This RFC proposes the addition of the tblgen-to-irdl tool to MLIR. Motivation ODS today allows us to write the necessary C++ boilerplate for dialects. However, it is not easy to introspect operation and attribute verifiers. Tools that could use these verifier definitions cannot introspect these def…

This is an important step towards using IRDL for more use cases, thanks for this work! [image] Groverkss: IRDL definitions of all upstream MLIR dialects can be found here: affine.irdl · GitHub , which were generated using this tool. Not all dialect operations are defined in TableGen. For ex…

Thanks for the proposal [image] Groverkss: Making the IR definitions more introspectable could lead to tools operating on IR definitions, like Fuzzers, Reducers, Mutators, etc., which use these definitions to improve their strategies. I am very much looking forward to an mlir-reduce tool tha…

[image] gysit: IIRC Op(?)DSL generates tablegen files under the hood. If we are lucky, these should be convertible to IRDL as well. Ah, indeed, into build/tools/mlir/include/mlir/Dialect/. I see three dialects in my build: OpenMP, AccCommon and Linalg. @Groverkss could you try and generate t…

I think this is useful. Although this reminds me more of a transpiler in the sense that the output is probably not something that a person writing IRDL would have written. [image] Groverkss: For example, let’s say we want a tool to know which operations generate a SSA Value of type i32 from a S…

[image] jpienaar: Looking at the output it wouldn’t seem this is straight forward in the form generated by this tool today. Having looked at it for some time, it is definitely possible to convert this to a “proper” IRDL definition. One solution is to either teach the tool to understand constr…

+1. Thanks for the RFC. Having a tool that can convert existing dialects to IRDL will make it very easy for folks to leverage its features! And +1 to Jacques’s point about teaching the tool more about MLIR methods over time, but I think that regex matching can have its limits. I think the natural t…

[image] jpienaar: I’d say this could even just be a different backend/output format for mlir-tblgen, but haven’t looked at code yet/could be good in own tool. The reason I did not make this a backend for mlir-tblgen is that this would make mlir-tblgen depend on MLIRIR. I’m not sure how this d…

[image] gysit: IIRC Op(?)DSL generates tablegen files under the hood. If we are lucky, these should be convertible to IRDL as well. This is correct. That piece is also due for a power wash at some point too, so it can be changed if needed to better adapt.

Thank you everyone for your comments and feedback. I just landed this. We will take the feedback provided in this discussion and try to integrate it as we move forward.

[RFC] tblgen-to-irdl tool

MLIR

Groverkss September 24, 2023, 7:26pm 8

I generated these dialects separately, since they seem to use different CMake configuration to generate op defs than other dialects: acc.irdl · GitHub

Topic		Replies	Views
[RFC] Upstream IRDL, a dialect for dynamic dialects MLIR	16	1401	March 7, 2023
Interfacing with xDSL etc. - tablegen, IRDL, Python? MLIR mlir	3	392	April 3, 2025
[RFC] Principles document for IRDL MLIR	13	599	October 6, 2025
Survey: Interested in discussing richer TableGen descriptions in MLIR? MLIR mlir	9	405	September 11, 2025
IRDL Roadmap Tracking Thread MLIR	0	315	September 11, 2025

[RFC] tblgen-to-irdl tool

Related topics