[petsc-dev] Building libCEED with -fsanitize=address

[Jeremy Luke Thompson]

Yeah, I misunderstood. That's what I get for reading emails before coffee.

Note, there are some known ASAN issues, specifically with one of our
diagonal assembly tests, but I haven't had the time to make a good fix and
it's not used in our Ratel or fluids work at this time.

On Tuesday, June 13, 2023, Matthew Knepley <knepley at gmail.com> wrote:
> On Tue, Jun 13, 2023 at 9:18 AM Jeremy Luke Thompson <
Jeremy.Thompson at colorado.edu> wrote:
>>
>> It's probably easier to trace that sort of thing when building libCEED
on its own, but it looks like ASAN is complaining about `char
*source_path_with_name = "";`
>>
>> Seems like an easy change I can do this morning.
>
> It is a warning. The fatal error was a link without the ASAN flag, but
somehow this is swallowed. Jacob found it.
>   Thanks,
>      Matt
>
>>
>> On Tuesday, June 13, 2023, Matthew Knepley <knepley at gmail.com> wrote:
>> > For some reason, libCEED seems broken with Address Sanitizer
>> >
*********************************************************************************************
>> >            UNABLE to CONFIGURE with GIVEN OPTIONS (see configure.log
for details):
>> >
---------------------------------------------------------------------------------------------
>> >   Error running make on libceed: Could not execute
"[['/PETSc3/petsc/apple/bin/make',
>> >   '-j13', '-l24.0',
'prefix=/PETSc3/petsc/petsc-pylith/arch-pylith-debug', 'V=1',
>> >   'CC=/PETSc3/petsc/apple/bin/mpicc', 'CFLAGS= -fPIC -Wall
-Wwrite-strings
>> >   -Wno-unknown-pragmas -fstack-protector -fno-stack-check
-Qunused-arguments
>> >   -fvisibility=hidden -fsanitize=address ',
'CXX=/PETSc3/petsc/apple/bin/mpicxx',
>> >   'CXXFLAGS= -Wall -Wwrite-strings -Wno-strict-aliasing
-Wno-unknown-pragmas
>> >   -fstack-protector -fno-stack-check -fvisibility=hidden
-fsanitize=address  -std=gnu++17
>> >   -fPIC ', '-B']]":
>> >
>> >   make: 'lib' with optional backends: /cpu/self/memcheck/serial
/cpu/self/memcheck/blocked
>> >   /PETSc3/petsc/apple/bin/mpicc -I./include -fPIC -Wall -Wwrite-strings
>> >   -Wno-unknown-pragmas -fstack-protector -fno-stack-check
-Qunused-arguments
>> >   -fvisibility=hidden -f
>> >   .... more output .....
>> >   mcheck/ceed-memcheck-qfunction.o
build/backends/memcheck/ceed-memcheck-qfunctioncontext.o
>> >   build/backends/memcheck/ceed-memcheck-serial.o
>> >   build/backends/memcheck/ceed-memcheck-vector.o
build/backends/ceed-backend-weak.o
>> >   build/gallery/ceed-gallery-weak.o
>> >
-lm/PETSc3/petsc/petsc-pylith/arch-pylith-debug/externalpackages/git.libceed/interface/ceed-preconditioning.c:44:9:
>> >   warning: initializing 'char *' with an expression of type 'const
char [1]' discards
>> >   qualifiers [-Wincompatible-pointer-types-discards-qualifier
>> >   .... more error .....
>> >   ceed-operator.o
>> >   _asan.module_ctor in ceed-preconditioning.o
>> >   ...
>> >   ld: symbol(s) not found for architecture x86_64
>> >   clang: error: linker command failed with exit code 1 (use -v to see
invocation)
>> >   make: *** [Makefile:517: lib/libceed.dylib] Error 1
>> >
>> > Does anyone else see this?
>> >
>> >   Thanks,
>> >      Matt
>> >
>> > --
>> > What most experimenters take for granted before they begin their
experiments is infinitely more interesting than any results to which their
experiments lead.
>> > -- Norbert Wiener
>> > https://www.cse.buffalo.edu/~knepley/
>> >
>
> --
> What most experimenters take for granted before they begin their
experiments is infinitely more interesting than any results to which their
experiments lead.
> -- Norbert Wiener
> https://www.cse.buffalo.edu/~knepley/
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.mcs.anl.gov/pipermail/petsc-dev/attachments/20230613/15e87d84/attachment-0001.html>