Re: [PATCH [RFC] v3 5/7] drm/atomic: Allow planes with NULL fb along with async flip

2026-01-12 Thread Murthy, Arun R

On 12-01-2026 21:33, Xaver Hugl wrote:

I see no request for being able to mix sync & async plane updates in a single 
commit.


Sorry maybe I might be creating more confusion or my words are not
giving clarity.

Let me try to put it in simple words.
The
comment(https://gitlab.freedesktop.org/drm/i915/kernel/-/issues/13834#note_2857640)
from Xaver says that disabling of cursor plane along with a sync flip
should work.
Uma in the
comment(https://gitlab.freedesktop.org/drm/i915/kernel/-/issues/13834#note_2863599)
has acknowledged the changes.

That's about a plane that was previously already disabled by a
non-async atomic commit. The issue is about the async commit failing
in a sequence like this:
- disable cursor plane in sync commit
- disable cursor plane in async commit + flip on the primary plane

I would *in general* like to be able to use the cursor plane while
doing async flips on the primary one, but I think doing that requires
more discussion on the high level API design first.


Few discussion around this in the v2 version of this series.

*https://lore.kernel.org/dri-devel/[email protected]/*

Thanks and Regards,
Arun R Murthy



Re: [PATCH [RFC] v3 5/7] drm/atomic: Allow planes with NULL fb along with async flip

2026-01-12 Thread Xaver Hugl
> > I see no request for being able to mix sync & async plane updates in a 
> > single commit.
> >
> Sorry maybe I might be creating more confusion or my words are not
> giving clarity.
>
> Let me try to put it in simple words.
> The
> comment(https://gitlab.freedesktop.org/drm/i915/kernel/-/issues/13834#note_2857640)
> from Xaver says that disabling of cursor plane along with a sync flip
> should work.
> Uma in the
> comment(https://gitlab.freedesktop.org/drm/i915/kernel/-/issues/13834#note_2863599)
> has acknowledged the changes.
That's about a plane that was previously already disabled by a
non-async atomic commit. The issue is about the async commit failing
in a sequence like this:
- disable cursor plane in sync commit
- disable cursor plane in async commit + flip on the primary plane

I would *in general* like to be able to use the cursor plane while
doing async flips on the primary one, but I think doing that requires
more discussion on the high level API design first.

- Xaver


Re: [PATCH [RFC] v3 5/7] drm/atomic: Allow planes with NULL fb along with async flip

2026-01-12 Thread Michel Dänzer
On 1/12/26 16:20, Murthy, Arun R wrote:
> 
> On 12-01-2026 16:54, Michel Dänzer wrote:
>> On 1/12/26 09:26, Murthy, Arun R wrote:
>>> On 09-01-2026 16:53, Michel Dänzer wrote:
 On 1/9/26 12:08, Murthy, Arun R wrote:
>> From: Michel Dänzer 
>> On 1/8/26 10:43, Arun R Murthy wrote:
>>> Along with async flip if there is a request to disable a sync plane by
>>> providing a NULL fb allow them.
>> That could result in async changes to other planes taking effect in an 
>> earlier
>> refresh cycle than the sync plane being disabled, couldn't it? In which 
>> case the
>> commit arguably wouldn't actually be "atomic".
>>
> This is the request from the community to allow disabling of a sync plane 
> in an async flip atomic ioctl.
> https://gitlab.freedesktop.org/drm/i915/kernel/-/issues/13834
 Can't see any such request there. I suspect there might be a 
 misunderstanding.
>>> Here cursor is a sync flip.
>>> https://gitlab.freedesktop.org/drm/i915/kernel/-/issues/13834#note_2855843
>> Not sure what you mean.
>>
>> The cursor plane is disabled in KWin's atomic commits with 
>> DRM_MODE_PAGE_FLIP_ASYNC, so it's irrelevant for them.
>>
>> In the comment following the one you referenced, Xaver (one of the main KWin 
>> developers) agreed that KWin not setting DRM_MODE_PAGE_FLIP_ASYNC in the 
>> commit which disables the cursor plane is fine.
>>
>> I see no request for being able to mix sync & async plane updates in a 
>> single commit.
>>
> Sorry maybe I might be creating more confusion or my words are not giving 
> clarity.
> 
> Let me try to put it in simple words.
> The 
> comment(https://gitlab.freedesktop.org/drm/i915/kernel/-/issues/13834#note_2857640)
>  from Xaver says that disabling of cursor plane along with a sync flip should 
> work.

Honestly can't see how that comment could be interpreted that way.

Maybe Xaver can clarify.


-- 
Earthling Michel Dänzer   \GNOME / Xwayland / Mesa developer
https://redhat.com \   Libre software enthusiast


Re: [PATCH [RFC] v3 5/7] drm/atomic: Allow planes with NULL fb along with async flip

2026-01-12 Thread Murthy, Arun R



On 12-01-2026 16:54, Michel Dänzer wrote:

On 1/12/26 09:26, Murthy, Arun R wrote:

On 09-01-2026 16:53, Michel Dänzer wrote:

On 1/9/26 12:08, Murthy, Arun R wrote:

From: Michel Dänzer 
On 1/8/26 10:43, Arun R Murthy wrote:

Along with async flip if there is a request to disable a sync plane by
providing a NULL fb allow them.

That could result in async changes to other planes taking effect in an earlier
refresh cycle than the sync plane being disabled, couldn't it? In which case the
commit arguably wouldn't actually be "atomic".


This is the request from the community to allow disabling of a sync plane in an 
async flip atomic ioctl.
https://gitlab.freedesktop.org/drm/i915/kernel/-/issues/13834

Can't see any such request there. I suspect there might be a misunderstanding.

Here cursor is a sync flip.
https://gitlab.freedesktop.org/drm/i915/kernel/-/issues/13834#note_2855843

Not sure what you mean.

The cursor plane is disabled in KWin's atomic commits with 
DRM_MODE_PAGE_FLIP_ASYNC, so it's irrelevant for them.

In the comment following the one you referenced, Xaver (one of the main KWin 
developers) agreed that KWin not setting DRM_MODE_PAGE_FLIP_ASYNC in the commit 
which disables the cursor plane is fine.

I see no request for being able to mix sync & async plane updates in a single 
commit.

Sorry maybe I might be creating more confusion or my words are not 
giving clarity.


Let me try to put it in simple words.
The 
comment(https://gitlab.freedesktop.org/drm/i915/kernel/-/issues/13834#note_2857640) 
from Xaver says that disabling of cursor plane along with a sync flip 
should work.
Uma in the 
comment(https://gitlab.freedesktop.org/drm/i915/kernel/-/issues/13834#note_2863599) 
has acknowledged the changes.


Thanks and Regards,
Arun R Murthy




Re: [PATCH [RFC] v3 5/7] drm/atomic: Allow planes with NULL fb along with async flip

2026-01-12 Thread Michel Dänzer
On 1/12/26 09:26, Murthy, Arun R wrote:
> On 09-01-2026 16:53, Michel Dänzer wrote:
>> On 1/9/26 12:08, Murthy, Arun R wrote:
 From: Michel Dänzer 
 On 1/8/26 10:43, Arun R Murthy wrote:
> Along with async flip if there is a request to disable a sync plane by
> providing a NULL fb allow them.
 That could result in async changes to other planes taking effect in an 
 earlier
 refresh cycle than the sync plane being disabled, couldn't it? In which 
 case the
 commit arguably wouldn't actually be "atomic".

>>> This is the request from the community to allow disabling of a sync plane 
>>> in an async flip atomic ioctl.
>>> https://gitlab.freedesktop.org/drm/i915/kernel/-/issues/13834
>> Can't see any such request there. I suspect there might be a 
>> misunderstanding.
> 
> Here cursor is a sync flip.
> https://gitlab.freedesktop.org/drm/i915/kernel/-/issues/13834#note_2855843

Not sure what you mean.

The cursor plane is disabled in KWin's atomic commits with 
DRM_MODE_PAGE_FLIP_ASYNC, so it's irrelevant for them.

In the comment following the one you referenced, Xaver (one of the main KWin 
developers) agreed that KWin not setting DRM_MODE_PAGE_FLIP_ASYNC in the commit 
which disables the cursor plane is fine.

I see no request for being able to mix sync & async plane updates in a single 
commit.


-- 
Earthling Michel Dänzer   \GNOME / Xwayland / Mesa developer
https://redhat.com \   Libre software enthusiast


Re: [PATCH [RFC] v3 5/7] drm/atomic: Allow planes with NULL fb along with async flip

2026-01-12 Thread Murthy, Arun R

On 09-01-2026 16:53, Michel Dänzer wrote:

On 1/9/26 12:08, Murthy, Arun R wrote:

From: Michel Dänzer 
On 1/8/26 10:43, Arun R Murthy wrote:

Along with async flip if there is a request to disable a sync plane by
providing a NULL fb allow them.

That could result in async changes to other planes taking effect in an earlier
refresh cycle than the sync plane being disabled, couldn't it? In which case the
commit arguably wouldn't actually be "atomic".


This is the request from the community to allow disabling of a sync plane in an 
async flip atomic ioctl.
https://gitlab.freedesktop.org/drm/i915/kernel/-/issues/13834

Can't see any such request there. I suspect there might be a misunderstanding.


Here cursor is a sync flip.
https://gitlab.freedesktop.org/drm/i915/kernel/-/issues/13834#note_2855843

Thanks and Regards,
Arun R Murthy




Re: [PATCH [RFC] v3 5/7] drm/atomic: Allow planes with NULL fb along with async flip

2026-01-09 Thread Michel Dänzer
On 1/9/26 12:08, Murthy, Arun R wrote:
>> From: Michel Dänzer 
>> On 1/8/26 10:43, Arun R Murthy wrote:
>>> Along with async flip if there is a request to disable a sync plane by
>>> providing a NULL fb allow them.
>>
>> That could result in async changes to other planes taking effect in an 
>> earlier
>> refresh cycle than the sync plane being disabled, couldn't it? In which case 
>> the
>> commit arguably wouldn't actually be "atomic".
>>
> This is the request from the community to allow disabling of a sync plane in 
> an async flip atomic ioctl.
> https://gitlab.freedesktop.org/drm/i915/kernel/-/issues/13834

Can't see any such request there. I suspect there might be a misunderstanding.


-- 
Earthling Michel Dänzer   \GNOME / Xwayland / Mesa developer
https://redhat.com \   Libre software enthusiast


RE: [PATCH [RFC] v3 5/7] drm/atomic: Allow planes with NULL fb along with async flip

2026-01-09 Thread Murthy, Arun R

> -Original Message-
> From: Michel Dänzer 
> Sent: Friday, January 9, 2026 3:14 PM
> To: Murthy, Arun R ; Maarten Lankhorst
> ; Maxime Ripard ;
> Thomas Zimmermann ; David Airlie
> ; Simona Vetter ; Jani Nikula
> ; Vivi, Rodrigo ; Joonas
> Lahtinen ; Tvrtko Ursulin
> ; [email protected]; [email protected]; Kumar,
> Naveen1 ; Syrjala, Ville ;
> Dmitry Baryshkov 
> Cc: [email protected]; [email protected]; intel-
> [email protected]
> Subject: Re: [PATCH [RFC] v3 5/7] drm/atomic: Allow planes with NULL fb along
> with async flip
> 
> On 1/8/26 10:43, Arun R Murthy wrote:
> > Along with async flip if there is a request to disable a sync plane by
> > providing a NULL fb allow them.
> 
> That could result in async changes to other planes taking effect in an earlier
> refresh cycle than the sync plane being disabled, couldn't it? In which case 
> the
> commit arguably wouldn't actually be "atomic".
> 
This is the request from the community to allow disabling of a sync plane in an 
async flip atomic ioctl.
https://gitlab.freedesktop.org/drm/i915/kernel/-/issues/13834

Thanks and Regards,
Arun R Murthy
---


Re: [PATCH [RFC] v3 5/7] drm/atomic: Allow planes with NULL fb along with async flip

2026-01-09 Thread Michel Dänzer
On 1/8/26 10:43, Arun R Murthy wrote:
> Along with async flip if there is a request to disable a sync plane by
> providing a NULL fb allow them.

That could result in async changes to other planes taking effect in an earlier 
refresh cycle than the sync plane being disabled, couldn't it? In which case 
the commit arguably wouldn't actually be "atomic".


-- 
Earthling Michel Dänzer   \GNOME / Xwayland / Mesa developer
https://redhat.com \   Libre software enthusiast