Re: + kernelh-make-abs-work-with-64-bit-types.patch added to -mm tree

[Michal Nazarewicz <mina86@mina86.com>]

On Thu, Sep 24 2015, Linus Torvalds wrote:
> One thing that *is* interesting is "what if 'long' and 's64' are the
> same size?" In particular, it means that right now Michal's patch
> *always* returns "long" on a 64-bit architecture, but will return
> "long" or "s64" on a 32-bit one.

That’s not accurate.  s64 is defined as long long so:
- on 64-bit architectures, the macro will return s64 (i.e. long long)
  for long arguments (because sizeof(long) == 8 == sizeof(s64) and the
  first path is taken), but
- on 32-bit architectures, it will return long for long arguments (since
  sizeof(long) == 4 != 8 == sizeof(long long) and the second path is
  taken).
But yes, the point remains, depending on architecture, the macro returns
different type for long arguments.

> The reason that is somewhat interesting is that while the sizes and
> values are the same, and the resulting C type expansions are
> "equivalent" types, i people *print* things, you have to use different
> modifiers for the two cases. So you might get warnings on 32-bit
> architectures and not get them on 64-bit, or vice versa.
>
> However, I don't see a good solution for that. And assuming we don't
> use "abs()" in an expression to printk(), I guess it doesn't much
> matter either.

This should do the trick:

#define abs(x) __builtin_choose_expr(				\
	__builtin_types_compatible_p(typeof(x), s64) ||		\
	__builtin_types_compatible_p(typeof(x), u64),		\
	({ s64 __x = (x); __x < 0 ? -__x : __x; }),		\
	__builtin_choose_expr(sizeof(x) <= sizeof(long), ({	\
		long ret;					\
		if (sizeof(x) == sizeof(long)) {		\
			long __x = (x);				\
			ret = (__x < 0) ? -__x : __x;		\
		} else {					\
			int __x = (x);				\
			ret = (__x < 0) ? -__x : __x;		\
		}						\
		ret;						\
	}), (void)(x)))

It’ll return s64 for s64 and u64 (i.e. long long and unsigned long long)
types, long for anything whose sizeof <= sizeof(long) and will bail out
with compile time error if used for any other type (if return value is
used).

I dunno whether added complexity is worth solving the problem though.

-- 
Best regards,                                            _     _
.o. | Liege of Serenely Enlightened Majesty of         o' \,=./ `o
..o | Computer Science,  ミハウ “mina86” ナザレヴイツ  (o o)
ooo +--<mpn@google.com>--<xmpp:mina86@jabber.org>-----ooO--(_)--Ooo--