Bug in semop() in multiprocess environment?
shcheklein at gmail.com
Sun Nov 18 09:23:53 PST 2007
It seems to me that there is a bug in the semop() code:
After wakeup we see the following:
* The semaphore is still alive. Readjust the count of
* waiting processes.
if (sopptr->sem_op == 0)
It seems to me that "semptr" can be invalid after wakeup. For example,
semaphore's base can be changed after remove called on another previously
I think semptr must be renewed after wakeup.
More information about the freebsd-bugs