On Tue, 20 Mar 2018, Jakub Jelinek wrote:
> On Tue, Mar 20, 2018 at 01:42:04PM +0100, Richard Biener wrote:
> > 2018-03-20 Richard Biener <[email protected]>
> >
> > * testsuite/libgomp.graphite/force-parallel-4.c: XFAIL one
> > parallelizable loop.
> >
> > Index: libgomp/testsuite/libgomp.graphite/force-parallel-4.c
> > ===================================================================
> > --- libgomp/testsuite/libgomp.graphite/force-parallel-4.c (revision
> > 258678)
> > +++ libgomp/testsuite/libgomp.graphite/force-parallel-4.c (working copy)
> > @@ -46,7 +46,10 @@ int main(void)
> > return 0;
> > }
> >
> > -/* Check that parallel code generation part make the right answer. */
> > -/* { dg-final { scan-tree-dump-times "2 loops carried no dependency" 1
> > "graphite" } } */
> > +/* Check that parallel code generation part make the right answer.
> > + ??? XFAILed for i1 because conditional store elimination wrecks
> > + our dependence representation. */
> > +/* { dg-final { scan-tree-dump-times "2 loops carried no dependency" 1
> > "graphite" { xfail *-*-* } } } */
> > +/* { dg-final { scan-tree-dump-times "1 loops carried no dependency" 1
> > "graphite" } } */
>
> Shouldn't this line be then "\[12] loops carried no dependency" 1 "graphite"
> } } */
> so that when the previous starts XPASSing, we don't actually get a new FAIL?
I think we want to know this and adjust the testcase then to not go
back to 1 loop with no dependence.
> > /* { dg-final { scan-tree-dump-times "loopfn.0" 4 "optimized" } } */
> > /* { dg-final { scan-tree-dump-times "loopfn.1" 4 "optimized" } } */
