AMD-GFX Archive on lore.kernel.org
 help / color / mirror / Atom feed
From: Leo Li <sunpeng.li-5C7GfCeVMHo@public.gmane.org>
To: amd-gfx-PD4FTy7X32lNgt0PjOBp9y5qC8QIuHrW@public.gmane.org
Cc: michel-otUistvHUpPR7s880joybQ@public.gmane.org,
	harry.wentland-5C7GfCeVMHo@public.gmane.org
Subject: Re: [PATCH xf86-video-amdgpu 00/13] Enabling Color Management - Round 2
Date: Mon, 14 May 2018 09:38:04 -0400	[thread overview]
Message-ID: <6a7754d8-158b-06c4-b301-6d8d932fec46@amd.com> (raw)
In-Reply-To: <1525372315-28462-1-git-send-email-sunpeng.li-5C7GfCeVMHo@public.gmane.org>

Ping :)

Leo

On 2018-05-03 02:31 PM, sunpeng.li@amd.com wrote:
> From: "Leo (Sunpeng) Li" <sunpeng.li@amd.com>
> 
> 
> This patchset ended up looking quite different from the first. To address some
> fundamental issues, the design had to be reworked.
> 
> Things gathered from previous review:
> 
> 1. User client should not have to handle DRM blob objects. That should be the
>     job of the DDX driver.
> 
> 2. Since legacy gamma sets the same properties within DRM as non-legacy gamma,
>     the previous implementation created conflicts when setting both.
> 
>      * We should at least support this use-case: Nightlight enabled (uses legacy
>        gamma), with monitor correction enabled (non-legacy gamma)
> 
> 3. Since color management properties are attached to the CRTC, the previous
>     revision has the properties hooked onto the CRTC life-cycle, not the output
>     life-cycle. This is problematic, since clients expect properties on an
>     output to stay consistent throughout its life.
> 
> 4. Although not mentioned during review, the color properties did not persist
>     across certain events, such as DPMS state changes or hotplugs.
> 
> 
> To address the above, the following was done:
> 
> 1. XRandR allows setting of array-like properties. This is used to directly
>     pass LUT/CTM data from the client to the DDX driver.
> 
> 2. Legacy and non-legacy gamma LUTs are now merged via composition. This will
>     allow both to be in effect, while only programming one LUT in kernel driver.
> 
> 3. The three color management properties (Degamma LUT, Color Transform Matrix
>     (CTM), and Gamma LUT) are hard-coded into the DDX driver, to be listed (as
>     disabled) regardless of whether a CRTC is attached on the output, or whether
>     the kernel driver supports it.
> 
>      * If kernel driver does not support color management, the properties will
>        remain disabled. A `xrandr --set` will then error.
> 
> 4. Color properties are now *staged* inside the driver-private CRTC object.
>     This allows us to *push* the properties into kernel DRM (and consequently
>     into hardware) whenever there is a need.
> 
>      * Along with staging and pushing, an *update* function is used to notify
>        RandR to update the color properties listed on outputs. This can be used
>        when `xrandr --auto` enables a CRTC on an output, and the output need to
>        reflect the CRTC's color properties.
> 
> 
> However, there are some things being done that aren't quite nice, to which I
> don't yet have a solution. Any thoughts and suggestions are welcome:
> 
> * When using libXrandr to set the CTM property, 16bit format is used. Ideally
>    we should use 64bit format, since the CTM consists of 9x64bit fixed-point
>    values. However, it isn't recognized by XRRChangeOutputProperty as a valid
>    format. 32 bit could work, since the CTM values are S31.32 fixed-point.
>    However, using this format corrupts the data once it gets to xserver. On
>    first glance, it may be the cast to long within XRRChangeOutputProperty, in
>    addition to compiling 64 bit (shouldn't it use int32_t instead?).
> 
> * Since LUTs can be quite long, the output of `xrandr --prop` isn't very nice.
> 
> * Setting these properties through the xrandr app isn't supported without
>    modifications to the app, due to the length of these properties. However,
>    setting through libXrandr works.
> 
> Support on the xrandr app side can come as a seperate patch set. For now,
> testing the new API can be done via libXrandr. I've made a sample application
> here: git://people.freedesktop.org/~hwentland/color-demo-app
> Clone, make, and it's ready to go.
> 
> 
> Leo (Sunpeng) Li (13):
>    Add color management properties to driver-private CRTC object
>    Push color properties to kernel DRM on CRTC init
>    List disabled color properties on RandR outputs without a CRTC
>    Use CRTC's color properties if output has a CRTC attached
>    Enable setting of color properties though RandR
>    Compose non-legacy with legacy gamma LUT before pushing to kernel DRM
>    Also compose LUT when setting legacy gamma.
>    Set driver-private CRTC's dpms mode on disable
>    Move drmmode_do_crtc_dpms
>    Push staged color properties when DPMS state toggles On
>    Push staged color properties on output detect
>    Update color properties on modeset major
>    Refactor pushing color management properties into a function
> 
>   src/drmmode_display.c | 878 ++++++++++++++++++++++++++++++++++++++++++++++----
>   src/drmmode_display.h |   8 +
>   2 files changed, 824 insertions(+), 62 deletions(-)
> 
_______________________________________________
amd-gfx mailing list
amd-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/amd-gfx

  parent reply	other threads:[~2018-05-14 13:38 UTC|newest]

Thread overview: 35+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2018-05-03 18:31 [PATCH xf86-video-amdgpu 00/13] Enabling Color Management - Round 2 sunpeng.li-5C7GfCeVMHo
     [not found] ` <1525372315-28462-1-git-send-email-sunpeng.li-5C7GfCeVMHo@public.gmane.org>
2018-05-03 18:31   ` [PATCH xf86-video-amdgpu 01/13] Add color management properties to driver-private CRTC object sunpeng.li-5C7GfCeVMHo
     [not found]     ` <1525372315-28462-2-git-send-email-sunpeng.li-5C7GfCeVMHo@public.gmane.org>
2018-05-16 17:07       ` Michel Dänzer
     [not found]         ` <0dda7c80-2bdc-52a7-8dcf-ad5a7f6e9346-otUistvHUpPR7s880joybQ@public.gmane.org>
2018-05-17 21:43           ` Leo Li
2018-05-03 18:31   ` [PATCH xf86-video-amdgpu 02/13] Push color properties to kernel DRM on CRTC init sunpeng.li-5C7GfCeVMHo
     [not found]     ` <1525372315-28462-3-git-send-email-sunpeng.li-5C7GfCeVMHo@public.gmane.org>
2018-05-16 17:08       ` Michel Dänzer
     [not found]         ` <a2b8ff58-edcc-bb95-f630-1926b1a8874b-otUistvHUpPR7s880joybQ@public.gmane.org>
2018-05-17 21:43           ` Leo Li
     [not found]             ` <db641905-9dea-015d-24d2-fa25d136abfc-5C7GfCeVMHo@public.gmane.org>
2018-05-18  7:52               ` Michel Dänzer
2018-05-03 18:31   ` [PATCH xf86-video-amdgpu 03/13] List disabled color properties on RandR outputs without a CRTC sunpeng.li-5C7GfCeVMHo
     [not found]     ` <1525372315-28462-4-git-send-email-sunpeng.li-5C7GfCeVMHo@public.gmane.org>
2018-05-16 17:09       ` Michel Dänzer
     [not found]         ` <62d1f2e3-d271-658a-7bf4-88b41b080266-otUistvHUpPR7s880joybQ@public.gmane.org>
2018-05-17 21:43           ` Leo Li
2018-05-03 18:31   ` [PATCH xf86-video-amdgpu 04/13] Use CRTC's color properties if output has a CRTC attached sunpeng.li-5C7GfCeVMHo
2018-05-03 18:31   ` [PATCH xf86-video-amdgpu 05/13] Enable setting of color properties though RandR sunpeng.li-5C7GfCeVMHo
2018-05-03 18:31   ` [PATCH xf86-video-amdgpu 06/13] Compose non-legacy with legacy gamma LUT before pushing to kernel DRM sunpeng.li-5C7GfCeVMHo
2018-05-03 18:31   ` [PATCH xf86-video-amdgpu 07/13] Also compose LUT when setting legacy gamma sunpeng.li-5C7GfCeVMHo
2018-05-03 18:31   ` [PATCH xf86-video-amdgpu 08/13] Set driver-private CRTC's dpms mode on disable sunpeng.li-5C7GfCeVMHo
     [not found]     ` <1525372315-28462-9-git-send-email-sunpeng.li-5C7GfCeVMHo@public.gmane.org>
2018-05-16 17:09       ` Michel Dänzer
     [not found]         ` <12553d4f-c171-112b-bad6-6f53d47aa8e3-otUistvHUpPR7s880joybQ@public.gmane.org>
2018-05-17 21:43           ` Leo Li
     [not found]             ` <6cf602ed-b43d-d446-74a0-a39d35042f5a-5C7GfCeVMHo@public.gmane.org>
2018-05-18 10:35               ` Michel Dänzer
2018-05-03 18:31   ` [PATCH xf86-video-amdgpu 09/13] Move drmmode_do_crtc_dpms sunpeng.li-5C7GfCeVMHo
2018-05-03 18:31   ` [PATCH xf86-video-amdgpu 10/13] Push staged color properties when DPMS state toggles On sunpeng.li-5C7GfCeVMHo
     [not found]     ` <1525372315-28462-11-git-send-email-sunpeng.li-5C7GfCeVMHo@public.gmane.org>
2018-05-16 17:10       ` Michel Dänzer
     [not found]         ` <23a9fe2a-df1c-6ecb-b60e-bdcb22d8179c-otUistvHUpPR7s880joybQ@public.gmane.org>
2018-05-17 21:44           ` Leo Li
     [not found]             ` <f323d1de-d517-805f-d373-a7310ec6496c-5C7GfCeVMHo@public.gmane.org>
2018-05-18  8:01               ` Michel Dänzer
     [not found]                 ` <6500b993-624a-45aa-6310-958ca8c1f21b-otUistvHUpPR7s880joybQ@public.gmane.org>
2018-05-24 19:01                   ` Leo Li
2018-05-03 18:31   ` [PATCH xf86-video-amdgpu 11/13] Push staged color properties on output detect sunpeng.li-5C7GfCeVMHo
2018-05-03 18:31   ` [PATCH xf86-video-amdgpu 12/13] Update color properties on modeset major sunpeng.li-5C7GfCeVMHo
2018-05-03 18:31   ` [PATCH xf86-video-amdgpu 13/13] Refactor pushing color management properties into a function sunpeng.li-5C7GfCeVMHo
2018-05-14 13:38   ` Leo Li [this message]
2018-05-14 14:17   ` [PATCH xf86-video-amdgpu 00/13] Enabling Color Management - Round 2 Michel Dänzer
2018-05-16 17:06   ` Michel Dänzer
     [not found]     ` <b3a7fcde-fe69-5944-1857-c2b43fa068de-otUistvHUpPR7s880joybQ@public.gmane.org>
2018-05-17 21:43       ` Leo Li
     [not found]         ` <62c350eb-8634-5bab-e14e-6d75315e3f56-5C7GfCeVMHo@public.gmane.org>
2018-05-18  8:10           ` Michel Dänzer
     [not found]             ` <8fe3423f-8d9e-120c-32e1-7a6a7fd81606-otUistvHUpPR7s880joybQ@public.gmane.org>
2018-05-24 20:29               ` Leo Li
     [not found]                 ` <7c183caf-d2de-cf56-ed31-7922b24d3a8e-5C7GfCeVMHo@public.gmane.org>
2018-05-25  7:51                   ` Michel Dänzer

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=6a7754d8-158b-06c4-b301-6d8d932fec46@amd.com \
    --to=sunpeng.li-5c7gfcevmho@public.gmane.org \
    --cc=amd-gfx-PD4FTy7X32lNgt0PjOBp9y5qC8QIuHrW@public.gmane.org \
    --cc=harry.wentland-5C7GfCeVMHo@public.gmane.org \
    --cc=michel-otUistvHUpPR7s880joybQ@public.gmane.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox