[Bug middle-end/126896] backward threader cannot duplicate a PHI-of-compares join block

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=126896

Richard Biener <rguenth at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
     Ever confirmed|0                           |1
   Last reconfirmed|                            |2026-08-17
             Status|UNCONFIRMED                 |NEW

--- Comment #1 from Richard Biener <rguenth at gcc dot gnu.org> ---
To clarify, DOM turns

  <bb 2> [local count: 1073741824]:
  if (x_3(D) != 0)
    goto <bb 3>; [50.00%]
  else
    goto <bb 4>; [50.00%]

  <bb 3> [local count: 536870912]:
  t_9 = a_7(D) < b_8(D);
  goto <bb 5>; [100.00%]

  <bb 4> [local count: 536870912]:
  t_6 = c_4(D) < d_5(D);

  <bb 5> [local count: 1073741824]:
  # t_1 = PHI <t_9(3), t_6(4)>
  if (t_1 != 0)
    goto <bb 6>; [33.00%]
  else
    goto <bb 7>; [67.00%]

  <bb 6> [local count: 354334800]:
  g ();

  <bb 7> [local count: 1073741824]:
  return;

into

  <bb 2> [local count: 1073741824]:
  if (x_3(D) != 0)
    goto <bb 3>; [50.00%]
  else
    goto <bb 4>; [50.00%]

  <bb 3> [local count: 536870912]:
  t_9 = a_7(D) < b_8(D);
  if (t_9 != 0)
    goto <bb 5>; [0.00%]
  else
    goto <bb 6>; [100.00%]

  <bb 4> [local count: 536870912]:
  t_6 = c_4(D) < d_5(D);
  if (t_6 != 0)
    goto <bb 5>; [66.00%]
  else
    goto <bb 6>; [34.00%]

  <bb 5> [local count: 354334800]:
  g ();

  <bb 6> [local count: 1073741824]:
  return;


This isn't jump threading.  It looks more like sth for phiopt or tail
duplication?

 t_9 = a_7(D) < b_8(D);
..
 t_6 = c_4(D) < d_5(D);
..
 # t_1 = PHI <t_9, t_6>
 if (t_1 != 0)


I suspect there's some odd special case code in DOM to register this as
"jump thread" just to get the if (t != 0) block duplicated.