tree-data-ref.c (dr_analyze_alias): Handle case smt is NULL.

	* tree-data-ref.c (dr_analyze_alias): Handle case smt is NULL.
	* tree-predcom.c (mark_virtual_ops_for_renaming): Exported.
	* tree-ssa-loop-prefetch.c: Include optabs.h.
	(FENCE_FOLLOWING_MOVNT): New macro.
	(struct mem_ref): Add independent_p and storent_p fields.
	(record_ref): Initalize the new fields.
	(gather_memory_references_ref): Return true if the reference
	could be analysed.
	(gather_memory_references): Check whether all memory accesses
	in loop were recorded.
	(should_issue_prefetch_p): Return false for nontemporal stores.
	(nontemporal_store_p, mark_nontemporal_store, emit_mfence_after_loop,
	may_use_storent_in_loop_p, mark_nontemporal_stores): New functions.
	(determine_loop_nest_reuse): Detect independent memory references.
	(loop_prefetch_arrays): Call mark_nontemporal_stores.
	* tree-flow.h (mark_virtual_ops_for_renaming): Declare.
	* Makefile.in (tree-ssa-loop-prefetch.o): Add OPTABS_H dependency.
	* config/i386/i386.h (x86_mfence): Declare.
	(FENCE_FOLLOWING_MOVNT): Return x86_mfence.
	* config/i386/i386.c (x86_mfence): New variable.
	(ix86_init_mmx_sse_builtins): Initialize x86_mfence.

	* tree-pretty-print.c (dump_generic_node): Mark nontemporal stores.
	* optabs.c (init_optabs): Initialize storent_optab.
	* optabs.h (enum optab_index): Add OTI_storent.
	(storent_optab): Declare.
	* genopinit.c (optabs): Add initialization for storent_optab.
	* tree.h (MOVE_NONTEMPORAL): New macro.
	* expr.c (expand_assignment, store_expr, store_constructor_field,
	store_constructor, store_field, expand_expr_real_1): Propagate
	nontemporality of the expanded store.
	(emit_storent_insn): New function.
	* expr.h (expand_assignment, store_expr): Declaration changed.
	* function.c (assign_parm_setup_reg): Pass false as nontemporality
	to expand_assignment.
	* stmt.c (expand_asm_expr): Ditto.
	* calls.c (initialize_argument_information): Pass false as
	nontemporality to store_expr.
	* config/i386/sse.md (storentv4sf, storentv2df, storentv2di,
	storentsi): New.

	* gcc.dg/tree-ssa/prefetch-7.c: New test.

From-SVN: r125604

1 files changed, 59 insertions, 0 deletions

diff --git a/gcc/testsuite/gcc.dg/tree-ssa/prefetch-7.c b/gcc/testsuite/gcc.dg/tree-ssa/prefetch-7.c
new file mode 100644
index 0000000..510dee0
--- /dev/null
+++ b/gcc/testsuite/gcc.dg/tree-ssa/prefetch-7.c
@@ -0,0 +1,59 @@
+/* { dg-do compile { target i?86-*-* x86_64-*-* } } */
+/* { dg-require-effective-target ilp32 } */
+/* { dg-options "-O2 -fprefetch-loop-arrays -march=athlon -msse2 -mfpmath=sse --param simultaneous-prefetches=100 --param max-unrolled-insns=1 -fdump-tree-aprefetch-details -fdump-tree-final_cleanup" } */
+
+#define K 1000000
+int a[K], b[K];
+
+void test(int *p)
+{
+  unsigned i;
+
+  /* Nontemporal store should be used for a.  */
+  for (i = 0; i < K; i++)
+    a[i] = 0;
+
+  /* Nontemporal store should be used for a, nontemporal prefetch for b.  */
+  for (i = 0; i < K; i++)
+    a[i] = b[i];
+
+  /* Nontemporal store should not be used here (only write and read temporal
+     prefetches).  */
+  for (i = 0; i < K - 10000; i++)
+    a[i + 10000] = a[i];
+
+  /* Nontemporal store should not be used here (only write and read nontemporal
+     prefetches).  */
+  for (i = 0; i < K - 100000; i++)
+    a[i + 100000] = a[i];
+
+  /* Nontemporal store should be used neither for a nor for p, as we do not know
+     whether they alias or not.  */
+  for (i = 0; i < K; i++)
+    {
+      a[i] = 0;
+      *p++ = 1;
+    }
+
+  /* Nontemporal store should not be used for a, as we do not know whether its
+     value will be reused or not.  */
+  for (i = 0; i < 1000; i++)
+    a[i] = 0;
+}
+
+/* { dg-final { scan-tree-dump-times "Issued prefetch" 5 "aprefetch" } } */
+/* { dg-final { scan-tree-dump-times "Issued nontemporal prefetch" 3 "aprefetch" } } */
+/* { dg-final { scan-tree-dump-times "nontemporal store" 2 "aprefetch" } } */
+
+/* { dg-final { scan-tree-dump-times "builtin_prefetch" 8 "final_cleanup" } } */
+/* { dg-final { scan-tree-dump-times "=\\{nt\\}" 2 "final_cleanup" } } */
+/* { dg-final { scan-tree-dump-times "__builtin_ia32_mfence" 2 "final_cleanup" } } */
+
+/* { dg-final { scan-assembler-times "prefetchw" 5 } } */
+/* { dg-final { scan-assembler-times "prefetcht" 1 } } */
+/* { dg-final { scan-assembler-times "prefetchnta" 2 } } */
+/* { dg-final { scan-assembler-times "movnti" 2 } } */
+/* { dg-final { scan-assembler-times "mfence" 2 } } */
+
+/* { dg-final { cleanup-tree-dump "aprefetch" } } */
+/* { dg-final { cleanup-tree-dump "final_cleanup" } } */