[Lldb-commits] [lldb] [lldb][debugserver] Read/write SME registers on arm64 (PR #119171)

[David Spickett via lldb-commits]

================
@@ -93,6 +93,55 @@ 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;
+      else
+        g_sme_max_svl = get_svl_bytes();
+    }
+  });
+  return g_sme_max_svl;
+}
+
+// This function can only be called on systems with hw.optional.arm.FEAT_SME
+// It will return the maximum SVL length for this process.
+uint16_t __attribute__((target("sme"))) DNBArchMachARM64::get_svl_bytes(void) {
+  uint64_t ret = 0;
+  asm volatile("rdsvl	%[ret], #1" : [ret] "=r"(ret));
----------------
DavidSpickett wrote:

That would be my concern, older debug server - newer CPU with a wider SME unit.

You could leave it in as a last resort but I wonder if it would even be understandable for users. If the kernel returns the registers in a buffer end to end, and they're 128 byte but debugserver thinks they're 64 byte. It'll likely 1. fail to read at all or 2. Show the first ZA row as the first two ZA rows.

As opposed to what might be useful in some very niche scenario, where you could see the bottom few bytes of each register.

Not that SME/SVE in its raw form is that useful to begin with.

https://github.com/llvm/llvm-project/pull/119171