After upgrading from MariaDB 5.2.9 to 5.3.3 (incl. upgrading and checking tables), MariaDB 5.3.3 has serious issues crashing with Aria tables and, it seems, specificly indexes (exception 0xc0000005). There is one specific table 'xadmin' that even causes MariaDB 5.2.10 (after downgrading from 5.3.3) to crash with exception 0xc0000005. That happened on a Windows 2003 32bit server w/ 12GB RAM. This one serves as master.

Some other Windows XP 32bit PC serves as slave (replication). It is running MariaDB 5.3.3, the databases were copied from the master the evening (10-1-2012) and the slave was reset. The master server crashed this morning (11-1-2012).

The slave was then synced/replicated this afternoon (11-1-2012) and stopped with:
(null) (null) Page 966656: Fot error 126 when reading index file

After downgrading the master to MariaDB 5.2.10, it crashed after a while with:
— MariaDB 5.2.10 log begin —
Thread pointer: 0x22993708
Attempting backtrace. You can use the following information to find out
where mysqld died. If you see no messages after this, something went
terribly wrong...
5F878422

Trying to get some variables.
Some pointers may be invalid and cause the dump to abort.
Query (22F17F00): =UPDATE xadmin SET rechnum = '' WHERE rechnum = '0' OR rechnum IS NULL
Connection ID (thread ID): 10
Status: NOT_KILLED
— MariaDB 5.2.10 log end —

I will upload the table on ftp.askmonty.org including the log files.