[LegalizeTypes][MSP430] Soften FAKE_USE operand #142714
Merged
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
|
@llvm/pr-subscribers-backend-msp430 Author: Acthink Yang (Acthink-Yang) ChangesAdds support for softening FAKE_USE operands. Fixes #137572 Full diff: https://github.com/llvm/llvm-project/pull/142714.diff 3 Files Affected:
diff --git a/llvm/lib/CodeGen/SelectionDAG/LegalizeFloatTypes.cpp b/llvm/lib/CodeGen/SelectionDAG/LegalizeFloatTypes.cpp
index 41e85521b41ea..81f7c8059760e 100644
--- a/llvm/lib/CodeGen/SelectionDAG/LegalizeFloatTypes.cpp
+++ b/llvm/lib/CodeGen/SelectionDAG/LegalizeFloatTypes.cpp
@@ -1143,6 +1143,9 @@ bool DAGTypeLegalizer::SoftenFloatOperand(SDNode *N, unsigned OpNo) {
Res = SoftenFloatOp_ATOMIC_STORE(N, OpNo);
break;
case ISD::FCOPYSIGN: Res = SoftenFloatOp_FCOPYSIGN(N); break;
+ case ISD::FAKE_USE:
+ Res = SoftenFloatOp_FAKE_USE(N);
+ break;
}
// If the result is null, the sub-method took care of registering results etc.
@@ -1477,6 +1480,12 @@ SDValue DAGTypeLegalizer::SoftenFloatOp_LLRINT(SDNode *N) {
RTLIB::LLRINT_PPCF128));
}
+SDValue DAGTypeLegalizer::SoftenFloatOp_FAKE_USE(SDNode *N) {
+ SDValue Op1 = BitConvertToInteger(N->getOperand(1));
+ return DAG.getNode(N->getOpcode(), SDLoc(N), N->getValueType(0),
+ N->getOperand(0), Op1);
+}
+
//===----------------------------------------------------------------------===//
// Float Result Expansion
//===----------------------------------------------------------------------===//
diff --git a/llvm/lib/CodeGen/SelectionDAG/LegalizeTypes.h b/llvm/lib/CodeGen/SelectionDAG/LegalizeTypes.h
index dd9af47da5287..a541833684f38 100644
--- a/llvm/lib/CodeGen/SelectionDAG/LegalizeTypes.h
+++ b/llvm/lib/CodeGen/SelectionDAG/LegalizeTypes.h
@@ -651,6 +651,7 @@ class LLVM_LIBRARY_VISIBILITY DAGTypeLegalizer {
SDValue SoftenFloatOp_STORE(SDNode *N, unsigned OpNo);
SDValue SoftenFloatOp_ATOMIC_STORE(SDNode *N, unsigned OpNo);
SDValue SoftenFloatOp_FCOPYSIGN(SDNode *N);
+ SDValue SoftenFloatOp_FAKE_USE(SDNode *N);
//===--------------------------------------------------------------------===//
// Float Expansion Support: LegalizeFloatTypes.cpp
diff --git a/llvm/test/CodeGen/MSP430/fake_use_float.ll b/llvm/test/CodeGen/MSP430/fake_use_float.ll
new file mode 100644
index 0000000000000..3983cea0f5939
--- /dev/null
+++ b/llvm/test/CodeGen/MSP430/fake_use_float.ll
@@ -0,0 +1,31 @@
+; RUN: llc -stop-after=finalize-isel -filetype=asm -o - < %s | FileCheck %s
+target datalayout = "e-m:e-p:16:16-i32:16-i64:16-f32:16-f64:16-a:8-n8:16-S16"
+target triple = "msp430"
+
+; CHECK: bb.0.entry:
+; CHECK: %0:gr16 = MOV16ri
+; CHECK-DAG: FAKE_USE killed %0
+; CHECK: %1:gr16 = MOV16ri
+; CHECK-DAG: FAKE_USE killed %1
+; CHECK: %2:gr16 = MOV16ri
+; CHECK-DAG: FAKE_USE killed %2
+; CHECK: %3:gr16 = MOV16ri
+; CHECK-DAG: FAKE_USE killed %3
+; CHECK: RET
+define void @test-double() {
+entry:
+ call void (...) @llvm.fake.use(double -8.765430e+02)
+ ret void
+}
+
+; CHECK: bb.0.entry:
+; CHECK: %0:gr16 = MOV16ri
+; CHECK-DAG: FAKE_USE killed %0
+; CHECK: %1:gr16 = MOV16ri
+; CHECK-DAG: FAKE_USE killed %1
+; CHECK: RET
+define void @test-float() {
+entry:
+ call void (...) @llvm.fake.use(float -8.76e+02)
+ ret void
+}
|
|
@llvm/pr-subscribers-llvm-selectiondag Author: Acthink Yang (Acthink-Yang) ChangesAdds support for softening FAKE_USE operands. Fixes #137572 Full diff: https://github.com/llvm/llvm-project/pull/142714.diff 3 Files Affected:
diff --git a/llvm/lib/CodeGen/SelectionDAG/LegalizeFloatTypes.cpp b/llvm/lib/CodeGen/SelectionDAG/LegalizeFloatTypes.cpp
index 41e85521b41ea..81f7c8059760e 100644
--- a/llvm/lib/CodeGen/SelectionDAG/LegalizeFloatTypes.cpp
+++ b/llvm/lib/CodeGen/SelectionDAG/LegalizeFloatTypes.cpp
@@ -1143,6 +1143,9 @@ bool DAGTypeLegalizer::SoftenFloatOperand(SDNode *N, unsigned OpNo) {
Res = SoftenFloatOp_ATOMIC_STORE(N, OpNo);
break;
case ISD::FCOPYSIGN: Res = SoftenFloatOp_FCOPYSIGN(N); break;
+ case ISD::FAKE_USE:
+ Res = SoftenFloatOp_FAKE_USE(N);
+ break;
}
// If the result is null, the sub-method took care of registering results etc.
@@ -1477,6 +1480,12 @@ SDValue DAGTypeLegalizer::SoftenFloatOp_LLRINT(SDNode *N) {
RTLIB::LLRINT_PPCF128));
}
+SDValue DAGTypeLegalizer::SoftenFloatOp_FAKE_USE(SDNode *N) {
+ SDValue Op1 = BitConvertToInteger(N->getOperand(1));
+ return DAG.getNode(N->getOpcode(), SDLoc(N), N->getValueType(0),
+ N->getOperand(0), Op1);
+}
+
//===----------------------------------------------------------------------===//
// Float Result Expansion
//===----------------------------------------------------------------------===//
diff --git a/llvm/lib/CodeGen/SelectionDAG/LegalizeTypes.h b/llvm/lib/CodeGen/SelectionDAG/LegalizeTypes.h
index dd9af47da5287..a541833684f38 100644
--- a/llvm/lib/CodeGen/SelectionDAG/LegalizeTypes.h
+++ b/llvm/lib/CodeGen/SelectionDAG/LegalizeTypes.h
@@ -651,6 +651,7 @@ class LLVM_LIBRARY_VISIBILITY DAGTypeLegalizer {
SDValue SoftenFloatOp_STORE(SDNode *N, unsigned OpNo);
SDValue SoftenFloatOp_ATOMIC_STORE(SDNode *N, unsigned OpNo);
SDValue SoftenFloatOp_FCOPYSIGN(SDNode *N);
+ SDValue SoftenFloatOp_FAKE_USE(SDNode *N);
//===--------------------------------------------------------------------===//
// Float Expansion Support: LegalizeFloatTypes.cpp
diff --git a/llvm/test/CodeGen/MSP430/fake_use_float.ll b/llvm/test/CodeGen/MSP430/fake_use_float.ll
new file mode 100644
index 0000000000000..3983cea0f5939
--- /dev/null
+++ b/llvm/test/CodeGen/MSP430/fake_use_float.ll
@@ -0,0 +1,31 @@
+; RUN: llc -stop-after=finalize-isel -filetype=asm -o - < %s | FileCheck %s
+target datalayout = "e-m:e-p:16:16-i32:16-i64:16-f32:16-f64:16-a:8-n8:16-S16"
+target triple = "msp430"
+
+; CHECK: bb.0.entry:
+; CHECK: %0:gr16 = MOV16ri
+; CHECK-DAG: FAKE_USE killed %0
+; CHECK: %1:gr16 = MOV16ri
+; CHECK-DAG: FAKE_USE killed %1
+; CHECK: %2:gr16 = MOV16ri
+; CHECK-DAG: FAKE_USE killed %2
+; CHECK: %3:gr16 = MOV16ri
+; CHECK-DAG: FAKE_USE killed %3
+; CHECK: RET
+define void @test-double() {
+entry:
+ call void (...) @llvm.fake.use(double -8.765430e+02)
+ ret void
+}
+
+; CHECK: bb.0.entry:
+; CHECK: %0:gr16 = MOV16ri
+; CHECK-DAG: FAKE_USE killed %0
+; CHECK: %1:gr16 = MOV16ri
+; CHECK-DAG: FAKE_USE killed %1
+; CHECK: RET
+define void @test-float() {
+entry:
+ call void (...) @llvm.fake.use(float -8.76e+02)
+ ret void
+}
|
Acthink-Yang
force-pushed
the
SoftenFloatOperand-FAKE_USE
branch
from
5b3c6f5 to
511d75d
Compare
arsenm
reviewed
Jun 4, 2025
Acthink-Yang
force-pushed
the
SoftenFloatOperand-FAKE_USE
branch
from
a6f993b to
d4a180d
Compare
arsenm
approved these changes
Jun 4, 2025
Adds support for softening FAKE_USE operands. Adds MSP430 tests that exercise the new softening code. Fixes llvm#137572
Co-authored-by: Matt Arsenault <arsenm2@gmail.com>
Co-authored-by: Matt Arsenault <arsenm2@gmail.com>
Acthink-Yang
force-pushed
the
SoftenFloatOperand-FAKE_USE
branch
from
d4a180d to
bec195c
Compare
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
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.
Adds support for softening FAKE_USE operands.
Adds MSP430 tests that exercise the new softening code.
Fixes #137572