Suppress new gcc-15 -Wunterminated-string-initialization

[denix0]

GCC 15 now warns when character arrays are being initialized by strings
and terminating NUL character doesn't fit. GCC 15.1 also allows marking
such arrays with nonstring attribute to suppress the warning. W/o such
attribute, the warning becomes error due to the global -Werror. Add the
attribute accordingly.

[mikkorapeli-linaro]

Fixes optee-test yocto builds from meta-arm.

[jforissier]

Thanks @denix0

Reviewed-by: Jerome Forissier <jerome.forissier@linaro.org>

[jenswikl]

Acked-by: Jens Wiklander <jens.wiklander@linaro.org>

[etienne-lms]

Acked-by: Etienne Carriere <etienne.carriere@foss.st.com>

[rossburton]

FWIW, our clang builds fail with this patch:

./regression_4000_data.h:7917:78: error: unknown attribute 'nonstring' ignored [-Werror,-Wunknown-attributes]

[mikkorapeli-linaro]

FWIW, our clang builds fail with this patch:

./regression_4000_data.h:7917:78: error: unknown attribute 'nonstring' ignored [-Werror,-Wunknown-attributes]

#797 should help by disabling Werror. Maybe we can backport this to meta-arm...

[rossburton]

#797 should help by disabling Werror. Maybe we can backport this to meta-arm...

Yes, I'd approve that.

[denix0]

Is clang behind gcc for once? :)

Anyway, according to the manual clang 21 does have support for this attribute, but in Yocto we have 20.1 for now:
https://clang.llvm.org/docs/AttributeReference.html#nonstring

[mikkorapeli-linaro]

I hope #ifdef's for gcc/clang and/or versions are not needed... or feature checks?

[jforissier]

I hope #ifdef's for gcc/clang and/or versions are not needed... or feature checks?

#if __has_attribute(nonstring) should work

[denix0]

This nonstring attribute has been added to clang only last week:
llvm/llvm-project#137705

[denix0]

I hope #ifdef's for gcc/clang and/or versions are not needed... or feature checks?

#if __has_attribute(nonstring) should work

I've just updated the commit with this check.

Should I collect all the early reviews in the commit message, or wait for the new ones?

[jforissier]

I hope #ifdef's for gcc/clang and/or versions are not needed... or feature checks?

#if __has_attribute(nonstring) should work

I've just updated the commit with this check.

Should I collect all the early reviews in the commit message, or wait for the new ones?

Acked-by's can usually be kept when changes are minor. Reviewed-by's are a different story, but you have mine anyways:

Reviewed-by: Jerome Forissier <jerome.forissier@linaro.org>

Thanks!

[denix0]

I've just updated the commit with this check.
Should I collect all the early reviews in the commit message, or wait for the new ones?

Acked-by's can usually be kept when changes are minor. Reviewed-by's are a different story, but you have mine anyways:

Reviewed-by: Jerome Forissier <jerome.forissier@linaro.org>

Thanks!

Ok, updated the commit one more time with reviews and acks. Thanks for all the feedback.