[Fwd: Re: [Suspend2-devel] Extent of suspend2 interactions with other kernel systems?]

classic Classic list List threaded Threaded
1 message Options
Reply | Threaded
Open this post in threaded view
|

[Fwd: Re: [Suspend2-devel] Extent of suspend2 interactions with other kernel systems?]

Karl Hegbloom
FYI...  ;-)

There is a person on the Suspend2 development list (Bojan Smojver) who
is patching a Rawhide kernel 2.6.14-1.1709_FC5 kernel.  I don't know if
he's with the Fedora project or not.  He says that kernel is based on
2.6.15-rc2-git3.

Nigel Cunningham has been posting incremental patches lately, perhaps
intending to make it easier to maintain a kernel source containing the
patches?

I'm finding that the 'hibernate' script works better than the
'acpi-support' hibernate and sleep does.  My Athlon-64 workstation will
hibernate to RAM only with the 'hibernate' script; it fails with stock
'acpi-support' scripts.  I really think that Ubuntu should be using it.
It works with ACPI, the present swsusp, and with the Suspend2 patch.

Also see:

 http://wiki.suspend2.net/DistroAndHardwareSetup/Ubuntu_Breezy_Badger

 http://lists.suspend2.net/lurker/message/20051030.144541.02493640.en.html

 http://www.laclinux.com/~karlheg/experimental/suspend2/?F=0&C=N&O=A&V=1&P=*&X=Go

 Suspend2 User-UI for Usplash (in development):

  http://lists.suspend2.net/lurker/message/20051031.223703.7764eb56.en.html

--
Karl Hegbloom <[hidden email]>

Hi Karl.

On Wed, 2005-11-23 at 23:50, Karl Hegbloom wrote:
> If I have a kernel source, and want to apply the suspend2 patch along
> with some arbitrary set of other patches, what do I need to know wrt how
> suspend2 interacts with other kernel subsystems?

Not much. The main changes that affect other subsystems relate to
freezing kernel threads. Suspend has a different freezer implementation
to the one in the unpatched kernel. It essentially involves modifying
calls to try_to_freeze into try_todo_list, and modifying the creation of
workqueues and kthreads that need to run during suspend (there aren't
many of them) to use nofreeze variants.

> What kinds of patches will need to be fixed up and in what way?  That
> is, patches that affect subsystem X could potentially Z unless you
> ensure that A, B, and C.
>
> What are the potential conflicts or pitfalls that could occur?
>
> Where can I read about each thing in the responses to the above?
>
> When will suspend2 be available via a git repository?  Please?

My general plan is:

- Add last new functionality to Suspend2 (done)
- Bug fix until the only reported problems relate to drivers (in
progress)
- Finish preparing the git tree (in progress)
- Seek to get Suspend2 into mm (quite possibly involving another round
of modifications to make X, Y and Z happy).

> I really want to _lobby_ for (and potentially help work on) getting
> Suspend2 into Ubuntu "Dapper Drake", because I know it works better and
> on a wider variety of hardware than the swsusp system.  I know that the
> debate over doing so will tend to come down to Maintainability and
> Stability.

Which is why I'm going for bug fixing too.

> Having the tree in Git will certainly make that maintenance into a much
> more natural process.  It's very difficult to maintain a kernel source
> package where you are continually applying and fixing up a huge patch
> like Suspend2.  How do you upgrade the patch, for instance?  Start over
> with fresh kernel source each time?  That's just totally impossible to
> manage without having Suspend2 in Git, so that incremental changesets
> can be merged at will.  On the stability front, Git would allow a stable
> and development branch to be maintained more easily, so that the
> distributions can ship with known stable releases of Suspend2 in their
> kernels.
I'm hoping that with 2.2, development will pretty much stop. I don't see
anything major that still needs to be added, and so expect the majority
of future changes to be fixing bugs and adapting to new kernel versions.
Oh, I would like to reinstate support for building as modules, but that
shouldn't be a big change since we've had it in the past.

Regards,

Nigel


--
kernel-team mailing list
[hidden email]
http://lists.ubuntu.com/mailman/listinfo/kernel-team