Hi!

As mentioned in the PR, this hunk is guarded with !wi::neg_p (r1val | r1mask, 
sgn)
which means if sgn is UNSIGNED, it is always true, but r1val | r1mask in
widest_int is still sign-extended.  That means wi::clz (arg) returns 0,
wi::get_precision (arg) returns some very large number
(WIDE_INT_MAX_PRECISION, on x86_64 576 bits) and width is 64, so we end up
with lzcount of -512 where the code afterwards expects a non-negative
lzcount.  For arg without the sign bit set the code works right, those
numbers are zero extended and so wi::clz must return wi::get_precision (arg) - 
width
plus number of leading zero bits within the width precision.
The patch fixes it by handling the sign-extension specially, either it could
be done through wi::neg_p (arg) check, but lzcount == 0 works identically.

Bootstrapped/regtested on x86_64-linux and i686-linux, ok for trunk?

2021-08-31  Jakub Jelinek  <ja...@redhat.com>

        PR tree-optimization/102134
        * tree-ssa-ccp.c (bit_value_binop) <case RSHIFT_EXPR>: If sgn is
        UNSIGNED and r1val | r1mask has MSB set, ensure lzcount doesn't
        become negative.

        * gcc.c-torture/execute/pr102134.c: New test.

--- gcc/tree-ssa-ccp.c.jj       2021-08-30 08:36:11.302515439 +0200
+++ gcc/tree-ssa-ccp.c  2021-08-30 22:49:21.957503630 +0200
@@ -1695,7 +1695,8 @@ bit_value_binop (enum tree_code code, si
              /* Logical right shift, or zero sign bit.  */
              widest_int arg = r1val | r1mask;
              int lzcount = wi::clz (arg);
-             lzcount -= wi::get_precision (arg) - width;
+             if (lzcount)
+               lzcount -= wi::get_precision (arg) - width;
              widest_int tmp = wi::mask <widest_int> (width, false);
              tmp = wi::lrshift (tmp, lzcount);
              tmp = wi::lrshift (tmp, wi::bit_and_not (r2val, r2mask));
--- gcc/testsuite/gcc.c-torture/execute/pr102134.c.jj   2021-08-30 
22:47:41.115920522 +0200
+++ gcc/testsuite/gcc.c-torture/execute/pr102134.c      2021-08-30 
22:47:20.811205820 +0200
@@ -0,0 +1,23 @@
+/* PR tree-optimization/102134 */
+
+typedef unsigned long long u64;
+
+u64 g;
+
+void
+foo (u64 a, u64 b, u64 c, u64 *r)
+{
+  b *= b;
+  u64 x = a && ((b >> (c & 63)) | ((b << (c & 63)) & g));
+  *r = x + a;
+}
+
+int
+main ()
+{
+  u64 x;
+  foo (1, 3000, 0, &x);
+  if (x != 2)
+    __builtin_abort ();
+  return 0;
+}

        Jakub

Reply via email to