Re: [EBS-Linux] Re: The semaphore problem w.r.t. "true background" tasks

[Peter W. <pe...@au...>]

John Lee wrote:
> On Fri, 27 Feb 2004, Peter Williams wrote:
> 
> 
>>I think that a "flag" based solution to this problem would probably be 
>>inadequate as there's no guarantee that a task only holds one semaphore 
>>at a time.  So I think that we should investigate a "count" based 
>>solution.
> 
>  
> Wouldn't a better solution to this problem be to ensure that a background 
> task running in kernel mode always get a highish priority? It can then 
> finish its work in the kernel, be given its background priority back and 
> call schedule().
> 
> I don't know whether the user_mode() macro could be used within 
> sched_normal_effective_prio(), but if it could solving this problem could 
> simply be:
> 
> static inline int sched_normal_effective_prio(task_t *p, runqueue_t *rq)
> {
>         uint32_t eps;
> 
>         if (unlikely(0 == p->cpu_rate_cap_per_share)) {
>                 if (user_mode(regs))
>                         return EBS_BGND_PRI;
> 		else
> 			return EBS_MEAN_PRI;
>         }
>         ....
> 
> Of course 'regs' is undefined above. I'll take a closer look at 
> user_mode() and struct pt_regs.

OK

BTW Should we think about doing something similar for capped tasks?  Or 
is it unnecessary?

Peter
-- 
Dr Peter Williams, Chief Scientist                pe...@au...
Aurema Pty Limited                                Tel:+61 2 9698 2322
PO Box 305, Strawberry Hills NSW 2012, Australia  Fax:+61 2 9699 9174
79 Myrtle Street, Chippendale NSW 2008, Australia http://www.aurema.com