Changes from the innodb-5.0-ss115 snapshot.

 Fixed bugs:
 BUG#15991: "innodb-file-per-table + symlink database + rename = crash"
 BUG#15650: "DELETE with LEFT JOIN crashes server"
 BUG#15308: "Problem of Order with Enum Column in Primary Key"
 BUG#14189: "VARBINARY and BINARY variables: trailing space ignored"


innobase/include/data0type.h:
  Changes from the innodb-5.0-ss115 snapshot.
innobase/include/data0type.ic:
  Changes from the innodb-5.0-ss115 snapshot.
   Fixed bug #14189. dtype_get_pad_char(): Do not pad VARBINARY
   or BINARY cloumns.
innobase/include/lock0lock.h:
  Changes from the innodb-5.0-ss115 snapshot.
innobase/include/os0file.h:
  Changes from the innodb-5.0-ss115 snapshot.
   os_file_handle_error(): Map the error codes EXDEV, ENOTDIR, and
   EISDIR to the new code OS_FILE_PATH_ERROR. Treat this code as
   OS_FILE_PATH_ERROR. This fixes the crash on RENAME TABLE when
   the .ibd file is a symbolic link to a different file system
   (bug#15991).
innobase/include/row0mysql.h:
  Changes from the innodb-5.0-ss115 snapshot.
innobase/lock/lock0lock.c:
  Changes from the innodb-5.0-ss115 snapshot.
   lock_rec_unlock(): Initialize local variable release_lock,
   in order to avoid dereferencing an uninitialized pointer
   when no lock exists on rec.
innobase/os/os0file.c:
  Changes from the innodb-5.0-ss115 snapshot.
   os_file_handle_error(): Map the error codes EXDEV, ENOTDIR, and
     EISDIR to the new code OS_FILE_PATH_ERROR. Treat this code as
     OS_FILE_PATH_ERROR. This fixes the crash on RENAME TABLE when
     the .ibd file is a symbolic link to a different file system
     (bug#15991).
   Protect the increment and decrement operations on the statistic
     variables os_n_pending_writes/reads with os_file_count_mutes.
innobase/row/row0ins.c:
  Changes from the innodb-5.0-ss115 snapshot.
   Fixed bug #14189. row_ins_cascade_calc_update_vec(): Refuse
   ON UPDATE_CASCADE when trying to change the length of of a
   VARBINARY column that refers to or is referenced by a BINARY
   column. BINARY columns are no longer padded on comparison,
   and thus they cannot be padded on storage either.
innobase/row/row0mysql.c:
  Changes from the innodb-5.0-ss115 snapshot.
   Fixed bug on unlock_row. In a unlock_row we may unlock
   only the latest lock granted to this transaction to the row.
innobase/row/row0sel.c:
  Changes from the innodb-5.0-ss115 snapshot.
   Fixed bug #15308.
   Fixed bug #14189: innobase_init(): Assert that
     DATA_MYSQL_BINARY_CHARSET_COLL == my_charset_bin.number.
   After review fixes for unlock bug where unlock released all
     locks transaction requested for a row. Only a latest requested
     lock to a row should be released. Update function comments to
     reflect current state. Persistent cursor should be stored
     whenever select lock type != LOCK_NONE.
innobase/trx/trx0trx.c:
  Changes from the innodb-5.0-ss115 snapshot.
   trx_commit_off_kernel(): Do not write empty trx->mysql_log_file_name.
mysql-test/r/innodb.result:
  Changes from the innodb-5.0-ss115 snapshot.
mysql-test/t/innodb.test:
  Changes from the innodb-5.0-ss115 snapshot.
sql/ha_innodb.cc:
  Changes from the innodb-5.0-ss115 snapshot.
   Fixed bug #15308.
   Fixed bug #14189: innobase_init(): Assert that
     DATA_MYSQL_BINARY_CHARSET_COLL == my_charset_bin.number.
   After review fixes for unlock bug where unlock released all
     locks transaction requested for a row. Only a latest requested
     lock to a row should be released. Update function comments to
     reflect current state. Persistent cursor should be stored
     whenever select lock type != LOCK_NONE.
mysql-test/r/innodb_unsafe_binlog.result:
  Changes from the innodb-5.0-ss115 snapshot.
mysql-test/t/innodb_unsafe_binlog-master.opt:
  Changes from the innodb-5.0-ss115 snapshot.
mysql-test/t/innodb_unsafe_binlog.test:
  Changes from the innodb-5.0-ss115 snapshot.
   Added testcases for bug #15650.

1 files changed, 6 insertions, 6 deletions

diff --git a/innobase/row/row0sel.c b/innobase/row/row0sel.c
index 1b66f14f5d7..d25023dc6be 100644
--- a/innobase/row/row0sel.c
+++ b/innobase/row/row0sel.c
@@ -2771,8 +2771,8 @@ row_sel_get_clust_rec_for_mysql(
 func_exit:
 	*out_rec = clust_rec;
 
-	if (prebuilt->select_lock_type == LOCK_X) {
-		/* We may use the cursor in update: store its position */
+	if (prebuilt->select_lock_type != LOCK_NONE) {
+		/* We may use the cursor in unlock: store its position */
 		
 		btr_pcur_store_position(prebuilt->clust_pcur, mtr);
 	}
@@ -3972,12 +3972,12 @@ got_row:
 	/* We have an optimization to save CPU time: if this is a consistent
 	read on a unique condition on the clustered index, then we do not
 	store the pcur position, because any fetch next or prev will anyway
-	return 'end of file'. An exception is the MySQL HANDLER command
-	where the user can move the cursor with PREV or NEXT even after
-	a unique search. */
+	return 'end of file'. Exceptions are locking reads and the MySQL 
+	HANDLER command where the user can move the cursor with PREV or NEXT 
+	even after a unique search. */
 
 	if (!unique_search_from_clust_index
-	    || prebuilt->select_lock_type == LOCK_X
+	    || prebuilt->select_lock_type != LOCK_NONE
 	    || prebuilt->used_in_HANDLER) {
 
 		/* Inside an update always store the cursor position */