All of lore.kernel.org
 help / color / mirror / Atom feed
From: "Michel Dänzer" <michel@daenzer.net>
To: "Christian König" <ckoenig.leichtzumerken@gmail.com>
Cc: amd-gfx@lists.freedesktop.org, dri-devel@lists.freedesktop.org
Subject: Re: Change how amdgpu stores fences in dma_resv objects
Date: Thu, 10 Jun 2021 18:34:47 +0200	[thread overview]
Message-ID: <f3788c24-bb42-4913-6334-fb3549c21fb9@daenzer.net> (raw)
In-Reply-To: <20210610091800.1833-1-christian.koenig@amd.com>

On 2021-06-10 11:17 a.m., Christian König wrote:
> Since we can't find a consensus on hot to move forward with the dma_resv object I concentrated on changing the approach for amdgpu first.
> 
> This new approach changes how the driver stores the command submission fence in the dma_resv object in DMA-buf exported BOs.
> 
> For exported BOs we now store the CS fence in a dma_fence_chain container and assign that one to the exclusive fences slot.
> 
> During synchronization this dma_fence_chain container is unpacked again and the containing fences handled individually.
> 
> This has a little bit more overhead than the old approach, but it allows for waiting for the exclusive slot for writes again.

Nice!

This seems to work as expected with https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1880: Some buffers now don't poll readable at first, until the GPU is done processing them.


Unfortunately, as expected, without a high priority context for the compositor which can preempt client drawing, this isn't enough to prevent slow clients from slowing down the compositor as well. But it should already help for fullscreen apps where the compositor can directly scan out the client buffers at least.


-- 
Earthling Michel Dänzer               |               https://redhat.com
Libre software enthusiast             |             Mesa and X developer

WARNING: multiple messages have this Message-ID (diff)
From: "Michel Dänzer" <michel@daenzer.net>
To: "Christian König" <ckoenig.leichtzumerken@gmail.com>
Cc: amd-gfx@lists.freedesktop.org, dri-devel@lists.freedesktop.org,
	daniel@ffwll.ch
Subject: Re: Change how amdgpu stores fences in dma_resv objects
Date: Thu, 10 Jun 2021 18:34:47 +0200	[thread overview]
Message-ID: <f3788c24-bb42-4913-6334-fb3549c21fb9@daenzer.net> (raw)
In-Reply-To: <20210610091800.1833-1-christian.koenig@amd.com>

On 2021-06-10 11:17 a.m., Christian König wrote:
> Since we can't find a consensus on hot to move forward with the dma_resv object I concentrated on changing the approach for amdgpu first.
> 
> This new approach changes how the driver stores the command submission fence in the dma_resv object in DMA-buf exported BOs.
> 
> For exported BOs we now store the CS fence in a dma_fence_chain container and assign that one to the exclusive fences slot.
> 
> During synchronization this dma_fence_chain container is unpacked again and the containing fences handled individually.
> 
> This has a little bit more overhead than the old approach, but it allows for waiting for the exclusive slot for writes again.

Nice!

This seems to work as expected with https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1880: Some buffers now don't poll readable at first, until the GPU is done processing them.


Unfortunately, as expected, without a high priority context for the compositor which can preempt client drawing, this isn't enough to prevent slow clients from slowing down the compositor as well. But it should already help for fullscreen apps where the compositor can directly scan out the client buffers at least.


-- 
Earthling Michel Dänzer               |               https://redhat.com
Libre software enthusiast             |             Mesa and X developer
_______________________________________________
amd-gfx mailing list
amd-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/amd-gfx

  parent reply	other threads:[~2021-06-10 16:34 UTC|newest]

Thread overview: 60+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2021-06-10  9:17 Change how amdgpu stores fences in dma_resv objects Christian König
2021-06-10  9:17 ` Christian König
2021-06-10  9:17 ` [PATCH 1/7] dma-buf: some dma_fence_chain improvements Christian König
2021-06-10  9:17   ` Christian König
2021-06-11  7:49   ` Daniel Vetter
2021-06-11  7:49     ` Daniel Vetter
2021-06-10  9:17 ` [PATCH 2/7] dma-buf: add dma_fence_chain_alloc/free Christian König
2021-06-10  9:17   ` Christian König
2021-06-11  7:54   ` Daniel Vetter
2021-06-11  7:54     ` Daniel Vetter
2021-06-11 11:48     ` Christian König
2021-06-11 11:48       ` Christian König
2021-06-11 14:40       ` Daniel Vetter
2021-06-11 14:40         ` Daniel Vetter
2021-06-10  9:17 ` [PATCH 3/7] dma-buf: add dma_fence_chain_alloc/free self tests Christian König
2021-06-10  9:17   ` Christian König
2021-06-11  7:58   ` Daniel Vetter
2021-06-11  7:58     ` Daniel Vetter
2021-06-11 10:04     ` Christian König
2021-06-11 10:04       ` Christian König
2021-06-10  9:17 ` [PATCH 4/7] dma-buf: add dma_fence_chain_garbage_collect Christian König
2021-06-10  9:17   ` Christian König
2021-06-11  8:58   ` Daniel Vetter
2021-06-11  8:58     ` Daniel Vetter
2021-06-11 10:07     ` Christian König
2021-06-11 10:07       ` Christian König
2021-06-11 15:11       ` Daniel Vetter
2021-06-11 15:11         ` Daniel Vetter
2021-06-16 17:29   ` kernel test robot
2021-06-16 17:29     ` kernel test robot
2021-06-16 17:29     ` kernel test robot
2021-06-16 18:50   ` kernel test robot
2021-06-16 18:50     ` kernel test robot
2021-06-16 18:50     ` kernel test robot
2021-06-10  9:17 ` [PATCH 5/7] drm/syncobj: drop the manual garbage collection Christian König
2021-06-10  9:17   ` Christian König
2021-06-10  9:17 ` [PATCH 6/7] drm/amdgpu: unwrap fence chains in the explicit sync fence Christian König
2021-06-10  9:17   ` Christian König
2021-06-11  9:07   ` Daniel Vetter
2021-06-11  9:07     ` Daniel Vetter
2021-06-11 10:09     ` Christian König
2021-06-11 10:09       ` Christian König
2021-06-11 15:18       ` Daniel Vetter
2021-06-11 15:18         ` Daniel Vetter
2021-06-14  7:25         ` Christian König
2021-06-14  7:25           ` Christian König
2021-06-17 16:51           ` Daniel Vetter
2021-06-17 16:51             ` Daniel Vetter
2021-06-10  9:18 ` [PATCH 7/7] drm/amdgpu: rework dma_resv handling Christian König
2021-06-10  9:18   ` Christian König
2021-06-11  9:17   ` Daniel Vetter
2021-06-11  9:17     ` Daniel Vetter
2021-06-11 10:12     ` Christian König
2021-06-11 10:12       ` Christian König
2021-06-11 15:21       ` Daniel Vetter
2021-06-11 15:21         ` Daniel Vetter
2021-06-10 16:34 ` Michel Dänzer [this message]
2021-06-10 16:34   ` Change how amdgpu stores fences in dma_resv objects Michel Dänzer
2021-06-10 16:44   ` Christian König
2021-06-10 16:44     ` Christian König

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=f3788c24-bb42-4913-6334-fb3549c21fb9@daenzer.net \
    --to=michel@daenzer.net \
    --cc=amd-gfx@lists.freedesktop.org \
    --cc=ckoenig.leichtzumerken@gmail.com \
    --cc=dri-devel@lists.freedesktop.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 an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.