Passing `SelfMultiply` information in `CodeGen/SelectionDAG/TargetLowering.cpp`

We pass the `SelfMultiply` information into `KnownBits::mul` here
https://github.com/llvm/llvm-project/blob/2db6978f724bfb0f9ae6f37b8ea585c707e65ce8/llvm/lib/CodeGen/SelectionDAG/SelectionDAG.cpp#L3672-L3691

We also test if we are self multiplying here (`Op.getOperand(0) == Op.getOperand(1)`):
https://github.com/llvm/llvm-project/blob/3a30af978df0b6f39505fd702f4bfd55b3adb96c/llvm/lib/CodeGen/SelectionDAG/TargetLowering.cpp#L2925-L2946
But this information is not passed later on into `KnownBits::mul` here, so some additional known bits related to squaring may be missed:
https://github.com/llvm/llvm-project/blob/3a30af978df0b6f39505fd702f4bfd55b3adb96c/llvm/lib/CodeGen/SelectionDAG/TargetLowering.cpp#L3059-L3066

Read the SelfMultiply handling in llvm/lib/CodeGen/SelectionDAG/TargetLowering.cpp around the referenced ranges, and compare it with the KnownBits::mul call in llvm/lib/CodeGen/SelectionDAG/SelectionDAG.cpp. Trace how the self-multiplication information is carried between these entry points; done means the later KnownBits::mul call receives the information and the related known bits are covered by validation.

Written by the indexing model from the issue text.