[Bug tree-optimization/61607] DOM missed jump threading and destroyed loops

[rguenth at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=61607

--- Comment #6 from Richard Biener <rguenth at gcc dot gnu.org> ---
The bogus loop cancelling is fixed as well as the equivalence recording.  Still
DOM does

  Registering jump thread: (3, 4) incoming edge;  (4, 5) joiner;  (5, 6)
normal;
  Registering jump thread: (5, 7) incoming edge;  (7, 9) normal;
  Registering jump thread: (2, 4) incoming edge;  (4, 5) joiner;  (5, 7)
normal;
  Cancelling jump thread: (2, 4) incoming edge;  (4, 5) joiner;  (5, 7) normal;
  Threaded jump 5 --> 7 to 10

with result

  <bb 5>:
  # inter_I_lsm.3_24 = PHI <inter_I_lsm.3_10(4)>
  # inter_I_lsm.4_25 = PHI <inter_I_lsm.4_11(4)>
  # inter_I_lsm.5_26 = PHI <inter_I_lsm.5_12(4)>
  # inter_I_lsm.6_27 = PHI <inter_I_lsm.4_11(4)>
  if (inter_I_lsm.4_25 != 0)
    goto <bb 6>;
  else
    goto <bb 10>;

  <bb 6>:
  inter[0] = inter_I_lsm.3_24;

  <bb 7>:
  if (inter_I_lsm.4_25 != 0)
    goto <bb 8>;
  else
    goto <bb 9>;

  <bb 8>:
  inter[1] = inter_I_lsm.5_26;

  <bb 9>:
  foo (&inter);
  inter ={v} {CLOBBER};
  return;

  <bb 10>:
  goto <bb 9>;

but no attempt at threading of

  (5, 6) incoming edge -> (6->7) joiner -> (7 -> 8) normal

VRP simplfies the conditional to true afterwards - so maybe this is just
a missed simplification-after-threading in DOM?  Without VRP the
redundant conditional stays.  It still looks like a trivial missing thing to
adjust the jump we thread through ...?