i386 local-alloc patch

John Wehle john@feith.com
Thu Jun 25 00:06:00 GMT 1998


ChangeLog:

Thu Jun 25 00:22:27 EDT 1998  John Wehle  (john@feith.com)

	* local-alloc.c (block_alloc): Don't adjust the lifetime of
	a qty if SMALL_REGISTER_CLASSES is in effect.

Enjoy!

-- John Wehle
------------------8<------------------------8<------------------------
*** gcc/local-alloc.c.ORIGINAL	Fri Jun 19 19:36:59 1998
--- gcc/local-alloc.c	Thu Jun 25 00:03:45 1998
***************
*** 1448,1454 ****
  		 This optimization is only appropriate when we will run
  		 a scheduling pass after reload and we are not optimizing
  		 for code size.  */
! 	      if (flag_schedule_insns_after_reload && !optimize_size)
  		{
  		
  		  qty_phys_reg[q] = find_free_reg (qty_min_class[q], 
--- 1448,1455 ----
  		 This optimization is only appropriate when we will run
  		 a scheduling pass after reload and we are not optimizing
  		 for code size.  */
! 	      if (! SMALL_REGISTER_CLASSES
! 		  && flag_schedule_insns_after_reload && !optimize_size)
  		{
  		
  		  qty_phys_reg[q] = find_free_reg (qty_min_class[q], 
***************
*** 1467,1473 ****
  
  #ifdef INSN_SCHEDULING
  	  /* Similarly, avoid false dependencies.  */
! 	  if (flag_schedule_insns_after_reload && !optimize_size
  	      && qty_alternate_class[q] != NO_REGS)
  	    qty_phys_reg[q] = find_free_reg (qty_alternate_class[q],
  					     qty_mode[q], q, 0, 0,
--- 1468,1475 ----
  
  #ifdef INSN_SCHEDULING
  	  /* Similarly, avoid false dependencies.  */
! 	  if (! SMALL_REGISTER_CLASSES
! 	      && flag_schedule_insns_after_reload && !optimize_size
  	      && qty_alternate_class[q] != NO_REGS)
  	    qty_phys_reg[q] = find_free_reg (qty_alternate_class[q],
  					     qty_mode[q], q, 0, 0,
-------------------------------------------------------------------------
|   Feith Systems  |   Voice: 1-215-646-8000  |  Email: john@feith.com  |
|    John Wehle    |     Fax: 1-215-540-5495  |                         |
-------------------------------------------------------------------------




More information about the Gcc-patches mailing list