[linux-pm] [Fwd: Re: PM messages]

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



With the correct list address this time...

BTW, When do we plan the next irc meeting ?

Ben.

-----Forwarded Message-----
From: Benjamin Herrenschmidt <benh@xxxxxxxxxxxxxxxxxxx>
To: Pavel Machek <pavel@xxxxxx>, linux-pm-request@xxxxxxxxxxxxxx
Cc: David Brownell <david-b@xxxxxxxxxxx>, Patrick Mochel <mochel@xxxxxxxxxxxxxxxxxx>, Paul Mackerras <paulus@xxxxxxxxx>
Subject: Re: PM messages
Date: Sun, 17 Oct 2004 19:27:17 +1000

On Sun, 2004-10-17 at 19:17, Pavel Machek wrote:

> Hmm, it is actually pretty nasty do powerdown devices between freeze
> and unfreeze... As long as you are not using modules, you can count on
> hardware being initialized by image being booted.. That does not work
> with modules :-(.

What do you mean ? I don't understand your point with modules.

In some cases, devices will be "left" from a frozen state and will
be "recovered" from a power-on-reset state, in some others (like
BIOS assisted S4), I suspect some device may still be in the sleeping
state when the unfreeze gets there.

Drivers will have to deal with that, but it may be of some help to
them if we could give some more precise indication at resume() time
of what actually happened: Is this a normal "live" unfreeze or is this
the real resume after suspend. I don't have a device that actually
cares at this point, so I'm not talking about a real life scenario.

I can make radeonfb figure out the HW state (or rather, I can get it
reprogram the mode on every resume whatever state we are coming from,
and it can 'test' if the chip needs a full low level re-initialization
or is in D2 state if necessary. But some indication could be useful
avoiding unnecessary steps. Maybe that can stay part of the "details"
thing we talked about tho.

On another note, while I do want a struct to be passed as the PM
message, I'd like that struct to contain a scalar, since I want it to
be easy for drivers to use switch/case constructs. I can foresee
various cases where the drivers will fold 2 cases in one (like do the
same thing for freeze/idle/suspend, or do the same thing for idle/suspend,
same goes if we ever add this parameter to resume()). I don't like
piling up if/else statements ...

Ben.
-- 
Benjamin Herrenschmidt <benh@xxxxxxxxxxxxxxxxxxx>



[Index of Archives]     [Linux ACPI]     [Netdev]     [Ethernet Bridging]     [Linux Wireless]     [CPU Freq]     [Kernel Newbies]     [Fedora Kernel]     [Security]     [Linux for Hams]     [Netfilter]     [Bugtraq]     [Yosemite News]     [MIPS Linux]     [ARM Linux]     [Linux RAID]     [Linux Admin]     [Samba]

  Powered by Linux