================
@@ -33,6 +33,37 @@ static void EmplaceSafeString(llvm::json::Object &obj,
llvm::StringRef key,
obj.try_emplace(key, llvm::json::fixUTF8(str));
}
+static void UpdateDwoFileCounts(SymbolFile *sym_file,
+ uint32_t &total_dwo_file_count,
+ uint32_t &total_loaded_dwo_file_count) {
+ // Count DWO files from this symbol file using GetSeparateDebugInfo
+ // For DWP files, this increments counts for both total and successfully
+ // loaded DWO CUs. For non split-dwarf files, these counts should not change
+ StructuredData::Dictionary separate_debug_info;
+ if (sym_file->GetSeparateDebugInfo(separate_debug_info,
----------------
jeffreytan81 wrote:
I would not recommend calling `GetSeparateDebugInfo` API for this task.
`GetSeparateDebugInfo` API constructs `StructuredData::Dictionary` JSON style
objects internally which is not cheap (I have seen profile traces that indicate
constructing `StructuredData::Dictionary` as hot paths). This can get worse
considering we are using `statistics dump` in lldb session logging by default.
(means we are paying the cost every time).
If we want to reuse it, I would suggest refactoring and reuse the necessary
underlying implementation without extra expensive burdens.
https://github.com/llvm/llvm-project/pull/144424
_______________________________________________
lldb-commits mailing list
lldb-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/lldb-commits
