Subj : Re: Challenge: Multithreading & Synchronization
To   : comp.programming.threads
From : David Butenhof
Date : Mon May 23 2005 02:51 pm

Giancarlo Niccolai wrote:
> doug wrote:
> 
> 
>>Actually, I have a (hopefully) quick question.  Is the above algorithm
>>safe in a system where you dynamically create the recursive mutex?
> 
> Theoretically yes, but then the problem is NOT in the mutex, but in your
> app, trying to access a non initialized mutex when it is still not
> available.

To put it another way that might make this more clear...

A mutex is SHARED DATA. Normally, mutexes are initialized statically, or 
at initialization time before threads are created, and people don't 
think about the implications of "mutex visibility". However when you 
initialize mutexes dynamically you need to concern yourself with the 
memory visibility rules just as with any other dynamic shared data.

If you have reliable visibility to the mutex, then you'll have no 
trouble reading the initialized data; including the owner field with 
(one presumes) an initial value correctly communicating to each possible 
thread that it is not the current owner. But if you do not have reliable 
visibility, then you have much worse problems than not seeing the 
initial owner value... all use of that mutex is completely 
indeterminate. (For example, the first attempt to lock from another 
thread might misread the LOCK status, and attempt to block on an 
uninitialized wait queue...)

-- 
Dave Butenhof, David.Butenhof@hp.com
HP Utility Pricing software, POSIX thread consultant
Manageability Solutions Lab (MSL), Hewlett-Packard Company
110 Spit Brook Road, ZK2/3-Q18, Nashua, NH 03062

.