RFR: 8324790: ifnode::fold_compares_helper cleanup

[Emanuel Peter]

On Fri, 26 Jan 2024 23:31:00 GMT, Joshua Cao <duke at openjdk.org> wrote:

> I hope my assumptions  in `filtered_int_type` are correct here:
> 
> * we assert that `if_proj` is an `IfTrue` or `IfFalse`, so it is safe to assume `if_proj->_in` is an  `IfNode`
> * the 1'th input of a CmpNode is a BoolNode
> * Tthe 1'th input of an IfNode is **not always a BoolNode**, it can be a constant. We need to leave this check in.
> 
> We also remove a some of the if-checks in `compare_folds_cleanup` which seem unnecessary.
> 
> Passes tier1 locally.

I have not yet looked at your code. But I'm not sure your assumptions are all correct:

> we assert that if_proj is an IfTrue or IfFalse, so it is safe to assume if_proj->_in is an IfNode

Generally, during IGVN, ctrl inputs could always be TOP, if we are in a dead-code place. Are you sure there is always an IfNode above the projections? Maybe that is true in this specific call chain?

> the 1'th input of a CmpNode is a BoolNode

Do you mean the 1st input of BoolNode is a CmpNode? I think we may also fold CmpNodes to constants, or TOP. Is that somehow excluded in your cases?

The problem with testing of this: getting TOP inputs is kinda rare. But you can make it more likely with `StressIGVN`.

-------------

PR Comment: https://git.openjdk.org/jdk/pull/17601#issuecomment-1931607095