maemo.org - Talk - fremantle on other platforms

maemo.org - Talk (https://talk.maemo.org/index.php)

- NeoFremantle (https://talk.maemo.org/forumdisplay.php?f=58)

- - fremantle on other platforms (https://talk.maemo.org/showthread.php?t=92464)

Re: Neo900 - finally a successor of N900

Quote:

Originally Posted by joerg_rw (Post 1407785)

seems you all are missing the rationale. It's for sure NOT for the NAME of something.
Use standard Debian ->

mafw: none -> Mediaplayer OpenMediaplayer, Gallery: gone
alsaped: none -> consistent handling of audio in all situations, for all sources and sinks: gone
PA XPROT: none -> sooner or later your speakers blow up (unless we fix that on hw level).
standard upstream kernel -> optimized power handling of maemo kernel: gone; standby time: abysmal
MCE (and all the stuff it needs, like dsme etc pp): none -> complete state control (ringtone, vibra, indicator LED, screen lock, orientation, backlight brightness): gone
HAL-addon-bme & friends: none -> proper battery monitoring, low-bat warnings, clean shutdown: gone
liblocation: none -> all location aware apps: gone

I understand that, but this is already an open distro, I understand that more work will have to be done in actually building the OS. Maybe I've been led to believe that the N900 is more closed source than it already is

Re: Neo900 - finally a successor of N900

Quote:

Originally Posted by joerg_rw (Post 1407785)

seems you all are missing the rationale. It's for sure NOT for the NAME of something.
Use standard Debian ->

mafw: none -> Mediaplayer OpenMediaplayer, Gallery: gone
alsaped: none -> consistent handling of audio in all situations, for all sources and sinks: gone
PA XPROT: none -> sooner or later your speakers blow up (unless we fix that on hw level).
standard upstream kernel -> optimized power handling of maemo kernel: gone; standby time: abysmal (watch out! oversimplified picture - power handling depends on whole system)
MCE (and all the stuff it needs, like dsme etc pp): none -> complete state control (ringtone, vibra, indicator LED, screen lock, orientation, backlight brightness): gone
HAL-addon-bme & friends: none -> proper battery monitoring, low-bat warnings, clean shutdown: gone
liblocation: none -> all location aware apps: gone

and so on and on (probably I missed to list 60..80% in that enumeration)

We use fremantle since it's mature and been tested for 4 years now, basing on chinook and diablo. We know it's working. We don't want to invent yet another system. If you want to do exactly that, you're of course free to do that but at least anything in context of Neo900 is clearly meant to give fremantle and all the apps made for fremantle a new life.
You might have missed that detail, but to me it seems Harmattan (N9) and even Sailfish (Jolla) are using pretty much every single detail of the subsystems I listed or missed to list above. Guess why! While other distros use exactly zero of all those.

PS: Maemo is Debian, with several special tweaks on top. So where's the rationale to use "standard Debian" (I.E. a maemo with all special maemo tweaks stripped) and do the tweaks again? There actually IS a rationale, when you call that process ReverseEngineering and do it for the closed blobs only, while keeping the FOSS parts that turn Debian into Maemo. "why buy a 4WD model? Wasn't it easier to buy the standard model and mount something to it that drives the front two wheels as well?"

Maemo may well be 4 years mature and tested but it has also accumulated of 4 years of known security vulnerabilities which have already been fixed in Debian. This, and instant access to the Debian repos are my main appeal for using standard Debian and I don't think I am not alone on this.

By no means do I want you to change what you are doing, I realise that we would have to build this ourselves. Obviously we won't want to use the standard upstream kernel and I would hope that we could construct a runtime environment to run the rest of the Maemo components until they can be ported to Debian.

Maybe I'm getting way ahead of myself and have overlooked the complexity involved in this but it's certainly something I would at least like to look into when I finally get the device in my hands.

Re: Neo900 - finally a successor of N900

Quote:

Originally Posted by wicket (Post 1407827)

Check this thread http://talk.maemo.org/showthread.php?t=91709

Re: Neo900 - finally a successor of N900

Just my personal view:

Quote:

Originally Posted by joerg_rw (Post 1407785)

mafw: none -> Mediaplayer OpenMediaplayer, Gallery: gone
alsaped: none -> consistent handling of audio in all situations, for all sources and sinks: gone

I don't care.

Quote:

Originally Posted by joerg_rw (Post 1407785)

PA XPROT: none -> sooner or later your speakers blow up (unless we fix that on hw level).

I hope you can fix that on the HW level somehow, otherwise is there reason why Debian's PA can't be told to work like Fremantle's? (this is so not my field of expertise)

Quote:

Originally Posted by joerg_rw (Post 1407785)

standard upstream kernel -> optimized power handling of maemo kernel: gone; standby time: abysmal (watch out! oversimplified picture - power handling depends on whole system)
MCE (and all the stuff it needs, like dsme etc pp): none -> complete state control (ringtone, vibra, indicator LED, screen lock, orientation, backlight brightness): gone
HAL-addon-bme & friends: none -> proper battery monitoring, low-bat warnings, clean shutdown: gone

Is there a reason why this can't be imlemented in Debian (custom kernel+some other packages)?

Quote:

Originally Posted by joerg_rw (Post 1407785)

liblocation: none -> all location aware apps: gone

Yay! :p

Quote:

Originally Posted by joerg_rw (Post 1407785)

and so on and on (probably I missed to list 60..80% in that enumeration)

Maybe someone would care to complete this list - especially when there are potentially HW-damaging items on it?

Quote:

Originally Posted by joerg_rw (Post 1407785)

PS: Maemo is Debian,

You don't know how much it pains me to hear that from you (again).
From a Debian maintainer's POV (I'm none) Fremantle was broken from the start. It was a crude mix of Lenny and Squeeze with even some parts coming from Etch.
Everybody who asks for help with a system like this in a Debian community will be asked to throw it away and start over again.
Fremantle is the impersonation of dependency hell. For Easy Debian we had to patch PA since the Squeeze images, and starting with Jessie we'll have to patch glibc (assuming it will still work when Jessie is released). Some packages are already broken (gparted on armel/hf, gimp on armel) and this is not due to the architecture (the same ED images work fine on my Cubieboard2). And it will only get worse.
Quite frankly I'm tired of this and it even made me stop my activity on TMO because in the long run I didn't see any future in the N900/Fremantle. The Neo900 with the potential to run Debian (the real thing) changed that.

Now, don't get me wrong, I highly appreciate what you are doing to keep Fremantle alive. But in its current state (the state it has been in from the start) I don't want it anymore. It is not compatible with Debian and therefore it is NOT Debian.

Enough with this whining. The rest I'd have to say (positive and negative) has already been said by wicket.

Re: Neo900 - finally a successor of N900

Quote:

Originally Posted by joerg_rw (Post 1407785)

[*]PA XPROT: none -> sooner or later your speakers blow up (unless we fix that on hw level).

I was under the impression that NeoFremantle would not be ready and the Neo900 would ship with another OS (probably Debian). Does this mean that we will potentially have to wait for NeoFremantle for a solution to this?

Re: fremantle on other platforms

The annoying thing is that right now, with the current resources, building a new device based on omap3 is easier and faster than noving fremantle to another device...

If more OS developer resources were available, probably the next hw platform with somewhat good linux driver availability would be the Allwinner series SoCs. I believe some of them have almost complete linux support, except for graphics. They're also mainly aimed for tablets and not quite as power lean..

Re: fremantle on other platforms

removing closed source packages and forward porting patches is for me the way to proceed at the moment. any idea of updating needs to address these.

cssu is already doing this where possible (pixman comes to mind). Aapo's work on rebasing against Debian 6 shows whats possible but is hindered by these blobs. pali and fmg (and those i forgot) are working hard on upstream kernel, just check out number of features added in 3.13, but again need lots of LD_PRELOAD shims IIRC to get it working with the rest of it. then there's omp, clock replacement, calculator....

debian stable releases have plenty of testing due to their release when ready attitude, something the debian maintainers should be proud of. switching to a base where we can track the upstream stable point releases would help aleviate (sorry if spelt wrong) some of the cssu guys work. switching to upstream kernel reduces number of backported fixes/features they need to maintain.

so i see why you may want a new updated version, but we need to sort out the one we've got first and that won't happen whilst we have to rely on closed source drivers/components.

Re: Neo900 - finally a successor of N900

Quote:

Originally Posted by Alecsandru (Post 1407550)

could you elaborate the missing hardware features , or in short : what ''hardware'' you must have to boot maemo?

Hi!

Could You contact me?
I would have some questions about N900!;)

Thanks a lot!

My email address: dedista@ and google mail dot com

Re: fremantle on other platforms

I would like to know how to port maemo or which rough steps are needed. This should allow us to start porting very quick when the Neo900 is released.
So I was looking for a cheap device which could be useful to make a proof of concept.
My problem is now I don't have a clue how to start.

Should I study linux from scratch or just build a kernel and copy the rootfs?
Which toolchain I need to use?
When I am using the same toolchain as N900 do I need to compile all packages?
So first thing is taking the N900 Kernel and add all hardware drivers for the new device?

I would appreciate if someone can put some light in this matter. I will also document the work that it could be used as a guide.

I choosed following hardware it has some kernel source (don't know about hardware status support) and you could boot from sd card:

lenovo A1_07 Kernel

TI OMAP 3622 1GHz Cortex-A8 single-core processor
MDDR 512MB
16 GB
GPU PowerVR SGX530
600 x 1024 pixels

Is it possible and feasible to port to that device? If somebody want to jump on this project and try to port maemo with me, I would donate 2 used devices for the first two seriously willing developers which have some experience in kernel work and porting.

The benefit of this project could be that some owners of the A1 will get interested in Maemo so the community can grow a little and we get new manpower.
Also the project will be teaching and training the community for the release day of Neo900 how to help porting maemo to it.

What you think?
Everyone is welcome to jump on and participate.

Re: fremantle on other platforms

If you've not already seen it, you may want to look at the main porting thread.

The BeagleBoard-xM might be a better candidate as it uses the same SoC as the Neo900.

Having said that, I believe the biggest porting issues right now are the closed components which must be reverse engineered and rewritten. You don't necessarily need a single-board computer to help with that.