This is the mail archive of the
gcc@gcc.gnu.org
mailing list for the GCC project.
Re: loop optimizer naivety
> The patch for PR opt/10087 was minimal and as non-intrusive as possible
> to fix the testcase. I wasn't very familiar with the loop optimizer and
> didn't want to change variables such as 'biv_count' and 'incremented'
> without fully understanding what effect it would have in other parts of
> the code.
Yes, I've attached a patch of the same flavor which fixes PR opt/12340.
> I'm willing to look into this, but unfortunately the testcase for PR
> opt/12340 is difficult to work with and apparently doesn't segfault on my
> system even with a vanilla 3.3 branch build.
Weird, it segfaults on mine. Do you use the right options?
--
Eric Botcazou
Index: unroll.c
===================================================================
RCS file: /cvs/gcc/gcc/gcc/unroll.c,v
retrieving revision 1.184.2.5
diff -u -p -r1.184.2.5 unroll.c
--- unroll.c 21 Jul 2003 17:49:59 -0000 1.184.2.5
+++ unroll.c 23 Sep 2003 10:22:41 -0000
@@ -2478,7 +2478,13 @@ biv_total_increment (bl)
if (v->always_computable && v->mult_val == const1_rtx
&& ! v->maybe_multiple
&& SCALAR_INT_MODE_P (v->mode))
- result = fold_rtx_mult_add (result, const1_rtx, v->add_val, v->mode);
+ {
+ /* If we have already counted it, skip it. */
+ if (v->same)
+ continue;
+
+ result = fold_rtx_mult_add (result, const1_rtx, v->add_val, v->mode);
+ }
else
return 0;
}
@@ -3538,6 +3544,9 @@ loop_iterations (loop)
REGNO (biv_inc->add_val));
return 0;
}
+ /* If we have already counted it, skip it. */
+ else if (biv_inc->same)
+ continue;
offset -= INTVAL (biv_inc->add_val);
}