Bug#17332 - changing key_buffer_size on a running server

            can crash under load

Resizing a key cache while it was in heavy use could crash the
server. There were several race conditions.

I reworked some of the algorithms to fix the race conditions.

No test case. Repeating the crashes requires heavy concurrent
load on the key cache. A test script is attached to the bug report.

More explanations to the changes are contained in a text file
attached to the bug report.
parent f05c519a
...@@ -44,6 +44,7 @@ typedef struct st_keycache_wqueue ...@@ -44,6 +44,7 @@ typedef struct st_keycache_wqueue
typedef struct st_key_cache typedef struct st_key_cache
{ {
my_bool key_cache_inited; my_bool key_cache_inited;
my_bool in_resize; /* true during resize operation */
my_bool resize_in_flush; /* true during flush of resize operation */ my_bool resize_in_flush; /* true during flush of resize operation */
my_bool can_be_used; /* usage of cache for read/write is allowed */ my_bool can_be_used; /* usage of cache for read/write is allowed */
uint key_cache_shift; uint key_cache_shift;
...@@ -72,6 +73,11 @@ typedef struct st_key_cache ...@@ -72,6 +73,11 @@ typedef struct st_key_cache
BLOCK_LINK *used_ins; /* ptr to the insertion block in LRU chain */ BLOCK_LINK *used_ins; /* ptr to the insertion block in LRU chain */
pthread_mutex_t cache_lock; /* to lock access to the cache structure */ pthread_mutex_t cache_lock; /* to lock access to the cache structure */
KEYCACHE_WQUEUE resize_queue; /* threads waiting during resize operation */ KEYCACHE_WQUEUE resize_queue; /* threads waiting during resize operation */
/*
Waiting for a zero resize count. Using a queue for symmetry though
only one thread can wait here.
*/
KEYCACHE_WQUEUE waiting_for_resize_cnt;
KEYCACHE_WQUEUE waiting_for_hash_link; /* waiting for a free hash link */ KEYCACHE_WQUEUE waiting_for_hash_link; /* waiting for a free hash link */
KEYCACHE_WQUEUE waiting_for_block; /* requests waiting for a free block */ KEYCACHE_WQUEUE waiting_for_block; /* requests waiting for a free block */
BLOCK_LINK *changed_blocks[CHANGED_BLOCKS_HASH]; /* hash for dirty file bl.*/ BLOCK_LINK *changed_blocks[CHANGED_BLOCKS_HASH]; /* hash for dirty file bl.*/
......
This source diff could not be displayed because it is too large. You can view the blob instead.
...@@ -48,9 +48,6 @@ struct st_remember _my_sig_remember[MAX_SIGNALS]={{0,0}}; ...@@ -48,9 +48,6 @@ struct st_remember _my_sig_remember[MAX_SIGNALS]={{0,0}};
sigset_t my_signals; /* signals blocked by mf_brkhant */ sigset_t my_signals; /* signals blocked by mf_brkhant */
#endif #endif
/* from mf_keycache.c */
my_bool key_cache_inited=0;
/* from mf_reccache.c */ /* from mf_reccache.c */
ulong my_default_record_cache_size=RECORD_CACHE_SIZE; ulong my_default_record_cache_size=RECORD_CACHE_SIZE;
......
...@@ -2716,8 +2716,8 @@ int ha_init_key_cache(const char *name, KEY_CACHE *key_cache) ...@@ -2716,8 +2716,8 @@ int ha_init_key_cache(const char *name, KEY_CACHE *key_cache)
if (!key_cache->key_cache_inited) if (!key_cache->key_cache_inited)
{ {
pthread_mutex_lock(&LOCK_global_system_variables); pthread_mutex_lock(&LOCK_global_system_variables);
long tmp_buff_size= (long) key_cache->param_buff_size; ulong tmp_buff_size= (ulong) key_cache->param_buff_size;
long tmp_block_size= (long) key_cache->param_block_size; uint tmp_block_size= (uint) key_cache->param_block_size;
uint division_limit= key_cache->param_division_limit; uint division_limit= key_cache->param_division_limit;
uint age_threshold= key_cache->param_age_threshold; uint age_threshold= key_cache->param_age_threshold;
pthread_mutex_unlock(&LOCK_global_system_variables); pthread_mutex_unlock(&LOCK_global_system_variables);
......
...@@ -4575,8 +4575,13 @@ int reassign_keycache_tables(THD *thd, KEY_CACHE *src_cache, ...@@ -4575,8 +4575,13 @@ int reassign_keycache_tables(THD *thd, KEY_CACHE *src_cache,
bool mysql_preload_keys(THD* thd, TABLE_LIST* tables) bool mysql_preload_keys(THD* thd, TABLE_LIST* tables)
{ {
DBUG_ENTER("mysql_preload_keys"); DBUG_ENTER("mysql_preload_keys");
/*
We cannot allow concurrent inserts. The storage engine reads
directly from the index file, bypassing the cache. It could read
outdated information if parallel inserts into cache blocks happen.
*/
DBUG_RETURN(mysql_admin_table(thd, tables, 0, DBUG_RETURN(mysql_admin_table(thd, tables, 0,
"preload_keys", TL_READ, 0, 0, 0, 0, "preload_keys", TL_READ_NO_INSERT, 0, 0, 0, 0,
&handler::preload_keys, 0)); &handler::preload_keys, 0));
} }
......
...@@ -858,6 +858,7 @@ int ha_myisam::preload_keys(THD* thd, HA_CHECK_OPT *check_opt) ...@@ -858,6 +858,7 @@ int ha_myisam::preload_keys(THD* thd, HA_CHECK_OPT *check_opt)
ulonglong map= ~(ulonglong) 0; ulonglong map= ~(ulonglong) 0;
TABLE_LIST *table_list= table->pos_in_table_list; TABLE_LIST *table_list= table->pos_in_table_list;
my_bool ignore_leaves= table_list->ignore_leaves; my_bool ignore_leaves= table_list->ignore_leaves;
char buf[ERRMSGSIZE+20];
DBUG_ENTER("ha_myisam::preload_keys"); DBUG_ENTER("ha_myisam::preload_keys");
...@@ -889,7 +890,6 @@ int ha_myisam::preload_keys(THD* thd, HA_CHECK_OPT *check_opt) ...@@ -889,7 +890,6 @@ int ha_myisam::preload_keys(THD* thd, HA_CHECK_OPT *check_opt)
errmsg= "Failed to allocate buffer"; errmsg= "Failed to allocate buffer";
break; break;
default: default:
char buf[ERRMSGSIZE+20];
my_snprintf(buf, ERRMSGSIZE, my_snprintf(buf, ERRMSGSIZE,
"Failed to read from index file (errno: %d)", my_errno); "Failed to read from index file (errno: %d)", my_errno);
errmsg= buf; errmsg= buf;
......
...@@ -58,12 +58,39 @@ int mi_preload(MI_INFO *info, ulonglong key_map, my_bool ignore_leaves) ...@@ -58,12 +58,39 @@ int mi_preload(MI_INFO *info, ulonglong key_map, my_bool ignore_leaves)
/* Check whether all indexes use the same block size */ /* Check whether all indexes use the same block size */
for (i= 1 ; i < keys ; i++) for (i= 1 ; i < keys ; i++)
{ {
#if !defined(INGO_TEST_LOADIDX_OFF)
/* Allow non-IGNORE-LEAVES index loading even with different block sizes. */
if (ignore_leaves && (keyinfo[i].block_length != block_length))
DBUG_RETURN(my_errno= HA_ERR_NON_UNIQUE_BLOCK_SIZE);
set_if_bigger(block_length, keyinfo[i].block_length);
#else
if (keyinfo[i].block_length != block_length) if (keyinfo[i].block_length != block_length)
DBUG_RETURN(my_errno= HA_ERR_NON_UNIQUE_BLOCK_SIZE); DBUG_RETURN(my_errno= HA_ERR_NON_UNIQUE_BLOCK_SIZE);
#endif
} }
#if !defined(INGO_TEST_LOADIDX_OFF)
/* Align non-IGNORE-LEAVES index loads. */
if (!ignore_leaves)
{
/* Round up to the next multiple of key_cache_block_size. */
length= ((info->preload_buff_size +
share->key_cache->key_cache_block_size - 1) /
share->key_cache->key_cache_block_size *
share->key_cache->key_cache_block_size);
/* Round down to the next multiple of key_cache_block_size. */
pos= (share->base.keystart / share->key_cache->key_cache_block_size *
share->key_cache->key_cache_block_size);
}
else
{
length= info->preload_buff_size/block_length * block_length;
set_if_bigger(length, block_length);
}
#else
length= info->preload_buff_size/block_length * block_length; length= info->preload_buff_size/block_length * block_length;
set_if_bigger(length, block_length); set_if_bigger(length, block_length);
#endif
if (!(buff= (uchar *) my_malloc(length, MYF(MY_WME)))) if (!(buff= (uchar *) my_malloc(length, MYF(MY_WME))))
DBUG_RETURN(my_errno= HA_ERR_OUT_OF_MEM); DBUG_RETURN(my_errno= HA_ERR_OUT_OF_MEM);
......
Markdown is supported
0%
or
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment