For what it's worth, I haven't seen this on any systems I manage when going from 5.0.22->5.0.45, which include permutations of master-slave and master-master.
Is there anything useful in /var/log/mysqld.log?
after I updated from mysql-5.0.22 CentOS 5.0 to mysqld-5.0.45 in CentOS 5.2, mysql looses master-slave sync after one node reboots. I've noticed that the slave doest not respect the informantion on master.info, instead, it tries to read the informantion from the master server file inc-index.index....