toolchain: Replace GCC_VERSION and BUILD_ASSERT macros in gcc.h

[keith-packard]

GCC_VERSION is defined in a few modules, and those headers are often included first, so we need to replace that definition with the Zephyr-specific one.

BUILD_ASSERT is also defined in include/toolchain/common.h, which might get included before gcc.h. We want to use the gcc-specific one instead of the general one.

Signed-off-by: Keith Packard keithp@keithp.com

[stephanosio]

As per #51963, the modules should either conditionally define (if the semantics are equivalent to that of Zephyr) or rename; not the other way around.

[keith-packard]

As per #51963, the modules should either conditionally define (if the semantics are equivalent to that of Zephyr) or rename; not the other way around.

#51963 seems to suggest that we use Z_GCC_VERSION in this file as it is purely internal in nature and not part of the Zephyr API. I'll run tests with BUILD_ASSERT left in place, but with Z_GCC_VERSION and see what happens. This patch is getting old enough that "things have changed".

[stephanosio]

#51963 seems to suggest that we use Z_GCC_VERSION in this file as it is purely internal in nature and not part of the Zephyr API. I'll run tests with BUILD_ASSERT left in place, but with Z_GCC_VERSION and see what happens. This patch is getting old enough that "things have changed".

Sure, GCC_VERSION is not part of any public API and is more or less only intended to be used by the "toolchain abstraction layer," so Z_GCC_VERSION would not be a bad choice; though, the Z_ prefix is often more associated with the kernel and arch-layer stuff.

Since this is a toolchain abstraction layer-specific macro, TOOLCHAIN_GCC_VERSION might not be a bad choice either -- prefixing subsystem-specific identifiers with a subsystem-specific keyword is always a good practice.

[keith-packard]

Since this is a toolchain abstraction layer-specific macro, TOOLCHAIN_GCC_VERSION might not be a bad choice either -- prefixing subsystem-specific identifiers with a subsystem-specific keyword is always a good practice.

Sounds good. I've rebased and pushed this version.

[stephanosio]

BUILD_ASSERT is also defined in include/toolchain/common.h, which might
get included before gcc.h. We want to use the gcc-specific one instead
of the general one.

toolchain/common.h is never supposed to be included before the toolchain-specific headers such as toolchain/gcc.h; in fact, it should not be included by any files other than the toolchain-specific headers inside toolchain/.

Properly documenting and enforcing that seems to be beyond the scope of this PR, however.