WTF, there is no DSP driver here?!? No video HW acceleration, very smart.

Now I looked on meego kernel: http://meego.gitorious.org/meego-dev...on/n900_kernel

I tried it boot on maemo but it failed. First problem is with getbootstate. It return MALF state.

Now I have reverse enginnered maemo version of getbootstate, so I added bootreason and component-version entries to /proc. But n900 still reboot.

Can anybody help me? Which components/special functions was removed in meego .37 kernel?

closed drivers maybe, like:

batterydriver
modem
gfx driver

there are no closed drivers in kernel space

There are a number of interfaces in the 2.6.28 kernel that have been removed and are no longer present in newer kernels. As a result, closed source components that are dependent upon these will not operate if you attempt to upgrade.

As for the graphics driver, its kernel stubs have always been open (by necessity) and all the good stuff is in a closed userspace blob.

bme will likely break and we'll need MeeGo's bme.

Turn off all the watchdogs and try again:
http://wiki.maemo.org/R&D_mode

Also enable the framebuffer console at bootup, so you'll see where exactly it's breaking.

Isn't the DSP driver easily obtainable from numerous places on the Internet, and didn't the DE team say that if you want it, you can simply drop it in?

Might be mixing some information with the ones for 720p playback on M5, but I think this is mostly correct.

I may be off course here but I think that in the recent thread about open drivers, he who has just left us mentioned that while the N900 might be fine to run 2.6.37 and associated drivers etc, Maemo5 isn't.

This answers the thread question, I think?

(Apologies for long-winded-ness of the next bit: thoughts in writing!)

However, the matter remains of how we can achieve an OS that runs well on the N900 for all users: general, power and developer; and, that meets the following reqirements:

• fix bugs which prevent the operation of device;
• fully implement hardware and provide complete access to its functionality, though freely available headers or full source;
• access the functionality of all other parts of the OS (non-hardware related), to allow its extension, customisation and application-interaction;
• clear documentation explaining the operation of all parts of the OS, including embedded drivers;
• and possibly, for those developers who are creating interactive s/w, a means to help them do so with consistent UIs and UXs.

To do so, there are at least three routes, but some seem to result in dead-ends...

Option 1 - Maemo5:
There is a lot right with Maemo5 and combined with it, the N900 is nearly perfect but there are some gaping holes! Indeed, From an end-user's point of view, the list is quite long because Nokia essentially created a blank canvas, leaving others to complete the picture. This would have been fine except Nokia "forgot" to stretch said canvas on a sufficiently sturdy frame or provide suffient brushes, paint or turps!

From bug-reports, dicussions and articles, I have come to accept that even releasing headers to many binaries, including drivers, would be pointless and, from what I can make out, the interfaces are so flakey that in fixing one problem in Maemo5, other things break. Thankfully, a lot of the bugs and problems with Maemo5 have been fixed and, from what I can make out, even though these fixes do not work on Maemo5, they do work on the N900; so there is hope. In short though, the requirements above are not possible with Maemo5, including 2.6.37 (dead-end 1).

Options 2 & 3 - Meego-Cordia or Harmattan:
Those working on Meego & MDE were instrumental to making these fixes, which are a part of Meego DE and do work on the N900 (again, as I understand sources I have read). However, I can see the arguments to keep Maemo5 over Meego, not least because Maemo5 and the N900 were designed with each other in mind, though it sometimes seems each one was done at opposite ends of the world. Also, Meego (even the DE) is a lot more hungry so is not likely to be be as reponsive as Maemo5 on the N900, which is why it might be worth considering an adaptation of the next iteration of Maemo, Harmattan to replace Fremantle - option 2.

I'm working on the assumption that Harmattan has been brought up to the same standard of driver and middle-ware operability as Meego and will come closer to meeting the above aims than M5 - N900. I might be wrong here. Also assume that 2.6.37 will work with H - again might be wrong here too, if so another dead-end.

Importantly, Nokia is already developing or has already developed Harmattan. Therefore, less work would be required, apart from a few adjustments for the N900. And, there is some chance it will happen with Nokia's help (even if via a CSSU and not a PR). A little pressure might aid our cause?

If Cordia can compete with Maemo5's UX on the N900 (including speed), it might be a viable option (option 3), depending on release time-scale. Also, can Cordia take advantage of the maturity of Maemo, through its various iterations?

Is it worth asking for an official adapation (or for help with a CSSU adapation) of Harmattan for the N900 that meets the aims and requirements as closely as possible? Or would Cordia achieve this, if so in the next year or so? One or both of these might also be dead-ends but only time will tell.

Either with Cordia or Harmattan, a change of OS will mean the catalogues will be a little dry for a while but we'll have a device that actually works in its entirity and can be properly built on!

I'd be interested to hear from others whether Harmattan is bug- fixed and might be something worth pushing for on the N900. Also, would Cordia be fluid enough for all to use? Is it viable time-wise?

Input welcome, move to another thread if you like. For me, these thoughts were just an extension of how to get 2.6.37 on the N900 ...

This is not a Meego-Maemo rant or contest so, please: no shouting. I think we all want to explore the possibilities to achive the optimal result for our device, don't we?

Mmm, Harmattan HE for the N900. I wonder if Nokia would oblige to that.

Personally, I'd be fine with Maemo 5 being reborn as Cordia+MeeGo and then development being continued on that.

Any Developments Here ? its been long time to see this thread :P