Skip site navigation (1)Skip section navigation (2)
Date:      Sat, 28 Oct 2006 10:12:15 -0400 (EDT)
From:      Daniel Eischen <deischen@freebsd.org>
To:        David Xu <davidxu@freebsd.org>
Cc:        Paul Allen <nospam@ugcs.caltech.edu>, freebsd-current@freebsd.org, Julian Elischer <julian@elischer.org>, "Alexandre \\Sunny\\ Kovalenko" <Alex.Kovalenko@verizon.net>, current@freebsd.org
Subject:   Re: Comments on the  KSE option
Message-ID:  <Pine.GSO.4.64.0610281006430.12299@sea.ntplx.net>
In-Reply-To: <200610281255.57135.davidxu@freebsd.org>
References:  <45425D92.8060205@elischer.org> <200610281206.13588.davidxu@freebsd.org> <4542DE59.5010500@elischer.org> <200610281255.57135.davidxu@freebsd.org>

next in thread | previous in thread | raw e-mail | index | archive | help
On Sat, 28 Oct 2006, David Xu wrote:

> On Saturday 28 October 2006 12:36, Julian Elischer wrote:
>>>> Julian
>>>
>>> As you are emphasizing fairness, I must say I don't believe fairness in
>>> libpthread either,
>>
>> you mean you don't think it is a good idea or that you don't think it
>> works? (sorry, I know that your english is way better than my
>> chinese ;-)
>>
> I meant I don't think libpthread's userland scheduler + ksegrp in kernel
> has implemented fairness between threads correctly.

I think it has, at least WRT POSIX RR and FIFO scheduling.
If there are more than one ksegrps, then it depends.
POSIX says that scheduling with multiple scheduling
allocation domains (I think that is the wording) is
implementation defined.  But one of the things I want
to do is to keep threads scheduled on the same
kse/ksegrp, so there would be one run queue for
each (libpthread's version of a) kse, and threads
would stay on the same kse.

-- 
DE



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?Pine.GSO.4.64.0610281006430.12299>