torch-mlir-opt doesn't support a case for onnx.QLinearConv conversion

[uazizTT]

What happened?

I converted an ONNX model resnet50-v1-12-int8.onnx to mlir using import_onnx tool that completed successfully.

python -m torch_mlir.tools.import_onnx resnet50-v1-12-int8.onnx -o restnet50-v1-12.int8.onnx.mlir

However when converting the torch-onnx to torch using the torch-mlir-opt tool, it fails to convert onnx.QLinearConv operator.

restnet50-v1-12.int8.onnx.mlir:371:12: error: failed to legalize operation 'torch.operator' that was explicitly marked illegal
    %367 = torch.operator "onnx.QLinearConv"(%366, %1, %0, %2, %3, %4, %7, %6, %5) {torch.onnx.auto_pad = "NOTSET", torch.onnx.dilations = [1 : si64, 1 : si64], torch.onnx.group = 1 : si64, torch.onnx.kernel_shape = [7 : si64, 7 : si64], torch.onnx.pads = [3 : si64, 3 : si64, 3 : si64, 3 : si64], torch.onnx.strides = [2 : si64, 2 : si64]} : (!torch.vtensor<[?,3,224,224],ui8>, !torch.vtensor<[],f32>, !torch.vtensor<[],ui8>, !torch.vtensor<[64,3,7,7],si8>, !torch.vtensor<[64],f32>, !torch.vtensor<[64],si8>, !torch.vtensor<[],f32>, !torch.vtensor<[],ui8>, !torch.vtensor<[64],si32>) -> !torch.vtensor<[?,64,112,112],ui8>

This operator is currently marked as completed on the op tracker nod-ai/SHARK-ModelDev#215

I can reproduce the same issue with iree-import-onnx tool followed by iree-compile.

Steps to reproduce your issue

python -m torch_mlir.tools.import_onnx resnet50-v1-12-int8.onnx -o restnet50-v1-12.int8.onnx.mlir

./build/bin/torch-mlir-opt --convert-torch-onnx-to-torch restnet50-v1-12.int8.onnx.mlir -o restnet50-v1-12.int8.mlir

The onnx model can be downloaded from https://github.com/onnx/models

What component(s) does this issue relate to?

Frontends, MLIR

Version information

No response

Additional context

No response