Re: sys_sched_yield fast path

[Davide Libenzi]

On 11-Mar-2001 Dave Zarzycki wrote:
> On Mon, 12 Mar 2001, Anton Blanchard wrote:
> 
>> Perhaps we need something like sched_yield that takes off some of
>> tsk->counter so the task with the spinlock will run earlier.
> 
> Personally speaking, I wish sched_yield() API was like so:
> 
> int sched_yield(pid_t pid);

Yes, You could do an API like this but it's not the mean of sched_yield().

> This would allow the thread wanting to acquire the spinlock to yield
> specifically to the thread holding the lock (assuming the pid of the lock
> holder was stored in the spinlock...) In fact, the the original lock owner
> could in theory yield back to the threading wanting to acquire the lock.

Everything happens inside a spinlock should be very fast otherwise the use of a
spinlock should be avoided.

- Davide