arch.7: deprecate __ILP32__ and __LP64__ macros

[brooks]

Summary

With CHERI there now more options than ILP32 and LP64 so lack of
one does not imply the other. Encourage the use of SIZEOF_* macro
tests in their place.

Sponsored by: Innovate UK

[kib]

Are ILP32/LP64 undefined for cheri? If yes, I think that this should be explicitly mentioned, and in fact these macros are still useful, to exclude the outliers from regular arches.

[brooks]

Attempt to approve wording around continued use of ILP32 and LP64.

[kib]

So are ILP32/LP64 defined for cheri? If not, please mention it explicitly.

[brooks]

Be explicit that the compiler does not define LP64 on CHERI
architectures. Don't bother mentioning ILP32 as FreeBSD will never
support 32-bit CHERI.

[brooks]

FWIW, we do currently add __LP64__ to CFLAGS in bsd.cpu.mk has a hack because there's a lot of code of the form:

#ifdef __LP64__
// 64-bit stuff
#else 
// 32-bit stuff
#endif

and we haven't cleaned it all up yet. The goal here is to get people stop adding more. It is usually less wrong to write:

#ifdef __ILP32__
// 32-bit stuff
#else 
// 64-bit stuff
#endif

since much of the time a the 64-bit stuff isn't about pointer size.

[emaste]

Latest update doesn't seem to include the mentioned change.

[brooks]

Actully add previous change I guess I hadn't rebased

[jhb]

share/man/man7/arch.7

346

357

Or perhaps "test only one"

361

Do we want to word this more strongly? Something like:

With the arrival of CHERI architectures, this is no longer the case.
.Dv __SIZEOF_*__
macros should be used instead.
New uses of
.Dv __ILP32__
and
.Dv __LP64__
should be avoided.

[kib]

share/man/man7/arch.7

361

Since cheri does not define either ILP32 nor LP64, both macros become esp. useful in fact.

[jrtc27]

share/man/man7/arch.7

361

No, if you want to ask "is not CHERI" then ask "is not CHERI". Don't use "is ILP32 or LP64" as a proxy for "is not CHERI". ILP32 and LP64 should only be used when you specifically want to check for that exact ABI's combination of types; anything else and you should be more precise and query the properties you actually care about (whether size of long, size of pointer, CHERI or not, etc).

[brooks]

Corrections and suggestions from @jhb