Hi:

On Fri, Dec 26, 2008 at 2:44 PM, ajit mote <[email protected]> wrote:

>  How can I know who own the semaphore at one moment?
>>>>>>
>>>>>
>>>>> ipcs -s
>>>>>
>>>>  'cat /proc/sysvipc/sem' should be same.
>>>>
>>>>
>>>>>
>>>>>
>>>>>
>>>>>> On Thu, Dec 25, 2008 at 12:26 AM, shark huang <[email protected]>wrote:
>>>>>>
>>>>>>> Hi:  is the value of cuid and uid always same? Looks like they only
>>>>>>> are given a same value in function "ipc_addid", which means the value 
>>>>>>> was
>>>>>>> initialized when the semaphore is created.
>>>>>>>
>>>>>>
>>>   When performing the first task, the process which calls *semget*becomes 
>>> the owner/creator, and  Remember, ownership can be changed, but the
>>> creating process always remains the creator. So values of uid and cuid may
>>> not be always same.
>>>
>>     Several user can operate one semaphore  at the same time, so how to
>> decide who is the owner  ? Could you show me in source code?
>>
>>> Unfortunately, there isn't really a way to know what pid is "owning" the
> critical region that is being protected by this semaphore.  This is because
> semaphores don't really enforce the notion of a critical section.
>  Semaphores aren't binary gates, they aren't locked or unlocked (although
> they can be used that way).  Its entirely possible for two processes to
> manipulate a semaphore in such a way as both complete the operation without
> blocking.  So its not like the semphore defines a critical region by virtue
> solely of the call sites where it referenced.
> So the answer is no, there isn't a way to get what pids which are operating
> on the semaphore
>
>>
Yes, Yes. So I confuse what does mean the uid for?

>
>>>
>>>
>>>>>>> br
>>>>>>> shark
>>>>>>>
>>>>>>>
>>> Note: Please send mail to mailing list, this way everybody will be
>>> reading the thread
>>>
>>>>
>>>>>>
>>>>>
>>>>>
>>>>> --
>>>>> Thanks & Regards,
>>>>> Ajit Subhash Mote
>>>>>
>>>>>
>>>>
>>>
>>>
>>> --
>>> Thanks & Regards,
>>> Ajit Subhash Mote
>>>
>>>
>>
>
>
> --
> Thanks & Regards,
> Ajit Subhash Mote
>
>

Reply via email to