Linux Kernel Source


Squidge

Certified Guru
Joined
Nov 16, 2003
Messages
8,493
Location
UK
Website
Visit site
Well, it seems that GPH don't want to release the kernel source to the GP2X, this is what Craigix had to say:

they won't give up the kernal, its like its a secret recepie of 11 herbs and spices or something

I've not looked in detail, but this may give us some useful information.
http://lxr.post-tech.com/source/linux-mmsp...linux-mmsp2dtk3

Eg.

/*
* drivers/sound/mp2520f-audio.c -- audio interface for the MP2520F
*
* © Copyright 2004
* DIGNSYS Inc. < www.dignsys.com >
* Kane Ahn < hbahn@dignsys.com >
*
* This program is free software; you can redistribute it and/or modify
* it under the terms of the GNU General Public License version 2 as
* published by the Free Software Foundation.
*
*/

Edit: I've now checked this out a bit more, and found certain code like the display initialisation, and it certainly seems to match the datasheet. So I think this is what I'm goin to start of with.
 
Hmm,

Thinking about this, what would people say if I grabbed the source and popped it into CVS somewhere so we could keep some sort of tabs on hacks to it and make it easy to grab a local copy? It's not like it's not GPL so we can fork it in effect and do what we want :D.
 
To be honest, since Dignsys & Mizi (the two main porters of Linux to MMSP it seems) have both GPL'd there code all the way apart from the register definitions, wouldn't it be better if our own SDK (which I'll be writing shortly after obtaining hardware) would also be GPL'd, but missing a few include files for register definitions/etc ? After all, the NDA covers the registers and what they do, so as long as you don't give out the source which tells people what register is at what address, it should be ok.

I'm mirroring a local copy now, will zip it up when it's done and post the linky here. You can then throw it in cvs if you really want to. Might make it easier to upgrade to Linux 2.6, which is far more suited to an embedded environment than 2.4.
 
Squidge posted on Sep 13 2005 at 11:44 AM said:
Might make it easier to upgrade to Linux 2.6, which is far more suited to an embedded environment than 2.4.

Thats the plan, upping to 2.6 is not a 1 man job so CVS or such would make a group push a reality (maybe) ;) and would 'show willing' to the Linux fanboys ;) not that I think Linux is a better option to real H/W, just a way to learn the H/W.
 
Last edited by a moderator:
Lets cut to the chase ..

We cannot support the device _at all_ if they're going to screw with the GPL. They want to use the Linux kernel, they have to follow the rules.. otherwise its essentially theft.. and we cannot support such silliness. (As software developers, don't we have the obligation to support each other?)

One thought is to just replace their kernel with a third party one.. but thats ignoring the larger problem .. like building on top of a broken foundation. I don't like it.

Can we stop the stupidity from spreading before it really bites us in the rear?

ie: Theres 100% no reason for them not to release source -- any custom code of theirs they could plop into a closed source kernel module and load it up; putting their code into the kernel proper isn't an excuse to break the law.

jeff

(Maybe I'm taking it too seriously, but I'm not the only one whose had his code stolen before, right? :)
 
GPH clearly are not the only people doing it...

Tritton Technologies NAS120 - contains a modified Linux kernel, and they refuse to provide the source for the modified kernel. They claim that the GPL does not
apply to those changes (No, it doesn't use modules)

Macsense HomePod - "I am sorry. The linux kernel for the HomePod is not available. The kernel is confidential and will not be made public. "

Compex Wireless Products

Inventel DW 200 wireless/ADSL router

Sweex DSL router

TRENDnet TEW-410APB

Game Park Holdings GP2X :)

and lots of others...

So, either we hassle GPH for the sources (unlikely, as Magiceyes don't allow a download either unless you sign an NDA), or we make something ourselves and ignore Linux completely, hacking the hardware directly.

I'm for the last option :)
 
Jeff,

Nope you not the only one to be ripped off in the past :angry:. I don't like GPH's stance any more then you do but the link Squidge posted IS to the kernel that is almost certainly used on the device (2.4.25 with low level BSP for the MP2520F).

Everything has to be taken with a pinch of salt until release, benefit of the doubt and all.

I see no good reason that GPH did not use the kernel produced by Dignsys, Mizi and Magic Eye as stock. That is GPL and the code is available to look at from the links above (with a full GPL header and corresponding copyleft etc.). Not saying it will build but it does look to be mostly code complete (missing some headers I think :eek:)

I am holding a lot of stuff close to my chest until I see real hardware (or at least hear about it) and what it comes bundled with, the only things I can see needing to really be closed modules are SD/SDIO support and maybe registers headers etc. (as there NDA from the chip maker, god knows why like anybody is going to rip it off for another chip maker).

The rest of it is GPL or linked to GPL by definition and I would expect GPH to release at least a patch to the above source if they make ANY changes to it outside linking loadable binary only modules and I believe it would be reasonable to make a statement to that effect.

If GPH want an open source console then they could do a lot worse then take there head out of the sand and support it as such. The same applies to any GPL’ed code used in the creation of the interface, toolsets, that mplayer hack etc. If they dont staticly link with GPL'ed code or use it in a executable then fair enough, they have no reason to release the source but if they do they should.

I am not ready to write it off just yet and the fact the kernel code is available publicly gives me some faith that it’s just some dumb fucks afraid of an NDA or not really understanding the questions ;). Time will tell.

My thoughts on CVS are simple, get the source somewhere public, open and publicised (say sourceforge) and use it going forward as a support side for the board. All I want to do is bring the code together in one place, nothing in the GPL says you have to make the code easy to find or use, just that it has to be avalable on request ;). If GPH did just use the stock kernel then there would be no need for them to explicitly release there source just tell us where it is, in this case indexed by LRX and worry about the GPL status of the rest of there stuff.

Over and out for now :blink: (man, I am tired).

Edit: If it sounds pro GPH it's not, I don't like the way there are even 'thinking' of trying to worm out of the GPL.
 
Agreed on all counts, but it is worrying; still, we needn't worry until we get it into our sweaty little hands and bust it open :)

I do think it amusing that the first thing Squidge wants to do is toss the OS anyway ;) Me, I'm rather fond of having a (slim) OS. JennyCraigOS :)

jeff
 
Me, I want to use the Linux source as a basis for 2.6 (for some other people to tackle, happy to do some ground work) and as a basis to make a BSP for RTEMS.

Become rather fond of it (Another World on the GP32 runs on top of it) and it can do some interesting things with 2 CPU cores, even if one does not have an MMU (or so I am told ;)). GPH's stance worries me but the code is ‘mostly’ available and correctly licensed (and found by Googleing MP2520F) and as Squidge says that's a fair bit better then some, we need a few more high profile Linksys type cases to make people realise the GPL is not a licence to rip code.
 
OS's suck ;) I've just been passed the ported gnuboy source, and damn, it's ugly. At least there's something to learn when hacking the hardware :)
 
Squidge posted on Sep 13 2005 at 09:33 PM said:
OS's suck ;) I've just been passed the ported gnuboy source, and damn, it's ugly. At least there's something to learn when hacking the hardware :)

Care to share?, I had a nasty feeling it would be a real hack-job.
 
Last edited by a moderator:
Squidge posted on Sep 13 2005 at 10:08 PM said:
It's not a hack-job as such, more of a "I can't see anything that's been changed".

http://ibiblio.org/paulc/gnuboy.tgz

Couldn't that mean that it isn't a hack job?

What if they just took the standard Linux SDL version, compiled it and it worked straight off? :lol:

Hey - I'm an eternal optimist!
Is there a binary that goes with it?
*edit* - is that 109k fbgnuboy the binary? Does it run on a GP2X?
 
Last edited by a moderator:
DJWillis posted on Sep 13 2005 at 08:51 AM said:
It's there ;).

http://lxr.post-tech.com/source/linux-mmsp...linux-mmsp2dtk3

Makes for a very interesting read. This combined with the documents we have starts to make things look a fair bit more informed :D. As usual GamePark leave the real work to the hobbyists :rolleyes:.

This is nice..!! Unfortunately some header files seems to be missing at least in this site.
 
Last edited by a moderator:
This is the only site I've found the stuff at. The missing header files just seem to be dedicated to the MMSP2 (like the register definitions), but we have these anyway, so it doesn't really matter :)

So many things to play with - the wait for hardware is killing me :(
 
Thanks for the gnuBoy source.

On the kernel front, the missing headers do seem to be the sort if thing that could be rebuilt from the docs but without hardware there is no reason to try yet. Still, all documentation is good documentation ;).
 
/me want’s to cross-compile ScummVM against SDL for a laugh and see what happens.

Hint: I would LOVE to be pleasantly surprised :D. Actually for POC work it's potentially freaking amazing, speed aside.

/me waits to be disappointed ;).

Hmmm, what about any hacks for key mapping? Controls etc. I guess you could pre-patch SDL to map what you want before you build.
 
Back
Top