LIKE was not optimized in then server when run against INFORMATION_SCHEMA tables and no wildcards were used. Databases/tables with '_' or '%' in their names (escaped or not) are handled by this code path, although slower, since it is rare to find these characters in table names in SQL. If there is a '_' or '%' in the string, LIKE takes care of that; otherwise, '=' is now used instead. The only exception is the information_schema database, which is handled separately. The patch covers both getTables() and getColumns(). (Bug #61332)

The first call to a stored procedure failed with “No Database Selected”. The workaround introduced in DatabaseMetaData.getCallStmtParameterTypes to fix the server bug where SHOW CREATE PROCEDURE was not respecting lowercase table names was misbehaving when the connection was not attached to a database and on case-insensitive operating systems. (Bug #61150)

There was a concurrency bottleneck in Java's character set encoding/decoding when converting bytes to/from String values.

(Bug #61105)

Connector/J now avoids a concurrent bottleneck in Java's character set encoding/decoding when converting bytes to and from instances of String. (Bug #61105, Bug #12622056)