[DAG] SelectionDAG::canCreateUndefOrPoison - add ISD::FABS/ handling + tests

[RKSimon]

Pulled out of #148796 - Add ISD::FABS to SelectionDAG::canCreateUndefOrPoison and add suitable test coverage

[Shoreshen]

Hi @RKSimon , I'm still trying to find a suitable test. Most of the ll test I tried reach the maximum search depth. I tried to add freeze in between but it optimized out. May I ask if you have any suggestion?? Thanks

[RKSimon]

How about this? https://zig.godbolt.org/z/q866cb744 (you need as target that doesn't lower fabs to bitlogic)

[Shoreshen]

How about this? https://zig.godbolt.org/z/q866cb744 (you need as target that doesn't lower fabs to bitlogic)

the freeze will be optimize out...... but I think I can try to build up a poisoned vector and try, thanks a lot ：)

[RKSimon]

define <4 x float> @src(<4 x float> %A, <4 x float> %B) {
  %A0 = call <4 x float> @llvm.fabs.v4f32(<4 x float> %A)
  %F1 = freeze <4 x float> %A0
  %A1 = call <4 x float> @llvm.fabs.v4f32(<4 x float> %F1)
  ret <4 x float> %A1
}
define <4 x float> @tgt(<4 x float> %A, <4 x float> %B) {
  %F = freeze <4 x float> %A
  %A0 = call <4 x float> @llvm.fabs.v4f32(<4 x float> %F)
  %A1 = call <4 x float> @llvm.fabs.v4f32(<4 x float> %A0)
  ret <4 x float> %A1
}

Isn't the point of the test that it will optimize out? By adding ISD::FABS to canCreateUndefOrPoison then src will fold to tgt (and the second fabs will then be removed)

[Shoreshen]

@RKSimon Hi you are right. Sorry my bad... I made myself confused...

[RKSimon]

Fixed by #149440 / 5fe9b52