[RISCV][GlobalISel] Lower calls to variadic functions

[nitinjohnraj]

Calls to variadic functions do not seem to require any special handling.

[llvmbot]

@llvm/pr-subscribers-llvm-globalisel

@llvm/pr-subscribers-backend-risc-v

Changes

Calls to variadic functions do not seem to require any special handling.

---

Full diff: https://github.com/llvm/llvm-project/pull/68271.diff

2 Files Affected:

• (modified) llvm/lib/Target/RISCV/GISel/RISCVCallLowering.cpp (-4)
• (added) llvm/test/CodeGen/RISCV/GlobalISel/call-lowering/variadic-call.ll (+33)

diff --git a/llvm/lib/Target/RISCV/GISel/RISCVCallLowering.cpp b/llvm/lib/Target/RISCV/GISel/RISCVCallLowering.cpp
index 5505f89a32f21be..a362a709329d5df 100644
--- a/llvm/lib/Target/RISCV/GISel/RISCVCallLowering.cpp
+++ b/llvm/lib/Target/RISCV/GISel/RISCVCallLowering.cpp
@@ -252,10 +252,6 @@ bool RISCVCallLowering::lowerCall(MachineIRBuilder &MIRBuilder,
   const Function &F = MF.getFunction();
   CallingConv::ID CC = F.getCallingConv();
 
-  // TODO: Support vararg functions.
-  if (Info.IsVarArg)
-    return false;
-
   // TODO: Support all argument types.
   for (auto &AInfo : Info.OrigArgs) {
     if (AInfo.Ty->isIntegerTy())
diff --git a/llvm/test/CodeGen/RISCV/GlobalISel/call-lowering/variadic-call.ll b/llvm/test/CodeGen/RISCV/GlobalISel/call-lowering/variadic-call.ll
new file mode 100644
index 000000000000000..232e7cf39ac634e
--- /dev/null
+++ b/llvm/test/CodeGen/RISCV/GlobalISel/call-lowering/variadic-call.ll
@@ -0,0 +1,33 @@
+; NOTE: Assertions have been autogenerated by utils/update_mir_test_checks.py
+; RUN: llc -mtriple=riscv64 -global-isel --stop-before=legalizer < %s -o - \
+; RUN:   | FileCheck %s
+
+declare noundef signext i32 @foo(i32 noundef signext, ...)
+
+define dso_local noundef signext i32 @main() {
+  ; CHECK-LABEL: name: main
+  ; CHECK: bb.1.entry:
+  ; CHECK-NEXT:   [[C:%[0-9]+]]:_(s32) = G_CONSTANT i32 0
+  ; CHECK-NEXT:   [[FRAME_INDEX:%[0-9]+]]:_(p0) = G_FRAME_INDEX %stack.0.retval
+  ; CHECK-NEXT:   G_STORE [[C]](s32), [[FRAME_INDEX]](p0) :: (store (s32) into %ir.retval)
+  ; CHECK-NEXT:   [[C1:%[0-9]+]]:_(s64) = G_CONSTANT i64 0
+  ; CHECK-NEXT:   [[C2:%[0-9]+]]:_(s64) = G_CONSTANT i64 1
+  ; CHECK-NEXT:   [[C3:%[0-9]+]]:_(s64) = G_CONSTANT i64 2
+  ; CHECK-NEXT:   [[C4:%[0-9]+]]:_(s64) = G_CONSTANT i64 3
+  ; CHECK-NEXT:   $x10 = COPY [[C1]](s64)
+  ; CHECK-NEXT:   $x11 = COPY [[C2]](s64)
+  ; CHECK-NEXT:   $x12 = COPY [[C3]](s64)
+  ; CHECK-NEXT:   $x13 = COPY [[C4]](s64)
+  ; CHECK-NEXT:   PseudoCALL target-flags(riscv-call) @_Z1fiz, implicit-def $x1, implicit $x10, implicit $x11, implicit $x12, implicit $x13, implicit-def $x10
+  ; CHECK-NEXT:   [[COPY:%[0-9]+]]:_(s64) = COPY $x10
+  ; CHECK-NEXT:   [[ASSERT_SEXT:%[0-9]+]]:_(s64) = G_ASSERT_SEXT [[COPY]], 32
+  ; CHECK-NEXT:   [[TRUNC:%[0-9]+]]:_(s32) = G_TRUNC [[ASSERT_SEXT]](s64)
+  ; CHECK-NEXT:   [[SEXT:%[0-9]+]]:_(s64) = G_SEXT [[TRUNC]](s32)
+  ; CHECK-NEXT:   $x10 = COPY [[SEXT]](s64)
+  ; CHECK-NEXT:   PseudoRET implicit $x10
+entry:
+  %retval = alloca i32, align 4
+  store i32 0, ptr %retval, align 4
+  %call = call noundef signext i32 (i32, ...) @(i32 noundef signext 0, i32 noundef signext 1, i32 noundef signext 2, i32 noundef signext 3)
+  ret i32 %call
+}

[michaelmaitland]

In ISelLowering it looks like we do give some care to varargs.

• If all registers are allocated, then all varargs must be passed on the stack and don't need to save any argregs
• record the frame index of the first variable argument which is a value necessary to VASTART.
• if saving an odd number of registers then create an extra stack slot to ensure that the frame pointer is correctly aligned
• copy the integer register that may have been used for passing varargs to the vararg save area

Why don't we need this in GISel?

[topperc]

In ISelLowering it looks like we do give some care to varargs.

• If all registers are allocated, then all varargs must be passed on the stack and don't need to save any argregs
• record the frame index of the first variable argument which is a value necessary to VASTART.
• if saving an odd number of registers then create an extra stack slot to ensure that the frame pointer is correctly aligned
• copy the integer register that may have been used for passing varargs to the vararg save area

Why don't we need this in GISel?

I believe the code you mention is in RISCVTargetLowering::LowerFormalArguments in SelectionDAG. That is used to codegen the beginning of a variadic function. That will need to be done in RISCVCallLowering::lowerFormalArguments for gisel.

This patch is for calls to variadic functions. The equivalent is RISCVTargetLowering::LowerCall in SelectionDAG.

[topperc]

LGTM