RE: [ACPI] s4bios

[Grover, A. <and...@in...>]

> From: Juliusz Chroboczek [mailto:jc...@pp...] 
> AG> What I meant was that the method of hibernation was an 
> implementation
> AG> detail, and as such should not be exposed through the 
> kernel<-->user
> AG> interface.
> 
> Isn't that a policy issue that should be handled by ``the'' userspace
> power-management daemon?

> Suspension does require a bunch of userspace operations (saving the
> clock, etc.), so you want to request suspension to the daemon, not
> directly to the kernel.

Yes you're absolutely right. So the interface I'm talking about is how the
daemon tells the kernel to go to S4.

The alternatives I see are 1) there is some way for the daemon to find out
what S4 implementations exist and then write the correct value to
/proc/acpi/sleep 2) Try them in order until one works or 3) have only 1
possible value to write regardless of S4 type, and the kernel does the right
thing.

In all 3 cases I don't see how the kernel can help but have knowledge about
which S4 states work on a given system. Given that, we might as well do #3,
which makes things a little easier for the daemon.

> (Now what I'd like to know: who is ``the'' userspace daemon today?)

While it's rather broken at the moment, the plan is to have ospmd fulfill
that role.

Regards -- Andy