Bug: loop incr optimised to a constant

[Mark Powell]

Franz,

Yes it fixes the bug on both my m68k-unknown-coff and sparc-sun-solaris2.7
configurations (gcc-2.95.2).

thank you for finding this.


-- 
Mark Powell, Senior Software Engineer, Primagraphics Limited
New Cambridge House, Litlington, nr.Royston, Herts, SG8 0SS, UK
Tel. +44 1763 852222, Fax. 853324, medp@primag.co.uk, http://www.primag.co.uk


On 09-Nov-99 Franz Sirl wrote:
>  At 15:02 08.11.99 , Franz Sirl wrote:
> >Unless one of the loop experts tackles this, I will look at that later 
> >today or tomorrow.
>  
>  The appended little patch from the mainline sources solves this problem
>  for 
>  me. Mark, please try the patch on your platform. If it fixes the problem 
>  for you too, the patch should go into the release branch.
>  
>  Franz.
>  
>  1999-08-31 12:20 -0700 Zack Weinberg <zack@bitmover.com>
>           * loop.c (check_dbra_loop): Check bl->biv->add_val is a
>           CONST_INT before using its INTVAL.
>  
>  Index: loop.c
>  ===================================================================
>  RCS file: /cvs/gcc/egcs/gcc/loop.c,v
>  retrieving revision 1.179
>  retrieving revision 1.180
>  diff -u -p -r1.179 -r1.180
>  --- loop.c      1999/08/29 10:09:28     1.179
>  +++ loop.c      1999/08/31 19:39:05     1.180
>  @@ -7797,7 +7797,8 @@ check_dbra_loop (loop_end, insn_count, l
>               }
>           }
>        }
>  -  else if (INTVAL (bl->biv->add_val) > 0)
>  +  else if (GET_CODE (bl->biv->add_val) == CONST_INT
>  +          && INTVAL (bl->biv->add_val) > 0)
>        {
>          /* Try to change inc to dec, so can apply above optimization.  */
>          /* Can do this if: