Re: patch cow-swapin [was Re: Very bad swap bug -- 2.0, 2.1 at least] (fwd)

Petri Kaukasoina (kaukasoi@elektroni.ee.tut.fi)
Fri, 25 Sep 1998 14:43:44 +0300


On Thu, Sep 24, 1998 at 12:08:46PM +0200, Andrea Arcangeli wrote:
> Subject: Re: patch cow-swapin [was Re: Very bad swap bug -- 2.0, 2.1 at least]
>
> On Wed, 23 Sep 1998, Andrea Arcangeli wrote:
>
> >I' ll do a 2.0.x backport soon.
>
> I include in this email my patch against 2.0.36-pre9.

> You can change the bufsize and the sleep time of course but if you don' t
> leave this proggy in the background you won' t test my code at all
> because my code never run in normal conditions.

I noticed the effect in my normal conditions.

I run LPRng as my lpd server and I have always been wondering why the lpq
requests from the network cause noticable disk activity. For example if I
asked the queue status once a second from the network the disk made sounds
once a second even though I thought that everything it needs should be
cached in RAM. (I even had to remove the samba server from this machine.
When the printer manager window, or whatever it's called, was forgotten
open, the windows machine asked the printer queue once every 5 seconds which
was very irritating.)

Now, I first purged the free memory so that the RSS of lpd went zero and
then run lpq from the network. Every lpq caused disk activity. Then I
applied Andrea's patch and did the same thing. The first lpq caused disk
activity but then it was silent! Excellent!

(tried it on linux-2.0.36pre11 and pre12)

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