[Mlir-commits] [mlir] [MLIR] VectorEmulateNarrowType to support loading of unaligned vectors (PR #113411)

[Han-Chung Wang]

================
@@ -102,6 +124,41 @@ static FailureOr<Operation *> getCompressedMaskOp(OpBuilder &rewriter,
   return newMask;
 }
 
+static std::optional<int64_t>
+getFrontPaddingSize(ConversionPatternRewriter &rewriter, Location loc,
+                    const memref::LinearizedMemRefInfo linearizedInfo,
+                    bool isUnalignedEmulation) {
+  if (!isUnalignedEmulation)
+    return 0;
+  auto foldedFrontPaddingSize = getValueOrCreateConstantIndexOp(
+      rewriter, loc, linearizedInfo.frontPaddingSize);
+  // try to fold the front padding size into a constant
+  if (auto frontPadding = dyn_cast_or_null<arith::ConstantIndexOp>(
+          foldedFrontPaddingSize.getDefiningOp())) {
+    return frontPadding.value();
+  }
+  return std::nullopt;
+}
+
+static Value extractSubvectorFrom(ConversionPatternRewriter &rewriter,
+                                  Location loc, VectorType extractType,
+                                  Value vector, int64_t frontOffset,
+                                  int64_t subvecSize) {
+  return rewriter
+      .create<vector::ExtractStridedSliceOp>(
+          loc, extractType, vector, rewriter.getI64ArrayAttr({frontOffset}),
+          rewriter.getI64ArrayAttr({subvecSize}), rewriter.getI64ArrayAttr({1}))
+      ->getResult(0);
----------------
hanhanW wrote:

optional: I'd declare few variables to make readability better. E.g.,

```mlir
auto offsets = rewriter.getI64ArrayAttr({frontOffset};
auto sizes = ...;
auto strides = ...;
return rewriter.create...; // without ->getResult(0).
```

https://github.com/llvm/llvm-project/pull/113411