Fixes to encoding/transcoding for ractors.

Not all ractor-related encoding issues were fixed by 1afc07e.
I found more by running my test-all branch with 3 ractors for each test.

Author: luke-gruber

depend

@@ -1401,7 +1401,6 @@ compile.$(OBJEXT): $(top_srcdir)/prism/pack.h
 compile.$(OBJEXT): $(top_srcdir)/prism/parser.h
 compile.$(OBJEXT): $(top_srcdir)/prism/prettyprint.h
 compile.$(OBJEXT): $(top_srcdir)/prism/prism.h
-compile.$(OBJEXT): $(top_srcdir)/prism/prism.h
 compile.$(OBJEXT): $(top_srcdir)/prism/regexp.h
 compile.$(OBJEXT): $(top_srcdir)/prism/static_literals.h
 compile.$(OBJEXT): $(top_srcdir)/prism/util/pm_buffer.h
@@ -17727,6 +17726,7 @@ transcode.$(OBJEXT): $(hdrdir)/ruby/ruby.h
 transcode.$(OBJEXT): $(top_srcdir)/internal/array.h
 transcode.$(OBJEXT): $(top_srcdir)/internal/class.h
 transcode.$(OBJEXT): $(top_srcdir)/internal/compilers.h
+transcode.$(OBJEXT): $(top_srcdir)/internal/encoding.h
 transcode.$(OBJEXT): $(top_srcdir)/internal/gc.h
 transcode.$(OBJEXT): $(top_srcdir)/internal/inits.h
 transcode.$(OBJEXT): $(top_srcdir)/internal/object.h

encoding.c

@@ -459,6 +459,16 @@ enc_registered(struct enc_table *enc_table, const char *name)
     return -1;
 }
 
+int
+rb_enc_registered(const char *name)
+{
+    int idx;
+    GLOBAL_ENC_TABLE_LOCKING(enc_table) {
+        idx = enc_registered(enc_table, name);
+    }
+    return idx;
+}
+
 void
 rb_encdb_declare(const char *name)
 {
@@ -1600,8 +1610,10 @@ enc_set_default_encoding(struct default_encoding *def, VALUE encoding, const cha
         /* Already set */
         overridden = TRUE;
 
+    int index = 0;
     if (!NIL_P(encoding)) {
         enc_check_encoding(encoding); // loads it if necessary. Needs to be done outside of VM lock.
+        index = rb_enc_to_index(rb_to_encoding(encoding));
     }
 
     GLOBAL_ENC_TABLE_LOCKING(enc_table) {
@@ -1619,7 +1631,7 @@ enc_set_default_encoding(struct default_encoding *def, VALUE encoding, const cha
                       (st_data_t)UNSPECIFIED_ENCODING);
         }
         else {
-            def->index = rb_enc_to_index(rb_to_encoding(encoding));
+            def->index = index;
             def->enc = 0;
             enc_alias_internal(enc_table, name, def->index);
         }

internal/encoding.h

@@ -29,6 +29,7 @@ void rb_encdb_declare(const char *name);
 void rb_enc_set_base(const char *name, const char *orig);
 int rb_enc_set_dummy(int index);
 void rb_enc_raw_set(VALUE obj, rb_encoding *enc);
+int rb_enc_registered(const char *name);
 
 PUREFUNC(int rb_data_is_encoding(VALUE obj));
 

test/ruby/test_encoding.rb

@@ -157,4 +157,23 @@ def test_ractor_lazy_load_encoding_concurrently
       assert rs.empty?
     end;
   end
+
+  def test_ractor_set_default_external_string
+    assert_ractor("#{<<~"begin;"}\n#{<<~'end;'}")
+    begin;
+    $-w = nil
+    rs = []
+    7.times do |i|
+      rs << Ractor.new(i) do |i|
+        Encoding.default_external = "us-ascii"
+      end
+    end
+
+    while rs.any?
+      r, _obj = Ractor.select(*rs)
+      rs.delete(r)
+    end
+    assert rs.empty?
+    end;
+  end
 end

test/ruby/test_transcode.rb

@@ -2361,6 +2361,52 @@ def test_ractor_lazy_load_encoding_random
     end;
   end
 
+  def test_ractor_asciicompat_encoding_exists
+    assert_ractor("#{<<~"begin;"}\n#{<<~'end;'}")
+    begin;
+      rs = []
+      7.times do
+        rs << Ractor.new do
+          string = "ISO-2022-JP"
+          encoding = Encoding.find(string)
+          20_000.times do
+            Encoding::Converter.asciicompat_encoding(string)
+            Encoding::Converter.asciicompat_encoding(encoding)
+          end
+        end
+      end
+
+      while rs.any?
+        r, _obj = Ractor.select(*rs)
+        rs.delete(r)
+      end
+      assert rs.empty?
+    end;
+  end
+
+  def test_ractor_asciicompat_encoding_doesnt_exist
+    assert_ractor("#{<<~"begin;"}\n#{<<~'end;'}")
+    begin;
+      rs = []
+      NO_EXIST = "I".freeze
+      7.times do
+        rs << Ractor.new do
+          50.times do
+            if (val = Encoding::Converter.asciicompat_encoding(NO_EXIST))
+              raise "Got #{val}, expected nil"
+            end
+          end
+        end
+      end
+
+      while rs.any?
+        r, _obj = Ractor.select(*rs)
+        rs.delete(r)
+      end
+      assert rs.empty?
+    end;
+  end
+
   private
 
   def assert_conversion_both_ways_utf8(utf8, raw, encoding)

transcode.c

@@ -19,6 +19,7 @@
 #include "internal/object.h"
 #include "internal/string.h"
 #include "internal/transcode.h"
+#include "internal/encoding.h"
 #include "ruby/encoding.h"
 #include "vm_sync.h"
 
@@ -1826,7 +1827,9 @@ rb_econv_asciicompat_encoding(const char *ascii_incompat_name)
     st_table *table2;
     struct asciicompat_encoding_t data = {0};
 
-    RB_VM_LOCKING() {
+    unsigned int lev;
+    RB_VM_LOCK_ENTER_LEV(&lev);
+    {
         if (st_lookup(transcoder_table, (st_data_t)ascii_incompat_name, &v)) {
             table2 = (st_table *)v;
             /*
@@ -1839,12 +1842,25 @@ rb_econv_asciicompat_encoding(const char *ascii_incompat_name)
             if (table2->num_entries == 1) {
                 data.ascii_incompat_name = ascii_incompat_name;
                 data.ascii_compat_name = NULL;
-                st_foreach(table2, asciicompat_encoding_i, (st_data_t)&data);
+                if (rb_multi_ractor_p()) {
+                    /*
+                     * We need to unlock in case `load_transcoder_entry` actually loads the encoding
+                     * and table2 could be inserted into when we unlock.
+                     */
+                    st_table *dup_table2 = st_copy(table2);
+                    RB_VM_LOCK_LEAVE_LEV(&lev);
+                    st_foreach(dup_table2, asciicompat_encoding_i, (st_data_t)&data);
+                    st_free_table(dup_table2);
+                    RB_VM_LOCK_ENTER_LEV(&lev);
+                }
+                else {
+                    st_foreach(table2, asciicompat_encoding_i, (st_data_t)&data);
+                }
             }
 
         }
-
     }
+    RB_VM_LOCK_LEAVE_LEV(&lev);
 
     return data.ascii_compat_name; // can be NULL
 }
@@ -2989,10 +3005,16 @@ static rb_encoding *
 make_encoding(const char *name)
 {
     rb_encoding *enc;
-    RB_VM_LOCKING() {
-        enc = rb_enc_find(name);
-        if (!enc)
-            enc = make_dummy_encoding(name);
+    enc = rb_enc_find(name);
+    if (!enc) {
+        RB_VM_LOCKING() {
+            if (rb_enc_registered(name)) {
+                enc = NULL;
+            }
+            else {
+                enc = make_dummy_encoding(name);
+            }
+        }
     }
     return enc;
 }
@@ -3029,14 +3051,10 @@ econv_s_asciicompat_encoding(VALUE klass, VALUE arg)
     VALUE enc = Qnil;
 
     enc_arg(&arg, &arg_name, &arg_enc);
-
-    RB_VM_LOCKING() {
-        result_name = rb_econv_asciicompat_encoding(arg_name);
-
-        if (result_name) {
-            result_enc = make_encoding(result_name);
-            enc = rb_enc_from_encoding(result_enc);
-        }
+    result_name = rb_econv_asciicompat_encoding(arg_name);
+    if (result_name) {
+        result_enc = make_encoding(result_name);
+        enc = rb_enc_from_encoding(result_enc);
     }
     return enc;
 }