gcc-4.3.2: add backport fix for PR 32044

Previous Topic Next Topic
 
classic Classic list List threaded Threaded
4 messages Options
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

gcc-4.3.2: add backport fix for PR 32044

Daniel Mack
Hi,

gcc-4.3.2 needs a backported patch for issue PR 32044, below is a patch
that adds this to buildroot2.

Many thanks to Uwe Kleine-König for the backport and to Robert Schwebel
for pointing me there.

The original posting is here:

http://www.nabble.com/Backport-fix-for-PR-32044-(final-value-replacement-too-aggressive)-%09to-gcc-4.3-td22199271.html                       

Not sure yet whether this solves a number of strange crashes on a PXA
board, but the patch is certainly needed.

Best regards,
Daniel


Index: toolchain/gcc/4.3.2/901-backport-fix-for-bug-32044.patch
===================================================================
--- toolchain/gcc/4.3.2/901-backport-fix-for-bug-32044.patch (revision 0)
+++ toolchain/gcc/4.3.2/901-backport-fix-for-bug-32044.patch (revision 0)
@@ -0,0 +1,188 @@
+Index: gcc-4.3.2/gcc/tree-scalar-evolution.c
+===================================================================
+--- gcc-4.3.2.orig/gcc/tree-scalar-evolution.c 2009-01-28 10:14:37.000000000 +0100
++++ gcc-4.3.2/gcc/tree-scalar-evolution.c 2009-01-28 10:17:50.000000000 +0100
+@@ -2716,6 +2716,50 @@
+   scalar_evolution_info = NULL;
+ }
+
++/* Returns true if the expression EXPR is considered to be too expensive
++   for scev_const_prop.  */
++
++bool
++expression_expensive_p (tree expr)
++{
++  enum tree_code code;
++
++  if (is_gimple_val (expr))
++    return false;
++
++  code = TREE_CODE (expr);
++  if (code == TRUNC_DIV_EXPR
++      || code == CEIL_DIV_EXPR
++      || code == FLOOR_DIV_EXPR
++      || code == ROUND_DIV_EXPR
++      || code == TRUNC_MOD_EXPR
++      || code == CEIL_MOD_EXPR
++      || code == FLOOR_MOD_EXPR
++      || code == ROUND_MOD_EXPR
++      || code == EXACT_DIV_EXPR)
++    {
++      /* Division by power of two is usually cheap, so we allow it.
++ Forbid anything else.  */
++      if (!integer_pow2p (TREE_OPERAND (expr, 1)))
++ return true;
++    }
++
++  switch (TREE_CODE_CLASS (code))
++    {
++    case tcc_binary:
++    case tcc_comparison:
++      if (expression_expensive_p (TREE_OPERAND (expr, 1)))
++ return true;
++
++      /* Fallthru.  */
++    case tcc_unary:
++      return expression_expensive_p (TREE_OPERAND (expr, 0));
++
++    default:
++      return true;
++    }
++}
++
+ /* Replace ssa names for that scev can prove they are constant by the
+    appropriate constants.  Also perform final value replacement in loops,
+    in case the replacement expressions are cheap.
+@@ -2802,12 +2846,6 @@
+ continue;
+
+       niter = number_of_latch_executions (loop);
+-      /* We used to check here whether the computation of NITER is expensive,
+- and avoided final value elimination if that is the case.  The problem
+- is that it is hard to evaluate whether the expression is too
+- expensive, as we do not know what optimization opportunities the
+- the elimination of the final value may reveal.  Therefore, we now
+- eliminate the final values of induction variables unconditionally.  */
+       if (niter == chrec_dont_know)
+ continue;
+
+@@ -2838,7 +2876,15 @@
+      /* Moving the computation from the loop may prolong life range
+ of some ssa names, which may cause problems if they appear
+ on abnormal edges.  */
+-      || contains_abnormal_ssa_name_p (def))
++      || contains_abnormal_ssa_name_p (def)
++      /* Do not emit expensive expressions.  The rationale is that
++ when someone writes a code like
++
++ while (n > 45) n -= 45;
++
++ he probably knows that n is not large, and does not want it
++ to be turned into n %= 45.  */
++      || expression_expensive_p (def))
+    continue;
+
+  /* Eliminate the PHI node and replace it by a computation outside
+Index: gcc-4.3.2/gcc/tree-scalar-evolution.h
+===================================================================
+--- gcc-4.3.2.orig/gcc/tree-scalar-evolution.h 2009-01-28 10:22:47.000000000 +0100
++++ gcc-4.3.2/gcc/tree-scalar-evolution.h 2009-01-28 10:23:10.000000000 +0100
+@@ -35,6 +35,7 @@
+ extern void scev_analysis (void);
+ unsigned int scev_const_prop (void);
+
++bool expression_expensive_p (tree);
+ extern bool simple_iv (struct loop *, tree, tree, affine_iv *, bool);
+
+ /* Returns the loop of the polynomial chrec CHREC.  */
+Index: gcc-4.3.2/gcc/testsuite/gcc.dg/pr34027-1.c
+===================================================================
+--- gcc-4.3.2.orig/gcc/testsuite/gcc.dg/pr34027-1.c 2009-01-28 10:24:09.000000000 +0100
++++ gcc-4.3.2/gcc/testsuite/gcc.dg/pr34027-1.c 2009-01-28 10:24:43.000000000 +0100
+@@ -8,5 +8,9 @@
+   return ns;
+ }
+
+-/* { dg-final { scan-tree-dump "ns % 10000" "optimized" } } */
++/* This test was originally introduced to test that we transform
++   to ns % 10000.  See the discussion of PR 32044 why we do not do
++   that anymore.  */
++/* { dg-final { scan-tree-dump-times "%" 0 "optimized" } } */
++/* { dg-final { scan-tree-dump-times "/" 0 "optimized" } } */
+ /* { dg-final { cleanup-tree-dump "optimized" } } */
+Index: gcc-4.3.2/gcc/testsuite/gcc.dg/tree-ssa/pr32044.c
+===================================================================
+--- /dev/null 1970-01-01 00:00:00.000000000 +0000
++++ gcc-4.3.2/gcc/testsuite/gcc.dg/tree-ssa/pr32044.c 2009-01-28 10:25:50.000000000 +0100
+@@ -0,0 +1,55 @@
++/* { dg-do compile } */
++/* { dg-options "-O2 -fdump-tree-empty -fdump-tree-final_cleanup" } */
++
++int foo (int n)
++{
++  while (n >= 45)
++    n -= 45;
++
++  return n;
++}
++
++int bar (int n)
++{
++  while (n >= 64)
++    n -= 64;
++
++  return n;
++}
++
++int bla (int n)
++{
++  int i = 0;
++
++  while (n >= 45)
++    {
++      i++;
++      n -= 45;
++    }
++
++  return i;
++}
++
++int baz (int n)
++{
++  int i = 0;
++
++  while (n >= 64)
++    {
++      i++;
++      n -= 64;
++    }
++
++  return i;
++}
++
++/* The loops computing division/modulo by 64 should be eliminated.  */
++/* { dg-final { scan-tree-dump-times "Removing empty loop" 2 "empty" } } */
++
++/* There should be no division/modulo in the final dump (division and modulo
++   by 64 are done using bit operations).  */
++/* { dg-final { scan-tree-dump-times "/" 0 "final_cleanup" } } */
++/* { dg-final { scan-tree-dump-times "%" 0 "final_cleanup" } } */
++
++/* { dg-final { cleanup-tree-dump "empty" } } */
++/* { dg-final { cleanup-tree-dump "final_cleanup" } } */
+Index: gcc-4.3.2/gcc/tree-ssa-loop-ivopts.c
+===================================================================
+--- gcc-4.3.2.orig/gcc/tree-ssa-loop-ivopts.c 2009-01-28 10:26:04.000000000 +0100
++++ gcc-4.3.2/gcc/tree-ssa-loop-ivopts.c 2009-01-28 10:27:09.000000000 +0100
+@@ -3778,7 +3778,12 @@
+     return false;
+
+   cand_value_at (loop, cand, use->stmt, nit, &bnd);
++
+   *bound = aff_combination_to_tree (&bnd);
++  /* It is unlikely that computing the number of iterations using division
++     would be more profitable than keeping the original induction variable.  */
++  if (expression_expensive_p (*bound))
++    return false;
+   return true;
+ }
_______________________________________________
buildroot mailing list
[hidden email]
http://lists.busybox.net/mailman/listinfo/buildroot
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Re: gcc-4.3.2: add backport fix for PR 32044

Peter Korsgaard-3
>>>>> "Daniel" == Daniel Mack <[hidden email]> writes:

 Daniel> Hi,
 Daniel> gcc-4.3.2 needs a backported patch for issue PR 32044, below is a patch
 Daniel> that adds this to buildroot2.

Thanks - And 4.3.1 / 4.3.3 as well I guess?

--
Bye, Peter Korsgaard
_______________________________________________
buildroot mailing list
[hidden email]
http://lists.busybox.net/mailman/listinfo/buildroot
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Re: gcc-4.3.2: add backport fix for PR 32044

Daniel Mack
On Tue, Mar 10, 2009 at 10:21:20AM +0100, Peter Korsgaard wrote:
> >>>>> "Daniel" == Daniel Mack <[hidden email]> writes:
>
>  Daniel> Hi,
>  Daniel> gcc-4.3.2 needs a backported patch for issue PR 32044, below is a patch
>  Daniel> that adds this to buildroot2.
>
> Thanks - And 4.3.1 / 4.3.3 as well I guess?

I can't tell - only checked 4.3.2, sorry.

Daniel
_______________________________________________
buildroot mailing list
[hidden email]
http://lists.busybox.net/mailman/listinfo/buildroot
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Re: gcc-4.3.2: add backport fix for PR 32044

Peter Korsgaard-3
In reply to this post by Daniel Mack
>>>>> "Daniel" == Daniel Mack <[hidden email]> writes:

 Daniel> Hi,
 Daniel> gcc-4.3.2 needs a backported patch for issue PR 32044, below is a patch
 Daniel> that adds this to buildroot2.

Committed as r25601.

--
Bye, Peter Korsgaard
_______________________________________________
buildroot mailing list
[hidden email]
http://lists.busybox.net/mailman/listinfo/buildroot
Loading...