>Number:         21414
>Category:       toolchain
>Synopsis:       double to unsigned int conversion broken
>Confidential:   no
>Severity:       serious
>Priority:       medium
>Responsible:    toolchain-manager
>State:          open
>Class:          sw-bug
>Submitter-Id:   net
>Arrival-Date:   Thu May 01 17:56:00 UTC 2003
>Closed-Date:
>Last-Modified:
>Originator:     Martin Husemann
>Release:        NetBSD 1.6R
>Organization:
>Environment:
System: NetBSD sunny-weather.duskware.de 1.6R NetBSD 1.6R (SUNNY) #4: Thu May 1 13:49:15 CEST 2003 martin@sunny-weather.duskware.de:/usr/src/sys/arch/sparc64/compile/SUNNY sparc64
Architecture: sparc64
Machine: sparc64
>Description:

The below test program outputs 

  ul = 567686, dtemp = 2438192804.397056, result = 0x7fffffff

on sparc64, while it (correctly) prints

  ul = 567686, dtemp = 2438192804.397056, result = 0x9153dea4

on alpha and i386.

This has been noticed by Frank Kardel in PR port-sparc64/21406 as ntpd
lossage (that PR is suspended until this is fixed, please reexamine it when
the toolchain is fixed)

>How-To-Repeat:

Compile and run this:

main()
{
  unsigned long ul = 567686;
  unsigned int  oul;
  double dtemp;

  dtemp = ul * 4294967296. / 1e6;

  oul = (unsigned int)dtemp;
 
  printf("ul = %ld, dtemp = %f, result = 0x%x\n", ul, dtemp, oul);
}

>Fix:
n/a
>Release-Note:
>Audit-Trail:
>Unformatted: