[AArch64] Add support for the GNU ILP32 ABI

[Amanieu]

Summary

Add the aarch64[_be]-*-gnu_ilp32 targets to support the GNU ILP32 ABI for AArch64.

The needed codegen changes were mostly already implemented in D61259, which added support for the watchOS ILP32 ABI. The main changes are:

• Wiring up the new target to enable ILP32 codegen and MC.
• ILP32 va_list support.
• ILP32 TLSDESC relocation support.

There was existing MC support for ELF ILP32 relocations from D25159 which could be enabled by passing "-target-abi ilp32" to llvm-mc. This was changed to check for "gnu_ilp32" in the target triple instead. This shouldn't cause any issues since the existing support was slightly broken: it was generating ELF64 objects instead of the ELF32 object files expected by the GNU ILP32 toolchain.

This target has been tested by running the full rustc testsuite on a big-endian ILP32 system based on the GCC ILP32 toolchain.

[joelkevinjones]

I'm happy to see this change. It has been on my TO DO list for a very long time. I will review it soon.

[Amanieu]

clang-format & clang-tidy fixes

[joelkevinjones]

In D16213 I introduced the MCTargetOptions argument to the MCAsmBackend constructor, as well as in constructor wrapper functions. This change touched multiple (every, probably) backend. Given the extension to embed the ABI in the Triple, can the MCTargetOptions argument be removed? That shouldn't be in this commit, I think.

I have't checked the new test cases for completeness. I also haven't checked whether there are existing test cases not in this commit that might usefully have "-target-abi=ilp32" changed to "-triple aarch64-none-linux-gnu_ilp32"

llvm/lib/Target/AArch64/AArch64ISelLowering.cpp

6952–6953

For all the following comments of the form:
// ... at offset n
perhaps change it to:
// ... at offset m * pointer size
so, for example, the next comment would be
// void *__gr_top at offset 1 * pointer size

llvm/lib/Target/AArch64/AArch64TargetMachine.cpp

229

I may be over using "?:", but I think using Explaining Variables (AKA Extract Variable) makes the code clearer.

std::string Endian(LittleEndian ? "e" : "E");
std::string Ptr32(TT.getEnvironment == Triple::GNUILP32 ? "-p:32:32" : "");
return Endian + "-m:e" + Ptr32 + "-i8:8:32-i16:16:32-i64:64-i128:128-n32:64-S128";

llvm/test/CodeGen/AArch64/ilp32-tlsdesc.ll

23

Should the CHECK-RELOC lines contain more complete checks like arm32-elf-relocs.s, for example, does? That is symbol name and offset.

[Amanieu]

In D94143#2493974, @joelkevinjones wrote:

In D16213 I introduced the MCTargetOptions argument to the MCAsmBackend constructor, as well as in constructor wrapper functions. This change touched multiple (every, probably) backend. Given the extension to embed the ABI in the Triple, can the MCTargetOptions argument be removed? That shouldn't be in this commit, I think.

That sounds reasonable. After this is accepted I will prepare another diff to revert D16213.

I have't checked the new test cases for completeness. I also haven't checked whether there are existing test cases not in this commit that might usefully have "-target-abi=ilp32" changed to "-triple aarch64-none-linux-gnu_ilp32"

I've converted every existing test case using -target-abi=ilp32 to use the new target triple.

llvm/test/CodeGen/AArch64/ilp32-tlsdesc.ll

23

I mostly copied this test from arm64-tls-dynamic.ll which also only checks the relocation name.

[joelkevinjones]

LGTM. I think someone else has to approve it.

llvm/lib/Support/Triple.cpp

244

(nit) I'd ignore the previous lint warning and combine lines here.

llvm/lib/Target/AArch64/AArch64ISelLowering.cpp

6952–6953

Nice. I like the separate accounting better than what I suggested.

[kristof.beyls]

My understanding is that AArch64 gnu_ilp32 support is present in mainline gcc, but not in mainline glibc nor in the mainline linux kernel. My understanding is also that it's unlikely that AArch64 gnu_ilp32 support will be added to mainline glibc or linux kernel anytime soon.
That makes me wonder how useful it is to support AArch64 gnu_ilp32 in mainline LLVM. It seems unlikely that it can be used for linux targets anytime soon. Are there non-linux targets using AArch64 gnu_ilp32?

[joelkevinjones]

There isn't any support for ILP32 in the mainline Linux kernel, and there isn't likely to be. However, mainline glibc has had ILP32 support. I don't know the current status.

[Amanieu]

Undo some unnecessary formatting changes.

[Amanieu]

In D94143#2500843, @kristof.beyls wrote:

My understanding is that AArch64 gnu_ilp32 support is present in mainline gcc, but not in mainline glibc nor in the mainline linux kernel. My understanding is also that it's unlikely that AArch64 gnu_ilp32 support will be added to mainline glibc or linux kernel anytime soon.
That makes me wonder how useful it is to support AArch64 gnu_ilp32 in mainline LLVM. It seems unlikely that it can be used for linux targets anytime soon. Are there non-linux targets using AArch64 gnu_ilp32?

ILP32 is used internally in Huawei for some network devices, which run a Linux-based RTOS. Until now they have been using a GCC-based toolchain but we would like to use Rust on the platform, which requires LLVM support.

While kernel and glibc support for ILP32 are unlikely to be supported upstream this is not a concern for LLVM: the ILP32 ABI has a published spec and is supported by mainline GCC.

[Amanieu]

Ping

[• financeaiaiai]

Hi experts:
I am working on compiler-rt to get asan runtime libs for aarch64 ilp32 and I get blocked with https://stackoverflow.com/questions/65806679/llvm-ompiler-rt-sanitizer-not-support-aarch64-ilp32-asan-interceptors-vfork-s-i
I can't tell you how much I appreciate for your reply.

llvm/include/llvm/ADT/Triple.h

enum EnvironmentType {
     UnknownEnvironment,
 
     GNU,
     GNUABIN32,
     GNUABI64,
     GNUEABI,
     GNUEABIHF,
     GNUX32,
+    GNUILP32,
     CODE16,
     EABI,
     EABIHF,
     Android,
     Musl,
     MuslEABI,
     MuslEABIHF,
 
     MSVC,
     Itanium,
     Cygnus,
     CoreCLR,
     Simulator, // Simulator variants of other systems, e.g., Apple's iOS
-    MacABI, // Mac Catalyst variant of Apple's iOS deployment target.
+    MacABI,    // Mac Catalyst variant of Apple's iOS deployment target.
     LastEnvironmentType = MacABI
   };
   enum ObjectFormatType {
     UnknownObjectFormat,
 
     COFF,
     ELF,
     GOFF,
public:
   }
 
   /// Tests whether the target is AArch64 and pointers are the size specified by
   /// \p PointerWidth.
   bool isAArch64(int PointerWidth) const {
     assert(PointerWidth == 64 || PointerWidth == 32);
     if (!isAArch64())
       return false;
-    return isArch64Bit() ? PointerWidth == 64 : PointerWidth == 32;
+    return getArch() == Triple::aarch64_32 ||
+                   getEnvironment() == Triple::GNUILP32
+               ? PointerWidth == 32
+               : PointerWidth == 64;
   }
 
   /// Tests whether the target is MIPS 32-bit (little and big endian).
   bool isMIPS32() const {
     return getArch() == Triple::mips || getArch() == Triple::mipsel;
   }
 
   /// Tests whether the target is MIPS 64-bit (little and big endian).

llvm/lib/CodeGen/TargetLoweringObjectFileImpl.cpp

case Triple::hexagon:
     break;
   case Triple::aarch64:
   case Triple::aarch64_be:
   case Triple::aarch64_32:
     // The small model guarantees static code/data size < 4GB, but not where it
     // will be in memory. Most of these could end up >2GB away so even a signed
     // pc-relative 32-bit address is insufficient, theoretically.
     if (isPositionIndependent()) {
+      // ILP32 uses sdata4 instead of sdata8
+      if (TgtM.getTargetTriple().getEnvironment() == Triple::GNUILP32) {
+        PersonalityEncoding = dwarf::DW_EH_PE_indirect | dwarf::DW_EH_PE_pcrel |
+                              dwarf::DW_EH_PE_sdata4;
+        LSDAEncoding = dwarf::DW_EH_PE_pcrel | dwarf::DW_EH_PE_sdata4;
+        TTypeEncoding = dwarf::DW_EH_PE_indirect | dwarf::DW_EH_PE_pcrel |
+                        dwarf::DW_EH_PE_sdata4;
+      } else {
         PersonalityEncoding = dwarf::DW_EH_PE_indirect | dwarf::DW_EH_PE_pcrel |
                               dwarf::DW_EH_PE_sdata8;
         LSDAEncoding = dwarf::DW_EH_PE_pcrel | dwarf::DW_EH_PE_sdata8;
         TTypeEncoding = dwarf::DW_EH_PE_indirect | dwarf::DW_EH_PE_pcrel |
                         dwarf::DW_EH_PE_sdata8;
+      }
     } else {
       PersonalityEncoding = dwarf::DW_EH_PE_absptr;
       LSDAEncoding = dwarf::DW_EH_PE_absptr;
       TTypeEncoding = dwarf::DW_EH_PE_absptr;
     }
     break;
   case Triple::lanai:
     LSDAEncoding = dwarf::DW_EH_PE_absptr;

llvm/lib/Support/Triple.cpp

StringRef Triple::getEnvironmentTypeName(EnvironmentType Kind) {
   case EABI: return "eabi";
   case EABIHF: return "eabihf";
   case GNU: return "gnu";
   case GNUABI64: return "gnuabi64";
   case GNUABIN32: return "gnuabin32";
   case GNUEABI: return "gnueabi";
   case GNUEABIHF: return "gnueabihf";
   case GNUX32: return "gnux32";
+  case GNUILP32:
+    return "gnu_ilp32";
   case Itanium: return "itanium";
   case MSVC: return "msvc";
   case MacABI: return "macabi";
   case Musl: return "musl";
   case MuslEABI: return "musleabi";
   case MuslEABIHF: return "musleabihf";
   case Simulator: return "simulator";
   }
return StringSwitch<Triple::OSType>(OSName)
     .StartsWith("hurd", Triple::Hurd)
     .StartsWith("wasi", Triple::WASI)
     .StartsWith("emscripten", Triple::Emscripten)
     .Default(Triple::UnknownOS);
 }
 
 static Triple::EnvironmentType parseEnvironment(StringRef EnvironmentName) {
   return StringSwitch<Triple::EnvironmentType>(EnvironmentName)
       .StartsWith("eabihf", Triple::EABIHF)
       .StartsWith("eabi", Triple::EABI)
       .StartsWith("gnuabin32", Triple::GNUABIN32)
       .StartsWith("gnuabi64", Triple::GNUABI64)
       .StartsWith("gnueabihf", Triple::GNUEABIHF)
       .StartsWith("gnueabi", Triple::GNUEABI)
       .StartsWith("gnux32", Triple::GNUX32)
+      .StartsWith("gnu_ilp32", Triple::GNUILP32)
       .StartsWith("code16", Triple::CODE16)
       .StartsWith("gnu", Triple::GNU)
       .StartsWith("android", Triple::Android)
       .StartsWith("musleabihf", Triple::MuslEABIHF)
       .StartsWith("musleabi", Triple::MuslEABI)
       .StartsWith("musl", Triple::Musl)
       .StartsWith("msvc", Triple::MSVC)
       .StartsWith("itanium", Triple::Itanium)
       .StartsWith("cygnus", Triple::Cygnus)
       .StartsWith("coreclr", Triple::CoreCLR)
       .StartsWith("simulator", Triple::Simulator)
       .StartsWith("macabi", Triple::MacABI)
       .Default(Triple::UnknownEnvironment);
 }
 
 static Triple::ObjectFormatType parseFormat(StringRef EnvironmentName) {
   return StringSwitch<Triple::ObjectFormatType>(EnvironmentName)
     // "xcoff" must come before "coff" because of the order-dependendent
     // pattern matching.
     .EndsWith("xcoff", Triple::XCOFF)
     .EndsWith("coff", Triple::COFF)

llvm/lib/Target/AArch64/AArch64AsmPrinter.cpp

case AArch64::TLSDESC_CALLSEQ: {
 
     MCInst Adrp;
     Adrp.setOpcode(AArch64::ADRP);
     Adrp.addOperand(MCOperand::createReg(AArch64::X0));
     Adrp.addOperand(SymTLSDesc);
     EmitToStreamer(*OutStreamer, Adrp);
 
     MCInst Ldr;
+    if (STI->isTargetILP32()) {
+      Ldr.setOpcode(AArch64::LDRWui);
+      Ldr.addOperand(MCOperand::createReg(AArch64::W1));
+    } else {
       Ldr.setOpcode(AArch64::LDRXui);
       Ldr.addOperand(MCOperand::createReg(AArch64::X1));
+    }
     Ldr.addOperand(MCOperand::createReg(AArch64::X0));
     Ldr.addOperand(SymTLSDescLo12);
     Ldr.addOperand(MCOperand::createImm(0));
     EmitToStreamer(*OutStreamer, Ldr);
 
     MCInst Add;
+    if (STI->isTargetILP32()) {
+      Add.setOpcode(AArch64::ADDWri);
+      Add.addOperand(MCOperand::createReg(AArch64::W0));
+      Add.addOperand(MCOperand::createReg(AArch64::W0));
+    } else {
       Add.setOpcode(AArch64::ADDXri);
       Add.addOperand(MCOperand::createReg(AArch64::X0));
       Add.addOperand(MCOperand::createReg(AArch64::X0));
+    }
     Add.addOperand(SymTLSDescLo12);
     Add.addOperand(MCOperand::createImm(AArch64_AM::getShiftValue(0)));
     EmitToStreamer(*OutStreamer, Add);
 
     // Emit a relocation-annotation. This expands to no code, but requests
     // the following instruction gets an R_AARCH64_TLSDESC_CALL.
     MCInst TLSDescCall;
     TLSDescCall.setOpcode(AArch64::TLSDESCCALL);

llvm/lib/Target/AArch64/AArch64ISelLowering.cpp

SDValue FR = DAG.getFrameIndex(FuncInfo->getVarArgsGPRSize() > 0
                                      : FuncInfo->getVarArgsStackIndex(),
                                  getPointerTy(DAG.getDataLayout()));
   const Value *SV = cast<SrcValueSDNode>(Op.getOperand(2))->getValue();
   return DAG.getStore(Op.getOperand(0), DL, FR, Op.getOperand(1),
                       MachinePointerInfo(SV));
 }
 
 SDValue AArch64TargetLowering::LowerAAPCS_VASTART(SDValue Op,
                                                   SelectionDAG &DAG) const {
   // The layout of the va_list struct is specified in the AArch64 Procedure Call
   // Standard, section B.3.
   MachineFunction &MF = DAG.getMachineFunction();
   AArch64FunctionInfo *FuncInfo = MF.getInfo<AArch64FunctionInfo>();
+  unsigned PtrSize = Subtarget->isTargetILP32() ? 4 : 8;
+  auto PtrMemVT = getPointerMemTy(DAG.getDataLayout());
   auto PtrVT = getPointerTy(DAG.getDataLayout());
   SDLoc DL(Op);
 
   SDValue Chain = Op.getOperand(0);
   SDValue VAList = Op.getOperand(1);
   const Value *SV = cast<SrcValueSDNode>(Op.getOperand(2))->getValue();
   SmallVector<SDValue, 4> MemOps;
 
   // void *__stack at offset 0
+  unsigned Offset = 0;
   SDValue Stack = DAG.getFrameIndex(FuncInfo->getVarArgsStackIndex(), PtrVT);
-  MemOps.push_back(
-      DAG.getStore(Chain, DL, Stack, VAList, MachinePointerInfo(SV), Align(8)));
+  Stack = DAG.getZExtOrTrunc(Stack, DL, PtrMemVT);
+  MemOps.push_back(DAG.getStore(Chain, DL, Stack, VAList,
+                                MachinePointerInfo(SV), Align(PtrSize)));
 
-  // void *__gr_top at offset 8
+  // void *__gr_top at offset 8 (4 on ILP32)
+  Offset += PtrSize;
   int GPRSize = FuncInfo->getVarArgsGPRSize();
   if (GPRSize > 0) {
     SDValue GRTop, GRTopAddr;
 
-    GRTopAddr =
-        DAG.getNode(ISD::ADD, DL, PtrVT, VAList, DAG.getConstant(8, DL, PtrVT));
+    GRTopAddr = DAG.getNode(ISD::ADD, DL, PtrVT, VAList,
+                            DAG.getConstant(Offset, DL, PtrVT));
 
     GRTop = DAG.getFrameIndex(FuncInfo->getVarArgsGPRIndex(), PtrVT);
     GRTop = DAG.getNode(ISD::ADD, DL, PtrVT, GRTop,
                         DAG.getConstant(GPRSize, DL, PtrVT));
+    GRTop = DAG.getZExtOrTrunc(GRTop, DL, PtrMemVT);
 
     MemOps.push_back(DAG.getStore(Chain, DL, GRTop, GRTopAddr,
-                                  MachinePointerInfo(SV, 8), Align(8)));
+                                  MachinePointerInfo(SV, Offset),
+                                  Align(PtrSize)));
   }
 
-  // void *__vr_top at offset 16
+  // void *__vr_top at offset 16 (8 on ILP32)
+  Offset += PtrSize;
   int FPRSize = FuncInfo->getVarArgsFPRSize();
   if (FPRSize > 0) {
     SDValue VRTop, VRTopAddr;
     VRTopAddr = DAG.getNode(ISD::ADD, DL, PtrVT, VAList,
-                            DAG.getConstant(16, DL, PtrVT));
+                            DAG.getConstant(Offset, DL, PtrVT));
 
     VRTop = DAG.getFrameIndex(FuncInfo->getVarArgsFPRIndex(), PtrVT);
     VRTop = DAG.getNode(ISD::ADD, DL, PtrVT, VRTop,
                         DAG.getConstant(FPRSize, DL, PtrVT));
+    VRTop = DAG.getZExtOrTrunc(VRTop, DL, PtrMemVT);
 
     MemOps.push_back(DAG.getStore(Chain, DL, VRTop, VRTopAddr,
-                                  MachinePointerInfo(SV, 16), Align(8)));
+                                  MachinePointerInfo(SV, Offset),
+                                  Align(PtrSize)));
   }
 
-  // int __gr_offs at offset 24
-  SDValue GROffsAddr =
-      DAG.getNode(ISD::ADD, DL, PtrVT, VAList, DAG.getConstant(24, DL, PtrVT));
+  // int __gr_offs at offset 24 (12 on ILP32)
+  Offset += PtrSize;
+  SDValue GROffsAddr = DAG.getNode(ISD::ADD, DL, PtrVT, VAList,
+                                   DAG.getConstant(Offset, DL, PtrVT));
   MemOps.push_back(
       DAG.getStore(Chain, DL, DAG.getConstant(-GPRSize, DL, MVT::i32),
-                   GROffsAddr, MachinePointerInfo(SV, 24), Align(4)));
+                   GROffsAddr, MachinePointerInfo(SV, Offset), Align(4)));
 
-  // int __vr_offs at offset 28
-  SDValue VROffsAddr =
-      DAG.getNode(ISD::ADD, DL, PtrVT, VAList, DAG.getConstant(28, DL, PtrVT));
+  // int __vr_offs at offset 28 (16 on ILP32)
+  Offset += 4;
+  SDValue VROffsAddr = DAG.getNode(ISD::ADD, DL, PtrVT, VAList,
+                                   DAG.getConstant(Offset, DL, PtrVT));
   MemOps.push_back(
       DAG.getStore(Chain, DL, DAG.getConstant(-FPRSize, DL, MVT::i32),
-                   VROffsAddr, MachinePointerInfo(SV, 28), Align(4)));
+                   VROffsAddr, MachinePointerInfo(SV, Offset), Align(4)));
 
   return DAG.getNode(ISD::TokenFactor, DL, MVT::Other, MemOps);
 }
 
 SDValue AArch64TargetLowering::LowerVASTART(SDValue Op,
                                             SelectionDAG &DAG) const {
   MachineFunction &MF = DAG.getMachineFunction();
 
   if (Subtarget->isCallingConvWin64(MF.getFunction().getCallingConv()))
     return LowerWin64_VASTART(Op, DAG);
   else if (Subtarget->isTargetDarwin())
     return LowerDarwin_VASTART(Op, DAG);
   else
     return LowerAAPCS_VASTART(Op, DAG);
 }
 
 SDValue AArch64TargetLowering::LowerVACOPY(SDValue Op,
                                            SelectionDAG &DAG) const {
   // AAPCS has three pointers and two ints (= 32 bytes), Darwin has single
   // pointer.
   SDLoc DL(Op);
   unsigned PtrSize = Subtarget->isTargetILP32() ? 4 : 8;
-  unsigned VaListSize = (Subtarget->isTargetDarwin() ||
-                         Subtarget->isTargetWindows()) ? PtrSize : 32;
+  unsigned VaListSize =
+      (Subtarget->isTargetDarwin() || Subtarget->isTargetWindows())
+          ? PtrSize
+          : Subtarget->isTargetILP32() ? 20 : 32;
   const Value *DestSV = cast<SrcValueSDNode>(Op.getOperand(3))->getValue();
   const Value *SrcSV = cast<SrcValueSDNode>(Op.getOperand(4))->getValue();
 
   return DAG.getMemcpy(Op.getOperand(0), DL, Op.getOperand(1), Op.getOperand(2),
                        DAG.getConstant(VaListSize, DL, MVT::i32),
                        Align(PtrSize), false, false, false,
                        MachinePointerInfo(DestSV), MachinePointerInfo(SrcSV));
 }

llvm/lib/Target/AArch64/AArch64Subtarget.h

public:
   bool isTargetWindows() const { return TargetTriple.isOSWindows(); }
   bool isTargetAndroid() const { return TargetTriple.isAndroid(); }
   bool isTargetFuchsia() const { return TargetTriple.isOSFuchsia(); }
 
   bool isTargetCOFF() const { return TargetTriple.isOSBinFormatCOFF(); }
   bool isTargetELF() const { return TargetTriple.isOSBinFormatELF(); }
   bool isTargetMachO() const { return TargetTriple.isOSBinFormatMachO(); }
 
-  bool isTargetILP32() const { return TargetTriple.isArch32Bit(); }
+  bool isTargetILP32() const {
+    return TargetTriple.isArch32Bit() ||
+           TargetTriple.getEnvironment() == Triple::GNUILP32;
+  }
 
   bool useAA() const override { return UseAA; }
 
   bool outlineAtomics() const { return OutlineAtomics; }
 
   bool hasVH() const { return HasVH; }
   bool hasPAN() const { return HasPAN; }
   bool hasLOR() const { return HasLOR; }

llvm/lib/Target/AArch64/AArch64TargetMachine.cpp

static std::unique_ptr<TargetLoweringObjectFile> createTLOF(const Triple &TT) {
 
   return std::make_unique<AArch64_ELFTargetObjectFile>();
 }
 
 // Helper function to build a DataLayout string
 static std::string computeDataLayout(const Triple &TT,
                                      const MCTargetOptions &Options,
                                      bool LittleEndian) {
-  if (Options.getABIName() == "ilp32")
-    return "e-m:e-p:32:32-i8:8-i16:16-i64:64-S128";
   if (TT.isOSBinFormatMachO()) {
     if (TT.getArch() == Triple::aarch64_32)
       return "e-m:o-p:32:32-i64:64-i128:128-n32:64-S128";
     return "e-m:o-i64:64-i128:128-n32:64-S128";
   }
   if (TT.isOSBinFormatCOFF())
     return "e-m:w-p:64:64-i32:32-i64:64-i128:128-n32:64-S128";
-  if (LittleEndian)
-    return "e-m:e-i8:8:32-i16:16:32-i64:64-i128:128-n32:64-S128";
-  return "E-m:e-i8:8:32-i16:16:32-i64:64-i128:128-n32:64-S128";
+  std::string Endian = LittleEndian ? "e" : "E";
+  std::string Ptr32 = TT.getEnvironment() == Triple::GNUILP32 ? "-p:32:32" : "";
+  return Endian + "-m:e" + Ptr32 +
+         "-i8:8:32-i16:16:32-i64:64-i128:128-n32:64-S128";
 }
 
 static StringRef computeDefaultCPU(const Triple &TT, StringRef CPU) {
   if (CPU.empty() && TT.isArm64e())
     return "apple-a12";
   return CPU;
 }
 
 static Reloc::Model getEffectiveRelocModel(const Triple &TT,
AArch64TargetMachine::AArch64TargetMachine(const Target &T, const Triple &TT,
   else if (getCodeModel() == CodeModel::Tiny && this->Options.TLSSize > 24)
     // for the tiny code model, the maximum TLS size is 1MiB (< 16MiB)
     this->Options.TLSSize = 24;
 
   // Enable GlobalISel at or below EnableGlobalISelAt0, unless this is
   // MachO/CodeModel::Large, which GlobalISel does not support.
   if (getOptLevel() <= EnableGlobalISelAtO &&
       TT.getArch() != Triple::aarch64_32 &&
+      TT.getEnvironment() != Triple::GNUILP32 &&
       !(getCodeModel() == CodeModel::Large && TT.isOSBinFormatMachO())) {
     setGlobalISel(true);
     setGlobalISelAbort(GlobalISelAbortMode::Disable);
   }
 
   // AArch64 supports the MachineOutliner.
   setMachineOutliner(true);
 

llvm/lib/Target/AArch64/AsmParser/AArch64AsmParser.cpp

 #define GET_OPERAND_DIAGNOSTIC_TYPES
 #include "AArch64GenAsmMatcher.inc"
   };
   bool IsILP32;
 
   AArch64AsmParser(const MCSubtargetInfo &STI, MCAsmParser &Parser,
                    const MCInstrInfo &MII, const MCTargetOptions &Options)
     : MCTargetAsmParser(Options, STI, MII) {
-    IsILP32 = Options.getABIName() == "ilp32";
+    IsILP32 = STI.getTargetTriple().getEnvironment() == Triple::GNUILP32;
     MCAsmParserExtension::Initialize(Parser);
     MCStreamer &S = getParser().getStreamer();
     if (S.getTargetStreamer() == nullptr)
       new AArch64TargetStreamer(S);
 
     // Alias .hword/.word/.[dx]word to the target-independent
     // .2byte/.4byte/.8byte directives as they have the same form and
     // semantics:

llvm/lib/Target/AArch64/MCTargetDesc/AArch64AsmBackend.cpp

MCAsmBackend *llvm::createAArch64leAsmBackend(const Target &T,
   }
 
   if (TheTriple.isOSBinFormatCOFF())
     return new COFFAArch64AsmBackend(T, TheTriple);
 
   assert(TheTriple.isOSBinFormatELF() && "Invalid target");
 
   uint8_t OSABI = MCELFObjectTargetWriter::getOSABI(TheTriple.getOS());
-  bool IsILP32 = Options.getABIName() == "ilp32";
+  bool IsILP32 = STI.getTargetTriple().getEnvironment() == Triple::GNUILP32;
   return new ELFAArch64AsmBackend(T, TheTriple, OSABI, /*IsLittleEndian=*/true,
                                   IsILP32);
 }
 
 MCAsmBackend *llvm::createAArch64beAsmBackend(const Target &T,
                                               const MCSubtargetInfo &STI,
                                               const MCRegisterInfo &MRI,
                                               const MCTargetOptions &Options) {
   const Triple &TheTriple = STI.getTargetTriple();
   assert(TheTriple.isOSBinFormatELF() &&
          "Big endian is only supported for ELF targets!");
   uint8_t OSABI = MCELFObjectTargetWriter::getOSABI(TheTriple.getOS());
-  bool IsILP32 = Options.getABIName() == "ilp32";
+  bool IsILP32 = STI.getTargetTriple().getEnvironment() == Triple::GNUILP32;
   return new ELFAArch64AsmBackend(T, TheTriple, OSABI, /*IsLittleEndian=*/false,
                                   IsILP32);
 }

llvm/lib/Target/AArch64/MCTargetDesc/AArch64ELFObjectWriter.cpp

protected:
   unsigned getRelocType(MCContext &Ctx, const MCValue &Target,
                         const MCFixup &Fixup, bool IsPCRel) const override;
   bool IsILP32;
 };
 
 } // end anonymous namespace
 
 AArch64ELFObjectWriter::AArch64ELFObjectWriter(uint8_t OSABI, bool IsILP32)
-    : MCELFObjectTargetWriter(/*Is64Bit*/ true, OSABI, ELF::EM_AARCH64,
+    : MCELFObjectTargetWriter(/*Is64Bit*/ !IsILP32, OSABI, ELF::EM_AARCH64,
                               /*HasRelocationAddend*/ true),
       IsILP32(IsILP32) {}
 
 #define R_CLS(rtype)                                                           \
   IsILP32 ? ELF::R_AARCH64_P32_##rtype : ELF::R_AARCH64_##rtype
 #define BAD_ILP32_MOV(lp64rtype)                                               \
   "ILP32 absolute MOV relocation not "                                         \
   "supported (LP64 eqv: " #lp64rtype ")"

llvm/lib/Target/AArch64/MCTargetDesc/AArch64MCAsmInfo.cpp

 AArch64MCAsmInfoELF::AArch64MCAsmInfoELF(const Triple &T) {
   if (T.getArch() == Triple::aarch64_be)
     IsLittleEndian = false;
 
   // We prefer NEON instructions to be printed in the generic form when
   // targeting ELF.
   AssemblerDialect = AsmWriterVariant == Default ? Generic : AsmWriterVariant;
 
-  CodePointerSize = 8;
+  CodePointerSize = T.getEnvironment() == Triple::GNUILP32 ? 4 : 8;
 
   // ".comm align is in bytes but .align is pow-2."
   AlignmentIsInBytes = false;
 
   CommentString = "//";
   PrivateGlobalPrefix = ".L";
   PrivateLabelPrefix = ".L";
   Code32Directive = ".code\t32";

llvm/test/CodeGen/AArch64/ilp32-tlsdesc.ll

+; RUN: llc -mtriple=aarch64-linux-gnu_ilp32 -relocation-model=pic %s -o - | FileCheck %s
+; RUN: llc -mtriple=aarch64-linux-gnu_ilp32 -relocation-model=pic -filetype=obj < %s | llvm-objdump -r - | FileCheck --check-prefix=CHECK-RELOC %s
+
+@var = thread_local global i32 zeroinitializer
+
+define i32 @test_thread_local() {
+; CHECK-LABEL: test_thread_local:
+
+  %val = load i32, i32* @var
+  ret i32 %val
+
+; CHECK: adrp x[[TLSDESC_HI:[0-9]+]], :tlsdesc:var
+; CHECK-NEXT: ldr w[[CALLEE:[0-9]+]], [x[[TLSDESC_HI]], :tlsdesc_lo12:var]
+; CHECK-NEXT: add w0, w[[TLSDESC_HI]], :tlsdesc_lo12:var
+; CHECK-NEXT: .tlsdesccall var
+; CHECK-NEXT: blr x[[CALLEE]]
+
+; CHECK-RELOC: R_AARCH64_P32_TLSDESC_ADR_PAGE21
+; CHECK-RELOC: R_AARCH64_P32_TLSDESC_LD32_LO12
+; CHECK-RELOC: R_AARCH64_P32_TLSDESC_ADD_LO12
+; CHECK-RELOC: R_AARCH64_P32_TLSDESC_CALL
+}

llvm/test/CodeGen/AArch64/ilp32-va.ll

+; RUN: llc -aarch64-load-store-renaming=true -verify-machineinstrs -mtriple=arm64-linux-gnu_ilp32 -pre-RA-sched=linearize -enable-misched=false -disable-post-ra < %s | FileCheck %s
+
+%va_list = type {i8*, i8*, i8*, i32, i32}
+
+@var = dso_local global %va_list zeroinitializer, align 8
+
+declare void @llvm.va_start(i8*)
+
+define dso_local void @test_simple(i32 %n, ...) {
+; CHECK-LABEL: test_simple:
+; CHECK: sub sp, sp, #[[STACKSIZE:[0-9]+]]
+; CHECK: add x[[STACK_TOP:[0-9]+]], sp, #[[STACKSIZE]]
+
+; CHECK: adrp x[[VA_LIST_HI:[0-9]+]], var
+; CHECK: add x[[VA_LIST:[0-9]+]], {{x[0-9]+}}, :lo12:var
+
+; CHECK-DAG: stp x6, x7, [sp, #
+; ... omit middle ones ...
+; CHECK-DAG: str x1, [sp, #[[GR_BASE:[0-9]+]]]
+
+; CHECK-DAG: stp q0, q1, [sp]
+; ... omit middle ones ...
+; CHECK-DAG: stp q6, q7, [sp, #
+
+; CHECK: str w[[STACK_TOP]], [x[[VA_LIST]]]
+
+; CHECK: add x[[GR_TOPTMP:[0-9]+]], sp, #[[GR_BASE]]
+; CHECK: add [[GR_TOP:w[0-9]+]], w[[GR_TOPTMP]], #56
+
+
+; CHECK: mov x[[VR_TOPTMP:[0-9]+]], sp
+; CHECK: add [[VR_TOP:w[0-9]+]], w[[VR_TOPTMP]], #128
+; CHECK: stp [[GR_TOP]], [[VR_TOP]], [x[[VA_LIST]], #4]
+
+; CHECK: mov     [[GRVR:x[0-9]+]], #-56
+; CHECK: movk    [[GRVR]], #65408, lsl #32
+; CHECK: stur    [[GRVR]], [x[[VA_LIST]], #12]
+
+  %addr = bitcast %va_list* @var to i8*
+  call void @llvm.va_start(i8* %addr)
+
+  ret void
+}
+
+define dso_local void @test_fewargs(i32 %n, i32 %n1, i32 %n2, float %m, ...) {
+; CHECK-LABEL: test_fewargs:
+; CHECK: sub sp, sp, #[[STACKSIZE:[0-9]+]]
+; CHECK: add x[[STACK_TOP:[0-9]+]], sp, #[[STACKSIZE]]
+
+; CHECK: adrp x[[VA_LIST_HI:[0-9]+]], var
+; CHECK: add x[[VA_LIST:[0-9]+]], {{x[0-9]+}}, :lo12:var
+
+; CHECK-DAG: stp x6, x7, [sp, #
+; ... omit middle ones ...
+; CHECK-DAG: str x3, [sp, #[[GR_BASE:[0-9]+]]]
+
+; CHECK-DAG: stp q6, q7, [sp, #80]
+; ... omit middle ones ...
+; CHECK-DAG: str q1, [sp]
+
+; CHECK: str w[[STACK_TOP]], [x[[VA_LIST]]]
+
+; CHECK: add x[[GR_TOPTMP:[0-9]+]], sp, #[[GR_BASE]]
+; CHECK: add [[GR_TOP:w[0-9]+]], w[[GR_TOPTMP]], #40
+
+; CHECK: mov x[[VR_TOPTMP:[0-9]+]], sp
+; CHECK: add [[VR_TOP:w[0-9]+]], w[[VR_TOPTMP]], #112
+; CHECK: stp [[GR_TOP]], [[VR_TOP]], [x[[VA_LIST]], #4]
+
+; CHECK: mov  [[GRVR_OFFS:x[0-9]+]], #-40
+; CHECK: movk [[GRVR_OFFS]], #65424, lsl #32
+; CHECK: stur [[GRVR_OFFS]], [x[[VA_LIST]], #12]
+
+  %addr = bitcast %va_list* @var to i8*
+  call void @llvm.va_start(i8* %addr)
+
+  ret void
+}
+
+define dso_local void @test_nospare([8 x i64], [8 x float], ...) {
+; CHECK-LABEL: test_nospare:
+
+  %addr = bitcast %va_list* @var to i8*
+  call void @llvm.va_start(i8* %addr)
+; CHECK-NOT: sub sp, sp
+; CHECK: mov x[[STACK:[0-9]+]], sp
+; CHECK: add x[[VAR:[0-9]+]], {{x[0-9]+}}, :lo12:var
+; CHECK: str w[[STACK]], [x[[VAR]]]
+
+  ret void
+}
+
+; If there are non-variadic arguments on the stack (here two i64s) then the
+; __stack field should point just past them.
+define dso_local void @test_offsetstack([8 x i64], [2 x i64], [3 x float], ...) {
+; CHECK-LABEL: test_offsetstack:
+
+; CHECK-DAG: stp {{q[0-9]+}}, {{q[0-9]+}}, [sp, #48]
+; CHECK-DAG: stp {{q[0-9]+}}, {{q[0-9]+}}, [sp, #16]
+; CHECK-DAG: str {{q[0-9]+}}, [sp]
+; CHECK-DAG: add x[[STACK_TOP:[0-9]+]], sp, #96
+; CHECK-DAG: add x[[VAR:[0-9]+]], {{x[0-9]+}}, :lo12:var
+; CHECK-DAG: str w[[STACK_TOP]], [x[[VAR]]]
+
+  %addr = bitcast %va_list* @var to i8*
+  call void @llvm.va_start(i8* %addr)
+  ret void
+}
+
+declare void @llvm.va_end(i8*)
+
+define dso_local void @test_va_end() nounwind {
+; CHECK-LABEL: test_va_end:
+; CHECK-NEXT: %bb.0
+
+  %addr = bitcast %va_list* @var to i8*
+  call void @llvm.va_end(i8* %addr)
+
+  ret void
+; CHECK-NEXT: ret
+}
+
+declare void @llvm.va_copy(i8* %dest, i8* %src)
+
+@second_list = dso_local global %va_list zeroinitializer
+
+define dso_local void @test_va_copy() {
+; CHECK-LABEL: test_va_copy:
+  %srcaddr = bitcast %va_list* @var to i8*
+  %dstaddr = bitcast %va_list* @second_list to i8*
+  call void @llvm.va_copy(i8* %dstaddr, i8* %srcaddr)
+
+; CHECK: add x[[SRC:[0-9]+]], {{x[0-9]+}}, :lo12:var
+
+; CHECK: ldr [[BLOCK:q[0-9]+]], [x[[SRC]]]
+; CHECK: add x[[DST:[0-9]+]], {{x[0-9]+}}, :lo12:second_list
+; CHECK: ldr [[BLOCK:w[0-9]+]], [x[[SRC]], #16]
+; CHECK: str [[BLOCK:q[0-9]+]], [x[[DST]]]
+; CHECK: str [[BLOCK:w[0-9]+]], [x[[DST]], #16]
+  ret void
+; CHECK: ret
+}

llvm/test/MC/AArch64/adrp-relocation.s

 // RUN: llvm-mc -triple=aarch64-linux-gnu -filetype=obj -o - %s| llvm-readobj -r - | FileCheck %s
-// RUN: llvm-mc -target-abi=ilp32 -triple=aarch64-linux-gnu -filetype=obj \
+// RUN: llvm-mc -triple=aarch64-linux-gnu_ilp32 -filetype=obj \
 // RUN: -o - %s| llvm-readobj -r - | FileCheck -check-prefix=CHECK-ILP32 %s
         .text
 // This tests that LLVM doesn't think it can deal with the relocation on the ADRP
 // itself (even though it knows everything about the relative offsets of sym and
 // the adrp instruction) because its value depends on where this object file's
 // .text section gets relocated in memory.
         adrp x0, sym
         adrp x0, :got:sym

llvm/test/MC/AArch64/arm32-elf-relocs.s

-// RUN: llvm-mc -target-abi=ilp32 -triple=arm64-linux-gnu -o - < %s | \
-// RUN:   FileCheck %s
-// RUN: llvm-mc -target-abi=ilp32 -triple=arm64-linux-gnu -show-encoding \
+// RUN: llvm-mc -triple=arm64-linux-gnu_ilp32 -o - < %s | FileCheck %s
+// RUN: llvm-mc -triple=arm64-linux-gnu_ilp32 -show-encoding \
 // RUN:    -o - < %s | FileCheck --check-prefix=CHECK-ENCODING %s
-// RUN: llvm-mc -target-abi=ilp32 -triple=arm64-linux-gnu -filetype=obj < %s | \
+// RUN: llvm-mc -triple=arm64-linux-gnu_ilp32 -filetype=obj < %s | \
 // RUN:   llvm-objdump --triple=arm64-linux-gnu - -r | \
 // RUN:   FileCheck %s --check-prefix=CHECK-OBJ-ILP32
 
    add x0, x2, #:lo12:sym
 // CHECK: add x0, x2, :lo12:sym
 // CHECK-OBJ-ILP32: 0 R_AARCH64_P32_ADD_ABS_LO12_NC sym
 
    add x5, x7, #:dtprel_lo12:sym

llvm/test/MC/AArch64/arm64-elf-reloc-condbr.s

 // RUN: llvm-mc -triple=arm64-none-linux-gnu -filetype=obj %s -o - | \
 // RUN:   llvm-readobj -r - | FileCheck -check-prefix=OBJ %s
-// RUN: llvm-mc -target-abi=ilp32 -triple=arm64-none-linux-gnu -filetype=obj \
-// RUN:   %s -o - | \
+// RUN: llvm-mc -triple=arm64-none-linux-gnu_ilp32 -filetype=obj %s -o - | \
 // RUN:   llvm-readobj -r - | FileCheck -check-prefix=OBJ-ILP32 %s
 
         b.eq somewhere
 
 // OBJ:      Relocations [
 // OBJ-NEXT:   Section {{.*}} .rela.text {
 // OBJ-NEXT:     0x0 R_AARCH64_CONDBR19 somewhere 0x0
 // OBJ-NEXT:   }
 // OBJ-NEXT: ]
 
 // OBJ-ILP32:      Relocations [
 // OBJ-ILP32-NEXT:   Section {{.*}} .rela.text {
 // OBJ-ILP32-NEXT:     0x0 R_AARCH64_P32_CONDBR19 somewhere 0x0
 // OBJ-ILP32-NEXT:   }
 // OBJ-ILP32-NEXT: ]

llvm/test/MC/AArch64/arm64-ilp32.s

-// RUN: llvm-mc -target-abi=ilp32 -triple aarch64-non-linux-gnu -filetype=obj \
+// RUN: llvm-mc -triple aarch64-non-linux-gnu_ilp32 -filetype=obj \
 // RUN:  %s | llvm-objdump -r - | FileCheck --check-prefix=CHECK-ILP32 %s
-// RUN: llvm-mc                   -triple aarch64-non-linux-gnu -filetype=obj \
+// RUN: llvm-mc -triple aarch64-non-linux-gnu -filetype=obj \
 // RUN:  %s | llvm-objdump -r - | FileCheck --check-prefix=CHECK-LP64 %s
   .text
   .file "../projects/clang/test/Driver/arm64-ilp32.c"
   .globl  foo
   .align  2
   .type foo,@function
 foo:                                    // @foo
 // %bb.0:                               // %entry
   sub sp, sp, #16             // =16
-// CHECK-ILP32: 0000000000000004 R_AARCH64_P32_ADR_PREL_PG_HI21 sizes
-// CHECK-ILP32: 0000000000000008 R_AARCH64_P32_ADD_ABS_LO12_NC sizes
+// CHECK-ILP32: 00000004 R_AARCH64_P32_ADR_PREL_PG_HI21 sizes
+// CHECK-ILP32: 00000008 R_AARCH64_P32_ADD_ABS_LO12_NC sizes
 // CHECK-LP64:  0000000000000004 R_AARCH64_ADR_PREL_PG_HI21 sizes
 // CHECK-LP64:  0000000000000008 R_AARCH64_ADD_ABS_LO12_NC sizes
   adrp  x8, sizes
   add x8, x8, :lo12:sizes
   str w0, [sp, #12]
   str w1, [sp, #8]
   ldr   w0, [x8]
   add sp, sp, #16             // =16

llvm/test/MC/AArch64/elf-reloc-ldrlit.s

 // RUN: llvm-mc -triple=aarch64-none-linux-gnu -filetype=obj %s -o - | \
 // RUN:   llvm-readobj -r - | FileCheck -check-prefix=OBJ %s
-// RUN: llvm-mc -target-abi=ilp32 -triple=aarch64-none-linux-gnu \
-// RUN:   -filetype=obj %s -o - | \
+// RUN: llvm-mc -triple=aarch64-none-linux-gnu_ilp32 -filetype=obj %s -o - | \
 // RUN:   llvm-readobj -r - | FileCheck -check-prefix=OBJ-ILP32 %s
 
         ldr x0, some_label
         ldr w3, some_label
         ldrsw x9, some_label
         prfm pldl3keep, some_label
 
 // OBJ:      Relocations [

llvm/test/MC/AArch64/elf-reloc-tstb.s

 // RUN: llvm-mc -triple=aarch64-none-linux-gnu -filetype=obj %s -o - | \
 // RUN:   llvm-readobj -r - | FileCheck -check-prefix=OBJ %s
-// RUN: llvm-mc -target-abi=ilp32 -triple=aarch64-none-linux-gnu \
-// RUN:   -filetype=obj %s -o - | \
+// RUN: llvm-mc -triple=aarch64-none-linux-gnu_ilp32 -filetype=obj %s -o - | \
 // RUN:   llvm-readobj -r - | FileCheck -check-prefix=OBJ-ILP32 %s
 
         tbz x6, #45, somewhere
         tbnz w3, #15, somewhere
 
 // OBJ:      Relocations [
 // OBJ-NEXT:   Section {{.*}} .rela.text {
 // OBJ-NEXT:     0x0  R_AARCH64_TSTBR14 somewhere 0x0

llvm/test/MC/AArch64/elf-reloc-uncondbrimm.s

 // RUN: llvm-mc -triple=aarch64-none-linux-gnu -filetype=obj %s -o - | \
 // RUN:   llvm-readobj -r - | FileCheck -check-prefix=OBJ %s
-// RUN: llvm-mc -target-abi=ilp32 -triple=aarch64-none-linux-gnu \
-// RUN:   -filetype=obj %s -o - | \
+// RUN: llvm-mc -triple=aarch64-none-linux-gnu_ilp32 -filetype=obj %s -o - | \
 // RUN:   llvm-readobj -r - | FileCheck -check-prefix=OBJ-ILP32 %s
 
         b somewhere
         bl somewhere
 
 // OBJ:      Relocations [
 // OBJ-NEXT:   Section {{.*}} .rela.text {
 // OBJ-NEXT:     0x0 R_AARCH64_JUMP26 somewhere 0x0

llvm/test/MC/AArch64/ilp32-diagnostics.s

-// RUN: not llvm-mc -triple aarch64-none-linux-gnu -target-abi=ilp32 \
+// RUN: not llvm-mc -triple aarch64-none-linux-gnu_ilp32 \
 // RUN:  < %s 2> %t2 -filetype=obj >/dev/null
 // RUN: FileCheck --check-prefix=CHECK-ERROR %s < %t2
 
         .xword sym-.
 // CHECK-ERROR: error: ILP32 8 byte PC relative data relocation not supported (LP64 eqv: PREL64)
 // CHECK-ERROR: ^
 
         .xword sym+16