[Mlir-commits] [mlir] [MLIR] Fix computeTiedSuccessorInputs for single-iteration scf.for (PR #188986)

[Matthias Springer]

matthias-springer wrote:

I took another look at this PR. I think it would mark `%0` and `%arg0` as "tied" in the following example, but I don't see why they would be tied. It depends on the semantics of the op.
```mlir
%0 = region_based_op(%val) ({
^bb0(%arg0):  // entry region
  yield %x    // back to parent
})
```

[Here's](https://github.com/llvm/llvm-project/pull/216627) another PR that may solve the issue. I'm not sure if it actually fixes all the test cases.


https://github.com/llvm/llvm-project/pull/188986