Merge branch 'main' into users/meinersbur/flang_runtime_split-headersusers/meinersbur/flang_runtime_split-headers

7 files changed, 1019 insertions, 269 deletions

diff --git a/lldb/tools/debugserver/source/DNBDefs.h b/lldb/tools/debugserver/source/DNBDefs.h
index dacee652b3eb..df8ca809d412 100644
--- a/lldb/tools/debugserver/source/DNBDefs.h
+++ b/lldb/tools/debugserver/source/DNBDefs.h
@@ -312,16 +312,21 @@ struct DNBRegisterValue {
     uint64_t uint64;
     float float32;
     double float64;
-    int8_t v_sint8[64];
-    int16_t v_sint16[32];
-    int32_t v_sint32[16];
-    int64_t v_sint64[8];
-    uint8_t v_uint8[64];
-    uint16_t v_uint16[32];
-    uint32_t v_uint32[16];
-    uint64_t v_uint64[8];
-    float v_float32[16];
-    double v_float64[8];
+    // AArch64 SME's ZA register max size is 64k, this object must be
+    // large enough to hold that much data.  The current Apple cores
+    // have a much smaller maximum ZA reg size, but there are not
+    // multiple copies of this object so increase the static size to
+    // maximum possible.
+    int8_t v_sint8[65536];
+    int16_t v_sint16[32768];
+    int32_t v_sint32[16384];
+    int64_t v_sint64[8192];
+    uint8_t v_uint8[65536];
+    uint16_t v_uint16[32768];
+    uint32_t v_uint32[16384];
+    uint64_t v_uint64[8192];
+    float v_float32[16384];
+    double v_float64[8192];
     void *pointer;
     char *c_str;
   } value;
diff --git a/lldb/tools/debugserver/source/MacOSX/MachProcess.mm b/lldb/tools/debugserver/source/MacOSX/MachProcess.mm
index 530c5b24b424..a2179bf2f91e 100644
--- a/lldb/tools/debugserver/source/MacOSX/MachProcess.mm
+++ b/lldb/tools/debugserver/source/MacOSX/MachProcess.mm
@@ -1417,15 +1417,17 @@ void MachProcess::RefineWatchpointStopInfo(
       continue;
     for (uint32_t reg = 0; reg < reg_sets[set].num_registers; ++reg) {
       if (strcmp(reg_sets[set].registers[reg].name, "esr") == 0) {
-        DNBRegisterValue reg_value;
-        if (GetRegisterValue(tid, set, reg, &reg_value)) {
-          esr = reg_value.value.uint64;
+        std::unique_ptr<DNBRegisterValue> reg_value =
+            std::make_unique<DNBRegisterValue>();
+        if (GetRegisterValue(tid, set, reg, reg_value.get())) {
+          esr = reg_value->value.uint64;
         }
       }
       if (strcmp(reg_sets[set].registers[reg].name, "far") == 0) {
-        DNBRegisterValue reg_value;
-        if (GetRegisterValue(tid, set, reg, &reg_value)) {
-          far = reg_value.value.uint64;
+        std::unique_ptr<DNBRegisterValue> reg_value =
+            std::make_unique<DNBRegisterValue>();
+        if (GetRegisterValue(tid, set, reg, reg_value.get())) {
+          far = reg_value->value.uint64;
         }
       }
     }
diff --git a/lldb/tools/debugserver/source/MacOSX/MachThread.cpp b/lldb/tools/debugserver/source/MacOSX/MachThread.cpp
index de2bebfcec70..69e1c9bb0e25 100644
--- a/lldb/tools/debugserver/source/MacOSX/MachThread.cpp
+++ b/lldb/tools/debugserver/source/MacOSX/MachThread.cpp
@@ -509,10 +509,12 @@ void MachThread::DumpRegisterState(nub_size_t regSet) {
     if (m_arch_up->RegisterSetStateIsValid((int)regSet)) {
       const size_t numRegisters = GetNumRegistersInSet(regSet);
       uint32_t regIndex = 0;
-      DNBRegisterValueClass reg;
+      std::unique_ptr<DNBRegisterValueClass> reg =
+          std::make_unique<DNBRegisterValueClass>();
       for (regIndex = 0; regIndex < numRegisters; ++regIndex) {
-        if (m_arch_up->GetRegisterValue((uint32_t)regSet, regIndex, &reg)) {
-          reg.Dump(NULL, NULL);
+        if (m_arch_up->GetRegisterValue((uint32_t)regSet, regIndex,
+                                        reg.get())) {
+          reg->Dump(NULL, NULL);
         }
       }
     } else {
diff --git a/lldb/tools/debugserver/source/MacOSX/arm64/DNBArchImplARM64.cpp b/lldb/tools/debugserver/source/MacOSX/arm64/DNBArchImplARM64.cpp
index b6f52cb5cf49..34a4ee21f850 100644
--- a/lldb/tools/debugserver/source/MacOSX/arm64/DNBArchImplARM64.cpp
+++ b/lldb/tools/debugserver/source/MacOSX/arm64/DNBArchImplARM64.cpp
@@ -93,6 +93,45 @@ DNBArchMachARM64::SoftwareBreakpointOpcode(nub_size_t byte_size) {
 
 uint32_t DNBArchMachARM64::GetCPUType() { return CPU_TYPE_ARM64; }
 
+static std::once_flag g_cpu_has_sme_once;
+bool DNBArchMachARM64::CPUHasSME() {
+  static bool g_has_sme = false;
+  std::call_once(g_cpu_has_sme_once, []() {
+    int ret = 0;
+    size_t size = sizeof(ret);
+    if (sysctlbyname("hw.optional.arm.FEAT_SME", &ret, &size, NULL, 0) != -1)
+      g_has_sme = ret == 1;
+  });
+  return g_has_sme;
+}
+
+static std::once_flag g_cpu_has_sme2_once;
+bool DNBArchMachARM64::CPUHasSME2() {
+  static bool g_has_sme2 = false;
+  std::call_once(g_cpu_has_sme2_once, []() {
+    int ret = 0;
+    size_t size = sizeof(ret);
+    if (sysctlbyname("hw.optional.arm.FEAT_SME2", &ret, &size, NULL, 0) != -1)
+      g_has_sme2 = ret == 1;
+  });
+  return g_has_sme2;
+}
+
+static std::once_flag g_sme_max_svl_once;
+unsigned int DNBArchMachARM64::GetSMEMaxSVL() {
+  static unsigned int g_sme_max_svl = 0;
+  std::call_once(g_sme_max_svl_once, []() {
+    if (CPUHasSME()) {
+      unsigned int ret = 0;
+      size_t size = sizeof(ret);
+      if (sysctlbyname("hw.optional.arm.sme_max_svl_b", &ret, &size, NULL, 0) !=
+          -1)
+        g_sme_max_svl = ret;
+    }
+  });
+  return g_sme_max_svl;
+}
+
 static uint64_t clear_pac_bits(uint64_t value) {
   uint32_t addressing_bits = 0;
   if (!DNBGetAddressingBits(addressing_bits))
@@ -415,6 +454,118 @@ kern_return_t DNBArchMachARM64::GetDBGState(bool force) {
   return kret;
 }
 
+kern_return_t DNBArchMachARM64::GetSVEState(bool force) {
+  int set = e_regSetSVE;
+  // Check if we have valid cached registers
+  if (!force && m_state.GetError(set, Read) == KERN_SUCCESS)
+    return KERN_SUCCESS;
+
+  if (!CPUHasSME())
+    return KERN_INVALID_ARGUMENT;
+
+  // If the processor is not in Streaming SVE Mode, these thread_get_states
+  // will fail, and we may return uninitialized data in the register context.
+  memset(&m_state.context.sve.z[0], 0,
+         ARM_SVE_Z_STATE_COUNT * sizeof(uint32_t));
+  memset(&m_state.context.sve.z[16], 0,
+         ARM_SVE_Z_STATE_COUNT * sizeof(uint32_t));
+  memset(&m_state.context.sve.p[0], 0,
+         ARM_SVE_P_STATE_COUNT * sizeof(uint32_t));
+
+  // Read the registers from our thread
+  mach_msg_type_number_t count = ARM_SVE_Z_STATE_COUNT;
+  kern_return_t kret =
+      ::thread_get_state(m_thread->MachPortNumber(), ARM_SVE_Z_STATE1,
+                         (thread_state_t)&m_state.context.sve.z[0], &count);
+  m_state.SetError(set, Read, kret);
+  DNBLogThreadedIf(LOG_THREAD, "Read SVE registers z0..z15 return value %d",
+                   kret);
+  if (kret != KERN_SUCCESS)
+    return kret;
+
+  count = ARM_SVE_Z_STATE_COUNT;
+  kret = thread_get_state(m_thread->MachPortNumber(), ARM_SVE_Z_STATE2,
+                          (thread_state_t)&m_state.context.sve.z[16], &count);
+  m_state.SetError(set, Read, kret);
+  DNBLogThreadedIf(LOG_THREAD, "Read SVE registers z16..z31 return value %d",
+                   kret);
+  if (kret != KERN_SUCCESS)
+    return kret;
+
+  count = ARM_SVE_P_STATE_COUNT;
+  kret = thread_get_state(m_thread->MachPortNumber(), ARM_SVE_P_STATE,
+                          (thread_state_t)&m_state.context.sve.p[0], &count);
+  m_state.SetError(set, Read, kret);
+  DNBLogThreadedIf(LOG_THREAD, "Read SVE registers p0..p15 return value %d",
+                   kret);
+
+  return kret;
+}
+
+kern_return_t DNBArchMachARM64::GetSMEState(bool force) {
+  int set = e_regSetSME;
+  // Check if we have valid cached registers
+  if (!force && m_state.GetError(set, Read) == KERN_SUCCESS)
+    return KERN_SUCCESS;
+
+  if (!CPUHasSME())
+    return KERN_INVALID_ARGUMENT;
+
+  // If the processor is not in Streaming SVE Mode, these thread_get_states
+  // will fail, and we may return uninitialized data in the register context.
+  memset(&m_state.context.sme.svcr, 0, ARM_SME_STATE_COUNT * sizeof(uint32_t));
+  memset(m_state.context.sme.za.data(), 0, m_state.context.sme.za.size());
+  if (CPUHasSME2())
+    memset(&m_state.context.sme.zt0, 0,
+           ARM_SME2_STATE_COUNT * sizeof(uint32_t));
+
+  // Read the registers from our thread
+  mach_msg_type_number_t count = ARM_SME_STATE_COUNT;
+  kern_return_t kret =
+      ::thread_get_state(m_thread->MachPortNumber(), ARM_SME_STATE,
+                         (thread_state_t)&m_state.context.sme.svcr, &count);
+  m_state.SetError(set, Read, kret);
+  DNBLogThreadedIf(LOG_THREAD, "Read ARM_SME_STATE return value %d", kret);
+  if (kret != KERN_SUCCESS)
+    return kret;
+
+  size_t za_size = m_state.context.sme.svl_b * m_state.context.sme.svl_b;
+  const size_t max_chunk_size = 4096;
+  int n_chunks;
+  size_t chunk_size;
+  if (za_size <= max_chunk_size) {
+    n_chunks = 1;
+    chunk_size = za_size;
+  } else {
+    n_chunks = za_size / max_chunk_size;
+    chunk_size = max_chunk_size;
+  }
+  for (int i = 0; i < n_chunks; i++) {
+    count = ARM_SME_ZA_STATE_COUNT;
+    arm_sme_za_state_t za_state;
+    kret = thread_get_state(m_thread->MachPortNumber(), ARM_SME_ZA_STATE1 + i,
+                            (thread_state_t)&za_state, &count);
+    m_state.SetError(set, Read, kret);
+    DNBLogThreadedIf(LOG_THREAD, "Read ARM_SME_STATE return value %d", kret);
+    if (kret != KERN_SUCCESS)
+      return kret;
+    memcpy(m_state.context.sme.za.data() + (i * chunk_size), &za_state,
+           chunk_size);
+  }
+
+  if (CPUHasSME2()) {
+    count = ARM_SME2_STATE_COUNT;
+    kret = thread_get_state(m_thread->MachPortNumber(), ARM_SME2_STATE,
+                            (thread_state_t)&m_state.context.sme.zt0, &count);
+    m_state.SetError(set, Read, kret);
+    DNBLogThreadedIf(LOG_THREAD, "Read ARM_SME2_STATE return value %d", kret);
+    if (kret != KERN_SUCCESS)
+      return kret;
+  }
+
+  return kret;
+}
+
 kern_return_t DNBArchMachARM64::SetGPRState() {
   int set = e_regSetGPR;
   kern_return_t kret = ::thread_set_state(
@@ -441,6 +592,80 @@ kern_return_t DNBArchMachARM64::SetVFPState() {
   return kret;                             // Return the error code
 }
 
+kern_return_t DNBArchMachARM64::SetSVEState() {
+  if (!CPUHasSME())
+    return KERN_INVALID_ARGUMENT;
+
+  int set = e_regSetSVE;
+  kern_return_t kret = thread_set_state(
+      m_thread->MachPortNumber(), ARM_SVE_Z_STATE1,
+      (thread_state_t)&m_state.context.sve.z[0], ARM_SVE_Z_STATE_COUNT);
+  m_state.SetError(set, Write, kret);
+  DNBLogThreadedIf(LOG_THREAD, "Write ARM_SVE_Z_STATE1 return value %d", kret);
+  if (kret != KERN_SUCCESS)
+    return kret;
+
+  kret = thread_set_state(m_thread->MachPortNumber(), ARM_SVE_Z_STATE2,
+                          (thread_state_t)&m_state.context.sve.z[16],
+                          ARM_SVE_Z_STATE_COUNT);
+  m_state.SetError(set, Write, kret);
+  DNBLogThreadedIf(LOG_THREAD, "Write ARM_SVE_Z_STATE2 return value %d", kret);
+  if (kret != KERN_SUCCESS)
+    return kret;
+
+  kret = thread_set_state(m_thread->MachPortNumber(), ARM_SVE_P_STATE,
+                          (thread_state_t)&m_state.context.sve.p[0],
+                          ARM_SVE_P_STATE_COUNT);
+  m_state.SetError(set, Write, kret);
+  DNBLogThreadedIf(LOG_THREAD, "Write ARM_SVE_P_STATE return value %d", kret);
+  if (kret != KERN_SUCCESS)
+    return kret;
+
+  return kret;
+}
+
+kern_return_t DNBArchMachARM64::SetSMEState() {
+  if (!CPUHasSME())
+    return KERN_INVALID_ARGUMENT;
+  kern_return_t kret;
+
+  int set = e_regSetSME;
+  size_t za_size = m_state.context.sme.svl_b * m_state.context.sme.svl_b;
+  const size_t max_chunk_size = 4096;
+  int n_chunks;
+  size_t chunk_size;
+  if (za_size <= max_chunk_size) {
+    n_chunks = 1;
+    chunk_size = za_size;
+  } else {
+    n_chunks = za_size / max_chunk_size;
+    chunk_size = max_chunk_size;
+  }
+  for (int i = 0; i < n_chunks; i++) {
+    arm_sme_za_state_t za_state;
+    memcpy(&za_state, m_state.context.sme.za.data() + (i * chunk_size),
+           chunk_size);
+    kret = thread_set_state(m_thread->MachPortNumber(), ARM_SME_ZA_STATE1 + i,
+                            (thread_state_t)&za_state, ARM_SME_ZA_STATE_COUNT);
+    m_state.SetError(set, Write, kret);
+    DNBLogThreadedIf(LOG_THREAD, "Write ARM_SME_STATE return value %d", kret);
+    if (kret != KERN_SUCCESS)
+      return kret;
+  }
+
+  if (CPUHasSME2()) {
+    kret = thread_set_state(m_thread->MachPortNumber(), ARM_SME2_STATE,
+                            (thread_state_t)&m_state.context.sme.zt0,
+                            ARM_SME2_STATE);
+    m_state.SetError(set, Write, kret);
+    DNBLogThreadedIf(LOG_THREAD, "Write ARM_SME2_STATE return value %d", kret);
+    if (kret != KERN_SUCCESS)
+      return kret;
+  }
+
+  return kret;
+}
+
 kern_return_t DNBArchMachARM64::SetEXCState() {
   int set = e_regSetEXC;
   kern_return_t kret = ::thread_set_state(
@@ -1531,6 +1756,59 @@ enum {
   vfp_d31
 };
 
+enum {
+  sve_z0,
+  sve_z1,
+  sve_z2,
+  sve_z3,
+  sve_z4,
+  sve_z5,
+  sve_z6,
+  sve_z7,
+  sve_z8,
+  sve_z9,
+  sve_z10,
+  sve_z11,
+  sve_z12,
+  sve_z13,
+  sve_z14,
+  sve_z15,
+  sve_z16,
+  sve_z17,
+  sve_z18,
+  sve_z19,
+  sve_z20,
+  sve_z21,
+  sve_z22,
+  sve_z23,
+  sve_z24,
+  sve_z25,
+  sve_z26,
+  sve_z27,
+  sve_z28,
+  sve_z29,
+  sve_z30,
+  sve_z31,
+  sve_p0,
+  sve_p1,
+  sve_p2,
+  sve_p3,
+  sve_p4,
+  sve_p5,
+  sve_p6,
+  sve_p7,
+  sve_p8,
+  sve_p9,
+  sve_p10,
+  sve_p11,
+  sve_p12,
+  sve_p13,
+  sve_p14,
+  sve_p15
+};
+
+enum { sme_svcr, sme_tpidr2, sme_svl_b, sme_za, sme_zt0 };
+
 enum { exc_far = 0, exc_esr, exc_exception };
 
 // These numbers from the "DWARF for the ARM 64-bit Architecture (AArch64)"
@@ -1681,7 +1959,60 @@ enum {
   debugserver_vfp_v30,
   debugserver_vfp_v31,
   debugserver_vfp_fpsr,
-  debugserver_vfp_fpcr
+  debugserver_vfp_fpcr,
+  debugserver_sve_z0,
+  debugserver_sve_z1,
+  debugserver_sve_z2,
+  debugserver_sve_z3,
+  debugserver_sve_z4,
+  debugserver_sve_z5,
+  debugserver_sve_z6,
+  debugserver_sve_z7,
+  debugserver_sve_z8,
+  debugserver_sve_z9,
+  debugserver_sve_z10,
+  debugserver_sve_z11,
+  debugserver_sve_z12,
+  debugserver_sve_z13,
+  debugserver_sve_z14,
+  debugserver_sve_z15,
+  debugserver_sve_z16,
+  debugserver_sve_z17,
+  debugserver_sve_z18,
+  debugserver_sve_z19,
+  debugserver_sve_z20,
+  debugserver_sve_z21,
+  debugserver_sve_z22,
+  debugserver_sve_z23,
+  debugserver_sve_z24,
+  debugserver_sve_z25,
+  debugserver_sve_z26,
+  debugserver_sve_z27,
+  debugserver_sve_z28,
+  debugserver_sve_z29,
+  debugserver_sve_z30,
+  debugserver_sve_z31,
+  debugserver_sve_p0,
+  debugserver_sve_p1,
+  debugserver_sve_p2,
+  debugserver_sve_p3,
+  debugserver_sve_p4,
+  debugserver_sve_p5,
+  debugserver_sve_p6,
+  debugserver_sve_p7,
+  debugserver_sve_p8,
+  debugserver_sve_p9,
+  debugserver_sve_p10,
+  debugserver_sve_p11,
+  debugserver_sve_p12,
+  debugserver_sve_p13,
+  debugserver_sve_p14,
+  debugserver_sve_p15,
+  debugserver_sme_svcr,
+  debugserver_sme_tpidr2,
+  debugserver_sme_svl_b,
+  debugserver_sme_za,
+  debugserver_sme_zt0
 };
 
 const char *g_contained_x0[]{"x0", NULL};
@@ -1906,38 +2237,74 @@ const char *g_contained_v29[]{"v29", NULL};
 const char *g_contained_v30[]{"v30", NULL};
 const char *g_contained_v31[]{"v31", NULL};
 
-const char *g_invalidate_v0[]{"v0", "d0", "s0", NULL};
-const char *g_invalidate_v1[]{"v1", "d1", "s1", NULL};
-const char *g_invalidate_v2[]{"v2", "d2", "s2", NULL};
-const char *g_invalidate_v3[]{"v3", "d3", "s3", NULL};
-const char *g_invalidate_v4[]{"v4", "d4", "s4", NULL};
-const char *g_invalidate_v5[]{"v5", "d5", "s5", NULL};
-const char *g_invalidate_v6[]{"v6", "d6", "s6", NULL};
-const char *g_invalidate_v7[]{"v7", "d7", "s7", NULL};
-const char *g_invalidate_v8[]{"v8", "d8", "s8", NULL};
-const char *g_invalidate_v9[]{"v9", "d9", "s9", NULL};
-const char *g_invalidate_v10[]{"v10", "d10", "s10", NULL};
-const char *g_invalidate_v11[]{"v11", "d11", "s11", NULL};
-const char *g_invalidate_v12[]{"v12", "d12", "s12", NULL};
-const char *g_invalidate_v13[]{"v13", "d13", "s13", NULL};
-const char *g_invalidate_v14[]{"v14", "d14", "s14", NULL};
-const char *g_invalidate_v15[]{"v15", "d15", "s15", NULL};
-const char *g_invalidate_v16[]{"v16", "d16", "s16", NULL};
-const char *g_invalidate_v17[]{"v17", "d17", "s17", NULL};
-const char *g_invalidate_v18[]{"v18", "d18", "s18", NULL};
-const char *g_invalidate_v19[]{"v19", "d19", "s19", NULL};
-const char *g_invalidate_v20[]{"v20", "d20", "s20", NULL};
-const char *g_invalidate_v21[]{"v21", "d21", "s21", NULL};
-const char *g_invalidate_v22[]{"v22", "d22", "s22", NULL};
-const char *g_invalidate_v23[]{"v23", "d23", "s23", NULL};
-const char *g_invalidate_v24[]{"v24", "d24", "s24", NULL};
-const char *g_invalidate_v25[]{"v25", "d25", "s25", NULL};
-const char *g_invalidate_v26[]{"v26", "d26", "s26", NULL};
-const char *g_invalidate_v27[]{"v27", "d27", "s27", NULL};
-const char *g_invalidate_v28[]{"v28", "d28", "s28", NULL};
-const char *g_invalidate_v29[]{"v29", "d29", "s29", NULL};
-const char *g_invalidate_v30[]{"v30", "d30", "s30", NULL};
-const char *g_invalidate_v31[]{"v31", "d31", "s31", NULL};
+const char *g_invalidate_v[32][4]{
+    {"v0", "d0", "s0", NULL},    {"v1", "d1", "s1", NULL},
+    {"v2", "d2", "s2", NULL},    {"v3", "d3", "s3", NULL},
+    {"v4", "d4", "s4", NULL},    {"v5", "d5", "s5", NULL},
+    {"v6", "d6", "s6", NULL},    {"v7", "d7", "s7", NULL},
+    {"v8", "d8", "s8", NULL},    {"v9", "d9", "s9", NULL},
+    {"v10", "d10", "s10", NULL}, {"v11", "d11", "s11", NULL},
+    {"v12", "d12", "s12", NULL}, {"v13", "d13", "s13", NULL},
+    {"v14", "d14", "s14", NULL}, {"v15", "d15", "s15", NULL},
+    {"v16", "d16", "s16", NULL}, {"v17", "d17", "s17", NULL},
+    {"v18", "d18", "s18", NULL}, {"v19", "d19", "s19", NULL},
+    {"v20", "d20", "s20", NULL}, {"v21", "d21", "s21", NULL},
+    {"v22", "d22", "s22", NULL}, {"v23", "d23", "s23", NULL},
+    {"v24", "d24", "s24", NULL}, {"v25", "d25", "s25", NULL},
+    {"v26", "d26", "s26", NULL}, {"v27", "d27", "s27", NULL},
+    {"v28", "d28", "s28", NULL}, {"v29", "d29", "s29", NULL},
+    {"v30", "d30", "s30", NULL}, {"v31", "d31", "s31", NULL}};
+
+const char *g_invalidate_z[32][5]{
+    {"z0", "v0", "d0", "s0", NULL},     {"z1", "v1", "d1", "s1", NULL},
+    {"z2", "v2", "d2", "s2", NULL},     {"z3", "v3", "d3", "s3", NULL},
+    {"z4", "v4", "d4", "s4", NULL},     {"z5", "v5", "d5", "s5", NULL},
+    {"z6", "v6", "d6", "s6", NULL},     {"z7", "v7", "d7", "s7", NULL},
+    {"z8", "v8", "d8", "s8", NULL},     {"z9", "v9", "d9", "s9", NULL},
+    {"z10", "v10", "d10", "s10", NULL}, {"z11", "v11", "d11", "s11", NULL},
+    {"z12", "v12", "d12", "s12", NULL}, {"z13", "v13", "d13", "s13", NULL},
+    {"z14", "v14", "d14", "s14", NULL}, {"z15", "v15", "d15", "s15", NULL},
+    {"z16", "v16", "d16", "s16", NULL}, {"z17", "v17", "d17", "s17", NULL},
+    {"z18", "v18", "d18", "s18", NULL}, {"z19", "v19", "d19", "s19", NULL},
+    {"z20", "v20", "d20", "s20", NULL}, {"z21", "v21", "d21", "s21", NULL},
+    {"z22", "v22", "d22", "s22", NULL}, {"z23", "v23", "d23", "s23", NULL},
+    {"z24", "v24", "d24", "s24", NULL}, {"z25", "v25", "d25", "s25", NULL},
+    {"z26", "v26", "d26", "s26", NULL}, {"z27", "v27", "d27", "s27", NULL},
+    {"z28", "v28", "d28", "s28", NULL}, {"z29", "v29", "d29", "s29", NULL},
+    {"z30", "v30", "d30", "s30", NULL}, {"z31", "v31", "d31", "s31", NULL}};
+
+const char *g_contained_z0[]{"z0", NULL};
+const char *g_contained_z1[]{"z1", NULL};
+const char *g_contained_z2[]{"z2", NULL};
+const char *g_contained_z3[]{"z3", NULL};
+const char *g_contained_z4[]{"z4", NULL};
+const char *g_contained_z5[]{"z5", NULL};
+const char *g_contained_z6[]{"z6", NULL};
+const char *g_contained_z7[]{"z7", NULL};
+const char *g_contained_z8[]{"z8", NULL};
+const char *g_contained_z9[]{"z9", NULL};
+const char *g_contained_z10[]{"z10", NULL};
+const char *g_contained_z11[]{"z11", NULL};
+const char *g_contained_z12[]{"z12", NULL};
+const char *g_contained_z13[]{"z13", NULL};
+const char *g_contained_z14[]{"z14", NULL};
+const char *g_contained_z15[]{"z15", NULL};
+const char *g_contained_z16[]{"z16", NULL};
+const char *g_contained_z17[]{"z17", NULL};
+const char *g_contained_z18[]{"z18", NULL};
+const char *g_contained_z19[]{"z19", NULL};
+const char *g_contained_z20[]{"z20", NULL};
+const char *g_contained_z21[]{"z21", NULL};
+const char *g_contained_z22[]{"z22", NULL};
+const char *g_contained_z23[]{"z23", NULL};
+const char *g_contained_z24[]{"z24", NULL};
+const char *g_contained_z25[]{"z25", NULL};
+const char *g_contained_z26[]{"z26", NULL};
+const char *g_contained_z27[]{"z27", NULL};
+const char *g_contained_z28[]{"z28", NULL};
+const char *g_contained_z29[]{"z29", NULL};
+const char *g_contained_z30[]{"z30", NULL};
+const char *g_contained_z31[]{"z31", NULL};
 
 #if defined(__arm64__) || defined(__aarch64__)
 #define VFP_V_OFFSET_IDX(idx)                                                  \
@@ -1948,141 +2315,18 @@ const char *g_invalidate_v31[]{"v31", "d31", "s31", NULL};
   (offsetof(DNBArchMachARM64::FPU, opaque) + (idx * 16) +                      \
    offsetof(DNBArchMachARM64::Context, vfp))
 #endif
-#define VFP_OFFSET_NAME(reg)                                                   \
-  (offsetof(DNBArchMachARM64::FPU, reg) +                                      \
-   offsetof(DNBArchMachARM64::Context, vfp))
 #define EXC_OFFSET(reg)                                                        \
   (offsetof(DNBArchMachARM64::EXC, reg) +                                      \
    offsetof(DNBArchMachARM64::Context, exc))
-
-//#define FLOAT_FORMAT Float
-#define DEFINE_VFP_V_IDX(idx)                                                  \
-  {                                                                            \
-    e_regSetVFP, vfp_v##idx, "v" #idx, "q" #idx, Vector, VectorOfUInt8, 16,    \
-        VFP_V_OFFSET_IDX(idx), INVALID_NUB_REGNUM, dwarf_v##idx,               \
-        INVALID_NUB_REGNUM, debugserver_vfp_v##idx, NULL, g_invalidate_v##idx  \
-  }
-#define DEFINE_PSEUDO_VFP_S_IDX(idx)                                           \
-  {                                                                            \
-    e_regSetVFP, vfp_s##idx, "s" #idx, NULL, IEEE754, Float, 4, 0,             \
-        INVALID_NUB_REGNUM, INVALID_NUB_REGNUM, INVALID_NUB_REGNUM,            \
-        INVALID_NUB_REGNUM, g_contained_v##idx, g_invalidate_v##idx            \
-  }
-#define DEFINE_PSEUDO_VFP_D_IDX(idx)                                           \
-  {                                                                            \
-    e_regSetVFP, vfp_d##idx, "d" #idx, NULL, IEEE754, Float, 8, 0,             \
-        INVALID_NUB_REGNUM, INVALID_NUB_REGNUM, INVALID_NUB_REGNUM,            \
-        INVALID_NUB_REGNUM, g_contained_v##idx, g_invalidate_v##idx            \
-  }
-
-// Floating point registers
-const DNBRegisterInfo DNBArchMachARM64::g_vfp_registers[] = {
-    DEFINE_VFP_V_IDX(0),
-    DEFINE_VFP_V_IDX(1),
-    DEFINE_VFP_V_IDX(2),
-    DEFINE_VFP_V_IDX(3),
-    DEFINE_VFP_V_IDX(4),
-    DEFINE_VFP_V_IDX(5),
-    DEFINE_VFP_V_IDX(6),
-    DEFINE_VFP_V_IDX(7),
-    DEFINE_VFP_V_IDX(8),
-    DEFINE_VFP_V_IDX(9),
-    DEFINE_VFP_V_IDX(10),
-    DEFINE_VFP_V_IDX(11),
-    DEFINE_VFP_V_IDX(12),
-    DEFINE_VFP_V_IDX(13),
-    DEFINE_VFP_V_IDX(14),
-    DEFINE_VFP_V_IDX(15),
-    DEFINE_VFP_V_IDX(16),
-    DEFINE_VFP_V_IDX(17),
-    DEFINE_VFP_V_IDX(18),
-    DEFINE_VFP_V_IDX(19),
-    DEFINE_VFP_V_IDX(20),
-    DEFINE_VFP_V_IDX(21),
-    DEFINE_VFP_V_IDX(22),
-    DEFINE_VFP_V_IDX(23),
-    DEFINE_VFP_V_IDX(24),
-    DEFINE_VFP_V_IDX(25),
-    DEFINE_VFP_V_IDX(26),
-    DEFINE_VFP_V_IDX(27),
-    DEFINE_VFP_V_IDX(28),
-    DEFINE_VFP_V_IDX(29),
-    DEFINE_VFP_V_IDX(30),
-    DEFINE_VFP_V_IDX(31),
-    {e_regSetVFP, vfp_fpsr, "fpsr", NULL, Uint, Hex, 4,
-     VFP_V_OFFSET_IDX(32) + 0, INVALID_NUB_REGNUM, INVALID_NUB_REGNUM,
-     INVALID_NUB_REGNUM, INVALID_NUB_REGNUM, NULL, NULL},
-    {e_regSetVFP, vfp_fpcr, "fpcr", NULL, Uint, Hex, 4,
-     VFP_V_OFFSET_IDX(32) + 4, INVALID_NUB_REGNUM, INVALID_NUB_REGNUM,
-     INVALID_NUB_REGNUM, INVALID_NUB_REGNUM, NULL, NULL},
-
-    DEFINE_PSEUDO_VFP_S_IDX(0),
-    DEFINE_PSEUDO_VFP_S_IDX(1),
-    DEFINE_PSEUDO_VFP_S_IDX(2),
-    DEFINE_PSEUDO_VFP_S_IDX(3),
-    DEFINE_PSEUDO_VFP_S_IDX(4),
-    DEFINE_PSEUDO_VFP_S_IDX(5),
-    DEFINE_PSEUDO_VFP_S_IDX(6),
-    DEFINE_PSEUDO_VFP_S_IDX(7),
-    DEFINE_PSEUDO_VFP_S_IDX(8),
-    DEFINE_PSEUDO_VFP_S_IDX(9),
-    DEFINE_PSEUDO_VFP_S_IDX(10),
-    DEFINE_PSEUDO_VFP_S_IDX(11),
-    DEFINE_PSEUDO_VFP_S_IDX(12),
-    DEFINE_PSEUDO_VFP_S_IDX(13),
-    DEFINE_PSEUDO_VFP_S_IDX(14),
-    DEFINE_PSEUDO_VFP_S_IDX(15),
-    DEFINE_PSEUDO_VFP_S_IDX(16),
-    DEFINE_PSEUDO_VFP_S_IDX(17),
-    DEFINE_PSEUDO_VFP_S_IDX(18),
-    DEFINE_PSEUDO_VFP_S_IDX(19),
-    DEFINE_PSEUDO_VFP_S_IDX(20),
-    DEFINE_PSEUDO_VFP_S_IDX(21),
-    DEFINE_PSEUDO_VFP_S_IDX(22),
-    DEFINE_PSEUDO_VFP_S_IDX(23),
-    DEFINE_PSEUDO_VFP_S_IDX(24),
-    DEFINE_PSEUDO_VFP_S_IDX(25),
-    DEFINE_PSEUDO_VFP_S_IDX(26),
-    DEFINE_PSEUDO_VFP_S_IDX(27),
-    DEFINE_PSEUDO_VFP_S_IDX(28),
-    DEFINE_PSEUDO_VFP_S_IDX(29),
-    DEFINE_PSEUDO_VFP_S_IDX(30),
-    DEFINE_PSEUDO_VFP_S_IDX(31),
-
-    DEFINE_PSEUDO_VFP_D_IDX(0),
-    DEFINE_PSEUDO_VFP_D_IDX(1),
-    DEFINE_PSEUDO_VFP_D_IDX(2),
-    DEFINE_PSEUDO_VFP_D_IDX(3),
-    DEFINE_PSEUDO_VFP_D_IDX(4),
-    DEFINE_PSEUDO_VFP_D_IDX(5),
-    DEFINE_PSEUDO_VFP_D_IDX(6),
-    DEFINE_PSEUDO_VFP_D_IDX(7),
-    DEFINE_PSEUDO_VFP_D_IDX(8),
-    DEFINE_PSEUDO_VFP_D_IDX(9),
-    DEFINE_PSEUDO_VFP_D_IDX(10),
-    DEFINE_PSEUDO_VFP_D_IDX(11),
-    DEFINE_PSEUDO_VFP_D_IDX(12),
-    DEFINE_PSEUDO_VFP_D_IDX(13),
-    DEFINE_PSEUDO_VFP_D_IDX(14),
-    DEFINE_PSEUDO_VFP_D_IDX(15),
-    DEFINE_PSEUDO_VFP_D_IDX(16),
-    DEFINE_PSEUDO_VFP_D_IDX(17),
-    DEFINE_PSEUDO_VFP_D_IDX(18),
-    DEFINE_PSEUDO_VFP_D_IDX(19),
-    DEFINE_PSEUDO_VFP_D_IDX(20),
-    DEFINE_PSEUDO_VFP_D_IDX(21),
-    DEFINE_PSEUDO_VFP_D_IDX(22),
-    DEFINE_PSEUDO_VFP_D_IDX(23),
-    DEFINE_PSEUDO_VFP_D_IDX(24),
-    DEFINE_PSEUDO_VFP_D_IDX(25),
-    DEFINE_PSEUDO_VFP_D_IDX(26),
-    DEFINE_PSEUDO_VFP_D_IDX(27),
-    DEFINE_PSEUDO_VFP_D_IDX(28),
-    DEFINE_PSEUDO_VFP_D_IDX(29),
-    DEFINE_PSEUDO_VFP_D_IDX(30),
-    DEFINE_PSEUDO_VFP_D_IDX(31)
-
-};
+#define SVE_OFFSET_Z_IDX(idx)                                                  \
+  (offsetof(DNBArchMachARM64::SVE, z[idx]) +                                   \
+   offsetof(DNBArchMachARM64::Context, sve))
+#define SVE_OFFSET_P_IDX(idx)                                                  \
+  (offsetof(DNBArchMachARM64::SVE, p[idx]) +                                   \
+   offsetof(DNBArchMachARM64::Context, sve))
+#define SME_OFFSET(reg)                                                        \
+  (offsetof(DNBArchMachARM64::SME, reg) +                                      \
+   offsetof(DNBArchMachARM64::Context, sme))
 
 //_STRUCT_ARM_EXCEPTION_STATE64
 //{
@@ -2106,29 +2350,220 @@ const DNBRegisterInfo DNBArchMachARM64::g_exc_registers[] = {
 // Number of registers in each register set
 const size_t DNBArchMachARM64::k_num_gpr_registers =
     sizeof(g_gpr_registers) / sizeof(DNBRegisterInfo);
-const size_t DNBArchMachARM64::k_num_vfp_registers =
-    sizeof(g_vfp_registers) / sizeof(DNBRegisterInfo);
 const size_t DNBArchMachARM64::k_num_exc_registers =
     sizeof(g_exc_registers) / sizeof(DNBRegisterInfo);
-const size_t DNBArchMachARM64::k_num_all_registers =
-    k_num_gpr_registers + k_num_vfp_registers + k_num_exc_registers;
-
-// Register set definitions. The first definitions at register set index
-// of zero is for all registers, followed by other registers sets. The
-// register information for the all register set need not be filled in.
-const DNBRegisterSetInfo DNBArchMachARM64::g_reg_sets[] = {
-    {"ARM64 Registers", NULL, k_num_all_registers},
-    {"General Purpose Registers", g_gpr_registers, k_num_gpr_registers},
-    {"Floating Point Registers", g_vfp_registers, k_num_vfp_registers},
-    {"Exception State Registers", g_exc_registers, k_num_exc_registers}};
-// Total number of register sets for this architecture
-const size_t DNBArchMachARM64::k_num_register_sets =
-    sizeof(g_reg_sets) / sizeof(DNBRegisterSetInfo);
 
+static std::vector<DNBRegisterInfo> g_sve_registers;
+static void initialize_sve_registers() {
+  static const char *g_z_regnames[32] = {
+      "z0",  "z1",  "z2",  "z3",  "z4",  "z5",  "z6",  "z7",
+      "z8",  "z9",  "z10", "z11", "z12", "z13", "z14", "z15",
+      "z16", "z17", "z18", "z19", "z20", "z21", "z22", "z23",
+      "z24", "z25", "z26", "z27", "z28", "z29", "z30", "z31"};
+  static const char *g_p_regnames[16] = {
+      "p0", "p1", "p2",  "p3",  "p4",  "p5",  "p6",  "p7",
+      "p8", "p9", "p10", "p11", "p12", "p13", "p14", "p15"};
+
+  if (DNBArchMachARM64::CPUHasSME()) {
+    uint32_t svl_bytes = DNBArchMachARM64::GetSMEMaxSVL();
+    for (uint32_t i = 0; i < 32; i++) {
+      g_sve_registers.push_back(
+          {DNBArchMachARM64::e_regSetSVE, (uint32_t)sve_z0 + i, g_z_regnames[i],
+           NULL, Vector, VectorOfUInt8, svl_bytes,
+           static_cast<uint32_t>(SVE_OFFSET_Z_IDX(i)), INVALID_NUB_REGNUM,
+           INVALID_NUB_REGNUM, INVALID_NUB_REGNUM,
+           (uint32_t)debugserver_sve_z0 + i, NULL, g_invalidate_z[i]});
+    }
+    for (uint32_t i = 0; i < 16; i++) {
+      g_sve_registers.push_back(
+          {DNBArchMachARM64::e_regSetSVE, (uint32_t)sve_p0 + i, g_p_regnames[i],
+           NULL, Vector, VectorOfUInt8, svl_bytes / 8,
+           (uint32_t)SVE_OFFSET_P_IDX(i), INVALID_NUB_REGNUM,
+           INVALID_NUB_REGNUM, INVALID_NUB_REGNUM,
+           (uint32_t)debugserver_sve_p0 + i, NULL, NULL});
+    }
+  }
+}
+
+static std::vector<DNBRegisterInfo> g_vfp_registers;
+static void initialize_vfp_registers() {
+  static const char *g_v_regnames[32] = {
+      "v0",  "v1",  "v2",  "v3",  "v4",  "v5",  "v6",  "v7",
+      "v8",  "v9",  "v10", "v11", "v12", "v13", "v14", "v15",
+      "v16", "v17", "v18", "v19", "v20", "v21", "v22", "v23",
+      "v24", "v25", "v26", "v27", "v28", "v29", "v30", "v31"};
+  static const char *g_q_regnames[32] = {
+      "q0",  "q1",  "q2",  "q3",  "q4",  "q5",  "q6",  "q7",
+      "q8",  "q9",  "q10", "q11", "q12", "q13", "q14", "q15",
+      "q16", "q17", "q18", "q19", "q20", "q21", "q22", "q23",
+      "q24", "q25", "q26", "q27", "q28", "q29", "q30", "q31"};
+
+  static const char *g_d_regnames[32] = {
+      "d0",  "d1",  "d2",  "d3",  "d4",  "d5",  "d6",  "d7",
+      "d8",  "d9",  "d10", "d11", "d12", "d13", "d14", "d15",
+      "d16", "d17", "d18", "d19", "d20", "d21", "d22", "d23",
+      "d24", "d25", "d26", "d27", "d28", "d29", "d30", "d31"};
+
+  static const char *g_s_regnames[32] = {
+      "s0",  "s1",  "s2",  "s3",  "s4",  "s5",  "s6",  "s7",
+      "s8",  "s9",  "s10", "s11", "s12", "s13", "s14", "s15",
+      "s16", "s17", "s18", "s19", "s20", "s21", "s22", "s23",
+      "s24", "s25", "s26", "s27", "s28", "s29", "s30", "s31"};
+
+  for (uint32_t i = 0; i < 32; i++)
+    if (DNBArchMachARM64::CPUHasSME())
+      g_vfp_registers.push_back(
+          {DNBArchMachARM64::e_regSetVFP, (uint32_t)vfp_v0 + i, g_v_regnames[i],
+           g_q_regnames[i], Vector, VectorOfUInt8, 16,
+           static_cast<uint32_t>(VFP_V_OFFSET_IDX(i)), INVALID_NUB_REGNUM,
+           (uint32_t)dwarf_v0 + i, INVALID_NUB_REGNUM,
+           (uint32_t)debugserver_vfp_v0 + i, NULL, g_invalidate_z[i]});
+    else
+      g_vfp_registers.push_back(
+          {DNBArchMachARM64::e_regSetVFP, (uint32_t)vfp_v0 + i, g_v_regnames[i],
+           g_q_regnames[i], Vector, VectorOfUInt8, 16,
+           static_cast<uint32_t>(VFP_V_OFFSET_IDX(i)), INVALID_NUB_REGNUM,
+           (uint32_t)dwarf_v0 + i, INVALID_NUB_REGNUM,
+           (uint32_t)debugserver_vfp_v0 + i, NULL, g_invalidate_v[i]});
+
+  g_vfp_registers.push_back(
+      {DNBArchMachARM64::e_regSetVFP, vfp_fpsr, "fpsr", NULL, Uint, Hex, 4,
+       VFP_V_OFFSET_IDX(32) + 0, INVALID_NUB_REGNUM, INVALID_NUB_REGNUM,
+       INVALID_NUB_REGNUM, INVALID_NUB_REGNUM, NULL, NULL});
+  g_vfp_registers.push_back(
+      {DNBArchMachARM64::e_regSetVFP, vfp_fpcr, "fpcr", NULL, Uint, Hex, 4,
+       VFP_V_OFFSET_IDX(32) + 4, INVALID_NUB_REGNUM, INVALID_NUB_REGNUM,
+       INVALID_NUB_REGNUM, INVALID_NUB_REGNUM, NULL, NULL});
+
+  for (uint32_t i = 0; i < 32; i++)
+    if (DNBArchMachARM64::CPUHasSME())
+      g_vfp_registers.push_back(
+          {DNBArchMachARM64::e_regSetVFP, (uint32_t)vfp_d0 + i, g_d_regnames[i],
+           NULL, IEEE754, Float, 8, 0, INVALID_NUB_REGNUM, INVALID_NUB_REGNUM,
+           INVALID_NUB_REGNUM, INVALID_NUB_REGNUM, NULL, g_invalidate_z[i]});
+    else
+      g_vfp_registers.push_back(
+          {DNBArchMachARM64::e_regSetVFP, (uint32_t)vfp_d0 + i, g_d_regnames[i],
+           NULL, IEEE754, Float, 8, 0, INVALID_NUB_REGNUM, INVALID_NUB_REGNUM,
+           INVALID_NUB_REGNUM, INVALID_NUB_REGNUM, NULL, g_invalidate_v[i]});
+
+  for (uint32_t i = 0; i < 32; i++)
+    if (DNBArchMachARM64::CPUHasSME())
+      g_vfp_registers.push_back(
+          {DNBArchMachARM64::e_regSetVFP, (uint32_t)vfp_s0 + i, g_s_regnames[i],
+           NULL, IEEE754, Float, 4, 0, INVALID_NUB_REGNUM, INVALID_NUB_REGNUM,
+           INVALID_NUB_REGNUM, INVALID_NUB_REGNUM, NULL, g_invalidate_z[i]});
+    else
+      g_vfp_registers.push_back(
+          {DNBArchMachARM64::e_regSetVFP, (uint32_t)vfp_s0 + i, g_s_regnames[i],
+           NULL, IEEE754, Float, 4, 0, INVALID_NUB_REGNUM, INVALID_NUB_REGNUM,
+           INVALID_NUB_REGNUM, INVALID_NUB_REGNUM, NULL, g_invalidate_v[i]});
+}
+
+static std::once_flag g_vfp_once;
+DNBRegisterInfo *
+DNBArchMachARM64::get_vfp_registerinfo(size_t &num_vfp_registers) {
+  std::call_once(g_vfp_once, []() { initialize_vfp_registers(); });
+  num_vfp_registers = g_vfp_registers.size();
+  if (num_vfp_registers > 0)
+    return g_vfp_registers.data();
+  else
+    return nullptr;
+}
+
+static std::once_flag g_sve_once;
+DNBRegisterInfo *
+DNBArchMachARM64::get_sve_registerinfo(size_t &num_sve_registers) {
+  std::call_once(g_sve_once, []() { initialize_sve_registers(); });
+  num_sve_registers = g_sve_registers.size();
+  if (num_sve_registers > 0)
+    return g_sve_registers.data();
+  else
+    return nullptr;
+}
+
+static std::vector<DNBRegisterInfo> g_sme_registers;
+static void initialize_sme_registers() {
+  if (DNBArchMachARM64::CPUHasSME()) {
+    uint32_t svl_bytes = DNBArchMachARM64::GetSMEMaxSVL();
+    g_sme_registers.push_back(
+        {DNBArchMachARM64::e_regSetSME, sme_svcr, "svcr", NULL, Uint, Hex, 8,
+         SME_OFFSET(svcr), INVALID_NUB_REGNUM, INVALID_NUB_REGNUM,
+         INVALID_NUB_REGNUM, INVALID_NUB_REGNUM, NULL, NULL});
+    g_sme_registers.push_back(
+        {DNBArchMachARM64::e_regSetSME, sme_tpidr2, "tpidr2", NULL, Uint, Hex,
+         8, SME_OFFSET(tpidr2), INVALID_NUB_REGNUM, INVALID_NUB_REGNUM,
+         INVALID_NUB_REGNUM, INVALID_NUB_REGNUM, NULL, NULL});
+    g_sme_registers.push_back(
+        {DNBArchMachARM64::e_regSetSME, sme_svl_b, "svl", NULL, Uint, Hex, 2,
+         SME_OFFSET(svl_b), INVALID_NUB_REGNUM, INVALID_NUB_REGNUM,
+         INVALID_NUB_REGNUM, INVALID_NUB_REGNUM, NULL, NULL});
+    uint32_t za_max_size = svl_bytes * svl_bytes;
+    g_sme_registers.push_back({DNBArchMachARM64::e_regSetSME, sme_za, "za",
+                               NULL, Vector, VectorOfUInt8, za_max_size,
+                               SME_OFFSET(za), INVALID_NUB_REGNUM,
+                               INVALID_NUB_REGNUM, INVALID_NUB_REGNUM,
+                               INVALID_NUB_REGNUM, NULL, NULL});
+  }
+  if (DNBArchMachARM64::CPUHasSME2()) {
+    g_sme_registers.push_back({DNBArchMachARM64::e_regSetSME, sme_zt0, "zt0",
+                               NULL, Vector, VectorOfUInt8, 64, SME_OFFSET(zt0),
+                               INVALID_NUB_REGNUM, INVALID_NUB_REGNUM,
+                               INVALID_NUB_REGNUM, INVALID_NUB_REGNUM, NULL,
+                               NULL});
+  }
+}
+
+static std::once_flag g_sme_once;
+DNBRegisterInfo *
+DNBArchMachARM64::get_sme_registerinfo(size_t &num_sme_registers) {
+  std::call_once(g_sme_once, []() { initialize_sme_registers(); });
+  num_sme_registers = g_sme_registers.size();
+  if (num_sme_registers > 0)
+    return g_sme_registers.data();
+  else
+    return nullptr;
+}
+
+static std::vector<DNBRegisterSetInfo> g_reg_sets;
+void DNBArchMachARM64::initialize_reg_sets() {
+  nub_size_t num_all_registers = DNBArchMachARM64::k_num_gpr_registers +
+                                 DNBArchMachARM64::k_num_exc_registers;
+  size_t num_vfp_registers = 0;
+  DNBRegisterInfo *vfp_reginfos =
+      DNBArchMachARM64::get_vfp_registerinfo(num_vfp_registers);
+  size_t num_sve_registers = 0;
+  DNBRegisterInfo *sve_reginfos =
+      DNBArchMachARM64::get_sve_registerinfo(num_sve_registers);
+  size_t num_sme_registers = 0;
+  DNBRegisterInfo *sme_reginfos =
+      DNBArchMachARM64::get_sme_registerinfo(num_sme_registers);
+  num_all_registers +=
+      num_vfp_registers + num_sve_registers + num_sme_registers;
+  g_reg_sets.push_back({"ARM64 Registers", NULL, num_all_registers});
+  g_reg_sets.push_back({"General Purpose Registers",
+                        DNBArchMachARM64::g_gpr_registers,
+                        DNBArchMachARM64::k_num_gpr_registers});
+  g_reg_sets.push_back(
+      {"Floating Point Registers", vfp_reginfos, num_vfp_registers});
+  g_reg_sets.push_back({"Exception State Registers",
+                        DNBArchMachARM64::g_exc_registers,
+                        DNBArchMachARM64::k_num_exc_registers});
+  if (DNBArchMachARM64::CPUHasSME()) {
+    g_reg_sets.push_back({"Scalable Vector Extension Registers", sve_reginfos,
+                          num_sve_registers});
+    g_reg_sets.push_back({"Scalable Matrix Extension Registers", sme_reginfos,
+                          num_sme_registers});
+  }
+}
+
+static std::once_flag g_initialize_register_set_info;
 const DNBRegisterSetInfo *
 DNBArchMachARM64::GetRegisterSetInfo(nub_size_t *num_reg_sets) {
-  *num_reg_sets = k_num_register_sets;
-  return g_reg_sets;
+  std::call_once(g_initialize_register_set_info,
+                 []() { initialize_reg_sets(); });
+  *num_reg_sets = g_reg_sets.size();
+  return g_reg_sets.data();
 }
 
 bool DNBArchMachARM64::FixGenericRegisterNumber(uint32_t &set, uint32_t &reg) {
@@ -2185,6 +2620,7 @@ bool DNBArchMachARM64::GetRegisterValue(uint32_t set, uint32_t reg,
 
   const DNBRegisterInfo *regInfo = m_thread->GetRegisterInfo(set, reg);
   if (regInfo) {
+    uint16_t max_svl_bytes = GetSMEMaxSVL();
     value->info = *regInfo;
     switch (set) {
     case e_regSetGPR:
@@ -2281,6 +2717,46 @@ bool DNBArchMachARM64::GetRegisterValue(uint32_t set, uint32_t reg,
       }
       break;
 
+    case e_regSetSVE:
+      if (GetRegisterState(e_regSetSVE, false) != KERN_SUCCESS)
+        return false;
+
+      if (reg >= sve_z0 && reg <= sve_z31) {
+        memset(&value->value.v_uint8, 0, max_svl_bytes);
+        memcpy(&value->value.v_uint8, &m_state.context.sve.z[reg - sve_z0],
+               max_svl_bytes);
+        return true;
+      } else if (reg >= sve_p0 && reg <= sve_p15) {
+        memset(&value->value.v_uint8, 0, max_svl_bytes / 8);
+        memcpy(&value->value.v_uint8, &m_state.context.sve.p[reg - sve_p0],
+               max_svl_bytes / 8);
+        return true;
+      }
+      break;
+
+    case e_regSetSME:
+      if (GetRegisterState(e_regSetSME, false) != KERN_SUCCESS)
+        return false;
+
+      if (reg == sme_svcr) {
+        value->value.uint64 = m_state.context.sme.svcr;
+        return true;
+      } else if (reg == sme_tpidr2) {
+        value->value.uint64 = m_state.context.sme.tpidr2;
+        return true;
+      } else if (reg == sme_svl_b) {
+        value->value.uint64 = m_state.context.sme.svl_b;
+        return true;
+      } else if (reg == sme_za) {
+        memcpy(&value->value.v_uint8, m_state.context.sme.za.data(),
+               max_svl_bytes * max_svl_bytes);
+        return true;
+      } else if (reg == sme_zt0) {
+        memcpy(&value->value.v_uint8, &m_state.context.sme.zt0, 64);
+        return true;
+      }
+      break;
+
     case e_regSetEXC:
       if (reg == exc_far) {
         value->value.uint64 = m_state.context.exc.__far;
@@ -2387,6 +2863,37 @@ bool DNBArchMachARM64::SetRegisterValue(uint32_t set, uint32_t reg,
       }
       break;
 
+    case e_regSetSVE:
+      if (reg >= sve_z0 && reg <= sve_z31) {
+        uint16_t max_svl_bytes = GetSMEMaxSVL();
+        memcpy(&m_state.context.sve.z[reg - sve_z0], &value->value.v_uint8,
+               max_svl_bytes);
+        success = true;
+      }
+      if (reg >= sve_p0 && reg <= sve_p15) {
+        uint16_t max_svl_bytes = GetSMEMaxSVL();
+        memcpy(&m_state.context.sve.p[reg - sve_p0], &value->value.v_uint8,
+               max_svl_bytes / 8);
+        success = true;
+      }
+      break;
+
+    case e_regSetSME:
+      // Cannot change ARM_SME_STATE registers with thread_set_state
+      if (reg == sme_svcr || reg == sme_tpidr2 || reg == sme_svl_b)
+        return false;
+      if (reg == sme_za) {
+        uint16_t max_svl_bytes = GetSMEMaxSVL();
+        memcpy(m_state.context.sme.za.data(), &value->value.v_uint8,
+               max_svl_bytes * max_svl_bytes);
+        success = true;
+      }
+      if (reg == sme_zt0) {
+        memcpy(&m_state.context.sme.zt0, &value->value.v_uint8, 64);
+        success = true;
+      }
+      break;
+
     case e_regSetEXC:
       if (reg == exc_far) {
         m_state.context.exc.__far = value->value.uint64;
@@ -2408,13 +2915,26 @@ bool DNBArchMachARM64::SetRegisterValue(uint32_t set, uint32_t reg,
 
 kern_return_t DNBArchMachARM64::GetRegisterState(int set, bool force) {
   switch (set) {
-  case e_regSetALL:
-    return GetGPRState(force) | GetVFPState(force) | GetEXCState(force) |
-           GetDBGState(force);
+  case e_regSetALL: {
+    kern_return_t retval = GetGPRState(force) | GetVFPState(force) |
+                           GetEXCState(force) | GetDBGState(force);
+    // If the processor is not in Streaming SVE Mode currently, these
+    // two will fail to read.  Don't return that as an error, it will
+    // be the most common case.
+    if (CPUHasSME()) {
+      GetSVEState(force);
+      GetSMEState(force);
+    }
+    return retval;
+  }
   case e_regSetGPR:
     return GetGPRState(force);
   case e_regSetVFP:
     return GetVFPState(force);
+  case e_regSetSVE:
+    return GetSVEState(force);
+  case e_regSetSME:
+    return GetSMEState(force);
   case e_regSetEXC:
     return GetEXCState(force);
   case e_regSetDBG:
@@ -2438,6 +2958,10 @@ kern_return_t DNBArchMachARM64::SetRegisterState(int set) {
     return SetGPRState();
   case e_regSetVFP:
     return SetVFPState();
+  case e_regSetSVE:
+    return SetSVEState();
+  case e_regSetSME:
+    return SetSMEState();
   case e_regSetEXC:
     return SetEXCState();
   case e_regSetDBG:
@@ -2455,6 +2979,15 @@ bool DNBArchMachARM64::RegisterSetStateIsValid(int set) const {
 nub_size_t DNBArchMachARM64::GetRegisterContext(void *buf, nub_size_t buf_len) {
   nub_size_t size = sizeof(m_state.context.gpr) + sizeof(m_state.context.vfp) +
                     sizeof(m_state.context.exc);
+  const bool cpu_has_sme = CPUHasSME();
+  if (cpu_has_sme) {
+    size += sizeof(m_state.context.sve);
+    // ZA register is in a std::vector<uint8_t> so we need to add
+    // the sizes of the SME manually.
+    size += ARM_SME_STATE_COUNT * sizeof(uint32_t);
+    size += m_state.context.sme.za.size();
+    size += ARM_SME2_STATE_COUNT * sizeof(uint32_t);
+  }
 
   if (buf && buf_len) {
     if (size > buf_len)
@@ -2463,6 +2996,13 @@ nub_size_t DNBArchMachARM64::GetRegisterContext(void *buf, nub_size_t buf_len) {
     bool force = false;
     if (GetGPRState(force) | GetVFPState(force) | GetEXCState(force))
       return 0;
+    // Don't error out if SME/SVE fail to read. These can only be read
+    // when the process is in Streaming SVE Mode, so the failure to read
+    // them will be common.
+    if (cpu_has_sme) {
+      GetSVEState(force);
+      GetSMEState(force);
+    }
 
     // Copy each struct individually to avoid any padding that might be between
     // the structs in m_state.context
@@ -2471,6 +3011,21 @@ nub_size_t DNBArchMachARM64::GetRegisterContext(void *buf, nub_size_t buf_len) {
     p += sizeof(m_state.context.gpr);
     ::memcpy(p, &m_state.context.vfp, sizeof(m_state.context.vfp));
     p += sizeof(m_state.context.vfp);
+    if (cpu_has_sme) {
+      ::memcpy(p, &m_state.context.sve, sizeof(m_state.context.sve));
+      p += sizeof(m_state.context.sve);
+
+      memcpy(p, &m_state.context.sme.svcr,
+             ARM_SME_STATE_COUNT * sizeof(uint32_t));
+      p += ARM_SME_STATE_COUNT * sizeof(uint32_t);
+      memcpy(p, m_state.context.sme.za.data(), m_state.context.sme.za.size());
+      p += m_state.context.sme.za.size();
+      if (CPUHasSME2()) {
+        memcpy(p, &m_state.context.sme.zt0,
+               ARM_SME2_STATE_COUNT * sizeof(uint32_t));
+        p += ARM_SME2_STATE_COUNT * sizeof(uint32_t);
+      }
+    }
     ::memcpy(p, &m_state.context.exc, sizeof(m_state.context.exc));
     p += sizeof(m_state.context.exc);
 
@@ -2490,6 +3045,15 @@ nub_size_t DNBArchMachARM64::SetRegisterContext(const void *buf,
                                                 nub_size_t buf_len) {
   nub_size_t size = sizeof(m_state.context.gpr) + sizeof(m_state.context.vfp) +
                     sizeof(m_state.context.exc);
+  if (CPUHasSME()) {
+    // m_state.context.za is three status registers, then a std::vector<uint8_t>
+    // for ZA, then zt0, so the size of the data is not statically knowable.
+    nub_size_t sme_size = ARM_SME_STATE_COUNT * sizeof(uint32_t);
+    sme_size += m_state.context.sme.za.size();
+    sme_size += ARM_SME2_STATE_COUNT * sizeof(uint32_t);
+
+    size += sizeof(m_state.context.sve) + sme_size;
+  }
 
   if (buf == NULL || buf_len == 0)
     size = 0;
@@ -2505,6 +3069,20 @@ nub_size_t DNBArchMachARM64::SetRegisterContext(const void *buf,
     p += sizeof(m_state.context.gpr);
     ::memcpy(&m_state.context.vfp, p, sizeof(m_state.context.vfp));
     p += sizeof(m_state.context.vfp);
+    if (CPUHasSME()) {
+      memcpy(&m_state.context.sve, p, sizeof(m_state.context.sve));
+      p += sizeof(m_state.context.sve);
+      memcpy(&m_state.context.sme.svcr, p,
+             ARM_SME_STATE_COUNT * sizeof(uint32_t));
+      p += ARM_SME_STATE_COUNT * sizeof(uint32_t);
+      memcpy(m_state.context.sme.za.data(), p, m_state.context.sme.za.size());
+      p += m_state.context.sme.za.size();
+      if (CPUHasSME2()) {
+        memcpy(&m_state.context.sme.zt0, p,
+               ARM_SME2_STATE_COUNT * sizeof(uint32_t));
+        p += ARM_SME2_STATE_COUNT * sizeof(uint32_t);
+      }
+    }
     ::memcpy(&m_state.context.exc, p, sizeof(m_state.context.exc));
     p += sizeof(m_state.context.exc);
 
@@ -2513,6 +3091,10 @@ nub_size_t DNBArchMachARM64::SetRegisterContext(const void *buf,
     assert(bytes_written == size);
     SetGPRState();
     SetVFPState();
+    if (CPUHasSME()) {
+      SetSVEState();
+      SetSMEState();
+    }
     SetEXCState();
   }
   DNBLogThreadedIf(
diff --git a/lldb/tools/debugserver/source/MacOSX/arm64/DNBArchImplARM64.h b/lldb/tools/debugserver/source/MacOSX/arm64/DNBArchImplARM64.h
index 0ea33d8e1c4c..11ad1f40c3ef 100644
--- a/lldb/tools/debugserver/source/MacOSX/arm64/DNBArchImplARM64.h
+++ b/lldb/tools/debugserver/source/MacOSX/arm64/DNBArchImplARM64.h
@@ -15,6 +15,10 @@
 #include <map>
 #include <vector>
 
+#if !defined(ARM_SME_STATE)
+#include "sme_thread_status.h"
+#endif
+
 #if defined(ARM_THREAD_STATE64_COUNT)
 
 #include "DNBArch.h"
@@ -93,7 +97,6 @@ public:
   bool DisableHardwareWatchpoint_helper(uint32_t hw_break_index,
                                         bool also_set_on_task);
 
-protected:
   kern_return_t EnableHardwareSingleStep(bool enable);
   static bool FixGenericRegisterNumber(uint32_t &set, uint32_t &reg);
 
@@ -102,6 +105,8 @@ protected:
     e_regSetGPR, // ARM_THREAD_STATE64,
     e_regSetVFP, // ARM_NEON_STATE64,
     e_regSetEXC, // ARM_EXCEPTION_STATE64,
+    e_regSetSVE, // ARM_SVE_Z_STATE1, ARM_SVE_Z_STATE2, ARM_SVE_P_STATE
+    e_regSetSME, // ARM_SME_STATE, ARM_SME_ZA_STATE1..16, ARM_SME2_STATE
     e_regSetDBG, // ARM_DEBUG_STATE64,
     kNumRegisterSets
   };
@@ -119,20 +124,39 @@ protected:
   typedef arm_neon_state64_t FPU;
   typedef arm_exception_state64_t EXC;
 
+  struct SVE {
+    uint8_t z[32][256]; // arm_sve_z_state_t z[2]
+    uint8_t p[16][32];  // arm_sve_p_state_t p
+  };
+
+  struct SME {
+    uint64_t svcr;   // arm_sme_state_t
+    uint64_t tpidr2; // arm_sme_state_t
+    uint16_t svl_b;  // arm_sme_state_t
+
+    std::vector<uint8_t> za;
+    uint8_t zt0[64];
+
+    SME() {
+      if (DNBArchMachARM64::CPUHasSME()) {
+        int svl = GetSMEMaxSVL();
+        za.resize(svl * svl, 0);
+      }
+    }
+  };
+
   static const DNBRegisterInfo g_gpr_registers[];
-  static const DNBRegisterInfo g_vfp_registers[];
   static const DNBRegisterInfo g_exc_registers[];
-  static const DNBRegisterSetInfo g_reg_sets[];
 
   static const size_t k_num_gpr_registers;
-  static const size_t k_num_vfp_registers;
   static const size_t k_num_exc_registers;
   static const size_t k_num_all_registers;
-  static const size_t k_num_register_sets;
 
   struct Context {
     GPR gpr;
     FPU vfp;
+    SVE sve;
+    SME sme;
     EXC exc;
   };
 
@@ -141,6 +165,8 @@ protected:
     arm_debug_state64_t dbg;
     kern_return_t gpr_errs[2]; // Read/Write errors
     kern_return_t vfp_errs[2]; // Read/Write errors
+    kern_return_t sve_errs[2]; // Read/Write errors
+    kern_return_t sme_errs[2]; // Read/Write errors
     kern_return_t exc_errs[2]; // Read/Write errors
     kern_return_t dbg_errs[2]; // Read/Write errors
     State() {
@@ -148,6 +174,8 @@ protected:
       for (i = 0; i < kNumErrors; i++) {
         gpr_errs[i] = -1;
         vfp_errs[i] = -1;
+        sve_errs[i] = -1;
+        sme_errs[i] = -1;
         exc_errs[i] = -1;
         dbg_errs[i] = -1;
       }
@@ -163,11 +191,15 @@ protected:
         // we got any kind of error.
         case e_regSetALL:
           return gpr_errs[err_idx] | vfp_errs[err_idx] | exc_errs[err_idx] |
-                 dbg_errs[err_idx];
+                 sve_errs[err_idx] | sme_errs[err_idx] | dbg_errs[err_idx];
         case e_regSetGPR:
           return gpr_errs[err_idx];
         case e_regSetVFP:
           return vfp_errs[err_idx];
+        case e_regSetSVE:
+          return sve_errs[err_idx];
+        case e_regSetSME:
+          return sme_errs[err_idx];
         case e_regSetEXC:
           return exc_errs[err_idx];
         // case e_regSetDBG:   return dbg_errs[err_idx];
@@ -183,6 +215,8 @@ protected:
         case e_regSetALL:
           gpr_errs[err_idx] = err;
           vfp_errs[err_idx] = err;
+          sve_errs[err_idx] = err;
+          sme_errs[err_idx] = err;
           dbg_errs[err_idx] = err;
           exc_errs[err_idx] = err;
           return true;
@@ -195,6 +229,14 @@ protected:
           vfp_errs[err_idx] = err;
           return true;
 
+        case e_regSetSVE:
+          sve_errs[err_idx] = err;
+          return true;
+
+        case e_regSetSME:
+          sme_errs[err_idx] = err;
+          return true;
+
         case e_regSetEXC:
           exc_errs[err_idx] = err;
           return true;
@@ -215,11 +257,15 @@ protected:
 
   kern_return_t GetGPRState(bool force);
   kern_return_t GetVFPState(bool force);
+  kern_return_t GetSVEState(bool force);
+  kern_return_t GetSMEState(bool force);
   kern_return_t GetEXCState(bool force);
   kern_return_t GetDBGState(bool force);
 
   kern_return_t SetGPRState();
   kern_return_t SetVFPState();
+  kern_return_t SetSVEState();
+  kern_return_t SetSMEState();
   kern_return_t SetEXCState();
   kern_return_t SetDBGState(bool also_set_on_task);
 
@@ -246,7 +292,16 @@ protected:
     uint32_t control;
   };
 
-protected:
+  static bool CPUHasSME();
+  static bool CPUHasSME2();
+  static unsigned int GetSMEMaxSVL();
+
+private:
+  static DNBRegisterInfo *get_vfp_registerinfo(size_t &num_vfp_registers);
+  static DNBRegisterInfo *get_sve_registerinfo(size_t &num_sve_registers);
+  static DNBRegisterInfo *get_sme_registerinfo(size_t &num_sme_registers);
+  static void initialize_reg_sets();
+
   MachThread *m_thread;
   State m_state;
   arm_debug_state64_t m_dbg_save;
@@ -264,6 +319,9 @@ protected:
 
   typedef std::map<uint32_t, Context> SaveRegisterStates;
   SaveRegisterStates m_saved_register_states;
+
+  DNBArchMachARM64(const DNBArchMachARM64 &) = delete;
+  DNBArchMachARM64 &operator=(const DNBArchMachARM64 &) = delete;
 };
 
 #endif // #if defined (ARM_THREAD_STATE64_COUNT)
diff --git a/lldb/tools/debugserver/source/MacOSX/arm64/sme_thread_status.h b/lldb/tools/debugserver/source/MacOSX/arm64/sme_thread_status.h
new file mode 100644
index 000000000000..f33b3202ccab
--- /dev/null
+++ b/lldb/tools/debugserver/source/MacOSX/arm64/sme_thread_status.h
@@ -0,0 +1,77 @@
+#include <mach/mach.h>
+#include <stdint.h>
+
+// Define the SVE/SME/SME2 thread status structures
+// flavors, and sizes so this can build against an
+// older SDK which does not have these definitions
+// yet.
+
+#if !defined(ARM_SME_STATE)
+
+#define _STRUCT_ARM_SME_STATE struct arm_sme_state
+_STRUCT_ARM_SME_STATE {
+  uint64_t svcr;
+  uint64_t tpidr2_el0;
+  uint16_t svl_b;
+};
+
+#define _STRUCT_ARM_SVE_Z_STATE struct arm_sve_z_state
+_STRUCT_ARM_SVE_Z_STATE { char z[16][256]; }
+__attribute__((aligned(alignof(unsigned int))));
+
+#define _STRUCT_ARM_SVE_P_STATE struct arm_sve_p_state
+_STRUCT_ARM_SVE_P_STATE { char p[16][256 / 8]; }
+__attribute__((aligned(alignof(unsigned int))));
+
+#define _STRUCT_ARM_SME_ZA_STATE struct arm_sme_za_state
+_STRUCT_ARM_SME_ZA_STATE { char za[4096]; }
+__attribute__((aligned(alignof(unsigned int))));
+
+#define _STRUCT_ARM_SME2_STATE struct arm_sme2_state
+_STRUCT_ARM_SME2_STATE { char zt0[64]; }
+__attribute__((aligned(alignof(unsigned int))));
+
+#define ARM_SME_STATE 28
+#define ARM_SVE_Z_STATE1 29
+#define ARM_SVE_Z_STATE2 30
+#define ARM_SVE_P_STATE 31
+#define ARM_SME_ZA_STATE1 32
+#define ARM_SME_ZA_STATE2 33
+#define ARM_SME_ZA_STATE3 34
+#define ARM_SME_ZA_STATE4 35
+#define ARM_SME_ZA_STATE5 36
+#define ARM_SME_ZA_STATE6 37
+#define ARM_SME_ZA_STATE7 38
+#define ARM_SME_ZA_STATE8 39
+#define ARM_SME_ZA_STATE9 40
+#define ARM_SME_ZA_STATE10 41
+#define ARM_SME_ZA_STATE11 42
+#define ARM_SME_ZA_STATE12 42
+#define ARM_SME_ZA_STATE13 44
+#define ARM_SME_ZA_STATE14 45
+#define ARM_SME_ZA_STATE15 46
+#define ARM_SME_ZA_STATE16 47
+#define ARM_SME2_STATE 48
+
+typedef _STRUCT_ARM_SME_STATE arm_sme_state_t;
+typedef _STRUCT_ARM_SVE_Z_STATE arm_sve_z_state_t;
+typedef _STRUCT_ARM_SVE_P_STATE arm_sve_p_state_t;
+typedef _STRUCT_ARM_SME_ZA_STATE arm_sme_za_state_t;
+typedef _STRUCT_ARM_SME2_STATE arm_sme2_state_t;
+
+#define ARM_SME_STATE_COUNT                                                    \
+  ((mach_msg_type_number_t)(sizeof(arm_sme_state_t) / sizeof(uint32_t)))
+
+#define ARM_SVE_Z_STATE_COUNT                                                  \
+  ((mach_msg_type_number_t)(sizeof(arm_sve_z_state_t) / sizeof(uint32_t)))
+
+#define ARM_SVE_P_STATE_COUNT                                                  \
+  ((mach_msg_type_number_t)(sizeof(arm_sve_p_state_t) / sizeof(uint32_t)))
+
+#define ARM_SME_ZA_STATE_COUNT                                                 \
+  ((mach_msg_type_number_t)(sizeof(arm_sme_za_state_t) / sizeof(uint32_t)))
+
+#define ARM_SME2_STATE_COUNT                                                   \
+  ((mach_msg_type_number_t)(sizeof(arm_sme2_state_t) / sizeof(uint32_t)))
+
+#endif // !defined(ARM_SME_STATE)
diff --git a/lldb/tools/debugserver/source/RNBRemote.cpp b/lldb/tools/debugserver/source/RNBRemote.cpp
index 07211c6e9db4..efa015920c0d 100644
--- a/lldb/tools/debugserver/source/RNBRemote.cpp
+++ b/lldb/tools/debugserver/source/RNBRemote.cpp
@@ -22,6 +22,7 @@
 #include <mach/mach_vm.h>
 #include <mach/task_info.h>
 #include <pwd.h>
+#include <string>
 #include <sys/stat.h>
 #include <sys/sysctl.h>
 #include <unistd.h>
@@ -2567,42 +2568,50 @@ rnb_err_t RNBRemote::HandlePacket_QSetProcessEvent(const char *p) {
   return SendPacket("OK");
 }
 
-void register_value_in_hex_fixed_width(std::ostream &ostrm, nub_process_t pid,
+// If a fail_value is provided, a correct-length reply is always provided,
+// even if the register cannot be read right now on this thread.
+bool register_value_in_hex_fixed_width(std::ostream &ostrm, nub_process_t pid,
                                        nub_thread_t tid,
                                        const register_map_entry_t *reg,
-                                       const DNBRegisterValue *reg_value_ptr) {
+                                       const DNBRegisterValue *reg_value_ptr,
+                                       std::optional<uint8_t> fail_value) {
   if (reg != NULL) {
-    DNBRegisterValue reg_value;
+    std::unique_ptr<DNBRegisterValue> reg_value =
+        std::make_unique<DNBRegisterValue>();
     if (reg_value_ptr == NULL) {
       if (DNBThreadGetRegisterValueByID(pid, tid, reg->nub_info.set,
-                                        reg->nub_info.reg, &reg_value))
-        reg_value_ptr = &reg_value;
+                                        reg->nub_info.reg, reg_value.get()))
+        reg_value_ptr = reg_value.get();
     }
 
     if (reg_value_ptr) {
       append_hex_value(ostrm, reg_value_ptr->value.v_uint8, reg->nub_info.size,
                        false);
-    } else {
-      // If we fail to read a register value, check if it has a default
-      // fail value. If it does, return this instead in case some of
-      // the registers are not available on the current system.
-      if (reg->nub_info.size > 0) {
-        std::vector<uint8_t> zeros(reg->nub_info.size, '\0');
-        append_hex_value(ostrm, zeros.data(), zeros.size(), false);
-      }
+      return true;
     }
+    if (!fail_value || reg->nub_info.size == 0)
+      return false;
+
+    // Pad out the reply to the correct size to maintain correct offsets,
+    // even if we could not read the register value.
+    std::vector<uint8_t> fail_result(reg->nub_info.size, *fail_value);
+    append_hex_value(ostrm, fail_result.data(), fail_result.size(), false);
+    return true;
   }
+  return false;
 }
 
 void debugserver_regnum_with_fixed_width_hex_register_value(
     std::ostream &ostrm, nub_process_t pid, nub_thread_t tid,
-    const register_map_entry_t *reg, const DNBRegisterValue *reg_value_ptr) {
+    const register_map_entry_t *reg, const DNBRegisterValue *reg_value_ptr,
+    std::optional<uint8_t> fail_value) {
   // Output the register number as 'NN:VVVVVVVV;' where NN is a 2 bytes HEX
   // gdb register number, and VVVVVVVV is the correct number of hex bytes
   // as ASCII for the register value.
   if (reg != NULL) {
     ostrm << RAWHEX8(reg->debugserver_regnum) << ':';
-    register_value_in_hex_fixed_width(ostrm, pid, tid, reg, reg_value_ptr);
+    register_value_in_hex_fixed_width(ostrm, pid, tid, reg, reg_value_ptr,
+                                      fail_value);
     ostrm << ';';
   }
 }
@@ -2651,15 +2660,16 @@ typedef std::map<nub_addr_t, StackMemory> StackMemoryMap;
 static void ReadStackMemory(nub_process_t pid, nub_thread_t tid,
                             StackMemoryMap &stack_mmap,
                             uint32_t backtrace_limit = 256) {
-  DNBRegisterValue reg_value;
+  std::unique_ptr<DNBRegisterValue> reg_value =
+      std::make_unique<DNBRegisterValue>();
   if (DNBThreadGetRegisterValueByID(pid, tid, REGISTER_SET_GENERIC,
-                                    GENERIC_REGNUM_FP, &reg_value)) {
+                                    GENERIC_REGNUM_FP, reg_value.get())) {
     uint32_t frame_count = 0;
     uint64_t fp = 0;
-    if (reg_value.info.size == 4)
-      fp = reg_value.value.uint32;
+    if (reg_value->info.size == 4)
+      fp = reg_value->value.uint32;
     else
-      fp = reg_value.value.uint64;
+      fp = reg_value->value.uint64;
     while (fp != 0) {
       // Make sure we never recurse more than 256 times so we don't recurse too
       // far or
@@ -2667,7 +2677,7 @@ static void ReadStackMemory(nub_process_t pid, nub_thread_t tid,
       if (++frame_count > backtrace_limit)
         break;
 
-      const nub_size_t read_size = reg_value.info.size * 2;
+      const nub_size_t read_size = reg_value->info.size * 2;
       StackMemory stack_memory;
       stack_memory.length = read_size;
       if (DNBProcessMemoryRead(pid, fp, read_size, stack_memory.bytes) !=
@@ -2679,7 +2689,7 @@ static void ReadStackMemory(nub_process_t pid, nub_thread_t tid,
       // Put the entry into the cache
       stack_mmap[fp] = stack_memory;
       // Dereference the frame pointer to get to the previous frame pointer
-      if (reg_value.info.size == 4)
+      if (reg_value->info.size == 4)
         fp = ((uint32_t *)stack_memory.bytes)[0];
       else
         fp = ((uint64_t *)stack_memory.bytes)[0];
@@ -2842,31 +2852,35 @@ rnb_err_t RNBRemote::SendStopReplyPacketForThread(nub_thread_t tid) {
     if (g_num_reg_entries == 0)
       InitializeRegisters();
 
-    if (g_reg_entries != NULL) {
-      auto interesting_regset = [](int regset) -> bool {
-#if defined(__arm64__) || defined(__aarch64__)
-        // GPRs and exception registers, helpful for debugging
-        // from packet logs.
-        return regset == 1 || regset == 3;
-#else
-        return regset == 1;
-#endif
-      };
-
-      DNBRegisterValue reg_value;
-      for (uint32_t reg = 0; reg < g_num_reg_entries; reg++) {
-        // Expedite all registers in the first register set that aren't
-        // contained in other registers
-        if (interesting_regset(g_reg_entries[reg].nub_info.set) &&
-            g_reg_entries[reg].nub_info.value_regs == NULL) {
-          if (!DNBThreadGetRegisterValueByID(
-                  pid, tid, g_reg_entries[reg].nub_info.set,
-                  g_reg_entries[reg].nub_info.reg, &reg_value))
-            continue;
-
-          debugserver_regnum_with_fixed_width_hex_register_value(
-              ostrm, pid, tid, &g_reg_entries[reg], &reg_value);
-        }
+    nub_size_t num_reg_sets = 0;
+    const DNBRegisterSetInfo *reg_sets = DNBGetRegisterSetInfo(&num_reg_sets);
+
+    std::unique_ptr<DNBRegisterValue> reg_value =
+        std::make_unique<DNBRegisterValue>();
+    for (uint32_t reg = 0; reg < g_num_reg_entries; reg++) {
+      int regset = g_reg_entries[reg].nub_info.set;
+      bool include_reg = false;
+      // Expedite interesting register sets, all registers not
+      // contained in other registers
+      if (g_reg_entries[reg].nub_info.value_regs == nullptr &&
+          (strcmp("General Purpose Registers", reg_sets[regset].name) == 0 ||
+           strcmp("Exception State Registers", reg_sets[regset].name) == 0))
+        include_reg = true;
+      // Include the SME state registers
+      if (strcmp("svcr", g_reg_entries[reg].nub_info.name) == 0 ||
+          strcmp("tpidr2", g_reg_entries[reg].nub_info.name) == 0 ||
+          strcmp("svl", g_reg_entries[reg].nub_info.name) == 0)
+        include_reg = true;
+
+      if (include_reg) {
+        if (!DNBThreadGetRegisterValueByID(pid, tid, regset,
+                                           g_reg_entries[reg].nub_info.reg,
+                                           reg_value.get()))
+          continue;
+
+        debugserver_regnum_with_fixed_width_hex_register_value(
+            ostrm, pid, tid, &g_reg_entries[reg], reg_value.get(),
+            std::nullopt);
       }
     }
 
@@ -3326,14 +3340,19 @@ rnb_err_t RNBRemote::HandlePacket_G(const char *p) {
   if (g_num_reg_entries == 0)
     InitializeRegisters();
 
-  StdStringExtractor packet(p);
-  packet.SetFilePos(1); // Skip the 'G'
+  p += 1; // Skip the 'G'
 
   nub_process_t pid = m_ctx.ProcessID();
   nub_thread_t tid = ExtractThreadIDFromThreadSuffix(p);
   if (tid == INVALID_NUB_THREAD)
     return HandlePacket_ILLFORMED(__FILE__, __LINE__, p,
                                   "No thread specified in p packet");
+  // Skip the thread specification in `G;thread:3488ea;[..data...]`
+  const char *last_semi = strrchr(p, ';');
+  if (last_semi)
+    p = last_semi + 1;
+
+  StdStringExtractor packet(p);
 
   // Get the register context size first by calling with NULL buffer
   nub_size_t reg_ctx_size = DNBThreadGetRegisterContext(pid, tid, NULL, 0);
@@ -4212,7 +4231,9 @@ rnb_err_t RNBRemote::HandlePacket_p(const char *p) {
       append_hex_value(ostrm, zeros.data(), zeros.size(), false);
     }
   } else {
-    register_value_in_hex_fixed_width(ostrm, pid, tid, reg_entry, NULL);
+    if (!register_value_in_hex_fixed_width(ostrm, pid, tid, reg_entry, NULL,
+                                           std::nullopt))
+      return SendErrorPacket("E97");
   }
   return SendPacket(ostrm.str());
 }
@@ -4266,9 +4287,10 @@ rnb_err_t RNBRemote::HandlePacket_P(const char *p) {
     return SendErrorPacket("E48");
   }
 
-  DNBRegisterValue reg_value;
-  reg_value.info = reg_entry->nub_info;
-  packet.GetHexBytes(reg_value.value.v_sint8, reg_entry->nub_info.size, 0xcc);
+  std::unique_ptr<DNBRegisterValue> reg_value =
+      std::make_unique<DNBRegisterValue>();
+  reg_value->info = reg_entry->nub_info;
+  packet.GetHexBytes(reg_value->value.v_sint8, reg_entry->nub_info.size, 0xcc);
 
   nub_thread_t tid = ExtractThreadIDFromThreadSuffix(p);
   if (tid == INVALID_NUB_THREAD)
@@ -4276,7 +4298,8 @@ rnb_err_t RNBRemote::HandlePacket_P(const char *p) {
                                   "No thread specified in p packet");
 
   if (!DNBThreadSetRegisterValueByID(pid, tid, reg_entry->nub_info.set,
-                                     reg_entry->nub_info.reg, &reg_value)) {
+                                     reg_entry->nub_info.reg,
+                                     reg_value.get())) {
     return SendErrorPacket("E32");
   }
   return SendPacket("OK");
@@ -5561,7 +5584,8 @@ RNBRemote::GetJSONThreadsInfo(bool threads_with_valid_stop_info_only) {
           }
         }
 
-        DNBRegisterValue reg_value;
+        std::unique_ptr<DNBRegisterValue> reg_value =
+            std::make_unique<DNBRegisterValue>();
 
         if (g_reg_entries != NULL) {
           JSONGenerator::DictionarySP registers_dict_sp(
@@ -5574,14 +5598,14 @@ RNBRemote::GetJSONThreadsInfo(bool threads_with_valid_stop_info_only) {
                 g_reg_entries[reg].nub_info.value_regs == NULL) {
               if (!DNBThreadGetRegisterValueByID(
                       pid, tid, g_reg_entries[reg].nub_info.set,
-                      g_reg_entries[reg].nub_info.reg, &reg_value))
+                      g_reg_entries[reg].nub_info.reg, reg_value.get()))
                 continue;
 
               std::ostringstream reg_num;
               reg_num << std::dec << g_reg_entries[reg].debugserver_regnum;
               // Encode native byte ordered bytes as hex ascii
               registers_dict_sp->AddBytesAsHexASCIIString(
-                  reg_num.str(), reg_value.value.v_uint8,
+                  reg_num.str(), reg_value->value.v_uint8,
                   g_reg_entries[reg].nub_info.size);
             }
           }