-
Notifications
You must be signed in to change notification settings - Fork 13.5k
[InferAddressSpaces] Handle llvm.lifetime #141045
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.
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
[InferAddressSpaces] Handle llvm.lifetime #141045
Conversation
Thank you for submitting a Pull Request (PR) to the LLVM Project! This PR will be automatically labeled and the relevant teams will be notified. If you wish to, you can add reviewers by using the "Reviewers" section on this page. If this is not working for you, it is probably because you do not have write permissions for the repository. In which case you can instead tag reviewers by name in a comment by using If you have received no comments on your PR for a week, you can request a review by "ping"ing the PR by adding a comment “Ping”. The common courtesy "ping" rate is once a week. Please remember that you are asking for valuable time from other developers. If you have further questions, they may be answered by the LLVM GitHub User Guide. You can also ask questions in a comment on this PR, on the LLVM Discord or on the forums. |
@llvm/pr-subscribers-backend-amdgpu Author: QiYue (QiYueFeiXue) ChangesFull diff: https://github.com/llvm/llvm-project/pull/141045.diff 3 Files Affected:
diff --git a/llvm/lib/Transforms/Scalar/InferAddressSpaces.cpp b/llvm/lib/Transforms/Scalar/InferAddressSpaces.cpp
index 1b7cecc7ceb6a..66836ef05d5db 100644
--- a/llvm/lib/Transforms/Scalar/InferAddressSpaces.cpp
+++ b/llvm/lib/Transforms/Scalar/InferAddressSpaces.cpp
@@ -428,6 +428,14 @@ bool InferAddressSpacesImpl::rewriteIntrinsicOperands(IntrinsicInst *II,
II->replaceUsesOfWith(OldV, NewV);
return true;
}
+ case Intrinsic::lifetime_start:
+ case Intrinsic::lifetime_end: {
+ Function *NewDecl = Intrinsic::getOrInsertDeclaration(
+ M, II->getIntrinsicID(), {NewV->getType()});
+ II->setArgOperand(1, NewV);
+ II->setCalledFunction(NewDecl);
+ return true;
+ }
default: {
Value *Rewrite = TTI->rewriteIntrinsicWithAddressSpace(II, OldV, NewV);
if (!Rewrite)
@@ -479,6 +487,12 @@ void InferAddressSpacesImpl::collectRewritableIntrinsicOperands(
break;
}
+ case Intrinsic::lifetime_start:
+ case Intrinsic::lifetime_end: {
+ appendsFlatAddressExpressionToPostorderStack(II->getArgOperand(1),
+ PostorderStack, Visited);
+ break;
+ }
default:
SmallVector<int, 2> OpIndexes;
if (TTI->collectFlatAddressOperands(OpIndexes, IID)) {
diff --git a/llvm/test/Transforms/InferAddressSpaces/AMDGPU/lifetime.ll b/llvm/test/Transforms/InferAddressSpaces/AMDGPU/lifetime.ll
new file mode 100644
index 0000000000000..291faa7ea9ef4
--- /dev/null
+++ b/llvm/test/Transforms/InferAddressSpaces/AMDGPU/lifetime.ll
@@ -0,0 +1,23 @@
+; RUN: opt -S -mtriple=amdgcn-amd-amdhsa -passes=infer-address-spaces %s | FileCheck %s
+
+
+define i32 @lifetime_flat_pointer() {
+; CHECK-LABEL: define i32 @lifetime_flat_pointer() {
+; CHECK-NEXT: [[ALLOCA:%.*]] = alloca i32, align 4, addrspace(5)
+; CHECK-NEXT: call void @llvm.lifetime.start.p5(i64 4, ptr addrspace(5) [[ALLOCA]])
+; CHECK-NEXT: store i32 1, ptr addrspace(5) [[ALLOCA]], align 4
+; CHECK-NEXT: %ret = load i32, ptr addrspace(5) [[ALLOCA]], align 4
+; CHECK-NEXT: call void @llvm.lifetime.end.p5(i64 4, ptr addrspace(5) [[ALLOCA]])
+; CHECK-NEXT: ret i32 %ret
+;
+ %alloca = alloca i32, align 4, addrspace(5)
+ %flat = addrspacecast ptr addrspace(5) %alloca to ptr
+ call void @llvm.lifetime.start.p0(i64 4 , ptr %flat)
+ store i32 1, ptr %flat, align 4
+ %ret = load i32, ptr %flat, align 4
+ call void @llvm.lifetime.end.p0(i64 4 , ptr %flat)
+ ret i32 %ret
+}
+
+declare void @llvm.lifetime.start.p0(i64 immarg, ptr nocapture)
+declare void @llvm.lifetime.end.p0(i64 immarg, ptr nocapture)
\ No newline at end of file
diff --git a/llvm/test/Transforms/InferAddressSpaces/NVPTX/lifetime.ll b/llvm/test/Transforms/InferAddressSpaces/NVPTX/lifetime.ll
new file mode 100644
index 0000000000000..05e46456c8bf9
--- /dev/null
+++ b/llvm/test/Transforms/InferAddressSpaces/NVPTX/lifetime.ll
@@ -0,0 +1,24 @@
+; RUN: opt -S -passes=infer-address-spaces %s | FileCheck %s
+
+target triple = "nvptx64-nvidia-cuda"
+
+define i32 @lifetime_flat_pointer() {
+; CHECK-LABEL: define i32 @lifetime_flat_pointer() {
+; CHECK-NEXT: [[ALLOCA:%.*]] = alloca i32, align 4, addrspace(5)
+; CHECK-NEXT: call void @llvm.lifetime.start.p5(i64 4, ptr addrspace(5) [[ALLOCA]])
+; CHECK-NEXT: store i32 1, ptr addrspace(5) [[ALLOCA]], align 4
+; CHECK-NEXT: %ret = load i32, ptr addrspace(5) [[ALLOCA]], align 4
+; CHECK-NEXT: call void @llvm.lifetime.end.p5(i64 4, ptr addrspace(5) [[ALLOCA]])
+; CHECK-NEXT: ret i32 %ret
+;
+ %alloca = alloca i32, align 4, addrspace(5)
+ %flat = addrspacecast ptr addrspace(5) %alloca to ptr
+ call void @llvm.lifetime.start.p0(i64 4 , ptr %flat)
+ store i32 1, ptr %flat, align 4
+ %ret = load i32, ptr %flat, align 4
+ call void @llvm.lifetime.end.p0(i64 4 , ptr %flat)
+ ret i32 %ret
+}
+
+declare void @llvm.lifetime.start.p0(i64 immarg, ptr nocapture)
+declare void @llvm.lifetime.end.p0(i64 immarg, ptr nocapture)
|
@llvm/pr-subscribers-backend-nvptx Author: QiYue (QiYueFeiXue) ChangesFull diff: https://github.com/llvm/llvm-project/pull/141045.diff 3 Files Affected:
diff --git a/llvm/lib/Transforms/Scalar/InferAddressSpaces.cpp b/llvm/lib/Transforms/Scalar/InferAddressSpaces.cpp
index 1b7cecc7ceb6a..66836ef05d5db 100644
--- a/llvm/lib/Transforms/Scalar/InferAddressSpaces.cpp
+++ b/llvm/lib/Transforms/Scalar/InferAddressSpaces.cpp
@@ -428,6 +428,14 @@ bool InferAddressSpacesImpl::rewriteIntrinsicOperands(IntrinsicInst *II,
II->replaceUsesOfWith(OldV, NewV);
return true;
}
+ case Intrinsic::lifetime_start:
+ case Intrinsic::lifetime_end: {
+ Function *NewDecl = Intrinsic::getOrInsertDeclaration(
+ M, II->getIntrinsicID(), {NewV->getType()});
+ II->setArgOperand(1, NewV);
+ II->setCalledFunction(NewDecl);
+ return true;
+ }
default: {
Value *Rewrite = TTI->rewriteIntrinsicWithAddressSpace(II, OldV, NewV);
if (!Rewrite)
@@ -479,6 +487,12 @@ void InferAddressSpacesImpl::collectRewritableIntrinsicOperands(
break;
}
+ case Intrinsic::lifetime_start:
+ case Intrinsic::lifetime_end: {
+ appendsFlatAddressExpressionToPostorderStack(II->getArgOperand(1),
+ PostorderStack, Visited);
+ break;
+ }
default:
SmallVector<int, 2> OpIndexes;
if (TTI->collectFlatAddressOperands(OpIndexes, IID)) {
diff --git a/llvm/test/Transforms/InferAddressSpaces/AMDGPU/lifetime.ll b/llvm/test/Transforms/InferAddressSpaces/AMDGPU/lifetime.ll
new file mode 100644
index 0000000000000..291faa7ea9ef4
--- /dev/null
+++ b/llvm/test/Transforms/InferAddressSpaces/AMDGPU/lifetime.ll
@@ -0,0 +1,23 @@
+; RUN: opt -S -mtriple=amdgcn-amd-amdhsa -passes=infer-address-spaces %s | FileCheck %s
+
+
+define i32 @lifetime_flat_pointer() {
+; CHECK-LABEL: define i32 @lifetime_flat_pointer() {
+; CHECK-NEXT: [[ALLOCA:%.*]] = alloca i32, align 4, addrspace(5)
+; CHECK-NEXT: call void @llvm.lifetime.start.p5(i64 4, ptr addrspace(5) [[ALLOCA]])
+; CHECK-NEXT: store i32 1, ptr addrspace(5) [[ALLOCA]], align 4
+; CHECK-NEXT: %ret = load i32, ptr addrspace(5) [[ALLOCA]], align 4
+; CHECK-NEXT: call void @llvm.lifetime.end.p5(i64 4, ptr addrspace(5) [[ALLOCA]])
+; CHECK-NEXT: ret i32 %ret
+;
+ %alloca = alloca i32, align 4, addrspace(5)
+ %flat = addrspacecast ptr addrspace(5) %alloca to ptr
+ call void @llvm.lifetime.start.p0(i64 4 , ptr %flat)
+ store i32 1, ptr %flat, align 4
+ %ret = load i32, ptr %flat, align 4
+ call void @llvm.lifetime.end.p0(i64 4 , ptr %flat)
+ ret i32 %ret
+}
+
+declare void @llvm.lifetime.start.p0(i64 immarg, ptr nocapture)
+declare void @llvm.lifetime.end.p0(i64 immarg, ptr nocapture)
\ No newline at end of file
diff --git a/llvm/test/Transforms/InferAddressSpaces/NVPTX/lifetime.ll b/llvm/test/Transforms/InferAddressSpaces/NVPTX/lifetime.ll
new file mode 100644
index 0000000000000..05e46456c8bf9
--- /dev/null
+++ b/llvm/test/Transforms/InferAddressSpaces/NVPTX/lifetime.ll
@@ -0,0 +1,24 @@
+; RUN: opt -S -passes=infer-address-spaces %s | FileCheck %s
+
+target triple = "nvptx64-nvidia-cuda"
+
+define i32 @lifetime_flat_pointer() {
+; CHECK-LABEL: define i32 @lifetime_flat_pointer() {
+; CHECK-NEXT: [[ALLOCA:%.*]] = alloca i32, align 4, addrspace(5)
+; CHECK-NEXT: call void @llvm.lifetime.start.p5(i64 4, ptr addrspace(5) [[ALLOCA]])
+; CHECK-NEXT: store i32 1, ptr addrspace(5) [[ALLOCA]], align 4
+; CHECK-NEXT: %ret = load i32, ptr addrspace(5) [[ALLOCA]], align 4
+; CHECK-NEXT: call void @llvm.lifetime.end.p5(i64 4, ptr addrspace(5) [[ALLOCA]])
+; CHECK-NEXT: ret i32 %ret
+;
+ %alloca = alloca i32, align 4, addrspace(5)
+ %flat = addrspacecast ptr addrspace(5) %alloca to ptr
+ call void @llvm.lifetime.start.p0(i64 4 , ptr %flat)
+ store i32 1, ptr %flat, align 4
+ %ret = load i32, ptr %flat, align 4
+ call void @llvm.lifetime.end.p0(i64 4 , ptr %flat)
+ ret i32 %ret
+}
+
+declare void @llvm.lifetime.start.p0(i64 immarg, ptr nocapture)
+declare void @llvm.lifetime.end.p0(i64 immarg, ptr nocapture)
|
@llvm/pr-subscribers-llvm-transforms Author: QiYue (QiYueFeiXue) ChangesFull diff: https://github.com/llvm/llvm-project/pull/141045.diff 3 Files Affected:
diff --git a/llvm/lib/Transforms/Scalar/InferAddressSpaces.cpp b/llvm/lib/Transforms/Scalar/InferAddressSpaces.cpp
index 1b7cecc7ceb6a..66836ef05d5db 100644
--- a/llvm/lib/Transforms/Scalar/InferAddressSpaces.cpp
+++ b/llvm/lib/Transforms/Scalar/InferAddressSpaces.cpp
@@ -428,6 +428,14 @@ bool InferAddressSpacesImpl::rewriteIntrinsicOperands(IntrinsicInst *II,
II->replaceUsesOfWith(OldV, NewV);
return true;
}
+ case Intrinsic::lifetime_start:
+ case Intrinsic::lifetime_end: {
+ Function *NewDecl = Intrinsic::getOrInsertDeclaration(
+ M, II->getIntrinsicID(), {NewV->getType()});
+ II->setArgOperand(1, NewV);
+ II->setCalledFunction(NewDecl);
+ return true;
+ }
default: {
Value *Rewrite = TTI->rewriteIntrinsicWithAddressSpace(II, OldV, NewV);
if (!Rewrite)
@@ -479,6 +487,12 @@ void InferAddressSpacesImpl::collectRewritableIntrinsicOperands(
break;
}
+ case Intrinsic::lifetime_start:
+ case Intrinsic::lifetime_end: {
+ appendsFlatAddressExpressionToPostorderStack(II->getArgOperand(1),
+ PostorderStack, Visited);
+ break;
+ }
default:
SmallVector<int, 2> OpIndexes;
if (TTI->collectFlatAddressOperands(OpIndexes, IID)) {
diff --git a/llvm/test/Transforms/InferAddressSpaces/AMDGPU/lifetime.ll b/llvm/test/Transforms/InferAddressSpaces/AMDGPU/lifetime.ll
new file mode 100644
index 0000000000000..291faa7ea9ef4
--- /dev/null
+++ b/llvm/test/Transforms/InferAddressSpaces/AMDGPU/lifetime.ll
@@ -0,0 +1,23 @@
+; RUN: opt -S -mtriple=amdgcn-amd-amdhsa -passes=infer-address-spaces %s | FileCheck %s
+
+
+define i32 @lifetime_flat_pointer() {
+; CHECK-LABEL: define i32 @lifetime_flat_pointer() {
+; CHECK-NEXT: [[ALLOCA:%.*]] = alloca i32, align 4, addrspace(5)
+; CHECK-NEXT: call void @llvm.lifetime.start.p5(i64 4, ptr addrspace(5) [[ALLOCA]])
+; CHECK-NEXT: store i32 1, ptr addrspace(5) [[ALLOCA]], align 4
+; CHECK-NEXT: %ret = load i32, ptr addrspace(5) [[ALLOCA]], align 4
+; CHECK-NEXT: call void @llvm.lifetime.end.p5(i64 4, ptr addrspace(5) [[ALLOCA]])
+; CHECK-NEXT: ret i32 %ret
+;
+ %alloca = alloca i32, align 4, addrspace(5)
+ %flat = addrspacecast ptr addrspace(5) %alloca to ptr
+ call void @llvm.lifetime.start.p0(i64 4 , ptr %flat)
+ store i32 1, ptr %flat, align 4
+ %ret = load i32, ptr %flat, align 4
+ call void @llvm.lifetime.end.p0(i64 4 , ptr %flat)
+ ret i32 %ret
+}
+
+declare void @llvm.lifetime.start.p0(i64 immarg, ptr nocapture)
+declare void @llvm.lifetime.end.p0(i64 immarg, ptr nocapture)
\ No newline at end of file
diff --git a/llvm/test/Transforms/InferAddressSpaces/NVPTX/lifetime.ll b/llvm/test/Transforms/InferAddressSpaces/NVPTX/lifetime.ll
new file mode 100644
index 0000000000000..05e46456c8bf9
--- /dev/null
+++ b/llvm/test/Transforms/InferAddressSpaces/NVPTX/lifetime.ll
@@ -0,0 +1,24 @@
+; RUN: opt -S -passes=infer-address-spaces %s | FileCheck %s
+
+target triple = "nvptx64-nvidia-cuda"
+
+define i32 @lifetime_flat_pointer() {
+; CHECK-LABEL: define i32 @lifetime_flat_pointer() {
+; CHECK-NEXT: [[ALLOCA:%.*]] = alloca i32, align 4, addrspace(5)
+; CHECK-NEXT: call void @llvm.lifetime.start.p5(i64 4, ptr addrspace(5) [[ALLOCA]])
+; CHECK-NEXT: store i32 1, ptr addrspace(5) [[ALLOCA]], align 4
+; CHECK-NEXT: %ret = load i32, ptr addrspace(5) [[ALLOCA]], align 4
+; CHECK-NEXT: call void @llvm.lifetime.end.p5(i64 4, ptr addrspace(5) [[ALLOCA]])
+; CHECK-NEXT: ret i32 %ret
+;
+ %alloca = alloca i32, align 4, addrspace(5)
+ %flat = addrspacecast ptr addrspace(5) %alloca to ptr
+ call void @llvm.lifetime.start.p0(i64 4 , ptr %flat)
+ store i32 1, ptr %flat, align 4
+ %ret = load i32, ptr %flat, align 4
+ call void @llvm.lifetime.end.p0(i64 4 , ptr %flat)
+ ret i32 %ret
+}
+
+declare void @llvm.lifetime.start.p0(i64 immarg, ptr nocapture)
+declare void @llvm.lifetime.end.p0(i64 immarg, ptr nocapture)
|
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
LGTM but I'd probably just delete the NVPTX tests. The first function there isn't representative IR, and the second is a no-op
call void @llvm.lifetime.start.p5(i64 4 , ptr %alloca) | ||
store i32 1, ptr %alloca, align 4 | ||
%ret = load i32, ptr %alloca, align 4 | ||
call void @llvm.lifetime.end.p5(i64 4 , ptr %alloca) |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
call void @llvm.lifetime.start.p5(i64 4 , ptr %alloca) | |
store i32 1, ptr %alloca, align 4 | |
%ret = load i32, ptr %alloca, align 4 | |
call void @llvm.lifetime.end.p5(i64 4 , ptr %alloca) | |
call void @llvm.lifetime.start.p5(i64 4, ptr %alloca) | |
store i32 1, ptr %alloca, align 4 | |
%ret = load i32, ptr %alloca, align 4 | |
call void @llvm.lifetime.end.p5(i64 4, ptr %alloca) |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
The second one is a typo. Please check if it is needed in this scene. The following is the difference between IR before and after this PR:
%alloca = alloca i32, align 4
call void @llvm.lifetime.start.p0(i64 4 , ptr %alloca)
store i32 1, ptr %alloca, align 4
%ret = load i32, ptr %alloca, align 4
call void @llvm.lifetime.end.p0(i64 4 , ptr %alloca)
ret i32 %ret
before PR:
%alloca = alloca i32, align 4
%1 = addrspacecast ptr %alloca to ptr addrspace(5)
%2 = addrspacecast ptr addrspace(5) %1 to ptr
%3 = addrspacecast ptr addrspace(5) %1 to ptr
call void @llvm.lifetime.start.p0(i64 4, ptr %2)
store i32 1, ptr addrspace(5) %1, align 4
%ret = load i32, ptr addrspace(5) %1, align 4
call void @llvm.lifetime.end.p0(i64 4, ptr %3)
ret i32 %ret
after PR:
%alloca = alloca i32, align 4
%1 = addrspacecast ptr %alloca to ptr addrspace(5)
call void @llvm.lifetime.start.p5(i64 4, ptr addrspace(5) %1)
store i32 1, ptr addrspace(5) %1, align 4
%ret = load i32, ptr addrspace(5) %1, align 4
call void @llvm.lifetime.end.p5(i64 4, ptr addrspace(5) %1)
ret i32 %ret
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
OK, so it demonstrates a possible phase ordering with the alloca lowering PTX does, I suppose that's useful, if you adjust the test to look like this "before PR" version
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Got it, done, thanks
@QiYueFeiXue Congratulations on having your first Pull Request (PR) merged into the LLVM Project! Your changes will be combined with recent changes from other authors, then tested by our build bots. If there is a problem with a build, you may receive a report in an email or a comment on this PR. Please check whether problems have been caused by your change specifically, as the builds can include changes from many authors. It is not uncommon for your change to be included in a build that fails due to someone else's changes, or infrastructure issues. How to do this, and the rest of the post-merge process, is covered in detail here. If your change does cause a problem, it may be reverted, or you can revert it yourself. This is a normal part of LLVM development. You can fix your changes and open a new PR to merge them again. If you don't get any reports, no action is required from you. Your changes are working as expected, well done! |
No description provided.