Fix Conv1d w8a32 operator

Summary:
#### Summary

This diff fixes the Conv1d w8a32 operator by adding a transformation to the `val` attribute of the `other_inputs[0].meta` dictionary. Specifically, the `permute` operation is applied to the `original_val` tensor with the `fake_mode` context, and the resulting `transposed_val` is assigned to `transposed_inputs.meta["val"]`.

Reviewed By: mcremon-meta

Differential Revision: D89863750

Author: Marco Giordano

backends/cadence/aot/compiler.py

@@ -22,6 +22,7 @@
     print_memory_planning_info,
 )
 from executorch.backends.cadence.aot.quantizer.fusion_pass import QuantFusion
+from executorch.exir.passes.spec_prop_pass import SpecPropPass
 from executorch.backends.cadence.aot.quantizer.quantizer import (
     CadenceDefaultQuantizer,
     CadenceQuantizer,
@@ -157,7 +158,9 @@ def apply_pre_edge_transform_passes(
     # Get patterns and apply fusion of dq -> op -> q to qop
     # pyre-ignore[16]: no attribute
     patterns = [q.pattern for q in quantizer.quantizers]
-    fused_program = _transform(converted_program, QuantFusion(patterns))
+    fused_program = _transform(
+        converted_program, QuantFusion(patterns), SpecPropPass()
+    )
 
     return fused_program
 

backends/cadence/aot/quantizer/patterns.py

@@ -625,7 +625,7 @@ def get_anchors(
             )
 
         cnn_weights = conv_layer.args[1]
-        if hasattr(cnn_weights.meta, "tensor_meta"):
+        if "tensor_meta" in cnn_weights.meta:
             cnn_weights_shape = cnn_weights.meta["tensor_meta"].shape
             # Bail if the channels are not multiple of 4 (SIMD)
             if cnn_weights_shape[0] % 4 != 0:
@@ -651,6 +651,18 @@ def get_anchors(
                     conv_layer,
                 )
 
+            inputs = conv_layer.args[0]
+            if "tensor_meta" in inputs.meta:
+                inputs_shape = inputs.meta["tensor_meta"].shape
+                # Bail if length != kernel size - Not yet supported
+                if inputs_shape[-1] != cnn_weights_shape[2]:
+                    return (
+                        PartitionAnchors(
+                            empty=True,
+                        ),
+                        conv_layer,
+                    )
+
         return (
             PartitionAnchors(
                 inputs=[],