_64-linux-selfhost-abi-test/stage2/lib/Support/CMakeFiles/LLVMSupport.dir/Errno.cpp.o -resource-dir /home/buildslave/slave_as-bldslv8/clang-x86_64-linux-selfhost-abi-test/stage1.install/bin/../lib/clang/3.7.0 -dependency-file lib/Support/CMakeFiles/LLVMSupport.dir/Errno.cpp.o.d -MT lib/Support/CMakeFiles/LLVMSupport.dir/Errno.cpp.o -D GTEST_HAS_RTTI=0 -D _DEBUG -D _GNU_SOURCE -D __STDC_CONSTANT_MACROS -D __STDC_FORMAT_MACROS -D __STDC_LIMIT_MACROS -U NDEBUG -I lib/Support -I /home/buildslave/slave_as-bldslv8/clang-x86_64-linux-selfhost-abi-test/llvm/lib/Support -I include -I /home/buildslave/slave_as-bldslv8/clang-x86_64-linux-selfhost-abi-test/llvm/include -internal-isystem /usr/lib/gcc/x86_64-linux-gnu/4.8/../../../../include/c++/4.8 -internal-isystem /usr/lib/gcc/x86_64-linux-gnu/4.8/../../../../include/x86_64-linux-gnu/c++/4.8 -internal-isystem /usr/lib/gcc/x86_64-linux-gnu/4.8/../../../../include/x86_64-linux-gnu/c++/4.8 -internal-isystem /usr/lib/gcc/x86_64-linux-gnu/4.8/../../../../include/c++/4.8/backward -internal-isystem /usr/local/include -internal-isystem /home/buildslave/slave_as-bldslv8/clang-x86_64-linux-selfhost-abi-test/stage1.install/bin/../lib/clang/3.7.0/include -internal-externc-isystem /usr/include/x86_64-linux-gnu -internal-externc-isystem /include -internal-externc-isystem /usr/include -O3 -Wall -W -Wno-unused-parameter -Wwrite-strings -Wcast-qual -Wmissing-field-initializers -Wno-long-long -Wcovered-switch-default -Wnon-virtual-dtor -pedantic -std=c++11 -fdeprecated-macro -fdebug-compilation-dir /home/buildslave/slave_as-bldslv8/clang-x86_64-linux-selfhost-abi-test/stage2 -ferror-limit 19 -fmessage-length 0 -fvisibility-inlines-hidden -mstackrealign -fno-rtti -fobjc-runtime=gcc -fdiagnostics-show-option -fcolor-diagnostics -vectorize-loops -vectorize-slp -o lib/Support/CMakeFiles/LLVMSupport.dir/Errno.cpp.o -x c++ /home/buildslave/slave_as-bldslv8/clang-x86_64-linux-selfhost-abi-test/llvm/lib/Support/Errno.cpp

I would expect that rebooting the bot will resolve the flakiness.

On Mon, Mar 23, 2015 at 1:26 PM, Kostya Serebryany <kcc at google.com<mailto:kcc at google.com>> wrote:


On Fri, Mar 20, 2015 at 9:59 AM, Galina Kistanova <gkistanova at accesssoftek.com<mailto:gkistanova at accesssoftek.com>> wrote:
Sure I can set you up with the access. That's one of the purposes of having the bots in the lab.

Just tell me what you are going to do there, so I could figure out the permissions, and when you will be accessing it, so I would stop the builder.

I will need the machine to build and run a single C++ test,
so all I need is the access to the clang binary built on this builder
and an ability to compile a test  using that clang.
No need to stop the builder.

Please also send me directly the ssh key you will be using.

ok!


Thank you

Galina


-----Original Message-----
From: hwennborg at google.com<mailto:hwennborg at google.com> [mailto:hwennborg at google.com<mailto:hwennborg at google.com>] On Behalf Of Hans Wennborg
Sent: Thursday, March 19, 2015 1:43 PM
To: David Blaikie
Cc: Kostya Serebryany; Galina Kistanova; llvm-commits at cs.uiuc.edu<mailto:llvm-commits at cs.uiuc.edu>
Subject: Re: [compiler-rt] r231367 - tsan: fix signal handling during stop-the-world

I'm hitting this when running 'ninja check-all' on my machine.

If I just run the individual run line, it doesn't crash though.

On Mon, Mar 16, 2015 at 11:37 AM, David Blaikie <dblaikie at gmail.com<mailto:dblaikie at gmail.com>> wrote:
> Not my bot (& I seem to have lost my local reproduction - it was failing
> pretty consistently for me last week - perhaps my machine instnalled a
> software update in the interim), +Galina who might have some more info about
> as-bldslv8.
>
> On Mon, Mar 16, 2015 at 11:34 AM, Kostya Serebryany <kcc at google.com<mailto:kcc at google.com>> wrote:
>>
>>
>>
>> On Sun, Mar 15, 2015 at 12:34 AM, Dmitry Vyukov <dvyukov at google.com<mailto:dvyukov at google.com>>
>> wrote:
>>>
>>> On Sun, Mar 15, 2015 at 12:32 AM, David Blaikie <dblaikie at gmail.com<mailto:dblaikie at gmail.com>>
>>> wrote:
>>> >> Added:
>>> >>     compiler-rt/trunk/test/asan/TestCases/Linux/leak_check_segv.cc
>>> >>
>>> >>
>>> >> compiler-rt/trunk/test/asan/TestCases/Linux/signal_during_stop_the_world.cc
>>> >>
>>> >>
>>> >> compiler-rt/trunk/test/sanitizer_common/TestCases/Linux/signal_segv_handler.cc
>>> >
>>> >
>>> > This test appears to be flaky. Failing every second run here:
>>> >
>>> > http://lab.llvm.org:8011/builders/clang-x86_64-linux-selfhost-abi-test/builds/2871
>>> > and failing for me locally.
>>>
>>>
>>> The test does not fail for a bunch of us locally (I run it 5000 times
>>> locally).
>>> Can you please run it locally _without_ any tool? Does it fail?
>>
>>
>> Yea, we tried reproducing it on multiple machines with no luck.
>> Can we have access to the bot?


-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20150324/25a92df4/attachment.html>


More information about the llvm-commits mailing list