Skip site navigation (1)Skip section navigation (2)
Date:      Fri, 05 Nov 2004 11:42:14 -0800
From:      Julian Elischer <julian@elischer.org>
To:        Alexander Leidinger <Alexander@Leidinger.net>
Cc:        current@freebsd.org
Subject:   Re: cvs commit: src/sys/sys buf.h
Message-ID:  <418BD796.3040706@elischer.org>
In-Reply-To: <1099650992.418b57b01c33b@netchild.homeip.net>
References:  <40541.1099561211@critter.freebsd.dk> <1099650992.418b57b01c33b@netchild.homeip.net>

next in thread | previous in thread | raw e-mail | index | archive | help


Alexander Leidinger wrote:

>Zitat von Poul-Henning Kamp <phk@phk.freebsd.dk>:
>
>  
>
>>We've reached a major milestone in the buf-junta work:  struct buf
>>is no longer an I/O request.
>>    
>>
>
>Do you have an outline where this heads to and why?
>

when systems were smaller the number of cached bufs was small, and bufs
represented 'buffers' likely to used soon fo rIO and IO requests were 
simple,
it made sence to combine the IO request and the storage descriptor (buf).

Since then, storage is done via the vm system, IO requests have gotten 
bigger,
and the number of IO requests needed at any time has remained small. it 
makes less
sense to have an IO request with every buf storage descriptor.
It also opens the way for:
changing the io descriptor to suit speical cases without having to worry 
about the buffer cache stuff.
addign more stuff to teh IO request without worrying about bloating teh 
buffer cache stuff.

>
>Bye,
>Alexander.
>
>  
>



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?418BD796.3040706>