Followup on 64-bit problem with Sun Ultra-5

[Paul Kinzelman <pkinz at timesync dot com>]

I'm running gcc version 2.8.1on:
Sun: Ultra 5 running Solaris 7 (uname says SunOS 5.7)
RedHat6: Pentium-2, 233MHz running Linux 2.2.5-15

I just got the same thing compiled under the sun standard
and
their professional compiler. So looks like maybe gcc is
working
correctly and the CPU is not. Next step is Sun...

If I do the following:
  unsigned long x, y;
  unsigned long long z;
  x = 123;
  y = 456;
  z = (((unsigned long long)y) * 1000000000) + x;
  printf ("y,x=%ld,%ld; z=%Ld\n", y, x, z);

The printf shows:
Sun:    y,x=456,123; z=106     [bogus]
RH6:   y,x=456,123; z=456000000123 [correct]

I tried these alternatives with the same result:
  z = (y * 100000000ULL) + x;
  z = (y * ((unsigned long long)100000000)) + x;

I got down this path because I was getting bogus answers
elsewhere in the program I was writing, so
I think the problem is more than just how they print.

Any help would be greatly appreciated!

--
-Paul Kinzelman, San Jose, CA, ISD Corp