Suspend2 (Dagobah's packages)

[Florian Diesch]

Matthew Garrett <mjg59 at srcf.ucam.org> wrote:

> On Tue, Jul 25, 2006 at 09:06:23PM -0600, Andrew Jorgensen wrote:
>
>> I'm really puzzled by this response.  suspend2 is significantly faster
>> than the current suspend (due to on the fly compression).  I've been
>> told before that the reason we won't do it is because it isn't
>> supported on all architectures but according to their website it's
>> supported on everything that the current suspend is supported on.
>
> It's a large body of code that isn't accepted by the upstream kernel 
> developers, and we don't have anyone with enough time to work through 
> the codebase and audit it ourselves. suspend2 touches a /lot/ of the 
> kernel, and risking it interacting badly with one of our other kernel 
> changes isn't worth it if all we get in return is some speed.

On my laptop[1] hibernating with the Ubuntu kernel means sitting for
about 15 minutes in front of a blank screen hoping it's not crashed and
no possibility to cancel.  Actually it was the first time now I did a
full hibernation cycle with the Ubuntu kernel as I just did a power-off
when it wasn't ready after 5 minutes when I tried it first after
installing Dapper. Is it worth a bug report?

Hibernating with suspend2[2] means waiting 55 seconds seeing a progress bar
with some messages what it is doing now and the possibility to press
<Esc> to cancel hibernation.




[1] Thinkpad R40-2722, Pentium M 1300, 768 MB RAM
[2] Vanilla 2.6.16.16 with suspend2-2.2.5-for-2.6.16.9 from http://suspend2.net


   Florian
-- 
<http://www.florian-diesch.de/>