database/mariadb
1
0
Fork 0
mirror of https://github.com/MariaDB/server.git synced 2025-07-30 16:24:05 +03:00
Code Activity

Bug#MDEV-4518 Server crashes in is_white_space when it's run

Browse Source 
with query cache, charset ucs2 and collation ucs2_unicode_ci

  @ mysql-test/r/ctype_ucs2_query_cache.result
  @ mysql-test/t/ctype_ucs2_query_cache-master.opt
  @ mysql-test/t/ctype_ucs2_query_cache.test
    Adding tests

  @ sql/sql_cache.cc
    Fixing not to use default_character_set->state_map,
    which can point to a non-ASCII character set (utc2, utf16, utf32)
   and thus have state_map undefined.
This commit is contained in:
Alexander Barkov
2013-05-17 10:16:56 +04:00
parent 0fa7729954
commit 3ef880cb59
4 changed files with 63 additions and 1 deletions
Download Patch File Download Diff File Expand all files Collapse all files

25
sql/sql_cache.cc
View File

@ -466,6 +466,8 @@ static void make_base_query(String *new_query,
/* The following is guaranteed by the query_cache interface */
DBUG_ASSERT(query[query_length] == 0);
DBUG_ASSERT(!is_white_space(query[0]));
/* We do not support UCS2, UTF16, UTF32 as a client character set */
DBUG_ASSERT(current_thd->variables.character_set_client->mbminlen == 1);
new_query->length(0); // Don't copy anything from old buffer
if (new_query->realloc(query_length + additional_length))
@ -2430,7 +2432,28 @@ void Query_cache::init()
m_cache_status= Query_cache::OK;
m_requests_in_progress= 0;
initialized = 1;
query_state_map= default_charset_info->state_map;
/*
Using state_map from latin1 should be fine in all cases:
1. We do not support UCS2, UTF16, UTF32 as a client character set.
2. The other character sets are compatible on the lower ASCII-range
0x00-0x20, and have the following characters marked as spaces:
0x09 TAB
0x0A LINE FEED
0x0B VERTICAL TAB
0x0C FORM FEED
0x0D CARRIAGE RETUR
0x20 SPACE
Additionally, only some of the ASCII-compatible character sets
(including latin1) can have 0xA0 mapped to "NON-BREAK SPACE"
and thus marked as space.
That should not be a problem for those charsets that map 0xA0
to something else: the parser will just return syntax error
if this character appears straight in the query
(i.e. not inside a string literal or comment).
*/
query_state_map= my_charset_latin1.state_map;
/*
If we explicitly turn off query cache from the command line query
cache will be disabled for the reminder of the server life