Skip to content

[mlir][EmitC]Expand the MemRefToEmitC pass - Adding scalars #148055

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
merged 9 commits into from
Jul 18, 2025
Merged

[mlir][EmitC]Expand the MemRefToEmitC pass - Adding scalars #148055

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
9 commits
Select commit Hold shift + click to select a range
a86bd7a
Initial work on memref ops
Jaddyen Jul 7, 2025
d62dd8d
Convert scalars to constants
Jaddyen Jul 10, 2025
263d6e0
global and getGlobal
Jaddyen Jul 10, 2025
5612d99
Use a helper when you can
Jaddyen Jul 14, 2025
2950fd0
Lower to pointer type
Jaddyen Jul 14, 2025
82bd2b4
Use an lvalue and apply op
Jaddyen Jul 17, 2025
1dc0e63
Update test file
Jaddyen Jul 17, 2025
bc7b8e2
remve unncessary auto
Jaddyen Jul 17, 2025
40e80d5
Merge branch 'main' into memref-ops
Jaddyen Jul 18, 2025
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
36 changes: 33 additions & 3 deletions mlir/lib/Conversion/MemRefToEmitC/MemRefToEmitC.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -16,7 +16,9 @@
#include "mlir/Dialect/EmitC/IR/EmitC.h"
#include "mlir/Dialect/MemRef/IR/MemRef.h"
#include "mlir/IR/Builders.h"
#include "mlir/IR/BuiltinTypes.h"
#include "mlir/IR/PatternMatch.h"
#include "mlir/IR/TypeRange.h"
#include "mlir/Transforms/DialectConversion.h"

using namespace mlir;
Expand Down Expand Up @@ -77,13 +79,23 @@ struct ConvertAlloca final : public OpConversionPattern<memref::AllocaOp> {
}
};

Type convertMemRefType(MemRefType opTy, const TypeConverter *typeConverter) {
Type resultTy;
if (opTy.getRank() == 0) {
resultTy = typeConverter->convertType(mlir::getElementTypeOrSelf(opTy));
} else {
resultTy = typeConverter->convertType(opTy);
}
return resultTy;
}

struct ConvertGlobal final : public OpConversionPattern<memref::GlobalOp> {
using OpConversionPattern::OpConversionPattern;

LogicalResult
matchAndRewrite(memref::GlobalOp op, OpAdaptor operands,
ConversionPatternRewriter &rewriter) const override {

MemRefType opTy = op.getType();
if (!op.getType().hasStaticShape()) {
return rewriter.notifyMatchFailure(
op.getLoc(), "cannot transform global with dynamic shape");
Expand All @@ -95,7 +107,9 @@ struct ConvertGlobal final : public OpConversionPattern<memref::GlobalOp> {
op.getLoc(), "global variable with alignment requirement is "
"currently not supported");
}
auto resultTy = getTypeConverter()->convertType(op.getType());

Type resultTy = convertMemRefType(opTy, getTypeConverter());

if (!resultTy) {
return rewriter.notifyMatchFailure(op.getLoc(),
"cannot convert result type");
Expand All @@ -114,6 +128,10 @@ struct ConvertGlobal final : public OpConversionPattern<memref::GlobalOp> {
bool externSpecifier = !staticSpecifier;

Attribute initialValue = operands.getInitialValueAttr();
if (opTy.getRank() == 0) {
auto elementsAttr = llvm::cast<ElementsAttr>(*op.getInitialValue());
initialValue = elementsAttr.getSplatValue<Attribute>();
Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

What is the initialValue before this vs splat value returned?

Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Without using getSplatValue, we have the initial value as: initial_value = dense<-1> : tensor<i32>
After getting the splat value, we have emitc.global static const @__constant_xi32 : i32 = -1

}
if (isa_and_present<UnitAttr>(initialValue))
initialValue = {};

Expand All @@ -132,11 +150,23 @@ struct ConvertGetGlobal final
matchAndRewrite(memref::GetGlobalOp op, OpAdaptor operands,
ConversionPatternRewriter &rewriter) const override {

auto resultTy = getTypeConverter()->convertType(op.getType());
MemRefType opTy = op.getType();
Type resultTy = convertMemRefType(opTy, getTypeConverter());

if (!resultTy) {
return rewriter.notifyMatchFailure(op.getLoc(),
"cannot convert result type");
}

if (opTy.getRank() == 0) {
emitc::LValueType lvalueType = emitc::LValueType::get(resultTy);
emitc::GetGlobalOp globalLValue = rewriter.create<emitc::GetGlobalOp>(
op.getLoc(), lvalueType, operands.getNameAttr());
emitc::PointerType pointerType = emitc::PointerType::get(resultTy);
rewriter.replaceOpWithNewOp<emitc::ApplyOp>(
op, pointerType, rewriter.getStringAttr("&"), globalLValue);
return success();
}
rewriter.replaceOpWithNewOp<emitc::GetGlobalOp>(op, resultTy,
operands.getNameAttr());
return success();
Expand Down
5 changes: 5 additions & 0 deletions mlir/test/Conversion/MemRefToEmitC/memref-to-emitc.mlir
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -41,6 +41,8 @@ func.func @memref_load(%buff : memref<4x8xf32>, %i: index, %j: index) -> f32 {
module @globals {
memref.global "private" constant @internal_global : memref<3x7xf32> = dense<4.0>
// CHECK-NEXT: emitc.global static const @internal_global : !emitc.array<3x7xf32> = dense<4.000000e+00>
memref.global "private" constant @__constant_xi32 : memref<i32> = dense<-1>
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I was expecting to see a corresponding change to memref-to-emitc-failed.mlir. Looking I suppose it isn't there, but are any of the cases, like

llvm-project/mlir/test/Conversion/MemRefToEmitC/memref-to-emitc-failed.mlir

Line 46 in 36b61a6

%0 = memref.alloca() : memref<0xf32>
going to work now?

Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Not yet. This would require changing those particular ops. I could circle back to this once we can completely lower the inliner model.

// CHECK-NEXT: emitc.global static const @__constant_xi32 : i32 = -1
memref.global @public_global : memref<3x7xf32>
// CHECK-NEXT: emitc.global extern @public_global : !emitc.array<3x7xf32>
memref.global @uninitialized_global : memref<3x7xf32> = uninitialized
Expand All @@ -50,6 +52,9 @@ module @globals {
func.func @use_global() {
// CHECK-NEXT: emitc.get_global @public_global : !emitc.array<3x7xf32>
%0 = memref.get_global @public_global : memref<3x7xf32>
// CHECK-NEXT: emitc.get_global @__constant_xi32 : !emitc.lvalue<i32>
// CHECK-NEXT: emitc.apply "&"(%1) : (!emitc.lvalue<i32>) -> !emitc.ptr<i32>
%1 = memref.get_global @__constant_xi32 : memref<i32>
Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Just to check, so previously memref<1xi32> worked, but memref<i32> didn't work before this?

Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Yes!

return
}
}
Loading