♟ seurer

FWIW it looks good to me

As examples on a 4.4 kernel memory was very roughly in this range (sampled from many test runs):
0x0000000100000984 to 0x000000013FFF0984 which is a range of about 3FFF0000

It is not specific to Ubuntu but is changes in how wide a range of addresses that the kernel chooses when ASLR is enabled. After some updates in the 4.x kernels (sorry, I am not sure which ones) the range is hundreds of times bigger which makes choosing the constants that the sanitizers use impossible.

TSAN/MSAN: Ahh, got it.

The similar change for aarch64 was done in

[PowerPC][asan] Fix asan tests to handle changed memory layouts

This patch fails on powerpc64

[PowerPC][asan] Update asan to handle changed memory layouts in newer kernels

[PowerPC][tsan] Update tsan to handle changed memory layouts in newer kernels

[PowerPC][msan] Update msan to handle changed memory layouts in newer kernels

[PowerPC][tsan] Update tsan to handle changed memory layouts in newer kernels

Code checked in as r317802.

[PowerPC][msan] Update msan to handle changed memory layouts in newer kernels

I tried this patch on powerpc64 (BE) and it fixed the problem.

[powerpc] deactivate tests combined.test and object.test on powerpc64 be

[powerpc] deactivate flakey test halt_on_error-torture.cc on powerpc64 be

[PowerPC] Fix test case sem_init_glibc.cc for powerpc64be

[powerpc] Remove XFAIL for sanitizer_coverage_no_prune.cc on powerpc64

[powerpc] Mark coverage-sample.cc as XFAIL on powerpc64le

[powerpc] Reactivate ppc64le lnt buildbot

[powerpc] deactivate flakey tests on powerpc64le

[powerpc] deactivate ManyThreadsTest asan test on powerpc64

[PowerPC] Fix sanitizer frame unwind on 32-bit ABIs

[PowerPC] Disable oggenc test case in testsuite

[PowerPC] mark xray test as unsupported on powerpcle in testsuite

[PowerPC] Disable oggenc test case in testsuite

[DAGCombiner] revert r295336

[powerpc] deactivate user_malloc.cc tsan test on powerpc64le

[powerpc] deactivate ThreadedOneSizeMallocStressTest asan test on powerpc64

[powerpc] deactivate ThreadedMallocStressTest asan test on powerpc64

[powerpc] deactivate readability-identifier-naming.cpp test on powerpc64le

In the new sets of tests the output being looked for is wrong.

This revision causes failures on power for both BE and LE:

[powerpc] Move #if to appease warnings-as-errors tests

[powerpc] Disable ManyThreadsWithStatsStressTest test for ppc64

Looks pretty good. For completeness sake how about adding CHECK_ASAN and CHECK_UBSAN and handling CHECK_LIBCXX and
CHECK_LLD the same way as the others?

[powerpc] Enable msan testing for ppc64le

[powerpc] Disable shared libs for powerpc64be

Yup. I can turn off this bypass easily enough if scudo gets support for
power.

[powerpc] Add ability to bypass scudo tests

[powerpc] mark static_tls.cc test as UNSUPPORTED on powerpc64

I just noticed this problem on powerpc as well:

[powerpc] mark JIT tests as UNSUPPORTED on powerpc64 big endian

[powerpc] mark JIT tests as XFAIL on powerpc64 big endian

These changes are no longer needed.

[powerpc] mark setuid.c and setuid2.c as unsupported

[powerpc] pacify lint for java_race_pc.cc

[powerpc] activate java_race_pc.cc on powerpc64le

[powerpc] reactivate ignore_lib4.cc on powerpc64le

[powerpc] build shared libraries

[powerpc] Add jobs parms to powerpc64 lnt buildbots

[powerpc] Switch powerpc64 lnt buildbots to use cmake/ninja

[powerpc] Alow skipping unsupported sanitizers

[powerpc] make test case as xfail

[powerpc] Correct erroneous builder name

[powerpc] Reorganize buildbots

Galina, once this is approved you will need to warn me when you are going to restart the master so I can create and start the bots with the correct names.

[power] Fix test case target checks

Fix case of path name

Committed revision 256351.

[powerpc64] Updates for powerpc64 buildbots

BTW, the earlier one I changed over worked very well after a few teething issues and runs in a fraction of the time it did previously.

Included in another revision

[powerpc64] Updates for powerpc64 buildbots

Fixed the extra parenthesis.

[power] Fix test case target checks

The use of the flag is causing errors in the sanitizer buildbot runs (for power at least) so assuming this patch fixes that it would be great to get this in ASAP.

Fix test case function name checks

Fix test case label check

[PowerPC]Test case re-activation and deactivation

I tested the --multisample change locally on the running bots and it took 1 hour off the run time of the big endian powerpc64 lnt test bot and 3 hours off the run time of the little endian one.

[PowerPC] Remove sqlite3 and tramp3d-v4 from the test suite for PowerPC

In D13524#262182, @rengolin wrote:

The multisample option is obvious, I'm not sure about the -j8. On all my LNT bots, all cores are busy all the time, so if you have 4 cores, I don't think -j8 will help you. If you have 8 cores, than you should change jobs=8, too.

seurer (Bill Seurer)User

Projects

User Details

Recent ActivityView All

Oct 19 2018

Oct 9 2018

Oct 8 2018

Jan 23 2018

Jan 17 2018

Dec 7 2017

Dec 6 2017

Nov 28 2017

Nov 13 2017

Nov 10 2017

Nov 9 2017

Nov 8 2017

Sep 21 2017

Jun 13 2017

May 25 2017

May 16 2017

May 12 2017

May 8 2017

May 5 2017

May 4 2017

May 1 2017

Mar 27 2017

Mar 20 2017

Mar 16 2017

Mar 10 2017

Mar 7 2017

Mar 1 2017

Feb 22 2017

Feb 2 2017

Jan 25 2017

Jan 24 2017

Dec 13 2016

Nov 14 2016

Sep 8 2016

Sep 7 2016

Jun 22 2016

Jun 21 2016

Jun 12 2016

Jun 8 2016

May 20 2016

May 9 2016

Apr 25 2016

Mar 30 2016

Mar 6 2016

Mar 3 2016

Feb 25 2016

Feb 2 2016

Feb 1 2016

Jan 29 2016

Jan 28 2016

Jan 25 2016

Jan 20 2016

Jan 13 2016

Jan 12 2016

Dec 24 2015

Dec 23 2015

Dec 21 2015

Dec 14 2015

Dec 10 2015

Dec 9 2015

Nov 20 2015

Nov 12 2015

Oct 22 2015

Oct 16 2015

Oct 9 2015

Oct 8 2015

seurer (Bill Seurer)
User

Recent Activity
View All