Incorrect results with XNOR/shift expression

[veripoolbot]

---

Author Name: Clifford Wolf (@cliffordwolf)
Original Redmine Issue: 776 from https://www.veripool.org
Original Date: 2014-05-25
Original Assignee: Wilson Snyder (@wsnyder)

---

This should always return y=4'b1111, but Verilator f705f9b only does this for a=0:

 module issue_053(a, y);
   input [3:0] a;
   output [3:0] y;
   assign y = (a >> a) ^~ (a >> a);
 endmodule

Self-contained test case:

http://svn.clifford.at/handicraft/2014/verilatortest/test019.v

http://svn.clifford.at/handicraft/2014/verilatortest/test019.cc

http://svn.clifford.at/handicraft/2014/verilatortest/test019.sh

[veripoolbot]

---

Original Redmine Comment
Author Name: Wilson Snyder (@wsnyder)
Original Date: 2014-05-26T22:19:15Z

---

Fixed in git towards 3.861.

There is a specific optimization for "(a>>b) op (c>>b)" -> "(a op c)>>b" which shouldn't have been applied to an XNOR.

[veripoolbot]

---

Original Redmine Comment
Author Name: Wilson Snyder (@wsnyder)
Original Date: 2014-06-11T00:58:18Z

---

In 3.862.