Intel Poulsbo DRM Proposed, But Rejected

[RealNC]

Is this about releasing the source or going open source? They could release the source (so users can compile it externally) but don't use an OSS license at all ("look but don't touch"), that would work just as well for packagers.

[chithanh]

The current recommendation is still that Linux users stay away from Intel Poulsbo/GMA500/US15W chipsets.

New developments in the meantime since the article was posted:

• A Facebook Intel Poulsbo (GMA500) Dissatisfaction Group has been created.
• Ubuntu has received a new version of the Poulsbo driver.
• Volunteers are still trying to collect the necessary bits and pieces for a basic (read: no video/3D acceleration) open source driver. But it could take some months.

If I understand correctly, no plans exist to open source the proprietary parts. And if things will go the same route as the last proprietary PowerVR Linux driver, then you will be stuck with basic 2D forever.

[danboid]

Any updates on the GMA500 saga/scam?

I'm one of the suckers who got stung by this by buying a Mini 10. "Intel graphics?" I thought "that'll work great with Linux!"- heh!

So I'm stuck on Dell's custom version of Hardy, I can't get xrandr to do any better than 576p on my HDTV, HDMI sound output doesn't work and I get frequent freezes. A very poor state of affairs indeed! I hope my Pandora will fare better when it finally arrives!.

I thought Intel were serious about supporting Linux? This has made all Linux users have to think twice now about recommending Intel IGPs for their xorg friendliness.

[chithanh]

Well basically it should not get rejected because the DRM part is full GPL and the Xorg part uses the binary blob. There is at least no licence issue with the kernel module.

From the article and airlied's comments, the reason for the NAK seems not to be the code or the license. It was insufficient explanations for the proposed drm changes, and lack of official statements from Intel regarding their plans/roadmap for that driver.

[Kano]

Well basically it should not get rejected because the DRM part is full GPL and the Xorg part uses the binary blob. There is at least no licence issue with the kernel module. If you want to maintain it when the Xorg part is based on closed source thats another question. To get the hardware working you need it, and if you find it in official drm, staging or distro specific addon it does not matter for the user. I would just work on more distros out of the box when it is in mainline or staging.

[stan]

Isn't it ironic that GKH was the developer who a couple of years ago proposed disallowing any non-GPL-compatible module to be loaded by Linux? And now he wants to include a kernel DRM driver whose sole purpose is to interact with a binary blob 3D driver.

[sreyan]

Seems like Greg Kroah Hartman might shove this in staging anyways by simply cloning a poulsbo specific copy of drm.

http://groups.google.com/group/fa.li...50330204cbc2af

[sreyan]

People are happy of docs, I'm happy of a working driver. Different points of view I guess.

If you're a gamer I understand your choice to go Nvidia. Wine simply works better with Nvidia in my experience and modern gaming doesnt come to linux in another way. Video rescaling and compiz are the closest things that my usage gets to real 3D so my view point as you said is quite different.

My experience with linux is that the hardware support is good out of the box. As time goes on less has to be installed manually.I suspect the same thing would be true for windows if drivers were continuously slipstreamed into the install disc. I have some older windows cds that don't even recognize network controllers so I can't even download the drivers to get the system to work.

For a gamer non working or slow 3d is a deal breaker. For me I just care that my system works well out of the box; Since fast 3d isn't one of my needs, but 2d and video is, i'm extremely interested in open source drivers going in.

[bulletxt]

1. Many of the open source drivers support other *nix operating systems.

2. PPC? Try using your beloved blob on that

3. My grandmother and I suspect most grandmothers don't have discrete graphics cards. I suspect most of them use IGPs. I'm fairly certain that Intel is the dominant player in the x86/amd64 graphics area, and IIRC they don't have discrete graphics yet.

4. Actually I do give money to companies that work. intel's igps work great out of the box and are entirely open. Thanks for the personal attack on my intelligence. I appreciate it.

My post was actually connected to this article. It was a way to say: "I'm happy there's a comany that does a working driver because I'm getting tired of all these problems, waiting, and so on".

Also, I have an ati 2600XT gpu. By the time my card works as it should with oss driver, it will be old. Old enough that for a gamer it's time to buy a new card. If I buy a new card it won't work yet with the oss driver cause 99% for sure it won't be ready. In other words I'll be stuck to FGRLX driver. No point reminind that FGLRX isn't a driver.

So conclusion is, I'm happy there is a company that just does a driver and makes my gpu work today, and tomorrow.

People are happy of docs, I'm happy of a working driver. Different points of view I guess.

[sreyan]

- they release a legacy driver and it's updated, maybe it doesn't have new features, but it gets updated. "someone else" doesn't do that.

- they support opensolaris and freebsd since years and years. "someone else" doesn't even know what solaris and freebsd is.

- 98% of desktop users are on x86 or amd64. if you want nvidia to release docs for 2% of desktop users, good luck.

final words: your grandmother has a "normal" computer and wants her gpu to work. she doesn't give a damn of your points because she payed for that damn card.

Are you able to understand this? no, you're not. give money to companies that sell you bread to get eaten in 5 months. I buy things that work today. good luck and stay safe.

1. Many of the open source drivers support other *nix operating systems.

2. PPC? Try using your beloved blob on that

3. My grandmother and I suspect most grandmothers don't have discrete graphics cards. I suspect most of them use IGPs. I'm fairly certain that Intel is the dominant player in the x86/amd64 graphics area, and IIRC they don't have discrete graphics yet.

4. Actually I do give money to companies that work. intel's igps work great out of the box and are entirely open. Thanks for the personal attack on my intelligence. I appreciate it.