Breakage in expand_end_case

[Jeffrey A Law]

Yes, this is fine for the testsuite.

  In message < Pine.LNX.3.96.1000319130038.15700A-100000@masala.cygnus.co.uk >you
 write:
  > On Thu, 16 Mar 2000, Richard Kenner wrote:
  > 
  > > + 	       /* RANGE may be signed, and really large ranges will sho
  > w up
  > > + 		  as negative numbers.  */
  > > + 	       || compare_tree_int (range, 0) < 0
  > > 
  > > Well, this should use tree_int_cst_sgn, but I'm confused: if RANGE is
  > > negative, compare_tree_int will already return -1.
  > > 
  > > Can you show me the precise code that shows this failure?
  > 
  > Here it is.  Can I install this in the testsuite (stripped down from the
  > Linux reboot function)?
  > 
  > 
  > long sys_reboot(int magic1, int magic2, int cmd, void * arg)
  > {
  > 	switch (cmd) {
  > 
  > 	case 0x89ABCDEF:
  > 		break;
  > 
  > 	case 0x00000000:
  > 		break;
  > 
  > 	case 0xCDEF0123:
  > 		break;
  > 
  > 	case 0x4321FEDC:
  > 		break;
  > 
  > 	case 0xA1B2C3D4:
  > 		break;
  > 
  > 	default:
  > 		break;
  > 	};
  > 	return 0;
  > }
  > 
  >  
  > 
  >