[gs-bugs] [Bug 690232] indeterminate rendering of 13-03 and 13-12.PS

[bugs.ghostscript.com-bugzilla-daemon at ghostscript.com]

http://bugs.ghostscript.com/show_bug.cgi?id=690232





------- Additional Comments From marcos.woehrmann at artifex.com  2009-02-07 23:02 -------
The problem is in gxccman.c; specifically the initializing of pair->hash in gx_add_fm_pair():

    pair->hash = (uint) (ulong) pair % 549;  /* scramble bits */

On my amd64 box the pair pointer is deterministic, on the cluster it's not.  

To demonstrate this I've instrumented the code thus:

    pair->hash = (uint) (ulong) pair % 549;  /* scramble bits */
    if_debug2('k', "[k] mhw 199 %d 0x%08lx\n",pair->hash,(ulong) pair);
    pair->mxx = mxx, pair->mxy = mxy;

If I run the code on my amd64 box multiple times, saving stdout to log.1..log.n, and then:

  foreach i ( log.* )
  grep "mhw 199" $i | head -1
  end

I get:

  [k] mhw 199 523 0x017adc78
  [k] mhw 199 523 0x017adc78
  [k] mhw 199 523 0x017adc78
  [k] mhw 199 523 0x017adc78
  [k] mhw 199 523 0x017adc78

On the cluster the same experiment produces:

  [k] mhw 199 537 0x09ee3524
  [k] mhw 199 391 0x092a5524
  [k] mhw 199 91 0x09c20524
  [k] mhw 199 517 0x098a8524
  [k] mhw 199 505 0x094eb524
  [k] mhw 199 49 0x0a9d9524






------- You are receiving this mail because: -------
You are the QA contact for the bug, or are watching the QA contact.