[llvm] [VPlan] Set debug location for VPReduction/VPWidenPHI/VPWidenIntrinsicRecipe. (PR #120054)
Elvis Wang via llvm-commits
llvm-commits at lists.llvm.org
Mon Dec 16 00:33:57 PST 2024
https://github.com/ElvisWang123 created https://github.com/llvm/llvm-project/pull/120054
This patch add missing debug location for VPReduction/VPWidenIntrinsicRecipe and align the debug location for
VPWidenPHIRecipe to its scalar counterpart.
Depends on #120053.
>From a7d91131c2e56651f4b9546bfc10287fdee415ce Mon Sep 17 00:00:00 2001
From: Elvis Wang <elvis.wang at sifive.com>
Date: Sun, 15 Dec 2024 19:45:13 -0800
Subject: [PATCH 1/2] Precommit test case.
---
.../LoopVectorize/RISCV/preserve-dbg-loc.ll | 36 +++++++++++++++++++
.../LoopVectorize/preserve-dbg-loc.ll | 33 +++++++++++++++++
2 files changed, 69 insertions(+)
create mode 100644 llvm/test/Transforms/LoopVectorize/RISCV/preserve-dbg-loc.ll
create mode 100644 llvm/test/Transforms/LoopVectorize/preserve-dbg-loc.ll
diff --git a/llvm/test/Transforms/LoopVectorize/RISCV/preserve-dbg-loc.ll b/llvm/test/Transforms/LoopVectorize/RISCV/preserve-dbg-loc.ll
new file mode 100644
index 00000000000000..5993dc26a0e8fe
--- /dev/null
+++ b/llvm/test/Transforms/LoopVectorize/RISCV/preserve-dbg-loc.ll
@@ -0,0 +1,36 @@
+; RUN: opt -passes=debugify,loop-vectorize \
+; RUN: -force-tail-folding-style=data-with-evl \
+; RUN: -prefer-predicate-over-epilogue=predicate-dont-vectorize \
+; RUN: -mtriple=riscv64 -mattr=+v -riscv-v-vector-bits-max=128 -S < %s 2>&1 | FileCheck --check-prefix=DEBUGLOC %s
+
+; Testing the debug locations of the generated vector intstruction are same as
+; their scalar instruction.
+
+; DEBUGLOC-LABEL: define void @vp_select(
+define void @vp_select(ptr noalias %a, ptr noalias %b, ptr noalias %c, i64 %N) {
+; DEBUGLOC: vector.body:
+; DEBUGLOC: %[[VPSel:[0-9]+]] = call <vscale x 4 x i32> @llvm.vp.select.nxv4i32(<vscale x 4 x i1> %15, <vscale x 4 x i32> %vp.op.load1, <vscale x 4 x i32> %vp.op, i32 %9)
+; DEBUGLOC: for.body:
+; DEBUGLOC: %cond.p = select i1 %cmp4, i32 %22, i32 %23, !dbg !39
+ entry:
+ br label %for.body
+
+ for.body:
+ %indvars.iv = phi i64 [ %indvars.iv.next, %for.body ], [ 0, %entry ]
+ %arrayidx = getelementptr inbounds i32, ptr %b, i64 %indvars.iv
+ %0 = load i32, ptr %arrayidx, align 4
+ %arrayidx3 = getelementptr inbounds i32, ptr %c, i64 %indvars.iv
+ %1 = load i32, ptr %arrayidx3, align 4
+ %cmp4 = icmp sgt i32 %0, %1
+ %2 = sub i32 0, %1
+ %cond.p = select i1 %cmp4, i32 %1, i32 %2
+ %cond = add i32 %cond.p, %0
+ %arrayidx15 = getelementptr inbounds i32, ptr %a, i64 %indvars.iv
+ store i32 %cond, ptr %arrayidx15, align 4
+ %indvars.iv.next = add nuw nsw i64 %indvars.iv, 1
+ %exitcond.not = icmp eq i64 %indvars.iv.next, %N
+ br i1 %exitcond.not, label %exit, label %for.body
+
+ exit:
+ ret void
+ }
diff --git a/llvm/test/Transforms/LoopVectorize/preserve-dbg-loc.ll b/llvm/test/Transforms/LoopVectorize/preserve-dbg-loc.ll
new file mode 100644
index 00000000000000..02cac45f1c5cc5
--- /dev/null
+++ b/llvm/test/Transforms/LoopVectorize/preserve-dbg-loc.ll
@@ -0,0 +1,33 @@
+; RUN: opt < %s -passes=debugify,loop-vectorize -force-vector-interleave=1 -force-vector-width=4 -prefer-inloop-reductions -S | FileCheck %s -check-prefix DEBUGLOC
+target datalayout = "e-m:e-i64:64-f80:128-n8:16:32:64-S128"
+
+; Testing the debug locations of the generated vector intstruction are same as
+; their scalar instruction.
+
+; DEBUGLOC-LABEL: define i32 @reduction_sum(
+define i32 @reduction_sum(ptr noalias nocapture %A, ptr noalias nocapture %B) {
+; DEBUGLOC: vector.body:
+; DEBUGLOC: %[[VecLoad:.*]] = load <4 x i32>, ptr %2, align 4, !dbg ![[LoadLoc0:[0-9]+]]
+; DEBUGLOC: %[[VecRed:.*]] = call i32 @llvm.vector.reduce.add.v4i32(<4 x i32> %wide.load), !dbg ![[LoadLoc0]]
+; DEBUGLOC: .lr.ph:
+; DEBUGLOC: %l3 = load i32, ptr %l2, align 4, !dbg ![[LoadLoc0]]
+; DEBUGLOC: %l7 = add i32 %sum.02, %l3, !dbg ![[RedLoc0:[0-9]+]]
+entry:
+ br label %.lr.ph
+
+.lr.ph: ; preds = %entry, %.lr.ph
+ %indvars.iv = phi i64 [ %indvars.iv.next, %.lr.ph ], [ 0, %entry ]
+ %sum.02 = phi i32 [ %l7, %.lr.ph ], [ 0, %entry ]
+ %l2 = getelementptr inbounds i32, ptr %A, i64 %indvars.iv
+ %l3 = load i32, ptr %l2, align 4
+ %l6 = trunc i64 %indvars.iv to i32
+ %l7 = add i32 %sum.02, %l3
+ %indvars.iv.next = add i64 %indvars.iv, 1
+ %lftr.wideiv = trunc i64 %indvars.iv.next to i32
+ %exitcond = icmp eq i32 %lftr.wideiv, 256
+ br i1 %exitcond, label %._crit_edge, label %.lr.ph
+
+._crit_edge: ; preds = %.lr.ph
+ %sum.0.lcssa = phi i32 [ %l7, %.lr.ph ]
+ ret i32 %sum.0.lcssa
+}
>From eb41b83c8799e68a1b79699d9511b5c64b1b6991 Mon Sep 17 00:00:00 2001
From: Elvis Wang <elvis.wang at sifive.com>
Date: Sun, 15 Dec 2024 21:45:32 -0800
Subject: [PATCH 2/2] [VPlan] Set debug loc for
VPReduction/VPWidenPHI/VPWidenIntrinsicRecipe.
This patch add missing debug location for
VPReduction/VPWidenIntrinsicRecipe and align the debug location for
VPWidenPHIRecipe to its scalar counterpart.
---
llvm/lib/Transforms/Vectorize/VPlan.h | 8 +++++---
llvm/lib/Transforms/Vectorize/VPlanRecipes.cpp | 2 ++
.../Transforms/LoopVectorize/RISCV/preserve-dbg-loc.ll | 4 ++--
llvm/test/Transforms/LoopVectorize/dbg-outer-loop-vect.ll | 4 ++--
llvm/test/Transforms/LoopVectorize/preserve-dbg-loc.ll | 4 ++--
5 files changed, 13 insertions(+), 9 deletions(-)
diff --git a/llvm/lib/Transforms/Vectorize/VPlan.h b/llvm/lib/Transforms/Vectorize/VPlan.h
index 8794517b777f3b..6ebf2ec36e3789 100644
--- a/llvm/lib/Transforms/Vectorize/VPlan.h
+++ b/llvm/lib/Transforms/Vectorize/VPlan.h
@@ -1662,7 +1662,7 @@ class VPWidenIntrinsicRecipe : public VPRecipeWithIRFlags {
VPWidenIntrinsicRecipe(Intrinsic::ID VectorIntrinsicID,
ArrayRef<VPValue *> CallArguments, Type *Ty,
DebugLoc DL = {})
- : VPRecipeWithIRFlags(VPDef::VPWidenIntrinsicSC, CallArguments),
+ : VPRecipeWithIRFlags(VPDef::VPWidenIntrinsicSC, CallArguments, DL),
VectorIntrinsicID(VectorIntrinsicID), ResultTy(Ty) {
LLVMContext &Ctx = Ty->getContext();
AttributeList Attrs = Intrinsic::getAttributes(Ctx, VectorIntrinsicID);
@@ -2288,7 +2288,8 @@ class VPWidenPHIRecipe : public VPSingleDefRecipe {
public:
/// Create a new VPWidenPHIRecipe for \p Phi with start value \p Start.
VPWidenPHIRecipe(PHINode *Phi, VPValue *Start = nullptr)
- : VPSingleDefRecipe(VPDef::VPWidenPHISC, ArrayRef<VPValue *>(), Phi) {
+ : VPSingleDefRecipe(VPDef::VPWidenPHISC, ArrayRef<VPValue *>(), Phi,
+ Phi->getDebugLoc()) {
if (Start)
addOperand(Start);
}
@@ -2590,7 +2591,8 @@ class VPReductionRecipe : public VPSingleDefRecipe {
VPReductionRecipe(const unsigned char SC, const RecurrenceDescriptor &R,
Instruction *I, ArrayRef<VPValue *> Operands,
VPValue *CondOp, bool IsOrdered)
- : VPSingleDefRecipe(SC, Operands, I), RdxDesc(R), IsOrdered(IsOrdered) {
+ : VPSingleDefRecipe(SC, Operands, I, I->getDebugLoc()), RdxDesc(R),
+ IsOrdered(IsOrdered) {
if (CondOp) {
IsConditional = true;
addOperand(CondOp);
diff --git a/llvm/lib/Transforms/Vectorize/VPlanRecipes.cpp b/llvm/lib/Transforms/Vectorize/VPlanRecipes.cpp
index e882368544e815..4179a9eebab3b7 100644
--- a/llvm/lib/Transforms/Vectorize/VPlanRecipes.cpp
+++ b/llvm/lib/Transforms/Vectorize/VPlanRecipes.cpp
@@ -2108,6 +2108,7 @@ void VPReductionRecipe::execute(VPTransformState &State) {
// Propagate the fast-math flags carried by the underlying instruction.
IRBuilderBase::FastMathFlagGuard FMFGuard(State.Builder);
State.Builder.setFastMathFlags(RdxDesc.getFastMathFlags());
+ State.setDebugLocFrom(getDebugLoc());
Value *NewVecOp = State.get(getVecOp());
if (VPValue *Cond = getCondOp()) {
Value *NewCond = State.get(Cond, State.VF.isScalar());
@@ -3424,6 +3425,7 @@ void VPWidenPHIRecipe::execute(VPTransformState &State) {
assert(EnableVPlanNativePath &&
"Non-native vplans are not expected to have VPWidenPHIRecipes.");
+ State.setDebugLocFrom(getDebugLoc());
Value *Op0 = State.get(getOperand(0));
Type *VecTy = Op0->getType();
Value *VecPhi = State.Builder.CreatePHI(VecTy, 2, "vec.phi");
diff --git a/llvm/test/Transforms/LoopVectorize/RISCV/preserve-dbg-loc.ll b/llvm/test/Transforms/LoopVectorize/RISCV/preserve-dbg-loc.ll
index 5993dc26a0e8fe..9998c44e6e3907 100644
--- a/llvm/test/Transforms/LoopVectorize/RISCV/preserve-dbg-loc.ll
+++ b/llvm/test/Transforms/LoopVectorize/RISCV/preserve-dbg-loc.ll
@@ -9,9 +9,9 @@
; DEBUGLOC-LABEL: define void @vp_select(
define void @vp_select(ptr noalias %a, ptr noalias %b, ptr noalias %c, i64 %N) {
; DEBUGLOC: vector.body:
-; DEBUGLOC: %[[VPSel:[0-9]+]] = call <vscale x 4 x i32> @llvm.vp.select.nxv4i32(<vscale x 4 x i1> %15, <vscale x 4 x i32> %vp.op.load1, <vscale x 4 x i32> %vp.op, i32 %9)
+; DEBUGLOC: %[[VPSel:[0-9]+]] = call <vscale x 4 x i32> @llvm.vp.select.nxv4i32(<vscale x 4 x i1> %15, <vscale x 4 x i32> %vp.op.load1, <vscale x 4 x i32> %vp.op, i32 %9), !dbg ![[SelLoc:[0-9]+]]
; DEBUGLOC: for.body:
-; DEBUGLOC: %cond.p = select i1 %cmp4, i32 %22, i32 %23, !dbg !39
+; DEBUGLOC: %cond.p = select i1 %cmp4, i32 %22, i32 %23, !dbg ![[SelLoc]]
entry:
br label %for.body
diff --git a/llvm/test/Transforms/LoopVectorize/dbg-outer-loop-vect.ll b/llvm/test/Transforms/LoopVectorize/dbg-outer-loop-vect.ll
index 66aceab9fb27c8..44afa34100c299 100644
--- a/llvm/test/Transforms/LoopVectorize/dbg-outer-loop-vect.ll
+++ b/llvm/test/Transforms/LoopVectorize/dbg-outer-loop-vect.ll
@@ -15,8 +15,8 @@ define void @foo(ptr %h) !dbg !4 {
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[FOR_COND_CLEANUP32:%.*]] ]
; CHECK-NEXT: br label [[FOR_COND5_PREHEADER1:%.*]], !dbg [[DBG21]]
; CHECK: for.cond5.preheader1:
-; CHECK-NEXT: [[VEC_PHI:%.*]] = phi <4 x i64> [ zeroinitializer, [[VECTOR_BODY]] ], [ [[TMP4:%.*]], [[FOR_COND5_PREHEADER1]] ], !dbg [[DBG21]]
-; CHECK-NEXT: [[TMP0:%.*]] = getelementptr i32, ptr [[H]], <4 x i64> [[VEC_PHI]], !dbg [[DBG21]]
+; CHECK-NEXT: [[VEC_PHI:%.*]] = phi <4 x i64> [ zeroinitializer, [[VECTOR_BODY]] ], [ [[TMP4:%.*]], [[FOR_COND5_PREHEADER1]] ]
+; CHECK-NEXT: [[TMP0:%.*]] = getelementptr i32, ptr [[H]], <4 x i64> [[VEC_PHI]]
; CHECK-NEXT: call void @llvm.masked.scatter.v4i32.v4p0(<4 x i32> zeroinitializer, <4 x ptr> [[TMP0]], i32 4, <4 x i1> splat (i1 true)), !dbg [[DBG22:![0-9]+]]
; CHECK-NEXT: [[TMP1:%.*]] = getelementptr i32, <4 x ptr> [[TMP0]], i64 1, !dbg [[DBG22]]
; CHECK-NEXT: call void @llvm.masked.scatter.v4i32.v4p0(<4 x i32> splat (i32 1), <4 x ptr> [[TMP1]], i32 4, <4 x i1> splat (i1 true)), !dbg [[DBG22]]
diff --git a/llvm/test/Transforms/LoopVectorize/preserve-dbg-loc.ll b/llvm/test/Transforms/LoopVectorize/preserve-dbg-loc.ll
index 02cac45f1c5cc5..12e8b92523d31f 100644
--- a/llvm/test/Transforms/LoopVectorize/preserve-dbg-loc.ll
+++ b/llvm/test/Transforms/LoopVectorize/preserve-dbg-loc.ll
@@ -8,10 +8,10 @@ target datalayout = "e-m:e-i64:64-f80:128-n8:16:32:64-S128"
define i32 @reduction_sum(ptr noalias nocapture %A, ptr noalias nocapture %B) {
; DEBUGLOC: vector.body:
; DEBUGLOC: %[[VecLoad:.*]] = load <4 x i32>, ptr %2, align 4, !dbg ![[LoadLoc0:[0-9]+]]
-; DEBUGLOC: %[[VecRed:.*]] = call i32 @llvm.vector.reduce.add.v4i32(<4 x i32> %wide.load), !dbg ![[LoadLoc0]]
+; DEBUGLOC: %[[VecRed:.*]] = call i32 @llvm.vector.reduce.add.v4i32(<4 x i32> %wide.load), !dbg ![[RedLoc0:[0-9]+]]
; DEBUGLOC: .lr.ph:
; DEBUGLOC: %l3 = load i32, ptr %l2, align 4, !dbg ![[LoadLoc0]]
-; DEBUGLOC: %l7 = add i32 %sum.02, %l3, !dbg ![[RedLoc0:[0-9]+]]
+; DEBUGLOC: %l7 = add i32 %sum.02, %l3, !dbg ![[RedLoc0]]
entry:
br label %.lr.ph
More information about the llvm-commits
mailing list