[petsc-dev] Extra compiler flags for debugging
Jacob Faibussowitsch
jacob.fai at gmail.com
Mon Mar 23 11:20:45 CDT 2020
> Because the code doesn't get recompiled when there are no changes.
Subsequent recompiles of the code with any changes also don’t reproduce the warning. The only way the warning pops up again is if I change the problematic code to correct code, compile, change it back to problematic then compile again. And then the warning only ever “stays around” for one compile.
Best regards,
Jacob Faibussowitsch
(Jacob Fai - booss - oh - vitch)
Cell: (312) 694-3391
> On Mar 23, 2020, at 11:16 AM, Satish Balay <balay at mcs.anl.gov> wrote:
>
> On Mon, 23 Mar 2020, Jacob Faibussowitsch wrote:
>
>> Hello all,
>>
>> I’m wondering if you guys use any extra compiler flags when debugging (other than -g3). I have noticed for example that when compiling I am only warned about something (i.e. wrong argument type for a function) once, so if I recompile the same program with no changes, the second compilation doesn’t reproduce the warning.
>
> Because the code doesn't get recompiled when there are no changes.
>
> You can use -Werror - this will force to fix all warnings. [This is what we do in the test pipelines for some of the jobs]
>
> Satish
>
>> Currently I use the default compiler flags that petsc generates for debugging support with MPICH:
>>
>> C Compiler: /Users/jacobfaibussowitsch/NoSync/petscpackages/bin/mpicc -Wall -Wwrite-strings -Wno-strict-aliasing -Wno-unknown-pragmas -fstack-protector -fno-stack-check -Qunused-arguments -fvisibility=hidden -g3
>>
>> Version: Apple clang version 11.0.0 (clang-1100.0.33.17)
>>
>> I’m mainly trying to find a 64-bit bug, so I have now added the “-ftrapv” flag to trap for integer overflows but there are probably more flags out there.
>>
>> Best regards,
>>
>> Jacob Faibussowitsch
>> (Jacob Fai - booss - oh - vitch)
>> Cell: (312) 694-3391
>>
>>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.mcs.anl.gov/pipermail/petsc-dev/attachments/20200323/52bab94b/attachment.html>
More information about the petsc-dev
mailing list