merge revision(s) 1f115f141d: [Backport #19246]

Speed up rebuilding the loaded feature index

	Rebuilding the loaded feature index slowed down with the bug fix
	for #17885 in 79a4484a07.  The
	slowdown was extreme if realpath emulation was used, but even when
	not emulated, it could be about 10x slower.

	This adds loaded_features_realpath_map to rb_vm_struct. This is a
	hidden hash mapping loaded feature paths to realpaths. When
	rebuilding the loaded feature index, look at this hash to get
	cached realpath values, and skip calling rb_check_realpath if a
	cached value is found.

	Fixes [Bug #19246]
	---
	 load.c    | 27 +++++++++++++++++++++++----
	 vm.c      |  2 ++
	 vm_core.h |  1 +
	 3 files changed, 26 insertions(+), 4 deletions(-)

.github/workflows/mingw.yml

@@ -124,16 +124,16 @@ jobs:
         timeout-minutes: 5
         run: |
           make test
+        shell: cmd
 
       - name: test-all
         timeout-minutes: 45
         run: |
-          # Actions uses UTF8, causes test failures, similar to normal OS setup
-          chcp.com 437
           make test-all
         env:
           RUBY_TESTOPTS: -j${{env.TEST_JOBS}} --retry --job-status=normal --show-skip --timeout-scale=1.5
           BUNDLER_VERSION:
+        shell: cmd
 
       - name: test-spec
         timeout-minutes: 10

load.c

@@ -162,6 +162,12 @@ get_loaded_features_realpaths(rb_vm_t *vm)
     return vm->loaded_features_realpaths;
 }
 
+static VALUE
+get_loaded_features_realpath_map(rb_vm_t *vm)
+{
+    return vm->loaded_features_realpath_map;
+}
+
 static VALUE
 get_LOADED_FEATURES(ID _x, VALUE *_y)
 {
@@ -327,13 +333,16 @@ get_loaded_features_index(rb_vm_t *vm)
 	st_foreach(vm->loaded_features_index, loaded_features_index_clear_i, 0);
 
         VALUE realpaths = vm->loaded_features_realpaths;
+        VALUE realpath_map = vm->loaded_features_realpath_map;
+        VALUE previous_realpath_map = rb_hash_dup(realpath_map);
         rb_hash_clear(realpaths);
+        rb_hash_clear(realpath_map);
         features = vm->loaded_features;
         for (i = 0; i < RARRAY_LEN(features); i++) {
             VALUE entry, as_str;
             as_str = entry = rb_ary_entry(features, i);
             StringValue(as_str);
-	    as_str = rb_fstring(rb_str_freeze(as_str));
+            as_str = rb_fstring(as_str);
             if (as_str != entry)
                 rb_ary_store(features, i, as_str);
             features_index_add(vm, as_str, INT2FIX(i));
@@ -344,9 +353,14 @@ get_loaded_features_index(rb_vm_t *vm)
         long j = RARRAY_LEN(features);
         for (i = 0; i < j; i++) {
             VALUE as_str = rb_ary_entry(features, i);
-            VALUE realpath = rb_check_realpath(Qnil, as_str, NULL);
+            VALUE realpath = rb_hash_aref(previous_realpath_map, as_str);
+            if (NIL_P(realpath)) {
+                realpath = rb_check_realpath(Qnil, as_str, NULL);
                 if (NIL_P(realpath)) realpath = as_str;
-            rb_hash_aset(realpaths, rb_fstring(realpath), Qtrue);
+                realpath = rb_fstring(realpath);
+            }
+            rb_hash_aset(realpaths, realpath, Qtrue);
+            rb_hash_aset(realpath_map, as_str, realpath);
         }
     }
     return vm->loaded_features_index;
@@ -1122,6 +1136,7 @@ require_internal(rb_execution_context_t *ec, VALUE fname, int exception, bool wa
     volatile VALUE saved_path;
     volatile VALUE realpath = 0;
     VALUE realpaths = get_loaded_features_realpaths(th->vm);
+    VALUE realpath_map = get_loaded_features_realpath_map(th->vm);
     volatile bool reset_ext_config = false;
     struct rb_ext_config prev_ext_config;
 
@@ -1213,7 +1228,9 @@ require_internal(rb_execution_context_t *ec, VALUE fname, int exception, bool wa
         rb_provide_feature(th2->vm, path);
         VALUE real = realpath;
         if (real) {
-            rb_hash_aset(realpaths, rb_fstring(real), Qtrue);
+            real = rb_fstring(real);
+            rb_hash_aset(realpaths, real, Qtrue);
+            rb_hash_aset(realpath_map, path, real);
         }
     }
     ec->errinfo = saved.errinfo;
@@ -1426,6 +1443,8 @@ Init_load(void)
     vm->loaded_features_index = st_init_numtable();
     vm->loaded_features_realpaths = rb_hash_new();
     rb_obj_hide(vm->loaded_features_realpaths);
+    vm->loaded_features_realpath_map = rb_hash_new();
+    rb_obj_hide(vm->loaded_features_realpath_map);
 
     rb_define_global_function("load", rb_f_load, -1);
     rb_define_global_function("require", rb_f_require, 1);

version.h

@@ -11,7 +11,7 @@
 # define RUBY_VERSION_MINOR RUBY_API_VERSION_MINOR
 #define RUBY_VERSION_TEENY 4
 #define RUBY_RELEASE_DATE RUBY_RELEASE_YEAR_STR"-"RUBY_RELEASE_MONTH_STR"-"RUBY_RELEASE_DAY_STR
-#define RUBY_PATCHLEVEL 237
+#define RUBY_PATCHLEVEL 238
 
 #define RUBY_RELEASE_YEAR 2023
 #define RUBY_RELEASE_MONTH 9

vm.c

@@ -2539,6 +2539,7 @@ rb_vm_update_references(void *ptr)
         vm->loaded_features = rb_gc_location(vm->loaded_features);
         vm->loaded_features_snapshot = rb_gc_location(vm->loaded_features_snapshot);
         vm->loaded_features_realpaths = rb_gc_location(vm->loaded_features_realpaths);
+        vm->loaded_features_realpath_map = rb_gc_location(vm->loaded_features_realpath_map);
         vm->top_self = rb_gc_location(vm->top_self);
         vm->orig_progname = rb_gc_location(vm->orig_progname);
 
@@ -2630,6 +2631,7 @@ rb_vm_mark(void *ptr)
         rb_gc_mark_movable(vm->loaded_features);
         rb_gc_mark_movable(vm->loaded_features_snapshot);
         rb_gc_mark_movable(vm->loaded_features_realpaths);
+        rb_gc_mark_movable(vm->loaded_features_realpath_map);
         rb_gc_mark_movable(vm->top_self);
         rb_gc_mark_movable(vm->orig_progname);
         RUBY_MARK_MOVABLE_UNLESS_NULL(vm->coverages);

vm_core.h

@@ -681,6 +681,7 @@ typedef struct rb_vm_struct {
     VALUE loaded_features;
     VALUE loaded_features_snapshot;
     VALUE loaded_features_realpaths;
+    VALUE loaded_features_realpath_map;
     struct st_table *loaded_features_index;
     struct st_table *loading_table;
 #if EXTSTATIC