fix @invokelatest performance regression (#58582)

vtjnash · topolarity · commit e35e82bc306a · 2025-06-30T20:44:59.000-04:00
A bit of hacking to get back near to the same performance as before by using the GlobalRef to optimize the getglobal lookup for now and avoiding the extra Vararg function indirection which forced some extra boxing and lookups. julia> @Btime foo(1.5) 22.892 ns (1 allocation: 16 bytes) # v1.11 141.543 ns (3 allocations: 48 bytes) # master 38.759 ns (2 allocations: 32 bytes) # PR The remaining difference is split about equally between the need now to box the world counter value for invoke_in_world and the extra cost of scanning the partition table for `Base.sin` to find the current entry. Fix #58334 (cherry picked from commit f12256b)
diff --git a/base/reflection.jl b/base/reflection.jl
@@ -1269,16 +1269,17 @@ macro invoke(ex)
     return esc(out)
 end
 
-apply_gr(gr::GlobalRef, @nospecialize args...) = getglobal(gr.mod, gr.name)(args...)
-apply_gr_kw(@nospecialize(kwargs::NamedTuple), gr::GlobalRef, @nospecialize args...) = Core.kwcall(kwargs, getglobal(gr.mod, gr.name), args...)
+getglobalref(gr::GlobalRef, world::UInt) = ccall(:jl_eval_globalref, Any, (Any, UInt), gr, world)
 
-function invokelatest_gr(gr::GlobalRef, @nospecialize args...; kwargs...)
+function invokelatest_gr(gr::GlobalRef, args...; kwargs...)
     @inline
     kwargs = merge(NamedTuple(), kwargs)
+    world = get_world_counter()
+    f = getglobalref(gr, world)
     if isempty(kwargs)
-        return invokelatest(apply_gr, gr, args...)
+        return invoke_in_world(world, f, args...)
     end
-    return invokelatest(apply_gr_kw, kwargs, gr, args...)
+    return invoke_in_world(world, Core.kwcall, kwargs, f, args...)
 end
 
 """
diff --git a/src/builtins.c b/src/builtins.c
@@ -1338,7 +1338,7 @@ JL_CALLABLE(jl_f_getglobal)
         jl_atomic_error("getglobal: module binding cannot be read non-atomically");
     else if (order >= jl_memory_order_seq_cst)
         jl_fence();
-    jl_value_t *v = jl_eval_global_var(mod, sym); // relaxed load
+    jl_value_t *v = jl_eval_global_var(mod, sym, jl_current_task->world_age); // relaxed load
     if (order >= jl_memory_order_acquire)
         jl_fence();
     return v;
diff --git a/src/gf.c b/src/gf.c
@@ -513,13 +513,13 @@ JL_DLLEXPORT jl_code_info_t *jl_gdbcodetyped1(jl_method_instance_t *mi, size_t w
     ct->world_age = jl_typeinf_world;
     jl_value_t **fargs;
     JL_GC_PUSHARGS(fargs, 4);
-    jl_module_t *CC = (jl_module_t*)jl_get_global_value(jl_core_module, jl_symbol("Compiler"));
+    jl_module_t *CC = (jl_module_t*)jl_get_global_value(jl_core_module, jl_symbol("Compiler"), ct->world_age);
     if (CC != NULL && jl_is_module(CC)) {
         JL_GC_PROMISE_ROOTED(CC);
-        fargs[0] = jl_get_global_value(CC, jl_symbol("NativeInterpreter"));;
+        fargs[0] = jl_get_global_value(CC, jl_symbol("NativeInterpreter"), ct->world_age);
         fargs[1] = jl_box_ulong(world);
         fargs[1] = jl_apply(fargs, 2);
-        fargs[0] = jl_get_global_value(CC, jl_symbol("typeinf_code"));
+        fargs[0] = jl_get_global_value(CC, jl_symbol("typeinf_code"), ct->world_age);
         fargs[2] = (jl_value_t*)mi;
         fargs[3] = jl_true;
         ci = (jl_code_info_t*)jl_apply(fargs, 4);
diff --git a/src/init.c b/src/init.c
@@ -249,7 +249,7 @@ JL_DLLEXPORT void jl_atexit_hook(int exitcode) JL_NOTSAFEPOINT_ENTER
     if (jl_base_module) {
         size_t last_age = ct->world_age;
         ct->world_age = jl_get_world_counter();
-        jl_value_t *f = jl_get_global_value(jl_base_module, jl_symbol("_atexit"));
+        jl_value_t *f = jl_get_global_value(jl_base_module, jl_symbol("_atexit"), ct->world_age);
         if (f != NULL) {
             jl_value_t **fargs;
             JL_GC_PUSHARGS(fargs, 2);
@@ -357,7 +357,7 @@ JL_DLLEXPORT void jl_postoutput_hook(void)
         jl_task_t *ct = jl_get_current_task();
         size_t last_age = ct->world_age;
         ct->world_age = jl_get_world_counter();
-        jl_value_t *f = jl_get_global_value(jl_base_module, jl_symbol("_postoutput"));
+        jl_value_t *f = jl_get_global_value(jl_base_module, jl_symbol("_postoutput"), ct->world_age);
         if (f != NULL) {
             JL_TRY {
                 JL_GC_PUSH1(&f);
diff --git a/src/interpreter.c b/src/interpreter.c
@@ -162,18 +162,18 @@ static jl_value_t *do_invoke(jl_value_t **args, size_t nargs, interpreter_state
 }
 
 // get the global (throwing if null) in the current world
-jl_value_t *jl_eval_global_var(jl_module_t *m, jl_sym_t *e)
+jl_value_t *jl_eval_global_var(jl_module_t *m, jl_sym_t *e, size_t world)
 {
-    jl_value_t *v = jl_get_global_value(m, e);
+    jl_value_t *v = jl_get_global_value(m, e, world);
     if (v == NULL)
         jl_undefined_var_error(e, (jl_value_t*)m);
     return v;
 }
 
 // get the global (throwing if null) in the current world, optimized
-jl_value_t *jl_eval_globalref(jl_globalref_t *g)
+jl_value_t *jl_eval_globalref(jl_globalref_t *g, size_t world)
 {
-    jl_value_t *v = jl_get_globalref_value(g);
+    jl_value_t *v = jl_get_globalref_value(g, world);
     if (v == NULL)
         jl_undefined_var_error(g->name, (jl_value_t*)g->mod);
     return v;
@@ -218,10 +218,10 @@ static jl_value_t *eval_value(jl_value_t *e, interpreter_state *s)
         return jl_quotenode_value(e);
     }
     if (jl_is_globalref(e)) {
-        return jl_eval_globalref((jl_globalref_t*)e);
+        return jl_eval_globalref((jl_globalref_t*)e, jl_current_task->world_age);
     }
     if (jl_is_symbol(e)) {  // bare symbols appear in toplevel exprs not wrapped in `thunk`
-        return jl_eval_global_var(s->module, (jl_sym_t*)e);
+        return jl_eval_global_var(s->module, (jl_sym_t*)e, jl_current_task->world_age);
     }
     if (jl_is_pinode(e)) {
         jl_value_t *val = eval_value(jl_fieldref_noalloc(e, 0), s);
diff --git a/src/jl_uv.c b/src/jl_uv.c
@@ -162,7 +162,8 @@ static void jl_uv_call_close_callback(jl_value_t *val)
     JL_GC_PUSHARGS(args, 2); // val is "rooted" in the finalizer list only right now
     args[0] = jl_eval_global_var(
             jl_base_relative_to(((jl_datatype_t*)jl_typeof(val))->name->module),
-            jl_symbol("_uv_hook_close")); // topmod(typeof(val))._uv_hook_close
+            jl_symbol("_uv_hook_close"),
+            jl_current_task->world_age); // topmod(typeof(val))._uv_hook_close
     args[1] = val;
     jl_apply(args, 2); // TODO: wrap in try-catch?
     JL_GC_POP();
diff --git a/src/jlapi.c b/src/jlapi.c
@@ -955,7 +955,7 @@ static NOINLINE int true_main(int argc, char *argv[])
     ct->world_age = jl_get_world_counter();
 
     jl_function_t *start_client = jl_base_module ?
-        (jl_function_t*)jl_get_global_value(jl_base_module, jl_symbol("_start")) : NULL;
+        (jl_function_t*)jl_get_global_value(jl_base_module, jl_symbol("_start"), ct->world_age) : NULL;
 
     if (start_client) {
         int ret = 1;
diff --git a/src/jltypes.c b/src/jltypes.c
@@ -3286,6 +3286,7 @@ void jl_init_types(void) JL_GC_DISABLED
                         jl_perm_symsvec(3, "mod", "name", "binding"),
                         jl_svec(3, jl_module_type, jl_symbol_type, jl_binding_type),
                         jl_emptysvec, 0, 0, 3);
+    jl_globalref_type->name->mayinlinealloc = 0; // not at all worthwhile, since the only constructor returns a boxed object
 
     jl_core_module = jl_new_module(jl_symbol("Core"), NULL);
 
diff --git a/src/julia.h b/src/julia.h
@@ -2117,9 +2117,7 @@ JL_DLLEXPORT jl_value_t *jl_get_existing_strong_gf(jl_binding_t *b JL_PROPAGATES
 JL_DLLEXPORT int jl_boundp(jl_module_t *m, jl_sym_t *var, int allow_import);
 JL_DLLEXPORT int jl_is_const(jl_module_t *m, jl_sym_t *var);
 JL_DLLEXPORT int jl_globalref_is_const(jl_globalref_t *gr);
-JL_DLLEXPORT jl_value_t *jl_get_globalref_value(jl_globalref_t *gr);
 JL_DLLEXPORT jl_value_t *jl_get_global(jl_module_t *m JL_PROPAGATES_ROOT, jl_sym_t *var);
-JL_DLLEXPORT jl_value_t *jl_get_global_value(jl_module_t *m, jl_sym_t *var);
 JL_DLLEXPORT void jl_set_global(jl_module_t *m JL_ROOTING_ARGUMENT, jl_sym_t *var, jl_value_t *val JL_ROOTED_ARGUMENT);
 JL_DLLEXPORT void jl_set_const(jl_module_t *m JL_ROOTING_ARGUMENT, jl_sym_t *var, jl_value_t *val JL_ROOTED_ARGUMENT);
 void jl_set_initial_const(jl_module_t *m JL_ROOTING_ARGUMENT, jl_sym_t *var, jl_value_t *val JL_ROOTED_ARGUMENT, int exported);
diff --git a/src/julia_internal.h b/src/julia_internal.h
@@ -901,7 +901,10 @@ STATIC_INLINE size_t module_usings_max(jl_module_t *m) JL_NOTSAFEPOINT {
 
 JL_DLLEXPORT jl_sym_t *jl_module_name(jl_module_t *m) JL_NOTSAFEPOINT;
 void jl_add_scanned_method(jl_module_t *m, jl_method_t *meth);
-jl_value_t *jl_eval_global_var(jl_module_t *m JL_PROPAGATES_ROOT, jl_sym_t *e);
+jl_value_t *jl_eval_global_var(jl_module_t *m JL_PROPAGATES_ROOT, jl_sym_t *e, size_t world);
+JL_DLLEXPORT jl_value_t *jl_eval_globalref(jl_globalref_t *g, size_t world);
+jl_value_t *jl_get_globalref_value(jl_globalref_t *gr, size_t world);
+jl_value_t *jl_get_global_value(jl_module_t *m, jl_sym_t *var, size_t world);
 jl_value_t *jl_interpret_opaque_closure(jl_opaque_closure_t *clos, jl_value_t **args, size_t nargs);
 jl_value_t *jl_interpret_toplevel_thunk(jl_module_t *m, jl_code_info_t *src);
 jl_value_t *jl_interpret_toplevel_expr_in(jl_module_t *m, jl_value_t *e,
diff --git a/src/module.c b/src/module.c
@@ -1539,20 +1539,20 @@ JL_DLLEXPORT jl_binding_t *jl_get_module_binding(jl_module_t *m, jl_sym_t *var,
 }
 
 
-// get the value (or null) in the current world
-JL_DLLEXPORT jl_value_t *jl_get_globalref_value(jl_globalref_t *gr)
+// get the value (or null) in the world
+jl_value_t *jl_get_globalref_value(jl_globalref_t *gr, size_t world)
 {
     jl_binding_t *b = gr->binding;
     if (!b)
         b = jl_get_module_binding(gr->mod, gr->name, 1);
-    return jl_get_binding_value_depwarn(b, jl_current_task->world_age);
+    return jl_get_binding_value_depwarn(b, world);
 }
 
-// get the value (or null) in the current world
-JL_DLLEXPORT jl_value_t *jl_get_global_value(jl_module_t *m, jl_sym_t *var)
+// get the value (or null) in the world
+jl_value_t *jl_get_global_value(jl_module_t *m, jl_sym_t *var, size_t world)
 {
     jl_binding_t *b = jl_get_module_binding(m, var, 1);
-    return jl_get_binding_value_depwarn(b, jl_current_task->world_age);
+    return jl_get_binding_value_depwarn(b, world);
 }
 
 // get the global (or null) in the latest world
diff --git a/src/precompile.c b/src/precompile.c
@@ -45,8 +45,8 @@ void write_srctext(ios_t *f, jl_array_t *udeps, int64_t srctextpos) {
         size_t last_age = ct->world_age;
         ct->world_age = jl_atomic_load_acquire(&jl_world_counter);
         JL_GC_PUSH4(&deptuple, &depots, &replace_depot_func, &normalize_depots_func);
-        replace_depot_func = jl_eval_global_var(jl_base_module, jl_symbol("replace_depot_path"));
-        normalize_depots_func = jl_eval_global_var(jl_base_module, jl_symbol("normalize_depots_for_relocation"));
+        replace_depot_func = jl_eval_global_var(jl_base_module, jl_symbol("replace_depot_path"), jl_current_task->world_age);
+        normalize_depots_func = jl_eval_global_var(jl_base_module, jl_symbol("normalize_depots_for_relocation"), jl_current_task->world_age);
         depots = jl_apply(&normalize_depots_func, 1);
         jl_datatype_t *deptuple_p[5] = {jl_module_type, jl_string_type, jl_uint64_type, jl_uint32_type, jl_float64_type};
         jl_value_t *jl_deptuple_type = jl_apply_tuple_type_v((jl_value_t**)deptuple_p, 5);
diff --git a/src/rtutils.c b/src/rtutils.c
@@ -1596,10 +1596,11 @@ void jl_log(int level, jl_value_t *module, jl_value_t *group, jl_value_t *id,
             jl_value_t *msg)
 {
     jl_value_t *logmsg_func = NULL;
+    jl_task_t *ct = jl_current_task;
     if (jl_base_module) {
-        jl_value_t *corelogging = jl_get_global_value(jl_base_module, jl_symbol("CoreLogging"));
+        jl_value_t *corelogging = jl_get_global_value(jl_base_module, jl_symbol("CoreLogging"), ct->world_age);
         if (corelogging && jl_is_module(corelogging)) {
-            logmsg_func = jl_get_global_value((jl_module_t*)corelogging, jl_symbol("logmsg_shim"));
+            logmsg_func = jl_get_global_value((jl_module_t*)corelogging, jl_symbol("logmsg_shim"), ct->world_age);
         }
     }
     if (!logmsg_func) {
diff --git a/src/scheduler.c b/src/scheduler.c
@@ -331,7 +331,7 @@ void jl_task_wait_empty(void)
         jl_wait_empty_begin();
         size_t lastage = ct->world_age;
         ct->world_age = jl_atomic_load_acquire(&jl_world_counter);
-        jl_value_t *f = jl_get_global_value(jl_base_module, jl_symbol("wait"));
+        jl_value_t *f = jl_get_global_value(jl_base_module, jl_symbol("wait"), ct->world_age);
         wait_empty = ct;
         if (f) {
             JL_GC_PUSH1(&f);
diff --git a/src/staticdata_utils.c b/src/staticdata_utils.c
@@ -544,21 +544,21 @@ static int64_t write_dependency_list(ios_t *s, jl_array_t* worklist, jl_array_t
     jl_value_t *get_compiletime_prefs_func = NULL;
     JL_GC_PUSH8(&depots, &prefs_list, &unique_func, &replace_depot_func, &normalize_depots_func, &toplevel, &prefs_hash_func, &get_compiletime_prefs_func);
 
-    jl_array_t *udeps = (jl_array_t*)jl_get_global_value(jl_base_module, jl_symbol("_require_dependencies"));
+    jl_array_t *udeps = (jl_array_t*)jl_get_global_value(jl_base_module, jl_symbol("_require_dependencies"), ct->world_age);
     *udepsp = udeps;
 
     // unique(udeps) to eliminate duplicates while preserving order:
     // we preserve order so that the topmost included .jl file comes first
     if (udeps) {
-        unique_func = jl_eval_global_var(jl_base_module, jl_symbol("unique"));
+        unique_func = jl_eval_global_var(jl_base_module, jl_symbol("unique"), ct->world_age);
         jl_value_t *uniqargs[2] = {unique_func, (jl_value_t*)udeps};
         udeps = (jl_array_t*)jl_apply(uniqargs, 2);
         *udepsp = udeps;
         JL_TYPECHK(write_dependency_list, array_any, (jl_value_t*)udeps);
     }
 
-    replace_depot_func = jl_get_global_value(jl_base_module, jl_symbol("replace_depot_path"));
-    normalize_depots_func = jl_eval_global_var(jl_base_module, jl_symbol("normalize_depots_for_relocation"));
+    replace_depot_func = jl_get_global_value(jl_base_module, jl_symbol("replace_depot_path"), ct->world_age);
+    normalize_depots_func = jl_eval_global_var(jl_base_module, jl_symbol("normalize_depots_for_relocation"), ct->world_age);
 
     depots = jl_apply(&normalize_depots_func, 1);
 
@@ -616,9 +616,9 @@ static int64_t write_dependency_list(ios_t *s, jl_array_t* worklist, jl_array_t
     // Calculate Preferences hash for current package.
     if (jl_base_module) {
         // Toplevel module is the module we're currently compiling, use it to get our preferences hash
-        toplevel = jl_get_global_value(jl_base_module, jl_symbol("__toplevel__"));
-        prefs_hash_func = jl_eval_global_var(jl_base_module, jl_symbol("get_preferences_hash"));
-        get_compiletime_prefs_func = jl_eval_global_var(jl_base_module, jl_symbol("get_compiletime_preferences"));
+        toplevel = jl_get_global_value(jl_base_module, jl_symbol("__toplevel__"), ct->world_age);
+        prefs_hash_func = jl_eval_global_var(jl_base_module, jl_symbol("get_preferences_hash"), ct->world_age);
+        get_compiletime_prefs_func = jl_eval_global_var(jl_base_module, jl_symbol("get_compiletime_preferences"), ct->world_age);
 
         if (toplevel) {
             // call get_compiletime_prefs(__toplevel__)
diff --git a/src/task.c b/src/task.c
@@ -335,7 +335,7 @@ void JL_NORETURN jl_finish_task(jl_task_t *ct)
     // let the runtime know this task is dead and find a new task to run
     jl_function_t *done = jl_atomic_load_relaxed(&task_done_hook_func);
     if (done == NULL) {
-        done = (jl_function_t*)jl_get_global_value(jl_base_module, jl_symbol("task_done_hook"));
+        done = (jl_function_t*)jl_get_global_value(jl_base_module, jl_symbol("task_done_hook"), ct->world_age);
         if (done != NULL)
             jl_atomic_store_release(&task_done_hook_func, done);
     }
diff --git a/src/threading.c b/src/threading.c
@@ -408,7 +408,7 @@ static void jl_init_task_lock(jl_task_t *ct)
     ct->world_age = jl_get_world_counter();
     jl_function_t *done = jl_atomic_load_relaxed(&init_task_lock_func);
     if (done == NULL) {
-        done = (jl_function_t*)jl_get_global_value(jl_base_module, jl_symbol("init_task_lock"));
+        done = (jl_function_t*)jl_get_global_value(jl_base_module, jl_symbol("init_task_lock"), ct->world_age);
         if (done != NULL)
             jl_atomic_store_release(&init_task_lock_func, done);
     }
diff --git a/src/toplevel.c b/src/toplevel.c

Original file line number	Diff line number	Diff line change
`@@ -162,18 +162,18 @@ static jl_value_t do_invoke(jl_value_t *args, size_t nargs, interpreter_state`
`162`	`162`	`}`
`163`	`163`
`164`	`164`	`// get the global (throwing if null) in the current world`
`165`		`-jl_value_t jl_eval_global_var(jl_module_t m, jl_sym_t *e)`
	`165`	`+jl_value_t jl_eval_global_var(jl_module_t m, jl_sym_t *e, size_t world)`
`166`	`166`	`{`
`167`		`- jl_value_t *v = jl_get_global_value(m, e);`
	`167`	`+ jl_value_t *v = jl_get_global_value(m, e, world);`
`168`	`168`	`if (v == NULL)`
`169`	`169`	`jl_undefined_var_error(e, (jl_value_t*)m);`
`170`	`170`	`return v;`
`171`	`171`	`}`
`172`	`172`
`173`	`173`	`// get the global (throwing if null) in the current world, optimized`
`174`		`-jl_value_t jl_eval_globalref(jl_globalref_t g)`
	`174`	`+jl_value_t jl_eval_globalref(jl_globalref_t g, size_t world)`
`175`	`175`	`{`
`176`		`- jl_value_t *v = jl_get_globalref_value(g);`
	`176`	`+ jl_value_t *v = jl_get_globalref_value(g, world);`
`177`	`177`	`if (v == NULL)`
`178`	`178`	`jl_undefined_var_error(g->name, (jl_value_t*)g->mod);`
`179`	`179`	`return v;`
`@@ -218,10 +218,10 @@ static jl_value_t eval_value(jl_value_t e, interpreter_state *s)`
`218`	`218`	`return jl_quotenode_value(e);`
`219`	`219`	`}`
`220`	`220`	`if (jl_is_globalref(e)) {`
`221`		`- return jl_eval_globalref((jl_globalref_t*)e);`
	`221`	`+ return jl_eval_globalref((jl_globalref_t*)e, jl_current_task->world_age);`
`222`	`222`	`}`
`223`	`223`	if (jl_is_symbol(e)) { // bare symbols appear in toplevel exprs not wrapped in `thunk`
`224`		`- return jl_eval_global_var(s->module, (jl_sym_t*)e);`
	`224`	`+ return jl_eval_global_var(s->module, (jl_sym_t*)e, jl_current_task->world_age);`
`225`	`225`	`}`
`226`	`226`	`if (jl_is_pinode(e)) {`
`227`	`227`	`jl_value_t *val = eval_value(jl_fieldref_noalloc(e, 0), s);`