MeeGo hardware adaptation for N8x0
 

Continuing from:

Originally Posted by Stskeeps:

On a silly sidenote: I had MeeGo base system running on my N810 with 2.6.33 kernel (+ Termanas, luke-jr's patches) some weeks ago.

and then by lma:

Cool :-) Obviously I've missed a couple of episodes because I thought we were still stuck in the state described here. Do you have a pointer to something describing the current situation, especially what bits are still missing?

and me:

Any takers to start a MeeGo hardware adaptation for N8x0?

and then lma:

Definitely interested (though I'm not a kernel hacker), where do we start?

-
So, this is a thread on starting work on a hardware adaptation for Nokia N8x0. (http://meego.com/developers/hardware-enabling-process). While I doubt we can be a first quality port due to need to upstream patches, we can try things and see how far we can get.

A starter is to grab the current kernel package from http://repo.meego.com/MeeGo/devel/tr....1-8.2.src.rpm , modify this with Termana and luke-jr's patches.

If we ask nicely, I think it should be possible to get repackaging done through my distmaster position so we can get the different firmwares, BME, etc provided through similar means (RPM repository) as tablets-dev.nokia.com does with the MeeGo on N900 things. So, we have a good start.

We can probably start with the chroot images (tar.gz) for N900 as they are fairly generic and the kickstart files.

Re: MeeGo hardware adaptation for N8x0
 

I can't code myself, but I have a N810 lying at home. How can I help?

Re: MeeGo hardware adaptation for N8x0
 

I can help out with the kernel work. Do we need to rebase everything onto the provided MeeGo kernel source?

Re: MeeGo hardware adaptation for N8x0
 

Yes. It is bare linus tree plus some patches.

Also, http://wiki.meego.com/Developing_in_a_Meego_Environment is a good start. I can recommend getting a Fedora VM to do this in.

Re: MeeGo hardware adaptation for N8x0
 

Hey this is a neat idea.. I have a spare N810 that I can use for testing this!

Re: MeeGo hardware adaptation for N8x0
 

We can also use the ones conveniently provided on the initfs, at least to start with.

What's the best strategy for dual-booting (KEXEC vs bootmenu) btw? Is there anything I can do on the Diablo community SSU side to make this easier?

Will those binaries run on OMAP2 or do we need to rebuild them with ARMv6-appropriate flags?

Re: MeeGo hardware adaptation for N8x0
 

Yes, they're built for ARMv5.

Re: MeeGo hardware adaptation for N8x0
 

excellent!
keeping n8x0 as workhorses is a great way to show power of the platform.

Re: MeeGo hardware adaptation for N8x0
 

Hm, chroot complains "FATAL: kernel too old", and LD_LIBRARY_PATH mangling results in segfaults. I guess we really do need a fresh kernel...

Re: MeeGo hardware adaptation for N8x0
 

i would test it on a n800 someone gave me instructions to install

Re: MeeGo hardware adaptation for N8x0
 

where are? .diff ? git repo? .deb ?

Re: MeeGo hardware adaptation for N8x0
 

I have an N800 I could loan to someone actively working on this or use it for testing.

Re: MeeGo hardware adaptation for N8x0
 

Crap, after various unsuccesful attempts to build & boot a modern kernel it seems I managed to fry my N810's camera:

(No, flashing vanilla Diablo, leaving the battery out for a while etc didn't help)

No one's to blame but me of course, but be careful out there!

Re: MeeGo hardware adaptation for N8x0
 

Or just messing up the kernel options? What does booting a Diablo say?

Re: MeeGo hardware adaptation for N8x0
 

That was Diablo :-(

Re: MeeGo hardware adaptation for N8x0
 

As we now have some kernels to play with on recent versions for N8x0, I suggest we start the real work.

Tools of the choice: Fedora chroot or VM. There's problems with MIC2 currently that needs to be fixed before we have proper support on Debian/Ubuntu. This guide should get you started building RPM packages for MeeGo.

I have some tasks we need to get started on in order to get N8x0 properly supported. Who will take these?

1) Adapt the kernel source package from MeeGo to make a kernel build for the N8x0. There is some work regarding 2.6.33 already we can probably piggyback on for ARM support. The patches Termana and tmr has provided should provide a good basis for this.

2) Package up Xomap for MeeGo - there is no good omapfb driver for N8x0 for Xorg sadly. Alternatively, someone can put in effort to work on this.

3) Integrators - someone good with RPM packaging that developers can contact regarding questions.

4) Testers - we need 2-3 committed testers to test the output, documentation provided, etc.

Testers so far: rlinfati

Discussion topics:

* What should be our deployment target? My suggestion is to target at least a 2GB (mini/micro)SD card and boot from there.
* What should be initial (realistic) hardware usable for MeeGo 1.0 with N8x0 HW adaptation? Remember, MeeGo 1.0 is in May
* Should we even care about the 3d drivers or should we look into OpenVG for Qt for instance?
* What would be useful to have Nokia provide in a tablets-dev.nokia.com RPM repository (closed source bits), so people can generate images with the bits.

Re: MeeGo hardware adaptation for N8x0
 

i have a n800, i can test... :)

Re: MeeGo hardware adaptation for N8x0
 

2GB sounds good to me.

I would like to have WiFi and/or Bluetooth and touchscreen. HW keyboard would be a bonus .

I haven't seen much activity regarding 3D driver here in TMO so I guess it's not really working? I would choose the easier route and if OpenVG for Qt is easier, I would vote for that. 3D acceleration is nice-to-have feature, not must-have.

Re: MeeGo hardware adaptation for N8x0
 

I am willing to adapt the n8x0 patches to the MeeGo kernel (though I'll need to have someone else test the n800 kernel), but as I mentioned in #mer, the kernel source is only available in RPM format. Can the MeeGo Project make the source for the kernel available in plain .tar.gz format? Or is anyone willing to do this independently?

Re: MeeGo hardware adaptation for N8x0
 

I may be missing your point entirely, but you should be able to extract the contents of the source RPM with rpm -i or rpm2cpio on any LSB-compliant distribution.

Re: MeeGo hardware adaptation for N8x0
 

n810 Rebased against 2.6.33.2 + MeeGo patches (all from here: http://repo.meego.com/MeeGo/devel/tr....2-7.3.src.rpm)
Includes zImage and patch.
http://bit.ly/aOzM0n

Re: MeeGo hardware adaptation for N8x0
 

Same here. I barely use my n810 since I bought my n900.

Just let me know if testers are needed.

Re: MeeGo hardware adaptation for N8x0
 

I Have a N800, would this kernel patch aply to my hardware? can i put it without fear?

this 1.0 img is without graphics ui, did i get it right? is only 2megas? wow!

will an USB keyboard work since n800´s don´t get a slide keyboard one?

Since i should install it to run tests (if it is compatible to n800)
what shoud be the best way to provide feedback to you and install on it?

other things, here we are at maemo.org community, but meego.com is growning and have their own forum, wiki, etc.
If we start posting this subject here, can it be a way to the meego project see this iniciative more atached to their project and be more supportive?

Re: MeeGo hardware adaptation for N8x0
 

I have some problem with this patch:
1. Internel MMC doesn't recognized
2. Omap Soc sound driver module doesn't create:
Device Drivers --->
Sound card support --->
Advanced Linux Sound Architecture --->
ALSA for SoC audio support --->
<M> SoC Audio for the Texas Instruments OMAP chips
<M> SoC Audio support for Nokia N810
<M> Build all ASoC CODEC drivers
....
ERROR: "i2c_register_board_info" [sound/soc/omap/snd-soc-n810.ko] undefined!

3. What's with WiFi ?

Thank you

Re: MeeGo hardware adaptation for N8x0
 

Instead of this hopping [in regards to N800/N810 OS] from Mer to Mer-whatver-version-2 to MeeGo maybe it would make sense to try to focus on sticking to one option and develop that.

Re: MeeGo hardware adaptation for N8x0
 

My job is also to help start development of things in the community - hence some of the hopping. From a personal preference point of view, MeeGo on N8x0 has most future. But I don't know when MeeGo is day-to-day usable either.

Re: MeeGo hardware adaptation for N8x0
 

No, this patch will not work with an n800, only an n810. There is a link in my signature for a .34-rc3 kernel for n800 which was done by tmr - however the n800 .34-rc3 patch is not against the MeeGo kernel (obviously).

No, this is a known problem. The menelaus driver has the code for the internal MMC, however it doesn't get recognized. Upstream (linux-omap) don't know why this is either.

I have not tested anything outside of my base configuration (arch/arm/configs/n810_defconfig), including sound. There was someone that got sound working for the n810, I will be seeing if I can contact him, if not I will try and get it working myself.

You need to include the p45spi driver into the config. If you look at the .33.1 kernel thread linked in my signature some people have got it working, I have not done any testing on this either and is not included in the base configuration.

I will be performing some testing on these components and making a new patch file soon.

Re: MeeGo hardware adaptation for N8x0
 

Does your job also include finishing a project because I haven't seen much of that?

Along those lines, since you say that MeeGo in your personal opinion has most future does that mean that Nokia is supporting Mer and if not then why is Mer being worked on if MeeGo in your opinion is a more plausible OS option?

I'm sorry to sound 'bitter' but honestly the lack of focus ('Mer for N8xx', 'No wait, we're also doing Mer for N900', 'No wait Mer is dead now it's Mer2 that will come', 'Oh wait. no MeeGo on N8xx is better') is indeed rather tiring and has become one huge disappointment.

If your job, as you mention above, is to start development of things in the community I think you've done that. But if you going forward expect people to beta test, etc you might also want to think about sticking to a project.

Re: MeeGo hardware adaptation for N8x0
 

While I understand the point you're making here, you have to understand that there has been a lot of churn and change that has forced plans (not just Mer, everywhere) to change in very big ways over time. MeeGo's very existence changes a lot of things, at least as far as I'm concerned, because it addresses a lot of the same problems Mer set out to accomplish (such as easier porting to devices).

And it's not just limited to new tech/etc introducing changes, just *doing* some of this stuff teaches lessons that force plans and other things to change in order to remain relevant and realistic. While I wasn't actively around for the earlier days of Mer, I've enjoyed watching it - and the people involved with it - evolve to meet the challenges of the growing requirements.

So no. It hasn't -yet- produced something of shipping quality, but I don't think there's been a better road to go along to learn how to get things done, and I remain hopeful that MeeGo will help to produce a stable and productive base platform for use on a wide variety of devices, including older NITs.

Re: MeeGo hardware adaptation for N8x0
 

Sometimes you have to look at the amount of support a project gets and say "not enough people care".
At least, not enough care enough to spend time making it happen.

As an example: How much time have you personally put into Mer silvermountain? Either development, docs, supporting others .. that kind of thing?
Probably not a lot - and that's fine, it's a volunteer thing :)

However, if not enough people volunteer then.... it doesn't happen.

It turns out that Merm as originally conceivedm was too hard for just a few core devs to deliver and, IMHO, the maemo community didn't engage enough.

Oh, and if you find it disappointing that it wasn't delivered then how disappointing do you think it is when you spend literally months of your life non-stop, 12+hours a day, 7 days a week working on a project and people typically say "Isn't it ready yet? Oh, I can't code, sorry." (I do sometimes wonder if it will come as a surprise to many of those people to find that coders learn to code by doing research and working at it!)

Anyhow...

Meego, by contrast, is much more open and reduces the effort needed to get the N8x0 up and running. There should be a reference UI too - lots of things to make it easier.

Re: MeeGo hardware adaptation for N8x0
 

You're right - Mer never ended up being finished. We started out as a research project and we ended as one. A lot of us who were involved ended up having a lot of knowledge of how a mobile system is put together, how to develop such a system in the open, etc. I'm sure there's a lot of people who will go on record to say the same. Some of us are even doing Maemo/MeeGo professionally now. Training ground for recruits, some might say :)

I'm sorry for the project ADHD I've caused. My professional life and plans regarding Mer certainly got thrown into the air when MeeGo came about. I discussed this in the post-mortem of Mer as an idealistic project.

What we did was that we in a year put together a Maemo compatible platform. Fully open source. Impressive, from a technical point of view. Good for users? Not always.

The (sad) part of my job is that what I do isn't always directly visible as having originated with me. If anything, we've had a central point for developing and hacking these devices on a system level - not just on applications. Again, I think a lot of people will go on record to say I've helped them along or found places for their work to be in.

Mer^2 was being worked on for the sole reason of a short-term solution for the N8x0 users to be able to run Fremantle applications. Maemo GTK+ does not currently exist in MeeGo, making this goal difficult. I have doubts if it makes sense to work on this, too.

If we're going to analyze this from a simple resource point of view, then let us go for MeeGo. It won't solve the problem of N8x0 being left behind short term, but long term it would, instead of maintaining a distro or a backport, we maintain a hardware adaptation. Which is a hell lot easier. MeeGo's as portable as Mer was.

We already have skilled people successfully getting a modern kernel on it. MeeGo, is - probably much to everyone's joy, not managed by me and is a platform a lot of different device vendors, communities, etc will use. This work on the platform would be directly transferable to our N8x0 devices.

I would actually like to ask you to start a thread on "Stskeeps's distmaster work - how has he been performing and what would we like to see him do in the future". Let's hear what community says I should be doing and what I've done wrong - let us clear the slate about my performance and mistakes once and for all.

What I do is fight for things benefiting the community and the users in MeeGo, such as openness, open devices, etc and a lot of random facilitation to activate people and make them feel they're contributing to something, not just a black hole (read: a bug report that noone ever merges the fix from). And doing it with skill and proving my point by technical means?

Now, is that not something worth having me doing?

Re: MeeGo hardware adaptation for N8x0
 

I haven't heard before that Mer was too much work to finish. What I observed is the difficulty they had getting Nokia (who say they support open source) to open needed components, even those which relate only to the legacy tablets, and then got blindsided along with the rest of us with MeeGo. Nokia obviously isn't really engaged with the maemo community and I don't see how the community can be faulted (except if by community you mean Council).

Re: MeeGo hardware adaptation for N8x0
 

No, that isn't what happened. It was hybris on our side and Nokia had been very cooperative in terms of getting us access to needed binaries, but we insisted on a open platform instead of taking a shortcut.

Re: MeeGo hardware adaptation for N8x0
 

Ouch, That's a bit harsh.

I agree slightly that the port of Mer to the N900 seemed to us on the outside to be a distraction, although I suspect that it was not a great amount of extra work compared with building Mer from scratch, and that for the Devs it may have been useful for debugging the desktop environment as the initial build was (and is) very slow on the N8x0.

I think that the Meego announcement caught most people on the outside of the Nokia inner sanctum by surprise. I can see why it makes a lot more sense to put the effort into porting meego to N8x0, than continuing to head down the mer path.

I'm no dev, but I can appereciate the effort that went into these projects, and am full of awe and respect for the people who largely give their time and experience freely to these projects.

Re: MeeGo hardware adaptation for N8x0
 

Notice I said open needed components, not agree to redistribute binaries. According to this page

http://wiki.maemo.org/Open_developme...losed_packages

nothing was opened in 2009.

You think it was hubris to ask Nokia to help in open source software development related to Maemo?

Re: MeeGo hardware adaptation for N8x0
 

No, it was a combination of technical (most components wouldn't work on a modern Debian system with the 'cuts' we made) and that we didn't feel we needed them.. It was a mistake and I have admitted to this before. It was hybris to think we could do it on our own with existing community strength. Open sourcing wasn't trivial either. Waiting for something to be open sourced could stall a project.

Re: MeeGo hardware adaptation for N8x0
 

I do apologize as it did come across as harsher than I meant. It's not that I don't appreciate and value the efforts put in - but after so many, at least perceived, sidetracks and abandoned efforts for the NEXT shiny solution I also can't help but to feel frustrated and wonder what could had been accomplished with a more structured/focused approach.

Re: MeeGo hardware adaptation for N8x0
 

However, I believe Ubuntu will also work.

Nice idea to get MeeGo on N8X0. I have an N800 I'm willing to lay down for testing. I've been thinking in this direction for a while now (even on http://www.meegoblog.net). I'd like to use this opportunity to try kernel development for the first time. And I'm hoping to do a comprehensive documentation of all my findings and eventual creation of a working MeeGo image for the N800.

Re: MeeGo hardware adaptation for N8x0
 

Well, at least the UI is the same on N900 and N810 for once:

http://www.daimi.au.dk/~cvm/meegoonn810.jpg

Will be publishing kernel and kickstart file (very close to N900 one) soon. Still a long way to go.

Re: MeeGo hardware adaptation for N8x0
 

No, open sourcing isn't trivial. But this isn't a matter of delay or something beyond their ability, Nokia hasn't opened any components for the tablets that they EOL'ed two years ago, despite professing a commitment to open source and to the community during that time period.