Improve how primary/top-level BIFs get initialized

src/Func.cc

@@ -902,6 +902,37 @@ void emit_builtin_exception(const char* msg, Obj* arg)
 	emit_builtin_error_common(msg, arg, true);
 	}
 
+void init_primary_bifs()
+	{
+	if ( did_builtin_init )
+		return;
+
+	ProcStats = id::find_type<RecordType>("ProcStats");
+	NetStats = id::find_type<RecordType>("NetStats");
+	MatcherStats = id::find_type<RecordType>("MatcherStats");
+	ConnStats = id::find_type<RecordType>("ConnStats");
+	ReassemblerStats = id::find_type<RecordType>("ReassemblerStats");
+	DNSStats = id::find_type<RecordType>("DNSStats");
+	GapStats = id::find_type<RecordType>("GapStats");
+	EventStats = id::find_type<RecordType>("EventStats");
+	TimerStats = id::find_type<RecordType>("TimerStats");
+	FileAnalysisStats = id::find_type<RecordType>("FileAnalysisStats");
+	ThreadStats = id::find_type<RecordType>("ThreadStats");
+	BrokerStats = id::find_type<RecordType>("BrokerStats");
+	ReporterStats = id::find_type<RecordType>("ReporterStats");
+
+	var_sizes = id::find_type("var_sizes")->AsTableType();
+
+#include "zeek.bif.func_init"
+#include "stats.bif.func_init"
+#include "reporter.bif.func_init"
+#include "strings.bif.func_init"
+#include "option.bif.func_init"
+#include "supervisor.bif.func_init"
+
+	did_builtin_init = true;
+	}
+
 } // namespace detail
 
 
@@ -936,31 +967,3 @@ void builtin_error(const char* msg, zeek::Obj* arg)
 	{
 	zeek::emit_builtin_error(msg, arg);
 	}
-
-void init_builtin_funcs()
-	{
-	ProcStats = zeek::id::find_type<zeek::RecordType>("ProcStats");
-	NetStats = zeek::id::find_type<zeek::RecordType>("NetStats");
-	MatcherStats = zeek::id::find_type<zeek::RecordType>("MatcherStats");
-	ConnStats = zeek::id::find_type<zeek::RecordType>("ConnStats");
-	ReassemblerStats = zeek::id::find_type<zeek::RecordType>("ReassemblerStats");
-	DNSStats = zeek::id::find_type<zeek::RecordType>("DNSStats");
-	GapStats = zeek::id::find_type<zeek::RecordType>("GapStats");
-	EventStats = zeek::id::find_type<zeek::RecordType>("EventStats");
-	TimerStats = zeek::id::find_type<zeek::RecordType>("TimerStats");
-	FileAnalysisStats = zeek::id::find_type<zeek::RecordType>("FileAnalysisStats");
-	ThreadStats = zeek::id::find_type<zeek::RecordType>("ThreadStats");
-	BrokerStats = zeek::id::find_type<zeek::RecordType>("BrokerStats");
-	ReporterStats = zeek::id::find_type<zeek::RecordType>("ReporterStats");
-
-	var_sizes = zeek::id::find_type("var_sizes")->AsTableType();
-
-#include "zeek.bif.func_init"
-#include "stats.bif.func_init"
-#include "reporter.bif.func_init"
-#include "strings.bif.func_init"
-#include "option.bif.func_init"
-#include "supervisor.bif.func_init"
-
-	zeek::detail::did_builtin_init = true;
-	}

src/Func.h

@@ -277,6 +277,7 @@ extern std::vector<CallInfo> call_stack;
 // This is set to true after the built-in functions have been initialized.
 extern bool did_builtin_init;
 extern std::vector<void (*)()> bif_initializers;
+extern void init_primary_bifs();
 
 inline void run_bif_initializers()
 	{
@@ -310,11 +311,6 @@ using function_ingredients [[deprecated("Remove in v4.1. Use zeek::detail::funct
 constexpr auto check_built_in_call [[deprecated("Remove in v4.1. Use zeek::detail::check_built_in_call.")]] = zeek::detail::check_built_in_call;
 constexpr auto render_call_stack [[deprecated("Remove in v4.1. Use zeek::render_call_stack.")]] = zeek::render_call_stack;
 
-// TODO: these are still here because of how all of the bif code gets included in Func.c. There could be a
-// renamed version inside the namespace, but the way that the code gets included complicates the matter. It
-// might need to be revisited after everything is namespaced everywhere else.
-void init_builtin_funcs();
-
 // TODO: do call_stack and did_builtin_init need to be aliased?
 
 // These have to be implemented as actual methods due to function overloading breaking the use of aliases.

src/scan.l

@@ -842,16 +842,6 @@ int yywrap()
 	if ( zeek::reporter->Errors() > 0 )
 		return 1;
 
-	if ( ! did_builtin_init && file_stack.length() == 1 )
-		{
-		// ### This is a gross hack - we know that the first file
-		// we parse is init-bare.zeek, and after it it's safe to initialize
-		// the built-ins.  Furthermore, we want to initialize the
-		// built-in's *right* after parsing bro.init, so that other
-		// source files can use built-in's when initializing globals.
-		init_builtin_funcs();
-		}
-
 	yy_delete_buffer(YY_CURRENT_BUFFER);
 
 	if ( file_stack.length() > 0 )

src/zeek-setup.cc

@@ -62,6 +62,7 @@ extern "C" {
 #include "broker/Manager.h"
 
 #include "binpac_zeek.h"
+#include "module_util.h"
 
 #include "3rdparty/sqlite3.h"
 
@@ -388,8 +389,7 @@ static std::vector<std::string> get_script_signature_files()
 	return rval;
 	}
 
-SetupResult setup(int argc, char** argv,
-                                Options* zopts)
+SetupResult setup(int argc, char** argv, Options* zopts)
 	{
 	ZEEK_LSAN_DISABLE();
 	std::set_new_handler(bro_new_handler);
@@ -628,6 +628,19 @@ SetupResult setup(int argc, char** argv,
 	HeapLeakChecker::Disabler disabler;
 #endif
 
+	auto ipbid = install_ID("__init_primary_bifs", GLOBAL_MODULE_NAME,
+	                        true, true);
+	auto ipbft = make_intrusive<FuncType>(make_intrusive<RecordType>(nullptr),
+	                                      base_type(TYPE_BOOL),
+	                                      FUNC_FLAVOR_FUNCTION);
+	ipbid->SetType(std::move(ipbft));
+	auto init_bifs = [](Frame* frame, const Args* args) -> BifReturnVal
+		{
+		init_primary_bifs();
+		return val_mgr->True();
+		};
+	auto ipbb = make_intrusive<BuiltinFunc>(init_bifs, ipbid->Name(), false);
+
 	run_state::is_parsing = true;
 	yyparse();
 	run_state::is_parsing = false;