Hi!
On Jul 31, Ramil Kalimullin wrote:
> #At file:///home/ram/mysql/b37756.5.1/
>
> 2681 Ramil Kalimullin 2008-07-31
> Fix for bug #37756: enabling fulltext indexes with
> myisam_repair_threads > 1 causes crash
>
> Problem: parallel repair (myisam_repair_threads > 1) of a myisam
> table with two or more fulltext keys that use the same parser may
> lead to a server crash. ALTER TABLE ENABLE KEYS is affected as well.
>
> Fix: properly initialize fulltext structures for parallel repair.
>
> Note: 1. there's no deterministic test case.
> 2. now we call parser->init() for each each fulltext key
> (not for each fulltext parser used).
> === modified file 'storage/myisam/ft_parser.c'
> --- a/storage/myisam/ft_parser.c 2007-11-14 14:32:03 +0000
> +++ b/storage/myisam/ft_parser.c 2008-07-31 04:50:44 +0000
> @@ -324,59 +324,58 @@ int ft_parse(TREE *wtree, uchar *doc, in
> }
>
> #define MAX_PARAM_NR 2
> -MYSQL_FTPARSER_PARAM *ftparser_call_initializer(MI_INFO *info,
> - uint keynr, uint paramnr)
> +
> +
> +MYSQL_FTPARSER_PARAM* ftparser_alloc_param(MI_INFO *info)
> {
> - uint32 ftparser_nr;
> - struct st_mysql_ftparser *parser;
> - if (! info->ftparser_param)
> + if (!info->ftparser_param)
> {
> - /* info->ftparser_param can not be zero after the initialization,
> - because it always includes built-in fulltext parser. And built-in
> - parser can be called even if the table has no fulltext indexes and
> - no varchar/text fields. */
> - if (! info->s->ftparsers)
> + /*
> + info->ftparser_param can not be zero after the initialization,
> + because it always includes built-in fulltext parser. And built-in
> + parser can be called even if the table has no fulltext indexes and
> + no varchar/text fields.
> + */
> + if (!info->s->ftparsers)
> {
> - /* It's ok that modification to shared structure is done w/o mutex
> - locks, because all threads would set the same variables to the
> - same values. */
> - uint i, j, keys= info->s->state.header.keys, ftparsers= 1;
> + /*
> + It's ok that modification to shared structure is done w/o mutex
> + locks, because all threads would set the same variables to the
> + same values.
> + */
> + uint i, keys= info->s->state.header.keys, ftparsers= 1;
> for (i= 0; i < keys; i++)
> {
> MI_KEYDEF *keyinfo= &info->s->keyinfo[i];
> +
> if (keyinfo->flag & HA_FULLTEXT)
> - {
> - for (j= 0;; j++)
> - {
> - if (j == i)
> - {
> - keyinfo->ftparser_nr= ftparsers++;
> - break;
> - }
> - if (info->s->keyinfo[j].flag & HA_FULLTEXT &&
> - keyinfo->parser == info->s->keyinfo[j].parser)
> - {
> - keyinfo->ftparser_nr= info->s->keyinfo[j].ftparser_nr;
> - break;
> - }
> - }
> - }
> + keyinfo->ftparser_nr= ftparsers++;
may be it shoud be then "ftindex_nr" ? and set in mi_open not here ?
> }
> info->s->ftparsers= ftparsers;
> }
> /*
> We have to allocate two MYSQL_FTPARSER_PARAM structures per plugin
> because in a boolean search a parser is called recursively
> - ftb_find_relevance* calls ftb_check_phrase*
> - (MAX_PARAM_NR=2)
> + ftb_find_relevance* calls ftb_check_phrase * (MAX_PARAM_NR == 2).
this doesn't make any sense. The comment should be
ftb_find_relevance... parser (ftb_find_relevance_parse,
ftb_find_relevance_add_word) calls ftb_check_phrase... parser
(ftb_check_phrase_internal, ftb_phrase_add_word). Thus MAX_PARAM_NR=2.
> */
> info->ftparser_param= (MYSQL_FTPARSER_PARAM *)
> my_malloc(MAX_PARAM_NR * sizeof(MYSQL_FTPARSER_PARAM) *
> info->s->ftparsers, MYF(MY_WME|MY_ZEROFILL));
> init_alloc_root(&info->ft_memroot, FTPARSER_MEMROOT_ALLOC_SIZE, 0);
> - if (! info->ftparser_param)
> - return 0;
> }
> + return info->ftparser_param;
> +}
Regards / Mit vielen Grüssen,
Sergei
--
__ ___ ___ ____ __
/ |/ /_ __/ __/ __ \/ / Sergei Golubchik <serg@stripped>
/ /|_/ / // /\ \/ /_/ / /__ Principal Software Engineer/Server Architect
/_/ /_/\_, /___/\___\_\___/ Sun Microsystems GmbH, HRB München 161028
<___/ Sonnenallee 1, 85551 Kirchheim-Heimstetten
Geschäftsführer: Thomas Schroeder, Wolfgang Engels, Dr. Roland Boemer
Vorsitzender des Aufsichtsrates: Martin Häring
