Re: [2.6.8-rc2-mm2] swsusp results on a hp compaq nx7000

Re: [2.6.8-rc2-mm2] swsusp results on a hp compaq nx7000

[crow]

On Fri, Aug 06, 2004 at 10:11:07PM +0200, Pavel Machek wrote:
> Hi!
> 
> > trying to get swsusp running on my hp compaq nx7000, I've noticed
> > following problems:
> > * echo 'platform' > /sys/power/disk
> >   creates a syslog message: switching to platform, but a
> >   cat /sys/power/disk
> >   still says firmware. Other comments regard disk-writes by 'shutdown'
> >   driver
> > * system calls acpi_*_shutdown to shutdown, but it's rebooting instead
> >   of doing a system shutdown
> > * locking with regard to preemption seems so be broken
> > * ohci1394 seems to generate sporadic OOPs on resume (could be
> >   preemption related)
> > * synaptics touchpad (PS/2 driver) stops working after resume
> >   (sporadic, could also be preemption related)
> > * the AGP port driver suspends, but system is a lot slower (e.g. xterm
> >   scrolling) after resuming
> > * enabled DRI drivers (radeon) doesn't work. I'm able to move the mouse
> >   after resuming for a short time, after that the system locks up hard.
> >   (the screen is generating snow crash even while moving the mouse).
> >   The same behaviour can be observered when suspending/resuming from
> >   vt, the computer crashes with the same symptoms after switchting to X
> >   after resuming. This happens with the in kernel drm driver as with
> >   the latest x.org dri drivers (X in question is x.org 6.7).
> > * radeonfb gives a "radeonfb: resumed" message on suspending. This may
> >   be correct (if you suspend the driver and resume it afterwards to
> >   display more on-screen), but it is a rather disturbing message.
> > 
> > After all swsusp is a lot more useable compared to last time ?'ve tested
> > it. It is useable as "suspend if battery power gets low" - measure, but
> > rather unuseable for me for other things, as (normal 2D-) graphics
> > operations crawl after resuming. thanks for the good work.
> > 
> Ugh, I was offline for a week and it will take me few days to catch
> up. This is valuable feedback, and you probably do want to mail it to
> lkml and Patrick Mochel.
> 
> If you have time results with preempt disabled would be nice,
> too.

After i got some unreproduceable hangups on system boot I disabled preempt.
It may take some time until some of the problems (i.e. ohci1394) resurface.

> psmouse.proto=bare might help with synaptics.

will try that soon, but it rather seems to be some kind of strange interaction
with other modules..

	--Andreas

Re: [2.6.8-rc2-mm2] swsusp results on a hp compaq nx7000

[Pavel Machek]

Hi!

> > > * locking with regard to preemption seems so be broken

I see it here, too.

> > > * ohci1394 seems to generate sporadic OOPs on resume (could be
> > >   preemption related)

I do not have firewire device to test with... There seem to be very
little of those beasts around, so I propose to ignore firewire for
now.

> > > * radeonfb gives a "radeonfb: resumed" message on suspending. This may
> > >   be correct (if you suspend the driver and resume it afterwards to
> > >   display more on-screen), but it is a rather disturbing message.

It is correct (but also completely useless) message. Feel free to
sumbit a patch to remove it (through radeonfb people).
								Pavel
-- 
People were complaining that M$ turns users into beta-testers...
...jr ghea gurz vagb qrirybcref, naq gurl frrz gb yvxr vg gung jnl!

Re: [2.6.8-rc2-mm2] swsusp results on a hp compaq nx7000

[Nigel Cunningham]

Hi.

On Mon, 2004-08-09 at 04:24, Pavel Machek wrote:
> Hi!
> 
> > > > * locking with regard to preemption seems so be broken
> 
> I see it here, too.
> 
> > > > * ohci1394 seems to generate sporadic OOPs on resume (could be
> > > >   preemption related)
> 
> I do not have firewire device to test with... There seem to be very
> little of those beasts around, so I propose to ignore firewire for
> now.

I have firewire hardware but no actual devices to plug in. Is that any
help at all when it comes to testing? (I'm not building any support in
or as modules at the moment).

Nigel

Re: [2.6.8-rc2-mm2] swsusp results on a hp compaq nx7000

[Pavel Machek]

Hi!

> > > > > * locking with regard to preemption seems so be broken
> > 
> > I see it here, too.
> > 
> > > > > * ohci1394 seems to generate sporadic OOPs on resume (could be
> > > > >   preemption related)
> > 
> > I do not have firewire device to test with... There seem to be very
> > little of those beasts around, so I propose to ignore firewire for
> > now.
> 
> I have firewire hardware but no actual devices to plug in. Is that any
> help at all when it comes to testing? (I'm not building any support in
> or as modules at the moment).

It seems to be broken even in that situation, but as firewire hardware
is pretty uncommon... I guess we have bigger problems to solve.

								Pavel
-- 
People were complaining that M$ turns users into beta-testers...
...jr ghea gurz vagb qrirybcref, naq gurl frrz gb yvxr vg gung jnl!