[mlir][LLVM] Add support for invariant group related intrinsics #115877
+84
−0
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Member
|
@llvm/pr-subscribers-mlir @llvm/pr-subscribers-mlir-llvm Author: Sirui Mu (Lancern) ChangesThis PR adds support for the following LLVM intrinsics:
Full diff: https://github.com/llvm/llvm-project/pull/115877.diff 5 Files Affected:
diff --git a/mlir/include/mlir/Dialect/LLVMIR/LLVMIntrinsicOps.td b/mlir/include/mlir/Dialect/LLVMIR/LLVMIntrinsicOps.td
index d07ebbacc60434..b0d6148726f61c 100644
--- a/mlir/include/mlir/Dialect/LLVMIR/LLVMIntrinsicOps.td
+++ b/mlir/include/mlir/Dialect/LLVMIR/LLVMIntrinsicOps.td
@@ -320,6 +320,24 @@ def LLVM_InvariantEndOp : LLVM_ZeroResultIntrOp<"invariant.end", [2],
"qualified(type($ptr))";
}
+def LLVM_LaunderInvariantGroupOp
+ : LLVM_OneResultIntrOp<"launder.invariant.group", [], [0],
+ [DeclareOpInterfaceMethods<PromotableOpInterface>,
+ SameOperandsAndResultType]> {
+ let arguments = (ins LLVM_AnyPointer:$ptr);
+ let results = (outs LLVM_AnyPointer:$res);
+ let assemblyFormat = "$ptr attr-dict `:` qualified(type($ptr))";
+}
+
+def LLVM_StripInvariantGroupOp
+ : LLVM_OneResultIntrOp<"strip.invariant.group", [], [0],
+ [DeclareOpInterfaceMethods<PromotableOpInterface>,
+ SameOperandsAndResultType]> {
+ let arguments = (ins LLVM_AnyPointer:$ptr);
+ let results = (outs LLVM_AnyPointer:$res);
+ let assemblyFormat = "$ptr attr-dict `:` qualified(type($ptr))";
+}
+
// Constrained Floating-Point Intrinsics.
class LLVM_ConstrainedIntr<string mnem, int numArgs,
diff --git a/mlir/lib/Dialect/LLVMIR/IR/LLVMMemorySlot.cpp b/mlir/lib/Dialect/LLVMIR/IR/LLVMMemorySlot.cpp
index 5dc506c14ef96f..086abdc55fd235 100644
--- a/mlir/lib/Dialect/LLVMIR/IR/LLVMMemorySlot.cpp
+++ b/mlir/lib/Dialect/LLVMIR/IR/LLVMMemorySlot.cpp
@@ -590,6 +590,30 @@ DeletionKind LLVM::InvariantEndOp::removeBlockingUses(
return DeletionKind::Delete;
}
+bool LLVM::LaunderInvariantGroupOp::canUsesBeRemoved(
+ const SmallPtrSetImpl<OpOperand *> &blockingUses,
+ SmallVectorImpl<OpOperand *> &newBlockingUses,
+ const DataLayout &dataLayout) {
+ return true;
+}
+
+DeletionKind LLVM::LaunderInvariantGroupOp::removeBlockingUses(
+ const SmallPtrSetImpl<OpOperand *> &blockingUses, OpBuilder &builder) {
+ return DeletionKind::Delete;
+}
+
+bool LLVM::StripInvariantGroupOp::canUsesBeRemoved(
+ const SmallPtrSetImpl<OpOperand *> &blockingUses,
+ SmallVectorImpl<OpOperand *> &newBlockingUses,
+ const DataLayout &dataLayout) {
+ return true;
+}
+
+DeletionKind LLVM::StripInvariantGroupOp::removeBlockingUses(
+ const SmallPtrSetImpl<OpOperand *> &blockingUses, OpBuilder &builder) {
+ return DeletionKind::Delete;
+}
+
bool LLVM::DbgDeclareOp::canUsesBeRemoved(
const SmallPtrSetImpl<OpOperand *> &blockingUses,
SmallVectorImpl<OpOperand *> &newBlockingUses,
diff --git a/mlir/test/Dialect/LLVMIR/roundtrip.mlir b/mlir/test/Dialect/LLVMIR/roundtrip.mlir
index aa558bad2299ce..f8cd23dc4e723c 100644
--- a/mlir/test/Dialect/LLVMIR/roundtrip.mlir
+++ b/mlir/test/Dialect/LLVMIR/roundtrip.mlir
@@ -672,6 +672,16 @@ llvm.func @invariant(%p: !llvm.ptr) {
llvm.return
}
+// CHECK-LABEL: @invariant_group_intrinsics
+// CHECK-SAME: %[[P:.+]]: !llvm.ptr
+llvm.func @invariant_group_intrinsics(%p: !llvm.ptr) {
+ // CHECK: %{{.+}} = llvm.intr.launder.invariant.group %[[P]] : !llvm.ptr
+ %1 = llvm.intr.launder.invariant.group %p : !llvm.ptr
+ // CHECK: %{{.+}} = llvm.intr.strip.invariant.group %[[P]] : !llvm.ptr
+ %2 = llvm.intr.strip.invariant.group %p : !llvm.ptr
+ llvm.return
+}
+
// CHECK-LABEL: @vararg_func
llvm.func @vararg_func(%arg0: i32, ...) {
// CHECK: %[[C:.*]] = llvm.mlir.constant(1 : i32)
diff --git a/mlir/test/Target/LLVMIR/Import/intrinsic.ll b/mlir/test/Target/LLVMIR/Import/intrinsic.ll
index 606b11175f572f..64911b09ebc85d 100644
--- a/mlir/test/Target/LLVMIR/Import/intrinsic.ll
+++ b/mlir/test/Target/LLVMIR/Import/intrinsic.ll
@@ -802,6 +802,15 @@ define void @invariant(ptr %0) {
ret void
}
+; CHECK-LABEL: llvm.func @invariant_group
+define void @invariant_group(ptr %0) {
+ ; CHECK: %{{.+}} = llvm.intr.launder.invariant.group %{{.*}} : !llvm.ptr
+ %2 = call ptr @llvm.launder.invariant.group.p0(ptr %0)
+ ; CHECK: %{{.+}} = llvm.intr.strip.invariant.group %{{.*}} : !llvm.ptr
+ %3 = call ptr @llvm.strip.invariant.group.p0(ptr %0)
+ ret void
+}
+
; CHECK-LABEL: llvm.func @vector_insert
define void @vector_insert(<vscale x 4 x float> %0, <4 x float> %1) {
; CHECK: llvm.intr.vector.insert %{{.*}}, %{{.*}}[4] : vector<4xf32> into !llvm.vec<? x 4 x f32>
@@ -1191,6 +1200,8 @@ declare void @llvm.lifetime.start.p0(i64 immarg, ptr nocapture)
declare void @llvm.lifetime.end.p0(i64 immarg, ptr nocapture)
declare ptr @llvm.invariant.start.p0(i64 immarg, ptr nocapture)
declare void @llvm.invariant.end.p0(ptr, i64 immarg, ptr nocapture)
+declare ptr @llvm.launder.invariant.group.p0(ptr nocapture)
+declare ptr @llvm.strip.invariant.group.p0(ptr nocapture)
declare void @llvm.assume(i1)
declare float @llvm.ssa.copy.f32(float returned)
diff --git a/mlir/test/Target/LLVMIR/llvmir-intrinsics.mlir b/mlir/test/Target/LLVMIR/llvmir-intrinsics.mlir
index cb712eb4e1262d..059abed34cd5d1 100644
--- a/mlir/test/Target/LLVMIR/llvmir-intrinsics.mlir
+++ b/mlir/test/Target/LLVMIR/llvmir-intrinsics.mlir
@@ -1020,6 +1020,15 @@ llvm.func @invariant(%p: !llvm.ptr) {
llvm.return
}
+// CHECK-LABEL: @invariant_group
+llvm.func @invariant_group(%p: !llvm.ptr) {
+ // CHECK: call ptr @llvm.launder.invariant.group
+ %1 = llvm.intr.launder.invariant.group %p : !llvm.ptr
+ // CHECK: call ptr @llvm.strip.invariant.group
+ %2 = llvm.intr.strip.invariant.group %p : !llvm.ptr
+ llvm.return
+}
+
// CHECK-LABEL: @ssa_copy
llvm.func @ssa_copy(%arg: f32) -> f32 {
// CHECK: call float @llvm.ssa.copy
|
gysit
reviewed
Nov 12, 2024
Contributor
gysit
left a comment
gysit
left a comment
There was a problem hiding this comment.
Thanks!
Looks good to me but I think we should have a test for the PromotableOpInterface. I tried to link the corresponding test case for the lifetime intrinsic below.
This patch adds support for the following LLVM intrinsics: - `llvm.launder.invariant.group` - `llvm.strip.invariant.group`
Lancern
force-pushed
the
mlir/llvmir/invariant-group-intrinsics
branch
from
61cbd49 to
07ceb0f
Compare
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
4 participants
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
This PR adds support for the following LLVM intrinsics:
llvm.launder.invariant.groupllvm.strip.invariant.group