All of lore.kernel.org
 help / color / mirror / Atom feed
From: Matthew Auld <matthew.auld@intel.com>
To: Maarten Lankhorst <dev@lankhorst.se>, intel-xe@lists.freedesktop.org
Cc: dri-devel@lists.freedesktop.org
Subject: Re: [PATCH 2/3] drm/xe/display: Avoid using stolen memory for framebuffer when media gt exists.
Date: Wed, 1 Jul 2026 11:27:31 +0100	[thread overview]
Message-ID: <06860a7f-0df8-4e8d-b4ff-c6ebabc5fa4c@intel.com> (raw)
In-Reply-To: <20260630135523.1775379-3-dev@lankhorst.se>

On 30/06/2026 14:55, Maarten Lankhorst wrote:
> On systems with media GT, extra latency is added when accessing stolen
> memory when the GT is in MC6.
> 
> Simply disable allocating stolen memory for framebuffers when media gt
> is found.
> 
> Link: https://gitlab.freedesktop.org/drm/xe/kernel/-/work_items/7513
> Signed-off-by: Maarten Lankhorst <dev@lankhorst.se>
> ---
>   drivers/gpu/drm/xe/display/xe_display_bo.c | 7 +++++++
>   1 file changed, 7 insertions(+)
> 
> diff --git a/drivers/gpu/drm/xe/display/xe_display_bo.c b/drivers/gpu/drm/xe/display/xe_display_bo.c
> index 8953da0136dcb..677ae3af9ac0c 100644
> --- a/drivers/gpu/drm/xe/display/xe_display_bo.c
> +++ b/drivers/gpu/drm/xe/display/xe_display_bo.c
> @@ -131,6 +131,13 @@ bool xe_display_bo_fbdev_prefer_stolen(struct xe_device *xe, unsigned int size)
>   	if (IS_DGFX(xe))
>   		return false;
>   
> +	/*
> +	 * Avoid stolen memory when the media_gt exists,
> +	 * because a lot of latency is added when media gt is in MC6
> +	 */
> +	if (xe_device_get_root_tile(xe)->media_gt)
> +		return false;
> +
>   	if (XE_DEVICE_WA(xe, 22019338487_display))
>   		return false;

There is also the question for why this WA only applies to LNL, but the 
GT side of the WA applies even to PTL+, which seems highly suspect. 
Might be worth mentioning that in the commit. Also wondering if can now 
just drop 22019338487_display?

Also not sure if should just go with platform check here:

if (info.platform >= LNL)

?

Either way:
Reviewed-by: Matthew Auld <matthew.auld@intel.com>

>   


  reply	other threads:[~2026-07-01 10:27 UTC|newest]

Thread overview: 14+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2026-06-30 13:55 [PATCH 0/3] drm/xe: Disable stolen memory for driver allocations Maarten Lankhorst
2026-06-30 13:55 ` [PATCH 1/3] drm/xe: Avoid using stolen memory for DPT Maarten Lankhorst
2026-06-30 14:05   ` sashiko-bot
2026-06-30 14:10   ` Matthew Auld
2026-07-01 10:14     ` Maarten Lankhorst
2026-07-01 10:20       ` Matthew Auld
2026-06-30 13:55 ` [PATCH 2/3] drm/xe/display: Avoid using stolen memory for framebuffer when media gt exists Maarten Lankhorst
2026-07-01 10:27   ` Matthew Auld [this message]
2026-07-01 10:28     ` Matthew Auld
2026-06-30 13:55 ` [PATCH 3/3] drm/xe/display: Do not allocate into stolen for new framebuffers Maarten Lankhorst
2026-07-01 10:33   ` Matthew Auld
2026-06-30 14:25 ` ✓ CI.KUnit: success for drm/xe: Disable stolen memory for driver allocations Patchwork
2026-06-30 14:59 ` ✓ Xe.CI.BAT: " Patchwork
2026-07-01  6:40 ` ✓ Xe.CI.FULL: " Patchwork

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=06860a7f-0df8-4e8d-b4ff-c6ebabc5fa4c@intel.com \
    --to=matthew.auld@intel.com \
    --cc=dev@lankhorst.se \
    --cc=dri-devel@lists.freedesktop.org \
    --cc=intel-xe@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.