Skip to content

[mlir][llvm] Add zeroinitializer constant #65508

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 3 commits into from
Sep 8, 2023
Merged

[mlir][llvm] Add zeroinitializer constant #65508

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
89cf3f8
[mlir][llvm] Add zeroinitializer constant
sitio-couto Aug 29, 2023
a82a416
Apply first round of reviews
sitio-couto Sep 6, 2023
ecedd7b
Fix test FileCheck variable usage
sitio-couto Sep 6, 2023
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
24 changes: 24 additions & 0 deletions mlir/include/mlir/Dialect/LLVMIR/LLVMOps.td
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1506,6 +1506,30 @@ def LLVM_PoisonOp : LLVM_Op<"mlir.poison", [Pure]>,
let assemblyFormat = "attr-dict `:` type($res)";
}

def LLVM_ZeroOp
: LLVM_Op<"mlir.zero", [Pure]>,
LLVM_Builder<"$res = llvm::Constant::getNullValue($_resultType);">
{
let summary = "Creates a zero-initialized value of LLVM dialect type.";
let description = [{
Unlike LLVM IR, MLIR does not have first-class zero-initialized values.
Such values must be created as SSA values using `llvm.mlir.zero`. This
operation has no operands or attributes. It creates a zero-initialized
value of the specified LLVM IR dialect type.

Example:

```mlir
// Create a zero-initialized value for a structure with a 32-bit integer
// followed by a float.
%0 = llvm.mlir.zero : !llvm.struct<(i32, f32)>
```
}];
let results = (outs LLVM_Type:$res);
let builders = [LLVM_OneResultOpBuilder];
let assemblyFormat = "attr-dict `:` type($res)";
}

def LLVM_ConstantOp
: LLVM_Op<"mlir.constant", [Pure, ConstantLike]>,
LLVM_Builder<[{$res = getLLVMConstant($_resultType, $value, $_location,
Expand Down
7 changes: 7 additions & 0 deletions mlir/test/Dialect/LLVMIR/roundtrip.mlir
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -334,6 +334,13 @@ func.func @null() {
llvm.return
}

// CHECK-LABEL: @zero
func.func @zero() {
// CHECK: llvm.mlir.zero : i8
%0 = llvm.mlir.zero : i8
llvm.return
}

// CHECK-LABEL: @atomic_load
func.func @atomic_load(%ptr : !llvm.ptr) {
// CHECK: llvm.load %{{.*}} atomic monotonic {alignment = 4 : i64} : !llvm.ptr -> f32
Expand Down
26 changes: 26 additions & 0 deletions mlir/test/Target/LLVMIR/llvmir.mlir
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -2292,3 +2292,29 @@ llvm.func @locally_streaming_func() attributes {arm_locally_streaming} {
}

// CHECK: attributes #[[ATTR]] = { "aarch64_pstate_sm_body" }

// -----

//
// Zero-initialize operation.
//

// CHECK: @partially_zeroinit_aggregate = linkonce global { i32, i64, [3 x i8] } { i32 0, i64 1, [3 x i8] zeroinitializer }
llvm.mlir.global linkonce @partially_zeroinit_aggregate() : !llvm.struct<(i32, i64, !llvm.array<3 x i8>)> {
%0 = llvm.mlir.zero : !llvm.struct<(i32, i64, !llvm.array<3 x i8>)>
%1 = llvm.mlir.constant(1 : i64) : i64
%2 = llvm.insertvalue %1, %0[1] : !llvm.struct<(i32, i64, !llvm.array<3 x i8>)>
llvm.return %2 : !llvm.struct<(i32, i64, !llvm.array<3 x i8>)>
}

llvm.func @zeroinit_complex_local_aggregate() {
// CHECK: %[[#VAR:]] = alloca [1000 x { i32, [3 x { double, <4 x ptr>, [2 x ptr] }], [6 x ptr] }], i64 1, align 32
Copy link
Member

Choose a reason for hiding this comment

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

I'm curious, how does %[[#VAR:]] work? Usually, we are using %[[#VAR:.*]] to capture the name.

Copy link
Contributor Author

Choose a reason for hiding this comment

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

It's called FileCheck Numeric Substitution Blocks! The # prefix enables it. It can be used to check more complex numeric relations, but it can also check basic numeric equalities.

%0 = llvm.mlir.constant(1 : i64) : i64
%1 = llvm.alloca %0 x !llvm.array<1000 x !llvm.struct<(i32, !llvm.array<3 x !llvm.struct<(f64, !llvm.vec<4 x ptr>, !llvm.array<2 x ptr>)>>, !llvm.array<6 x ptr>)>> : (i64) -> !llvm.ptr

// CHECK: store [1000 x { i32, [3 x { double, <4 x ptr>, [2 x ptr] }], [6 x ptr] }] zeroinitializer, ptr %[[#VAR]], align 32
%2 = llvm.mlir.zero : !llvm.array<1000 x !llvm.struct<(i32, !llvm.array<3 x !llvm.struct<(f64, !llvm.vec<4 x ptr>, !llvm.array<2 x ptr>)>>, !llvm.array<6 x ptr>)>>
llvm.store %2, %1 : !llvm.array<1000 x !llvm.struct<(i32, !llvm.array<3 x !llvm.struct<(f64, !llvm.vec<4 x ptr>, !llvm.array<2 x ptr>)>>, !llvm.array<6 x ptr>)>>, !llvm.ptr

llvm.return
}