[InferAddressSpaces] Initialize op(uninit const, uninit const, ...) -> flat #172143
+44
−1
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-llvm-transforms Author: Hongyu Chen (XChy) ChangesFixes #171890 Full diff: https://github.com/llvm/llvm-project/pull/172143.diff 2 Files Affected:
diff --git a/llvm/lib/Transforms/Scalar/InferAddressSpaces.cpp b/llvm/lib/Transforms/Scalar/InferAddressSpaces.cpp
index 352a1b331001a..d7caa60055750 100644
--- a/llvm/lib/Transforms/Scalar/InferAddressSpaces.cpp
+++ b/llvm/lib/Transforms/Scalar/InferAddressSpaces.cpp
@@ -1048,7 +1048,8 @@ bool InferAddressSpacesImpl::updateAddressSpace(
} else {
// Otherwise, infer the address space from its pointer operands.
SmallVector<Constant *, 2> ConstantPtrOps;
- for (Value *PtrOperand : getPointerOperands(V, *DL, TTI)) {
+ auto PtrOps = getPointerOperands(V, *DL, TTI);
+ for (Value *PtrOperand : PtrOps) {
auto I = InferredAddrSpace.find(PtrOperand);
unsigned OperandAS;
if (I == InferredAddrSpace.end()) {
@@ -1079,12 +1080,18 @@ bool InferAddressSpacesImpl::updateAddressSpace(
if (NewAS == FlatAddrSpace)
break;
}
+
if (NewAS != FlatAddrSpace && NewAS != UninitializedAddressSpace) {
if (any_of(ConstantPtrOps, [=](Constant *C) {
return !isSafeToCastConstAddrSpace(C, NewAS);
}))
NewAS = FlatAddrSpace;
}
+
+ // operator(uninit const, uninit const, ...) -> flat
+ if (NewAS == UninitializedAddressSpace &&
+ PtrOps.size() == ConstantPtrOps.size())
+ NewAS = FlatAddrSpace;
}
unsigned OldAS = InferredAddrSpace.lookup(&V);
diff --git a/llvm/test/Transforms/InferAddressSpaces/NVPTX/nullptr.ll b/llvm/test/Transforms/InferAddressSpaces/NVPTX/nullptr.ll
new file mode 100644
index 0000000000000..516d90d282f78
--- /dev/null
+++ b/llvm/test/Transforms/InferAddressSpaces/NVPTX/nullptr.ll
@@ -0,0 +1,36 @@
+; NOTE: Assertions have been autogenerated by utils/update_test_checks.py UTC_ARGS: --version 5
+; RUN: opt -S -passes=infer-address-spaces %s | FileCheck %s
+
+target triple = "nvptx64-unknown-nvidiacl"
+
+define ptr @pr171890() {
+; CHECK-LABEL: define ptr @pr171890() {
+; CHECK-NEXT: [[ENTRY:.*]]:
+; CHECK-NEXT: [[STACK:%.*]] = alloca i16, align 2
+; CHECK-NEXT: [[TMP0:%.*]] = addrspacecast ptr [[STACK]] to ptr addrspace(5)
+; CHECK-NEXT: [[TMP1:%.*]] = addrspacecast ptr addrspace(5) [[TMP0]] to ptr
+; CHECK-NEXT: br label %[[IF:.*]]
+; CHECK: [[IF]]:
+; CHECK-NEXT: [[NULLPHI:%.*]] = phi ptr [ null, %[[ENTRY]] ]
+; CHECK-NEXT: br i1 false, label %[[THEN:.*]], label %[[ELSE:.*]]
+; CHECK: [[ELSE]]:
+; CHECK-NEXT: br label %[[THEN]]
+; CHECK: [[THEN]]:
+; CHECK-NEXT: [[PHI2:%.*]] = phi ptr [ [[TMP1]], %[[IF]] ], [ [[NULLPHI]], %[[ELSE]] ]
+; CHECK-NEXT: ret ptr [[PHI2]]
+;
+entry:
+ %stack = alloca i16, align 2
+ br label %if
+
+if:
+ %nullphi = phi ptr [ null, %entry ]
+ br i1 false, label %then, label %else
+
+else:
+ br label %then
+
+then:
+ %phi2 = phi ptr [ %stack, %if ], [ %nullphi, %else ]
+ ret ptr %phi2
+}
|
Member
|
@llvm/pr-subscribers-backend-nvptx Author: Hongyu Chen (XChy) ChangesFixes #171890 Full diff: https://github.com/llvm/llvm-project/pull/172143.diff 2 Files Affected:
diff --git a/llvm/lib/Transforms/Scalar/InferAddressSpaces.cpp b/llvm/lib/Transforms/Scalar/InferAddressSpaces.cpp
index 352a1b331001a..d7caa60055750 100644
--- a/llvm/lib/Transforms/Scalar/InferAddressSpaces.cpp
+++ b/llvm/lib/Transforms/Scalar/InferAddressSpaces.cpp
@@ -1048,7 +1048,8 @@ bool InferAddressSpacesImpl::updateAddressSpace(
} else {
// Otherwise, infer the address space from its pointer operands.
SmallVector<Constant *, 2> ConstantPtrOps;
- for (Value *PtrOperand : getPointerOperands(V, *DL, TTI)) {
+ auto PtrOps = getPointerOperands(V, *DL, TTI);
+ for (Value *PtrOperand : PtrOps) {
auto I = InferredAddrSpace.find(PtrOperand);
unsigned OperandAS;
if (I == InferredAddrSpace.end()) {
@@ -1079,12 +1080,18 @@ bool InferAddressSpacesImpl::updateAddressSpace(
if (NewAS == FlatAddrSpace)
break;
}
+
if (NewAS != FlatAddrSpace && NewAS != UninitializedAddressSpace) {
if (any_of(ConstantPtrOps, [=](Constant *C) {
return !isSafeToCastConstAddrSpace(C, NewAS);
}))
NewAS = FlatAddrSpace;
}
+
+ // operator(uninit const, uninit const, ...) -> flat
+ if (NewAS == UninitializedAddressSpace &&
+ PtrOps.size() == ConstantPtrOps.size())
+ NewAS = FlatAddrSpace;
}
unsigned OldAS = InferredAddrSpace.lookup(&V);
diff --git a/llvm/test/Transforms/InferAddressSpaces/NVPTX/nullptr.ll b/llvm/test/Transforms/InferAddressSpaces/NVPTX/nullptr.ll
new file mode 100644
index 0000000000000..516d90d282f78
--- /dev/null
+++ b/llvm/test/Transforms/InferAddressSpaces/NVPTX/nullptr.ll
@@ -0,0 +1,36 @@
+; NOTE: Assertions have been autogenerated by utils/update_test_checks.py UTC_ARGS: --version 5
+; RUN: opt -S -passes=infer-address-spaces %s | FileCheck %s
+
+target triple = "nvptx64-unknown-nvidiacl"
+
+define ptr @pr171890() {
+; CHECK-LABEL: define ptr @pr171890() {
+; CHECK-NEXT: [[ENTRY:.*]]:
+; CHECK-NEXT: [[STACK:%.*]] = alloca i16, align 2
+; CHECK-NEXT: [[TMP0:%.*]] = addrspacecast ptr [[STACK]] to ptr addrspace(5)
+; CHECK-NEXT: [[TMP1:%.*]] = addrspacecast ptr addrspace(5) [[TMP0]] to ptr
+; CHECK-NEXT: br label %[[IF:.*]]
+; CHECK: [[IF]]:
+; CHECK-NEXT: [[NULLPHI:%.*]] = phi ptr [ null, %[[ENTRY]] ]
+; CHECK-NEXT: br i1 false, label %[[THEN:.*]], label %[[ELSE:.*]]
+; CHECK: [[ELSE]]:
+; CHECK-NEXT: br label %[[THEN]]
+; CHECK: [[THEN]]:
+; CHECK-NEXT: [[PHI2:%.*]] = phi ptr [ [[TMP1]], %[[IF]] ], [ [[NULLPHI]], %[[ELSE]] ]
+; CHECK-NEXT: ret ptr [[PHI2]]
+;
+entry:
+ %stack = alloca i16, align 2
+ br label %if
+
+if:
+ %nullphi = phi ptr [ null, %entry ]
+ br i1 false, label %then, label %else
+
+else:
+ br label %then
+
+then:
+ %phi2 = phi ptr [ %stack, %if ], [ %nullphi, %else ]
+ ret ptr %phi2
+}
|
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 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.
Fixes #171890
If the pointer operands of an instruction are all constants, we may infer the AS of the instruction as uninitialized finally. And the rewrite process will skip cloning the instruction, producing invalid IR.
This patch fixes it by inferring the AS of this kind of instruction as flat. Maybe we can fold the operator with all constants to get better performance, but I think this case is rare in the real world.