backport: re PR tree-optimization/58246 (wrong code at -O1 and above)

2014-05-07  Richard Biener  <rguenther@suse.de>

	Backport from mainline
	2013-08-29  Richard Biener  <rguenther@suse.de>

	PR tree-optimization/58246
	* tree-ssa-dce.c (mark_aliased_reaching_defs_necessary_1): Properly
	handle the dominance check inside a basic-block.

	* gcc.dg/torture/pr58246.c: New testcase.

From-SVN: r210141
This commit is contained in:
Richard Biener 2014-05-07 07:59:26 +00:00 committed by Richard Biener
parent b4cd4627a0
commit 7a6547fc50
4 changed files with 43 additions and 0 deletions

View File

@ -1,3 +1,12 @@
2014-05-07 Richard Biener <rguenther@suse.de>
Backport from mainline
2013-08-29 Richard Biener <rguenther@suse.de>
PR tree-optimization/58246
* tree-ssa-dce.c (mark_aliased_reaching_defs_necessary_1): Properly
handle the dominance check inside a basic-block.
2014-05-06 Richard Biener <rguenther@suse.de>
Backport from mainline

View File

@ -1,3 +1,11 @@
2014-05-07 Richard Biener <rguenther@suse.de>
Backport from mainline
2013-08-29 Richard Biener <rguenther@suse.de>
PR tree-optimization/58246
* gcc.dg/torture/pr58246.c: New testcase.
2014-05-06 Richard Biener <rguenther@suse.de>
Backport from mainline

View File

@ -0,0 +1,21 @@
/* { dg-do run } */
extern void abort (void);
int a, b;
int main ()
{
int t[2] = {1,1};
for (a = 0; a < 2; a++)
{
b ^= t[a];
t[a] = t[1] = 0;
}
if (b != 1)
abort ();
return 0;
}

View File

@ -576,6 +576,11 @@ mark_aliased_reaching_defs_necessary_1 (ao_ref *ref, tree vdef, void *data)
in the references (gcc.c-torture/execute/pr42142.c).
The simplest way is to check if the kill dominates
the use. */
/* But when both are in the same block we cannot
easily tell whether we came from a backedge
unless we decide to compute stmt UIDs
(see PR58246). */
&& (basic_block) data != gimple_bb (def_stmt)
&& dominated_by_p (CDI_DOMINATORS, (basic_block) data,
gimple_bb (def_stmt))
&& operand_equal_p (ref->ref, lhs, 0))