On 7/14/2026 11:22 AM, Ilya Leoshkevich wrote:
>
>
> On 7/14/26 17:43, Pierrick Bouvier wrote:
>> On 7/14/2026 1:54 AM, Ilya Leoshkevich wrote:
>>>
>>>
>>> On 7/10/26 23:43, Pierrick Bouvier wrote:
>>>> Signed-off-by: Pierrick Bouvier
>>>> <[email protected]> --- tests/tcg/
>>>> meson.build | 1 + tests/tcg/s390x/Makefile.target
>>>> | 127 ----------------------------- tests/tcg/s390x/
>>>> meson.build | 138 +++++++++++++++++++++++++++ +++++ 3
>>>> files changed, 139 insertions(+), 127 deletions(-) delete mode
>>>> 100644 tests/tcg/s390x/Makefile.target create mode 100644
>>>> tests/tcg/s390x/meson.build
>>>
>>> In general the s390x tests work now, however, I get:
>>>
>>> 155/164 tcg+tcg-s390x-linux-user - qemu:s390x-linux-user-fma
>>> SKIP 0.09s exit status 77 156/164 tcg+tcg-s390x-
>>> linux-user - qemu:s390x-linux-user-vfminmax SKIP
>>> 0.07s exit status 77 157/164 tcg+tcg-s390x-linux-user -
>>> qemu:s390x-linux-user-vxeh2_vs SKIP 0.06s exit
>>> status 77 158/164 tcg+tcg-s390x-linux-user - qemu:s390x-linux-
>>> user-vxeh2_vcvt SKIP 0.06s exit status 77 159/164
>>> tcg+tcg-s390x-linux-user - qemu:s390x-linux-user-vxeh2_vlstr
>>> SKIP 0.05s exit status 77 160/164 tcg+tcg-s390x-
>>> linux-user - qemu:s390x-linux-user-vxeh2_vstrs SKIP
>>> 0.04s exit status 77
>>>
>>> with the series. Without the series these tests do run:
>>>
>>> TEST fma on s390x TEST vfminmax on s390x TEST
>>> vxeh2_vcvt on s390x TEST vxeh2_vlstr on s390x TEST
>>> vxeh2_vstrs on s390x TEST vxeh2_vs on s390x
>>>
>>
>> Can you please run one of the test in verbose? It will print the
>> reason why it was skipped. $ ./build/pyvenv/bin/meson test -C
>> build \ s390x-linux-user-vxeh2_vcvt --verbose
>
> $ ./build/pyvenv/bin/meson test -C build \ s390x-linux-user-
> vxeh2_vcvt --verbose 1/1 tcg+tcg-s390x-linux-user - qemu:s390x-linux-
> user-vxeh2_vcvt RUNNING
>>>>
> UBSAN_OPTIONS=halt_on_error=1:abort_on_error=1:print_summary=1:print_stacktrace=1
> MSAN_OPTIONS=halt_on_error=1:abort_on_error=1:print_summary=1:print_stacktrace=1
> ASAN_OPTIONS=halt_on_error=1:abort_on_error=1:print_summary=1
> MESON_TEST_ITERATION=1 RUST_BACKTRACE=1 MALLOC_PERTURB_=146 /usr/bin/
> env sh -c 'echo "compiler feature z15 not supported" && exit 77'
> ―――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――
> ✀
> ――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――
> compiler feature z15 not supported
>
>> I guess the cross compiler is not detected, and we need to find
>> why. With following questions, we should know what's missing.
>>
>> - Do you have s390x-linux-gnu-gcc installed?
>
> Yes.
>
>>
>> - Does this command work? $ s390x-linux-gnu-gcc -static -o /dev/
>> null tests/tcg/test_cc.c
>
> $ s390x-linux-gnu-gcc -static -o /dev/null tests/tcg/test_cc.c In
> file included from tests/tcg/test_cc.c:3: /usr/lib/gcc/s390x-linux-
> gnu/16/include/stdint.h:11:16: fatal error: stdint.h: No such file
> or directory 11 | # include_next <stdint.h> |
> ^~~~~~~~~~
>
> Apparently it's the same issue. Fedora doesn't ship cross glibc
> headers.
>
And we should skip using it thus.
>> - Do you have podman or docker installed?
>
> Yes.
>
>>
>> - What is the output of this command? $ ./tests/docker/docker.py
>> probe
>
> $ ./tests/docker/docker.py probe docker
>
>>
>> Regards, Pierrick
>
It seems like local compiler was taken even though it's not working,
which should not happen.
Double checking the part checking this, there was a typo on my side.
Sorry, I checked this in fedora, refactored after and missed it.
Could you please try with following change?
It should correctly use container version and compile binaries without
checking if feature is supported.
```
--- a/tests/tcg/meson.build
+++ b/tests/tcg/meson.build
@@ -192,7 +192,7 @@ foreach target, plan: tcg_tests
cmd = run_command([cc, files('test_cc.c'), check_flags, '-o', '/
dev/null'],
check: false)
avail = cmd.returncode() == 0
- cc_from_system = has_cc
+ cc_from_system = avail
has_cc = avail
endif
```
Regards,
Pierrick
