[Bug tree-optimization/106126] [12 Regression] tree check fail in useless_type_conversion_p, at gimple-expr.cc:87 since r13-1184-g57424087e82db140

[marxin at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=106126

Martin Liška <marxin at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |ASSIGNED

--- Comment #14 from Martin Liška <marxin at gcc dot gnu.org> ---
(In reply to luoxhu from comment #13)
> Otherwise we need record first_bb when conditions_in_bbs->is_empty, then
> check that in is_beneficial, ordered_remove the info entry if that bb is not
> the first "if condition" with side_effect statement in it, the fix would be

No, we need to record if a BB has a side effect and allow only side effects for
a first BB in the main loop in pass_if_to_switch::execute. I'm testing a patch
candidate.

> as below, but I am not sure whether it is worth&correct way doing this to
> handle both PR105740 and PR106126?

Well, I replied in PR105740 where the problem is one needs to run the pass
later.