2000-11-07 20:50:57

by Michael Rothwell

[permalink] [raw]
Subject: continuing VM madness

Should kswapd and klogd ever get "do_try_to_free_pages failed"? when
this happens my machine is destabilized, and pauses briefly from time to
time before locking up or otherwise becoming inert. This is 2.2.16+USB.

Nov 7 14:51:36 cartman kernel: VM: do_try_to_free_pages failed for
kswapd...
Nov 7 15:46:39 cartman kernel: VM: do_try_to_free_pages failed for
panel...
Nov 7 15:46:39 cartman kernel: VM: do_try_to_free_pages failed for
klogd...
Nov 7 15:46:40 cartman kernel: VM: do_try_to_free_pages failed for
identd...
Nov 7 15:46:40 cartman kernel: VM: do_try_to_free_pages failed for
mini_commander_...
Nov 7 15:46:40 cartman kernel: VM: do_try_to_free_pages failed for
gpm...
Nov 7 15:46:40 cartman kernel: VM: do_try_to_free_pages failed for
mozilla-bin...
Nov 7 15:46:40 cartman kernel: VM: do_try_to_free_pages failed for
mozilla-bin...
Nov 7 15:46:40 cartman kernel: VM: do_try_to_free_pages failed for
init...
Nov 7 15:46:40 cartman kernel: VM: do_try_to_free_pages failed for
xntpd...
Nov 7 15:46:40 cartman kernel: VM: do_try_to_free_pages failed for
nmbd...
Nov 7 15:46:40 cartman kernel: VM: do_try_to_free_pages failed for
vmware...
Nov 7 15:46:40 cartman kernel: VM: do_try_to_free_pages failed for
vmware...


2000-11-07 22:33:23

by J.A. Magallon

[permalink] [raw]
Subject: Re: continuing VM madness


On Tue, 07 Nov 2000 21:50:32 Michael Rothwell wrote:
> Should kswapd and klogd ever get "do_try_to_free_pages failed"? when
> this happens my machine is destabilized, and pauses briefly from time to
> time before locking up or otherwise becoming inert. This is 2.2.16+USB.
>
> Nov 7 14:51:36 cartman kernel: VM: do_try_to_free_pages failed for
> kswapd...
> Nov 7 15:46:39 cartman kernel: VM: do_try_to_free_pages failed for
> panel...

That seems to be the place for Andrea Arcangeli VM patch. Get it at:

http://www.kernel.org/pub/linux/kernel/people/andrea/patches/v2.2/2.2.16/mm-fix-*

Or even better, get kernel 2.2.17 and
http://www.kernel.org/pub/linux/kernel/people/alan/2.2.18pre/pre-patch-2.2.18-18.bz2
http://www.kernel.org/pub/linux/kernel/people/andrea/patches/v2.2/2.2.18pre18/VM-global-2.2.18pre18-7.bz2

and get a 2.2.18-pre18-vm, with USB support included.

There is a 2.2.18-pre20 out, but I have not still checked if
VM-global-2.2.18pre18-7.bz2
works on it. It worked for me in -pre19.

--
Juan Antonio Magallon Lacarta #> cd /pub
mailto:[email protected] #> more beer

2000-11-08 16:59:12

by Mikulas Patocka

[permalink] [raw]
Subject: Re: continuing VM madness

> > Should kswapd and klogd ever get "do_try_to_free_pages failed"? when
> > this happens my machine is destabilized, and pauses briefly from time to
> > time before locking up or otherwise becoming inert. This is 2.2.16+USB.
> >
> > Nov 7 14:51:36 cartman kernel: VM: do_try_to_free_pages failed for
> > kswapd...
> > Nov 7 15:46:39 cartman kernel: VM: do_try_to_free_pages failed for
> > panel...
>
> That seems to be the place for Andrea Arcangeli VM patch. Get it at:
>
> http://www.kernel.org/pub/linux/kernel/people/andrea/patches/v2.2/2.2.16/mm-fix-*
>
> Or even better, get kernel 2.2.17 and
> http://www.kernel.org/pub/linux/kernel/people/alan/2.2.18pre/pre-patch-2.2.18-18.bz2
> http://www.kernel.org/pub/linux/kernel/people/andrea/patches/v2.2/2.2.18pre18/VM-global-2.2.18pre18-7.bz2
>
> and get a 2.2.18-pre18-vm, with USB support included.
>
> There is a 2.2.18-pre20 out, but I have not still checked if
> VM-global-2.2.18pre18-7.bz2
> works on it. It worked for me in -pre19.

Sadly it is not a bug but a VM misdesign (and people are just making
different workarounds that more or less work). I believe that this
solution will break again, as it happened in 2.2.15 and 2.2.16.

Go back to Linux 2.0 - it has the swapper implemented correctly :-)

Mikulas


2000-11-08 17:42:31

by Alan Cox

[permalink] [raw]
Subject: Re: continuing VM madness

> Sadly it is not a bug but a VM misdesign (and people are just making
> different workarounds that more or less work). I believe that this
> solution will break again, as it happened in 2.2.15 and 2.2.16.

2.2.15->16 was the major transition in getting stuff right. 2.2.18 should be
pretty reasonable. With Andrea's additional patches its quite nice. If we
add page aging then in theory it'll be as good as 2.2 but in practice who
knows

2000-11-08 18:51:09

by Mikulas Patocka

[permalink] [raw]
Subject: Re: continuing VM madness

> > Sadly it is not a bug but a VM misdesign (and people are just making
> > different workarounds that more or less work). I believe that this
> > solution will break again, as it happened in 2.2.15 and 2.2.16.
>
> 2.2.15->16 was the major transition in getting stuff right. 2.2.18 should be
> pretty reasonable. With Andrea's additional patches its quite nice. If we
> add page aging then in theory it'll be as good as 2.2 but in practice who
> knows

What about the possibility that kernel shoots processes when the machine
is receiving too much packets and runs out of atomic memory? It didn't
seem to go away in 2.2.16. 2.0 behaved correctly in this case.

Mikulas

2000-11-08 22:13:23

by Andrea Arcangeli

[permalink] [raw]
Subject: Re: continuing VM madness

On Wed, Nov 08, 2000 at 05:42:33PM +0000, Alan Cox wrote:
> add page aging then in theory it'll be as good as 2.2 but in practice who

Agreed.

Andrea