database/mariadb
1
0
Fork 0
mirror of https://github.com/MariaDB/server.git synced 2026-01-06 05:22:24 +03:00
Code Activity

Fix MySQL-5.1 bug http://bugs.mysql.com/bug.php?id=17992 and its

Browse Source 
duplicate http://bugs.mysql.com/bug.php?id=17134 : MySQL's partitioned 
table code does not set prebuilt->sql_stat_start right if it does an 
insert in the same statement after doing a search first in the same 
partition table. We now write trx id always to the buffer, not just when 
flag sql_stat_start is on. This will waste CPU time very slightly.
This commit is contained in:
heikki
2006-03-21 15:57:55 +00:00
parent df5ec8b2f3
commit 2f2c5642f6
1 changed files with 10 additions and 4 deletions
Download Patch File Download Diff File Expand all files Collapse all files

14
row/row0ins.c
View File

@@ -2451,7 +2451,15 @@ row_ins_step(
/* If this is the first time this node is executed (or when
execution resumes after wait for the table IX lock), set an
IX lock on the table and reset the possible select node. */
IX lock on the table and reset the possible select node. MySQL's
partitioned table code may also call an insert within the same
SQL statement AFTER it has used this tbale handle to do a search.
This happens, for example, when a row update moves it to another
partition. In that case, we have already set the IX lock on the
table during the search operation, and there is no need to set
it again here. But we must write thx->id to to node->trx_id_buf. */
trx_write_trx_id(node->trx_id_buf, trx->id);
if (node->state == INS_NODE_SET_IX_LOCK) {
@@ -2459,13 +2467,11 @@ row_ins_step(
its transaction, or it has been committed: */
if (UT_DULINT_EQ(trx->id, node->trx_id)) {
/* No need to do IX-locking or write trx id to buf */
/* No need to do IX-locking */
goto same_trx;
}
trx_write_trx_id(node->trx_id_buf, trx->id);
err = lock_table(0, node->table, LOCK_IX, thr);
if (err != DB_SUCCESS) {