This is the mail archive of the gcc-patches@gcc.gnu.org mailing list for the GCC project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

[patch] h8300.c: Revise a comment.


Hi,

Attached is a patch to revise a comment in two_insn_adds_subs_operand
config/h8300/h8300.c.  The patch also changes "mode != QImode" to
"mode == HImode" as this function is used only in HImode.

Tested on h8300 port.  Committed.

Kazu Hirata

2002-02-07  Kazu Hirata  <kazu@hxi.com>

	* config/h8300/h8300.c (two_insn_adds_subs_operand): Revise a
	comment.  Accept HImode only if TARGET_H8300.

Index: h8300.c
===================================================================
RCS file: /cvs/gcc/gcc/gcc/config/h8300/h8300.c,v
retrieving revision 1.109
diff -c -r1.109 h8300.c
*** h8300.c	2002/02/03 01:36:13	1.109
--- h8300.c	2002/02/08 00:22:13
***************
*** 632,643 ****
  	}
        else
  	{
! 	  /* A constant addition/subtraction takes 2 states in
! 	     QImode. It takes 6 states in HImode, requiring the
! 	     constant to be loaded to a register first, and a lot more
! 	     in SImode.  Thus the only case we can win is when either
! 	     HImode or SImode is used.  */
! 	  if (mode != QImode
  	      && (value == 2 + 1
  		  || value == 2 + 2))
  	    return 1;
--- 632,642 ----
  	}
        else
  	{
! 	  /* We do not profit directly by splitting addition or
! 	     subtraction of 3 and 4.  However, since these are
! 	     implemented as a sequence of adds or subs, they do not
! 	     clobber (cc0) unlike a sequence of add.b and add.x.  */
! 	  if (mode == HImode
  	      && (value == 2 + 1
  		  || value == 2 + 2))
  	    return 1;


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]