[ARM] add target arch definitions for 8.1-M and MVE.

[SjoerdMeijer]

Summary

This adds:

• LLVM subtarget features to make all the new instructions conditional on
• CPU and FPU names for use on armclang's command line, with default FPUs set so that "armv8.1-m.main+fp" and "armv8.1-m.main+fp.dp" will select the right FPU features
• architecture extension names "mve" and "mve.fp"
• ABI build attribute support for v8.1-M (a new value for Tag_CPU_arch) and MVE (a new actual tag)

[t.p.northover]

Could probably also do with some testing in unittests/Support/ARMAttributeParser.cpp and TargetParserTest.cpp.

[SjoerdMeijer]

This adds:

• LLVM subtarget features to make all the new instructions conditional on
• CPU and FPU names for use on armclang's command line, with default FPUs set so that "armv8.1-m.main+fp" and "armv8.1-m.main+fp.dp" will select the right FPU features
• architecture extension names "mve" and "mve.fp"
• ABI build attribute support for v8.1-M (a new value for Tag_CPU_arch) and MVE (a new actual tag)

Looks like they could be separate commits. But if that makes things too complicated now, I think we can deal with this initial plumbing work.

llvm/include/llvm/Support/ARMTargetParser.h

48 ↗

(On Diff #195156)

I think we need to regression test this new extension in unittests/Support/TargetParserTest.cpp too.

llvm/lib/Support/ARMAttributeParser.cpp

137 ↗

(On Diff #195156)

what are these nullptrs, and how many do we need? :-)
More seriously, I haven't looked into this bit before, but at least a comment would be good here I guess.

[ostannard]

llvm/lib/Support/ARMAttributeParser.cpp

137 ↗

(On Diff #195156)

The nullptrs are missing strings for "Arm v8-R", "Arm v8.1-A", "Arm v8.2-A" and "Arm v8.3-A", we should add these, but probably as a separate patch.

llvm/lib/Target/ARM/ARMSubtarget.h

114 ↗

(On Diff #195156)

Nit: can we add a trailing comma to tidy up the next diff?

[SjoerdMeijer]

I am helping Simon a bit with upstreaming.

I have:

• rebased this change,
• added a targetparser test,
• added a build attribute test,
• addressed a nit.

[SjoerdMeijer]

More targetparser and buildattribute tests!

[ostannard]

LGTM

[mstojanovic]

llvm/trunk/unittests/Support/TargetParserTest.cpp

578 ↗

(On Diff #202167)

With this addition the test runs out of memory on a MIPS board. Shouldn't this loop go through each flag in the Extensions set or at least through every combination of the flags. It seems like overkill to loop from 0 to 16916342 and push mostly the same elements into the vector each time. Even it the platform doesn't run out of memory the test will be unnecessarily slow.

include/llvm/ADT/Triple.h

enum SubArchType {
     ARMSubArch_v8_4a,
     ARMSubArch_v8_3a,
     ARMSubArch_v8_2a,
     ARMSubArch_v8_1a,
     ARMSubArch_v8,
     ARMSubArch_v8r,
     ARMSubArch_v8m_baseline,
     ARMSubArch_v8m_mainline,
+    ARMSubArch_v8_1m_mainline,
     ARMSubArch_v7,
     ARMSubArch_v7em,
     ARMSubArch_v7m,
     ARMSubArch_v7s,
     ARMSubArch_v7k,
     ARMSubArch_v7ve,
     ARMSubArch_v6,
     ARMSubArch_v6m,

include/llvm/Support/ARMAttributeParser.h

class ARMAttributeParser {
   void THUMB_ISA_use(ARMBuildAttrs::AttrType Tag, const uint8_t *Data,
                      uint32_t &Offset);
   void FP_arch(ARMBuildAttrs::AttrType Tag, const uint8_t *Data,
                uint32_t &Offset);
   void WMMX_arch(ARMBuildAttrs::AttrType Tag, const uint8_t *Data,
                  uint32_t &Offset);
   void Advanced_SIMD_arch(ARMBuildAttrs::AttrType Tag, const uint8_t *Data,
                           uint32_t &Offset);
+  void MVE_arch(ARMBuildAttrs::AttrType Tag, const uint8_t *Data,
+                uint32_t &Offset);
   void PCS_config(ARMBuildAttrs::AttrType Tag, const uint8_t *Data,
                   uint32_t &Offset);
   void ABI_PCS_R9_use(ARMBuildAttrs::AttrType Tag, const uint8_t *Data,
                       uint32_t &Offset);
   void ABI_PCS_RW_data(ARMBuildAttrs::AttrType Tag, const uint8_t *Data,
                        uint32_t &Offset);
   void ABI_PCS_RO_data(ARMBuildAttrs::AttrType Tag, const uint8_t *Data,
                        uint32_t &Offset);

include/llvm/Support/ARMBuildAttributes.h

enum AttrType {
   ABI_FP_optimization_goals = 31,
   compatibility             = 32,
   CPU_unaligned_access      = 34,
   FP_HP_extension           = 36,
   ABI_FP_16bit_format       = 38,
   MPextension_use           = 42, // recoded from 70 (ABI r2.08)
   DIV_use                   = 44,
   DSP_extension             = 46,
+  MVE_arch                  = 48,
   also_compatible_with      = 65,
   conformance               = 67,
   Virtualization_use        = 68,
 
   /// Legacy Tags
   Section                   = 2,  // deprecated (ABI r2.09)
   Symbol                    = 3,  // deprecated (ABI r2.09)
   ABI_align8_needed         = 24, // renamed to ABI_align_needed (ABI r2.09)
enum CPUArch {
   v7       = 10,  // e.g. Cortex A8, Cortex M3
   v6_M     = 11,  // e.g. Cortex M1
   v6S_M    = 12,  // v6_M with the System extensions
   v7E_M    = 13,  // v7_M with DSP extensions
   v8_A     = 14,  // v8_A AArch32
   v8_R     = 15,  // e.g. Cortex R52
   v8_M_Base= 16,  // v8_M_Base AArch32
   v8_M_Main= 17,  // v8_M_Main AArch32
+  v8_1_M_Main=21, // v8_1_M_Main AArch32
 };
 
 enum CPUArchProfile {               // (=7), uleb128
   Not_Applicable          = 0,      // pre v7, or cross-profile code
   ApplicationProfile      = (0x41), // 'A' (e.g. for Cortex A8)
   RealTimeProfile         = (0x52), // 'R' (e.g. for Cortex R4)
   MicroControllerProfile  = (0x4D), // 'M' (e.g. for Cortex M3)
   SystemProfile           = (0x53)  // 'S' Application or real-time profile
enum {
   AllowWMMXv2 = 2,  // The user permitted this entity to use WMMX v2
 
   // Tag_Advanced_SIMD_arch, (=12), uleb128
   AllowNeon = 1,      // SIMDv1 was permitted
   AllowNeon2 = 2,     // SIMDv2 was permitted (Half-precision FP, MAC operations)
   AllowNeonARMv8 = 3, // ARM v8-A SIMD was permitted
   AllowNeonARMv8_1a = 4,// ARM v8.1-A SIMD was permitted (RDMA)
 
+  // Tag_MVE_arch, (=48), uleb128
+  AllowMVEInteger = 1, // integer-only MVE was permitted
+  AllowMVEIntegerAndFloat = 2, // both integer and floating point MVE were permitted
+
   // Tag_ABI_PCS_R9_use, (=14), uleb128
   R9IsGPR = 0,        // R9 used as v6 (just another callee-saved register)
   R9IsSB = 1,         // R9 used as a global static base rgister
   R9IsTLSPointer = 2, // R9 used as a thread local storage pointer
   R9Reserved = 3,     // R9 not used by code associated with attributed entity
 
   // Tag_ABI_PCS_RW_data, (=15), uleb128
   AddressRWPCRel = 1, // Address RW static data PC-relative

include/llvm/Support/ARMTargetParser.h

enum ArchExtKind : unsigned {
   AEK_AES     =     1 << 16,
   AEK_FP16FML =     1 << 17,
   AEK_SB      =     1 << 18,
   AEK_SVE2 =        1 << 19,
   AEK_SVE2AES =     1 << 20,
   AEK_SVE2SM4 =     1 << 21,
   AEK_SVE2SHA3 =    1 << 22,
   AEK_BITPERM =     1 << 23,
+  AEK_FP_DP   =     1 << 24,
   // Unsupported extensions.
   AEK_OS = 0x8000000,
   AEK_IWMMXT = 0x10000000,
   AEK_IWMMXT2 = 0x20000000,
   AEK_MAVERICK = 0x40000000,
   AEK_XSCALE = 0x80000000,
 };
 

 
 // FPU Version
 enum class FPUVersion {
   NONE,
   VFPV2,
   VFPV3,
   VFPV3_FP16,
   VFPV4,
-  VFPV5
+  VFPV5,
+  VFPV5_FULLFP16,
 };
 
 // An FPU name restricts the FPU in one of three ways:
 enum class FPURestriction {
   None = 0, ///< No restriction
   D16,      ///< Only 16 D registers
   SP_D16    ///< Only single-precision instructions, with 16 D registers
 };

include/llvm/Support/ARMTargetParser.def

 ARM_FPU("vfpv3xd", FK_VFPV3XD, FPUVersion::VFPV3, NeonSupportLevel::None, FPURestriction::SP_D16)
 ARM_FPU("vfpv3xd-fp16", FK_VFPV3XD_FP16, FPUVersion::VFPV3_FP16, NeonSupportLevel::None, FPURestriction::SP_D16)
 ARM_FPU("vfpv4", FK_VFPV4, FPUVersion::VFPV4, NeonSupportLevel::None, FPURestriction::None)
 ARM_FPU("vfpv4-d16", FK_VFPV4_D16, FPUVersion::VFPV4, NeonSupportLevel::None, FPURestriction::D16)
 ARM_FPU("fpv4-sp-d16", FK_FPV4_SP_D16, FPUVersion::VFPV4, NeonSupportLevel::None, FPURestriction::SP_D16)
 ARM_FPU("fpv5-d16", FK_FPV5_D16, FPUVersion::VFPV5, NeonSupportLevel::None, FPURestriction::D16)
 ARM_FPU("fpv5-sp-d16", FK_FPV5_SP_D16, FPUVersion::VFPV5, NeonSupportLevel::None, FPURestriction::SP_D16)
 ARM_FPU("fp-armv8", FK_FP_ARMV8, FPUVersion::VFPV5, NeonSupportLevel::None, FPURestriction::None)
+ARM_FPU("fp-armv8-fullfp16-d16", FK_FP_ARMV8_FULLFP16_D16, FPUVersion::VFPV5_FULLFP16, NeonSupportLevel::None, FPURestriction::D16)
+ARM_FPU("fp-armv8-fullfp16-sp-d16", FK_FP_ARMV8_FULLFP16_SP_D16, FPUVersion::VFPV5_FULLFP16, NeonSupportLevel::None, FPURestriction::SP_D16)
 ARM_FPU("neon", FK_NEON, FPUVersion::VFPV3, NeonSupportLevel::Neon, FPURestriction::None)
 ARM_FPU("neon-fp16", FK_NEON_FP16, FPUVersion::VFPV3_FP16, NeonSupportLevel::Neon, FPURestriction::None)
 ARM_FPU("neon-vfpv4", FK_NEON_VFPV4, FPUVersion::VFPV4, NeonSupportLevel::Neon, FPURestriction::None)
 ARM_FPU("neon-fp-armv8", FK_NEON_FP_ARMV8, FPUVersion::VFPV5, NeonSupportLevel::Neon, FPURestriction::None)
 ARM_FPU("crypto-neon-fp-armv8", FK_CRYPTO_NEON_FP_ARMV8, FPUVersion::VFPV5, NeonSupportLevel::Crypto,
         FPURestriction::None)
 ARM_FPU("softvfp", FK_SOFTVFP, FPUVersion::NONE, NeonSupportLevel::None, FPURestriction::None)
 #undef ARM_FPU

 ARM_ARCH("armv8-r", ARMV8R, "8-R", "v8r", ARMBuildAttrs::CPUArch::v8_R,
           FK_NEON_FP_ARMV8,
           (ARM::AEK_MP | ARM::AEK_VIRT | ARM::AEK_HWDIVARM | ARM::AEK_HWDIVTHUMB |
            ARM::AEK_DSP | ARM::AEK_CRC))
 ARM_ARCH("armv8-m.base", ARMV8MBaseline, "8-M.Baseline", "v8m.base",
           ARMBuildAttrs::CPUArch::v8_M_Base, FK_NONE, ARM::AEK_HWDIVTHUMB)
 ARM_ARCH("armv8-m.main", ARMV8MMainline, "8-M.Mainline", "v8m.main",
           ARMBuildAttrs::CPUArch::v8_M_Main, FK_FPV5_D16, ARM::AEK_HWDIVTHUMB)
+ARM_ARCH("armv8.1-m.main", ARMV8_1MMainline, "8.1-M.Mainline", "v8.1m.main",
+         ARMBuildAttrs::CPUArch::v8_1_M_Main, FK_FP_ARMV8_FULLFP16_SP_D16, ARM::AEK_HWDIVTHUMB | ARM::AEK_RAS)
 // Non-standard Arch names.
 ARM_ARCH("iwmmxt", IWMMXT, "iwmmxt", "", ARMBuildAttrs::CPUArch::v5TE,
           FK_NONE, ARM::AEK_NONE)
 ARM_ARCH("iwmmxt2", IWMMXT2, "iwmmxt2", "", ARMBuildAttrs::CPUArch::v5TE,
           FK_NONE, ARM::AEK_NONE)
 ARM_ARCH("xscale", XSCALE, "xscale", "v5e", ARMBuildAttrs::CPUArch::v5TE,
           FK_NONE, ARM::AEK_NONE)
 ARM_ARCH("armv7s", ARMV7S, "7-S", "v7s", ARMBuildAttrs::CPUArch::v7,

 ARM_ARCH_EXT_NAME("none",     ARM::AEK_NONE,     nullptr,  nullptr)
 ARM_ARCH_EXT_NAME("crc",      ARM::AEK_CRC,      "+crc",   "-crc")
 ARM_ARCH_EXT_NAME("crypto",   ARM::AEK_CRYPTO,   "+crypto","-crypto")
 ARM_ARCH_EXT_NAME("sha2",     ARM::AEK_SHA2,     "+sha2",  "-sha2")
 ARM_ARCH_EXT_NAME("aes",      ARM::AEK_AES,      "+aes",   "-aes")
 ARM_ARCH_EXT_NAME("dotprod",  ARM::AEK_DOTPROD,  "+dotprod","-dotprod")
 ARM_ARCH_EXT_NAME("dsp",      ARM::AEK_DSP,      "+dsp",   "-dsp")
 ARM_ARCH_EXT_NAME("fp",       ARM::AEK_FP,       nullptr,  nullptr)
+ARM_ARCH_EXT_NAME("mve",      ARM::AEK_SIMD,     "+mve",   "-mve")
+ARM_ARCH_EXT_NAME("mve.fp",   (ARM::AEK_SIMD | ARM::AEK_FP), "+mve.fp", "-mve.fp")
 ARM_ARCH_EXT_NAME("idiv",     (ARM::AEK_HWDIVARM | ARM::AEK_HWDIVTHUMB), nullptr, nullptr)
 ARM_ARCH_EXT_NAME("mp",       ARM::AEK_MP,       nullptr,  nullptr)
 ARM_ARCH_EXT_NAME("simd",     ARM::AEK_SIMD,     nullptr,  nullptr)
 ARM_ARCH_EXT_NAME("sec",      ARM::AEK_SEC,      nullptr,  nullptr)
 ARM_ARCH_EXT_NAME("virt",     ARM::AEK_VIRT,     nullptr,  nullptr)
 ARM_ARCH_EXT_NAME("fp16",     ARM::AEK_FP16,     "+fullfp16",  "-fullfp16")
 ARM_ARCH_EXT_NAME("ras",      ARM::AEK_RAS,      "+ras", "-ras")
 ARM_ARCH_EXT_NAME("os",       ARM::AEK_OS,       nullptr,  nullptr)

lib/Object/ELFObjectFile.cpp

case ARMBuildAttrs::AllowNeon:
       break;
     case ARMBuildAttrs::AllowNeon2:
       Features.AddFeature("neon");
       Features.AddFeature("fp16");
       break;
     }
   }
 
+  if (Attributes.hasAttribute(ARMBuildAttrs::MVE_arch)) {
+    switch(Attributes.getAttributeValue(ARMBuildAttrs::MVE_arch)) {
+    default:
+      break;
+    case ARMBuildAttrs::Not_Allowed:
+      Features.AddFeature("mve", false);
+      Features.AddFeature("mve.fp", false);
+      break;
+    case ARMBuildAttrs::AllowMVEInteger:
+      Features.AddFeature("mve.fp", false);
+      Features.AddFeature("mve");
+      break;
+    case ARMBuildAttrs::AllowMVEIntegerAndFloat:
+      Features.AddFeature("mve.fp");
+      break;
+    }
+  }
+
   if (Attributes.hasAttribute(ARMBuildAttrs::DIV_use)) {
     switch(Attributes.getAttributeValue(ARMBuildAttrs::DIV_use)) {
     default:
       break;
     case ARMBuildAttrs::DisallowDIV:
       Features.AddFeature("hwdiv", false);
       Features.AddFeature("hwdiv-arm", false);
       break;

lib/Support/ARMAttributeParser.cpp

ARMAttributeParser::DisplayRoutines[] = {
   { ARMBuildAttrs::CPU_name, &ARMAttributeParser::StringAttribute },
   ATTRIBUTE_HANDLER(CPU_arch),
   ATTRIBUTE_HANDLER(CPU_arch_profile),
   ATTRIBUTE_HANDLER(ARM_ISA_use),
   ATTRIBUTE_HANDLER(THUMB_ISA_use),
   ATTRIBUTE_HANDLER(FP_arch),
   ATTRIBUTE_HANDLER(WMMX_arch),
   ATTRIBUTE_HANDLER(Advanced_SIMD_arch),
+  ATTRIBUTE_HANDLER(MVE_arch),
   ATTRIBUTE_HANDLER(PCS_config),
   ATTRIBUTE_HANDLER(ABI_PCS_R9_use),
   ATTRIBUTE_HANDLER(ABI_PCS_RW_data),
   ATTRIBUTE_HANDLER(ABI_PCS_RO_data),
   ATTRIBUTE_HANDLER(ABI_PCS_GOT_use),
   ATTRIBUTE_HANDLER(ABI_PCS_wchar_t),
   ATTRIBUTE_HANDLER(ABI_FP_rounding),
   ATTRIBUTE_HANDLER(ABI_FP_denormal),
void ARMAttributeParser::PrintAttribute(unsigned Tag, unsigned Value,
   }
 }
 
 void ARMAttributeParser::CPU_arch(AttrType Tag, const uint8_t *Data,
                                   uint32_t &Offset) {
   static const char *const Strings[] = {
     "Pre-v4", "ARM v4", "ARM v4T", "ARM v5T", "ARM v5TE", "ARM v5TEJ", "ARM v6",
     "ARM v6KZ", "ARM v6T2", "ARM v6K", "ARM v7", "ARM v6-M", "ARM v6S-M",
-    "ARM v7E-M", "ARM v8"
+    "ARM v7E-M", "ARM v8", nullptr,
+    "ARM v8-M Baseline", "ARM v8-M Mainline", nullptr, nullptr, nullptr,
+    "ARM v8.1-M Mainline"
   };
 
   uint64_t Value = ParseInteger(Data, Offset);
   StringRef ValueDesc =
     (Value < array_lengthof(Strings)) ? Strings[Value] : nullptr;
   PrintAttribute(Tag, Value, ValueDesc);
 }
 
void ARMAttributeParser::Advanced_SIMD_arch(AttrType Tag, const uint8_t *Data,
   };
 
   uint64_t Value = ParseInteger(Data, Offset);
   StringRef ValueDesc =
     (Value < array_lengthof(Strings)) ? Strings[Value] : nullptr;
   PrintAttribute(Tag, Value, ValueDesc);
 }
 
+void ARMAttributeParser::MVE_arch(AttrType Tag, const uint8_t *Data,
+                                  uint32_t &Offset) {
+  static const char *const Strings[] = {
+    "Not Permitted", "MVE integer", "MVE integer and float"
+  };
+
+  uint64_t Value = ParseInteger(Data, Offset);
+  StringRef ValueDesc =
+    (Value < array_lengthof(Strings)) ? Strings[Value] : nullptr;
+  PrintAttribute(Tag, Value, ValueDesc);
+}
+
 void ARMAttributeParser::PCS_config(AttrType Tag, const uint8_t *Data,
                                     uint32_t &Offset) {
   static const char *const Strings[] = {
     "None", "Bare Platform", "Linux Application", "Linux DSO", "Palm OS 2004",
     "Reserved (Palm OS)", "Symbian OS 2004", "Reserved (Symbian OS)"
   };
 
   uint64_t Value = ParseInteger(Data, Offset);

lib/Support/ARMBuildAttrs.cpp

const struct {
   { ARMBuildAttrs::CPU_name, "Tag_CPU_name" },
   { ARMBuildAttrs::CPU_arch, "Tag_CPU_arch" },
   { ARMBuildAttrs::CPU_arch_profile, "Tag_CPU_arch_profile" },
   { ARMBuildAttrs::ARM_ISA_use, "Tag_ARM_ISA_use" },
   { ARMBuildAttrs::THUMB_ISA_use, "Tag_THUMB_ISA_use" },
   { ARMBuildAttrs::FP_arch, "Tag_FP_arch" },
   { ARMBuildAttrs::WMMX_arch, "Tag_WMMX_arch" },
   { ARMBuildAttrs::Advanced_SIMD_arch, "Tag_Advanced_SIMD_arch" },
+  { ARMBuildAttrs::MVE_arch, "Tag_MVE_arch" },
   { ARMBuildAttrs::PCS_config, "Tag_PCS_config" },
   { ARMBuildAttrs::ABI_PCS_R9_use, "Tag_ABI_PCS_R9_use" },
   { ARMBuildAttrs::ABI_PCS_RW_data, "Tag_ABI_PCS_RW_data" },
   { ARMBuildAttrs::ABI_PCS_RO_data, "Tag_ABI_PCS_RO_data" },
   { ARMBuildAttrs::ABI_PCS_GOT_use, "Tag_ABI_PCS_GOT_use" },
   { ARMBuildAttrs::ABI_PCS_wchar_t, "Tag_ABI_PCS_wchar_t" },
   { ARMBuildAttrs::ABI_FP_rounding, "Tag_ABI_FP_rounding" },
   { ARMBuildAttrs::ABI_FP_denormal, "Tag_ABI_FP_denormal" },

lib/Support/ARMTargetParser.cpp

unsigned ARM::parseArchVersion(StringRef Arch) {
   case ArchKind::ARMV8_1A:
   case ArchKind::ARMV8_2A:
   case ArchKind::ARMV8_3A:
   case ArchKind::ARMV8_4A:
   case ArchKind::ARMV8_5A:
   case ArchKind::ARMV8R:
   case ArchKind::ARMV8MBaseline:
   case ArchKind::ARMV8MMainline:
+  case ArchKind::ARMV8_1MMainline:
     return 8;
   case ArchKind::INVALID:
     return 0;
   }
   llvm_unreachable("Unhandled architecture");
 }
 
 // Profile A/R/M
 ARM::ProfileKind ARM::parseArchProfile(StringRef Arch) {
   Arch = getCanonicalArchName(Arch);
   switch (parseArch(Arch)) {
   case ArchKind::ARMV6M:
   case ArchKind::ARMV7M:
   case ArchKind::ARMV7EM:
   case ArchKind::ARMV8MMainline:
   case ArchKind::ARMV8MBaseline:
+  case ArchKind::ARMV8_1MMainline:
     return ProfileKind::M;
   case ArchKind::ARMV7R:
   case ArchKind::ARMV8R:
     return ProfileKind::R;
   case ArchKind::ARMV7A:
   case ArchKind::ARMV7VE:
   case ArchKind::ARMV7K:
   case ArchKind::ARMV8A:
return StringSwitch<StringRef>(Arch)
       .Case("v8.1a", "v8.1-a")
       .Case("v8.2a", "v8.2-a")
       .Case("v8.3a", "v8.3-a")
       .Case("v8.4a", "v8.4-a")
       .Case("v8.5a", "v8.5-a")
       .Case("v8r", "v8-r")
       .Case("v8m.base", "v8-m.base")
       .Case("v8m.main", "v8-m.main")
+      .Case("v8.1m.main", "v8.1-m.main")
       .Default(Arch);
 }
 
 bool ARM::getFPUFeatures(unsigned FPUKind, std::vector<StringRef> &Features) {
 
   if (FPUKind >= FK_LAST || FPUKind == FK_INVALID)
     return false;
 
   // FPU version subtarget features are inclusive of lower-numbered ones, so
   // enable the one corresponding to this version and disable all that are
   // higher. We also have to make sure to disable fp16 when vfp4 is disabled,
   // as +vfp4 implies +fp16 but -vfp4 does not imply -fp16.
   switch (FPUNames[FPUKind].FPUVer) {
+  case FPUVersion::VFPV5_FULLFP16:
+    Features.push_back("+fp-armv8");
+    Features.push_back("+fullfp16");
+    break;
   case FPUVersion::VFPV5:
     Features.push_back("+fp-armv8");
     break;
   case FPUVersion::VFPV4:
     Features.push_back("+vfp4");
     Features.push_back("-fp-armv8");
     break;
   case FPUVersion::VFPV3_FP16:

lib/Support/Triple.cpp

static Triple::SubArchType parseSubArch(StringRef SubArchName) {
   case ARM::ArchKind::ARMV8_5A:
     return Triple::ARMSubArch_v8_5a;
   case ARM::ArchKind::ARMV8R:
     return Triple::ARMSubArch_v8r;
   case ARM::ArchKind::ARMV8MBaseline:
     return Triple::ARMSubArch_v8m_baseline;
   case ARM::ArchKind::ARMV8MMainline:
     return Triple::ARMSubArch_v8m_mainline;
+  case ARM::ArchKind::ARMV8_1MMainline:
+    return Triple::ARMSubArch_v8_1m_mainline;
   default:
     return Triple::NoSubArch;
   }
 }
 
 static StringRef getObjectFormatTypeName(Triple::ObjectFormatType Kind) {
   switch (Kind) {
   case Triple::UnknownObjectFormat: return "";

lib/Target/ARM/ARM.td

 def HasV8_4aOps   : SubtargetFeature<"v8.4a", "HasV8_4aOps", "true",
                                    "Support ARM v8.4a instructions",
                                    [HasV8_3aOps, FeatureDotProd]>;
 
 def HasV8_5aOps   : SubtargetFeature<"v8.5a", "HasV8_5aOps", "true",
                                    "Support ARM v8.5a instructions",
                                    [HasV8_4aOps, FeatureSB]>;
 
+def HasV8_1MMainlineOps : SubtargetFeature<
+               "v8.1m.main", "HasV8_1MMainlineOps", "true",
+               "Support ARM v8-1M Mainline instructions",
+               [HasV8MMainlineOps]>;
+def HasMVEIntegerOps : SubtargetFeature<
+               "mve", "HasMVEIntegerOps", "true",
+               "Support M-Class Vector Extension with integer ops",
+               [HasV8_1MMainlineOps, FeatureDSP, FeatureFPRegs16, FeatureFPRegs64]>;
+def HasMVEFloatOps : SubtargetFeature<
+               "mve.fp", "HasMVEFloatOps", "true",
+               "Support M-Class Vector Extension with integer and floating ops",
+               [HasMVEIntegerOps, FeatureFPARMv8_D16_SP, FeatureFullFP16]>;
+
 //===----------------------------------------------------------------------===//
 // ARM Processor subtarget features.
 //
 
 def ProcA5      : SubtargetFeature<"a5", "ARMProcFamily", "CortexA5",
                                    "Cortex-A5 ARM processors", []>;
 def ProcA7      : SubtargetFeature<"a7", "ARMProcFamily", "CortexA7",
                                    "Cortex-A7 ARM processors", []>;
def ARMv8mMainline : Architecture<"armv8-m.main", "ARMv8mMainline",
                                                        FeatureNoARM,
                                                        ModeThumb,
                                                        FeatureDB,
                                                        FeatureHWDivThumb,
                                                        Feature8MSecExt,
                                                        FeatureAcquireRelease,
                                                        FeatureMClass]>;
 
+def ARMv81mMainline : Architecture<"armv8.1-m.main", "ARMv81mMainline",
+                                                      [HasV8_1MMainlineOps,
+                                                       FeatureNoARM,
+                                                       ModeThumb,
+                                                       FeatureDB,
+                                                       FeatureHWDivThumb,
+                                                       Feature8MSecExt,
+                                                       FeatureAcquireRelease,
+                                                       FeatureMClass,
+                                                       FeatureRAS]>;
+
 // Aliases
 def IWMMXT   : Architecture<"iwmmxt",      "ARMv5te",  [ARMv5te]>;
 def IWMMXT2  : Architecture<"iwmmxt2",     "ARMv5te",  [ARMv5te]>;
 def XScale   : Architecture<"xscale",      "ARMv5te",  [ARMv5te]>;
 def ARMv6j   : Architecture<"armv6j",      "ARMv7a",   [ARMv6]>;
 def ARMv7k   : Architecture<"armv7k",      "ARMv7a",   [ARMv7a]>;
 def ARMv7s   : Architecture<"armv7s",      "ARMv7a",   [ARMv7a]>;
 

lib/Target/ARM/ARMPredicates.td

def HasV6M           : Predicate<"Subtarget->hasV6MOps()">,
                                  AssemblerPredicate<"HasV6MOps",
                                                     "armv6m or armv6t2">;
 def HasV8MBaseline   : Predicate<"Subtarget->hasV8MBaselineOps()">,
                                  AssemblerPredicate<"HasV8MBaselineOps",
                                                     "armv8m.base">;
 def HasV8MMainline   : Predicate<"Subtarget->hasV8MMainlineOps()">,
                                  AssemblerPredicate<"HasV8MMainlineOps",
                                                     "armv8m.main">;
+def HasV8_1MMainline : Predicate<"Subtarget->hasV8_1MMainlineOps()">,
+                                 AssemblerPredicate<"HasV8_1MMainlineOps",
+                                                    "armv8.1m.main">;
+def HasMVEInt        : Predicate<"Subtarget->hasMVEIntegerOps()">,
+                                 AssemblerPredicate<"HasMVEIntegerOps",
+                                                    "mve">;
+def HasMVEFloat      : Predicate<"Subtarget->hasMVEFloatOps()">,
+                                 AssemblerPredicate<"HasMVEFloatOps",
+                                                    "mve.fp">;
 def HasFPRegs        : Predicate<"Subtarget->hasFPRegs()">,
                                  AssemblerPredicate<"FeatureFPRegs",
                                                     "fp registers">;
 def HasFPRegs16      : Predicate<"Subtarget->hasFPRegs16()">,
                                  AssemblerPredicate<"FeatureFPRegs16",
                                                     "16-bit fp registers">;
 def HasFPRegs64      : Predicate<"Subtarget->hasFPRegs64()">,
                                  AssemblerPredicate<"FeatureFPRegs64",
                                                     "64-bit fp registers">;
+def HasFPRegsV8_1M   : Predicate<"Subtarget->hasFPRegs() && Subtarget->hasV8_1MMainlineOps()">,
+                                 AssemblerPredicate<"FeatureFPRegs,HasV8_1MMainlineOps",
+                                                    "armv8.1m.main with FP or MVE">;
 def HasV6T2          : Predicate<"Subtarget->hasV6T2Ops()">,
                                  AssemblerPredicate<"HasV6T2Ops", "armv6t2">;
 def NoV6T2           : Predicate<"!Subtarget->hasV6T2Ops()">;
 def HasV6K           : Predicate<"Subtarget->hasV6KOps()">,
                                  AssemblerPredicate<"HasV6KOps", "armv6k">;
 def NoV6K            : Predicate<"!Subtarget->hasV6KOps()">;
 def HasV7            : Predicate<"Subtarget->hasV7Ops()">,
                                  AssemblerPredicate<"HasV7Ops", "armv7">;

lib/Target/ARM/ARMSubtarget.h

enum ARMArchEnum {
     ARMv81a,
     ARMv82a,
     ARMv83a,
     ARMv84a,
     ARMv85a,
     ARMv8a,
     ARMv8mBaseline,
     ARMv8mMainline,
-    ARMv8r
+    ARMv8r,
+    ARMv81mMainline,
   };
 
 public:
   /// What kind of timing do load multiple/store multiple instructions have.
   enum ARMLdStMultipleTiming {
     /// Can load/store 2 registers/cycle.
     DoubleIssue,
     /// Can load/store 2 registers/cycle, but needs an extra cycle if the access
protected:
   bool HasV8Ops = false;
   bool HasV8_1aOps = false;
   bool HasV8_2aOps = false;
   bool HasV8_3aOps = false;
   bool HasV8_4aOps = false;
   bool HasV8_5aOps = false;
   bool HasV8MBaselineOps = false;
   bool HasV8MMainlineOps = false;
+  bool HasV8_1MMainlineOps = false;
+  bool HasMVEIntegerOps = false;
+  bool HasMVEFloatOps = false;
 
   /// HasVFPv2, HasVFPv3, HasVFPv4, HasFPARMv8, HasNEON - Specify what
   /// floating point ISAs are supported.
   bool HasVFPv2 = false;
   bool HasVFPv3 = false;
   bool HasVFPv4 = false;
   bool HasFPARMv8 = false;
   bool HasNEON = false;
public:
   bool hasV8Ops()   const { return HasV8Ops;  }
   bool hasV8_1aOps() const { return HasV8_1aOps; }
   bool hasV8_2aOps() const { return HasV8_2aOps; }
   bool hasV8_3aOps() const { return HasV8_3aOps; }
   bool hasV8_4aOps() const { return HasV8_4aOps; }
   bool hasV8_5aOps() const { return HasV8_5aOps; }
   bool hasV8MBaselineOps() const { return HasV8MBaselineOps; }
   bool hasV8MMainlineOps() const { return HasV8MMainlineOps; }
+  bool hasV8_1MMainlineOps() const { return HasV8_1MMainlineOps; }
+  bool hasMVEIntegerOps() const { return HasMVEIntegerOps; }
+  bool hasMVEFloatOps() const { return HasMVEFloatOps; }
   bool hasFPRegs() const { return HasFPRegs; }
   bool hasFPRegs16() const { return HasFPRegs16; }
   bool hasFPRegs64() const { return HasFPRegs64; }
 
   /// @{
   /// These functions are obsolete, please consider adding subtarget features
   /// or properties instead of calling them.
   bool isCortexA5() const { return ARMProcFamily == CortexA5; }

lib/Target/ARM/MCTargetDesc/ARMTargetStreamer.cpp

 static ARMBuildAttrs::CPUArch getArchForCPU(const MCSubtargetInfo &STI) {
   if (STI.getCPU() == "xscale")
     return ARMBuildAttrs::v5TEJ;
 
   if (STI.hasFeature(ARM::HasV8Ops)) {
     if (STI.hasFeature(ARM::FeatureRClass))
       return ARMBuildAttrs::v8_R;
     return ARMBuildAttrs::v8_A;
-  } else if (STI.hasFeature(ARM::HasV8MMainlineOps))
+  } else if (STI.hasFeature(ARM::HasV8_1MMainlineOps))
+    return ARMBuildAttrs::v8_1_M_Main;
+  else if (STI.hasFeature(ARM::HasV8MMainlineOps))
     return ARMBuildAttrs::v8_M_Main;
   else if (STI.hasFeature(ARM::HasV7Ops)) {
     if (STI.hasFeature(ARM::FeatureMClass) && STI.hasFeature(ARM::FeatureDSP))
       return ARMBuildAttrs::v7E_M;
     return ARMBuildAttrs::v7;
   } else if (STI.hasFeature(ARM::HasV6T2Ops))
     return ARMBuildAttrs::v6T2;
   else if (STI.hasFeature(ARM::HasV8MBaselineOps))
emitAttribute(ARMBuildAttrs::ABI_HardFP_use,
                   ARMBuildAttrs::HardFPSinglePrecision);
 
   if (STI.hasFeature(ARM::FeatureFP16))
     emitAttribute(ARMBuildAttrs::FP_HP_extension, ARMBuildAttrs::AllowHPFP);
 
   if (STI.hasFeature(ARM::FeatureMP))
     emitAttribute(ARMBuildAttrs::MPextension_use, ARMBuildAttrs::AllowMP);
 
+  if (STI.hasFeature(ARM::HasMVEFloatOps))
+    emitAttribute(ARMBuildAttrs::MVE_arch, ARMBuildAttrs::AllowMVEIntegerAndFloat);
+  else if (STI.hasFeature(ARM::HasMVEIntegerOps))
+    emitAttribute(ARMBuildAttrs::MVE_arch, ARMBuildAttrs::AllowMVEInteger);
+
   // Hardware divide in ARM mode is part of base arch, starting from ARMv8.
   // If only Thumb hwdiv is present, it must also be in base arch (ARMv7-R/M).
   // It is not possible to produce DisallowDIV: if hwdiv is present in the base
   // arch, supplying -hwdiv downgrades the effective arch, via ClearImpliedBits.
   // AllowDIVExt is only emitted if hwdiv isn't available in the base arch;
   // otherwise, the default value (AllowDIVIfExists) applies.
   if (STI.hasFeature(ARM::FeatureHWDivARM) && !STI.hasFeature(ARM::HasV8Ops))
     emitAttribute(ARMBuildAttrs::DIV_use, ARMBuildAttrs::AllowDIVExt);

test/CodeGen/ARM/build-attributes.ll

 ; RUN: llc < %s -mtriple=arm-none-none-eabi -mcpu=cortex-r52 | FileCheck %s --check-prefix=ARMv8R --check-prefix=ARMv8R-NEON
 
 ; ARMv8-M
 ; RUN: llc < %s -mtriple=thumbv8-none-none-eabi -mcpu=cortex-m23 | FileCheck %s --check-prefix=STRICT-ALIGN
 ; RUN: llc < %s -mtriple=thumbv8-none-none-eabi -mcpu=cortex-m33 | FileCheck %s --check-prefix=NO-STRICT-ALIGN
 ; RUN: llc < %s -mtriple=thumbv8-none-none-eabi -mcpu=cortex-m33 -mattr=+strict-align | FileCheck %s --check-prefix=STRICT-ALIGN
 ; RUN: llc < %s -mtriple=thumbv8-none-none-eabi -mcpu=cortex-m35p | FileCheck %s --check-prefix=NO-STRICT-ALIGN
 ; RUN: llc < %s -mtriple=thumbv8-none-none-eabi -mcpu=cortex-m35p -mattr=+strict-align | FileCheck %s --check-prefix=STRICT-ALIGN
+; RUN: llc < %s -mtriple=thumbv8.1m.main-none-none-eabi | FileCheck %s --check-prefix=ARMv81M-MAIN
+; RUN: llc < %s -mtriple=thumbv8.1m.main-none-none-eabi -mattr=+mve | FileCheck %s --check-prefix=ARMv81M-MAIN-MVEINT
+; RUN: llc < %s -mtriple=thumbv8.1m.main-none-none-eabi -mattr=+mve.fp | FileCheck %s --check-prefix=ARMv81M-MAIN-MVEFP
 
 ; CPU-SUPPORTED-NOT: is not a recognized processor for this target
 
 ; XSCALE:      .eabi_attribute 6, 5
 ; XSCALE:      .eabi_attribute 8, 1
 ; XSCALE:      .eabi_attribute 9, 1
 
 ; DYN-ROUNDING: .eabi_attribute 19, 1

 ; ARMv8R-NOFPU-NOT: .eabi_attribute 36
 ; ARMv8R-SP: .eabi_attribute 36, 1   @ Tag_FP_HP_extension
 ; ARMv8R-NEON: .eabi_attribute 36, 1   @ Tag_FP_HP_extension
 ; ARMv8R: .eabi_attribute 42, 1   @ Tag_MPextension_use
 ; ARMv8R: .eabi_attribute 68, 2   @ Tag_Virtualization_use
 ; ARMv8R: .eabi_attribute 38, 1   @ Tag_ABI_FP_16bit_format
 ; ARMv8R: .eabi_attribute 14, 0   @ Tag_ABI_PCS_R9_use
 
+; ARMv81M-MAIN: .eabi_attribute 6, 21 @ Tag_CPU_arch
+; ARMv81M-MAIN-NOT: .eabi_attribute 48
+; ARMv81M-MAIN-MVEINT: .eabi_attribute 6, 21 @ Tag_CPU_arch
+; ARMv81M-MAIN-MVEINT: .eabi_attribute 48, 1 @ Tag_MVE_arch
+; ARMv81M-MAIN-MVEFP: .eabi_attribute 6, 21 @ Tag_CPU_arch
+; ARMv81M-MAIN-MVEFP: .eabi_attribute 48, 2 @ Tag_MVE_arch
 define i32 @f(i64 %z) {
     ret i32 0
 }

unittests/Support/ARMAttributeParser.cpp

 TEST(FPHPBuildAttr, testBuildAttr) {
   EXPECT_TRUE(testTagString(36, "Tag_FP_HP_extension"));
   EXPECT_TRUE(testBuildAttr(36, 0, ARMBuildAttrs::FP_HP_extension,
                             ARMBuildAttrs::Not_Allowed));
   EXPECT_TRUE(testBuildAttr(36, 1, ARMBuildAttrs::FP_HP_extension,
                             ARMBuildAttrs::AllowHPFP));
 }
 
+TEST(MVEBuildAttr, testBuildAttr) {
+  EXPECT_TRUE(testTagString(48, "Tag_MVE_arch"));
+  EXPECT_TRUE(testBuildAttr(48, 0, ARMBuildAttrs::MVE_arch,
+                            ARMBuildAttrs::Not_Allowed));
+  EXPECT_TRUE(testBuildAttr(48, 1, ARMBuildAttrs::MVE_arch,
+                            ARMBuildAttrs::AllowMVEInteger));
+  EXPECT_TRUE(testBuildAttr(48, 2, ARMBuildAttrs::MVE_arch,
+                            ARMBuildAttrs::AllowMVEIntegerAndFloat));
+}
+
 TEST(CPUAlignBuildAttr, testBuildAttr) {
   EXPECT_TRUE(testTagString(34, "Tag_CPU_unaligned_access"));
   EXPECT_TRUE(testBuildAttr(34, 0, ARMBuildAttrs::CPU_unaligned_access,
                             ARMBuildAttrs::Not_Allowed));
   EXPECT_TRUE(testBuildAttr(34, 1, ARMBuildAttrs::CPU_unaligned_access,
                             ARMBuildAttrs::Allowed));
 }
 

unittests/Support/TargetParserTest.cpp

TEST(TargetParserTest, ARMFPURestriction) {
   }
 }
 
 TEST(TargetParserTest, ARMExtensionFeatures) {
   std::vector<StringRef> Features;
   unsigned Extensions = ARM::AEK_CRC | ARM::AEK_CRYPTO | ARM::AEK_DSP |
                         ARM::AEK_HWDIVARM | ARM::AEK_HWDIVTHUMB | ARM::AEK_MP |
                         ARM::AEK_SEC | ARM::AEK_VIRT | ARM::AEK_RAS | ARM::AEK_FP16 |
-                        ARM::AEK_FP16FML;
+                        ARM::AEK_FP16FML | ARM::AEK_FP_DP;
 
   for (unsigned i = 0; i <= Extensions; i++)
     EXPECT_TRUE(i == 0 ? !ARM::getExtensionFeatures(i, Features)
                        : ARM::getExtensionFeatures(i, Features));
 }
 
 TEST(TargetParserTest, ARMFPUFeatures) {
   std::vector<StringRef> Features;
const char *ArchExt[][4] = {{"crc", "nocrc", "+crc", "-crc"},
                               {"fp16fml", "nofp16fml", "+fp16fml", "-fp16fml"},
                               {"ras", "noras", "+ras", "-ras"},
                               {"dotprod", "nodotprod", "+dotprod", "-dotprod"},
                               {"os", "noos", nullptr, nullptr},
                               {"iwmmxt", "noiwmmxt", nullptr, nullptr},
                               {"iwmmxt2", "noiwmmxt2", nullptr, nullptr},
                               {"maverick", "maverick", nullptr, nullptr},
                               {"xscale", "noxscale", nullptr, nullptr},
-                              {"sb", "nosb", "+sb", "-sb"}};
+                              {"sb", "nosb", "+sb", "-sb"},
+                              {"mve", "nomve", "+mve", "-mve"},
+                              {"mve.fp", "nomve.fp", "+mve.fp", "-mve.fp"}};
 
   for (unsigned i = 0; i < array_lengthof(ArchExt); i++) {
     EXPECT_EQ(StringRef(ArchExt[i][2]), ARM::getArchExtFeature(ArchExt[i][0]));
     EXPECT_EQ(StringRef(ArchExt[i][3]), ARM::getArchExtFeature(ArchExt[i][1]));
   }
 }
 
 TEST(TargetParserTest, ARMparseHWDiv) {