[Lldb-commits] [lldb] [lldb] Add summary for NSIndirectTaggedPointerString (PR #136025)

[Adrian Prantl via lldb-commits]

================
@@ -368,3 +373,37 @@ bool lldb_private::formatters::NSTaggedString_SummaryProvider(
   stream << suffix;
   return true;
 }
+
+bool lldb_private::formatters::NSIndirectTaggedString_SummaryProvider(
+    ValueObject &valobj, ObjCLanguageRuntime::ClassDescriptorSP descriptor,
+    Stream &stream, const TypeSummaryOptions &summary_options) {
+  if (!descriptor)
+    return false;
+
+  uint64_t payload = 0;
+  if (!descriptor->GetTaggedPointerInfo(nullptr, nullptr, &payload))
+    return false;
+
+  // First 47 bits are the address of the contents.
+  addr_t ptr = payload & 0x7fffffffffffULL;
+  // Next 13 bits are the string's length.
+  size_t size = (payload >> 47) & 0x1fff;
+
+  Status status;
+  std::vector<char> buf(size);
+  if (auto process_sp = valobj.GetProcessSP())
+    if (process_sp->ReadMemory(ptr, buf.data(), size, status)) {
+      llvm::StringRef prefix, suffix;
+      if (auto *language = Language::FindPlugin(summary_options.GetLanguage()))
+        std::tie(prefix, suffix) =
+            language->GetFormatterPrefixSuffix("NSString");
+      stream << prefix << '"';
+      stream.PutCString({buf.data(), size});
----------------
adrian-prantl wrote:

Does PutCString scan through the buffer until it finds a `NUL` byte or does it copy `size` bytes (which would be faster).

I doubt this affects the performance measurably. Just being curious.

https://github.com/llvm/llvm-project/pull/136025