AMD Is Hoping To Upstream DAL For Linux 4.7, But A Lot Of Work Is Needed

[phoronix]

Phoronix: AMD Is Hoping To Upstream DAL For Linux 4.7, But A Lot Of Work Is Needed

Along with open-sourcing the next-gen Polaris GPU driver code yesterday, Alex Deucher of AMD laid out their plans for aiming to get the DAL display code into the Linux 4.7 kernel...

http://www.phoronix.com/scan.php?page=news_item&px=AMD-DAL-Hopes-For-Linux-4.7

[chuckula]

They need to add an extended khronos (EK) specification to DAL. I'm sure the new DALEK code would exterminate any attempts to keep it out of the kernel.

[Pecisk]

Well, it is smart just to add at staging level and got some consensus over reworking it for fitting better in mainstream kernel. It seems AMD team is on time pressure here, so tossing it out won't help anybody. I understand both sides here, and I hope they will come to unified solution, as this is very important for future of Linux gaming and AMD graphics support on Linux in general.

[remm]

Well, it is smart just to add at staging level and got some consensus over reworking it for fitting better in mainstream kernel. It seems AMD team is on time pressure here, so tossing it out won't help anybody. I understand both sides here, and I hope they will come to unified solution, as this is very important for future of Linux gaming and AMD graphics support on Linux in general.

People always have great excuses to justify inclusion of stuff that was done behind closed doors. It then never gets fixed due to another set of great excuses.

[Nille_kungen]

Jerome Glisse is also pretty harsh about upstreaming DAL.
https://lists.freedesktop.org/archiv...ch/103520.html

[Guest]

Well, it is smart just to add at staging level and got some consensus over reworking it for fitting better in mainstream kernel. It seems AMD team is on time pressure here, so tossing it out won't help anybody. I understand both sides here, and I hope they will come to unified solution, as this is very important for future of Linux gaming and AMD graphics support on Linux in general.

The major problem with these people is that rather than trying to intergrate into existing environment, they just develop a thing behind closed doors and then drop it once they think it's good enough or well past deadline. If they employed some drm devs into their ranks or made their own people transparently contribute then these things wouldn't be happening nearly as often. Both AMD and NVIDIA are well behind the Intel on this.

[eydee]

This is where open source shines. Code made by a company (most of this DAL code is rumored to be taken from Catalyst) is designed to minimize cost and just provide functionality needed. Open code accepted into the kernel is about quality and usefulness.

[adler187]

This is where open source shines. Code made by a company (most of this DAL code is rumored to be taken from Catalyst) is designed to minimize cost and just provide functionality needed. Open code accepted into the kernel is about quality and usefulness.

That rumor is unfounded:

The DAL we are working on upstreaming was written from scratch for amdgpu. The fglrx display code was shared with windows.

[darkbasic]

Jerome Glisse is also pretty harsh about upstreaming DAL.
https://lists.freedesktop.org/archiv...ch/103520.html

I will laugh for days at Torvald's response if the drm maintainers will dare to submit it for push into staging

[brent]

The rather extreme criticism seems a bit arbitrary and arrogant. The kernel and especially the DRM subsystem aren't stellar examples of great and perfectly designed code. I don't see much reason to impose these arbitrary high standards here when in other cases pretty crappy code ends up being included without much uproar. Sure the code needs work, but get off your high horses please.