Re: the Fremantle Porting Task Force, or "how to run maemo on Neo900"
 

The kernel has changed enormously. Vast swathes of low level functionality simply will not work without heavy kernel hacking. Many of the Nokia n900 kernel drivers simply will not work, or maybe even build, on a more modern kernel. Much of the work might be to put back old kernel interfaces, or even bugs, for userspace compatibility too. Selecting the best kernel to use is a very difficult task.

[OT: N9 stuck on .32, because that was absolutely current when we were working on it. Early N9 work was done on .29, as .32 didn't even exist yet. I tried to persuade them to let me move forwards to .33, as that was also LTS, but, even though the .29->.32 migration was described as the cleanest kernel migration ever in any of the projects, they were too scared to permit another one, as we were doing too many invasive things with the kernel, and they were risk-averse.]

Re: the Fremantle Porting Task Force, or "how to run maemo on Neo900"
 

For sure there are lots of changes in the kernel, but I see no reason why we shouldn't even try. After all we have KP and git bisect as a reference if something doesn't work.

The upstreaming efforts of important parts of N900 drivers seem to finally give fruits[0,1], and if we manage to do it right, with the incoming DT n900 will be supported upstream for as long as OMAP3 is supported IIUC.

If I correctly understand the kernel code, clock and regulator frameworks should give the same if not better results in power management. And if it is worse somewhere, well, we still have that reference ^^^ :)

I am not saying it is easy (it took me a month and a bisect 3.5<->3.8 to find why SSI doesn't work with linux >=3.7 ) but I feel it worth's the time if we manage to have fremantle on top of a recent kernel at the end.

So far we have fremantle booting on top of 3.12rc1 with the following HW working - DSP, DSS2(though there are some problems with the framebuffer), SSI(modem+GPS), SGX (with stock n900 drivers and HW acceleration), touchscreen(disabling doesn't work), bluetooth, wifi, keyboard, most of the sensors and LEDs, charging, HSMMC, etc.

Cameras doesn't work (drivers are loaded but no image could be captured) as is audio (no sound but the driver loads)


We just lack the manpower to speedup the porting process ;)

[0] https://lkml.org/lkml/2013/10/6/127
[1] https://lkml.org/lkml/2013/9/8/33

EDIT:
I talk n900 here, as I have no other piece of HW to test fremantle with recent kernels :)

Re: the Fremantle Porting Task Force, or "how to run maemo on Neo900"
 

great news!

Re: the Fremantle Porting Task Force, or "how to run maemo on Neo900"
 

I have been investigating the pulseaudio stuff and I have yet to figure out just whats going on in pulseaudio-nokia or how we can get module-nokia-voice to do what we need it to for the cellular modem audio.

One idea suggested was to pretend that the Option modem audio was actually VoIP. What is unknown is:
1.How pulseaudio-nokia-voice is told "hey this is cellular audio, please pass it through the cmtspeech path"
2.What we need to do to the Fremantle system so that the stock dialer (when making a phone call through telepathy-ring) wont trigger anything that causes the cmtspeech path to activate and will instead feed all its audio through the VoIP path.
and 3.How the audio is processed (in the VoIP path and the cmtspeech path), whether the cmtspeech path does processing the VoIP path does not (and whether that processing is processing we need) and whether any processing that gets done (in either path) that (due to processing done by the Option modem) is processing we dont need (or worse, is going to get in the way)

There are also questions about whether there is anything in the pulseaudio blobs that is somehow dependent on the specifics and specifications of the audio hardware of the N900 (audio chip, speaker amplifier, speakers, headphone amplifier, FM transmitter, microphone, bluetooth hardware, FM receiver or otherwise) and whether using different parts in that situation will have issues.

Re: the Fremantle Porting Task Force, or "how to run maemo on Neo900"
 

a nice (though dusty and aged) page: http://wiki.maemo.org/Free_Maemo

Re: the Fremantle Porting Task Force, or "how to run maemo on Neo900"
 

Yeah its nice although I dont think there is any info on that page that isn't covered by http://wiki.maemo.org/Fremantle_closed_packages

Re: the Fremantle Porting Task Force, or "how to run maemo on Neo900"
 

Has anyone done any testing on hardware?
Could someone provide me with a Maemo uImage?? I do not have my build environment anymore, need to redo it.

If you have any files you want tested, let me know, Just need a kernel (uImage)

I have taken a run at getting Maemo (from a nand dump) to run on at DM3730. I could not get the kernel to boot (zImage), I was using boot scripts from Ubuntu. Kernel >=3.0

Device is Open Pandora 1GHZ. Good device and fully open (Graphics drivers are closed IIRC).

Device Specs:
DM3730 @1 GHz with u-boot, HWKB, touch screen @ 800*480 I can test on. (RAM is at 512MB, but no real diff.)

Edit: I'm going to work on bringing the Pre-Alpha over, I didn't want to use that (SDK). Will try get Pre-Alph to boot on Pandora and then go from there...

Re: the Fremantle Porting Task Force, or "how to run maemo on Neo900"
 

See my post above, for the "real hardware"

Maybe it is better to join #maemo-ssu and ping me or Pali, as I am not sure what uImage you need, etc. Or you can try to play with the kernel on https://gitorious.org/linux-n900/lin...be382baa82268: and build your own kernel.

In order to boot Maemo with recent kernels, you'll need all the packages from cssu-devel repo, as well as some other fixes.

Re: the Fremantle Porting Task Force, or "how to run maemo on Neo900"
 

Thanks mate.
All good on the uImage.
I have the kernel, and I have a few of the old ones I made a few years back.
I'm going to start by trying to get the Pre-Alpha to boot on the DM3730 and if I can get to a prompt or better, I'll go from there.

Re: the Fremantle Porting Task Force, or "how to run maemo on Neo900"
 

The N900 has strong similaties to the Commodore 64, it simply refuses to die.
And I like that, sometimes a new machine comes up that just stands out so much that its users never abandon it.
Sorry if I'm off topic here, but I hope you guys will succeed with the hardware work, the porting and not the least the donations.
All respect to the work you put in to the Maemo and N900 community!