[AArch64] Add support for dllimport of values and functions

[mstorsjo]

Summary

Previously, the dllimport attribute did the right thing in terms of treating it as a pointer to a value, but this makes sure the names get mangled properly, and calls to such functions load the function from the __imp_ pointer.

This is based on SVN r212431 and r212430 where the same was implemented for ARM.

[mstorsjo]

Ping

[compnerd]

Seems fine with the adjustments.

lib/Target/AArch64/AArch64ISelLowering.cpp

3714

Can you lower this to ~L3725.

lib/Target/AArch64/AArch64MCInstLower.cpp

41

Early return please:

if (!TheTriple.isOSBinFormatCOFF())
  return Printer.getSymbol(MO.getGlobal());
...

test/CodeGen/AArch64/dllimport.ll

1

I prefer the canonicalized triple aarch64-unknown-windows-msvc

6

Are the arm_aapcs_vfpcc correct here? The variant of AAPCS64 isn't marked differently I thought. The same through out.

[mstorsjo]

lib/Target/AArch64/AArch64ISelLowering.cpp

3714

Ok, moving all of these three new variables down closer to where they're set and used.

lib/Target/AArch64/AArch64MCInstLower.cpp

41

Sure

test/CodeGen/AArch64/dllimport.ll

1

Ok, will change

6

Indeed, this is just a leftover since I based this on the ARM dllimport test - will remove.

lib/Target/AArch64/AArch64ISelLowering.h

private:
   SDValue getTargetNode(GlobalAddressSDNode *N, EVT Ty, SelectionDAG &DAG,
                         unsigned Flag) const;
   SDValue getTargetNode(JumpTableSDNode *N, EVT Ty, SelectionDAG &DAG,
                         unsigned Flag) const;
   SDValue getTargetNode(ConstantPoolSDNode *N, EVT Ty, SelectionDAG &DAG,
                         unsigned Flag) const;
   SDValue getTargetNode(BlockAddressSDNode *N, EVT Ty, SelectionDAG &DAG,
                         unsigned Flag) const;
-  template <class NodeTy> SDValue getGOT(NodeTy *N, SelectionDAG &DAG) const;
   template <class NodeTy>
-  SDValue getAddrLarge(NodeTy *N, SelectionDAG &DAG) const;
-  template <class NodeTy> SDValue getAddr(NodeTy *N, SelectionDAG &DAG) const;
+  SDValue getGOT(NodeTy *N, SelectionDAG &DAG, unsigned Flags = 0) const;
+  template <class NodeTy>
+  SDValue getAddrLarge(NodeTy *N, SelectionDAG &DAG, unsigned Flags = 0) const;
+  template <class NodeTy>
+  SDValue getAddr(NodeTy *N, SelectionDAG &DAG, unsigned Flags = 0) const;
   SDValue LowerGlobalAddress(SDValue Op, SelectionDAG &DAG) const;
   SDValue LowerGlobalTLSAddress(SDValue Op, SelectionDAG &DAG) const;
   SDValue LowerDarwinGlobalTLSAddress(SDValue Op, SelectionDAG &DAG) const;
   SDValue LowerELFGlobalTLSAddress(SDValue Op, SelectionDAG &DAG) const;
   SDValue LowerELFTLSDescCallSeq(SDValue SymAddr, const SDLoc &DL,
                                  SelectionDAG &DAG) const;
   SDValue LowerSETCC(SDValue Op, SelectionDAG &DAG) const;
   SDValue LowerBR_CC(SDValue Op, SelectionDAG &DAG) const;

lib/Target/AArch64/AArch64ISelLowering.cpp

   // direct call is) turn it into a TargetGlobalAddress/TargetExternalSymbol
   // node so that legalize doesn't hack it.
   if (auto *G = dyn_cast<GlobalAddressSDNode>(Callee)) {
     auto GV = G->getGlobal();
     if (Subtarget->classifyGlobalFunctionReference(GV, getTargetMachine()) ==
         AArch64II::MO_GOT) {
       Callee = DAG.getTargetGlobalAddress(GV, DL, PtrVT, 0, AArch64II::MO_GOT);
       Callee = DAG.getNode(AArch64ISD::LOADgot, DL, PtrVT, Callee);
+    } else if (Subtarget->isTargetCOFF() && GV->hasDLLImportStorageClass()) {
+      assert(Subtarget->isTargetWindows() &&
+             "Windows is the only supported COFF target");
+      Callee = getGOT(G, DAG, AArch64II::MO_DLLIMPORT);
     } else {
       const GlobalValue *GV = G->getGlobal();
       Callee = DAG.getTargetGlobalAddress(GV, DL, PtrVT, 0, 0);
     }
   } else if (auto *S = dyn_cast<ExternalSymbolSDNode>(Callee)) {
     if (getTargetMachine().getCodeModel() == CodeModel::Large &&
         Subtarget->isTargetMachO()) {
       const char *Sym = S->getSymbol();

 SDValue AArch64TargetLowering::getTargetNode(BlockAddressSDNode* N, EVT Ty,
                                              SelectionDAG &DAG,
                                              unsigned Flag) const {
   return DAG.getTargetBlockAddress(N->getBlockAddress(), Ty, 0, Flag);
 }
 
 // (loadGOT sym)
 template <class NodeTy>
-SDValue AArch64TargetLowering::getGOT(NodeTy *N, SelectionDAG &DAG) const {
+SDValue AArch64TargetLowering::getGOT(NodeTy *N, SelectionDAG &DAG,
+                                      unsigned Flags) const {
   DEBUG(dbgs() << "AArch64TargetLowering::getGOT\n");
   SDLoc DL(N);
   EVT Ty = getPointerTy(DAG.getDataLayout());
-  SDValue GotAddr = getTargetNode(N, Ty, DAG, AArch64II::MO_GOT);
+  SDValue GotAddr = getTargetNode(N, Ty, DAG, AArch64II::MO_GOT | Flags);
   // FIXME: Once remat is capable of dealing with instructions with register
   // operands, expand this into two nodes instead of using a wrapper node.
   return DAG.getNode(AArch64ISD::LOADgot, DL, Ty, GotAddr);
 }
 
 // (wrapper %highest(sym), %higher(sym), %hi(sym), %lo(sym))
 template <class NodeTy>
-SDValue AArch64TargetLowering::getAddrLarge(NodeTy *N, SelectionDAG &DAG)
-  const {
+SDValue AArch64TargetLowering::getAddrLarge(NodeTy *N, SelectionDAG &DAG,
+                                            unsigned Flags) const {
   DEBUG(dbgs() << "AArch64TargetLowering::getAddrLarge\n");
   SDLoc DL(N);
   EVT Ty = getPointerTy(DAG.getDataLayout());
   const unsigned char MO_NC = AArch64II::MO_NC;
   return DAG.getNode(
-        AArch64ISD::WrapperLarge, DL, Ty,
-        getTargetNode(N, Ty, DAG, AArch64II::MO_G3),
-        getTargetNode(N, Ty, DAG, AArch64II::MO_G2 | MO_NC),
-        getTargetNode(N, Ty, DAG, AArch64II::MO_G1 | MO_NC),
-        getTargetNode(N, Ty, DAG, AArch64II::MO_G0 | MO_NC));
+      getTargetNode(N, Ty, DAG, AArch64II::MO_G3 | Flags),
+      getTargetNode(N, Ty, DAG, AArch64II::MO_G2 | MO_NC | Flags),
+      getTargetNode(N, Ty, DAG, AArch64II::MO_G1 | MO_NC | Flags),
+      getTargetNode(N, Ty, DAG, AArch64II::MO_G0 | MO_NC | Flags));
 }
 
 // (addlow (adrp %hi(sym)) %lo(sym))
 template <class NodeTy>
-SDValue AArch64TargetLowering::getAddr(NodeTy *N, SelectionDAG &DAG) const {
+SDValue AArch64TargetLowering::getAddr(NodeTy *N, SelectionDAG &DAG,
+                                       unsigned Flags) const {
   DEBUG(dbgs() << "AArch64TargetLowering::getAddr\n");
   SDLoc DL(N);
   EVT Ty = getPointerTy(DAG.getDataLayout());
-  SDValue Hi = getTargetNode(N, Ty, DAG, AArch64II::MO_PAGE);
+  SDValue Hi = getTargetNode(N, Ty, DAG, AArch64II::MO_PAGE | Flags);
   SDValue Lo = getTargetNode(N, Ty, DAG,
-                             AArch64II::MO_PAGEOFF | AArch64II::MO_NC);
+                             AArch64II::MO_PAGEOFF | AArch64II::MO_NC | Flags);
   SDValue ADRP = DAG.getNode(AArch64ISD::ADRP, DL, Ty, Hi);
   return DAG.getNode(AArch64ISD::ADDlow, DL, Ty, ADRP, Lo);
 }
 
 SDValue AArch64TargetLowering::LowerGlobalAddress(SDValue Op,
                                                   SelectionDAG &DAG) const {
   GlobalAddressSDNode *GN = cast<GlobalAddressSDNode>(Op);
   const GlobalValue *GV = GN->getGlobal();
+  const AArch64II::TOF TargetFlags =
+      (GV->hasDLLImportStorageClass() ? AArch64II::MO_DLLIMPORT
+                                      : AArch64II::MO_NO_FLAG);
   unsigned char OpFlags =
       Subtarget->ClassifyGlobalReference(GV, getTargetMachine());
+  EVT PtrVT = getPointerTy(DAG.getDataLayout());
+  SDValue Result;
+  SDLoc DL(GN);
 
   assert(cast<GlobalAddressSDNode>(Op)->getOffset() == 0 &&
          "unexpected offset in global node");
 
   // This also catches the large code model case for Darwin.
   if ((OpFlags & AArch64II::MO_GOT) != 0) {
-    return getGOT(GN, DAG);
+    return getGOT(GN, DAG, TargetFlags);
   }
 
   if (getTargetMachine().getCodeModel() == CodeModel::Large) {
-    return getAddrLarge(GN, DAG);
+    Result = getAddrLarge(GN, DAG, TargetFlags);
   } else {
-    return getAddr(GN, DAG);
+    Result = getAddr(GN, DAG, TargetFlags);
   }
+  if (GV->hasDLLImportStorageClass())
+    Result = DAG.getLoad(PtrVT, DL, DAG.getEntryNode(), Result,
+                         MachinePointerInfo::getGOT(DAG.getMachineFunction()));
+  return Result;
 }
 
 /// \brief Convert a TLS address reference into the correct sequence of loads
 /// and calls to compute the variable's address (for Darwin, currently) and
 /// return an SDValue containing the final node.
 
 /// Darwin only has one TLS scheme which must be capable of dealing with the
 /// fully general situation, in the worst case. This means:

lib/Target/AArch64/AArch64MCInstLower.cpp

 
 #include "AArch64MCInstLower.h"
 #include "MCTargetDesc/AArch64MCExpr.h"
 #include "Utils/AArch64BaseInfo.h"
 #include "llvm/CodeGen/AsmPrinter.h"
 #include "llvm/CodeGen/MachineBasicBlock.h"
 #include "llvm/CodeGen/MachineInstr.h"
 #include "llvm/IR/Mangler.h"
+#include "llvm/MC/MCContext.h"
 #include "llvm/MC/MCExpr.h"
 #include "llvm/MC/MCInst.h"
 #include "llvm/Support/CodeGen.h"
 #include "llvm/Support/CommandLine.h"
+#include "llvm/Target/TargetLoweringObjectFile.h"
 #include "llvm/Target/TargetMachine.h"
 using namespace llvm;
 
 extern cl::opt<bool> EnableAArch64ELFLocalDynamicTLSGeneration;
 
 AArch64MCInstLower::AArch64MCInstLower(MCContext &ctx, AsmPrinter &printer)
     : Ctx(ctx), Printer(printer) {}
 
 MCSymbol *
 AArch64MCInstLower::GetGlobalAddressSymbol(const MachineOperand &MO) const {
-  return Printer.getSymbol(MO.getGlobal());
+  const GlobalValue *GV = MO.getGlobal();
+  unsigned TargetFlags = MO.getTargetFlags();
+  const Triple &TheTriple = Printer.TM.getTargetTriple();
+  if (TheTriple.isOSBinFormatCOFF()) {
+    assert(TheTriple.isOSWindows() &&
+           "Windows is the only supported COFF target");
+
+    bool IsIndirect = (TargetFlags & AArch64II::MO_DLLIMPORT);
+    if (!IsIndirect)
+      return Printer.getSymbol(GV);
+
+    SmallString<128> Name;
+    Name = "__imp_";
+    Printer.TM.getNameWithPrefix(Name, GV,
+                                 Printer.getObjFileLowering().getMangler());
+
+    return Ctx.getOrCreateSymbol(Name);
+  } else {
+    return Printer.getSymbol(GV);
+  }
 }
 
 MCSymbol *
 AArch64MCInstLower::GetExternalSymbolSymbol(const MachineOperand &MO) const {
   return Printer.GetExternalSymbolSymbol(MO.getSymbolName());
 }
 
 MCOperand AArch64MCInstLower::lowerSymbolOperandDarwin(const MachineOperand &MO,

lib/Target/AArch64/Utils/AArch64BaseInfo.h

enum TOF {
     /// reference for overflow. For example in an ADRP/ADD pair of relocations
     /// the ADRP usually does check, but not the ADD.
     MO_NC = 0x20,
 
     /// MO_TLS - Indicates that the operand being accessed is some kind of
     /// thread-local symbol. On Darwin, only one type of thread-local access
     /// exists (pre linker-relaxation), but on ELF the TLSModel used for the
     /// referee will affect interpretation.
-    MO_TLS = 0x40
+    MO_TLS = 0x40,
+
+    /// MO_DLLIMPORT - On a symbol operand, this represents that the reference
+    /// to the symbol is for an import stub.  This is used for DLL import
+    /// storage class indication on Windows.
+    MO_DLLIMPORT = 0x80,
   };
 } // end namespace AArch64II
 
 } // end namespace llvm
 
 #endif

test/CodeGen/AArch64/dllimport.ll

+; RUN: llc -mtriple aarch64-windows -filetype asm -o - %s | FileCheck %s
+
+@var = external dllimport global i32
+@ext = external global i32
+declare dllimport arm_aapcs_vfpcc i32 @external()
+declare arm_aapcs_vfpcc i32 @internal()
+
+define arm_aapcs_vfpcc i32 @get_var() {
+  %1 = load i32, i32* @var, align 4
+  ret i32 %1
+}
+
+; CHECK-LABEL: get_var
+; CHECK: adrp x8, __imp_var
+; CHECK: ldr x8, [x8, __imp_var]
+; CHECK: ldr w0, [x8]
+; CHECK: ret
+
+define arm_aapcs_vfpcc i32 @get_ext() {
+  %1 = load i32, i32* @ext, align 4
+  ret i32 %1
+}
+
+; CHECK-LABEL: get_ext
+; CHECK: adrp x8, ext
+; CHECK: ldr w0, [x8, ext]
+; CHECK: ret
+
+define arm_aapcs_vfpcc i32* @get_var_pointer() {
+  ret i32* @var
+}
+
+; CHECK-LABEL: get_var_pointer
+; CHECK: adrp x0, __imp_var
+; CHECK: ldr x0, [x0, __imp_var]
+; CHECK: ret
+
+define arm_aapcs_vfpcc i32 @call_external() {
+  %call = tail call arm_aapcs_vfpcc i32 @external()
+  ret i32 %call
+}
+
+; CHECK-LABEL: call_external
+; CHECK: adrp x8, __imp_external
+; CHECK: ldr x8, [x8, __imp_external]
+; CHECK: blr x8
+; CHECK: ret
+
+define arm_aapcs_vfpcc i32 @call_internal() {
+  %call = tail call arm_aapcs_vfpcc i32 @internal()
+  ret i32 %call
+}
+
+; CHECK-LABEL: call_internal
+; CHECK: bl internal