GNU/LINUX and new devices

[Eddie]

Hard to describe the direction that I want to go but fill it would be relevant to ask it here.

I would like to better understand linux building to the point of possibly designing and maintaing a flavor. I have been looking at LFS guides but it seems that even they are not quite what I'm looking for.

Holding the pandora and downloading the panda files I can see that there is still so much that I lack understanding of. Like being able to edit firmware and knowing what kernal setting to select. Spending hours of compliling time to learn that you should have add or not added something then going back to recomplie it...

Looking at new hardware as it comes out, how can I make linux work on this the way I want it to. That way when the new hardware comes out when mine is outdated I only have to update the stuff needed for Linux to comunicate with this new hardware. There are so many devices comming out and there all changing, How do I get linux to work with the new tablet processors and graphics chips. I don't want to boot to SD, I want to replace the flash with a lightweight linux and kick android and all of its exsictince off of the device.

What direction do I need to head in.. Downloading data sheets on embedded processors and the source hasnt been released for the graphics chip, is this a dead end until they release it?

Any guides that are going to realisticly empower me to do this things, in the shortest amount of time, would be benifical.

I currently reading guides but feel a bit like I'm running naked through bear forest without a map. I've noticed linux is much like playing some of the older rpg's without a guide or friend who has played it before. Right now I'm playing StarOcean and I put all my skillpoints into cooking.. I don't know if it was a bad choice and wonder if I should start over. But in the case of old RPG's I don't want to rely on guides only friends. It makes the game more enjoyable and keeps me in the game, I don't seem to finish them when I use the guides. May just have to run blind first couple of times through and become smarter second time through I guess.

 

[levi]

Porting any OS to new hardware without sufficient docs is going to be at least as hard as it is to do using Linux. It's a task that takes teams of very smart guys with lots of experience months to even get the simplest code to boot in the examples I've followed. Even with the NDAed docs, porting an OS to new chips is a very specialist role and still takes an awful lot of trial and error, as usually the docs don't quite match the reality.

 

[Eddie]

Thanks Levi, I was wondering about that. I have been waiting for the mele quad to come out.. http://liliputing.com/2013/04/mele-goes-quad-core-with-mele-a1000g-android-tv-box.html  But they were saying something about how the graphics chip does not have released support yet. It looks like it is a slighly newer graphics then the pandora and with the quad and 2 gigs ram, I get pretty excited about the possiblity of reflashing one of these with linux and using the Icontrol2 on it as I don't belive the pandora's video out is worth the trouble for me. (don't think I'm bashing the pandora, I love mine)

These are tablet hardware devices and run android. Can someone not root the android os and see the drivers and such to the copy over to the linux side or is it not that simple?

 

[klapse]

Save yourself a lot of time and forget about it. I've been using Linux for nineteen years and wouldn't have a realistic shot at creating a distro for a new device. I'm not that smart.

 

[Eddie]

 From what I have been reading a distro is really more the package manager. So, If I have to, I will use a spin off another distro and figure out something

 

[pmprog]

Save yourself a lot of time and forget about it. I've been using Linux for nineteen years and wouldn't have a realistic shot at creating a distro for a new device. I'm not that smart.

I don't want to sound rude, but why should he forget it just because you're "not that smart" (your words)?
People I work with have been using Windows for 10+ years, and half of them still struggle to navigate around properly, and none of them would know how to write code. Unless you spent all nineteen years trying to build your own Linux distro, it's not all that relevant.

Sorry if this sounds pretty negative, but I always think it's better to encourage people who want to try these things. Sure, it may be a lot of work, and yes, it may not come to fruition, but even through the act of trying, they learn something, whether it's about what they're doing, or about themselves. Learning is good.

 

[klapse]

Well the shortest route I could suggest would be to get familiar with multiple Linux distros made for portable devices. Pandora is one. Also there's a Linux for Archos tablets: here is a good doc with some background info on what it takes to make an embedded Linux distro http://www.openaos.org/~kevin/archos_angstrom_docs/

 

[Eddie]

https://www.youtube.com/embed/Ub-HMBkN6ts?feature=oembed

 

[Binky]

Wrong thread?

 

[dreamer]

 From what I have been reading a distro is really more the package manager.

Yeah, you're not reading the right things bro.


There are plenty of distros without package managers.

 

[ZXDunny]

Hard to describe the direction that I want to go but fill it would be relevant to ask it here.

I would like to better understand linux building to the point of possibly designing and maintaing a flavor. I have been looking at LFS guides but it seems that even they are not quite what I'm looking for.

Holding the pandora and downloading the panda files I can see that there is still so much that I lack understanding of. Like being able to edit firmware and knowing what kernal setting to select. Spending hours of compliling time to learn that you should have add or not added something then going back to recomplie it...

Looking at new hardware as it comes out, how can I make linux work on this the way I want it to. That way when the new hardware comes out when mine is outdated I only have to update the stuff needed for Linux to comunicate with this new hardware. There are so many devices comming out and there all changing, How do I get linux to work with the new tablet processors and graphics chips. I don't want to boot to SD, I want to replace the flash with a lightweight linux and kick android and all of its exsictince off of the device.

What direction do I need to head in.. Downloading data sheets on embedded processors and the source hasnt been released for the graphics chip, is this a dead end until they release it?

Any guides that are going to realisticly empower me to do this things, in the shortest amount of time, would be benifical.

I currently reading guides but feel a bit like I'm running naked through bear forest without a map. I've noticed linux is much like playing some of the older rpg's without a guide or friend who has played it before. Right now I'm playing StarOcean and I put all my skillpoints into cooking.. I don't know if it was a bad choice and wonder if I should start over. But in the case of old RPG's I don't want to rely on guides only friends. It makes the game more enjoyable and keeps me in the game, I don't seem to finish them when I use the guides. May just have to run blind first couple of times through and become smarter second time through I guess.

OMG don't put all your skill points into cooking. Seriously.

D.

 

[levi]

 From what I have been reading a distro is really more the package manager.

Yeah, you're not reading the right things bro.


There are plenty of distros without package managers.

Well, there are distros without PMs, but that's a pretty old-school way of doing it imo.  You could build a disc image from source and call it a distro, but why would you do that in this day and age?   Simpler, as Eddie says is to pick a Debian or a Fedora (or whatever the favourite distros are for your architecture) and create a different spin on that using the package manager they come with.

 

[Eddie]

ya, lfs talks about that, it dosnt have a package manager. you have to compile everything from source. then once you want to go to blfs you get into a simpler,faster pagage management type setup.  http://www.linuxfromscratch.org/

I am going to master cooking, so at the end we can have a shushi party!  :lol:

 

[skeezix]

Come now people.. don't discourage the lad   Nowadays we unix nerds are getting fairly common, but go back even 5 years back, or 10, and BSD, Linux were still young, and not so many of us Unixnerds around. You think all those OSes and distros came around on their own? Peopel started them. And those people mostly came from Windows (or DOS..) backgrounds as well. Everyone starts somewhere, and the lucky (or unlucky..) start where they want to go.. most people transition through life, figuring shit out So yeah, this is a forum that was based on hacking (well, in the gp32x.de days anyway), and coders, and doing crazy shit.. hence the Pandora.

So keep it up guy, depsite any negativity. It is hard

I'd say building a distro is both hard and _tedious_; most of it is tedious boring slogging to get through piles of stuff that just needs doing, but is not hard.. just not well documented, and boring as heck.

The other half, is hard to very hard; like if you need to port a kernel to a kernmel to a new device, it can be a big challenge depending what peripherals are already supported (if any) and how confgurable those drivers are, etc; how stable is your target platform, etc. If you're a coder, or not a coder? Know any embedded?

Certainly its doable, especially if you leverage existing distors (be they big, or small ones like Angstrom..), and how those distros are built to support porting or not (ie: x86 centric, or ARM centric, or agnostic?)

Theres too many questions, but gotta start somewhere. Start asking questions (a wide pile of them), and get an idea in your head of where you're going and where you're coming from; start figuring out the target platform and whats all in it in detail (in _detail), so you can start lining up features.. what features needed, what are offered by the Linux kernel (and your distro is not so relevent.. first job is finding out about bootloaders, and if the kernel is even viable; the distro userland is all secondary.. once you get a kernel up, the rest comes 'easy' (just tedious)).

Anyway I've no time for this big discussion now.. but keep asking, and draw the right people in, and you'll go; you may want to find the real serious linux nerd forums, and the distro mailing lists and such... people have done this before, but here mostly are gamers and coders.

jeff

 

[klapse]

Look, if you could point to any cases of amateurs, acting alone, put linux on portable devices lacking manufacturer support (including reverse-engineering and writing kernel drivers), then his goal might be doable. Since nobody has ever done this to my knowledge I am suggesting he doesn't have a realistic chance at doing it.

 

[levi]

Yeah, we're confusing two goals here.  Skeezix is talking about creating a new distro on a platform which is well supported, which is one direction this thread took.  On the other hand the OP (as I interpreted it) was talking about porting to a new platform with unknown and unsupported bits of the chipset, which is a whole other kettle of fish.

 

[skeezix]

Lots of distros .. in fact, nearly all of them.. start out where they have to write a pile of code to get up and going; very few I imagine start life on platforms that are undocumented, is true .. is that the case here? he does not actually provide any real detail.. except to hint at maybe wanting to replace the firmware in devices he owns or wants to own; thats a noble cause. He's not anywhere near that, but hey, look ay Cyanogen.

He hasn't said he wants to target new totally undocumented platforms; at hsi level of expertise, sounds totally implausable that.. but why discourage? Lets find out more.

I'm designing my own retro-styled old school 8bit game console/computer from scratch, without any modern hardware; all old school parts, no cheating. A few months back, I only knew a little electronics, and read some encouraging sites suggesting 'I didn't know jack, but over a few months, I build this <show impressive as hell pictures>', and I thought 'no way, you're full of shit'; but I've been meaning to do this project for years (like 20 years ), and finally got around to it; and a few months later, I' knee deep in wiring, but this shit is _happening_. Who knew? Its not has hard as I though actually.. so, yeah, totally different than his story, but just saying .. if someone wants to bite off too much -- good for them; some people are the kind that thrive on impossible, while others just give up. Lets see the measure of the man yet

(and if anyone cartes.. should I open up a thread to go on about the adventures with my shitty console? should I blog about it? its not pandora related at all, so I figured most people here do not care.)

jeff

 

[Binky]

and if anyone cartes.. should I open up a thread to go on about the adventures with my shitty console? should I blog about it?

An epic geek electronics retroconsole project?

Absolutely!

 

[levi]

Seconded!  I'd forgotten about Jeff's 8-bit computer, but if he feels like posting the odd photo and update that'd be awesome!

 

[Eddie]

Hell ya :lol: