The kernel versions including the following commits are referenced:

DRM_CLIENT_CAP_STEREO_3D
61d8e328 ("drm: Add a STEREO_3D capability to the SET_CLIENT_CAP ioctl")

DRM_CLIENT_CAP_UNIVERSAL_PLANES
681e7ec7 ("drm: Allow userspace to ask for universal plane list (v2)")
c7dbc6c9 ("drm: Remove command line guard for universal planes")

DRM_CLIENT_CAP_ATOMIC
88a48e29 ("drm: add atomic properties")
8b72ce15 ("drm: Always enable atomic API")

DRM_CLIENT_CAP_ASPECT_RATIO
7595bda2 ("drm: Add DRM client cap for aspect-ratio")

DRM_CLIENT_CAP_WRITEBACK_CONNECTORS
d67b6a20 ("drm: writeback: Add client capability for exposing writeback connectors")
Signed-off-by: NSimon Ser <contact@emersion.fr>
Reviewed-by: NDaniel Vetter <daniel.vetter@ffwll.ch>
Acked-by: NDaniel Stone <daniels@collabora.com>
Acked-by: NPekka Paalanen <pekka.paalanen@collabora.com>
Link: https://patchwork.freedesktop.org/patch/434202/

Make it clear that the client is responsible for enabling ATOMIC
prior to enabling WRITEBACK_CONNECTORS. Linkify the reference to
ATOMIC.
Signed-off-by: NSimon Ser <contact@emersion.fr>
Reviewed-by: NDaniel Vetter <daniel.vetter@ffwll.ch>
Acked-by: NDaniel Stone <daniels@collabora.com>
Acked-by: NPekka Paalanen <pekka.paalanen@collabora.com>
Link: https://patchwork.freedesktop.org/patch/434200/

In the docs for DRM_CLIENT_CAP_STEREO_3D and
DRM_CLIENT_CAP_ASPECT_RATIO, reference the DRM_MODE_FLAG_* defines
that get set when the cap is enabled.
Signed-off-by: NSimon Ser <contact@emersion.fr>
Reviewed-by: NDaniel Vetter <daniel.vetter@ffwll.ch>
Acked-by: NDaniel Stone <daniels@collabora.com>
Acked-by: NPekka Paalanen <pekka.paalanen@collabora.com>
Link: https://patchwork.freedesktop.org/patch/434201/

pm_runtime_get_sync will increment pm usage counter even it failed.
Forgetting to putting operation will result in reference leak here.
Fix it by replacing it with pm_runtime_resume_and_get to keep usage
counter balanced.
Reported-by: NHulk Robot <hulkci@huawei.com>
Signed-off-by: NYu Kuai <yukuai3@huawei.com>
Reviewed-by: NRobert Foss <robert.foss@linaro.org>
Signed-off-by: NRobert Foss <robert.foss@linaro.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20210531135622.3348252-1-yukuai3@huawei.com

The Lontium bridge is including legacy header files for GPIO
but not using them. Delete the includes.

Cc: Adrien Grassein <adrien.grassein@gmail.com>
Cc: Robert Foss <robert.foss@linaro.org>
Signed-off-by: NLinus Walleij <linus.walleij@linaro.org>
Reviewed-by: NAdren Grassein <adrien.grassein@gmail.com>
Signed-off-by: NRobert Foss <robert.foss@linaro.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20210529002759.468964-1-linus.walleij@linaro.org

To reduce memory various Rockchip VOP versions share
common reg structures. However more recent added SoCs not
always have to same futures as the old ones.
Add PX30 missing version info, so all VOP version checks
work correct if needed in the future.
Signed-off-by: NJohan Jonker <jbx6244@gmail.com>
Signed-off-by: NHeiko Stuebner <heiko@sntech.de>
Link: https://patchwork.freedesktop.org/patch/msgid/20210131125016.10837-1-jbx6244@gmail.com

The variable bit_per_pix is a u8 and is promoted in the multiplication
to an int type and then sign extended to a u64. If the result of the
int multiplication is greater than 0x7fffffff then the upper 32 bits will
be set to 1 as a result of the sign extension. Avoid this by casting
tu_size_reg to u64 to avoid sign extension and also a potential overflow.

Fixes: 1a0f7ed3 ("drm/rockchip: cdn-dp: add cdn DP support for rk3399")
Signed-off-by: NColin Ian King <colin.king@canonical.com>
Reviewed-by: NGuenter Roeck <groeck@chromium.org>
Signed-off-by: NHeiko Stuebner <heiko@sntech.de>
Link: https://patchwork.freedesktop.org/patch/msgid/20200915162049.36434-1-colin.king@canonical.com

'ret' is know to be 0 a this point. Checking the return value of
'phy_init()' and 'phy_set_mode()' was intended instead.

So add the missing assignments.

Fixes: cca1705c ("drm/rockchip: lvds: Add PX30 support")
Signed-off-by: NChristophe JAILLET <christophe.jaillet@wanadoo.fr>
Signed-off-by: NHeiko Stuebner <heiko@sntech.de>
Link: https://patchwork.freedesktop.org/patch/msgid/248220d4815dc8c8088cebfab7d6df5f70518438.1619881852.git.christophe.jaillet@wanadoo.fr

Fix the following clang warning:

drivers/gpu/drm/rockchip/dw-mipi-dsi-rockchip.c:320:20: warning: unused
function 'dsi_set' [-Wunused-function].
Reported-by: NAbaci Robot <abaci@linux.alibaba.com>
Signed-off-by: NJiapeng Chong <jiapeng.chong@linux.alibaba.com>
Signed-off-by: NHeiko Stuebner <heiko@sntech.de>
Link: https://patchwork.freedesktop.org/patch/msgid/1618476421-114429-1-git-send-email-jiapeng.chong@linux.alibaba.com

commit cf6d100d ("drm/rockchip: dsi: add dual mipi support") added
this devcnt field and call to component_del(). However, these both
appear to be erroneous changes left over from an earlier version of the
patch. In the version merged, nothing ever modifies devcnt, meaning
component_del() runs unconditionally and in addition to the
component_del() calls in dw_mipi_dsi_rockchip_host_detach(). The second
call fails to delete anything and produces a warning in dmesg.

If we look at the previous version of the patch[1], however, we see that
it had logic to calculate devcnt and call component_add() in certain
situations. This was removed in v6, and the fact that the deletion code
was not appears to have been an oversight.

[1] https://patchwork.kernel.org/project/dri-devel/patch/20180821140515.22246-8-heiko@sntech.de/

Fixes: cf6d100d ("drm/rockchip: dsi: add dual mipi support")
Cc: stable@vger.kernel.org
Signed-off-by: NThomas Hebb <tommyhebb@gmail.com>
Signed-off-by: NHeiko Stuebner <heiko@sntech.de>
Link: https://patchwork.freedesktop.org/patch/msgid/201385acb0eeb5dfb037afdc6a94bfbcdab97f99.1618797778.git.tommyhebb@gmail.com

When we first enable the DSI encoder, we currently program some per-chip
configuration that we look up in rk3399_chip_data based on the device
tree compatible we match. This data configures various parameters of the
MIPI lanes, including on RK3399 whether DSI1 is slaved to DSI0 in a
dual-mode configuration. It also selects which LCDC (i.e. VOP) to scan
out from.

This causes a problem in RK3399 dual-mode configurations, though: panel
prepare() callbacks run before the encoder gets enabled and expect to be
able to write commands to the DSI bus, but the bus isn't fully
functional until the lane and master/slave configuration have been
programmed. As a result, dual-mode panels (and possibly others too) fail
to turn on when the rockchipdrm driver is initially loaded.

Because the LCDC mux is the only thing we don't know until enable time
(and is the only thing that can ever change), we can actually move most
of the initialization to bind() and get it out of the way early. That's
what this change does. (Rockchip's 4.4 BSP kernel does it in mode_set(),
which also avoids the issue, but bind() seems like the more correct
place to me.)

Tested on a Google Scarlet board (Acer Chromebook Tab 10), which has a
Kingdisplay KD097D04 dual-mode panel. Prior to this change, the panel's
backlight would turn on but no image would appear when initially loading
rockchipdrm. If I kept rockchipdrm loaded and reloaded the panel driver,
it would come on. With this change, the panel successfully turns on
during initial rockchipdrm load as expected.

Fixes: 2d4f7bda ("drm/rockchip: dsi: migrate to use dw-mipi-dsi bridge driver")
Signed-off-by: NThomas Hebb <tommyhebb@gmail.com>
Tested-by: NJonathan Liu <net147@gmail.com>
Signed-off-by: NHeiko Stuebner <heiko@sntech.de>
Link: https://patchwork.freedesktop.org/patch/msgid/55fe7f3454d8c91dc3837ba5aa741d4a0e67378f.1618797813.git.tommyhebb@gmail.com

The CDP DP component of the rockchip GPU driver requires firmware
so define MODULE_FIRMWARE for rockchip/dptx.bin so the details
are available in the module info.
Signed-off-by: NPeter Robinson <pbrobinson@gmail.com>
Signed-off-by: NHeiko Stuebner <heiko@sntech.de>
Link: https://patchwork.freedesktop.org/patch/msgid/20210506092951.515813-1-pbrobinson@gmail.com

After calling clk_prepare_enable(), clk_disable_unprepare() need
be called when calling regmap_write() failed.

Fixes: 1a0f7ed3 ("drm/rockchip: cdn-dp: add cdn DP support for rk3399")
Reported-by: NHulk Robot <hulkci@huawei.com>
Signed-off-by: NYang Yingliang <yangyingliang@huawei.com>
Signed-off-by: NHeiko Stuebner <heiko@sntech.de>
Link: https://patchwork.freedesktop.org/patch/msgid/20210519134928.2696617-1-yangyingliang@huawei.com

alpha_en should be set to 0 if it is not used, i.e. to disable alpha
blending if it was enabled before and should be disabled now.

Fixes: 2aae8ed1 ("drm/rockchip: Add per-pixel alpha support for the PX30 VOP")
Signed-off-by: NAlex Bee <knaerzche@gmail.com>
Signed-off-by: NHeiko Stuebner <heiko@sntech.de>
Link: https://patchwork.freedesktop.org/patch/msgid/20210528130554.72191-6-knaerzche@gmail.com

With
commit 2aae8ed1 ("drm/rockchip: Add per-pixel alpha support for the PX30 VOP")
alpha support was introduced for PX30's VOP.
RK3036, RK3066, RK3126 and RK3188 VOPs support alpha blending in the
same manner.
With the exception of RK3066 all of them support pre-multiplied alpha.
Signed-off-by: NAlex Bee <knaerzche@gmail.com>
Signed-off-by: NHeiko Stuebner <heiko@sntech.de>
Link: https://patchwork.freedesktop.org/patch/msgid/20210528130554.72191-5-knaerzche@gmail.com

Add dither_up, dsp_lut_en and data_blank registers to enable their
respective functionality for RK3066's VOP.

While at that also fix .rb_swap and .format registers for all windows,
which have to be set though RK3066_SYS_CTRL1 register.
Also remove .scl from win1: Scaling is only supported on the primary
plane.
Signed-off-by: NAlex Bee <knaerzche@gmail.com>
Signed-off-by: NHeiko Stuebner <heiko@sntech.de>
Link: https://patchwork.freedesktop.org/patch/msgid/20210528130554.72191-4-knaerzche@gmail.com

Add dither_up, dsp_lut_en and data_blank registers to enable their
respective functionality for RK3188's VOP.
While at that also fix .dsp_blank register which is (only) set with
BIT24 (same as RK3066)
Signed-off-by: NAlex Bee <knaerzche@gmail.com>
Signed-off-by: NHeiko Stuebner <heiko@sntech.de>
Link: https://patchwork.freedesktop.org/patch/msgid/20210528130554.72191-3-knaerzche@gmail.com

Add the registers needed to make scaling work on RK3036's win1.
Signed-off-by: NAlex Bee <knaerzche@gmail.com>
Signed-off-by: NHeiko Stuebner <heiko@sntech.de>
Link: https://patchwork.freedesktop.org/patch/msgid/20210528130554.72191-2-knaerzche@gmail.com

There are drivers that register framebuffer devices very early in the boot
process and make use of the existing framebuffer as setup by the firmware.

If one of those drivers has registered a fbdev, then the fallback fbdev of
the DRM driver won't be bound to the framebuffer console. To avoid that,
remove any existing generic driver and take over the graphics device.

By doing that, the fb mapped to the console is switched correctly from the
early fbdev to the one registered by the rockchip DRM driver:

    [   40.752420] fb0: switching to rockchip-drm-fb from EFI VGA
Signed-off-by: NJavier Martinez Canillas <javierm@redhat.com>
Signed-off-by: NThomas Zimmermann <tzimmermann@suse.de>
Tested-by: NPeter Robinson <pbrobinson@gmail.com>
Acked-by: NThomas Zimmermann <tzimmermann@suse.de>
Acked-by: NHeiko Stuebner <heiko@sntech.de>
Link: https://patchwork.freedesktop.org/patch/msgid/20210516074833.451643-1-javierm@redhat.com

The previous commit wrongly assumed that dma_mmap_wc() could be replaced
by pgprot_writecombine() + dma_mmap_pages(). It did work on my setup,
but did not work everywhere.

Use dma_mmap_wc() when the buffer has the write-combine cache attribute,
and dma_mmap_pages() when it has the non-coherent cache attribute.
Signed-off-by: NPaul Cercueil <paul@crapouillou.net>
Reported-by: NTomi Valkeinen <tomi.valkeinen@ideasonboard.com>
Reviewed-by: NTomi Valkeinen <tomi.valkeinen@ideasonboard.com>
Tested-by: NTomi Valkeinen <tomi.valkeinen@ideasonboard.com>
Fixes: cf8ccbc7 ("drm: Add support for GEM buffers backed by non-coherent memory")
Link: https://patchwork.freedesktop.org/patch/msgid/20210527230334.151947-1-paul@crapouillou.net

[why]
Link rate in kHz is what is eventually required to calculate the link
bandwidth, which makes kHz a more generic unit. This should also make
forward-compatibility with new DP standards easier.

[how]
- Replace 'link rate DPCD code' with 'link rate in kHz' when used with
drm_dp_mst_topology_mgr_init()
- Add/remove related DPCD code conversion from/to kHz where applicable
Signed-off-by: NNikola Cornij <nikola.cornij@amd.com>
Acked-by: NJani Nikula <jani.nikula@intel.com>
Reviewed-by: NLyude Paul <lyude@redhat.com>
Signed-off-by: NLyude Paul <lyude@redhat.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20210512210011.8425-2-nikola.cornij@amd.com

i915 is broken without -rc3, let's bring that tag in to fix it.
Signed-off-by: NMaxime Ripard <maxime@cerno.tech>

The ChromeOS EC ANX7688 bridge is connected to a ChromeOS Embedded
Controller, and is accessed using I2C tunneling through the Embedded
Controller.  Hence add a dependency on I2C_CROS_EC_TUNNEL, to prevent
asking the user about this driver when configuring a kernel without
support for the ChromeOS EC tunnel I2C bus.

Fixes: 44602b10 ("drm/bridge: Add ChromeOS EC ANX7688 bridge driver support")
Signed-off-by: NGeert Uytterhoeven <geert+renesas@glider.be>
Reviewed-by: NEnric Balletbo i Serra <enric.balletbo@collabora.com>
Signed-off-by: NRobert Foss <robert.foss@linaro.org>
Link: https://patchwork.freedesktop.org/patch/msgid/d107d1840b83607baee8571cc5d88973fc32b519.1622015323.git.geert+renesas@glider.be

./drivers/gpu/drm/bridge/lontium-lt8912b.c:758:3-8: No need to set .owner here. The core will do it.

 Remove .owner field if calls are used which set it automatically

Generated by: scripts/coccinelle/api/platform_no_drv_owner.cocci
Reported-by: NHulk Robot <hulkci@huawei.com>
Signed-off-by: NZou Wei <zou_wei@huawei.com>
Reviewed-by: NRobert Foss <robert.foss@linaro.org>
Signed-off-by: NRobert Foss <robert.foss@linaro.org>
Link: https://patchwork.freedesktop.org/patch/msgid/1622109018-54648-1-git-send-email-zou_wei@huawei.com

Eliminate the follow smatch warning:

drivers/gpu/drm/bridge/cadence/cdns-mhdp8546-core.c:2143
cdns_mhdp_bridge_atomic_reset() warn: inconsistent indenting.
Reported-by: NAbaci Robot <abaci@linux.alibaba.com>
Signed-off-by: NJiapeng Chong <jiapeng.chong@linux.alibaba.com>
Reviewed-by: NRobert Foss <robert.foss@linaro.org>
Signed-off-by: NRobert Foss <robert.foss@linaro.org>
Link: https://patchwork.freedesktop.org/patch/msgid/1621852953-51325-1-git-send-email-jiapeng.chong@linux.alibaba.com

pm_runtime_get_sync will increment pm usage counter even it failed.
Forgetting to putting operation will result in reference leak here.
Fix it by replacing it with pm_runtime_resume_and_get to keep usage
counter balanced.
Reported-by: NHulk Robot <hulkci@huawei.com>
Signed-off-by: NZou Wei <zou_wei@huawei.com>
Reviewed-by: NRobert Foss <robert.foss@linaro.org>
Signed-off-by: NRobert Foss <robert.foss@linaro.org>
Link: https://patchwork.freedesktop.org/patch/msgid/1621840862-106024-1-git-send-email-zou_wei@huawei.com

drm/i915 is extremely on fire without the below revert from -rc3:

commit 293837b9
Author: Linus Torvalds <torvalds@linux-foundation.org>
Date:   Wed May 19 05:55:57 2021 -1000

    Revert "i915: fix remap_io_sg to verify the pgprot"

Backmerge so we don't have a too wide bisect window for anything
that's a more involved workload than booting the driver.
Signed-off-by: NDaniel Vetter <daniel.vetter@ffwll.ch>

The structure is unused. Remove it.
Signed-off-by: NThomas Zimmermann <tzimmermann@suse.de>
Reviewed-by: NSakari Ailus <sakari.ailus@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20210516121315.30321-5-tzimmermann@suse.de

The %p4cc printk format modifier was recently added to print fourcc codes,
replacing drm_get_format_name(). The function is no longer needed, so
remove it.
Signed-off-by: NSakari Ailus <sakari.ailus@linux.intel.com>
Signed-off-by: NThomas Zimmermann <tzimmermann@suse.de>
Reviewed-by: NPetr Mladek <pmladek@suse.com>
Reviewed-by: NAndy Shevchenko <andriy.shevchenko@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20210516121315.30321-4-tzimmermann@suse.de

Replace use of struct drm_format_name_buf with %p4cc for printing
4CC formats.
Signed-off-by: NThomas Zimmermann <tzimmermann@suse.de>
Reviewed-by: NSakari Ailus <sakari.ailus@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20210516121315.30321-3-tzimmermann@suse.de

Replace use of struct drm_format_name_buf with %p4cc for printing
4CC formats.
Signed-off-by: NThomas Zimmermann <tzimmermann@suse.de>
Reviewed-by: NSakari Ailus <sakari.ailus@linux.intel.com>
Acked-by: NChristian König <christian.koenig@amd.com>
Acked-by: NAlex Deucher <alexander.deucher@amd.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20210516121315.30321-2-tzimmermann@suse.de

This is one of the last drivers with a global init_module() function
instead of the modern module_init() annotation. Convert it over.
Signed-off-by: NArnd Bergmann <arnd@arndb.de>
Signed-off-by: NThomas Zimmermann <tzimmermann@suse.de>
Link: https://patchwork.freedesktop.org/patch/msgid/20210514213316.635070-1-arnd@kernel.org

Framebuffer devices that are registered by DRM drivers for fbdev emulation
have a "drmfb" suffix in their name. But makes them to be quite confusing
for drivers that already have "drm" in their name:

$ cat /proc/fb
0 rockchipdrmdrmfb

$ cat /proc/fb
0 simpledrmdrmfb

Also, there isn't a lot of value in adding these "drmfb" suffices to their
names, since users shouldn't really care if the FB devices were registered
by a real fbdev driver or a DRM driver using the fbdev emulation.

What programs should be interested about is if there's a DRM device, and
there are better ways to query that info than reading this procfs entry.

So let's just remove the suffix, which leads to much better device names:

$ cat /proc/fb
0 rockchipdrm

$ cat /proc/fb
0 simpledrm
Suggested-by: NThomas Zimmermann <tzimmermann@suse.de>
Signed-off-by: NJavier Martinez Canillas <javierm@redhat.com>
Signed-off-by: NThomas Zimmermann <tzimmermann@suse.de>
Link: https://patchwork.freedesktop.org/patch/msgid/20210525151313.3379622-1-javierm@redhat.com

s/Hisilicon/HiSilicon/.
It should use capital S, according to
https://www.hisilicon.com/en.
Signed-off-by: NHao Fang <fanghao11@huawei.com>
Acked-by: NTian Tao <tiantao6@hisilicon.com>
Signed-off-by: NThomas Zimmermann <tzimmermann@suse.de>
Link: https://patchwork.freedesktop.org/patch/msgid/1621678529-14389-1-git-send-email-fanghao11@huawei.com

After removing drm_pci_alloc/free, some instances where drm_pci_free()
would have kfreed the dma handle were skipped.

Ensure these handles are freed properly.
Signed-off-by: NJoseph Kogut <joseph.kogut@gmail.com>
Signed-off-by: NThomas Zimmermann <tzimmermann@suse.de>
Link: https://patchwork.freedesktop.org/patch/msgid/20210518212859.4148903-1-joseph.kogut@gmail.com

The option DRM_KMS_FB_HELPER has been removed. Also remove the last
remaining select statement for it.
Signed-off-by: NThomas Zimmermann <tzimmermann@suse.de>
Fixes: 91185d55 ("drm: Remove DRM_KMS_FB_HELPER Kconfig option")
Acked-by: NMaxime Ripard <maxime@cerno.tech>
Reported-by: NGeert Uytterhoeven <geert@linux-m68k.org>
Cc: Daniel Vetter <daniel.vetter@ffwll.ch>
Cc: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Cc: Maxime Ripard <mripard@kernel.org>
Cc: David Airlie <airlied@linux.ie>
Cc: Daniel Vetter <daniel@ffwll.ch>
Cc: dri-devel@lists.freedesktop.org
Link: https://patchwork.freedesktop.org/patch/msgid/20210526100825.29450-1-tzimmermann@suse.de

The function ttm_bo_mmap is unused. Remove it and it's helpers; including
the verify_access callback in struct ttm_device_funcs.
Signed-off-by: NThomas Zimmermann <tzimmermann@suse.de>
Reviewed-by: NChristian König <christian.koenig@amd.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20210525151055.8174-8-tzimmermann@suse.de

Vmwgfx is the only user of the TTM's verify_access callback. Inline
the call and avoid the indirection through the function pointer.
Signed-off-by: NThomas Zimmermann <tzimmermann@suse.de>
Reviewed-by: NZack Rusin <zackr@vmware.com>
Acked-by: NChristian König <christian.koenig@amd.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20210525151055.8174-7-tzimmermann@suse.de

The vmwgfx driver is the only remaining user of ttm_bo_mmap(). Inline
the code. The internal helper ttm_bo_vm_lookup() is now also part of
vmwgfx as vmw_bo_vm_lookup().

v2:
	* replace pr_err() with drm_err() (Zack)
Signed-off-by: NThomas Zimmermann <tzimmermann@suse.de>
Reviewed-by: NZack Rusin <zackr@vmware.com>
Acked-by: NChristian König <christian.koenig@amd.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20210525151055.8174-6-tzimmermann@suse.de

Moving the driver-specific mmap code into a GEM object function allows
for using DRM helpers for various mmap callbacks.

The GEM object function is provided by GEM TTM helpers. Nouveau's
implementation of verify_access is unused and has been removed. Access
permissions are validated by the DRM helpers.

As a side effect, nouveau_ttm_vm_ops and nouveau_ttm_fault() are now
implemented in nouveau's GEM code.
Signed-off-by: NThomas Zimmermann <tzimmermann@suse.de>
Reviewed-by: NChristian König <christian.koenig@amd.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20210525151055.8174-5-tzimmermann@suse.de