On Thu, Oct 08, 2020 at 11:46:36AM +0200, Aldy Hernandez via Gcc-patches wrote:
> This was a pasto by yours truly while porting the __builtin code from
> vr_values::extract_range_basic.
>
> I've tested Andrew's patch, and pushed it.
Note for next time, I think you should use --author when committing a patch
on someone else's behalf.
And no testcase was included, I'm including one below.
Anyway, this PR and the other CTZ related discussions led me to discover a
bug I've made earlier, CLZ/CTZ builtins have unsigned arguments and e.g.
both the vr-values.cc and now gimple-range.cc code heavily relies on that,
but __builtin_ffs has a signed operand and this optimization was incorrectly
making the operand signed too, so I guess it would greatly confuse VRP in
some cases.
Ok for trunk if it passes bootstrap/regtest?
2020-10-08 Jakub Jelinek <ja...@redhat.com>
PR tree-optimization/97325
* match.pd (FFS(nonzero) -> CTZ(nonzero) + 1): Cast argument to
corresponding unsigned type.
* gcc.c-torture/execute/pr97325.c: New test.
--- gcc/match.pd.jj 2020-10-01 10:40:10.057756683 +0200
+++ gcc/match.pd 2020-10-08 14:39:27.347249559 +0200
@@ -6196,7 +6196,8 @@ (define_operator_list COND_TERNARY
(if (INTEGRAL_TYPE_P (TREE_TYPE (@0))
&& direct_internal_fn_supported_p (IFN_CTZ, TREE_TYPE (@0),
OPTIMIZE_FOR_SPEED))
- (plus (CTZ:type @0) { build_one_cst (type); })))
+ (with { tree utype = unsigned_type_for (TREE_TYPE (@0)); }
+ (plus (CTZ:type (convert:utype @0)) { build_one_cst (type); }))))
#endif
(for ffs (BUILT_IN_FFS BUILT_IN_FFSL BUILT_IN_FFSLL
--- gcc/testsuite/gcc.c-torture/execute/pr97325.c.jj 2020-10-08
14:29:59.740440126 +0200
+++ gcc/testsuite/gcc.c-torture/execute/pr97325.c 2020-10-08
14:27:22.518708841 +0200
@@ -0,0 +1,15 @@
+/* PR tree-optimization/97325 */
+
+unsigned long long
+foo (unsigned long long c)
+{
+ return c ? __builtin_ffs (-(unsigned short) c) : 0;
+}
+
+int
+main ()
+{
+ if (foo (2) != 2)
+ __builtin_abort ();
+ return 0;
+}
Jakub
