dialects: add linalg.fill

[kayode-gif]

No description provided.

[superlopuh]

As you can see from the CI output, the fill operation in MLIR expects a scalar input, not a tensor. We might want to implement the same verification. The best thing to do would be to look at the MLIR source code to find the verifier for the fill op

[codecov]

Codecov Report

Attention: 4 lines in your changes are missing coverage. Please review.

Comparison is base (1cdfdf2) 89.36% compared to head (118e60e) 89.38%.
Report is 9 commits behind head on main.

Files	Patch %	Lines
xdsl/dialects/linalg.py	77.77%	4 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #2189      +/-   ##
==========================================
+ Coverage   89.36%   89.38%   +0.01%     
==========================================
  Files         316      317       +1     
  Lines       38413    38522     +109     
  Branches     5678     5693      +15     
==========================================
+ Hits        34328    34433     +105     
- Misses       3282     3287       +5     
+ Partials      803      802       -1     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[gabrielrodcanal]

Haven't really checked the custom format in the operation definition in MLIR, but I've found an example in which -> type($res) is optional (I'm referring to

xdsl/xdsl/dialects/linalg.py

Line 439 in ae2054d

"`outs` `(` $outputs `:` type($outputs) `)` `->` type($res) attr-dict"

). See the example here: https://github.com/Xilinx/mlir-air/blob/82d23361cb3843e5d8fc01d746d513f25a224125/test/airhost/51_air_mmult_2x2_channel_broadcast/air.mlir#L30

I'm not sure if this was a mistake or it's just provisional until the cutom format API supports optional groups, but just in case I thought I'd have you know!

[superlopuh]

Oh interesting, should be easy to fix, can you please file an issue with this info?

[gabrielrodcanal]

See #2228.