ACK: [SRU][Kinetic][PATCH 0/1] vmwgfx fails to reserve graphics buffer on aarch64 leading to blank display

[Andrei Gherzan]

On 23/04/20 12:21PM, Matthew Ruffell wrote:
> BugLink: https://bugs.launchpad.net/bugs/2007001
> 
> [Impact]
> 
> Numerous VMWare users have reported that vmwgfx cannot reserve the memory
> region for the graphics framebuffer, leading their VMs to have blank screens.
> 
> They see the following in dmesg:
> 
> [ 11.135360] vmwgfx 0000:00:0f.0: BAR 2: can't reserve [mem 0x70000000-0x77ffffff 64bit pref]
> [ 11.135366] vmwgfx: probe of 0000:00:0f.0 failed with error -16
> 
> And a cat /proc/iomem shows this:
> 
> 50000000-7fffffff : PCI Bus 0000:00
>   70000000-77ffffff : 0000:00:0f.0
>     70000000-702fffff : BOOTFB
> 
> The kernel has failed to release this memory region for vmwgfx to occupy.
> 
> Most affected users are on aarch64, with the host being Apple silicon systems.
> 
> [Fix]
> 
> The regression was introduced by the below commit in 5.19.0-30-generic:
> 
> commit 5e01376124309b4dbd30d413f43c0d9c2f60edea
> Author: Thomas Zimmermann <tzimmermann at suse.de>
> Date: Mon Jul 18 09:23:18 2022 +0200
> Subject: video/aperture: Disable and unregister sysfb devices via aperture helpers
> Link: https://github.com/torvalds/linux/commit/5e01376124309b4dbd30d413f43c0d9c2f60edea
> 
> This commit was part of a larger refactoring of the video subsystem, and
> requires the entire series to function correctly. You can review the whole
> series below:
> 
> https://patchwork.freedesktop.org/series/106040/
> 
> The patch series also requires quite a few additional fixups to fix bugs
> introduced by the series, making the size about 15 commits in total. The
> contents of the series don't really fix any bugs, and their purpose is to
> refactor the code for future changes to the fbdev subsystem, and really aren't
> appropriate to be backported to a stable kernel series.
> 
> "video/aperture: Disable and unregister sysfb devices via aperture helpers"
> seems to have been selected for -stable by mistake by its fixes: tag, and was
> pulled into upstream stable by a robot with little human review.
> 
> The best course of action is to revert. No action needed for Lunar, as the
> entire series is present in that release.
> 
> [Testcase]
> 
> This bug affects users running Ubuntu in VMWare VMs, notably on aarch64 devices,
> like modern Apple computers. 
> 
> Start a Kinetic or Jammy-HWE Server or Desktop VM in VMWare Fusion on Apple
> silicon, and see if the display comes up or not.
> 
> Affected users will see a blank screen.
> 
> There is a test kernel available in the following ppa:
> 
> https://launchpad.net/~mruffell/+archive/ubuntu/lp2007001-test
> 
> If you install the test kernel and reboot, you will be able to see the screen
> on your VM like normal.
> 
> [Where problems could occur]
> 
> This commit changes when the sysfb is disabled and memory region for the
> graphics framebuffer is released to the proper device driver.
> 
> If a regression were to occur, then graphics drivers may fail to reserve the
> framebuffer memory, and fail to start, leaving users with a blank screen.
> 
> There are no workarounds, other than booting a previous kernel.
> 
> Matthew Ruffell (1):
>   UBUNTU: SAUCE: Revert "video/aperture: Disable and unregister sysfb
>     devices via aperture helpers"
> 
>  drivers/gpu/drm/drm_aperture.c   | 14 --------------
>  drivers/video/fbdev/core/fbmem.c | 12 ++++++++++++
>  2 files changed, 12 insertions(+), 14 deletions(-)
> 
> -- 
> 2.39.2

Acked-by: Andrei Gherzan <andrei.gherzan at canonical.com>

-- 
Andrei Gherzan
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 833 bytes
Desc: not available
URL: <https://lists.ubuntu.com/archives/kernel-team/attachments/20230420/2ed50e4a/attachment.sig>