Re: Blocking read() policy?

From: Kasper Dupont (kasperd_at_daimi.au.dk)
Date: 07/24/05 

Date: Sun, 24 Jul 2005 11:40:54 +0200

Robert Redelmeier wrote:
>
> Kasper Dupont <kasperd@daimi.au.dk> wrote:
> > Preemption happens in entry.S and always did, those you
>
> entry.S has changed substantially, and the rescheduling check
> is very different. In 2.2.5 it is in the process' entry in
> the need_resched table. In 2.6.9 a call to the scheduler is
> in the direct return path, skipped only on different flags.

2.2.5 may call the scheduler on return to user mode,
2.6.9 may call the scheduler on return to user mode,
I don't see much of a difference.

>
> Linux' responsiveness has improved substantially.

Probably, but not so much of a difference that I
notice it when using the system.

> I believe
> quicker unblocking (at intr vs tick) has been one of the
> improvements.

That improvement happened before Linux 1.0 was
released. 

-- 
Kasper Dupont -- der bruger for meget tid på usenet.
Note to self: Don't try to allocate 256000 pages
with GFP_KERNEL on x86.

Relevant Pages

  • Re: [ck] Re: Linus 2.6.23-rc1
    ... They replaced code because he would have liked to have taken scheduler code ... He's strong in this area and weak in general kernel ... day in Linux because of people's collective lack of foresight. ... community admitted that there was a problem earlier and permitted more ...
    (Linux-Kernel)
  • Re: [OT] Interview with Con Kolivas on Linux failures
    ... Con Kolivas, the kernel hacker who authored a better scheduler, recently decided to quit. ... Loss for Linux ... The former option means you have a CPU scheduler which is difficult to model, and the behaviour is right 95% of the time and ebbs and flows in its metering out of CPU and latency. ...
    (Debian-User)
  • Re: whats the different between sched_core.c and sched_ule.c?
    ... maintains a set of KSE together, ... For Linux kernel 2.6 there were two competing ... Now, for FreeBSD, several benchmarks about threaded programs have circulated, ... And this is *not* the case of the BSD scheduler. ...
    (comp.unix.bsd.freebsd.misc)
  • Re: Unable to mount root fs on unknown-block(1,0)
    ... Sorry, the version of linux should be 2.6.15 for MPC8xx, which is included in ELDK. ... Load address: 0x100000 ... Image Name: ramdisk ... io scheduler anticipatory registered ...
    (Linux-Kernel)
  • Re: [patch 00/13] Syslets, "Threadlets", generic AIO support, v3
    ... With rescheduling there are _too_ many things to ... request in the future too]'. ... I did not run reschedulingtests with kernel thread, ... We do user-space register saving not in the scheduler ...
    (Linux-Kernel)