Re: Cromwell Hibernate

--- Timothy Fontaine wrote:
> On 11/8/05, Ben Dash wrote:
> > My wife was complaining about how long it took for
> my
> > xBox to boot and I was thinking this morning about
> > what it would take to implement hibernation in
> > Cromwell.
> >
> > Has anyone spent any time looking into this
> already?
> >
> > I'm no expert on what would be done during the
> > hibernate process but the concept sounds simple
> > enough. I would imagine that the contents of the
> RAM
> > would be serialized to a file inside hda55 upon
> > entering hibernation. When the xBox is rebooted
> the
> > bios could check for the existence of the
> hibernations
> > file. If the file does not exist then it could
> > proceed with a normal boot. If the hibernate file
> did
> > exist then it could be restored and so "awaken"
> the
> > xBox and take it to its fully running state almost
> > immediately.
> >
> > As I said, I am not at all sure what would be
> involved
> > for me to implement this, however, I think that it
> > would be an interesting project and not beyond the
> > realms of possibility.
> One option is to use a software suspend package like
> http://www.suspend2.net

I just had a look at suspend2 and it looks VERY
interesting.  I'm going to look into what may be
needed to port that to the xbox patched 2.6 kernel. 
It looks like it doesn't support NFS hibernation yet,
and I'm sure it doesn't support FatX hibernation, but
I'm sure it'll be possible to add.
 
> Of course another option is to fulfill the potential
> of your Xbox as a
> Console/Set-Top-Box, an area that unfortunately
> Xbox-Linux hasn't really
> tapped into. It is possible for Xbox-Linux to
> instead of using cromwell as
> another boot loader replacement, be a more
> appropriate firmware for your
> Xbox. It is a difficult task, on the outside there
> is a full meg of storage
> for a firmware, and on the inside only 256K,
> obviously this is problematic
> for the task I'm about to propose, though not wholly
> out of the question.
> 
> It is possible that Cromwell could return to its
> roots (crom) and bundle in
> a stripped down kernel (and strip out all the other
> gui/etherboot stuff).
> The kernel would only need to be responsible for
> handling initialization of
> the bare minimum of hardware (i.e. no ethernet, usb,
> or other initialization
> that can happen 'as needed') and from their copy
> straight into memory an
> image located somewhere on the harddrive that at
> it's largest would be
> 64megs (ok 128 for those advantageous enough to
> solder in more) All that's
> left is to jump (kexec?) into the environment and
> start devices on demand.
> This would more closely emulate how 'real'
> Console/Set-Top-Boxes function.
> 
> Now, I know most of the arguments against this sort
> of functionality.
> Firstly, it scares off distributions, as it's
> different from how most
> desktops work, but this is a GoodThing(tm) because
> your Xbox isn't a
> Desktop. Secondly, Linus is already little more than
> squeamish about
> including Xbox-Linux patches upstream, let alone
> after said reorganization.
> Thirdly, one already has software suspend options,
> and can trim down
> kernels, init scripts, and splash screens to more
> quickly get into Linux,
> why on earth would we do anyting so drastic? And
> finally, that sounds great
> and all, but who's going to do it? These are all
> fair points/questions, that
> indeed need to be raised.
> 
> 1) Your Xbox (Despite what the wiki may say, and
> what I may have said) is
> not a PC/Desktop, and you shouldn't treat it as
> such, it is a low memory
> Set-Top-Box for your multimedia disposal, and
> distributions have had years
> now to tie in Xbox-Linux and very few have adopted
> it, and the ones who
> tried did so with very little fanfare/success.
> 
> 2) This doesn't have to go into the main Xbox-Linux
> cromwell/kernel tree, it
> can be seperate. But also note that Linus isn't
> likely to accept Xbox-Linux
> patches upstream, just something we need to accept.
> 
> 3) Software Susped, small kernels, less services,
> and splash screens are
> simply not enough to fool most of the users that
> they are getting the most
> out of their Xbox while they wait >1min for
> everything (including X) to
> load. The point is we could/can do it faster.
> 
> 4) I'll certainly lend a hand to anyone who
> sincerely makes this effort,
> I've been behind such a project since Dash-POC (see
> cvs) was first dreamt
> up. No, my C is not at a level that would enable me
> to complete this project
> on my own. No, my Kernel and embedded systems
> knowledge is not at a level
> that would enable be to complete this project. Does
> that mean I wouldn't
> help? Of course not, this is a project I've been
> dying to see take of for
> some time.
> 
> I think this topic/idea is due for some rehashing as
> the rest of us watch
> the 360 come out and yearn for their soon to be
> anachronistic v1's to be of
> use.

The more comprehensive option sounds much better,
however, I think that I'd probably get acceptable boot
times with suspend2 and that would certainly be
simpler.  If I could cut the boot time to 20 seconds
that would be perfect.  Instant on would be great but
perhaps not possible.


	
		
__________________________________ 
Yahoo! Mail - PC Magazine Editors' Choice 2005 
http://mail.yahoo.com


-------------------------------------------------------
SF.Net email is sponsored by:
Tame your development challenges with Apache's Geronimo App Server. Download
it for free - -and be entered to win a 42" plasma tv or your very own
Sony(tm)PSP.  Click here to play: http://sourceforge.net/geronimo.php