Re: [Boost-users] owns_lock on lock not on mutex

Subject: Re: [Boost-users] owns_lock on lock not on mutex
From: Agustín K-ballo Bergé (kaballo86_at_[hidden])
Date: 2015-10-12 15:42:52

On 10/12/2015 4:17 PM, gast128 wrote:
>> Knowing whether a mutex is locked is of little use, it does not tell you
>> whether you own a lock on that mutex. Not to mention that the
>> information would already be stale by the time it is returned.
>
> Yes that's what I mean: if the mutex is locked by the calling thread,a
> although again I didn't specify my self correctly. No race condition
> possibility out of perspective of the calling thread
>
>> could be anything (object instance, scope, function call, underlying
>> thread or execution agent, etc), so the mutex can't tell you that.
>
> Mostly this is done for debugging purposes. Maybe it would a possibility to
> add some extra information to the mutex. Windows critical section for
> example keeps track which thread has lock it. This helps tracking down
> deadlocks in the debugger.

For debugging purposes, you could try creating your own `mutex` wrapper 
that keeps track of the calling `thread::id`. Note that access to this 
extra information is potentially concurrent, so you will need to 
synchronize it internally.

> I wasn't the first one who was looking for this:
> http://stackoverflow.com/questions/21892934/how-to-assert-if-a-stdmutex-is-locked

That looks like an obvious attempt to misuse the existing facilities.

Regards,
-- 
Agustín K-ballo Bergé.-
http://talesofcpp.fusionfenix.com