[Mlir-commits] [mlir] [mlir][IntRange] Poison support in int-range analysis (PR #152932)

[Ivan Butygin]

Hardcode84 wrote:

> And am curious about how ValueTracking.cpp or KnownBits down in LLVM handle this sort of thing.

It's not on-to-one mapping, but I think they also model poison with empty range and at least out intersect/union definitions are consistent https://github.com/llvm/llvm-project/blob/main/llvm/lib/IR/ConstantRange.cpp#L702

https://github.com/llvm/llvm-project/pull/152932