Hello Khaled,.
When a thread tries to lock a record and that record is already locked,the thread should give up the cpu and consume no more cpu cycles until the lock is available...
IMO speaking about cpu cycles at such high programming level is cleary an ambiguous requirement Sun want us to face to.
VM provides such hardware isolation that the requirement has no sense.
I decided to interpret it as:
When a thread tries to lock a record and that record is already locked,
the thread should enter in a wait state and should not be notified until the lock is available.
Regards, Ori