2007/2/11, Aubrey Li <[EMAIL PROTECTED]>:
On 2/11/07, Matvejchikov Ilya <[EMAIL PROTECTED]> wrote:
> 2007/2/9, Aubrey Li <[EMAIL PROTECTED]>:
> > On 2/9/07, Matvejchikov Ilya <[EMAIL PROTECTED]> wrote:
> > > 2007/2/9, Aubrey Li <[EMAIL PROTECTED]>:
> > > > On 2/9/07, Matvejchikov Ilya <[EMAIL PROTECTED]> wrote:
> > > > > Good Day!
> > > > >
> > > > > We use the linux 2.4.32-uc0 on M68360 based board. Would it be safety
> > > > > to disable kswapd in this configuration and what who can tell what
> > > > > lies beyond.
> > > > >
> > > > Kswapd is responsible for periodic memory reclaiming. It can't be
disabled.
> > > >
> > > > -Aubrey
> > > >
> > >
> > > Strange as it may seem, but I don't have any problems with
> > > applications (telnetd for ex.)... System memory gets freed after
> > > telnet session has quit.
> >
> > You misunderstood. kswapd startup for periodic memory page reclaiming
> > only occurs when the system memory is low.
> >
>
> I see your point. But I can't understand in which way any memory
> reclaiming can take place in the swapless system.
Certainly it does. There are a few types of memory which can be reclaimed.
For example, page cache, disk buffer, inode cache, slab object, etc.
> As I see it if memory has been allocated with kmalloc it can be freed only
with
> kfree.
There are much of allocation which will get memory from buddy zone directly.
OK, I take on trust it.. :)
> And the main point is that the thing which makes me think about
> disabling kswapd is that my system begins working with greater
> performance (at least in userspace).
>
That's Interesting. How did you disable kswapd? what made you think
performance is improved?
I've just put the 'return' command instead of kspwad thread
initialization (mmnommu/vmscan.c). It makes the system boot process
proceed slightly more faster. Various userspace applications works
little faster than before too. What I've doing wrong? :)
Thanks.
_______________________________________________
uClinux-dev mailing list
[email protected]
http://mailman.uclinux.org/mailman/listinfo/uclinux-dev
This message was resent by [email protected]
To unsubscribe see:
http://mailman.uclinux.org/mailman/options/uclinux-dev
