Re: schedule()

[Davide Libenzi <davidel@xmailserver.org>]

On Tue, 26 Feb 2002, Richard B. Johnson wrote:

> On Tue, 26 Feb 2002, Davide Libenzi wrote:
>
> > On Tue, 26 Feb 2002, Richard B. Johnson wrote:
> >
> > >
> > >
> > > I just read on this list that:
> > >
> > >     while(something)
> > >     {
> > >       current->policy |= SCHED_YIELD;
> > >       schedule();
> > >     }
> > >
> > > Will no longer be allowed in a kernel module! If this is true, how
> > > do I loop, waiting for a bit in a port, without wasting CPU time?
> > >
> > > A lot of hardware does not generate interrupts upon a condition,
> > > there is no CPU activity that could send a wake_up_interruptible()
> > > to something sleeping.
> > >
> > > For instance, I need to write data to a hardware FIFO, one long-word
> > > at a time, but I can't just write. I have to wait for a bit to be
> > > set or reset for each and every write. I'm going to be burning a
> > > lot of CPU cycles if I can't schedule() while the trickle-down-effect
> > > of the hardware is happening.
> >
> > What did it do yield() to you ? Doesn't it work for your case ?
> >
> There isn't one in 2.4.x  I'll modify my drivers to use YIELD
> and #define it depending upon version.

In 2.5 yield() maps to sys_sched_yield(). You can handle it in the same
way in your includes if version <= 2.4.




- Davide