Re: [PATCH] gas value truncation warning reports truncated values, doesn't look at signedness.

[Bart Samwel]

Alan Modra wrote:
> On Mon, Jan 12, 2004 at 02:20:25AM +0100, Bart Samwel wrote:
>
> > In 64-bit, ~0x80000000 is 0xffffffff7fffffff.
>
> True if int is 64 bits.  Note that most (all?) 64 bit gcc targets
> have a 32 bit int.  I'm harping on about this because it's a common
> error to think ~0x80000000 is something other than 0x7fffffff when
> writing C for a 64 bit target.  Try this:
>
> #include <stdio.h>
> int main (void)
> {
>   long long x = ~0x80000000;
>   printf ("%llx\n", x);
>   return 0;
> }

I understand why that doesn't work -- if 0x80000000 is an int, 
~0x80000000 is going to use 32-bit arithmetic. It should be 
~0x80000000LL, of course. I'm working on a 32-bit architecture however 
-- i386, to be precise. So I don't know why Debian compiles with a 
64-bit bfd for a 32-bit arch. Would there be any reason for them to do 
so, or might this just be something that slipped between the cracks when 
they added 64-bit architectures?

-- Bart