Remove unneeded include from mlir/lib/Conversion/LinalgToLLVM/LinalgToLLVM.cpp

Remove the pattern from ConvertLinalgToLLVMPass.

Oh wait me a sec, I think I should populate the pattern at populateStdToLLVMConversionPatterns, not in LinalgToLLVM.cpp file. Let me fix it.

It's fixed by adding a missing dep in mlir/lib/Conversion/LinalgToLLVM/CMakeLists.txt

In D81618#2093524, @mehdi_amini wrote:

Reverted in a9a21bb4b682474248dc85f9e7db4b260d249ab9 ; the standalone test was broken

Address comments.

Add missing files...

Thanks for the advice, I think it's better now. :)

Add a test

Add expanding tanh pattern to StandardOps/Transforms/

In D81317#2083665, @antiagainst wrote:

Should we drop this for now given this is not the preferred way to go?

rebase

rebase

Address comments.

In D80348#2052290, @mravishankar wrote:

Good to know that this minimal change works as expected (I hoped it would). Thanks for adding tests.
Based on the previous patch in this stack, if we can combine the methods for fusion of generic/indexed-generic producer/consumer then I think this change will not be needed. The tests are still useful though. Will wait to review this again after that.

Update parents

Update parents.

Merge methods together for generic ops and indexed_generic ops.

address comments

Add check on more unsigned ops.

Add a TODO.

In D79931#2037189, @hanchung wrote:

Still working on it, will reply once it's ready to review.

Handle bitwidth emulation on unsigned operator.

Still working on it, will reply once it's ready to review.

Add test cases for casting between i32 and index.

I think indices could be negative? so switch to using SConvert.

Rebases to master.

In D79373#2026852, @antiagainst wrote:

In D79373#2023053, @mravishankar wrote:

Looks good to me. Please wait for River to accept.

I think River's concerns were addressed. He is OOO for the next week; let's land this and we can always fix afterwards if he has more concerns.

Address comments.

Address comments

Address comments.

Address comments. Also found that scalar is not the case, so we can remove some checks and make the logic simpler.