Issue about buildpath introduced by -fintrinsic-modules-path option

[ChenQi]

On 10/18/24 14:17, Andrew Pinski wrote:
> On Thu, Oct 17, 2024 at 7:52 PM ChenQi <Qi.Chen@windriver.com> wrote:
>> Hi Fortran community,
>>
>> I recently got a buildpath issue when building libgfortran. I'm sending
>> out email to this mailing list first, but if you think it's a bug
>> instead of some expected behavior, please let me know and I'll file a bug.
>>
>> Here's the issue: the generated binary has buildpath in the
>> DW_AT_producer section. Below are some output from 'readelf --debug-dump
>> libgfortran.so.5.0.0'.
>>
>> """
>>
>> <0><1a37d3>: Abbrev Number: 4 (DW_TAG_compile_unit)
>>           <1a37d4>   DW_AT_producer    : (indirect string, offset:
>> 0xd653): GNU Fortran2008 14.2.0 -m64 -march=core2 -mtune=core2 -msse3
>> -mfpmath=sse -mshstk -g -O2 -O2 -fstack-protector-strong -fimplicit-none
>> -fno-repack-arrays -fno-underscoring -fcf-protection=full
>> -fallow-leading-underscore -fbuilding-libgfortran -fPIC
>> -fintrinsic-modules-path
>> /ala-lpggp72/qichen/Yocto/builds/build-poky/tmp/work/core2-64-poky-linux/libgfortran/14.2.0/recipe-sysroot-native/usr/bin/x86_64-poky-linux/../../lib/x86_64-poky-linux/gcc/x86_64-poky-linux/14.2.0/finclude
>> -fpre-include=../../../../recipe-sysroot/usr/include/finclude/math-vector-fortran.h
>>
>> """
>>
>> The problem is that this '-fintrinsic-modules-path' is not passed by me.
>> Use some .o file as an example. Here's command to build out
>> .libs/_mod_r17.o:
>>
>> """
>>
>> libtool: compile:  x86_64-poky-linux-gfortran -m64 -march=core2
>> -mtune=core2 -msse3 -mfpmath=sse -fstack-protector-strong -O2
>> -D_FORTIFY_SOURCE=2 -Wformat -Wformat-security -Werror=format-security
>> --sysroot=../../../../recipe-sysroot -fcanon-prefix-map
>> -fdebug-prefix-map=/ala-lpggp72/qichen/Yocto/builds/build-poky/tmp/work-shared/gcc-14.2.0-r0/gcc-14.2.0=/usr/src/debug/libgfortran/14.2.0
>> -fdebug-prefix-map=/ala-lpggp72/qichen/Yocto/builds/build-poky/tmp/work/core2-64-poky-linux/libgfortran/14.2.0/gcc-14.2.0/build.x86_64-poky-linux.x86_64-poky-linux=/usr/src/debug/libgfortran/14.2.0
>> -fdebug-prefix-map=/ala-lpggp72/qichen/Yocto/builds/build-poky/tmp/work/core2-64-poky-linux/libgfortran/14.2.0/recipe-sysroot-native=
>> -DHAVE_CONFIG_H -I. -I../../../../../../../../work-shared/gcc-14.2.\
>> 0-r0/gcc-14.2.0/libgfortran
>> -iquote../../../../../../../../work-shared/gcc-14.2.0-r0/gcc-14.2.0/libgfortran/io
>> -I../../../../../../../../work-shared/gcc-14.2.0-r0/gcc-14.2.0/libgfortran/../gcc
>> -I../../../../../../../../work-shared/gcc-14.2.0-r0/gcc-14.2.0/libgfortran/../gcc/config
>> -I/ala-lpggp72/qichen/Yocto/builds/build-poky/tmp/work/core2-64-poky-linux/libgfortran/14.2.0/gcc-14.2.0/build.x86_64-poky-linux.x86_64-poky-linux/x86_64-poky-linux/libgfortran/../.././gcc
>> -I../../../../../../../../work-shared/gcc-14.2.0-r0/gcc-14.2.0/libgfortran/../libgcc
>> -I/ala-lpggp72/qichen/Yocto/builds/build-poky/tmp/work/core2-64-poky-linux/libgfortran/14.2.0/gcc-14.2.0/build.x86_64-poky-linux.x86_64-poky-linux/x86_64-poky-linux/libgfortran/../libgcc
>> -I../../../../../../../../work-shared/gcc-14.2.0-r0/gcc-14.2.0/libgfortran/../libbacktrace
>> -I/ala-lpggp72/qichen/Yocto/builds/build-poky/tmp/work/core2-64-poky-linux/libgfortran/14.2.0/gcc-14.2.0/build.x86_64-poky-linux.x86_64-poky-linux/x86_64-poky-linux/libgfortran/../libbacktrace
>> -I../libbacktrace -I . -Wall -Werror -fimplicit-none -fno-repack-arrays
>> -fno-underscoring -fcf-protection -mshstk -fallow-leading-underscore
>> -fbuilding-libgfortran -g -O2 -c
>> ../../../../../../../../work-shared/gcc-14.2.0-r0/gcc-14.2.0/libgfortran/generated/_mod_r17.F90
>> -fPIC -o .libs/_mod_r17.o
>>
>> """
>>
>> You can see that there's no '-fintrinsic-modules-path' there. This means
>> that this '-fintrinsic-modules-path' is added automatically by gfortran.
>> And the generated _mod_r17.o has buildpath encoded.
>>
>> I tried to add '-fintrinsic-modules-path=/not/exist', hoping that no
>> extra '-fintrinsic-modules-path' option would be added, but it does not
>> work.
>>
>> I searched the whole gcc source codes, and really cannot find any codes
>> that adds this option automatically.
> It comes from the fortran specs:
> ```
> #define F951_OPTIONS        "%(cc1_options) %{J*} \
>                               %{!nostdinc:-fintrinsic-modules-path finclude%s}" \
>                               TARGET_F951_OPTIONS \
>                               "%{!fsyntax-only:%(invoke_as)}"
> ...
> {"@f95",
>      "f951 %i %{E:%{!cpp:%egfortran does not support -E without -cpp}}\
>            %{cpp:" F951_CPP_OPTIONS ";:  " CPP_FORWARD_OPTIONS  "} \
>            %{!E:" F951_OPTIONS "}", 0, 0, 0},
>
> ```
>
> I see the producer string should only have the saved options if
> dwarf_record_gcc_switches is true which is the default.
> You can turn that off with -gno-record-gcc-switches so build
> directories don't show up.
>
> Thanks,
> Andrew Pinski

Thanks a lot Andrew! This option works.

For anyone who is having a similar issue, here's the patch I sent to 
Yocto to fix this buildpath issue (libgfortran build, cross compilation 
environment): 
https://lists.openembedded.org/g/openembedded-core/message/206040

Regards,

Qi

>
>> Does someone know what's going on? Why is this
>> '-fintrinsic-modules-path' added automatically? Is there any way to get
>> rid of this buildpath?
>>
>> Thanks in advance.
>>
>> Regards,
>>
>> Qi
>>