Re: Patch for block write clustering

Andi Kleen (ak@muc.de)
06 Mar 1998 17:25:53 +0100


Rik van Riel <H.H.vanRiel@fys.ruu.nl> writes:

> On 5 Mar 1998, Matthias Urlichs wrote:
>
> > Rik van Riel <H.H.vanRiel@fys.ruu.nl> writes:
> > > Kswapd doesn't do:
> > > - swapins
> > > - reading of files/metadata
> >
> > I wonder why you would want to. The process waiting for these things is
> > blocking on them anyway. It makes no sense to incur the additional cost of
> > a context switch.
>
> It would give us the possibility of:
> - async I/O
> - easier I/O prioritizing (block badblocks -w for a while)
> - better request reordering (since we now program the sorting
> algorithm in it's own thread)
> - blocking a swapin when memory is tight (we first have to
> free some, so let the process wait a while)

.. See drivers/block.new in the vger tree.

-Andi

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu