[gcc(refs/vendors/redhat/heads/gcc-8-branch)] tree-optimization/99954 - fix loop distribution memcpy classification

[Jakub Jelinek <jakub@gcc.gnu.org>]

https://gcc.gnu.org/g:42f6f70f0a81257f715a96bb7f075d2485af4a1f

commit 42f6f70f0a81257f715a96bb7f075d2485af4a1f
Author: Richard Biener <rguenther@suse.de>
Date:   Wed Apr 7 13:17:05 2021 +0200

    tree-optimization/99954 - fix loop distribution memcpy classification
    
    This fixes bogus classification of a copy as memcpy.  We cannot use
    plain dependence analysis to decide between memcpy and memmove when
    it computes no dependence.  Instead we have to try harder later which
    the patch does for the gcc.dg/tree-ssa/ldist-24.c testcase by resorting
    to tree-affine to compute the difference between src and dest and
    compare against the copy size.
    
    2021-04-07  Richard Biener  <rguenther@suse.de>
    
            PR tree-optimization/99954
            * tree-loop-distribution.c: Include tree-affine.h.
            (generate_memcpy_builtin): Try using tree-affine to prove
            non-overlap.
            (loop_distribution::classify_builtin_ldst): Always classify
            as PKIND_MEMMOVE.
    
            * gcc.dg/torture/pr99954.c: New testcase.
    
    (cherry picked from commit b091cb1efa1881e93fb2e264daaab8876acf6800)

Diff:
---
 gcc/testsuite/gcc.dg/torture/pr99954.c | 30 ++++++++++++++++++++++++++++++
 gcc/tree-loop-distribution.c           | 17 +++++++++++++++--
 2 files changed, 45 insertions(+), 2 deletions(-)

diff --git a/gcc/testsuite/gcc.dg/torture/pr99954.c b/gcc/testsuite/gcc.dg/torture/pr99954.c
new file mode 100644
index 00000000000..7d447035912
--- /dev/null
+++ b/gcc/testsuite/gcc.dg/torture/pr99954.c
@@ -0,0 +1,30 @@
+/* { dg-do run } */
+
+#include <assert.h>
+
+#define CONTAINER_KIND union
+
+typedef CONTAINER_KIND container { int value; } container;
+
+void move(container* end, container* start) {
+    container* p;
+    for (p = end; p > start; p--) {
+	(p)->value = (p-1)->value;
+    }
+}
+
+#define N 100
+
+int main(int argc, char* argv[]) {
+    container vals[N];
+    int i;
+    for (i=0; i<N; i++) {
+        vals[i].value = argc + i;
+    }
+    move(&vals[N-1], &vals[0]);
+    assert(vals[0].value == argc + 0);
+    for (i=1; i<N; i++) {
+        assert(vals[i].value == argc + i - 1);
+    }
+    return 0;
+}
diff --git a/gcc/tree-loop-distribution.c b/gcc/tree-loop-distribution.c
index e257c9bc433..761d577f390 100644
--- a/gcc/tree-loop-distribution.c
+++ b/gcc/tree-loop-distribution.c
@@ -116,6 +116,7 @@ along with GCC; see the file COPYING3.  If not see
 #include "params.h"
 #include "tree-vectorizer.h"
 #include "tree-eh.h"
+#include "tree-affine.h"
 
 
 #define MAX_DATAREFS_NUM \
@@ -1060,6 +1061,18 @@ generate_memcpy_builtin (struct loop *loop, partition *partition)
     kind = BUILT_IN_MEMCPY;
   else
     kind = BUILT_IN_MEMMOVE;
+  /* Try harder if we're copying a constant size.  */
+  if (kind == BUILT_IN_MEMMOVE && poly_int_tree_p (nb_bytes))
+    {
+      aff_tree asrc, adest;
+      tree_to_aff_combination (src, ptr_type_node, &asrc);
+      tree_to_aff_combination (dest, ptr_type_node, &adest);
+      aff_combination_scale (&adest, -1);
+      aff_combination_add (&asrc, &adest);
+      if (aff_comb_cannot_overlap_p (&asrc, wi::to_poly_widest (nb_bytes),
+				     wi::to_poly_widest (nb_bytes)))
+	kind = BUILT_IN_MEMCPY;
+    }
 
   dest = force_gimple_operand_gsi (&gsi, dest, true, NULL_TREE,
 				   false, GSI_CONTINUE_LINKING);
@@ -1591,11 +1604,11 @@ classify_builtin_ldst (loop_p loop, struct graph *rdg, partition *partition,
   /* Now check that if there is a dependence.  */
   ddr_p ddr = get_data_dependence (rdg, src_dr, dst_dr);
 
-  /* Classify as memcpy if no dependence between load and store.  */
+  /* Classify as memmove if no dependence between load and store.  */
   if (DDR_ARE_DEPENDENT (ddr) == chrec_known)
     {
       partition->builtin = alloc_builtin (dst_dr, src_dr, base, src_base, size);
-      partition->kind = PKIND_MEMCPY;
+      partition->kind = PKIND_MEMMOVE;
       return;
     }