RE: [PATCH] Abstracted Priority Inheritance for RT

[Perez-Gonzalez, Inaky]

>From: Daniel Walker [mailto:dwalker@xxxxxxxxxx]
>On Wed, 2005-06-01 at 16:07 +0200, Esben Nielsen wrote:
>
>> Do you plan to use that callback for priority inheritance?
>> If so: It would lead to an recursive algorithm. That is not very nice
in
>> the kernel with a limited call-stack. It is not so much a problem if
the
>> mechanism is used in the kernel only, but if it is used for
user-space
>> locking, which can have unlimited neesting, it is potential problem.
>
>There is an API for for priority inheritance, the call by is strictly
>for the PI mechanism to signal when it changes a waiters priority , as
>the result of PI.
>
>It's somewhat explained in linux/pi.h . Currently the rt_mutex uses
this
>callback to move the waiter depending on it's new priority.
>
>I'm not sure I see how this could become recursive, could you explain
>more?

Maybe he is referring to the case?

A owns M
B owns N and is waiting for M
A is trying to wait for N

These deadlocking cases can be tricky during PI.

-- Inaky
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/