Re: schedule()

[Davide Libenzi]

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:
> >
> > > 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.
> >
> >
>
> Can't find it in any header file.
>
> Script started on Tue Feb 26 15:45:06 2002
> # grep "sys_sched_yield()" /usr/src/linux/include/linux/*.h
> # grep "sys_sched_yield()" /usr/src/linux/include/asm/*.h
> You have new mail in /var/spool/mail/root
> # exit
> exit
>
> Script done on Tue Feb 26 15:46:46 2002
>
> I'll #define a macro and be done with it.

It'd be too easy if it'd be inside an header file :-)

asmlinkage long sys_sched_yield(void);

in kernel/sched.c

- Davide