Hi!
I've noticed this PR is still open, but has been fixed, though
apparently Bernd's patch from the PR has been applied as is with
the needed tab eaten (instead 7 spaces) and no testcase has been added.
Tested with cross-compiler and the r240625 change reverted and as is
and running the resulting assembler on ppc64le.
Ok for trunk?
2016-11-09 Jakub Jelinek <ja...@redhat.com>
PR target/77718
* builtins.c (expand_builtin_memcmp): Formatting fix.
* gcc.c-torture/execute/pr77718.c: New test.
--- gcc/ChangeLog.jj 2016-11-09 15:22:28.000000000 +0100
+++ gcc/ChangeLog 2016-11-09 16:29:35.152056326 +0100
@@ -5027,7 +5027,8 @@
2016-09-29 Bernd Schmidt <bschm...@redhat.com>
- * builtins.c (expand_builtin_memcmp): don't swap args unless
+ PR target/77718
+ * builtins.c (expand_builtin_memcmp): Don't swap args unless
result is only being compared with zero.
2016-09-29 Marek Polacek <pola...@redhat.com>
--- gcc/builtins.c.jj 2016-10-31 13:28:06.000000000 +0100
+++ gcc/builtins.c 2016-11-09 16:19:08.886909150 +0100
@@ -3754,7 +3754,7 @@ expand_builtin_memcmp (tree exp, rtx tar
{
src_str = c_getstr (arg1);
if (src_str != NULL)
- std::swap (arg1_rtx, arg2_rtx);
+ std::swap (arg1_rtx, arg2_rtx);
}
/* If SRC is a string constant and block move would be done
--- gcc/testsuite/gcc.c-torture/execute/pr77718.c.jj 2016-11-09
16:28:30.414868074 +0100
+++ gcc/testsuite/gcc.c-torture/execute/pr77718.c 2016-11-09
16:22:06.000000000 +0100
@@ -0,0 +1,25 @@
+/* PR middle-end/77718 */
+
+char a[64] __attribute__((aligned (8)));
+
+__attribute__((noinline, noclone)) int
+foo (void)
+{
+ return __builtin_memcmp ("bbbbbb", a, 6);
+}
+
+__attribute__((noinline, noclone)) int
+bar (void)
+{
+ return __builtin_memcmp (a, "bbbbbb", 6);
+}
+
+int
+main ()
+{
+ __builtin_memset (a, 'a', sizeof (a));
+ if (((foo () < 0) ^ ('a' > 'b'))
+ || ((bar () < 0) ^ ('a' < 'b')))
+ __builtin_abort ();
+ return 0;
+}
Jakub
