[compiler-rt] r315777 - [asan] Deflake one test by running it 3 times.

Mon Oct 16 14:42:22 PDT 2017

Could the device be removed from testing? (is it likely the device that's
at fault, more than the test - or is it a truly representative, valid
device where any device in that configuration would have this problem and
it's a supported configuration (not too old, etc)?)

On Mon, Oct 16, 2017 at 2:38 PM Evgenii Stepanov via llvm-commits <
llvm-commits at lists.llvm.org> wrote:

> For some value of x, p^x is low enough :)
>
> Seriously speaking, the test is failing on a particular device and I
> suspect it could be a kernel issue: it is running 3.10, which is quite
> old. I don't have time to debug it, but I could make the test as
> failing on (android && i686) which is not the real cause of the error,
> but it would make the bot green and keep the test exercised on the
> arm-android device.
>
> On Mon, Oct 16, 2017 at 2:09 PM, Rui Ueyama via llvm-commits
> <llvm-commits at lists.llvm.org> wrote:
> > You cannot fix a flakey test by just running it more than once, can you?
> I
> > mean, the probability of test failure decreases from p to p^3, but it is
> > still not zero.
> >
> > On Mon, Oct 16, 2017 at 2:03 PM, Vitaly Buka via llvm-commits
> > <llvm-commits at lists.llvm.org> wrote:
> >>
> >> It does not makes test slower as it passes on all platforms from the
> first
> >> time.
> >>
> >> We don't suspect issues with functionally, rather some specific platform
> >> bug.
> >> Also on Android we know only one device were it fails. And I guess it
> >> fails only in Asan mode?
> >> So if we mark it unsupported, it should be very specific.
> >> E.g. // UNSUPPORTED: android && asan && x86 (not sure if that one is
> 64bit
> >> and how we exactly recognize x86 on android)
> >>
> >> On Mon, Oct 16, 2017 at 1:49 PM, David Blaikie via llvm-commits
> >> <llvm-commits at lists.llvm.org> wrote:
> >>>
> >>> If this functionality isn't reliable on Android, I think
> >>> UNSUPPORTED/XFAIL is more appropriate than deflaking like this. The
> >>> functionality's no good if it fails that much, so why test/support it,
> >>> making tests slower and risk flakiness creeping in on other platforms
> due to
> >>> the more generous testing?
> >>>
> >>> On Fri, Oct 13, 2017 at 4:57 PM Evgeniy Stepanov via llvm-commits
> >>> <llvm-commits at lists.llvm.org> wrote:
> >>>>
> >>>> Author: eugenis
> >>>> Date: Fri Oct 13 16:57:08 2017
> >>>> New Revision: 315777
> >>>>
> >>>> URL: http://llvm.org/viewvc/llvm-project?rev=315777&view=rev
> >>>> Log:
> >>>> [asan] Deflake one test by running it 3 times.
> >>>>
> >>>> The test seems to trigger an android platform bug under load.
> >>>>
> >>>> Modified:
> >>>>
> >>>>
> compiler-rt/trunk/test/sanitizer_common/TestCases/Linux/allow_user_segv.cc
> >>>>
> >>>> Modified:
> >>>>
> compiler-rt/trunk/test/sanitizer_common/TestCases/Linux/allow_user_segv.cc
> >>>> URL:
> >>>>
> http://llvm.org/viewvc/llvm-project/compiler-rt/trunk/test/sanitizer_common/TestCases/Linux/allow_user_segv.cc?rev=315777&r1=315776&r2=315777&view=diff
> >>>>
> >>>>
> ==============================================================================
> >>>> ---
> >>>>
> compiler-rt/trunk/test/sanitizer_common/TestCases/Linux/allow_user_segv.cc
> >>>> (original)
> >>>> +++
> >>>>
> compiler-rt/trunk/test/sanitizer_common/TestCases/Linux/allow_user_segv.cc
> >>>> Fri Oct 13 16:57:08 2017
> >>>> @@ -4,16 +4,34 @@
> >>>>  // clang-format off
> >>>>  // RUN: %clangxx -O0 %s -o %t
> >>>>
> >>>> +// RUN: %env_tool_opts=handle_segv=0 not %run %t 2>&1 | FileCheck %s
> >>>> --check-prefix=CHECK0 || \
> >>>> +// RUN: %env_tool_opts=handle_segv=0 not %run %t 2>&1 | FileCheck %s
> >>>> --check-prefix=CHECK0 || \
> >>>>  // RUN: %env_tool_opts=handle_segv=0 not %run %t 2>&1 | FileCheck %s
> >>>> --check-prefix=CHECK0
> >>>> +// RUN: %env_tool_opts=handle_segv=1 not %run %t 2>&1 | FileCheck %s
> >>>> --check-prefix=CHECK1 || \
> >>>> +// RUN: %env_tool_opts=handle_segv=1 not %run %t 2>&1 | FileCheck %s
> >>>> --check-prefix=CHECK1 || \
> >>>>  // RUN: %env_tool_opts=handle_segv=1 not %run %t 2>&1 | FileCheck %s
> >>>> --check-prefix=CHECK1
> >>>> +// RUN: %env_tool_opts=handle_segv=2 not %run %t 2>&1 | FileCheck %s
> >>>> --check-prefix=CHECK2 || \
> >>>> +// RUN: %env_tool_opts=handle_segv=2 not %run %t 2>&1 | FileCheck %s
> >>>> --check-prefix=CHECK2 || \
> >>>>  // RUN: %env_tool_opts=handle_segv=2 not %run %t 2>&1 | FileCheck %s
> >>>> --check-prefix=CHECK2
> >>>>
> >>>> +// RUN: %env_tool_opts=handle_segv=0:allow_user_segv_handler=0 not
> %run
> >>>> %t 2>&1 | FileCheck %s --check-prefix=CHECK0 || \
> >>>> +// RUN: %env_tool_opts=handle_segv=0:allow_user_segv_handler=0 not
> %run
> >>>> %t 2>&1 | FileCheck %s --check-prefix=CHECK0 || \
> >>>>  // RUN: %env_tool_opts=handle_segv=0:allow_user_segv_handler=0 not
> %run
> >>>> %t 2>&1 | FileCheck %s --check-prefix=CHECK0
> >>>> +// RUN: %env_tool_opts=handle_segv=1:allow_user_segv_handler=0 not
> %run
> >>>> %t 2>&1 | FileCheck %s --check-prefix=CHECK2 || \
> >>>> +// RUN: %env_tool_opts=handle_segv=1:allow_user_segv_handler=0 not
> %run
> >>>> %t 2>&1 | FileCheck %s --check-prefix=CHECK2 || \
> >>>>  // RUN: %env_tool_opts=handle_segv=1:allow_user_segv_handler=0 not
> %run
> >>>> %t 2>&1 | FileCheck %s --check-prefix=CHECK2
> >>>> +// RUN: %env_tool_opts=handle_segv=2:allow_user_segv_handler=0 not
> %run
> >>>> %t 2>&1 | FileCheck %s --check-prefix=CHECK2 || \
> >>>> +// RUN: %env_tool_opts=handle_segv=2:allow_user_segv_handler=0 not
> %run
> >>>> %t 2>&1 | FileCheck %s --check-prefix=CHECK2 || \
> >>>>  // RUN: %env_tool_opts=handle_segv=2:allow_user_segv_handler=0 not
> %run
> >>>> %t 2>&1 | FileCheck %s --check-prefix=CHECK2
> >>>>
> >>>> +// RUN: %env_tool_opts=handle_segv=0:allow_user_segv_handler=1 not
> %run
> >>>> %t 2>&1 | FileCheck %s --check-prefix=CHECK0 || \
> >>>> +// RUN: %env_tool_opts=handle_segv=0:allow_user_segv_handler=1 not
> %run
> >>>> %t 2>&1 | FileCheck %s --check-prefix=CHECK0 || \
> >>>>  // RUN: %env_tool_opts=handle_segv=0:allow_user_segv_handler=1 not
> %run
> >>>> %t 2>&1 | FileCheck %s --check-prefix=CHECK0
> >>>> +// RUN: %env_tool_opts=handle_segv=1:allow_user_segv_handler=1 not
> %run
> >>>> %t 2>&1 | FileCheck %s --check-prefix=CHECK1 || \
> >>>> +// RUN: %env_tool_opts=handle_segv=1:allow_user_segv_handler=1 not
> %run
> >>>> %t 2>&1 | FileCheck %s --check-prefix=CHECK1 || \
> >>>>  // RUN: %env_tool_opts=handle_segv=1:allow_user_segv_handler=1 not
> %run
> >>>> %t 2>&1 | FileCheck %s --check-prefix=CHECK1
> >>>> +// RUN: %env_tool_opts=handle_segv=2:allow_user_segv_handler=1 not
> %run
> >>>> %t 2>&1 | FileCheck %s --check-prefix=CHECK2 || \
> >>>> +// RUN: %env_tool_opts=handle_segv=2:allow_user_segv_handler=1 not
> %run
> >>>> %t 2>&1 | FileCheck %s --check-prefix=CHECK2 || \
> >>>>  // RUN: %env_tool_opts=handle_segv=2:allow_user_segv_handler=1 not
> %run
> >>>> %t 2>&1 | FileCheck %s --check-prefix=CHECK2
> >>>>  // clang-format on
> >>>>
> >>>> @@ -21,9 +39,6 @@
> >>>>  // XFAIL: msan
> >>>>  // XFAIL: tsan
> >>>>
> >>>> -// Flaky errors in debuggerd with "waitpid returned unexpected pid
> (0)"
> >>>> in logcat.
> >>>> -// UNSUPPORTED: android
> >>>> -
> >>>>  #include <signal.h>
> >>>>  #include <stdio.h>
> >>>>  #include <stdlib.h>
> >>>>
> >>>>
> >>>> _______________________________________________
> >>>> llvm-commits mailing list
> >>>> llvm-commits at lists.llvm.org
> >>>> http://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-commits
> >>>
> >>>
> >>> _______________________________________________
> >>> llvm-commits mailing list
> >>> llvm-commits at lists.llvm.org
> >>> http://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-commits
> >>>
> >>
> >>
> >> _______________________________________________
> >> llvm-commits mailing list
> >> llvm-commits at lists.llvm.org
> >> http://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-commits
> >>
> >
> >
> > _______________________________________________
> > llvm-commits mailing list
> > llvm-commits at lists.llvm.org
> > http://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-commits
> >
> _______________________________________________
> llvm-commits mailing list
> llvm-commits at lists.llvm.org
> http://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-commits
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20171016/3463bd2b/attachment.html>