x86/cpufeature: Fix various quality problems in the <asm/cpu_device_hd.h> header
Thomas noticed that the new arch/x86/include/asm/cpu_device_id.h header is
a train-wreck that didn't incorporate review feedback like not using __u8
in kernel-only headers.
While at it also fix all the *other* problems this header has:
- Use canonical names for the header guards. It's inexplicable why a non-standard
guard was used.
- Don't define the header guard to 1. Plus annotate the closing #endif as done
absolutely every other header. Again, an inexplicable source of noise.
- Move the kernel API calls provided by this header next to each other, there's
absolutely no reason to have them spread apart in the header.
- Align the INTEL_CPU_DESC() macro initializations vertically, this is easier to
read and it's also the canonical style.
- Actually name the macro arguments properly: instead of 'mod, step, rev',
spell out 'model, stepping, revision' - it's not like we have a lack of
characters in this header.
- Actually make arguments macro-safe - again it's inexplicable why it wasn't
done properly to begin with.
Quite amazing how many problems a 41 lines header can contain.
This kind of code quality is unacceptable, and it slipped through the
review net of 2 developers and 2 maintainers, including myself, until
Thomas noticed it. :-/
Reported-by: Thomas Gleixner <firstname.lastname@example.org>
Cc: Andi Kleen <email@example.com>
Cc: Kan Liang <firstname.lastname@example.org>
Cc: Peter Zijlstra (Intel) <email@example.com>
Cc: Borislav Petkov <firstname.lastname@example.org>
Cc: Linus Torvalds <email@example.com>
Signed-off-by: Ingo Molnar <firstname.lastname@example.org>
1 file changed