Re: Suspend to disk

> 
> On Sun, 29 Jun 2003 [email protected] wrote:
> > I've recently got it to work somewhat reliably on a Z1A,
> > using the 2.4.21-rc4-laptop1 (from linux kernel for laptops)
> > kernel.  
> 
> Why do you say "somewhat reliably"?  What are the symptoms of the
> failures?  How often does it fail?
> 

Without shutting down those rc.d services, it was very unreliable:
about 50% of the time it would "wakeup" from suspend, but
then crash soon after.

With that suspend script, it has worked correctly 100% of the time, but
that's only over a few days of use, so maybe 6-8 "resumes".  I haven't
been using the laptop for any real work yet.  I'm worried that I will
soon discover an application that loads a module that cannot survive
being suspended/resumed.

I've found that sound (i810_audio) module has to be unloaded
and reloaded before it will work after a suspend/resume cycle.  
I'd like to do that in my script, except that "rmmod i810_audio"
fails unless I first kill any application that may be using
sound (such as gnome-volume-control). Any ideas on this?

> 
> > I had to add: "resume=/dev/hda6"
> > as a boot option to the kernel, and /dev/hda6 is my swap partition
> > which is 50% larger than total RAM.  
> 
> Was this a requirement of the swsuspend stuff or are you just reporting
> your configuration?  My GRX570 has 512MB of ram in it, but I usually run
> it without paging space.  If I decide to run VMware, then I enable a
> filesystem-based paging space prior to VMware and turn it off afterward.
> I hate having partitions reserved for something I never use. :-)
> 
> I'm betting that I can't use filesystem-based paging space, either. :-(

You need swap space about 150% of RAM, and there are some restrictions
as to what type it can be, but I dont remember what they are.  I also
didn't have a swap partition since I also didn't think it was necessary
with 512MB - I had to create one with paritition magic.


> 
> > I run this script (probably RH9 specific) to put it to sleep:
> > 
> >   umount /proc/bus/usb
> >   /etc/init.d/network stop
> >   /etc/init.d/pcmcia stop
> >   /etc/init.d/xinetd stop
> >   rmmod ehci-hcd
> >   rmmod usb-uhci
> >   echo -n 4 > /proc/acpi/sleep
> 
> It probably doesn't matter because you're just killing the daemons, but it
> would probably be best to shut down the daemons in the reverse of the
> order they were started in.  So you'd be better off with a different
> order:
> 
>     xinetd
>     pcmcia
>     network
> 
> > and this script after a sucessfull resume:
> > 
> > /etc/init.d/network start
> > /etc/init.d/pcmcia start   
> > /etc/init.d/xinetd start
> > hwclock --hctosys
> > 
Thanks - that does seem like a good principal to follow.

Mark