Shifted the file opening utility and environment variable

Shifted opening file utility to RunnerUtils.cpp

Using environment variable and llvm::itostr

Trivial nit changes.

@ftynse @herhut Can you review this patch?

Can someone review this patch?
I've addressed the Windows build failures resulting earlier.

Shifted execution test case within mlir/test/mlir-cpu-runner

Updated integer to string conversion logic.

Trivial changes to test case

I've rebased the current revision on HEAD and resolved the assert failures.

Rebased to latest HEAD.
createBlock API requires argument location too and hence lead to assertion failure.
Fixed the same and brought a couple more declarations closer to their uses.

If there are no more comments, can this revision be landed soon?

Handled review comments.

Making early exit as early as possible before starting to build vectors/maps.

I've addressed all the useful review comments made so far.

Hi @mehdi_amini @ftynse
The review comments were useful and I've made the said updates in the revision.
You may re-review/accept/land this revision.

Addressed review comments.
Two major changes to notice are :-

1. Updates were made to use yield op's block argument number instead of resorting to loops.
2. Test cases CHECK lines were made more compact/strict by not comparing types.

Addressed review comments.
Major change is w.r.t the way loop invariants were decided.
This update traverses each scf.yield operands and back traces to query whether the
corresponding before block argument should be removed.

Hi @ftynse
Thank you for your input. I clearly missed out on the entire scf.while construct earlier.
I've tried addressing the same in this new update.
I've added 2 patterns (also demonstrated with necessary test cases) :-

• Added two patterns to take care of various cases that may arise.
• Not using RAUW rather resorting to the rewrite driver.

A few minor comments.

Hi @imaihal, thank you for this patch.
The function normalizeFuncOpMemRefs has become quite long and I believe it'd be better to split the logic into modular sub-functions.
AllocOps, function arguments, result types, and external functions are being addressed within and maybe we can have separate functions for each of them.
Not sure if this change should go into this patch, maybe @bondhugula can say better.

Minor code refactor

Updated test case to match loop body/yield

Thank you for raising this.
Solution looks good to me. Please add the test case though.

Handled review comments and updated test case

Updated getter/setter functions for lower and upper bound

Added test case for build method and bug fix

Bug fix for build function.

Tackled initial review comments.

Looks good to me.

Added TODO section for DimOp.

Addressed review changes.

Minor review changes

Clang-format error fixed

Rebased to latest master branch

Resolved merge conflicts and involved another test case

Handled minor comments

Minor formatting of code comment

@AlexEichenberger : Such operation handling can be a part of subsequent revision.
Thank you.

Review changes and code formatting

Iterating over function arguments

Reformatted test cases

Memref map layout normalization for function argument and AllocOps

Using setOperand to change memref operand of an operation.

Review changes :-

Code refactoring:

Reformatted the code using clang-format