Porting Various Operating Systems To Pandora?


you got it pretty right, though one must say, that there are very different opinions of what a derived work really is.
As said before, a lot of distributions are spreading nvidia/ATI/atheros-drivers though it's likely that they might violate the gpl.
Though I don't see a reason for this, commercial distributions like madriva or suse enterprise will have the drivers within their basepackage, while -most- free distributions spread precompiled proprietary drivers as packages in an online repository.

So, as long as you don't deliver the pandora with the proprietary driver installed, you will be pretty save but you might violate the gpl.
 
conso said:
So, as long as you don't deliver the pandora with the proprietary driver installed, you will be pretty save but you might violate the gpl.

Then who will provide the binary blob for SGX? Imgtec doesn't.
 
Last edited by a moderator:
Laurent said:
conso said:
So, as long as you don't deliver the pandora with the proprietary driver installed, you will be pretty save but you might violate the gpl.

Then who will provide the binary blob for SGX? Imgtec doesn't.
In this case, it would be the vendor. However, having said this, the issue isn't the blob itself, it's where it ties in. If it's what MWeston and I think it is, it's a thin shim that is legit and the blob lies in userspace. That's how, by the way, that those binary drivers get shipped with Mandriva and whatnot in a more legitimate manner (though it still remains to be seen how close they're skirting to breaching license with it...) as it's not a clear violation at the least. As for the "don't get to use your hardware" comment, unfortunately YES. There's a reason why Intel and AMD have given out technical data. It's why I'd like to see if we can sweet-talk ImgTec into doing the same thing. It's in their best interests, really. Claiming they're protecting their IP is a bit bogus when the players that out performed them in the desktop space are giving it away now- they've not got anything special that Intel doesn't have a variant thereof in the X3xxx series of GMA parts just for starters. If I thought I could find an ATOM SOC like the OMAP3 at this point with that GPU on it, I'd look at choosing it if it's power profile was right- just to avoid any more of this sillybuggers.

A few things allso worth observing about "junking Linux"- best of luck getting to this stage. I suggest BSD, but you won't get the same class of overall performance. BSD is strictly geared for server work. You could use RTEMS, but you'd be nearly in the same boat as with Linux. You could use eCos, but it's not in the same class as everything else and ImgTec would have to come up with drivers for you- we're not big enough to merit that without a LOT of money. You could use LynxOS, VRTX, or a host of other embedded OSes without this problem, but you're going to shell out lots and lots of cash to DO that- and have to get ImgTec to support you unless you use Symbian which is the only OS they support other than Linux right at the moment. The problem is the royalty price (You have to comply with the license terms) for the OS. Either we come up with an answer for the issue that won't cause problems on licensing or we come up with an answer that costs CraigX and MWeston hundreds of thousands of dollars. It's actually that simple.
 
Last edited by a moderator:
From MWeston's tone earlier, it sounds like ImgTec isn't likely to open up their driver.

OK, so what if Pandora doesn't ship with the offending driver. Could the user add it himself? Could Pandora use the same loophole emulator authors use and just let the end user obtain the "legally questionable" bits themselves?
 
Svartalf said:
Either we come up with an answer for the issue that won't cause problems on licensing or we come up with an answer that costs CraigX and MWeston hundreds of thousands of dollars. It's actually that simple.
MWeston is the Pandora hardware engineer, so I very much doubt it'll cost him anything, unless he has to start redesigning PCBs.
 
Last edited by a moderator:
How about we all just wait until Monday, to find out from MWeston whether there's anything to get "excited" about? ;) This issue is a very old one, and I'd be a little surprised if ImgTec didn't already code it properly.

That said, it seems David Airlie found some funky behavior from them in the past on his livejournal. Using GPL exports by having a binary driver claim to be GPL? Schwa? He doesn't blog on it again, so maybe it got resolved... (Intel released some register documentation of the 2700G, so nobody cared?)
 
Chip said:
Let me make sure I understand the situation. From what Svartalf is saying, we may be in a position where we have two options:
  1. Artificially cripple the Pandora platform in order to comply with the GPL license
  2. Violate the GPL in order to actually use the hardware we're paying for
I would like to get confirmation that this is the case before I start getting "excited".


I have no clue about all this GPL stuff. But it sounds some kind of serious. :-||. Of course I want a Handheld, where Devs have Hardware-Acess. Who needs such a powerful Device, if the Devs have no acess to the Power and have to use only unoptimized software-solutions for all the programs? :lol:
Is this "GPL" that serious? Does it have mean Lawyers like Sony or MS?

Well, I hope the Homebrew Scene can use the full hardware of the Pandora. ^^""
 
Last edited by a moderator:
fusion_power said:
I have no clue about all this GPL stuff. But it sounds some kind of serious. :-||. Of course I want a Handheld, where Devs have Hardware-Acess. Who needs such a powerful Device, if the Devs have no acess to the Power and have to use only unoptimized software-solutions for all the programs? :lol:
Is this "GPL" that serious? Does it have mean Lawyers like Sony or MS?
The GPL is a software copyright license. It dictates what rights someone has to copying and distributing the developer's work, in addition to whether you're allowed to modify it. The official version is in English, but you can also read an unofficial German translation here.

Violating a software license is against the law and an offense with some heavy penalties. The GPL is kind of unique beast in that the original developer forces anyone that modifies or uses his code to also publish the source code of his modifications. Since the Linux kernel is licensed under the GPL, any code that calls its functions (with system calls excepted) must be licensed under the GPL or compatible license (that is, its code must be published).

This causes some problems for hardware vendors that don't want to make open source drivers (to protect their IP, or some such nonsense). You can't create a driver without at least playing with interrupts and the bus the kernel mitigates access to, so at least that part must be open source.

However, you can get around the problem by creating an OS-independent binary that could perform the meat of the functions in your driver (but doesn't call any kernel functions). Since you didn't interact with or link to the kernel, this can be a closed source blob.

All the kernel interactions could then go into an open source shim so you comply with the GPL's terms. This is totally against the intent of a significant number of kernel devs, lol, but it's necessary for a practical Linux system (for now).

I'm pretty sure this is the correct interpretation, but I am not a lawyer. Someone correct me if I'm wrong. :)
 
Last edited by a moderator:
I'm a bit worried about the terms on the SGX. I'd speculate that it's not really *their* intellectual property that's an issue, it's what patent trolls might find by looking through their driver source.

In my view the SGX should be opened in order for the platform to remain relevant in the next years. Intel is a huge "threat" to this whole market and has a history now of totally opening their hardware. And they do have the ability to develop an embedded GL core if they so desire.

I'm not sure who I'm more upset with: the IP holder here or the corrupted patent system.
 
Chip said:
From MWeston's tone earlier, it sounds like ImgTec isn't likely to open up their driver.

OK, so what if Pandora doesn't ship with the offending driver. Could the user add it himself? Could Pandora use the same loophole emulator authors use and just let the end user obtain the "legally questionable" bits themselves?
That would be one of the ways to obtain the whole thing. It's not an issue if you get the thing legitimately and then wedge in proprietary drivers through a patching process (Whatever form that takes...) as long as you put it back the way it came when you gave it to someone else or sold it. Otherwise the person doing it that way would be guilty of infringement.

And, I'd say they're not likely in the short-term. Longer term, however, may be a differing story. They're about to face two competitors in their space that don't seem to be overly concerned about "protecting their IP" in the same manner as they do- and giving out technical programming information to the bulk of their GPU offering. We'll just have to see what all comes of things in the next couple of days at least. :D

proflogic said:
I'm pretty sure this is the correct interpretation, but I am not a lawyer. Someone correct me if I'm wrong. :)
It's close. But it's not "forces" as the right term to use. Everything, IP rights-wise, when it comes to licensing, requires some sort of remuneration for the rights given. It might be nada in some contexts. It might be cash up-front, and a recurring payment based off a percentage of sales (Games and things like MP3 and DVD players falls under that space...). In the end, a "royalty" of some sort is typically paid for the right to make copies or derivative works of a protected work.

In the case of the GPL and LGPL, the "royalty" paid to the rights holders is for you to abide by the terms of the license, namely providing the source code used to produce the binaries you might be distributing in their entirety. If you don't give out the code used, including any mods, you're not licensed to distribute or make derivative works thereof- and any attempt to do so is an act of infringement. If you are unable to do so because of other licensing, you are in violation of the terms as much as if you willfully did it like Verizon and Actiontec did. If you know you can't- it's as willful as their stunt was. If the Copyrights are registered, even for small quantities of infringements, you can be held actionable for the statutory damages, which is something like up to $150k per EACH verified instance of infringement done, per US law- the laws vary from country to country, but they've similar teeth in most of the Bourne Convention countries.

You don't want to go there. Nobody really does. :D

bnolsen said:
I'm a bit worried about the terms on the SGX. I'd speculate that it's not really *their* intellectual property that's an issue, it's what patent trolls might find by looking through their driver source.
Heh... They really need to realize that this won't protect them either way. The patent trolls will just do their thing- and in some cases have enough resources to RE their way into finding out you're "infringing" on their rights. :D

Hiding technical info doesn't protect them. It's that some IP attorney trying to make himself more important than he/she really is told them that they had to do this stuff for their due diligence (which is bogus in and of itself...).
 
Last edited by a moderator:
Svartalf said:
It's close. But it's not "forces" as the right term to use.
Bah, rhetoric! :p

Though, admittedly, "forces" is an element of the word set that makes people angry. I'll be more careful from now on. ;)

I hope all turns out well. I'm really anxious for a device like the Pandora combined with a community like this one. If I don't kill myself with classes (and hopefully research) next year, I'd like to develop some games too. Or at least... frameworks for games. I'm not sure of my abilities in the art department...
 
Last edited by a moderator:
proflogic said:
How about we all just wait until Monday, to find out from MWeston whether there's anything to get "excited" about? ;) This issue is a very old one, and I'd be a little surprised if ImgTec didn't already code it properly.
Being that they were, unfortunately, something of a niche player in the GPU space, save for the Dreamcast (And it's a shame, the thing was actually impressive to catch 4/5ths of the performance of a TCL capable card in the same class WITHOUT the transform, clipping, lighting (Vertex shaders...) support...)- since it's mostly mobile phones that use the suspect GPU/CPU combo in the link you provided and many people aren't watching it like a hawk like the other spaces until recently, it might have slipped notice past what was caught by Dave Airlie and the Pepper Pad community.

QUOTE

That said, it seems David Airlie found some funky behavior from them in the past on his livejournal. Using GPL exports by having a binary driver claim to be GPL? Schwa? He doesn't blog on it again, so maybe it got resolved... (Intel released some register documentation of the 2700G, so nobody cared?)



It's probably more of the case that the people that owned the Pepper Pads didn't care much and since there's no N800/N810 drivers done at this time that it just got forgotten. His read's not QUITE the one I have. Lying about the GPL status has pretty much only one purpose unless it's BSD licensed (which, I suppose would be okay- it's a more liberal license so it'd probably go anyhow...)- it's to cheat on the GPL licensing on a binary-only module. Distributing that would be an infringement by any parties doing it, not a DMCA violation.

As it stands, I'm a bit concerned about MWeston not coming back with an answer to this concern...
 
Last edited by a moderator:
I'm certain this is done in the same nVidia style of binary object + stub kernel code.

It technically does not violate the GPL (enough at least) to cause any issues. (Binary Blobs are bad for support issues, and well transparency... and portability, ect)

I view it as very unlikely that any driver source will ever be released, perhaps register specs but little more than that.

Good News is that next generation systems like this will probably be x86 SoC with fully open sourced GPU modules made by the big players (AMD/Intel/VIA).

Bad News is the blobs are here to stay, for now.
 
What? Who said x86 is a trend for general-purpose mobile devices? It's not even a good general-purpose architecture in several aspects, and it just keeps itself alive on top of millions of compatibility/improvement layers for 30 years.

Sorry, but I didn't see anything about that or even can find it remotely reasonable (apart from laptops or even UMPCs which are aimed at compatibility with existing software). OK if you do a UMlaptop or something, but I can't see a gaming device or cellphone running x86 in a sane world.

Please correct me if I'm wrong in my facts or point me to somewhere that states otherwise about the current trend for mobile architectures.
 
Series-8 said:
Good News is that next generation systems like this will probably be x86 SoC with fully open sourced GPU modules made by the big players (AMD/Intel/VIA).
Wanna place a wager on that one?
 
Last edited by a moderator:
Tinnus said:
Please correct me if I'm wrong in my facts or point me to somewhere that states otherwise about the current trend for mobile architectures.
Several different things disprove your position:

eeePC
OQO
Intel Atom
VIA Isaiah

The reality is, ARM and MIPS aren't the only plays these days. And if you have an ultra-low-power x86 machine with Linux and a GPU in the same class as the PowerVR in hand, you have alternatives in the mobile space that don't even need "porting" past what would be needed for Linux versions.
 
Last edited by a moderator:
Svartalf said:
As it stands, I'm a bit concerned about MWeston not coming back with an answer to this concern...
I haven't gotten any response back yet. It's a serious issue that needs to be cleared up one way or another so I hope they take providing a response serious. It might be useful to post the question on their forums but I will wait a while longer. I promise to provide details of what I hear back IF I hear back!
 
Last edited by a moderator:
Tinnus said:
Please correct me if I'm wrong in my facts or point me to somewhere that states otherwise about the current trend for mobile architectures.
I think Intel would probably like to correct you, but then again they have a lot riding on the expansion of x86 into the ultramobile and handheld markets. AMD is also working on a low power CPU/GPU combo chip for the UMPC market, but whether it will be efficient enough to work in a Pandora-sized device remains to be seen.

An x86-powered Pandora2 would not be out of the question in a year or two. Whether it would be desirable is obviously debatable. x86 does have some serious advantages as far as compatibility with existing software. If the big players can overcome the drawbacks of the architecture (as they are obviously trying to do), then there is no reason to think we won't see a handheld Windows gaming rig in the near future.

EDIT: Svartalf beat me to the punch by 4 min.
 
Last edited by a moderator:
Back
Top