Hi! Usually I bootstrap/regtest without graphite, so haven't seen these 3 test FAILs with my recent patch. Fixed thusly, tested on x86_64-linux, committed to trunk as obvious.
2013-03-14 Jakub Jelinek <ja...@redhat.com> PR tree-optimization/53265 * gcc.dg/graphite/scop-3.c (toto): Increase array size to avoid undefined behavior. * gcc.dg/graphite/id-6.c (test): Likewise. * gcc.dg/graphite/pr35356-2.c: Adjust regexp patterns to only look for MIN_EXPR and MAX_EXPR in GIMPLE stmts. --- gcc/testsuite/gcc.dg/graphite/scop-3.c.jj 2009-08-03 09:42:19.000000000 +0200 +++ gcc/testsuite/gcc.dg/graphite/scop-3.c 2013-03-14 11:37:25.719648369 +0100 @@ -1,7 +1,7 @@ int toto() { int i, j, k; - int a[100][100]; + int a[100][200]; int b[100]; for (i = 1; i < 100; i++) --- gcc/testsuite/gcc.dg/graphite/id-6.c.jj 2009-08-03 09:42:19.000000000 +0200 +++ gcc/testsuite/gcc.dg/graphite/id-6.c 2013-03-14 11:37:36.628586994 +0100 @@ -2,7 +2,7 @@ void foo (int); int test () { - int a[N]; + int a[N + 8]; unsigned i; for (i = 0; i < N; i++) --- gcc/testsuite/gcc.dg/graphite/pr35356-2.c.jj 2011-02-15 15:31:56.000000000 +0100 +++ gcc/testsuite/gcc.dg/graphite/pr35356-2.c 2013-03-14 11:47:54.655777100 +0100 @@ -39,6 +39,6 @@ foo (int bar, int n, int k) */ -/* { dg-final { scan-tree-dump-times "MIN_EXPR" 4 "graphite" } } */ -/* { dg-final { scan-tree-dump-times "MAX_EXPR" 4 "graphite" } } */ +/* { dg-final { scan-tree-dump-times "MIN_EXPR\[^\\n\\r]*;" 4 "graphite" } } */ +/* { dg-final { scan-tree-dump-times "MAX_EXPR\[^\\n\\r]*;" 4 "graphite" } } */ /* { dg-final { cleanup-tree-dump "graphite" } } */ Jakub