[llvm] [CVP][SCCP] Add support for `uitofp nneg` (PR #86154)
via llvm-commits
llvm-commits at lists.llvm.org
Sun May 5 15:06:53 PDT 2024
https://github.com/goldsteinn updated https://github.com/llvm/llvm-project/pull/86154
>From 8353c4301ef77b82f37db9b23926bbb382c20978 Mon Sep 17 00:00:00 2001
From: Noah Goldstein <goldstein.w.n at gmail.com>
Date: Thu, 21 Mar 2024 01:28:09 -0500
Subject: [PATCH 1/3] [CVP] Add tests for adding `nneg` flag to `uitofp` and
converting `sitofp` -> `uitofp nneg`; NFC
---
.../CorrelatedValuePropagation/sitofp.ll | 203 ++++++++++++++++++
.../CorrelatedValuePropagation/uitofp.ll | 203 ++++++++++++++++++
2 files changed, 406 insertions(+)
create mode 100644 llvm/test/Transforms/CorrelatedValuePropagation/sitofp.ll
create mode 100644 llvm/test/Transforms/CorrelatedValuePropagation/uitofp.ll
diff --git a/llvm/test/Transforms/CorrelatedValuePropagation/sitofp.ll b/llvm/test/Transforms/CorrelatedValuePropagation/sitofp.ll
new file mode 100644
index 00000000000000..7a49a33fbf9e6c
--- /dev/null
+++ b/llvm/test/Transforms/CorrelatedValuePropagation/sitofp.ll
@@ -0,0 +1,203 @@
+; NOTE: Assertions have been autogenerated by utils/update_test_checks.py
+; RUN: opt < %s -passes=correlated-propagation -S | FileCheck %s
+
+declare void @use.f32(float)
+
+define void @test1_fptosi(i32 %n) {
+; CHECK-LABEL: @test1_fptosi(
+; CHECK-NEXT: entry:
+; CHECK-NEXT: br label [[FOR_COND:%.*]]
+; CHECK: for.cond:
+; CHECK-NEXT: [[A:%.*]] = phi i32 [ [[N:%.*]], [[ENTRY:%.*]] ], [ [[EXT:%.*]], [[FOR_BODY:%.*]] ]
+; CHECK-NEXT: [[CMP:%.*]] = icmp sgt i32 [[A]], -1
+; CHECK-NEXT: br i1 [[CMP]], label [[FOR_BODY]], label [[FOR_END:%.*]]
+; CHECK: for.body:
+; CHECK-NEXT: [[EXT_WIDE:%.*]] = sitofp i32 [[A]] to float
+; CHECK-NEXT: call void @use.f32(float [[EXT_WIDE]])
+; CHECK-NEXT: [[EXT]] = fptosi float [[EXT_WIDE]] to i32
+; CHECK-NEXT: br label [[FOR_COND]]
+; CHECK: for.end:
+; CHECK-NEXT: ret void
+;
+entry:
+ br label %for.cond
+
+for.cond: ; preds = %for.body, %entry
+ %a = phi i32 [ %n, %entry ], [ %ext, %for.body ]
+ %cmp = icmp sgt i32 %a, -1
+ br i1 %cmp, label %for.body, label %for.end
+
+for.body: ; preds = %for.cond
+ %ext.wide = sitofp i32 %a to float
+ call void @use.f32(float %ext.wide)
+ %ext = fptosi float %ext.wide to i32
+ br label %for.cond
+
+for.end: ; preds = %for.cond
+ ret void
+}
+
+define void @test1_fptoui(i32 %n) {
+; CHECK-LABEL: @test1_fptoui(
+; CHECK-NEXT: entry:
+; CHECK-NEXT: br label [[FOR_COND:%.*]]
+; CHECK: for.cond:
+; CHECK-NEXT: [[A:%.*]] = phi i32 [ [[N:%.*]], [[ENTRY:%.*]] ], [ [[EXT:%.*]], [[FOR_BODY:%.*]] ]
+; CHECK-NEXT: [[CMP:%.*]] = icmp sgt i32 [[A]], -1
+; CHECK-NEXT: br i1 [[CMP]], label [[FOR_BODY]], label [[FOR_END:%.*]]
+; CHECK: for.body:
+; CHECK-NEXT: [[EXT_WIDE:%.*]] = sitofp i32 [[A]] to float
+; CHECK-NEXT: call void @use.f32(float [[EXT_WIDE]])
+; CHECK-NEXT: [[EXT]] = fptoui float [[EXT_WIDE]] to i32
+; CHECK-NEXT: br label [[FOR_COND]]
+; CHECK: for.end:
+; CHECK-NEXT: ret void
+;
+entry:
+ br label %for.cond
+
+for.cond: ; preds = %for.body, %entry
+ %a = phi i32 [ %n, %entry ], [ %ext, %for.body ]
+ %cmp = icmp sgt i32 %a, -1
+ br i1 %cmp, label %for.body, label %for.end
+
+for.body: ; preds = %for.cond
+ %ext.wide = sitofp i32 %a to float
+ call void @use.f32(float %ext.wide)
+ %ext = fptoui float %ext.wide to i32
+ br label %for.cond
+
+for.end: ; preds = %for.cond
+ ret void
+}
+
+;; Negative test to show transform doesn't happen unless n >= 0.
+define void @test2(i32 %n) {
+; CHECK-LABEL: @test2(
+; CHECK-NEXT: entry:
+; CHECK-NEXT: br label [[FOR_COND:%.*]]
+; CHECK: for.cond:
+; CHECK-NEXT: [[A:%.*]] = phi i32 [ [[N:%.*]], [[ENTRY:%.*]] ], [ [[EXT:%.*]], [[FOR_BODY:%.*]] ]
+; CHECK-NEXT: [[CMP:%.*]] = icmp sgt i32 [[A]], -2
+; CHECK-NEXT: br i1 [[CMP]], label [[FOR_BODY]], label [[FOR_END:%.*]]
+; CHECK: for.body:
+; CHECK-NEXT: [[EXT_WIDE:%.*]] = sitofp i32 [[A]] to float
+; CHECK-NEXT: call void @use.f32(float [[EXT_WIDE]])
+; CHECK-NEXT: [[EXT]] = fptosi float [[EXT_WIDE]] to i32
+; CHECK-NEXT: br label [[FOR_COND]]
+; CHECK: for.end:
+; CHECK-NEXT: ret void
+;
+entry:
+ br label %for.cond
+
+for.cond: ; preds = %for.body, %entry
+ %a = phi i32 [ %n, %entry ], [ %ext, %for.body ]
+ %cmp = icmp sgt i32 %a, -2
+ br i1 %cmp, label %for.body, label %for.end
+
+for.body: ; preds = %for.cond
+ %ext.wide = sitofp i32 %a to float
+ call void @use.f32(float %ext.wide)
+ %ext = fptosi float %ext.wide to i32
+ br label %for.cond
+
+for.end: ; preds = %for.cond
+ ret void
+}
+
+;; Non looping test case.
+define void @test3(i32 %n) {
+; CHECK-LABEL: @test3(
+; CHECK-NEXT: entry:
+; CHECK-NEXT: [[CMP:%.*]] = icmp sgt i32 [[N:%.*]], -1
+; CHECK-NEXT: br i1 [[CMP]], label [[BB:%.*]], label [[EXIT:%.*]]
+; CHECK: bb:
+; CHECK-NEXT: [[EXT_WIDE:%.*]] = sitofp i32 [[N]] to float
+; CHECK-NEXT: call void @use.f32(float [[EXT_WIDE]])
+; CHECK-NEXT: br label [[EXIT]]
+; CHECK: exit:
+; CHECK-NEXT: ret void
+;
+entry:
+ %cmp = icmp sgt i32 %n, -1
+ br i1 %cmp, label %bb, label %exit
+
+bb:
+ %ext.wide = sitofp i32 %n to float
+ call void @use.f32(float %ext.wide)
+ br label %exit
+
+exit:
+ ret void
+}
+
+;; Non looping negative test case.
+define void @test4(i32 %n) {
+; CHECK-LABEL: @test4(
+; CHECK-NEXT: entry:
+; CHECK-NEXT: [[CMP:%.*]] = icmp sgt i32 [[N:%.*]], -2
+; CHECK-NEXT: br i1 [[CMP]], label [[BB:%.*]], label [[EXIT:%.*]]
+; CHECK: bb:
+; CHECK-NEXT: [[EXT_WIDE:%.*]] = sitofp i32 [[N]] to float
+; CHECK-NEXT: call void @use.f32(float [[EXT_WIDE]])
+; CHECK-NEXT: br label [[EXIT]]
+; CHECK: exit:
+; CHECK-NEXT: ret void
+;
+entry:
+ %cmp = icmp sgt i32 %n, -2
+ br i1 %cmp, label %bb, label %exit
+
+bb:
+ %ext.wide = sitofp i32 %n to float
+ call void @use.f32(float %ext.wide)
+ br label %exit
+
+exit:
+ ret void
+}
+
+define float @may_including_undef(i1 %c.1, i1 %c.2) {
+; CHECK-LABEL: @may_including_undef(
+; CHECK-NEXT: br i1 [[C_1:%.*]], label [[TRUE_1:%.*]], label [[FALSE:%.*]]
+; CHECK: true.1:
+; CHECK-NEXT: br i1 [[C_2:%.*]], label [[TRUE_2:%.*]], label [[EXIT:%.*]]
+; CHECK: true.2:
+; CHECK-NEXT: br label [[EXIT]]
+; CHECK: false:
+; CHECK-NEXT: br label [[EXIT]]
+; CHECK: exit:
+; CHECK-NEXT: [[P:%.*]] = phi i32 [ 0, [[TRUE_1]] ], [ 1, [[TRUE_2]] ], [ undef, [[FALSE]] ]
+; CHECK-NEXT: [[EXT:%.*]] = sitofp i32 [[P]] to float
+; CHECK-NEXT: ret float [[EXT]]
+;
+ br i1 %c.1, label %true.1, label %false
+
+true.1:
+ br i1 %c.2, label %true.2, label %exit
+
+true.2:
+ br label %exit
+
+false:
+ br label %exit
+
+exit:
+ %p = phi i32 [ 0, %true.1 ], [ 1, %true.2], [ undef, %false ]
+ %ext = sitofp i32 %p to float
+ ret float %ext
+}
+
+define double @test_infer_at_use(i32 noundef %n) {
+; CHECK-LABEL: @test_infer_at_use(
+; CHECK-NEXT: [[CMP:%.*]] = icmp sgt i32 [[N:%.*]], -1
+; CHECK-NEXT: [[EXT:%.*]] = sitofp i32 [[N]] to double
+; CHECK-NEXT: [[SELECT:%.*]] = select i1 [[CMP]], double [[EXT]], double 0.000000e+00
+; CHECK-NEXT: ret double [[SELECT]]
+;
+ %cmp = icmp sgt i32 %n, -1
+ %ext = sitofp i32 %n to double
+ %select = select i1 %cmp, double %ext, double 0.0
+ ret double %select
+}
diff --git a/llvm/test/Transforms/CorrelatedValuePropagation/uitofp.ll b/llvm/test/Transforms/CorrelatedValuePropagation/uitofp.ll
new file mode 100644
index 00000000000000..1c9dc21c26feb7
--- /dev/null
+++ b/llvm/test/Transforms/CorrelatedValuePropagation/uitofp.ll
@@ -0,0 +1,203 @@
+; NOTE: Assertions have been autogenerated by utils/update_test_checks.py
+; RUN: opt < %s -passes=correlated-propagation -S | FileCheck %s
+
+declare void @use.f32(float)
+
+define void @test1_fptosi(i32 %n) {
+; CHECK-LABEL: @test1_fptosi(
+; CHECK-NEXT: entry:
+; CHECK-NEXT: br label [[FOR_COND:%.*]]
+; CHECK: for.cond:
+; CHECK-NEXT: [[A:%.*]] = phi i32 [ [[N:%.*]], [[ENTRY:%.*]] ], [ [[EXT:%.*]], [[FOR_BODY:%.*]] ]
+; CHECK-NEXT: [[CMP:%.*]] = icmp sgt i32 [[A]], -1
+; CHECK-NEXT: br i1 [[CMP]], label [[FOR_BODY]], label [[FOR_END:%.*]]
+; CHECK: for.body:
+; CHECK-NEXT: [[EXT_WIDE:%.*]] = uitofp i32 [[A]] to float
+; CHECK-NEXT: call void @use.f32(float [[EXT_WIDE]])
+; CHECK-NEXT: [[EXT]] = fptosi float [[EXT_WIDE]] to i32
+; CHECK-NEXT: br label [[FOR_COND]]
+; CHECK: for.end:
+; CHECK-NEXT: ret void
+;
+entry:
+ br label %for.cond
+
+for.cond: ; preds = %for.body, %entry
+ %a = phi i32 [ %n, %entry ], [ %ext, %for.body ]
+ %cmp = icmp sgt i32 %a, -1
+ br i1 %cmp, label %for.body, label %for.end
+
+for.body: ; preds = %for.cond
+ %ext.wide = uitofp i32 %a to float
+ call void @use.f32(float %ext.wide)
+ %ext = fptosi float %ext.wide to i32
+ br label %for.cond
+
+for.end: ; preds = %for.cond
+ ret void
+}
+
+define void @test1_fptoui(i32 %n) {
+; CHECK-LABEL: @test1_fptoui(
+; CHECK-NEXT: entry:
+; CHECK-NEXT: br label [[FOR_COND:%.*]]
+; CHECK: for.cond:
+; CHECK-NEXT: [[A:%.*]] = phi i32 [ [[N:%.*]], [[ENTRY:%.*]] ], [ [[EXT:%.*]], [[FOR_BODY:%.*]] ]
+; CHECK-NEXT: [[CMP:%.*]] = icmp sgt i32 [[A]], -1
+; CHECK-NEXT: br i1 [[CMP]], label [[FOR_BODY]], label [[FOR_END:%.*]]
+; CHECK: for.body:
+; CHECK-NEXT: [[EXT_WIDE:%.*]] = uitofp i32 [[A]] to float
+; CHECK-NEXT: call void @use.f32(float [[EXT_WIDE]])
+; CHECK-NEXT: [[EXT]] = fptoui float [[EXT_WIDE]] to i32
+; CHECK-NEXT: br label [[FOR_COND]]
+; CHECK: for.end:
+; CHECK-NEXT: ret void
+;
+entry:
+ br label %for.cond
+
+for.cond: ; preds = %for.body, %entry
+ %a = phi i32 [ %n, %entry ], [ %ext, %for.body ]
+ %cmp = icmp sgt i32 %a, -1
+ br i1 %cmp, label %for.body, label %for.end
+
+for.body: ; preds = %for.cond
+ %ext.wide = uitofp i32 %a to float
+ call void @use.f32(float %ext.wide)
+ %ext = fptoui float %ext.wide to i32
+ br label %for.cond
+
+for.end: ; preds = %for.cond
+ ret void
+}
+
+;; Negative test to show transform doesn't happen unless n >= 0.
+define void @test2(i32 %n) {
+; CHECK-LABEL: @test2(
+; CHECK-NEXT: entry:
+; CHECK-NEXT: br label [[FOR_COND:%.*]]
+; CHECK: for.cond:
+; CHECK-NEXT: [[A:%.*]] = phi i32 [ [[N:%.*]], [[ENTRY:%.*]] ], [ [[EXT:%.*]], [[FOR_BODY:%.*]] ]
+; CHECK-NEXT: [[CMP:%.*]] = icmp sgt i32 [[A]], -2
+; CHECK-NEXT: br i1 [[CMP]], label [[FOR_BODY]], label [[FOR_END:%.*]]
+; CHECK: for.body:
+; CHECK-NEXT: [[EXT_WIDE:%.*]] = uitofp i32 [[A]] to float
+; CHECK-NEXT: call void @use.f32(float [[EXT_WIDE]])
+; CHECK-NEXT: [[EXT]] = fptosi float [[EXT_WIDE]] to i32
+; CHECK-NEXT: br label [[FOR_COND]]
+; CHECK: for.end:
+; CHECK-NEXT: ret void
+;
+entry:
+ br label %for.cond
+
+for.cond: ; preds = %for.body, %entry
+ %a = phi i32 [ %n, %entry ], [ %ext, %for.body ]
+ %cmp = icmp sgt i32 %a, -2
+ br i1 %cmp, label %for.body, label %for.end
+
+for.body: ; preds = %for.cond
+ %ext.wide = uitofp i32 %a to float
+ call void @use.f32(float %ext.wide)
+ %ext = fptosi float %ext.wide to i32
+ br label %for.cond
+
+for.end: ; preds = %for.cond
+ ret void
+}
+
+;; Non looping test case.
+define void @test3(i32 %n) {
+; CHECK-LABEL: @test3(
+; CHECK-NEXT: entry:
+; CHECK-NEXT: [[CMP:%.*]] = icmp sgt i32 [[N:%.*]], -1
+; CHECK-NEXT: br i1 [[CMP]], label [[BB:%.*]], label [[EXIT:%.*]]
+; CHECK: bb:
+; CHECK-NEXT: [[EXT_WIDE:%.*]] = uitofp i32 [[N]] to float
+; CHECK-NEXT: call void @use.f32(float [[EXT_WIDE]])
+; CHECK-NEXT: br label [[EXIT]]
+; CHECK: exit:
+; CHECK-NEXT: ret void
+;
+entry:
+ %cmp = icmp sgt i32 %n, -1
+ br i1 %cmp, label %bb, label %exit
+
+bb:
+ %ext.wide = uitofp i32 %n to float
+ call void @use.f32(float %ext.wide)
+ br label %exit
+
+exit:
+ ret void
+}
+
+;; Non looping negative test case.
+define void @test4(i32 %n) {
+; CHECK-LABEL: @test4(
+; CHECK-NEXT: entry:
+; CHECK-NEXT: [[CMP:%.*]] = icmp sgt i32 [[N:%.*]], -2
+; CHECK-NEXT: br i1 [[CMP]], label [[BB:%.*]], label [[EXIT:%.*]]
+; CHECK: bb:
+; CHECK-NEXT: [[EXT_WIDE:%.*]] = uitofp i32 [[N]] to float
+; CHECK-NEXT: call void @use.f32(float [[EXT_WIDE]])
+; CHECK-NEXT: br label [[EXIT]]
+; CHECK: exit:
+; CHECK-NEXT: ret void
+;
+entry:
+ %cmp = icmp sgt i32 %n, -2
+ br i1 %cmp, label %bb, label %exit
+
+bb:
+ %ext.wide = uitofp i32 %n to float
+ call void @use.f32(float %ext.wide)
+ br label %exit
+
+exit:
+ ret void
+}
+
+define float @may_including_undef(i1 %c.1, i1 %c.2) {
+; CHECK-LABEL: @may_including_undef(
+; CHECK-NEXT: br i1 [[C_1:%.*]], label [[TRUE_1:%.*]], label [[FALSE:%.*]]
+; CHECK: true.1:
+; CHECK-NEXT: br i1 [[C_2:%.*]], label [[TRUE_2:%.*]], label [[EXIT:%.*]]
+; CHECK: true.2:
+; CHECK-NEXT: br label [[EXIT]]
+; CHECK: false:
+; CHECK-NEXT: br label [[EXIT]]
+; CHECK: exit:
+; CHECK-NEXT: [[P:%.*]] = phi i32 [ 0, [[TRUE_1]] ], [ 1, [[TRUE_2]] ], [ undef, [[FALSE]] ]
+; CHECK-NEXT: [[EXT:%.*]] = uitofp i32 [[P]] to float
+; CHECK-NEXT: ret float [[EXT]]
+;
+ br i1 %c.1, label %true.1, label %false
+
+true.1:
+ br i1 %c.2, label %true.2, label %exit
+
+true.2:
+ br label %exit
+
+false:
+ br label %exit
+
+exit:
+ %p = phi i32 [ 0, %true.1 ], [ 1, %true.2], [ undef, %false ]
+ %ext = uitofp i32 %p to float
+ ret float %ext
+}
+
+define double @test_infer_at_use(i32 noundef %n) {
+; CHECK-LABEL: @test_infer_at_use(
+; CHECK-NEXT: [[CMP:%.*]] = icmp sgt i32 [[N:%.*]], -1
+; CHECK-NEXT: [[EXT:%.*]] = uitofp i32 [[N]] to double
+; CHECK-NEXT: [[SELECT:%.*]] = select i1 [[CMP]], double [[EXT]], double 0.000000e+00
+; CHECK-NEXT: ret double [[SELECT]]
+;
+ %cmp = icmp sgt i32 %n, -1
+ %ext = uitofp i32 %n to double
+ %select = select i1 %cmp, double %ext, double 0.0
+ ret double %select
+}
>From 3b5fdd5ef369bfb85d146478840961487123345a Mon Sep 17 00:00:00 2001
From: Noah Goldstein <goldstein.w.n at gmail.com>
Date: Wed, 20 Mar 2024 17:00:47 -0500
Subject: [PATCH 2/3] [CVP] Convert `sitofp` -> `uitofp nneg` and add `nneg`
flag to `uitofp`
Similiar to the `InstCombine` changes, just furthering the scope of
the canonicalization/`uitofp nneg` support
---
.../Scalar/CorrelatedValuePropagation.cpp | 50 ++++++++++++++++---
.../CorrelatedValuePropagation/sitofp.ll | 8 +--
.../CorrelatedValuePropagation/uitofp.ll | 8 +--
3 files changed, 51 insertions(+), 15 deletions(-)
diff --git a/llvm/lib/Transforms/Scalar/CorrelatedValuePropagation.cpp b/llvm/lib/Transforms/Scalar/CorrelatedValuePropagation.cpp
index 715cdaff972796..50b5fdb5672074 100644
--- a/llvm/lib/Transforms/Scalar/CorrelatedValuePropagation.cpp
+++ b/llvm/lib/Transforms/Scalar/CorrelatedValuePropagation.cpp
@@ -62,6 +62,7 @@ STATISTIC(NumAShrsConverted, "Number of ashr converted to lshr");
STATISTIC(NumAShrsRemoved, "Number of ashr removed");
STATISTIC(NumSRems, "Number of srem converted to urem");
STATISTIC(NumSExt, "Number of sext converted to zext");
+STATISTIC(NumSIToFP, "Number of sitofp converted to uitofp");
STATISTIC(NumSICmps, "Number of signed icmp preds simplified to unsigned");
STATISTIC(NumAnd, "Number of ands removed");
STATISTIC(NumNW, "Number of no-wrap deductions");
@@ -89,7 +90,7 @@ STATISTIC(NumSMinMax,
"Number of llvm.s{min,max} intrinsics simplified to unsigned");
STATISTIC(NumUDivURemsNarrowedExpanded,
"Number of bound udiv's/urem's expanded");
-STATISTIC(NumZExt, "Number of non-negative deductions");
+STATISTIC(NumNNeg, "Number of zext/uitofp non-negative deductions");
static Constant *getConstantAt(Value *V, Instruction *At, LazyValueInfo *LVI) {
if (Constant *C = LVI->getConstant(V, At))
@@ -1075,20 +1076,49 @@ static bool processSExt(SExtInst *SDI, LazyValueInfo *LVI) {
return true;
}
-static bool processZExt(ZExtInst *ZExt, LazyValueInfo *LVI) {
- if (ZExt->getType()->isVectorTy())
+static bool processPossibleNonNeg(PossiblyNonNegInst *I, LazyValueInfo *LVI) {
+ if (I->getType()->isVectorTy())
return false;
- if (ZExt->hasNonNeg())
+ if (I->hasNonNeg())
return false;
- const Use &Base = ZExt->getOperandUse(0);
+ const Use &Base = I->getOperandUse(0);
if (!LVI->getConstantRangeAtUse(Base, /*UndefAllowed*/ false)
.isAllNonNegative())
return false;
- ++NumZExt;
- ZExt->setNonNeg();
+ ++NumNNeg;
+ I->setNonNeg();
+
+ return true;
+}
+
+static bool processZExt(ZExtInst *ZExt, LazyValueInfo *LVI) {
+ return processPossibleNonNeg(cast<PossiblyNonNegInst>(ZExt), LVI);
+}
+
+static bool processUIToFP(UIToFPInst *UIToFP, LazyValueInfo *LVI) {
+ return processPossibleNonNeg(cast<PossiblyNonNegInst>(UIToFP), LVI);
+}
+
+static bool processSIToFP(SIToFPInst *SIToFP, LazyValueInfo *LVI) {
+ if (SIToFP->getType()->isVectorTy())
+ return false;
+
+ const Use &Base = SIToFP->getOperandUse(0);
+ if (!LVI->getConstantRangeAtUse(Base, /*UndefAllowed*/ false)
+ .isAllNonNegative())
+ return false;
+
+ ++NumSIToFP;
+ auto *UIToFP = CastInst::Create(Instruction::UIToFP, Base, SIToFP->getType(),
+ "", SIToFP->getIterator());
+ UIToFP->takeName(SIToFP);
+ UIToFP->setDebugLoc(SIToFP->getDebugLoc());
+ UIToFP->setNonNeg();
+ SIToFP->replaceAllUsesWith(UIToFP);
+ SIToFP->eraseFromParent();
return true;
}
@@ -1197,6 +1227,12 @@ static bool runImpl(Function &F, LazyValueInfo *LVI, DominatorTree *DT,
case Instruction::ZExt:
BBChanged |= processZExt(cast<ZExtInst>(&II), LVI);
break;
+ case Instruction::UIToFP:
+ BBChanged |= processUIToFP(cast<UIToFPInst>(&II), LVI);
+ break;
+ case Instruction::SIToFP:
+ BBChanged |= processSIToFP(cast<SIToFPInst>(&II), LVI);
+ break;
case Instruction::Add:
case Instruction::Sub:
case Instruction::Mul:
diff --git a/llvm/test/Transforms/CorrelatedValuePropagation/sitofp.ll b/llvm/test/Transforms/CorrelatedValuePropagation/sitofp.ll
index 7a49a33fbf9e6c..db8d0a8b80e3bc 100644
--- a/llvm/test/Transforms/CorrelatedValuePropagation/sitofp.ll
+++ b/llvm/test/Transforms/CorrelatedValuePropagation/sitofp.ll
@@ -12,7 +12,7 @@ define void @test1_fptosi(i32 %n) {
; CHECK-NEXT: [[CMP:%.*]] = icmp sgt i32 [[A]], -1
; CHECK-NEXT: br i1 [[CMP]], label [[FOR_BODY]], label [[FOR_END:%.*]]
; CHECK: for.body:
-; CHECK-NEXT: [[EXT_WIDE:%.*]] = sitofp i32 [[A]] to float
+; CHECK-NEXT: [[EXT_WIDE:%.*]] = uitofp nneg i32 [[A]] to float
; CHECK-NEXT: call void @use.f32(float [[EXT_WIDE]])
; CHECK-NEXT: [[EXT]] = fptosi float [[EXT_WIDE]] to i32
; CHECK-NEXT: br label [[FOR_COND]]
@@ -46,7 +46,7 @@ define void @test1_fptoui(i32 %n) {
; CHECK-NEXT: [[CMP:%.*]] = icmp sgt i32 [[A]], -1
; CHECK-NEXT: br i1 [[CMP]], label [[FOR_BODY]], label [[FOR_END:%.*]]
; CHECK: for.body:
-; CHECK-NEXT: [[EXT_WIDE:%.*]] = sitofp i32 [[A]] to float
+; CHECK-NEXT: [[EXT_WIDE:%.*]] = uitofp nneg i32 [[A]] to float
; CHECK-NEXT: call void @use.f32(float [[EXT_WIDE]])
; CHECK-NEXT: [[EXT]] = fptoui float [[EXT_WIDE]] to i32
; CHECK-NEXT: br label [[FOR_COND]]
@@ -113,7 +113,7 @@ define void @test3(i32 %n) {
; CHECK-NEXT: [[CMP:%.*]] = icmp sgt i32 [[N:%.*]], -1
; CHECK-NEXT: br i1 [[CMP]], label [[BB:%.*]], label [[EXIT:%.*]]
; CHECK: bb:
-; CHECK-NEXT: [[EXT_WIDE:%.*]] = sitofp i32 [[N]] to float
+; CHECK-NEXT: [[EXT_WIDE:%.*]] = uitofp nneg i32 [[N]] to float
; CHECK-NEXT: call void @use.f32(float [[EXT_WIDE]])
; CHECK-NEXT: br label [[EXIT]]
; CHECK: exit:
@@ -192,7 +192,7 @@ exit:
define double @test_infer_at_use(i32 noundef %n) {
; CHECK-LABEL: @test_infer_at_use(
; CHECK-NEXT: [[CMP:%.*]] = icmp sgt i32 [[N:%.*]], -1
-; CHECK-NEXT: [[EXT:%.*]] = sitofp i32 [[N]] to double
+; CHECK-NEXT: [[EXT:%.*]] = uitofp nneg i32 [[N]] to double
; CHECK-NEXT: [[SELECT:%.*]] = select i1 [[CMP]], double [[EXT]], double 0.000000e+00
; CHECK-NEXT: ret double [[SELECT]]
;
diff --git a/llvm/test/Transforms/CorrelatedValuePropagation/uitofp.ll b/llvm/test/Transforms/CorrelatedValuePropagation/uitofp.ll
index 1c9dc21c26feb7..8cb3bb366be100 100644
--- a/llvm/test/Transforms/CorrelatedValuePropagation/uitofp.ll
+++ b/llvm/test/Transforms/CorrelatedValuePropagation/uitofp.ll
@@ -12,7 +12,7 @@ define void @test1_fptosi(i32 %n) {
; CHECK-NEXT: [[CMP:%.*]] = icmp sgt i32 [[A]], -1
; CHECK-NEXT: br i1 [[CMP]], label [[FOR_BODY]], label [[FOR_END:%.*]]
; CHECK: for.body:
-; CHECK-NEXT: [[EXT_WIDE:%.*]] = uitofp i32 [[A]] to float
+; CHECK-NEXT: [[EXT_WIDE:%.*]] = uitofp nneg i32 [[A]] to float
; CHECK-NEXT: call void @use.f32(float [[EXT_WIDE]])
; CHECK-NEXT: [[EXT]] = fptosi float [[EXT_WIDE]] to i32
; CHECK-NEXT: br label [[FOR_COND]]
@@ -46,7 +46,7 @@ define void @test1_fptoui(i32 %n) {
; CHECK-NEXT: [[CMP:%.*]] = icmp sgt i32 [[A]], -1
; CHECK-NEXT: br i1 [[CMP]], label [[FOR_BODY]], label [[FOR_END:%.*]]
; CHECK: for.body:
-; CHECK-NEXT: [[EXT_WIDE:%.*]] = uitofp i32 [[A]] to float
+; CHECK-NEXT: [[EXT_WIDE:%.*]] = uitofp nneg i32 [[A]] to float
; CHECK-NEXT: call void @use.f32(float [[EXT_WIDE]])
; CHECK-NEXT: [[EXT]] = fptoui float [[EXT_WIDE]] to i32
; CHECK-NEXT: br label [[FOR_COND]]
@@ -113,7 +113,7 @@ define void @test3(i32 %n) {
; CHECK-NEXT: [[CMP:%.*]] = icmp sgt i32 [[N:%.*]], -1
; CHECK-NEXT: br i1 [[CMP]], label [[BB:%.*]], label [[EXIT:%.*]]
; CHECK: bb:
-; CHECK-NEXT: [[EXT_WIDE:%.*]] = uitofp i32 [[N]] to float
+; CHECK-NEXT: [[EXT_WIDE:%.*]] = uitofp nneg i32 [[N]] to float
; CHECK-NEXT: call void @use.f32(float [[EXT_WIDE]])
; CHECK-NEXT: br label [[EXIT]]
; CHECK: exit:
@@ -192,7 +192,7 @@ exit:
define double @test_infer_at_use(i32 noundef %n) {
; CHECK-LABEL: @test_infer_at_use(
; CHECK-NEXT: [[CMP:%.*]] = icmp sgt i32 [[N:%.*]], -1
-; CHECK-NEXT: [[EXT:%.*]] = uitofp i32 [[N]] to double
+; CHECK-NEXT: [[EXT:%.*]] = uitofp nneg i32 [[N]] to double
; CHECK-NEXT: [[SELECT:%.*]] = select i1 [[CMP]], double [[EXT]], double 0.000000e+00
; CHECK-NEXT: ret double [[SELECT]]
;
>From 13af0e0e89e495dfc0c59d66a64a871eb31b9472 Mon Sep 17 00:00:00 2001
From: Noah Goldstein <goldstein.w.n at gmail.com>
Date: Wed, 20 Mar 2024 17:05:03 -0500
Subject: [PATCH 3/3] [SCCP] Add `nneg` flag to `uitofp` if its operand is
non-negative
Similiar to the `InstCombine` changes, just furthering the support of
the `uitofp nneg` support.
---
llvm/lib/Transforms/Utils/SCCPSolver.cpp | 12 +++++++-----
llvm/test/Transforms/SCCP/ip-ranges-casts.ll | 6 +++---
llvm/test/Transforms/SCCP/sitofp.ll | 8 ++++----
3 files changed, 14 insertions(+), 12 deletions(-)
diff --git a/llvm/lib/Transforms/Utils/SCCPSolver.cpp b/llvm/lib/Transforms/Utils/SCCPSolver.cpp
index c6029b428ed398..ce40e8b31b767a 100644
--- a/llvm/lib/Transforms/Utils/SCCPSolver.cpp
+++ b/llvm/lib/Transforms/Utils/SCCPSolver.cpp
@@ -143,7 +143,7 @@ static bool refineInstruction(SCCPSolver &Solver,
Changed = true;
}
}
- } else if (isa<ZExtInst>(Inst) && !Inst.hasNonNeg()) {
+ } else if (isa<PossiblyNonNegInst>(Inst) && !Inst.hasNonNeg()) {
auto Range = GetRange(Inst.getOperand(0));
if (Range.isAllNonNegative()) {
Inst.setNonNeg();
@@ -191,14 +191,16 @@ static bool replaceSignedInst(SCCPSolver &Solver,
Instruction *NewInst = nullptr;
switch (Inst.getOpcode()) {
- // Note: We do not fold sitofp -> uitofp here because that could be more
- // expensive in codegen and may not be reversible in the backend.
+ case Instruction::SIToFP:
case Instruction::SExt: {
- // If the source value is not negative, this is a zext.
+ // If the source value is not negative, this is a zext/uitofp.
Value *Op0 = Inst.getOperand(0);
if (InsertedValues.count(Op0) || !isNonNegative(Op0))
return false;
- NewInst = new ZExtInst(Op0, Inst.getType(), "", Inst.getIterator());
+ NewInst = CastInst::Create(Inst.getOpcode() == Instruction::SExt
+ ? Instruction::ZExt
+ : Instruction::UIToFP,
+ Op0, Inst.getType(), "", Inst.getIterator());
NewInst->setNonNeg();
break;
}
diff --git a/llvm/test/Transforms/SCCP/ip-ranges-casts.ll b/llvm/test/Transforms/SCCP/ip-ranges-casts.ll
index 05fa04a9fbe06f..e8d417546def85 100644
--- a/llvm/test/Transforms/SCCP/ip-ranges-casts.ll
+++ b/llvm/test/Transforms/SCCP/ip-ranges-casts.ll
@@ -167,7 +167,7 @@ define i1 @caller.sext() {
define internal i1 @f.fptosi(i32 %x) {
; CHECK-LABEL: define internal i1 @f.fptosi(
; CHECK-SAME: i32 [[X:%.*]]) {
-; CHECK-NEXT: [[TO_DOUBLE:%.*]] = sitofp i32 [[X]] to double
+; CHECK-NEXT: [[TO_DOUBLE:%.*]] = uitofp nneg i32 [[X]] to double
; CHECK-NEXT: [[ADD:%.*]] = fadd double 0.000000e+00, [[TO_DOUBLE]]
; CHECK-NEXT: [[TO_I32:%.*]] = fptosi double [[ADD]] to i32
; CHECK-NEXT: [[C_1:%.*]] = icmp sgt i32 [[TO_I32]], 300
@@ -209,7 +209,7 @@ define i1 @caller.fptosi() {
define internal i1 @f.fpext(i16 %x) {
; CHECK-LABEL: define internal i1 @f.fpext(
; CHECK-SAME: i16 [[X:%.*]]) {
-; CHECK-NEXT: [[TO_FLOAT:%.*]] = sitofp i16 [[X]] to float
+; CHECK-NEXT: [[TO_FLOAT:%.*]] = uitofp nneg i16 [[X]] to float
; CHECK-NEXT: [[TO_DOUBLE:%.*]] = fpext float [[TO_FLOAT]] to double
; CHECK-NEXT: [[TO_I64:%.*]] = fptoui float [[TO_FLOAT]] to i64
; CHECK-NEXT: [[C_1:%.*]] = icmp sgt i64 [[TO_I64]], 300
@@ -293,7 +293,7 @@ define i1 @int_range_to_double_cast(i32 %a) {
; CHECK-LABEL: define i1 @int_range_to_double_cast(
; CHECK-SAME: i32 [[A:%.*]]) {
; CHECK-NEXT: [[R:%.*]] = and i32 [[A]], 255
-; CHECK-NEXT: [[T4:%.*]] = sitofp i32 [[R]] to double
+; CHECK-NEXT: [[T4:%.*]] = uitofp nneg i32 [[R]] to double
; CHECK-NEXT: [[T10:%.*]] = fadd double 0.000000e+00, [[T4]]
; CHECK-NEXT: [[T11:%.*]] = fcmp olt double [[T4]], [[T10]]
; CHECK-NEXT: ret i1 [[T11]]
diff --git a/llvm/test/Transforms/SCCP/sitofp.ll b/llvm/test/Transforms/SCCP/sitofp.ll
index b635263a5726c2..24f04ae1fccb91 100644
--- a/llvm/test/Transforms/SCCP/sitofp.ll
+++ b/llvm/test/Transforms/SCCP/sitofp.ll
@@ -4,7 +4,7 @@
define float @sitofp_and(i8 %x) {
; CHECK-LABEL: @sitofp_and(
; CHECK-NEXT: [[PX:%.*]] = and i8 [[X:%.*]], 127
-; CHECK-NEXT: [[R:%.*]] = sitofp i8 [[PX]] to float
+; CHECK-NEXT: [[R:%.*]] = uitofp nneg i8 [[PX]] to float
; CHECK-NEXT: ret float [[R]]
;
%px = and i8 %x, 127
@@ -23,7 +23,7 @@ define half @sitofp_const(i8 %x) {
define double @sitofp_zext(i7 %x) {
; CHECK-LABEL: @sitofp_zext(
; CHECK-NEXT: [[PX:%.*]] = zext i7 [[X:%.*]] to i8
-; CHECK-NEXT: [[R:%.*]] = sitofp i8 [[PX]] to double
+; CHECK-NEXT: [[R:%.*]] = uitofp nneg i8 [[PX]] to double
; CHECK-NEXT: ret double [[R]]
;
%px = zext i7 %x to i8
@@ -52,7 +52,7 @@ define float @dominating_condition(i32 %x) {
; CHECK-NEXT: [[CMP:%.*]] = icmp sge i32 [[X:%.*]], 0
; CHECK-NEXT: br i1 [[CMP]], label [[T:%.*]], label [[F:%.*]]
; CHECK: t:
-; CHECK-NEXT: [[A:%.*]] = sitofp i32 [[X]] to float
+; CHECK-NEXT: [[A:%.*]] = uitofp nneg i32 [[X]] to float
; CHECK-NEXT: br label [[EXIT:%.*]]
; CHECK: f:
; CHECK-NEXT: br label [[EXIT]]
@@ -86,7 +86,7 @@ define float @dominating_condition_alt(i32 %x) {
; CHECK: t:
; CHECK-NEXT: br label [[EXIT:%.*]]
; CHECK: f:
-; CHECK-NEXT: [[A:%.*]] = sitofp i32 [[X]] to float
+; CHECK-NEXT: [[A:%.*]] = uitofp nneg i32 [[X]] to float
; CHECK-NEXT: br label [[EXIT]]
; CHECK: exit:
; CHECK-NEXT: [[COND:%.*]] = phi float [ -4.200000e+01, [[T]] ], [ [[A]], [[F]] ]
More information about the llvm-commits
mailing list