Drop xf86-video-intel, use modesetting driver instead

[lrusak]

Thanks for the tip @fritsch

apparently,

1. the intel devs suggest using the modesetting driver + glamor
2. Ubuntu 16.04 may ship with the modesetting driver + glamor

I think I have done this correctly, but I don't have any intel graphics to test with.

see, https://www.reddit.com/r/archlinux/comments/4cojj9/it_is_probably_time_to_ditch_xf86videointel/

[fritsch]

If you can provide me an image - I will test it on intel hw. v17 based would be really cool.

[fritsch]

And to add: I am not sure this works with the very old i915 gpus.

[MilhouseVH]

@fritsch: Please try http://milhouse.libreelec.tv/builds/testing/Generic/LibreELEC-Generic.x86_64-8.0-Milhouse-20160411100935-%230410x-gcead6d4.tar

[fritsch]

Will do tonight on my Braswell box. Thx much.

[MilhouseVH]

Note that this build includes your "explicitely disable TearFree and TripleBuffer" commit: fritsch/OpenELEC.tv@e7cbe9c

[fritsch]

Yeah - I will overwrite /storage/.config/xorg.conf with just modesetting
driver. As without in deed intel would be in use.

2016-04-11 16:04 GMT+02:00 MilhouseVH notifications@github.com:

Note that this build includes your "explicitely disable TearFree and
TripleBuffer" commit: fritsch/OpenELEC.tv@e7cbe9c
fritsch/OpenELEC.tv@e7cbe9c

—
You are receiving this because you were mentioned.
Reply to this email directly or view it on GitHub
#136 (comment)

               Key-ID:     0x1A995A9B
               keyserver: pgp.mit.edu

Fingerprint: 4606 DA19 EC2E 9A0B 0157 C81B DA07 CF63 1A99 5A9B

[lrusak]

So should we keep the xf86-video-intel driver in the build but explicitly use the modesetting driver? That way the user can override it with a change to the udev rules. At least until more testing is done

[fritsch]

That's a good idea. If user supplied xorg.conf with intel would still work,
that would also be cool.

2016-04-17 21:16 GMT+02:00 Lukas Rusak notifications@github.com:

So should we keep the xf86-video-intel driver in the build but explicitly
use the modesetting driver? That way the user can override it with a change
to the udev rules. At least until more testing is done

—
You are receiving this because you were mentioned.
Reply to this email directly or view it on GitHub
#136 (comment)

               Key-ID:     0x1A995A9B
               keyserver: pgp.mit.edu

Fingerprint: 4606 DA19 EC2E 9A0B 0157 C81B DA07 CF63 1A99 5A9B

[MilhouseVH]

I've been including some Intel patches in my builds that are apparently useful - could they be added here, or should they no longer be used?

fritsch/OpenELEC.tv@ed77472 #Intel: Implement Passthrough-Range
fritsch/OpenELEC.tv@98c65e6 #Kodi: Use Limited Range by default
fritsch/OpenELEC.tv@e7cbe9c #xf86-video-intel: explicitely disable TearFree and TripleBuffer

[lrusak]

Apparently those commits are only good for intel and not good for the
others, so I've left them out
On Sun, Apr 17, 2016 at 12:36 PM MilhouseVH notifications@github.com
wrote:

I've been including some Intel patches in my builds that are apparently
useful - could they be added here, or should they no longer be used?

fritsch/OpenELEC.tv@ed77472
fritsch/OpenELEC.tv@ed77472
#Intel: Implement Passthrough-Range
fritsch/OpenELEC.tv@98c65e6
fritsch/OpenELEC.tv@98c65e6
#Kodi: Use Limited Range by default
fritsch/OpenELEC.tv@e7cbe9c
fritsch/OpenELEC.tv@e7cbe9c
#xf86-video-intel: explicitely disable TearFree and TripleBuffer

—
You are receiving this because you authored the thread.
Reply to this email directly or view it on GitHub
#136 (comment)

[lrusak]

updated, so we will leave xf86-video-intel in the build for now but it will default to the modesetting driver. For a user to override this they will need custom udev rules and a custom xorg.conf

@fritsch what do you think about radeon and amdgpu drivers? should they be kept?

[fritsch]

It needs testing, but according to the MLs they need to be kept. Also some
people say, that only starting with HSW or later modesetting should be
chosen. I only tested HSW and BSW. So please take great care. In doubt ...
let's do a step back and ship both.

2016-04-18 20:00 GMT+02:00 Lukas Rusak notifications@github.com:

updated, so we will leave xf86-video-intel in the build for now but it
will default to the modesetting driver. For a user to override this they
will need custom udev rules and a custom xorg.conf

@fritsch https://github.com/fritsch what do you think about radeon and
amdgpu drivers? should they be kept?

—
You are receiving this because you were mentioned.
Reply to this email directly or view it on GitHub
#136 (comment)

               Key-ID:     0x1A995A9B
               keyserver: pgp.mit.edu

Fingerprint: 4606 DA19 EC2E 9A0B 0157 C81B DA07 CF63 1A99 5A9B

[lrusak]

I'm not sure what HSW and BSW are?

[fritsch]

The chip generations. Haswell and Braswell.

[lrusak]

Should be good to go 👍

[MilhouseVH]

I'll merge this - but keep an eye on my test builds from tonight! :)

Just so you all know, my test builds will continue to include these extra patches:

fritsch/OpenELEC.tv@ed77472 #Intel: Implement Passthrough-Range
fritsch/OpenELEC.tv@98c65e6 #Kodi: Use Limited Range by default
fritsch/OpenELEC.tv@e7cbe9c #xf86-video-intel: explicitely disable TearFree and TripleBuffer
http://nmacleod.com/public/oebuild/patches/00_bump_intel_video_le.txt #xf86-video-intel: bump to 3fafabe

Let me know if you have any concerns about these additional patches (beyond it affecting AMD/Nvidia, which doesn't seem to be an issue, and may actually be of benefit).

[MilhouseVH]

@fritsch: Is the following patch still useful/needed (I've had it in my x86 test builds for a while):

https://raw.githubusercontent.com/fritsch/OpenELEC.tv/9fcfd13524a8cc1d2803242a5c8ab400254aac9d/packages/mediacenter/kodi/patches/kodi-999.999-use-2-buffers-to-match-xorg.patch

[fritsch]

That one was never merged upstream and fits on fernet's master branch. It
is useful yes, but only makes sense in combination.

2016-04-19 20:03 GMT+02:00 MilhouseVH notifications@github.com:

@fritsch https://github.com/fritsch: Is the following patch still
useful/needed (I've had it in my x86 test builds for a while):

https://raw.githubusercontent.com/fritsch/OpenELEC.tv/9fcfd13524a8cc1d2803242a5c8ab400254aac9d/packages/mediacenter/kodi/patches/kodi-999.999-use-2-buffers-to-match-xorg.patch

—
You are receiving this because you were mentioned.
Reply to this email directly or view it on GitHub
#136 (comment)

               Key-ID:     0x1A995A9B
               keyserver: pgp.mit.edu

Fingerprint: 4606 DA19 EC2E 9A0B 0157 C81B DA07 CF63 1A99 5A9B

[MilhouseVH]

Yes, it is used in combination with Fernet's branch (which configures triple-buffering by default) so I'll keep it in - thanks.

[kq01526]

@lrusak wrote:

Thanks for the tip @fritsch

apparently,

1. the intel devs suggest using the modesetting driver + glamor
   

2. Ubuntu 16.04 may ship with the modesetting driver + glamor
   

Not true:

https://bugs.freedesktop.org/show_bug.cgi?id=94843

?

Where are the sources for these statements?

[kq01526]

By the way:

@lrusak wrote:

@fritsch what do you think about radeon and amdgpu drivers? should they be kept?

Here are all the links for:

"Ditching xf86-video-amdgpu in favor of xf86-video-modesetting?"
https://bugs.freedesktop.org/show_bug.cgi?id=94841

"Ditching xf86-video-ati in favor of xf86-video-modesetting?"
https://bugs.freedesktop.org/show_bug.cgi?id=94842

"Ditching xf86-video-intel in favor of xf86-video-modesetting?"
https://bugs.freedesktop.org/show_bug.cgi?id=94843

"Ditching xf86-video-nouveau in favor of xf86-video-modesetting?"
https://bugs.freedesktop.org/show_bug.cgi?id=94844

"Ditching xf86-video-openchrome in favor of xf86-video-modesetting?"
https://bugs.freedesktop.org/show_bug.cgi?id=95524

😉