[RISCV][GISEL] Legalize G_VAARG through expansion. (#73065)

G_VAARG can be expanded similiar to SelectionDAG::expandVAArg through
LegalizerHelper::lower. This patch implements the lowering through this
style of expansion.

The expansion gets the head of the va_list by loading the pointer to
va_list. Then, the head of the list is adjusted depending on argument
alignment information. This gives a pointer to the element to be read
out of the va_list. Next, the head of the va_list is bumped to the next
element in the list. The new head of the list is stored back to the
original pointer to the head of the va_list so that subsequent G_VAARG
instructions get the next element in the list. Lastly, the element is
loaded from the alignment adjusted pointer constructed earlier.

This change is stacked on #73062.

5 files changed, 224 insertions, 0 deletions

diff --git a/llvm/include/llvm/CodeGen/GlobalISel/LegalizerHelper.h b/llvm/include/llvm/CodeGen/GlobalISel/LegalizerHelper.h
index 365d222..711ba10 100644
--- a/llvm/include/llvm/CodeGen/GlobalISel/LegalizerHelper.h
+++ b/llvm/include/llvm/CodeGen/GlobalISel/LegalizerHelper.h
@@ -432,6 +432,7 @@ public:
   LegalizeResult lowerVectorReduction(MachineInstr &MI);
   LegalizeResult lowerMemcpyInline(MachineInstr &MI);
   LegalizeResult lowerMemCpyFamily(MachineInstr &MI, unsigned MaxLen = 0);
+  LegalizeResult lowerVAArg(MachineInstr &MI);
 };
 
 /// Helper function that creates a libcall to the given \p Name using the given
diff --git a/llvm/lib/CodeGen/GlobalISel/LegalizerHelper.cpp b/llvm/lib/CodeGen/GlobalISel/LegalizerHelper.cpp
index 045fc78..37e7153 100644
--- a/llvm/lib/CodeGen/GlobalISel/LegalizerHelper.cpp
+++ b/llvm/lib/CodeGen/GlobalISel/LegalizerHelper.cpp
@@ -3793,6 +3793,8 @@ LegalizerHelper::lower(MachineInstr &MI, unsigned TypeIdx, LLT LowerHintTy) {
     return lowerTRUNC(MI);
   GISEL_VECREDUCE_CASES_NONSEQ
     return lowerVectorReduction(MI);
+  case G_VAARG:
+    return lowerVAArg(MI);
   }
 }
 
@@ -7887,6 +7889,56 @@ LegalizerHelper::lowerVectorReduction(MachineInstr &MI) {
   return UnableToLegalize;
 }
 
+static Type *getTypeForLLT(LLT Ty, LLVMContext &C);
+
+LegalizerHelper::LegalizeResult LegalizerHelper::lowerVAArg(MachineInstr &MI) {
+  MachineFunction &MF = *MI.getMF();
+  const DataLayout &DL = MIRBuilder.getDataLayout();
+  LLVMContext &Ctx = MF.getFunction().getContext();
+  Register ListPtr = MI.getOperand(1).getReg();
+  LLT PtrTy = MRI.getType(ListPtr);
+
+  // LstPtr is a pointer to the head of the list. Get the address
+  // of the head of the list.
+  Align PtrAlignment = DL.getABITypeAlign(getTypeForLLT(PtrTy, Ctx));
+  MachineMemOperand *PtrLoadMMO = MF.getMachineMemOperand(
+      MachinePointerInfo(), MachineMemOperand::MOLoad, PtrTy, PtrAlignment);
+  auto VAList = MIRBuilder.buildLoad(PtrTy, ListPtr, *PtrLoadMMO).getReg(0);
+
+  const Align A(MI.getOperand(2).getImm());
+  LLT PtrTyAsScalarTy = LLT::scalar(PtrTy.getSizeInBits());
+  if (A > TLI.getMinStackArgumentAlignment()) {
+    Register AlignAmt =
+        MIRBuilder.buildConstant(PtrTyAsScalarTy, A.value() - 1).getReg(0);
+    auto AddDst = MIRBuilder.buildPtrAdd(PtrTy, VAList, AlignAmt);
+    auto AndDst = MIRBuilder.buildMaskLowPtrBits(PtrTy, AddDst, Log2(A));
+    VAList = AndDst.getReg(0);
+  }
+
+  // Increment the pointer, VAList, to the next vaarg
+  // The list should be bumped by the size of element in the current head of
+  // list.
+  Register Dst = MI.getOperand(0).getReg();
+  LLT LLTTy = MRI.getType(Dst);
+  Type *Ty = getTypeForLLT(LLTTy, Ctx);
+  auto IncAmt =
+      MIRBuilder.buildConstant(PtrTyAsScalarTy, DL.getTypeAllocSize(Ty));
+  auto Succ = MIRBuilder.buildPtrAdd(PtrTy, VAList, IncAmt);
+
+  // Store the increment VAList to the legalized pointer
+  MachineMemOperand *StoreMMO = MF.getMachineMemOperand(
+      MachinePointerInfo(), MachineMemOperand::MOStore, PtrTy, PtrAlignment);
+  MIRBuilder.buildStore(Succ, ListPtr, *StoreMMO);
+  // Load the actual argument out of the pointer VAList
+  Align EltAlignment = DL.getABITypeAlign(Ty);
+  MachineMemOperand *EltLoadMMO = MF.getMachineMemOperand(
+      MachinePointerInfo(), MachineMemOperand::MOLoad, LLTTy, EltAlignment);
+  MIRBuilder.buildLoad(Dst, VAList, *EltLoadMMO);
+
+  MI.eraseFromParent();
+  return Legalized;
+}
+
 static bool shouldLowerMemFuncForSize(const MachineFunction &MF) {
   // On Darwin, -Os means optimize for size without hurting performance, so
   // only really optimize for size when -Oz (MinSize) is used.
diff --git a/llvm/lib/Target/RISCV/GISel/RISCVLegalizerInfo.cpp b/llvm/lib/Target/RISCV/GISel/RISCVLegalizerInfo.cpp
index 570b980..d71a6cf 100644
--- a/llvm/lib/Target/RISCV/GISel/RISCVLegalizerInfo.cpp
+++ b/llvm/lib/Target/RISCV/GISel/RISCVLegalizerInfo.cpp
@@ -317,6 +317,14 @@ RISCVLegalizerInfo::RISCVLegalizerInfo(const RISCVSubtarget &ST)
 
   getActionDefinitionsBuilder(G_VASTART).customFor({p0});
 
+  // va_list must be a pointer, but most sized types are pretty easy to handle
+  // as the destination.
+  getActionDefinitionsBuilder(G_VAARG)
+      // TODO: Implement narrowScalar and widenScalar for G_VAARG for types
+      // outside the [s32, sXLen] range.
+      .clampScalar(0, s32, sXLen)
+      .lowerForCartesianProduct({s32, sXLen, p0}, {p0});
+
   getLegacyLegalizerInfo().computeTables();
 }
 
diff --git a/llvm/test/CodeGen/RISCV/GlobalISel/legalizer/legalize-vaarg-rv32.mir b/llvm/test/CodeGen/RISCV/GlobalISel/legalizer/legalize-vaarg-rv32.mir
new file mode 100644
index 0000000..b67691d
--- /dev/null
+++ b/llvm/test/CodeGen/RISCV/GlobalISel/legalizer/legalize-vaarg-rv32.mir
@@ -0,0 +1,70 @@
+# NOTE: Assertions have been autogenerated by utils/update_mir_test_checks.py
+# RUN: llc -mtriple=riscv32 -run-pass=legalizer %s -o - | FileCheck %s
+
+# On RISC-V, the MinStackArgumentAlignment is 1 and the ABI Alignment for p0 is
+# greater than 1, so we will always generate code to adjust for this alignment.
+
+---
+name:            va_arg_i32
+legalized:       false
+tracksRegLiveness: true
+fixedStack:
+  - { id: 0, type: default, offset: 0, size: 4, alignment: 16,
+      isImmutable: true, isAliased: false }
+stack:
+  - { id: 0, type: default, offset: 0, size: 4, alignment: 4 }
+machineFunctionInfo:
+  varArgsFrameIndex: -1
+  varArgsSaveSize: 0
+body:             |
+  bb.1:
+    liveins: $x10, $x11, $x12, $x13, $x14, $x15, $x16, $x17
+    ; CHECK-LABEL: name: va_arg_i32
+    ; CHECK: liveins: $x10, $x11, $x12, $x13, $x14, $x15, $x16, $x17
+    ; CHECK-NEXT: {{  $}}
+    ; CHECK-NEXT: [[FRAME_INDEX:%[0-9]+]]:_(p0) = G_FRAME_INDEX %stack.0
+    ; CHECK-NEXT: [[LOAD:%[0-9]+]]:_(p0) = G_LOAD [[FRAME_INDEX]](p0) :: (load (p0))
+    ; CHECK-NEXT: [[C:%[0-9]+]]:_(s32) = G_CONSTANT i32 3
+    ; CHECK-NEXT: [[PTR_ADD:%[0-9]+]]:_(p0) = G_PTR_ADD [[LOAD]], [[C]](s32)
+    ; CHECK-NEXT: [[C1:%[0-9]+]]:_(s32) = G_CONSTANT i32 -4
+    ; CHECK-NEXT: [[PTRMASK:%[0-9]+]]:_(p0) = G_PTRMASK [[PTR_ADD]], [[C1]](s32)
+    ; CHECK-NEXT: [[C2:%[0-9]+]]:_(s32) = G_CONSTANT i32 4
+    ; CHECK-NEXT: [[PTR_ADD1:%[0-9]+]]:_(p0) = G_PTR_ADD [[PTRMASK]], [[C2]](s32)
+    ; CHECK-NEXT: G_STORE [[PTR_ADD1]](p0), [[FRAME_INDEX]](p0) :: (store (p0))
+    ; CHECK-NEXT: PseudoRET
+    %0:_(p0) = G_FRAME_INDEX %stack.0
+    %1:_(s32) = G_VAARG %0(p0), 4
+    PseudoRET
+...
+---
+name:            va_arg_ptr
+legalized:       false
+tracksRegLiveness: true
+fixedStack:
+  - { id: 0, type: default, offset: 0, size: 4, alignment: 16,
+      isImmutable: true, isAliased: false }
+stack:
+  - { id: 0, type: default, offset: 0, size: 4, alignment: 4 }
+machineFunctionInfo:
+  varArgsFrameIndex: -1
+  varArgsSaveSize: 0
+body:             |
+  bb.1:
+    liveins: $x10, $x11, $x12, $x13, $x14, $x15, $x16, $x17
+    ; CHECK-LABEL: name: va_arg_ptr
+    ; CHECK: liveins: $x10, $x11, $x12, $x13, $x14, $x15, $x16, $x17
+    ; CHECK-NEXT: {{  $}}
+    ; CHECK-NEXT: [[FRAME_INDEX:%[0-9]+]]:_(p0) = G_FRAME_INDEX %stack.0
+    ; CHECK-NEXT: [[LOAD:%[0-9]+]]:_(p0) = G_LOAD [[FRAME_INDEX]](p0) :: (load (p0))
+    ; CHECK-NEXT: [[C:%[0-9]+]]:_(s32) = G_CONSTANT i32 3
+    ; CHECK-NEXT: [[PTR_ADD:%[0-9]+]]:_(p0) = G_PTR_ADD [[LOAD]], [[C]](s32)
+    ; CHECK-NEXT: [[C1:%[0-9]+]]:_(s32) = G_CONSTANT i32 -4
+    ; CHECK-NEXT: [[PTRMASK:%[0-9]+]]:_(p0) = G_PTRMASK [[PTR_ADD]], [[C1]](s32)
+    ; CHECK-NEXT: [[C2:%[0-9]+]]:_(s32) = G_CONSTANT i32 4
+    ; CHECK-NEXT: [[PTR_ADD1:%[0-9]+]]:_(p0) = G_PTR_ADD [[PTRMASK]], [[C2]](s32)
+    ; CHECK-NEXT: G_STORE [[PTR_ADD1]](p0), [[FRAME_INDEX]](p0) :: (store (p0))
+    ; CHECK-NEXT: PseudoRET
+    %0:_(p0) = G_FRAME_INDEX %stack.0
+    %1:_(p0) = G_VAARG %0(p0), 4
+    PseudoRET
+...
diff --git a/llvm/test/CodeGen/RISCV/GlobalISel/legalizer/legalize-vaarg-rv64.mir b/llvm/test/CodeGen/RISCV/GlobalISel/legalizer/legalize-vaarg-rv64.mir
new file mode 100644
index 0000000..537c0fb
--- /dev/null
+++ b/llvm/test/CodeGen/RISCV/GlobalISel/legalizer/legalize-vaarg-rv64.mir
@@ -0,0 +1,93 @@
+# NOTE: Assertions have been autogenerated by utils/update_mir_test_checks.py
+# RUN: llc -mtriple=riscv64 -run-pass=legalizer %s -o - | FileCheck %s
+
+# On RISC-V, the MinStackArgumentAlignment is 1 and the ABI Alignment for p0 is
+# greater than 1, so we will always generate code to adjust for this alignment.
+
+---
+name:            va_arg_i32
+legalized:       false
+tracksRegLiveness: true
+fixedStack:
+  - { id: 0, type: default, offset: 0, size: 8, alignment: 16,
+      isImmutable: true, isAliased: false }
+stack:
+  - { id: 0, type: default, offset: 0, size: 8, alignment: 8 }
+machineFunctionInfo:
+  varArgsFrameIndex: -1
+  varArgsSaveSize: 0
+body:             |
+  bb.1:
+    ; CHECK-LABEL: name: va_arg_i32
+    ; CHECK: [[FRAME_INDEX:%[0-9]+]]:_(p0) = G_FRAME_INDEX %stack.0
+    ; CHECK-NEXT: [[LOAD:%[0-9]+]]:_(p0) = G_LOAD [[FRAME_INDEX]](p0) :: (load (p0))
+    ; CHECK-NEXT: [[C:%[0-9]+]]:_(s64) = G_CONSTANT i64 3
+    ; CHECK-NEXT: [[PTR_ADD:%[0-9]+]]:_(p0) = G_PTR_ADD [[LOAD]], [[C]](s64)
+    ; CHECK-NEXT: [[C1:%[0-9]+]]:_(s64) = G_CONSTANT i64 -4
+    ; CHECK-NEXT: [[PTRMASK:%[0-9]+]]:_(p0) = G_PTRMASK [[PTR_ADD]], [[C1]](s64)
+    ; CHECK-NEXT: [[C2:%[0-9]+]]:_(s64) = G_CONSTANT i64 4
+    ; CHECK-NEXT: [[PTR_ADD1:%[0-9]+]]:_(p0) = G_PTR_ADD [[PTRMASK]], [[C2]](s64)
+    ; CHECK-NEXT: G_STORE [[PTR_ADD1]](p0), [[FRAME_INDEX]](p0) :: (store (p0))
+    ; CHECK-NEXT: PseudoRET
+    %0:_(p0) = G_FRAME_INDEX %stack.0
+    %1:_(s32) = G_VAARG %0(p0), 4
+    PseudoRET
+...
+---
+name:            va_arg_i64
+legalized:       false
+tracksRegLiveness: true
+fixedStack:
+  - { id: 0, type: default, offset: 0, size: 8, alignment: 16,
+      isImmutable: true, isAliased: false }
+stack:
+  - { id: 0, type: default, offset: 0, size: 8, alignment: 8 }
+machineFunctionInfo:
+  varArgsFrameIndex: -1
+  varArgsSaveSize: 0
+body:             |
+  bb.1:
+    ; CHECK-LABEL: name: va_arg_i64
+    ; CHECK: [[FRAME_INDEX:%[0-9]+]]:_(p0) = G_FRAME_INDEX %stack.0
+    ; CHECK-NEXT: [[LOAD:%[0-9]+]]:_(p0) = G_LOAD [[FRAME_INDEX]](p0) :: (load (p0))
+    ; CHECK-NEXT: [[C:%[0-9]+]]:_(s64) = G_CONSTANT i64 3
+    ; CHECK-NEXT: [[PTR_ADD:%[0-9]+]]:_(p0) = G_PTR_ADD [[LOAD]], [[C]](s64)
+    ; CHECK-NEXT: [[C1:%[0-9]+]]:_(s64) = G_CONSTANT i64 -4
+    ; CHECK-NEXT: [[PTRMASK:%[0-9]+]]:_(p0) = G_PTRMASK [[PTR_ADD]], [[C1]](s64)
+    ; CHECK-NEXT: [[C2:%[0-9]+]]:_(s64) = G_CONSTANT i64 8
+    ; CHECK-NEXT: [[PTR_ADD1:%[0-9]+]]:_(p0) = G_PTR_ADD [[PTRMASK]], [[C2]](s64)
+    ; CHECK-NEXT: G_STORE [[PTR_ADD1]](p0), [[FRAME_INDEX]](p0) :: (store (p0))
+    ; CHECK-NEXT: PseudoRET
+    %0:_(p0) = G_FRAME_INDEX %stack.0
+    %1:_(s64) = G_VAARG %0(p0), 4
+    PseudoRET
+...
+---
+name:            va_arg_ptr
+legalized:       false
+tracksRegLiveness: true
+fixedStack:
+  - { id: 0, type: default, offset: 0, size: 8, alignment: 16,
+      isImmutable: true, isAliased: false }
+stack:
+  - { id: 0, type: default, offset: 0, size: 8, alignment: 8 }
+machineFunctionInfo:
+  varArgsFrameIndex: -1
+  varArgsSaveSize: 0
+body:             |
+  bb.1:
+    ; CHECK-LABEL: name: va_arg_ptr
+    ; CHECK: [[FRAME_INDEX:%[0-9]+]]:_(p0) = G_FRAME_INDEX %stack.0
+    ; CHECK-NEXT: [[LOAD:%[0-9]+]]:_(p0) = G_LOAD [[FRAME_INDEX]](p0) :: (load (p0))
+    ; CHECK-NEXT: [[C:%[0-9]+]]:_(s64) = G_CONSTANT i64 3
+    ; CHECK-NEXT: [[PTR_ADD:%[0-9]+]]:_(p0) = G_PTR_ADD [[LOAD]], [[C]](s64)
+    ; CHECK-NEXT: [[C1:%[0-9]+]]:_(s64) = G_CONSTANT i64 -4
+    ; CHECK-NEXT: [[PTRMASK:%[0-9]+]]:_(p0) = G_PTRMASK [[PTR_ADD]], [[C1]](s64)
+    ; CHECK-NEXT: [[C2:%[0-9]+]]:_(s64) = G_CONSTANT i64 8
+    ; CHECK-NEXT: [[PTR_ADD1:%[0-9]+]]:_(p0) = G_PTR_ADD [[PTRMASK]], [[C2]](s64)
+    ; CHECK-NEXT: G_STORE [[PTR_ADD1]](p0), [[FRAME_INDEX]](p0) :: (store (p0))
+    ; CHECK-NEXT: PseudoRET
+    %0:_(p0) = G_FRAME_INDEX %stack.0
+    %1:_(s64) = G_VAARG %0(p0), 4
+    PseudoRET
+...