MDEV-34288 SET NAMES DEFAULT crashes mariadbd --collation-server=utf8mb4_unicode_ci

The @@global.character_set_client variable could erroneously be set to a non-default collation of its character set, which further made the `SET NAMES DEFAULT` statement crash the server. Fixing the code to make sure that the global value these variables: @@character_set_client @@character_set_connection @@character_set_server @@character_set_database @@character_set_connection point to the default compiled collations of the character set.
2025-01-15 19:42:28 +01:00 · 2024-06-04 12:00:20 +04:00 · 2024-06-04 12:00:20 +04:00 · 76e0dc18b6
commit 76e0dc18b6
parent d61947550a
9 changed files with 156 additions and 16 deletions
--- a/mysql-test/main/ctype_utf8mb4_unicode_ci_def.result
+++ b/mysql-test/main/ctype_utf8mb4_unicode_ci_def.result
@ -9,3 +9,48 @@ DROP TABLE t1;
 #
 # End of 10.3 tests
 #
+#
+# Start of 10.11 tests
+#
+#
+# MDEV-34288 SET NAMES DEFAULT crashes `mariadbd --collation-server=utf8mb4_unicode_ci`
+#
+SET NAMES DEFAULT COLLATE latin1_bin;
+ERROR 42000: COLLATION 'latin1_bin' is not valid for CHARACTER SET 'utf8mb4'
+SELECT @@character_set_connection, @@collation_connection, @@character_set_results;
+@@character_set_connection	@@collation_connection	@@character_set_results
+latin1	latin1_swedish_ci	latin1
+SET NAMES DEFAULT COLLATE utf8mb4_bin;
+SELECT @@character_set_connection, @@collation_connection, @@character_set_results;
+@@character_set_connection	@@collation_connection	@@character_set_results
+utf8mb4	utf8mb4_bin	utf8mb4
+SET NAMES DEFAULT COLLATE uca1400_ai_ci;
+SELECT @@character_set_connection, @@collation_connection, @@character_set_results;
+@@character_set_connection	@@collation_connection	@@character_set_results
+utf8mb4	utf8mb4_uca1400_ai_ci	utf8mb4
+SET @@global.character_set_client=latin1;
+SET NAMES DEFAULT;
+SELECT @@character_set_connection, @@collation_connection, @@character_set_results;
+@@character_set_connection	@@collation_connection	@@character_set_results
+latin1	latin1_swedish_ci	latin1
+SET @@global.character_set_client=utf8mb3;
+SET NAMES DEFAULT;
+SELECT @@character_set_connection, @@collation_connection, @@character_set_results;
+@@character_set_connection	@@collation_connection	@@character_set_results
+utf8mb3	utf8mb3_general_ci	utf8mb3
+SET @@global.character_set_client=DEFAULT;
+SET NAMES DEFAULT;
+SELECT @@character_set_connection, @@collation_connection, @@character_set_results;
+@@character_set_connection	@@collation_connection	@@character_set_results
+utf8mb4	utf8mb4_general_ci	utf8mb4
+SET NAMES DEFAULT;
+SELECT @@character_set_connection, @@collation_connection, @@character_set_results;
+@@character_set_connection	@@collation_connection	@@character_set_results
+utf8mb4	utf8mb4_general_ci	utf8mb4
+SET NAMES DEFAULT COLLATE DEFAULT;
+SELECT @@character_set_connection, @@collation_connection, @@character_set_results;
+@@character_set_connection	@@collation_connection	@@character_set_results
+utf8mb4	utf8mb4_general_ci	utf8mb4
+#
+# End of 10.11 tests
+#
--- a/mysql-test/main/ctype_utf8mb4_unicode_ci_def.test
+++ b/mysql-test/main/ctype_utf8mb4_unicode_ci_def.test
@ -13,3 +13,45 @@ DROP TABLE t1;
 --echo #
 --echo # End of 10.3 tests
 --echo #
+
+
+--echo #
+--echo # Start of 10.11 tests
+--echo #
+
+--echo #
+--echo # MDEV-34288 SET NAMES DEFAULT crashes `mariadbd --collation-server=utf8mb4_unicode_ci`
+--echo #
+
+--error ER_COLLATION_CHARSET_MISMATCH
+SET NAMES DEFAULT COLLATE latin1_bin;
+SELECT @@character_set_connection, @@collation_connection, @@character_set_results;
+
+SET NAMES DEFAULT COLLATE utf8mb4_bin;
+SELECT @@character_set_connection, @@collation_connection, @@character_set_results;
+
+SET NAMES DEFAULT COLLATE uca1400_ai_ci;
+SELECT @@character_set_connection, @@collation_connection, @@character_set_results;
+
+SET @@global.character_set_client=latin1;
+SET NAMES DEFAULT;
+SELECT @@character_set_connection, @@collation_connection, @@character_set_results;
+
+SET @@global.character_set_client=utf8mb3;
+SET NAMES DEFAULT;
+SELECT @@character_set_connection, @@collation_connection, @@character_set_results;
+
+SET @@global.character_set_client=DEFAULT;
+SET NAMES DEFAULT;
+SELECT @@character_set_connection, @@collation_connection, @@character_set_results;
+
+SET NAMES DEFAULT;
+SELECT @@character_set_connection, @@collation_connection, @@character_set_results;
+
+SET NAMES DEFAULT COLLATE DEFAULT;
+SELECT @@character_set_connection, @@collation_connection, @@character_set_results;
+
+
+--echo #
+--echo # End of 10.11 tests
+--echo #
--- a/sql/lex_charset.h
+++ b/sql/lex_charset.h
@ -288,6 +288,12 @@ public:
    DBUG_ASSERT(0);
    return m_ci->coll_name;
  }
+  static Lex_extended_collation_st collate_default()
+  {
+    Lex_extended_collation_st res;
+    res.set_collate_default();
+    return res;
+  }
  void set_collate_default()
  {
    m_ci= &my_collation_contextually_typed_default;
--- a/sql/mysqld.cc
+++ b/sql/mysqld.cc
@ -4329,8 +4329,10 @@ static int init_common_variables()
  if (is_supported_parser_charset(default_charset_info))
  {
    global_system_variables.collation_connection= default_charset_info;
-    global_system_variables.character_set_results= default_charset_info;
-    global_system_variables.character_set_client= default_charset_info;
+    global_system_variables.character_set_results=
+    global_system_variables.character_set_client=
+      Lex_exact_charset_opt_extended_collate(default_charset_info, true).
+        find_default_collation();
  }
  else
  {
--- a/sql/sql_lex.cc
+++ b/sql/sql_lex.cc
@ -12131,10 +12131,31 @@ bool LEX::sp_create_set_password_instr(THD *thd,
 }


+/*
+  Handle the SET NAMES statement variants, e.g.:
+    SET NAMES DEFAULT;
+    SET NAMES DEFAULT COLLATE DEFAULT;
+    SET NAMES DEFAULT COLLATE latin1_bin;
+    SET NAMES latin1;
+    SET NAMES latin1 COLLATE DEFAULT;
+    SET NAMES latin1 COLLATE latin1_bin;
+    SET NAMES utf8mb4 COLLATE uca1400_ai_ci;
+
+  @param pos          - The position of the keyword `NAMES` inside the query
+  @param cs           - The character set part, or nullptr if DEFAULT
+  @param cl           - The collation (explicit or contextually typed)
+  @param no_lookahead - The tokinizer lookahead state
+*/
 bool LEX::set_names(const char *pos,
-                    const Lex_exact_charset_opt_extended_collate &cscl,
+                    CHARSET_INFO *cs,
+                    const Lex_extended_collation_st &cl,
                    bool no_lookahead)
 {
+  CHARSET_INFO *def= global_system_variables.character_set_client;
+  Lex_exact_charset_opt_extended_collate cscl(cs ? cs : def, true);
+  if (cscl.merge_collation_override(cl))
+    return true;
+
  if (sp_create_assignment_lex(thd, pos))
    return true;
  CHARSET_INFO *ci= cscl.collation().charset_info();
--- a/sql/sql_lex.h
+++ b/sql/sql_lex.h
@ -3873,7 +3873,8 @@ public:
  int case_stmt_action_then();
  bool setup_select_in_parentheses();
  bool set_names(const char *pos,
-                 const Lex_exact_charset_opt_extended_collate &cs,
+                 CHARSET_INFO *cs,
+                 const Lex_extended_collation_st &coll,
                 bool no_lookahead);
  bool set_trigger_new_row(const LEX_CSTRING *name, Item *val);
  bool set_trigger_field(const LEX_CSTRING *name1, const LEX_CSTRING *name2,
--- a/sql/sql_yacc.yy
+++ b/sql/sql_yacc.yy
@ -16671,18 +16671,15 @@ option_value_no_option_type:
          }
        | NAMES_SYM charset_name_or_default
          {
-            CHARSET_INFO *def= global_system_variables.character_set_client;
-            Lex_exact_charset_opt_extended_collate tmp($2 ? $2 : def, false);
-            if (Lex->set_names($1.pos(), tmp, yychar == YYEMPTY))
+            if (Lex->set_names($1.pos(), $2,
+                               Lex_extended_collation_st::collate_default(),
+                               yychar == YYEMPTY))
              MYSQL_YYABORT;
          }
        | NAMES_SYM charset_name_or_default
                    COLLATE_SYM collation_name_or_default
          {
-            CHARSET_INFO *def= global_system_variables.character_set_client;
-            Lex_exact_charset_opt_extended_collate tmp($2 ? $2 : def, false);
-            if (tmp.merge_collation($4) ||
-                Lex->set_names($1.pos(), tmp, yychar == YYEMPTY))
+            if (Lex->set_names($1.pos(), $2, $4, yychar == YYEMPTY))
              MYSQL_YYABORT;
          }
        | DEFAULT ROLE_SYM grant_role
--- a/sql/sys_vars.cc
+++ b/sql/sys_vars.cc
@ -824,7 +824,7 @@ static Sys_var_struct Sys_character_set_system(
       READ_ONLY GLOBAL_VAR(system_charset_info), NO_CMD_LINE,
       offsetof(CHARSET_INFO, cs_name.str), DEFAULT(0));

-static Sys_var_struct Sys_character_set_server(
+static Sys_var_charset Sys_character_set_server(
       "character_set_server", "The default character set",
       SESSION_VAR(collation_server), NO_CMD_LINE,
       offsetof(CHARSET_INFO, cs_name.str), DEFAULT(&default_charset_info),
@ -838,7 +838,7 @@ static bool check_charset_db(sys_var *self, THD *thd, set_var *var)
    var->save_result.ptr= thd->db_charset;
  return false;
 }
-static Sys_var_struct Sys_character_set_database(
+static Sys_var_charset Sys_character_set_database(
       "character_set_database",
       "The character set used by the default database",
       SESSION_VAR(collation_database), NO_CMD_LINE,
@ -862,7 +862,8 @@ static bool fix_thd_charset(sys_var *self, THD *thd, enum_var_type type)
    thd->update_charset();
  return false;
 }
-static Sys_var_struct Sys_character_set_client(
+
+static Sys_var_charset Sys_character_set_client(
       "character_set_client", "The character set for statements "
       "that arrive from the client",
       NO_SET_STMT SESSION_VAR(character_set_client), NO_CMD_LINE,
@ -872,7 +873,7 @@ static Sys_var_struct Sys_character_set_client(
 // for check changing
 export sys_var *Sys_character_set_client_ptr= &Sys_character_set_client;

-static Sys_var_struct Sys_character_set_connection(
+static Sys_var_charset Sys_character_set_connection(
       "character_set_connection", "The character set used for "
       "literals that do not have a character set introducer and for "
       "number-to-string conversion",
@ -883,7 +884,7 @@ static Sys_var_struct Sys_character_set_connection(
 // for check changing
 export sys_var *Sys_character_set_connection_ptr= &Sys_character_set_connection;

-static Sys_var_struct Sys_character_set_results(
+static Sys_var_charset Sys_character_set_results(
       "character_set_results", "The character set used for returning "
       "query results to the client",
       SESSION_VAR(character_set_results), NO_CMD_LINE,
--- a/sql/sys_vars.inl
+++ b/sql/sys_vars.inl
@ -2166,6 +2166,31 @@ public:
  { return valptr(thd, *(uchar**)option.def_value); }
 };

+
+/**
+  The class to store character sets.
+*/
+class Sys_var_charset: public Sys_var_struct
+{
+public:
+  using Sys_var_struct::Sys_var_struct;
+  void global_save_default(THD *thd, set_var *var)
+  {
+    /*
+      The default value can point to an arbitrary collation,
+      e.g. default_charset_info.
+      Let's convert it to the compiled default collation.
+      This makes the code easier in various places such as SET NAMES.
+    */
+    void **default_value= reinterpret_cast<void**>(option.def_value);
+    var->save_result.ptr=
+      Lex_exact_charset_opt_extended_collate((CHARSET_INFO *) *default_value,
+                                             true).
+        find_default_collation();
+  }
+};
+
+
 /**
  The class for variables that store time zones