[BPF] add support for 32 bit registers in inline asm

[alessandrod]

Summary

Add "w" constraint type which allows selecting 32 bit registers.

32 bit registers were added in https://reviews.llvm.org/rGca31c3bb3ff149850b664838fbbc7d40ce571879.

[yonghong-song]

Generally looks good. One suggestion and a few nits. Thanks!

clang/lib/Basic/Targets/BPF.cpp

54

-mcpu=v3 also enables alu32. See below llvm code

void BPFSubtarget::initSubtargetFeatures(StringRef CPU, StringRef FS) {
  if (CPU == "probe")
    CPU = sys::detail::getHostCPUNameForBPF();
  if (CPU == "generic" || CPU == "v1")
    return;
  if (CPU == "v2") {
    HasJmpExt = true;
    return;
  }
  if (CPU == "v3") {
    HasJmpExt = true;
    HasJmp32 = true;
    HasAlu32 = true;
    return;
  }
}

Could you add a check with setCPU() to set HasAlu32 = true if it is v3? This way, when people directly use -mcpu=v3 in clang command line, 'w' constraint is also supported.

60

empty line here.

clang/test/CodeGen/bpf-inline-asm.c

7

I assume test_generic_constraints() will pass even without this patch, but it is a good addition for clang side of bpf inline-asm support, so I think it is okay to be included in this patch.

32

empty line at the end of file.

llvm/test/CodeGen/BPF/inlineasm-wreg.ll

2

Let us remove the empty line in the top of the test file.

19

empty line at the end of file.

[alessandrod]

Enable "w" constraint when -mcpu=v3 and fix whitespace.

[yonghong-song]

LGTM except one minor issue. Please do address the nit before merging. Thanks!

clang/test/CodeGen/bpf-inline-asm.c

6

The var is not used and can be removed.

[alessandrod]

Remove unused global from test

[alessandrod]

In D102118#2761476, @yonghong-song wrote:

LGTM except one minor issue. Please do address the nit before merging. Thanks!

Done, thanks for the review! I don't have commit access so someone will have to merge this for me :)

[yonghong-song]

Sure. I will merge the patch for you.

clang/lib/Basic/Targets/BPF.h

 #include "llvm/ADT/Triple.h"
 #include "llvm/Support/Compiler.h"
 
 namespace clang {
 namespace targets {
 
 class LLVM_LIBRARY_VISIBILITY BPFTargetInfo : public TargetInfo {
   static const Builtin::Info BuiltinInfo[];
+  bool HasAlu32 = false;
 
 public:
   BPFTargetInfo(const llvm::Triple &Triple, const TargetOptions &)
       : TargetInfo(Triple) {
     LongWidth = LongAlign = PointerWidth = PointerAlign = 64;
     SizeType = UnsignedLong;
     PtrDiffType = SignedLong;
     IntPtrType = SignedLong;
public:
   bool hasFeature(StringRef Feature) const override {
     return Feature == "bpf" || Feature == "alu32" || Feature == "dwarfris";
   }
 
   void setFeatureEnabled(llvm::StringMap<bool> &Features, StringRef Name,
                          bool Enabled) const override {
     Features[Name] = Enabled;
   }
+  bool handleTargetFeatures(std::vector<std::string> &Features,
+                            DiagnosticsEngine &Diags) override;
 
   ArrayRef<Builtin::Info> getTargetBuiltins() const override;
 
   const char *getClobbers() const override { return ""; }
 
   BuiltinVaListKind getBuiltinVaListKind() const override {
     return TargetInfo::VoidPtrBuiltinVaList;
   }
 
   bool isValidGCCRegisterName(StringRef Name) const override { return true; }
   ArrayRef<const char *> getGCCRegNames() const override { return None; }
 
   bool validateAsmConstraint(const char *&Name,
-                             TargetInfo::ConstraintInfo &info) const override {
+                             TargetInfo::ConstraintInfo &Info) const override {
+    switch (*Name) {
+    default:
+      break;
+    case 'w':
+      if (HasAlu32) {
+        Info.setAllowsRegister();
+      }
+      break;
+    }
     return true;
   }
 
   ArrayRef<TargetInfo::GCCRegAlias> getGCCRegAliases() const override {
     return None;
   }
 
   bool allowDebugInfoForExternalRef() const override { return true; }
 
   CallingConvCheckResult checkCallingConvention(CallingConv CC) const override {
     switch (CC) {
     default:
       return CCCR_Warning;
     case CC_C:
     case CC_OpenCLKernel:
       return CCCR_OK;
     }
   }
 
   bool isValidCPUName(StringRef Name) const override;
 
   void fillValidCPUList(SmallVectorImpl<StringRef> &Values) const override;
 
   bool setCPU(const std::string &Name) override {
+    if (Name == "v3") {
+      HasAlu32 = true;
+    }
+
     StringRef CPUName(Name);
     return isValidCPUName(CPUName);
   }
 };
 } // namespace targets
 } // namespace clang
 #endif // LLVM_CLANG_LIB_BASIC_TARGETS_BPF_H

clang/lib/Basic/Targets/BPF.cpp

 void BPFTargetInfo::fillValidCPUList(SmallVectorImpl<StringRef> &Values) const {
   Values.append(std::begin(ValidCPUNames), std::end(ValidCPUNames));
 }
 
 ArrayRef<Builtin::Info> BPFTargetInfo::getTargetBuiltins() const {
   return llvm::makeArrayRef(BuiltinInfo, clang::BPF::LastTSBuiltin -
                                              Builtin::FirstTSBuiltin);
 }
+
+bool BPFTargetInfo::handleTargetFeatures(std::vector<std::string> &Features,
+                                         DiagnosticsEngine &Diags) {
+  for (const auto &Feature : Features) {
+    if (Feature == "+alu32") {
+      HasAlu32 = true;
+    }
+  }
+
+  return true;
+}

clang/test/CodeGen/bpf-inline-asm.c

+// REQUIRES: bpf-registered-target
+// RUN: %clang -target bpf -emit-llvm -S -Xclang -target-feature -Xclang +alu32 %s -o - | FileCheck %s
+// RUN: %clang -target bpf -emit-llvm -S -mcpu=v3 %s -o - | FileCheck %s
+
+void test_generic_constraints(int var32, long var64) {
+  asm("%0 = %1"
+      : "=r"(var32)
+      : "0"(var32));
+  // CHECK: [[R32_ARG:%[a-zA-Z0-9]+]] = load i32, i32*
+  // CHECK: call i32 asm "$0 = $1", "=r,0"(i32 [[R32_ARG]])
+
+  asm("%0 = %1"
+      : "=r"(var64)
+      : "0"(var64));
+  // CHECK: [[R64_ARG:%[a-zA-Z0-9]+]] = load i64, i64*
+  // CHECK: call i64 asm "$0 = $1", "=r,0"(i64 [[R64_ARG]])
+
+  asm("%0 = %1"
+      : "=r"(var64)
+      : "r"(var64));
+  // CHECK: [[R64_ARG:%[a-zA-Z0-9]+]] = load i64, i64*
+  // CHECK: call i64 asm "$0 = $1", "=r,r"(i64 [[R64_ARG]])
+}
+
+void test_constraint_w(int a) {
+  asm("%0 = %1"
+      : "=w"(a)
+      : "w"(a));
+  // CHECK: [[R32_ARG:%[a-zA-Z0-9]+]] = load i32, i32*
+  // CHECK: call i32 asm "$0 = $1", "=w,w"(i32 [[R32_ARG]])
+}

llvm/lib/Target/BPF/BPFISelLowering.h

public:
 
   // This method returns the name of a target specific DAG node.
   const char *getTargetNodeName(unsigned Opcode) const override;
 
   // This method decides whether folding a constant offset
   // with the given GlobalAddress is legal.
   bool isOffsetFoldingLegal(const GlobalAddressSDNode *GA) const override;
 
+  BPFTargetLowering::ConstraintType
+  getConstraintType(StringRef Constraint) const override;
+
   std::pair<unsigned, const TargetRegisterClass *>
   getRegForInlineAsmConstraint(const TargetRegisterInfo *TRI,
                                StringRef Constraint, MVT VT) const override;
 
   MachineBasicBlock *
   EmitInstrWithCustomInserter(MachineInstr &MI,
                               MachineBasicBlock *BB) const override;
 

llvm/lib/Target/BPF/BPFISelLowering.cpp

 bool BPFTargetLowering::isZExtFree(EVT VT1, EVT VT2) const {
   if (!getHasAlu32() || !VT1.isInteger() || !VT2.isInteger())
     return false;
   unsigned NumBits1 = VT1.getSizeInBits();
   unsigned NumBits2 = VT2.getSizeInBits();
   return NumBits1 == 32 && NumBits2 == 64;
 }
 
+BPFTargetLowering::ConstraintType
+BPFTargetLowering::getConstraintType(StringRef Constraint) const {
+  if (Constraint.size() == 1) {
+    switch (Constraint[0]) {
+    default:
+      break;
+    case 'w':
+      return C_RegisterClass;
+    }
+  }
+
+  return TargetLowering::getConstraintType(Constraint);
+}
+
 std::pair<unsigned, const TargetRegisterClass *>
 BPFTargetLowering::getRegForInlineAsmConstraint(const TargetRegisterInfo *TRI,
                                                 StringRef Constraint,
                                                 MVT VT) const {
   if (Constraint.size() == 1)
     // GCC Constraint Letters
     switch (Constraint[0]) {
     case 'r': // GENERAL_REGS
       return std::make_pair(0U, &BPF::GPRRegClass);
+    case 'w':
+      if (HasAlu32)
+        return std::make_pair(0U, &BPF::GPR32RegClass);
+      break;
     default:
       break;
     }
 
   return TargetLowering::getRegForInlineAsmConstraint(TRI, Constraint, VT);
 }
 
 void BPFTargetLowering::ReplaceNodeResults(

llvm/test/CodeGen/BPF/inlineasm-wreg.ll

+; RUN: llc < %s -march=bpfel -mattr=+alu32 -verify-machineinstrs | FileCheck %s
+; RUN: llc < %s -march=bpfeb -mattr=+alu32 -verify-machineinstrs | FileCheck %s
+
+; Test that %w works as input constraint
+; CHECK-LABEL: test_inlineasm_w_input_constraint
+define dso_local i32 @test_inlineasm_w_input_constraint() {
+  tail call void asm sideeffect "w0 = $0", "w"(i32 42)
+; CHECK: w0 = w1
+  ret i32 42
+}
+
+; Test that %w works as output constraint
+; CHECK-LABEL: test_inlineasm_w_output_constraint
+define dso_local i32 @test_inlineasm_w_output_constraint() {
+  %1 = tail call i32 asm sideeffect "$0 = $1", "=w,i"(i32 42)
+; CHECK: w0 = 42
+  ret i32 %1
+}