ns_mutex - Operate on mutexes
This command provides a mechanism to manipulate mutexes.
Initializes a new mutual exclusion (mutex) lock and returns a handle to it. If name is provided the mutex name will be set to this value. The command returns a mutex ID which used as argument for the commands operating on this mutex.
Destroys the mutex and frees any resources it was using. NOTE: The mutex must be unlocked, or else the behavior is undefined and will likely crash the server. Before using this, you should probably look at
ns_mutex lock acquires the specified mutual exclusion lock.
A convenience function to lock the specified mutex, run the script and unlock the mutex finally. The result and error condition of the script are returned by the command.
ns_mutex trylock tries to acquire the specified mutual exclusion lock. If the mutex is locked successfully, return value of 0 is returned. Non-zero return value indicates that the mutex is already locked by someone else.
ns_mutex unlock unlocks the specified mutual exclusion lock.
At startup (for example, in your init.tcl procedure), create a mutex named "foo" and remember it in a shared variable.
nsv_set mutex foo [ns_mutex create]
Later (for example, in a request procedure), you might use this mutex to lock the access to the file:
set mutex [nsv_get mutex foo]
...
ns_mutex lock $mutex
catch {
... access file foo ...
}
ns_mutex unlock $mutex
Note: The catch in the snipped above is important so the lock isn't held if Tcl unwinds due to an error accessing the file. Alternatively, one can use the following shorter version, where the eval subcommand takes care of locking/unlocking of the mutex without the need for the catch operation.
ns_mutex eval $mutex {
... access file foo ...
}
At shutdown (for example, in your shutdown procedure registered with ns_atshutdown), one can destroy the mutex (not strictly necessary).
ns_mutex destroy [nsv_get mutex foo] nsv_unset mutex foo