Pandora Pandora Kernel Git Server

[EvilDragon]

If anybody wants to take a sneak peek at the current kernel, we've got a read-only GIT setup which features the most recent version of the MK2 OpenPandora Kernel.

The git is:

git://git.openpandora.org/pandora-kernel.git

Use the pandora-27-omap1 branch for the moment, there will soon be a 2.6.28.

Also, if any dev or project team needs a git server or a subdomain with webspace, feel free to send me a PM or an eMail. All Pandora Projects can be hosted by us!
Just be sure to not have any copyrighted stuff (images, sprites, audio, etc.) in your websites or projects.

 

[nat1192]

Nice, can't wait to take a look (cloning now)

 

[skeezix]

Slap a link to this in the wiki somewhere? (hmm, I forget if that other post notaz made about kernel functions got linked or not as well.)

I've not taken a look yet, but I imagine theres a stack of blob's in there for all the chipset stuf we can't have source to eh? Or any licensing issues left to work out?

jeff

edit: I tossed a note onto the wiki, but someone verify my formatting; using a phone to add to a wiki is hell, I tell you

http://pandorawiki.org/Development_Tools

 

[EvilDragon]

Nope, nothing in there which is not supposed to be public, don't worry

 

[Chip]

I tossed a note onto the wiki, but someone verify my formatting; using a phone to add to a wiki is hell, I tell you
Made some minor formatting changes. You didn't do half bad for phoning it in. I end up doing that a lot myself.

 

[MWeston]

We're trying to do everything without requiring any of this NDA protected code. Right now, Wifi is that last step and that one could be tough. Interestingly enough, the G1 phone uses the same chip so we are trying to snag their code first, but this confidential driver is always waiting here if it becomes our only shot.


ED: I'm going to need to send you a list of people who need access to the private dev section soon. I'm just waiting for replies and confirmation of commitment.


P.S. I don't count the 3D driver as part of this open source initiative because we know that will never, ever happen.

 

[ldesnogu]

P.S. I don't count the 3D driver as part of this open source initiative because we know that will never, ever happen.

I hope they didn't change their mind on the kernel module itself which was supposed to be open source...

 

[fusion_power]

P.S. I don't count the 3D driver as part of this open source initiative because we know that will never, ever happen.

Why Actualy not? SGX would benefit from a optimized Driver too I think and I've heard that their own 3D Driver is not the best. So what do they fear?
Well, for the worst case, I still count on maciek_urbanski and his interesting Ideas.

 

[Tor]

$ git clone git://openpandora.org/pandora-kernel.git
[...]
remote: 100% (1015134/1015134) done
Receiving objects: 0% (1/1015134)

It's been sitting there 1 hour 20 minutes by now.. trouble with the server? Slashdotted?

 

[notaz]

http://pandorawiki.org/Kernel_build_instructions

P.S. I don't count the 3D driver as part of this open source initiative because we know that will never, ever happen.

Why Actualy not? SGX would benefit from a optimized Driver too I think and I've heard that their own 3D Driver is not the best. So what do they fear?

This was probably about original SGX driver, as someone might think SGX opened up from the rest of his post. The community reversed driver is great idea, but I don't think it can be done anytime soon due to amount of work needed for this.

It's been sitting there 1 hour 20 minutes by now.. trouble with the server? Slashdotted?

Just tried it and it clones fine. You might have firewall/ISP blocking problems.

 

[Tor]

Just tried it and it clones fine. You might have firewall/ISP blocking problems.

Nope. I clone git servers everywhere.

Updated: It got through eventually, after a few more retries. Once it failed with 'fatal: early EOF' during the objects transfer, but the last time it got through fine (took a while though). So there probably was some load on the server or some such. I've got the kernel now though so I'm fine.

 

[ldesnogu]

SGX would benefit from a optimized Driver too I think and I've heard that their own 3D Driver is not the best.

It looks like SGX are doing quite advanced things to optimize generated code, that a user community would have a hard time to match, even with appropriate documentation.
One should not underestimate the work needed to write fast 3d drivers

PS - This is not to say that what maciek_urbanski is doing is useless, it will certainly be interesting. It's just that people should not expect a faster OpenGL ES driver than SGX one as the outcome of his work within a reasonable amount of time

 

[maciek_urbanski]

SGX would benefit from a optimized Driver too I think and I've heard that their own 3D Driver is not the best.

It looks like SGX are doing quite advanced things to optimize generated code, that a user community would have a hard time to match, even with appropriate documentation.
One should not underestimate the work needed to write fast 3d drivers

PS - This is not to say that what maciek_urbanski is doing is useless, it will certainly be interesting. It's just that people should not expect a faster OpenGL ES driver than SGX one as the outcome of his work within a reasonable amount of time

I've heard my name mentioned.

I can tell that in fact there is large amount of optimizations happening in PowerVR OpenGL ES driver. But most important thing that people assume about my project is that I'm going to write optimized OpenGL ES driver.
I'm not!

I'm writing low-level direct-to-hardware driver that has no internal compiler, it merely exposes internal assembler and state of SGX. Users of this driver will write into internal SGX assembly language, provide SGX state, and driver will merely pass this into SGX hardware. Of course there will be a lot of documentation about SGX assembler, state (and so on) to ease development process.
The most value comes from fact that developers of 'cutting edge' software will be able optimize shaders themselves, or write shaders that use capabilities unique to SGX hardware. In spirit this project is very similar to libscg - low-level API that exposes grpahics functionality of RSX (graphics chip insode Playstation 3).

Of course someone else can write OpenGL ES driver using my driver as a base. But this 'someone' will not be me.

 

[EvilDragon]

$ git clone git://openpandora.org/pandora-kernel.git
[...]
remote: 100% (1015134/1015134) done
Receiving objects: 0% (1/1015134)

It's been sitting there 1 hour 20 minutes by now.. trouble with the server? Slashdotted?

Hi,

whoops, seems to have had wrong permissions set.
Should work now (at least, it does from my server at home without login data)

 

[zodttd]

Great work Notaz!

I'm in the process of cloning. So far at 15% indexed. Looks like it will finish smoothly, though others are reporting hangups during the counting process. It's definitely a large repo with 1015134 objects, but should be able to be handled without a hitch. I'll check what version of git I installed.

Hi,

whoops, seems to have had wrong permissions set.
Should work now (at least, it does from my server at home without login data)

Ah! Oops! Thanks for the heads up EvilDragon.

 

[maciek_urbanski]

EvilDragon, I didn't found SGX driver in this kernel, only framebuffer.

Is there a schedule or release date for Texas Instruments to provide it to Pandora team ?

 

[notaz]

EvilDragon, I didn't found SGX driver in this kernel, only framebuffer.

What made you think this kernel has it? It is still not known when it will be released from what I understand.

 

[efegea]

EvilDragon, I didn't found SGX driver in this kernel, only framebuffer.

What made you think this kernel has it? It is still not known when it will be released from what I understand.


And that's what he is asking, why didn't you quote the full post? :huh:

 

[maciek_urbanski]

EvilDragon, I didn't found SGX driver in this kernel, only framebuffer.

What made you think this kernel has it? It is still not known when it will be released from what I understand.


Quoting from blog:
QUOTE

You've all been waiting for a status update, eh? Well, here it comes:

1. Almost ALL drivers have been finished and work fine so far!
2. 20 MK2 devboards have been produced and have been sent out to some devs!
3. PXML Fileformat definition is finished.
4. Prototype keypad and case have been ordered (nope, we don't know yet exactly when they're finished)
5. Chip from the gp32x.de boards is now officially helping us with tiny news and informations. Thanks a bunch and welcome to the team!



After reading text in bold I was optimistic that SGX driver will be there, but I guess it in 'almost' category. B)

It wasn't, so I asked about schedule.

 

[Svartalf]

After reading text in bold I was optimistic that SGX driver will be there, but I guess it in 'almost' category. B)

Yep. That it is...

You're probably going to have to wait on someone like MWeston to chime in with an answer... I'm waiting right at the moment. They have drivers. Unfortunately right now, if you want them, you have to sign away your first and second born in a contract with Imagination Technologies, signed in your own blood... My impression on it is that they're on the way and we're just going to have to wait a bit longer for the publicly accessable stuff from them.