lkml.org 
[lkml]   [2002]   [Jun]   [19]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
Date
From
SubjectRe: Question about sched_yield()
On Thu, 20 Jun 2002, Rusty Russell wrote:

> In message <Pine.LNX.3.96.1020619072548.1119D-100000@gatekeeper.tmr.com> you wr
> ite:
> > On Wed, 19 Jun 2002, Rusty Russell wrote:
> >
> > > On Mon, 17 Jun 2002 17:46:29 -0700
> > > David Schwartz <davids@webmaster.com> wrote:
> > > > "The sched_yield() function shall force the running thread to relinquish
> the
> > > > processor until it again becomes the head of its thread list. It takes no
>
> > > > arguments."
> > >
> > > Notice how incredibly useless this definition is. It's even defined in ter
> ms
> > > of UP.
> >
> > I think you parse this differently than I, I see no reference to UP. The
> > term "the processor" clearly (to me at least) means the processor running
> > in that thread at the time of the yeild.
> >
> > The number of processors running in a single thread at any one time is an
> > integer number in the range zero to one.
>
> It's the word "until": "relinquish the processor until".
>
> It's pretty clearly implied that it's going to "unrelinquish" *the
> processor* at the end of this process.
>
> So, by your definition, it can be scheduled on another CPU before it
> becomes head of the thread list?

I have to read "head of the thread list" broadly, and assume it means the
thread will be run when it is the most appropriate thread to be run. I
don't read the wording as requiring or forbidding SMP, uni, or a strict
round-robin scheduler. The term "head of the thread list" doesn't state
that all other processes must get a time slice.

I believe I clarified my concerns in another post, I don't want to repeat
them. I don't want a process with threads contending for a resource to get
all or none of the CPU, each of which is possible with various schedulers
and patches recently available. I'd like to see threads of a single
process be able to get, use, and share a timeslice before some cpu hog
comes in and get his timeslice.

I don't read the text you quoted as requiring much more than 'run someone
else," because it's comfortably vague. To me anyway. I'm not knocking
anyone who reads it more strictly, I just don't see what you did.

--
bill davidsen <davidsen@tmr.com>
CTO, TMR Associates, Inc
Doing interesting things with little computers since 1979.

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

\
 
 \ /
  Last update: 2005-03-22 13:26    [W:0.070 / U:0.692 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site