[compiler-rt] r298476 - XFAIL CFI stats test while LLD produces corrupt PDBs that confuse DIA

Peter Collingbourne via llvm-commits llvm-commits at lists.llvm.org
Wed Mar 22 14:15:13 PDT 2017


I see. I wonder if it would be better to use /debugpdb to control PDB
creation entirely for now (and remove that flag once we're in better
shape). It looks like right now it just disables the TPI stream.

Peter

On Wed, Mar 22, 2017 at 2:02 PM, Reid Kleckner <rnk at google.com> wrote:

> Yes, https://reviews.llvm.org/D31224 / r298467.
>
> On Wed, Mar 22, 2017 at 1:45 PM, Peter Collingbourne <peter at pcc.me.uk>
> wrote:
>
>> Do you know which change caused this test to start failing?
>>
>> Peter
>>
>> On Wed, Mar 22, 2017 at 1:23 PM, Reid Kleckner <rnk at google.com> wrote:
>>
>>> Both failures should be fixed with the changes to add a /nopdb flag to
>>> LLD and use it in these sanitizer tests:
>>> http://lab.llvm.org:8011/builders/sanitizer-windows/builds/8459
>>>
>>> In the future, we probably want to disentagle /debug from symbol table
>>> generation so that /debug can be more faithful to link.exe's command line
>>> interface.
>>>
>>> On Tue, Mar 21, 2017 at 10:36 PM, Vitaly Buka <vitalybuka at google.com>
>>> wrote:
>>>
>>>> Now it fails another test.
>>>> http://lab.llvm.org:8011/builders/sanitizer-windows/builds/8
>>>> 411/steps/run%20tests/logs/stdio
>>>>
>>>> Looks like windows bot stops on the first failure, so fuse-lld was
>>>> introduced before but hidden by CFI bug.
>>>> Would be nice always continue to run all tests even after an error.
>>>>
>>>> Maybe using "ninja -k N flag" or just by splitting "argv: ['ninja',
>>>> '-j', '16', 'check-asan', 'check-asan-dynamic', 'check-sanitizer',
>>>> 'check-cfi']" into multiple steps.
>>>>
>>>> On Tue, Mar 21, 2017 at 6:57 PM Reid Kleckner via llvm-commits <
>>>> llvm-commits at lists.llvm.org> wrote:
>>>>
>>>>> Author: rnk
>>>>> Date: Tue Mar 21 20:45:08 2017
>>>>> New Revision: 298476
>>>>>
>>>>> URL: http://llvm.org/viewvc/llvm-project?rev=298476&view=rev
>>>>> Log:
>>>>> XFAIL CFI stats test while LLD produces corrupt PDBs that confuse DIA
>>>>>
>>>>> Modified:
>>>>>     compiler-rt/trunk/test/cfi/stats.cpp
>>>>>
>>>>> Modified: compiler-rt/trunk/test/cfi/stats.cpp
>>>>> URL: http://llvm.org/viewvc/llvm-project/compiler-rt/trunk/test/c
>>>>> fi/stats.cpp?rev=298476&r1=298475&r2=298476&view=diff
>>>>> ============================================================
>>>>> ==================
>>>>> --- compiler-rt/trunk/test/cfi/stats.cpp (original)
>>>>> +++ compiler-rt/trunk/test/cfi/stats.cpp Tue Mar 21 20:45:08 2017
>>>>> @@ -2,6 +2,8 @@
>>>>>  // RUN: env SANITIZER_STATS_PATH=%t.stats %t
>>>>>  // RUN: sanstats %t.stats | FileCheck %s
>>>>>
>>>>> +// XFAIL: win32
>>>>> +
>>>>>  // FIXME: We currently emit the wrong debug info under
>>>>> devirtualization.
>>>>>  // UNSUPPORTED: devirt
>>>>>
>>>>>
>>>>>
>>>>> _______________________________________________
>>>>> llvm-commits mailing list
>>>>> llvm-commits at lists.llvm.org
>>>>> http://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-commits
>>>>>
>>>>
>>>
>>
>>
>> --
>> --
>> Peter
>>
>
>


-- 
-- 
Peter
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20170322/2378dd95/attachment.html>


More information about the llvm-commits mailing list