[llvm] AArch64: Use reportFatalUsageError for unsupported calling conv (PR #144840)

[via llvm-commits]

llvmbot wrote:


<!--LLVM PR SUMMARY COMMENT-->

@llvm/pr-subscribers-backend-aarch64

Author: Matt Arsenault (arsenm)

<details>
<summary>Changes</summary>

This probably should emit a DiagnosticInfoUnsupported and use the
default calling convention instead, but then we would need to pass
in the context.

Also move where CCAssignFnForCall is called. It was unnecessarily
called for each argument, so the error wouldn't trigger for functions
with 0 arguments.

This only ensures the error occurs for functions defined with the
calling convention. The error is still missed for outgoing calls
with no arguments. The lowering logic here is convoluted, calling
CCAssignFnForCall for each argument and it does not mirror
LowerFormalArguments so I'm not sure what's going on here.

---
Full diff: https://github.com/llvm/llvm-project/pull/144840.diff


3 Files Affected:

- (modified) llvm/lib/Target/AArch64/AArch64ISelLowering.cpp (+9-5) 
- (added) llvm/test/CodeGen/AArch64/unsupported-cc-call.ll (+8) 
- (added) llvm/test/CodeGen/AArch64/unsupported-cc-func.ll (+6) 


``````````diff
diff --git a/llvm/lib/Target/AArch64/AArch64ISelLowering.cpp b/llvm/lib/Target/AArch64/AArch64ISelLowering.cpp
index 581f152776026..69f8c6c21a556 100644
--- a/llvm/lib/Target/AArch64/AArch64ISelLowering.cpp
+++ b/llvm/lib/Target/AArch64/AArch64ISelLowering.cpp
@@ -7632,7 +7632,7 @@ CCAssignFn *AArch64TargetLowering::CCAssignFnForCall(CallingConv::ID CC,
                                                      bool IsVarArg) const {
   switch (CC) {
   default:
-    report_fatal_error("Unsupported calling convention.");
+    reportFatalUsageError("unsupported calling convention");
   case CallingConv::GHC:
     return CC_AArch64_GHC;
   case CallingConv::PreserveNone:
@@ -7741,6 +7741,12 @@ SDValue AArch64TargetLowering::LowerFormalArguments(
   unsigned NumArgs = Ins.size();
   Function::const_arg_iterator CurOrigArg = F.arg_begin();
   unsigned CurArgIdx = 0;
+  bool UseVarArgCC = false;
+  if (IsWin64)
+    UseVarArgCC = isVarArg;
+
+  CCAssignFn *AssignFn = CCAssignFnForCall(CallConv, UseVarArgCC);
+
   for (unsigned i = 0; i != NumArgs; ++i) {
     MVT ValVT = Ins[i].VT;
     if (Ins[i].isOrigArg()) {
@@ -7757,10 +7763,6 @@ SDValue AArch64TargetLowering::LowerFormalArguments(
       else if (ActualMVT == MVT::i16)
         ValVT = MVT::i16;
     }
-    bool UseVarArgCC = false;
-    if (IsWin64)
-      UseVarArgCC = isVarArg;
-    CCAssignFn *AssignFn = CCAssignFnForCall(CallConv, UseVarArgCC);
     bool Res =
         AssignFn(i, ValVT, ValVT, CCValAssign::Full, Ins[i].Flags, CCInfo);
     assert(!Res && "Call operand has unhandled type");
@@ -8429,6 +8431,8 @@ static void analyzeCallOperands(const AArch64TargetLowering &TLI,
         ArgVT = MVT::i16;
     }
 
+    // FIXME: CCAssignFnForCall should be called once, for the call and not per
+    // argument. This logic should exactly mirror LowerFormalArguments.
     CCAssignFn *AssignFn = TLI.CCAssignFnForCall(CalleeCC, UseVarArgCC);
     bool Res = AssignFn(i, ArgVT, ArgVT, CCValAssign::Full, ArgFlags, CCInfo);
     assert(!Res && "Call operand has unhandled type");
diff --git a/llvm/test/CodeGen/AArch64/unsupported-cc-call.ll b/llvm/test/CodeGen/AArch64/unsupported-cc-call.ll
new file mode 100644
index 0000000000000..3e03ac81baf5a
--- /dev/null
+++ b/llvm/test/CodeGen/AArch64/unsupported-cc-call.ll
@@ -0,0 +1,8 @@
+; FIXME: This should error:
+; RUN: llc -mtriple=aarch64-- -filetype=null %s
+declare amdgpu_gfx void @amdgpu_gfx_func()
+
+define void @call_amdgpu_gfx_func() {
+  call amdgpu_gfx void @amdgpu_gfx_func()
+  ret void
+}
diff --git a/llvm/test/CodeGen/AArch64/unsupported-cc-func.ll b/llvm/test/CodeGen/AArch64/unsupported-cc-func.ll
new file mode 100644
index 0000000000000..2d0fb12af0d4a
--- /dev/null
+++ b/llvm/test/CodeGen/AArch64/unsupported-cc-func.ll
@@ -0,0 +1,6 @@
+; RUN: not llc -mtriple=aarch64-- -filetype=null %s 2>&1 | FileCheck %s
+
+; CHECK: LLVM ERROR: unsupported calling convention
+define amdgpu_gfx void @amdgpu_gfx_func_definition() {
+  ret void
+}

``````````

</details>


https://github.com/llvm/llvm-project/pull/144840