misc/176628: [stdint.h] use safer way of definint __WORDSIZE

[Dmitry Marakasov]

The following reply was made to PR kern/176628; it has been noted by GNATS.

From: Dmitry Marakasov <amdmi3 at amdmi3.ru>
To: Bruce Evans <brde at optusnet.com.au>
Cc: FreeBSD-gnats-submit at FreeBSD.org, freebsd-bugs at FreeBSD.org
Subject: Re: misc/176628: [stdint.h] use safer way of definint __WORDSIZE
Date: Thu, 7 Mar 2013 00:38:44 +0400

 * Bruce Evans (brde at optusnet.com.au) wrote:
 
 > > __WORDSIZE is always defined as 32, which is wrong on 64bit systems.
 > >
 > > I have two solutions for the problem.
 > > First one uses the same way of testing for 64bit pointers, but doesn't define __WORDSIZE if it can't be detected reliably.
 > 
 > C++ code should be happier with __WORDSIZE being undefined that with it
 > being defined to garbage.
 > 
 > > Second one uses different way of testing for 64bit pointers with checking for __LP64__.
 > >
 > > The second one looks much more useful, but I'm not sure if __LP64__ has the right semantics and will work in all platforms.
 > 
 > Not right, but better than the UINTPTR_MAX test.  Both are broken and default
 > to __WORDSIZE == 32 if pointers are not precisely 64 bits.  Actually, the
 > __LP64__ test is more fragile, since it fails if __L32_P64__ and many systems
 > use that.  Just not any FreeBSD systems, so it is valid to use a hackish
 > ifdef that only works on current FreeBSD systems.
 
 So, what solution do you think is the best in current situation?
 
 -- 
 Dmitry Marakasov   .   55B5 0596 FF1E 8D84 5F56  9510 D35A 80DD F9D2 F77D
 amdmi3 at amdmi3.ru  ..:  jabber: amdmi3 at jabber.ru    http://www.amdmi3.ru