diff options
author | unknown <kroki/tomash@moonlight.intranet> | 2006-08-22 11:47:52 +0400 |
---|---|---|
committer | unknown <kroki/tomash@moonlight.intranet> | 2006-08-22 11:47:52 +0400 |
commit | 877477f173419933f1d7c2b9e06dc7cfe925eb6e (patch) | |
tree | 6da805e8e859d377d16c5f512dfe034ba5c7ecb2 /include | |
parent | d8180d44b81519ee1d1db8fe79caa2a6d1d6a5eb (diff) | |
download | mariadb-git-877477f173419933f1d7c2b9e06dc7cfe925eb6e.tar.gz |
BUG#21051: RESET QUERY CACHE very slow when query_cache_type=0
There were two problems: RESET QUERY CACHE took a long time to complete
and other threads were blocked during this time.
The patch does three things:
1 fixes a bug with improper use of test-lock-test_again technique.
AKA Double-Checked Locking is applicable here only in few places.
2 Somewhat improves performance of RESET QUERY CACHE.
Do my_hash_reset() instead of deleting elements one by one. Note
however that the slowdown also happens when inserting into sorted
list of free blocks, should be rewritten using balanced tree.
3 Makes RESET QUERY CACHE non-blocking.
The patch adjusts the locking protocol of the query cache in the
following way: it introduces a flag flush_in_progress, which is
set when Query_cache::flush_cache() is in progress. This call
sets the flag on enter, and then releases the lock. Every other
call is able to acquire the lock, but does nothing if
flush_in_progress is set (as if the query cache is disabled).
The only exception is the concurrent calls to
Query_cache::flush_cache(), that are blocked until the flush is
over. When leaving Query_cache::flush_cache(), the lock is
acquired and the flag is reset, and one thread waiting on
Query_cache::flush_cache() (if any) is notified that it may
proceed.
include/mysql_com.h:
Add comment for NET::query_cache_query.
sql/net_serv.cc:
Use query_cache_init_query() for initialization of
NET::query_cache_query if query cache is used.
Do not access net->query_cache_query without a lock.
sql/sql_cache.cc:
Fix bug with accessing query_cache_size, Query_cache_query::wri and
thd->net.query_cache_query before acquiring the lock---leave
double-check locking only in safe places.
Wherever we check that cache is usable (query_cache_size > 0) we now
also check that flush_in_progress is false, i.e. we are not in the
middle of cache flush.
Add Query_cache::not_in_flush_or_wait() method and use it in
Query_cache::flush_cache(), so that threads doing cache flush will
wait it to finish, while other threads will bypass the cache as if
it is disabled.
Extract Query_cache::free_query_internal() from Query_cache::free_query(),
which does not removes elements from the hash, and use it together with
my_hash_reset() in Query_cache::flush_cache().
sql/sql_cache.h:
Add declarations for new members and methods.
Make is_cacheable() a static method.
Add query_cache_init_query() function.
sql/sql_class.cc:
Use query_cache_init_query() for initialization of
NET::query_cache_query.
Diffstat (limited to 'include')
-rw-r--r-- | include/mysql_com.h | 6 |
1 files changed, 6 insertions, 0 deletions
diff --git a/include/mysql_com.h b/include/mysql_com.h index ec1c133799f..aed32064dfe 100644 --- a/include/mysql_com.h +++ b/include/mysql_com.h @@ -210,7 +210,13 @@ typedef struct st_net { char last_error[MYSQL_ERRMSG_SIZE], sqlstate[SQLSTATE_LENGTH+1]; unsigned int last_errno; unsigned char error; + + /* + 'query_cache_query' should be accessed only via query cache + functions and methods to maintain proper locking. + */ gptr query_cache_query; + my_bool report_error; /* We should report error (we have unreported error) */ my_bool return_errno; } NET; |