Re: [patch] Real-Time Preemption, -RT-2.6.12-rc1-V0.7.41-07

[Ingo Molnar <mingo@elte.hu>]

* Steven Rostedt <rostedt@goodmis.org> wrote:

> > I was going to say the opposit. I know that there are many more rt_locks
> > locks around and the fields thus will take more memory when put there but
> > I believe it is more logical to have the fields there.
> 
> It seems logical to be there, but in practicality, it's not.
> 
> The problem is that the flags represent a state of the task with 
> respect to a single lock.  When the task loses ownership of a lock, 
> the state of the task changes. But the the lock has a different state 
> at that moment (it has a new onwner).  Now when it releases the lock, 
> it might give the lock to another task, and that becomes the pending 
> owner. Now the state of the lock is the same as in the beginning. But 
> the first task needs to see this change.
> 
> You can still pull this off by testing the state of the lock and 
> compare it to the current owner, but I too like the fact that you 
> don't increase the size of the kernel statically.  There are a lot 
> more locks in the kernel than tasks on most systems. And those systems 
> that will have more tasks than locks, need a lot of memory anyway.  So 
> we only punish the big systems (that expect to be punished) and keep 
> the little guys safe.

no system is punished. Since task_struct embedds 2 locks already, moving 
the field(s) into task_struct is already a win.

	Ingo