ACK/Cmnt: [SRU][resolute][PATCH 0/1] drm/xe/display: allow external dma-buf framebuffers

Hui Wang hui.wang at canonical.com
Tue Jun 30 08:01:58 UTC 2026


Please copy the SRU template to the tracking bug, and create linux 
(resolute) target in the tracking bug.

Acked-by: Hui Wang <hui.wang at canonical.com>

On 6/30/26 12:49, ChunAn Wu via kernel-team wrote:
> BugLink: https://bugs.launchpad.net/bugs/2158605
>
> [Impact]
> Screen garbage can show when running glxgears fullscreen on affected
> Intel graphics systems.
>
> The Launchpad report says it happens about 1 in 3 runs.
>
> [Fix]
> Skip the SCANOUT flag and VM_BIND checks for external dma-buf imports
> in the xe display framebuffer setup and pin paths.
>
> For imported sg buffers, CPU caching is controlled by the exporter.
> The patch keeps the existing checks for local buffers.
>
> This patch is from upstream Patchwork, not yet merged:
> https://patchwork.freedesktop.org/patch/732636/
>
> [Test Plan]
> Run the fullscreen glxgears test 10 times:
> $ glxgears -fullscreen
>
> Or run the matching checkbox test for the affected GPU:
> $ checkbox-cli run graphics/2_valid_glxgears_fullscreen_PCI_ID_<affected-gpu>
>
> Without patch: screen shows garbage during the fullscreen glxgears test.
> With patch: fullscreen glxgears completes without screen garbage.
>
> [Where problems could occur]
> Could break external dma-buf framebuffer setup in the xe display driver.
>
> If the sg-buffer exception is wrong, imported framebuffers could be
> accepted when they should be rejected. This could show as display
> corruption, failed framebuffer creation, or xe framebuffer pinning
> warnings.
>
> [Other Info]
> SRU target: linux-oem-6.17.
> The patch is not in an upstream kernel release yet.
>
> Matthew Auld (1):
>    UBUNTU: SAUCE: drm/xe/display: skip FORCE_WC and vm_bound check for external dma-bufs
>
>   drivers/gpu/drm/xe/display/intel_fb_bo.c | 3 ++-
>   drivers/gpu/drm/xe/display/xe_fb_pin.c   | 3 ++-
>   2 files changed, 4 insertions(+), 2 deletions(-)
>



More information about the kernel-team mailing list