platform/kernel/linux-starfive.git
21 months agodrm: exynos: dsi: Drop explicit call to bridge detach
Jagan Teki [Wed, 8 Mar 2023 16:39:38 +0000 (22:09 +0530)]
drm: exynos: dsi: Drop explicit call to bridge detach

Exynos DSI already converted into a bridge driver, so bridge
detach will suppose happened during bridge chain removal done
by the bridge core.

Drop the explicit call chain to detach the bridge.

Tested-by: Marek Szyprowski <m.szyprowski@samsung.com>
Reviewed-by: Marek Vasut <marex@denx.de>
Signed-off-by: Jagan Teki <jagan@amarulasolutions.com>
Signed-off-by: Inki Dae <inki.dae@samsung.com>
21 months agoMerge tag 'drm-rcar-next-20230325' of git://git.kernel.org/pub/scm/linux/kernel/git...
Daniel Vetter [Mon, 27 Mar 2023 07:31:20 +0000 (09:31 +0200)]
Merge tag 'drm-rcar-next-20230325' of git://git./linux/kernel/git/pinchartl/linux into drm-next

Miscellaneous fixes and improvements for rcar-du

Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
From: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20230325204922.GD19335@pendragon.ideasonboard.com
21 months agodrm: rcar-du: Fix a NULL vs IS_ERR() bug
Dan Carpenter [Mon, 27 Feb 2023 10:06:59 +0000 (13:06 +0300)]
drm: rcar-du: Fix a NULL vs IS_ERR() bug

The drmm_encoder_alloc() function returns error pointers.  It never
returns NULL.  Fix the check accordingly.

Fixes: 7a1adbd23990 ("drm: rcar-du: Use drmm_encoder_alloc() to manage encoder")
Signed-off-by: Dan Carpenter <error27@gmail.com>
Reviewed-by: Kieran Bingham <kieran.bingham+renesas@ideasonboard.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart+renesas@ideasonboard.com>
Signed-off-by: Laurent Pinchart <laurent.pinchart+renesas@ideasonboard.com>
21 months agodrm: rcar-du: Write correct values in DORCR reserved fields
Laurent Pinchart [Wed, 22 Feb 2023 03:49:39 +0000 (05:49 +0200)]
drm: rcar-du: Write correct values in DORCR reserved fields

The DORCR register controls the routing of clocks and data between DU
channels within a group. For groups that contain a single channel,
there's no routing option to control, and some fields of the register
are then reserved. On Gen2 those reserved fields are documented as
required to be set to 0, while on Gen3 and newer the PG1T, DK1S and PG1D
reserved fields must be set to 1.

The DU driver initializes the DORCR register in rcar_du_group_setup(),
where it ignores the PG1T, DK1S and PG1D, and then configures those
fields to the correct value in rcar_du_group_set_routing(). This hasn't
been shown to cause any issue, but prevents certifying that the driver
complies with the documentation in safety-critical use cases.

As there is no reasonable change that the documentation will be updated
to clarify that those reserved fields can be written to 0 temporarily
before starting the hardware, make sure that the registers are always
set to valid values.

Signed-off-by: Laurent Pinchart <laurent.pinchart+renesas@ideasonboard.com>
Reviewed-by: Tomi Valkeinen <tomi.valkeinen@ideasonboard.com>
21 months agodrm: rcar-du: Rename DORCR fields to make them 0-based
Laurent Pinchart [Wed, 22 Feb 2023 22:08:14 +0000 (00:08 +0200)]
drm: rcar-du: Rename DORCR fields to make them 0-based

The DORCR fields were documented in the R-Car H1 datasheet with 1-based
named, and then got renamed to 0-based in Gen2. The 0-based names are
used for Gen3 and Gen4, making H1 an outlier. Rename the field macros to
make them 0-based, in order to increase readability of the code when
comparing it with the documentation.

Signed-off-by: Laurent Pinchart <laurent.pinchart+renesas@ideasonboard.com>
Reviewed-by: Tomi Valkeinen <tomi.valkeinen@ideasonboard.com>
21 months agodrm: rcar-du: Disable alpha blending for DU planes used with VSP
Laurent Pinchart [Wed, 22 Feb 2023 03:54:03 +0000 (05:54 +0200)]
drm: rcar-du: Disable alpha blending for DU planes used with VSP

When the input to a DU channel comes from a VSP, the DU doesn't perform
any blending operation. Select XRGB8888 instead of ARGB8888 to ensure
that the corresponding registers don't get written with invalid values.

Signed-off-by: Laurent Pinchart <laurent.pinchart+renesas@ideasonboard.com>
Reviewed-by: Tomi Valkeinen <tomi.valkeinen@ideasonboard.com>
21 months agodrm: rcar-du: Don't write unimplemented ESCR and OTAR registers on Gen3
Laurent Pinchart [Wed, 22 Feb 2023 03:49:39 +0000 (05:49 +0200)]
drm: rcar-du: Don't write unimplemented ESCR and OTAR registers on Gen3

The ESCR and OTAR registers are not present in all DU channels on Gen3
SoCs. ESCR only exists in channels that can be routed to an LVDS or
DPAD, and OTAR in channels that can be routed to a DPAD. Skip writing
those registers for other channels. This replaces the DU gen check, as
Gen4 doesn't have LVDS or DPAD outputs.

Signed-off-by: Laurent Pinchart <laurent.pinchart+renesas@ideasonboard.com>
Reviewed-by: Tomi Valkeinen <tomi.valkeinen@ideasonboard.com>
21 months agodrm: rcar-du: lvds: Fix LVDS PLL disable on D3/E3
Laurent Pinchart [Mon, 13 Feb 2023 13:25:15 +0000 (15:25 +0200)]
drm: rcar-du: lvds: Fix LVDS PLL disable on D3/E3

On R-Car D3 and E3, the LVDS encoder provides the dot (pixel) clock to
the DU, regardless of whether the LVDS output is used or not. When using
the DPAD (RGB) output, the DU driver thus enables and disables the LVDS
PLL manually, while when using the LVDS output, it lets the LVDS bridge
driver handle the PLL configuration internally as part of the atomic
enable and disable operations.

This causes an issue when using the LVDS output. As bridges are disabled
before CRTCs, the current implementation violates the enable/disable
sequences documented in the hardware datasheet, which requires the dot
clock to be enabled before the CRTC is started and disabled after it
gets stopped.

Fix the problem by enabling/disabling the LVDS PLL manually from the DU
regardless of which output is used, and skipping the PLL handling in the
LVDS bridge atomic enable and disable operations.

This is however not enough. Disabling the LVDS encoder while leaving the
PLL on still results in a vertical blanking wait timeout when disabling
the DU. Investigation showed that the culprit is the LVEN bit. For an
unclear reason, clearing the bit when disabling the LVDS encoder blocks
vertical blanking interrupts. We thus have to delay disabling the whole
LVDS encoder, not just disabling the PLL, until the DU is disabled.

We could split the LVDS disable sequence by clearing the LVRES bit in
the LVDS bridge atomic disable handler, and delaying the rest of the
operations, in order to disable the LVDS output at bridge atomic disable
time, before stopping the CRTC. This would make the code more complex,
without a clear benefit, so keep the implementation simple(r).

Signed-off-by: Laurent Pinchart <laurent.pinchart+renesas@ideasonboard.com>
Reviewed-by: Tomi Valkeinen <tomi.valkeinen+renesas@ideasonboard.com>
21 months agodrm: rcar-du: lvds: Move LVDS enable code to separate code section
Laurent Pinchart [Mon, 13 Feb 2023 23:19:17 +0000 (01:19 +0200)]
drm: rcar-du: lvds: Move LVDS enable code to separate code section

To prepare for a rework of the LVDS disable code, which will need to be
called from rcar_lvds_pclk_disable(), move the LVDS enable code,
currently stored in the __rcar_lvds_atomic_enable() function, to a
separate code section separate from bridge operations. It will be then
extended with the LVDS disable code.

As part of this rework the __rcar_lvds_atomic_enable() function is
renamed to rcar_lvds_enable() to more clearly indicate its purpose.

No functional change intended.

Signed-off-by: Laurent Pinchart <laurent.pinchart+renesas@ideasonboard.com>
Reviewed-by: Geert Uytterhoeven <geert+renesas@glider.be>
Reviewed-by: Tomi Valkeinen <tomi.valkeinen+renesas@ideasonboard.com>
21 months agodrm: rcar-du: lvds: Call function directly instead of through pointer
Laurent Pinchart [Mon, 13 Feb 2023 23:07:18 +0000 (01:07 +0200)]
drm: rcar-du: lvds: Call function directly instead of through pointer

When disabling the companion bridge in rcar_lvds_atomic_disable(),
there's no need to go through the bridge's operations to call
.atomic_disable(). Call rcar_lvds_atomic_disable() on the companion
directly.

Signed-off-by: Laurent Pinchart <laurent.pinchart+renesas@ideasonboard.com>
Reviewed-by: Tomi Valkeinen <tomi.valkeinen+renesas@ideasonboard.com>
21 months agoMerge tag 'drm-intel-next-2023-03-23' of git://anongit.freedesktop.org/drm/drm-intel...
Daniel Vetter [Fri, 24 Mar 2023 19:22:02 +0000 (20:22 +0100)]
Merge tag 'drm-intel-next-2023-03-23' of git://anongit.freedesktop.org/drm/drm-intel into drm-next

Core Changes:
- drm: Add SDP Error Detection Configuration Register (Arun)

Driver Changes:
- Meteor Lake enabling and fixes (RK, Jose, Madhumitha)
- Lock the fbdev obj before vma pin (Tejas)
- DSC fixes (Stanislav)
- Fixes and clean-up on opregion code (Imre)
- More wm/vblank stuff (Ville)
- More general display code organization (Jani)
- DP Fixes (Stanislav, Ville)
- Introduce flags to ignore long HPD and link training issues \
  for handling spurious issues on CI (Vinod)
- Plane cleanups and extra registers (Ville)
- Update audio keepalive clock values (Clint)
- Rename find_section to bdb_find_section (Maarten)
- DP SDP CRC16 for 128b132b link layer (Arun)
- Fix various issues with noarm register writes (Ville)
- Fix a few TypeC / MST issues (Imre)
- Create GSC submission targeting HDCP and PXP usages on MTL+ (Suraj)
- Enable HDCP2.x via GSC CS (Suraj)

Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
From: Rodrigo Vivi <rodrigo.vivi@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/ZBy56qc9C00tCLOY@intel.com
21 months agoMerge tag 'drm-misc-next-2023-03-23' of git://anongit.freedesktop.org/drm/drm-misc...
Daniel Vetter [Fri, 24 Mar 2023 18:35:37 +0000 (19:35 +0100)]
Merge tag 'drm-misc-next-2023-03-23' of git://anongit.freedesktop.org/drm/drm-misc into drm-next

drm-misc-next for v6.4-rc1:

Core Changes:
- Add unit test for xrgb8888 to mono.
- Assorted small fixes to format helper selftests.
- Assorted documentation updates.
- Drop drm_dev_set_unique.
- Always use shadow buffer in generic fbdev emulation helpers, and
  improve error handling.

Driver Changes:
- Assorted small fixes to malidp, hdlcd, gma500, lima, bridge, rockchip.
- Move fbdev in gma500 to use drm_client.
- Convert bridge platform callbacks to void return.
- Drop leftover from vgem to shmem helper conversion.

Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
From: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/5a4c438e-7047-c044-fc77-5a3597000264@linux.intel.com
21 months agodrm/i915/mtl: Add HDCP GSC interface
Suraj Kandpal [Thu, 16 Mar 2023 09:29:27 +0000 (14:59 +0530)]
drm/i915/mtl: Add HDCP GSC interface

MTL uses GSC command streamer i.e gsc cs to send HDCP/PXP commands
to GSC f/w. It requires to keep hdcp display driver
agnostic to content protection f/w (ME/GSC fw) in the form of
i915_hdcp_fw_ops generic ops.

Adding HDCP GSC CS interface by leveraging the i915_hdcp_fw_ops generic
ops instead of I915_HDCP_COMPONENT as integral part of i915.

Adding checks to see if GSC is loaded and proxy is setup

--v6
-dont change the license date in same patch series [Jani]
-fix the license year {Jani]

--v8
-remove stale comment [Ankit]
-get headers in alphabetical order [Ankit]
-fix hdcp2_supported check [Ankit]

--v9
-remove return statement from hdcp_gsc_fini [Ankit]

Cc: Tomas Winkler <tomas.winkler@intel.com>
Cc: Rodrigo Vivi <rodrigo.vivi@intel.com>
Cc: Uma Shankar <uma.shankar@intel.com>
Cc: Ankit Nautiyal <ankit.k.nautiyal@intel.com>
Signed-off-by: Anshuman Gupta <anshuman.gupta@intel.com>
Signed-off-by: Suraj Kandpal <suraj.kandpal@intel.com>
Reviewed-by: Ankit Nautiyal <ankit.k.nautiyal@intel.com>
Reviewed-by: Uma Shankar <uma.shankar@intel.com>
Signed-off-by: Uma Shankar <uma.shankar@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20230316092927.668980-7-suraj.kandpal@intel.com
21 months agodrm/i915/mtl: Add function to send command to GSC CS
Suraj Kandpal [Thu, 16 Mar 2023 09:29:26 +0000 (14:59 +0530)]
drm/i915/mtl: Add function to send command to GSC CS

Add function that takes care of sending command to gsc cs. We start
of with allocation of memory for our command intel_hdcp_gsc_message that
contains gsc cs memory header as directed in specs followed by the
actual payload hdcp message that we want to send.
Spec states that we need to poll pending bit of response header around
20 times each try being 50ms apart hence adding that to current
gsc_msg_send function
Also we use the same function to take care of both sending and receiving
hence no separate function to get the response.

--v4
-Create common function to fill in gsc_mtl_header [Alan]
-define host session bitmask [Alan]

--v5
-use i915 directly instead of gt->i915 [Alan]
-No need to make fields NULL as we are already
using kzalloc [Alan]

--v8
-change mechanism to reuse the same memory for one hdcp session[Alan]
-fix header ordering
-add comments to explain flags and host session mask [Alan]

--v9
-remove gem obj from hdcp message as we can use
i915_vma_unpin_and_release [Alan]
-move hdcp message allocation and deallocation from hdcp2_enable and
hdcp2_disable to init and teardown of HDCP [Alan]

--v10
-remove unnecessary i915_vma_unpin [Alan]

--v11
-fix comment style [Uma]

Cc: Ankit Nautiyal <ankit.k.nautiyal@intel.com>
Cc: Daniele Ceraolo Spurio <daniele.ceraolospurio@intel.com>
Cc: Alan Pervin Teres <alan.previn.teres.alexis@intel.com>
Cc: Uma Shankar <uma.shankar@intel.com>
Cc: Anshuman Gupta <anshuman.gupta@intel.com>
Signed-off-by: Suraj Kandpal <suraj.kandpal@intel.com>
Reviewed-by: Alan Previn <alan.previn.teres.alexis@intel.com>
Reviewed-by: Uma Shankar <uma.shankar@intel.com>
Signed-off-by: Uma Shankar <uma.shankar@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20230316092927.668980-6-suraj.kandpal@intel.com
21 months agodrm/i915/hdcp: Refactor HDCP API structures
Suraj Kandpal [Thu, 16 Mar 2023 09:29:25 +0000 (14:59 +0530)]
drm/i915/hdcp: Refactor HDCP API structures

It requires to move intel specific HDCP API structures to
i915_hdcp_interface.h from driver/misc/mei/hdcp/mei_hdcp.h
so that any content protection fw interfaces can use these
structures.

Cc: Tomas Winkler <tomas.winkler@intel.com>
Cc: Rodrigo Vivi <rodrigo.vivi@intel.com>
Cc: Uma Shankar <uma.shankar@intel.com>
Cc: Ankit Nautiyal <ankit.k.nautiyal@intel.com>
Signed-off-by: Anshuman Gupta <anshuman.gupta@intel.com>
Signed-off-by: Suraj Kandpal <suraj.kandpal@intel.com>
Reviewed-by: Ankit Nautiyal <ankit.k.nautiyal@intel.com>
Reviewed-by: Uma Shankar <uma.shankar@intel.com>
Signed-off-by: Uma Shankar <uma.shankar@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20230316092927.668980-5-suraj.kandpal@intel.com
21 months agodrm/i915/hdcp: HDCP2.x Refactoring to agnostic hdcp
Suraj Kandpal [Thu, 16 Mar 2023 09:29:24 +0000 (14:59 +0530)]
drm/i915/hdcp: HDCP2.x Refactoring to agnostic hdcp

There are more than 1 type of content protection security firmware.
Make the name generic
%s/_mei_/_

--v3
-Changing names to drop cp_fw to make naming more agnostic[Jani]

--v4
-remove header reference in intel_display_core.h [Uma]
-fix commit message and prefix drm [Uma]

Cc: Tomas Winkler <tomas.winkler@intel.com>
Cc: Rodrigo Vivi <rodrigo.vivi@intel.com>
Cc: Uma Shankar <uma.shankar@intel.com>
Cc: Ankit Nautiyal <ankit.k.nautiyal@intel.com>
Signed-off-by: Anshuman Gupta <anshuman.gupta@intel.com>
Signed-off-by: Suraj Kandpal <suraj.kandpal@intel.com>
Reviewed-by: Ankit Nautiyal <ankit.k.nautiyal@intel.com>
Reviewed-by: Uma Shankar <uma.shankar@intel.com>
Signed-off-by: Uma Shankar <uma.shankar@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20230316092927.668980-4-suraj.kandpal@intel.com
21 months agodrm/i915/hdcp: Use generic names for HDCP helpers and structs
Anshuman Gupta [Thu, 16 Mar 2023 09:29:23 +0000 (14:59 +0530)]
drm/i915/hdcp: Use generic names for HDCP helpers and structs

pre MTL we interact with mei interface to talk to
firmware and enable CP but going forward we will talk to gsc cs
because of which we are making all names for HDCP helpers and
structures generic as either mei or gsc cs maybe used.

Change the include/drm/i915_mei_hdcp_interface.h to
include/drm/i915_hdcp_interface.h

Change the i915_hdcp_interface.h header naming convention to
suit generic f/w type.
%s/MEI_/HDCP_
%s/mei_dev/hdcp_dev

Change structure name Accordingly.
%s/i915_hdcp_comp_master/i915_hdcp_master
%s/i915_hdcp_component_ops/i915_hdcp_ops

--v6
-make each patch build individually [Jani]

--v8
-change ME FW to ME/GSC FW [Ankit]
-fix formatting issue [Ankit]

--v9
-fix commit message and header [Uma]

--v10
-rename comp variable [Uma]

Cc: Tomas Winkler <tomas.winkler@intel.com>
Cc: Rodrigo Vivi <rodrigo.vivi@intel.com>
Cc: Uma Shankar <uma.shankar@intel.com>
Cc: Ankit Nautiyal <ankit.k.nautiyal@intel.com>
Signed-off-by: Anshuman Gupta <anshuman.gupta@intel.com>
Signed-off-by: Suraj Kandpal <suraj.kandpal@intel.com>
Reviewed-by: Ankit Nautiyal <ankit.k.nautiyal@intel.com>
Reviewed-by: Uma Shankar <uma.shankar@intel.com>
Acked-by: Tomas Winkler <tomas.winkler@intel.com>
Signed-off-by: Uma Shankar <uma.shankar@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20230316092927.668980-3-suraj.kandpal@intel.com
21 months agodrm/i915/gsc: Create GSC request submission mechanism
Suraj Kandpal [Thu, 16 Mar 2023 09:29:22 +0000 (14:59 +0530)]
drm/i915/gsc: Create GSC request submission mechanism

HDCP and PXP will require a common function to allow it to
submit commands to the gsc cs. Also adding the gsc mtl header
that needs to be added on to the existing payloads of HDCP
and PXP.

--v4
-Seprate gsc load and heci cmd submission into different
functions in different files for better scalability [Alan]
-Rename gsc address field [Alan]

--v5
-remove extra line is intel_gsc_fw.h [Uma]

Cc: Daniele Ceraolo Spurio <daniele.ceraolospurio@intel.com>
Cc: Alan Previn <alan.previn.teres.alexis@intel.com>
Signed-off-by: Suraj Kandpal<suraj.kandpal@intel.com>
Reviewed-by: Alan Previn <alan.previn.teres.alexis@intel.com>
Reviewed-by: Uma Shankar <uma.shankar@intel.com>
Signed-off-by: Uma Shankar <uma.shankar@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20230316092927.668980-2-suraj.kandpal@intel.com
21 months agodrm/rockchip: vop2: Add error check to devm_regmap_init_mmio
Alfredo Cruz [Wed, 22 Mar 2023 22:44:11 +0000 (23:44 +0100)]
drm/rockchip: vop2: Add error check to devm_regmap_init_mmio

devm_regmap_init_mmio() may return an invalid pointer in case of an error.
This patch adds the corresponding IS_ERR check to vop2->map.

Signed-off-by: Alfredo Cruz <alfredo.carlon@gmail.com>
Signed-off-by: Heiko Stuebner <heiko@sntech.de>
Link: https://patchwork.freedesktop.org/patch/msgid/20230322224411.15612-1-alfredo.carlon@gmail.com
21 months agodrm/rockchip: dw_hdmi: cleanup drm encoder during unbind
Toby Chen [Fri, 17 Mar 2023 00:51:26 +0000 (17:51 -0700)]
drm/rockchip: dw_hdmi: cleanup drm encoder during unbind

This fixes a use-after-free crash during rmmod.

The DRM encoder is embedded inside the larger rockchip_hdmi,
which is allocated with the component. The component memory
gets freed before the main drm device is destroyed. Fix it
by running encoder cleanup before tearing down its container.

Signed-off-by: Toby Chen <tobyc@nvidia.com>
[moved encoder cleanup above clk_disable, similar to bind-error-path]
Signed-off-by: Heiko Stuebner <heiko@sntech.de>
Link: https://patchwork.freedesktop.org/patch/msgid/20230317005126.496-1-tobyc@nvidia.com
21 months agodrm/rockchip: vop2: fix uninitialized variable possible_crtcs
Tom Rix [Thu, 16 Mar 2023 13:23:02 +0000 (09:23 -0400)]
drm/rockchip: vop2: fix uninitialized variable possible_crtcs

clang reportes this error
drivers/gpu/drm/rockchip/rockchip_drm_vop2.c:2322:8: error:
  variable 'possible_crtcs' is used uninitialized whenever 'if'
  condition is false [-Werror,-Wsometimes-uninitialized]
                        if (vp) {
                            ^~
drivers/gpu/drm/rockchip/rockchip_drm_vop2.c:2336:36: note:
  uninitialized use occurs here
                ret = vop2_plane_init(vop2, win, possible_crtcs);
                                                 ^~~~~~~~~~~~~~
drivers/gpu/drm/rockchip/rockchip_drm_vop2.c:2322:4:
  note: remove the 'if' if its condition is always true
                        if (vp) {
                        ^~~~~~~~

The else-statement changes the win->type to OVERLAY without setting the
possible_crtcs variable.  Rework the block, initialize possible_crtcs to
0 to remove the else-statement.  Split the else-if-statement out to its
own if-statement so the OVERLAY check will catch when the win-type has
been changed.

Fixes: 368419a2d429 ("drm/rockchip: vop2: initialize possible_crtcs properly")
Signed-off-by: Tom Rix <trix@redhat.com>
Acked-by: Michael Riesch <michael.riesch@wolfvision.net>
Signed-off-by: Heiko Stuebner <heiko@sntech.de>
Link: https://patchwork.freedesktop.org/patch/msgid/20230316132302.531724-1-trix@redhat.com
21 months agodrm/i915/tc: Check the PLL type used by an enabled TC port
Imre Deak [Tue, 21 Mar 2023 22:01:01 +0000 (00:01 +0200)]
drm/i915/tc: Check the PLL type used by an enabled TC port

The current way to determine during HW state sanitization if a PHY is
connected in the expected way doesn't work in all cases. The check for
this considers only the PHY ready/owned state and the initial TC mode
which was determined earlier by the TC port HW readout - using the
sink's HPD and the same PHY ready/owned states.

For instance for an enabled DP-alt/TBT port without the PHY ready/owned
flags set the initial mode will be TBT, and this will be regarded as a
valid PHY state. However it's possible that the port is actually enabled
in DP-alt mode, but for some reason the PHY ownership was not acquired.

Make sure the driver can detect invalid PHY states as in the above
example by checking the PHY ready/owned state wrt. the PLL type used.
This should be the TBT PLL if the PHY is not owned and the MG (non-TBT)
PLL if the PHY is owned.

v2: Rebased on change passing crtc_state in the previous patch.

Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Imre Deak <imre.deak@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20230321220101.983366-3-imre.deak@intel.com
21 months agodrm/i915/tc: Factor out a function querying active links on a TC port
Imre Deak [Tue, 21 Mar 2023 22:01:00 +0000 (00:01 +0200)]
drm/i915/tc: Factor out a function querying active links on a TC port

For clarity factor out the function to determine if there are active
links on a TC port. This prepares for the next patch also checking the
port's PLL type.

While at it pass crtc_state to intel_tc_port_sanitize_mode(), and check
hw.active in that, instead of the deprecated crtc->active flag.

v2: Check crtc_state->hw.active instead of crtc->active. (Ville)

Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Imre Deak <imre.deak@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20230321220101.983366-2-imre.deak@intel.com
21 months agodrm/i915: Add encoder hook to get the PLL type used by TC ports
Imre Deak [Thu, 16 Mar 2023 13:17:22 +0000 (15:17 +0200)]
drm/i915: Add encoder hook to get the PLL type used by TC ports

Add an encoder hook, which can be called on enabled TC ports to
determine if the port uses a TBT or a non-TBT PLL. An upcoming patch
will use this to sanity check active TC port's PHY state wrt. the PLL
type used by the port.

Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Imre Deak <imre.deak@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20230316131724.359612-13-imre.deak@intel.com
21 months agodrm/i915/tc: Assume a TC port is legacy if VBT says the port has HDMI
Imre Deak [Tue, 21 Mar 2023 22:00:59 +0000 (00:00 +0200)]
drm/i915/tc: Assume a TC port is legacy if VBT says the port has HDMI

Since an HDMI output can only be enabled in legacy mode on TC ports,
assume that VBT is wrong and the port is legacy if VBT says the port is
non-legacy and has HDMI. If VBT says to enable DP as well leave the
non-legacy flag enabled, relying on the flag getting fixed up based on
the HPD status during sink detection.

v2: Fix the legacy port flag only if DP is not enabled.

Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Imre Deak <imre.deak@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20230321220101.983366-1-imre.deak@intel.com
21 months agodrm/i915/tc: Make the TC mode readout consistent in all PHY states
Imre Deak [Thu, 16 Mar 2023 13:17:20 +0000 (15:17 +0200)]
drm/i915/tc: Make the TC mode readout consistent in all PHY states

For consistency detect the initial TC mode in the PHY owned state the
same way this is done in the not owned state (w/o changing the
behavior). While at it, add more details to the PHY state debug print.

Signed-off-by: Imre Deak <imre.deak@intel.com>
Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20230316131724.359612-11-imre.deak@intel.com
21 months agodrm/i915/tc: Fix initial TC mode on disabled legacy ports
Imre Deak [Thu, 16 Mar 2023 13:17:19 +0000 (15:17 +0200)]
drm/i915/tc: Fix initial TC mode on disabled legacy ports

Atm, a TC port's initial mode will be read out as TBT mode in any case
the PHY ownership is not held. This isn't correct for legacy ports which
should be used only in legacy mode.

Fix the above initial mode to be disconnected mode for a legacy port and
TBT mode for DP-alt/TBT ports. Determine the port type by checking first
the HPD state and then the legacy VBT flag (so the HPD state can correct
a bogus VBT flag). If a sink is connected on a disabled port the PHY
will get also connected (switching it to legacy mode on a legacy port).

Also connect the PHY on a legacy port if it's enabled but BIOS
incorrectly left it in the disconnected state for some reason.

Signed-off-by: Imre Deak <imre.deak@intel.com>
Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20230316131724.359612-10-imre.deak@intel.com
21 months agodrm/i915/tc: Fix TC mode for a legacy port if the PHY is not ready
Imre Deak [Thu, 16 Mar 2023 13:17:18 +0000 (15:17 +0200)]
drm/i915/tc: Fix TC mode for a legacy port if the PHY is not ready

A legacy TC port can't be switched to TBT mode, even if the PHY
initialization wasn't ready yet for some reason, so prevent this.

This shouldn't normally happen as the driver waits for the IOM/TCSS
PHY initialization during driver loading and system resume.

Signed-off-by: Imre Deak <imre.deak@intel.com>
Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20230316131724.359612-9-imre.deak@intel.com
21 months agodrm/i915/tc: Fix target TC mode for a disconnected legacy port
Imre Deak [Thu, 16 Mar 2023 13:17:17 +0000 (15:17 +0200)]
drm/i915/tc: Fix target TC mode for a disconnected legacy port

Atm, the target TC mode - which the PHY should be switched to at any
point it's used - is TBT in case there is no sink connected. However
legacy ports are only used in the legacy mode regardless of the sink
connected state. Fix the mode returned by
intel_tc_port_get_target_mode() accordingly.

Despite of the above issue, the PHY got disconnected as expected in
response to a sink disconnect event, causing only a redundant
PHY disconnect->reconnect sequence whenever the port was used.

Signed-off-by: Imre Deak <imre.deak@intel.com>
Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20230316131724.359612-8-imre.deak@intel.com
21 months agodrm/i915/tc: Factor out helpers converting HPD mask to TC mode
Imre Deak [Thu, 16 Mar 2023 13:17:16 +0000 (15:17 +0200)]
drm/i915/tc: Factor out helpers converting HPD mask to TC mode

Factor out helpers used later in the patchset to convert an HPD
status mask to TC mode or target TC mode.

No functional changes.

Signed-off-by: Imre Deak <imre.deak@intel.com>
Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20230316131724.359612-7-imre.deak@intel.com
21 months agodrm/i915/tc: Wait for IOM/FW PHY initialization of legacy TC ports
Imre Deak [Thu, 16 Mar 2023 13:17:15 +0000 (15:17 +0200)]
drm/i915/tc: Wait for IOM/FW PHY initialization of legacy TC ports

During boot-up/system resume, the TC PHY on legacy ports will be
initialized by the IOM/TCSS firmware regardless of a sink being
connected or not (as opposed to DP-alt/TBT ports, which the FW only
inits once a sink is connected).

Wait for the above initialization to complete during HW readout, so that
connecting the PHY later will already see the expected PHY ready state.

Signed-off-by: Imre Deak <imre.deak@intel.com>
Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20230316131724.359612-6-imre.deak@intel.com
21 months agodrm/i915/tc: Fix system resume MST mode restore for DP-alt sinks
Imre Deak [Thu, 16 Mar 2023 13:17:14 +0000 (15:17 +0200)]
drm/i915/tc: Fix system resume MST mode restore for DP-alt sinks

At least restoring the MST topology during system resume needs to use
AUX before the display HW readout->sanitization sequence is complete,
but on TC ports the PHY may be in the wrong mode for this, resulting in
the AUX transfers to fail.

The initial TC port mode is kept fixed as BIOS left it for the above HW
readout sequence (to prevent changing the mode on an enabled port).  If
the port is disabled this initial mode is TBT - as in any case the PHY
ownership is not held - even if a DP-alt sink is connected. Thus, the
AUX transfers during this time will use TBT mode instead of the expected
DP-alt mode and so time out.

Fix the above by connecting the PHY during port initialization if the
port is disabled, which will switch to the expected mode (DP-alt in the
above case).

As the encoder/pipe HW state isn't read-out yet at this point, check if
the port is enabled based on the DDI_BUF enabled flag. Save the read-out
initial mode, so intel_tc_port_sanitize_mode() can check this wrt. the
read-out encoder HW state.

Signed-off-by: Imre Deak <imre.deak@intel.com>
Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20230316131724.359612-5-imre.deak@intel.com
21 months agodrm/i915/tc: Fix the ICL PHY ownership check in TC-cold state
Imre Deak [Thu, 16 Mar 2023 13:17:13 +0000 (15:17 +0200)]
drm/i915/tc: Fix the ICL PHY ownership check in TC-cold state

The commit renaming icl_tc_phy_is_in_safe_mode() to
icl_tc_phy_take_ownership() didn't flip the function's return value
accordingly, fix this up.

This didn't cause an actual problem besides state check errors, since
the function is only used during HW readout.

Cc: José Roberto de Souza <jose.souza@intel.com>
Fixes: f53979d68a77 ("drm/i915/display/tc: Rename safe_mode functions ownership")
Reviewed-by: José Roberto de Souza <jose.souza@intel.com>
Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Imre Deak <imre.deak@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20230316131724.359612-4-imre.deak@intel.com
21 months agodrm/i915/tc: Fix TC port link ref init for DP MST during HW readout
Imre Deak [Thu, 16 Mar 2023 13:17:12 +0000 (15:17 +0200)]
drm/i915/tc: Fix TC port link ref init for DP MST during HW readout

An enabled TC MST port holds one TC port link reference, regardless of
the number of enabled streams on it, but the TC port HW readout takes
one reference for each active MST stream.

Fix the HW readout, taking only one reference for MST ports.

This didn't cause an actual problem, since the encoder HW readout doesn't
yet support reading out the MST HW state.

Signed-off-by: Imre Deak <imre.deak@intel.com>
Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Reviewed-by: Mika Kahola <mika.kahola@intel.com>
Reviewed-by: Andrzej Hajda <andrzej.hajda@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20230316131724.359612-3-imre.deak@intel.com
21 months agodrm/i915/tc: Abort DP AUX transfer on a disconnected TC port
Imre Deak [Thu, 16 Mar 2023 13:17:11 +0000 (15:17 +0200)]
drm/i915/tc: Abort DP AUX transfer on a disconnected TC port

On TC ports the 4ms AUX timeout combined with the 5 * 32 retry
attempts during DPCD accesses adds a 640ms delay to each access if the
sink is disconnected. This in turn slows down a modeset during which the
sink is disconnected (for instance a disabling modeset).

Prevent the above delay by aborting AUX transfers on a TC port with a
disconnected sink.

The DP 1.4a link CTS (4.2.1.5 Source Device Inactive HPD / Inactive AUX
Test") also requires not to initiate AUX transfers on disconnected DP
ports in general, however this patch doesn't change the behavior on
non-TC ports, leaving that for a follow-up.

Reported-and-tested-by: Chris Chiu <chris.chiu@canonical.com>
Closes: https://gitlab.freedesktop.org/drm/intel/-/issues/8279
Reviewed-by: Mika Kahola <mika.kahola@intel.com>
Signed-off-by: Imre Deak <imre.deak@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20230316131724.359612-2-imre.deak@intel.com
21 months agodrm/fbdev-generic: Rename symbols
Thomas Zimmermann [Mon, 20 Mar 2023 15:07:51 +0000 (16:07 +0100)]
drm/fbdev-generic: Rename symbols

Rename symbols to match the style of other fbdev-emulation source
code. No functional changes.

Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Reviewed-by: Javier Martinez Canillas <javierm@redhat.com>
Acked-by: Zack Rusin <zackr@vmware.com>
Tested-by: Sui Jingfeng <suijingfeng@loongson.cn>
Link: https://patchwork.freedesktop.org/patch/msgid/20230320150751.20399-9-tzimmermann@suse.de
21 months agodrm/fb-helper: Consolidate CONFIG_DRM_FBDEV_LEAK_PHYS_SMEM
Thomas Zimmermann [Mon, 20 Mar 2023 15:07:50 +0000 (16:07 +0100)]
drm/fb-helper: Consolidate CONFIG_DRM_FBDEV_LEAK_PHYS_SMEM

Consolidate all handling of CONFIG_DRM_FBDEV_LEAK_PHYS_SMEM by
making the module parameter optional in drm_fb_helper.c.

Without the config option, modules can set smem_start in struct
fb_info for internal usage, but not export if to userspace. The
address can only be exported by enabling the option and setting
the module parameter. Also update the comment.

Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Reviewed-by: Javier Martinez Canillas <javierm@redhat.com>
Acked-by: Zack Rusin <zackr@vmware.com>
Tested-by: Sui Jingfeng<suijingfeng@loongson.cn>
Link: https://patchwork.freedesktop.org/patch/msgid/20230320150751.20399-8-tzimmermann@suse.de
21 months agodrm/fbdev-generic: Clean up after failed probing
Thomas Zimmermann [Mon, 20 Mar 2023 15:07:49 +0000 (16:07 +0100)]
drm/fbdev-generic: Clean up after failed probing

Clean up fbdev and client state if the probe function fails. It
used to leak allocated resources. Also reorder the individual steps
to simplify cleanup.

v2:
* move screen_size update into separate patches

Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Reviewed-by: Javier Martinez Canillas <javierm@redhat.com>
Acked-by: Zack Rusin <zackr@vmware.com>
Tested-by: Sui Jingfeng <suijingfeng@loongson.cn>
Link: https://patchwork.freedesktop.org/patch/msgid/20230320150751.20399-7-tzimmermann@suse.de
21 months agodrm/fbdev-generic: Set screen size to size of GEM buffer
Thomas Zimmermann [Mon, 20 Mar 2023 15:07:48 +0000 (16:07 +0100)]
drm/fbdev-generic: Set screen size to size of GEM buffer

The size of the screen memory should be equivalent to the size of
the screen's GEM buffer. Don't recalculate the value.

Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Reviewed-by: Javier Martinez Canillas <javierm@redhat.com>
Tested-by: Sui Jingfeng <suijingfeng@loongson.cn>
Link: https://patchwork.freedesktop.org/patch/msgid/20230320150751.20399-6-tzimmermann@suse.de
21 months agodrm/fb-helper: Support smem_len in deferred I/O
Thomas Zimmermann [Mon, 20 Mar 2023 15:07:47 +0000 (16:07 +0100)]
drm/fb-helper: Support smem_len in deferred I/O

The size of the framebuffer can either be stored in screen_info or
smem_len. Take both into account in the deferred I/O code.

Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Reviewed-by: Javier Martinez Canillas <javierm@redhat.com>
Tested-by: Sui Jingfeng <suijingfeng@loongson.cn>
Link: https://patchwork.freedesktop.org/patch/msgid/20230320150751.20399-5-tzimmermann@suse.de
21 months agodrm/fb-helper: Export drm_fb_helper_release_info()
Thomas Zimmermann [Mon, 20 Mar 2023 15:07:46 +0000 (16:07 +0100)]
drm/fb-helper: Export drm_fb_helper_release_info()

Export the fb_info release code as drm_fb_helper_release_info(). Will
help with cleaning up failed fbdev probing.

Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Reviewed-by: Javier Martinez Canillas <javierm@redhat.com>
Acked-by: Zack Rusin <zackr@vmware.com>
Tested-by: Sui Jingfeng<suijingfeng@loongson.cn>
Link: https://patchwork.freedesktop.org/patch/msgid/20230320150751.20399-4-tzimmermann@suse.de
21 months agodrm/fbdev-generic: Remove unused prefer_shadow_fbdev flag
Thomas Zimmermann [Mon, 20 Mar 2023 15:07:45 +0000 (16:07 +0100)]
drm/fbdev-generic: Remove unused prefer_shadow_fbdev flag

Remove the flag prefer_shadow_fbdev from struct drm_mode_config.
Drivers set this flag to enable shadow buffering in the generic
fbdev emulation. Such shadow buffering is now mandatory, so the
flag is unused.

Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Reviewed-by: Javier Martinez Canillas <javierm@redhat.com>
Reviewed-by: Zack Rusin <zackr@vmware.com>
Tested-by: Sui Jingfeng <suijingfeng@loongson.cn>
Link: https://patchwork.freedesktop.org/patch/msgid/20230320150751.20399-3-tzimmermann@suse.de
21 months agodrm/fbdev-generic: Always use shadow buffering
Thomas Zimmermann [Mon, 20 Mar 2023 15:07:44 +0000 (16:07 +0100)]
drm/fbdev-generic: Always use shadow buffering

Remove all codepaths that implement fbdev output directly on GEM
buffers. Always allocate a shadow buffer in system memory and set
up deferred I/O for mmap.

The fbdev code that operated directly on GEM buffers was used by
drivers based on GEM DMA helpers. Those drivers have been migrated
to use fbdev-dma, a dedicated fbdev emulation for DMA memory. All
remaining users of fbdev-generic require shadow buffering.

Memory management of the remaining callers uses TTM, GEM SHMEM
helpers or a variant of GEM DMA helpers that is incompatible with
fbdev-dma. Therefore remove the unused codepaths from fbdev-generic
and simplify the code.

Using a shadow buffer with deferred I/O is probably the best case
for most remaining callers. Some of the TTM-based drivers might
benefit from a dedicated fbdev emulation that operates directly on
the driver's video memory.

Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Reviewed-by: Javier Martinez Canillas <javierm@redhat.com>
Acked-by: Zack Rusin <zackr@vmware.com>
Tested-by: Sui Jingfeng <suijingfeng@loongson.cn>
Link: https://patchwork.freedesktop.org/patch/msgid/20230320150751.20399-2-tzimmermann@suse.de
21 months agodrm: bridge: simple-bridge: Use dev_err_probe()
Ye Xingchen [Wed, 22 Mar 2023 08:22:51 +0000 (16:22 +0800)]
drm: bridge: simple-bridge: Use dev_err_probe()

Replace the open-code with dev_err_probe() to simplify the code.

Signed-off-by: Ye Xingchen <ye.xingchen@zte.com.cn>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: Neil Armstrong <neil.armstrong@linaro.org>
Link: https://patchwork.freedesktop.org/patch/msgid/202303221622511915615@zte.com.cn
21 months agodrm/bridge: display-connector: Use dev_err_probe()
Ye Xingchen [Wed, 22 Mar 2023 08:21:33 +0000 (16:21 +0800)]
drm/bridge: display-connector: Use dev_err_probe()

Replace the open-code with dev_err_probe() to simplify the code.

Signed-off-by: Ye Xingchen <ye.xingchen@zte.com.cn>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: Neil Armstrong <neil.armstrong@linaro.org>
Link: https://patchwork.freedesktop.org/patch/msgid/202303221621336645576@zte.com.cn
21 months agodrm: remove drm_dev_set_unique
Christian König [Fri, 17 Feb 2023 12:54:27 +0000 (13:54 +0100)]
drm: remove drm_dev_set_unique

Not used by any drivers any more, the only use case in drm_dev_init()
can be inlined now.

Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Jeffrey Hugo <quic_jhugo@quicinc.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20230316082035.567520-2-christian.koenig@amd.com
21 months agoMerge tag 'drm-habanalabs-next-2023-03-20' of https://git.kernel.org/pub/scm/linux...
Dave Airlie [Wed, 22 Mar 2023 00:35:45 +0000 (10:35 +1000)]
Merge tag 'drm-habanalabs-next-2023-03-20' of https://git./linux/kernel/git/ogabbay/linux into drm-next

This tag contains habanalabs driver and accel changes for v6.4:

- uAPI changes:

  - Add opcodes to the CS ioctl to allow user to stall/resume specific engines
    inside Gaudi2. This is to allow the user to perform power
    testing/measurements when training different topologies.

  - Expose in the INFO ioctl the amount of device memory that the driver
    and f/w reserve for themselves.

  - Expose in the INFO ioctl a bit-mask of the available rotator engines
    in Gaudi2. This is to align with other engines that are already exposed.

  - Expose in the INFO ioctl the register's address of the f/w that should
    be used to trigger interrupts from within the user's code running in the
    compute engines.

  - Add a critical-event bit in the eventfd bitmask so the user will know the
    event that was received was critical, and a reset will now occur

  - Expose in the INFO ioctl two new opcodes to fetch information on h/w and
    f/w events. The events recorded are the events that were reported in the
    eventfd.

- New features and improvements:

  - Add a dedicated interrupt ID in MSI-X in the device to the notification of
    an unexpected user-related event in Gaudi2. Handle it in the driver by
    reporting this event.

  - Allow the user to fetch the device memory current usage even when the
    device is undergoing compute-reset (a reset type that only clears the
    compute engines).

  - Enable graceful reset mechanism for compute-reset. This will give the
    user a few seconds before the device is reset. For example, the user can,
    during that time, perform certain device operations (dump data for debug)
    or close the device in an orderly fashion.

  - Align the decoder with the rest of the engines in regard to notification
    to the user about interrupts and in regard to performing graceful reset
    when needed (instead of immediate reset).

  - Add support for assert interrupt from the TPC engine.

  - Get the reset type that is necessary to perform per event from the
    auto-generated irq_map array.

  - Print the specific reason why a device is still in use when notifying to
    the user about it (after the user closed the device's FD).

  - Move to threaded IRQ when handling interrupts of workload completions.

- Firmware related fixes:

  - Fix RAZWI event handler to match newest f/w version.

  - Read error cause register in dma core events because the f/w doesn't
    do that.

  - Increase maximum time to wait for completion of Gaudi2 reset due to f/w
    bug.

  - Align to the latest firmware specs.

- Enforce the release order of the compute device and dma-buf.
  i.e increment the device file refcount for any dma-buf that was exported
  for that device. This will make sure the compute device release function
  won't be called until the user closes all the FDs of the relevant
  dma-bufs. Without this change, closing the device's FD before/without
  closing the dma-buf's FD would always lead to hard-reset of the device.

- Fix a link in the drm documentation to correctly point to the accel section.

- Compilation warnings cleanups

- Misc bug fixes and code cleanups

Signed-off-by: Dave Airlie <airlied@redhat.com>
# -----BEGIN PGP SIGNATURE-----
#
# iQEzBAABCgAdFiEE7TEboABC71LctBLFZR1NuKta54AFAmQYfcAACgkQZR1NuKta
54DB4Af/SuiHZkVXwr+yHPv9El726rz9ZQD7mQtzNmehWGonwAvz15yqocNMUSbF
# JbqE/vrZjvbXrP1Uv5UrlRVdnFHSPV18VnHU4BMS/WOm19SsR6vZ0QOXOoa6/AUb
# w+kF3D//DbFI4/mTGfpH5/pzwu51ti8aVktosPFlHIa8iI8CB4/4IV+ivQ8UW4oK
# HyDRkIvHdRmER7vGOfhwhsr4zdqSlJBYrv3C3Z1dkSYBPW/5ICbiM1UlKycwdYKI
# cajQBSdUQwUCWnI+i8RmSy3kjNO6OE4XRUvTv89F2bQeyK/1rJLG2m2xZR/Ml/o5
# 7Cgvbn0hWZyeqe7OObYiBlSOBSehCA==
# =wclm
# -----END PGP SIGNATURE-----
# gpg: Signature made Tue 21 Mar 2023 01:37:36 AEST
# gpg:                using RSA key ED311BA00042EF52DCB412C5651D4DB8AB5AE780
# gpg: Can't check signature: No public key
From: Oded Gabbay <ogabbay@kernel.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20230320154026.GA766126@ogabbay-vm-u20.habana-labs.com
21 months agoMerge tag 'drm-intel-gt-next-2023-03-16' of git://anongit.freedesktop.org/drm/drm...
Dave Airlie [Tue, 21 Mar 2023 20:49:01 +0000 (06:49 +1000)]
Merge tag 'drm-intel-gt-next-2023-03-16' of git://anongit.freedesktop.org/drm/drm-intel into drm-next

Driver Changes:

- Fix issue #6333: "list_add corruption" and full system lockup from
  performance monitoring (Janusz)
- Give the punit time to settle before fatally failing (Aravind, Chris)
- Don't use stolen memory or BAR for ring buffers on LLC platforms (John)
- Add missing ecodes and correct timeline seqno on GuC error captures (John)
- Make sure DSM size has correct 1MiB granularity on Gen12+ (Nirmoy,
  Lucas)
- Fix potential SSEU max_subslices array-index-out-of-bounds access on Gen11 (Andrea)
- Whitelist COMMON_SLICE_CHICKEN3 for UMD access on Gen12+ (Matt R.)
- Apply Wa_1408615072/Wa_1407596294 correctly on Gen11 (Matt R)
- Apply LNCF/LBCF workarounds correctly on XeHP SDV/PVC/DG2 (Matt R)
- Implement Wa_1606376872 for Xe_LP (Gustavo)
- Consider GSI offset when doing MCR lookups on Meteorlake+ (Matt R.)
- Add engine TLB invalidation for Meteorlake (Matt R.)
- Fix GSC Driver-FLR completion on Meteorlake (Alan)
- Fix GSC races on driver load/unload on Meteorlake+ (Daniele)
- Disable MC6 for MTL A step (Badal)

- Consolidate TLB invalidation flow (Tvrtko)
- Improve debug GuC/HuC debug messages (Michal Wa., John)
- Move fd_install after last use of fence (Rob)
- Initialize the obj flags for shmem objects (Aravind)
- Fix missing debug object activation (Nirmoy)
- Probe lmem before the stolen portion (Matt A)
- Improve clean up of GuC busyness stats worker (John)
- Fix missing return code checks in GuC submission init (John)
- Annotate two more workaround/tuning registers as MCR on PVC (Matt R)
- Fix GEN8_MISCCPCTL definition and remove unused INF_UNIT_LEVEL_CLKGATE (Lucas)
- Use sysfs_emit() and sysfs_emit_at() (Nirmoy)
- Make kobj_type structures constant (Thomas W.)
- make kobj attributes const on gt/ (Jani)
- Remove the unused virtualized start hack on buddy allocator (Matt A)
- Remove redundant check for DG1 (Lucas)
- Move DG2 tuning to the right function (Lucas)
- Rename dev_priv to i915 for private data naming consistency in gt/ (Andi)
- Remove unnecessary whitelisting of CS_CTX_TIMESTAMP on Xe_HP platforms (Matt R.)
-

- Escape wildcard in method names in kerneldoc (Bagas)
- Selftest improvements (Chris, Jonathan, Tvrtko, Anshuman, Tejas)
- Fix sparse warnings (Jani)

[airlied: fix unused variable in intel_workarounds]
Signed-off-by: Dave Airlie <airlied@redhat.com>
From: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/ZBMSb42yjjzczRhj@jlahtine-mobl.ger.corp.intel.com
21 months agodrm/lima: Use drm_sched_job_add_syncobj_dependency()
Maíra Canal [Fri, 24 Feb 2023 21:41:32 +0000 (18:41 -0300)]
drm/lima: Use drm_sched_job_add_syncobj_dependency()

As lima_gem_add_deps() performs the same steps as
drm_sched_job_add_syncobj_dependency(), replace the open-coded
implementation in Lima in order to simply use the DRM function.

Signed-off-by: Maíra Canal <mcanal@igalia.com>
Reviewed-by: Qiang Yu <yuq825@gmail.com>
Signed-off-by: Maíra Canal <mairacanal@riseup.net>
Link: https://patchwork.freedesktop.org/patch/msgid/20230224214133.411966-1-mcanal@igalia.com
21 months agoMerge tag 'drm-misc-next-2023-03-16' of git://anongit.freedesktop.org/drm/drm-misc...
Dave Airlie [Tue, 21 Mar 2023 01:03:16 +0000 (11:03 +1000)]
Merge tag 'drm-misc-next-2023-03-16' of git://anongit.freedesktop.org/drm/drm-misc into drm-next

drm-misc-next for v6.4-rc1:

Cross-subsystem Changes:
- Add drm_bridge.h to drm_bridge maintainers.

Core Changes:
- Assorted fixes to TTM, tests, format-helper, accel.
- Assorted Makefile fixes to drivers and accel.
- Implement fbdev emulation for GEM DMA drivers, and convert a lot of
  drivers to use it.
- Use tgid instead of pid for tracking clients.

Driver Changes:
- Assorted fixes in rockchip, vmwgfx, nouveau, cirrus.
- Add imx25 driver.
- Add Elida KD50T048A, Sony TD4353, Novatek NT36523, STARRY 2081101QFH032011-53G panels.
- Add 4K mode support to rockchip.
- Convert cirrus to use regular atomic helpers, and more cirrus
  improvements.
- Add damage clipping to cirrus, virtio.

[airlied: add drm_bridge.h include to imx]
Signed-off-by: Dave Airlie <airlied@redhat.com>
From: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/f7b765c7-d49d-edb5-2a6a-4f7a7be16a59@linux.intel.com
21 months agodrm/i915: Disable DC states for all commits
Ville Syrjälä [Mon, 20 Mar 2023 18:35:32 +0000 (20:35 +0200)]
drm/i915: Disable DC states for all commits

Keeping DC states enabled is incompatible with the _noarm()/_arm()
split we use for writing pipe/plane registers. When DC5 and PSR
are enabled, all pipe/plane registers effectively become self-arming
on account of DC5 exit arming the update, and PSR exit latching it.

What probably saves us most of the time is that (with PIPE_MISC[21]=0)
all pipe register writes themselves trigger PSR exit, and then
we don't re-enter PSR until the idle frame count has elapsed.
So it may be that the PSR exit happens already before we've
updated the state too much.

Also the PSR1 panel (at least on this KBL) seems to discard the first
frame we trasmit, presumably still scanning out from its internal
framebuffer at that point. So only the second frame we transmit is
actually visible. But I suppose that could also be panel specific
behaviour. I haven't checked out how other PSR panels behave, nor
did I bother to check what the eDP spec has to say about this.

And since this really is all about DC states, let's switch from
the MODESET domain to the DC_OFF domain. Functionally they are
100% identical. We should probably remove the MODESET domain...

And for good measure let's toss in an assert to the place where
we do the _noarm() register writes to make sure DC states are
in fact off.

v2: Just use intel_display_power_is_enabled() (Imre)

Cc: <stable@vger.kernel.org> #v5.17+
Cc: Manasi Navare <navaremanasi@google.com>
Cc: Drew Davenport <ddavenport@chromium.org>
Cc: Jouni Högander <jouni.hogander@intel.com>
Reviewed-by: Imre Deak <imre.deak@intel.com>
Fixes: d13dde449580 ("drm/i915: Split pipe+output CSC programming to noarm+arm pair")
Fixes: f8a005eb8972 ("drm/i915: Optimize icl+ universal plane programming")
Fixes: 890b6ec4a522 ("drm/i915: Split skl+ plane update into noarm+arm pair")
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20230320183532.17727-1-ville.syrjala@linux.intel.com
21 months agodrm/i915: Workaround ICL CSC_MODE sticky arming
Ville Syrjälä [Mon, 20 Mar 2023 09:54:36 +0000 (11:54 +0200)]
drm/i915: Workaround ICL CSC_MODE sticky arming

Unlike SKL/GLK the ICL CSC unit suffers from a new issue where
CSC_MODE arming is sticky. That is, once armed it remains armed
causing the CSC coeff/offset registers to become effectively
self-arming.

CSC coeff/offset registers writes no longer disarm the CSC,
but fortunately register read still do. So we can use that
to disarm the CSC unit once the registers for the current
frame have been latched. This avoid s the self-arming behaviour
from persisting into the next frame's .color_commit_noarm()
call.

Cc: <stable@vger.kernel.org> #v5.19+
Cc: Manasi Navare <navaremanasi@google.com>
Cc: Drew Davenport <ddavenport@chromium.org>
Cc: Imre Deak <imre.deak@intel.com>
Cc: Jouni Högander <jouni.hogander@intel.com>
Fixes: d13dde449580 ("drm/i915: Split pipe+output CSC programming to noarm+arm pair")
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20230320095438.17328-5-ville.syrjala@linux.intel.com
Reviewed-by: Imre Deak <imre.deak@intel.com>
21 months agodrm/i915: Add a .color_post_update() hook
Ville Syrjälä [Mon, 20 Mar 2023 09:54:35 +0000 (11:54 +0200)]
drm/i915: Add a .color_post_update() hook

We're going to need stuff after the color management
register latching has happened. Add a corresponding hook.

Cc: <stable@vger.kernel.org> #v5.19+
Cc: Manasi Navare <navaremanasi@google.com>
Cc: Drew Davenport <ddavenport@chromium.org>
Cc: Imre Deak <imre.deak@intel.com>
Cc: Jouni Högander <jouni.hogander@intel.com>
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20230320095438.17328-4-ville.syrjala@linux.intel.com
Reviewed-by: Imre Deak <imre.deak@intel.com>
21 months agodrm/i915: Move CSC load back into .color_commit_arm() when PSR is enabled on skl/glk
Ville Syrjälä [Mon, 20 Mar 2023 09:54:34 +0000 (11:54 +0200)]
drm/i915: Move CSC load back into .color_commit_arm() when PSR is enabled on skl/glk

SKL/GLK CSC unit suffers from a nasty issue where a CSC
coeff/offset register read or write between DC5 exit and
PSR exit will undo the CSC arming performed by DMC, and
then during PSR exit the hardware will latch zeroes into
the active CSC registers. This causes any plane going
through the CSC to output all black.

We can sidestep the issue by making sure the PSR exit has
already actually happened before we touch the CSC coeff/offset
registers. Easiest way to guarantee that is to just move the
CSC programming back into the .color_commir_arm() as we force
a PSR exit (and crucially wait for it to actually happen)
prior to touching the arming registers.

When PSR (and thus also DC states) are disabled we don't
have anything to worry about, so we can keep using the
more optional _noarm() hook for writing the CSC registers.

Cc: <stable@vger.kernel.org> #v5.19+
Cc: Manasi Navare <navaremanasi@google.com>
Cc: Drew Davenport <ddavenport@chromium.org>
Cc: Imre Deak <imre.deak@intel.com>
Cc: Jouni Högander <jouni.hogander@intel.com>
Closes: https://gitlab.freedesktop.org/drm/intel/-/issues/8283
Fixes: d13dde449580 ("drm/i915: Split pipe+output CSC programming to noarm+arm pair")
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20230320095438.17328-3-ville.syrjala@linux.intel.com
Reviewed-by: Imre Deak <imre.deak@intel.com>
21 months agodrm/i915: Split icl_color_commit_noarm() from skl_color_commit_noarm()
Ville Syrjälä [Mon, 20 Mar 2023 09:54:33 +0000 (11:54 +0200)]
drm/i915: Split icl_color_commit_noarm() from skl_color_commit_noarm()

We're going to want different behavior for skl/glk vs. icl
in .color_commit_noarm(), so split the hook into two. Arguably
we already had slightly different behaviour since
csc_enable/gamma_enable are never set on icl+, so the old
code was perhaps a bit confusing as well.

Cc: <stable@vger.kernel.org> #v5.19+
Cc: Manasi Navare <navaremanasi@google.com>
Cc: Drew Davenport <ddavenport@chromium.org>
Cc: Imre Deak <imre.deak@intel.com>
Cc: Jouni Högander <jouni.hogander@intel.com>
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20230320095438.17328-2-ville.syrjala@linux.intel.com
Reviewed-by: Imre Deak <imre.deak@intel.com>
21 months agoi915/display/dp: SDP CRC16 for 128b132b link layer
Arun R Murthy [Thu, 2 Mar 2023 08:15:32 +0000 (13:45 +0530)]
i915/display/dp: SDP CRC16 for 128b132b link layer

Enable SDP error detection configuration, this will set CRC16 in
128b/132b link layer.
For Display version 13 a hardware bit31 in register VIDEO_DIP_CTL is
added to enable/disable SDP CRC applicable for DP2.0 only, but the
default value of this bit will enable CRC16 in 128b/132b hence
skipping this write.
Corrective actions on SDP corruption is yet to be defined.

v2: Moved the CRC enable to link training init(Jani N)
v3: Moved crc enable to ddi pre enable <Jani N>
v4: Separate function for SDP CRC16 (Jani N)

Signed-off-by: Arun R Murthy <arun.r.murthy@intel.com>
Reviewed-by: Jani Nikula <jani.nikula@intel.com>
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20230302081532.765821-3-arun.r.murthy@intel.com
21 months agodrm: Add SDP Error Detection Configuration Register
Arun R Murthy [Thu, 2 Mar 2023 08:15:31 +0000 (13:45 +0530)]
drm: Add SDP Error Detection Configuration Register

DP2.0 E11 defines a new register to facilitate SDP error detection by a
128B/132B capable DPRX device.

v2: Update the macro name to reflect the DP spec(Harry)

Signed-off-by: Arun R Murthy <arun.r.murthy@intel.com>
Reviewed-by: Harry Wentland <harry.wentland@amd.com>
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
Acked-by: Thomas Zimmermann <tzimmermann@suse.de>
Link: https://patchwork.freedesktop.org/patch/msgid/20230302081532.765821-2-arun.r.murthy@intel.com
21 months agodrm/i915/bios: Rename find_section to find_bdb_section
Maarten Lankhorst [Wed, 15 Mar 2023 12:19:24 +0000 (14:19 +0200)]
drm/i915/bios: Rename find_section to find_bdb_section

This prevents a namespace collision on other archs.

Signed-off-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20230315121924.2314693-1-jani.nikula@intel.com
21 months agodrm/i915/debugfs: add crtc i915_pipe debugfs file
Jani Nikula [Mon, 20 Mar 2023 12:44:29 +0000 (14:44 +0200)]
drm/i915/debugfs: add crtc i915_pipe debugfs file

The pipe may differ from crtc index if pipes are fused off. For testing
purposes, IGT needs to know the pipe.

There's already a I915_GET_PIPE_FROM_CRTC_ID IOCTL for this. However,
the upcoming Xe driver won't have that IOCTL, and going forward, we'll
want a unified interface for testing i915 and Xe, as they share the
display code. Thus add the debugfs for i915 display.

v2: User letters for pipe names (Ville)

Cc: Bhanuprakash Modem <bhanuprakash.modem@intel.com>
Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Tested-by: Bhanuprakash Modem <bhanuprakash.modem@intel.com>
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20230320124429.786985-2-jani.nikula@intel.com
21 months agodrm/i915/debugfs: switch crtc debugfs to struct intel_crtc
Jani Nikula [Mon, 20 Mar 2023 12:44:28 +0000 (14:44 +0200)]
drm/i915/debugfs: switch crtc debugfs to struct intel_crtc

Convert the crtc debugfs code to use struct intel_crtc instead of struct
drm_crtc.

v2: Fix build for CONFIG_DRM_I915_DEBUG_VBLANK_EVADE=y (kernel test robot)

Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20230320124429.786985-1-jani.nikula@intel.com
21 months agodrm/gma500: remove unused gma_pipe_event function
Tom Rix [Sun, 19 Mar 2023 14:23:20 +0000 (10:23 -0400)]
drm/gma500: remove unused gma_pipe_event function

clang with W=1 reports
drivers/gpu/drm/gma500/psb_irq.c:35:19: error: unused function
  'gma_pipe_event' [-Werror,-Wunused-function]
static inline u32 gma_pipe_event(int pipe)
                  ^
This function is not used, so remove it.

Signed-off-by: Tom Rix <trix@redhat.com>
Signed-off-by: Patrik Jakobsson <patrik.r.jakobsson@gmail.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20230319142320.1704336-1-trix@redhat.com
21 months agodrm/i915/audio: update audio keepalive clock values
Clint Taylor [Thu, 16 Mar 2023 23:46:54 +0000 (16:46 -0700)]
drm/i915/audio: update audio keepalive clock values

BSPEC has updated the cdclk audio keepalives AUD_TS_CDCLK_M value to 60
for all supported platforms and refclks.

BSPEC: 54034
BSPEC: 55409
BSPEC: 65243
Cc: Kai Vehmanen <kai.vehmanen@linux.intel.com>
Cc: Uma Shankar <uma.shankar@intel.com>
Cc: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Clint Taylor <clinton.a.taylor@intel.com>
Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20230316234654.3797572-1-clinton.a.taylor@intel.com
21 months agoaccel/habanalabs: remove redundant TODOs
Ofir Bitton [Thu, 16 Mar 2023 15:24:45 +0000 (17:24 +0200)]
accel/habanalabs: remove redundant TODOs

As mmu refactor and nic resume are not relevant anymore, remove
their TODO comments.

Signed-off-by: Ofir Bitton <obitton@habana.ai>
Reviewed-by: Oded Gabbay <ogabbay@kernel.org>
Signed-off-by: Oded Gabbay <ogabbay@kernel.org>
21 months agoaccel/habanalabs: change razwi handle after fw fix
Dani Liberman [Thu, 16 Mar 2023 13:03:12 +0000 (15:03 +0200)]
accel/habanalabs: change razwi handle after fw fix

FW had one data route for tpc0 and tpc1 when running in secured mode
and a different one when running without secured mode. After fw fixed
this issue, both mode have the same data path.

Signed-off-by: Dani Liberman <dliberman@habana.ai>
Reviewed-by: Oded Gabbay <ogabbay@kernel.org>
Signed-off-by: Oded Gabbay <ogabbay@kernel.org>
21 months agoaccel/habanalabs: add handling for unexpected user event
Ofir Bitton [Wed, 8 Mar 2023 11:34:52 +0000 (13:34 +0200)]
accel/habanalabs: add handling for unexpected user event

In order for the user to be aware of unexpected events in Gaudi2 that
aren't assigned to a specific engine, we are adding the handling of
this dedicated interrupt.

Signed-off-by: Ofir Bitton <obitton@habana.ai>
Reviewed-by: Oded Gabbay <ogabbay@kernel.org>
Signed-off-by: Oded Gabbay <ogabbay@kernel.org>
21 months agoaccel/habanalabs: fix a missing-braces compilation warning
Tomer Tayar [Wed, 15 Mar 2023 11:13:54 +0000 (13:13 +0200)]
accel/habanalabs: fix a missing-braces compilation warning

Replace initialization of "struct cpucp_packet" from "{0} to "{}" to
avoid a "missing braces around initializer" compilation warning.

Signed-off-by: Tomer Tayar <ttayar@habana.ai>
Reviewed-by: Oded Gabbay <ogabbay@kernel.org>
Signed-off-by: Oded Gabbay <ogabbay@kernel.org>
21 months agoaccel/habanalabs: fix a maybe-uninitialized compilation warnings
Tomer Tayar [Wed, 15 Mar 2023 09:46:18 +0000 (11:46 +0200)]
accel/habanalabs: fix a maybe-uninitialized compilation warnings

Initialize 'index' in gaudi2_handle_qman_err() and 'offset' in
gaudi2_get_nic_idle_status() to avoid "maybe-uninitialized" compilation
warnings.

Signed-off-by: Tomer Tayar <ttayar@habana.ai>
Reviewed-by: Oded Gabbay <ogabbay@kernel.org>
Signed-off-by: Oded Gabbay <ogabbay@kernel.org>
21 months agoaccel/habanalabs: fix page fault event clear
Dani Liberman [Wed, 15 Mar 2023 08:00:58 +0000 (10:00 +0200)]
accel/habanalabs: fix page fault event clear

After getting page fault in gaudi2, we need to clear the valid bit
instead of the address.

Signed-off-by: Dani Liberman <dliberman@habana.ai>
Reviewed-by: Oded Gabbay <ogabbay@kernel.org>
Signed-off-by: Oded Gabbay <ogabbay@kernel.org>
21 months agoaccel/habanalabs: expose rotator mask to userspace
Ofir Bitton [Mon, 13 Mar 2023 20:30:23 +0000 (22:30 +0200)]
accel/habanalabs: expose rotator mask to userspace

All engine masks are exposed to user, make sure user gets the
correct rotator enabled mask in gaudi2.

Signed-off-by: Ofir Bitton <obitton@habana.ai>
Reviewed-by: Oded Gabbay <ogabbay@kernel.org>
Signed-off-by: Oded Gabbay <ogabbay@kernel.org>
21 months agoaccel/habanalabs: regenerate gaudi2 ids_map_extended
Ohad Sharabi [Mon, 13 Mar 2023 08:56:59 +0000 (10:56 +0200)]
accel/habanalabs: regenerate gaudi2 ids_map_extended

Some names of events has been modified/added.

Signed-off-by: Ohad Sharabi <osharabi@habana.ai>
Reviewed-by: Oded Gabbay <ogabbay@kernel.org>
Signed-off-by: Oded Gabbay <ogabbay@kernel.org>
21 months agoaccel: Link to compute accelerator subsystem intro
Bagas Sanjaya [Tue, 7 Mar 2023 04:35:26 +0000 (11:35 +0700)]
accel: Link to compute accelerator subsystem intro

Commit 2c204f3d53218d ("accel: add dedicated minor for accelerator
devices") adds link to accelerator nodes section of DRM internals doc
(Documentation/gpu/drm-internals.rst), but the target doesn't exist.
Instead, there is only an introduction doc for computer accelerator
subsytem.

Link to that doc until there is documentation of accelerator internals.

Fixes: 2c204f3d53218d ("accel: add dedicated minor for accelerator devices")
Signed-off-by: Bagas Sanjaya <bagasdotme@gmail.com>
Reviewed-by: Jeffrey Hugo <quic_jhugo@quicinc.com>
Signed-off-by: Oded Gabbay <ogabbay@kernel.org>
21 months agoaccel/habanalabs: expose dram reserved size by kmd
Ofir Bitton [Mon, 13 Mar 2023 15:10:28 +0000 (17:10 +0200)]
accel/habanalabs: expose dram reserved size by kmd

We expose this in order for user applications to know how much dram
is reserved for internal use.

Signed-off-by: Ofir Bitton <obitton@habana.ai>
Reviewed-by: Oded Gabbay <ogabbay@kernel.org>
Signed-off-by: Oded Gabbay <ogabbay@kernel.org>
21 months agoaccel/habanalabs: remove '\n' when passing strings to gaudi2_print_event()
Tomer Tayar [Sun, 12 Mar 2023 15:15:03 +0000 (17:15 +0200)]
accel/habanalabs: remove '\n' when passing strings to gaudi2_print_event()

Remove all '\n' from strings which are passed as arguments to
gaudi2_print_event(), because the newline character is added internally
in this function.

Signed-off-by: Tomer Tayar <ttayar@habana.ai>
Reviewed-by: Oded Gabbay <ogabbay@kernel.org>
Signed-off-by: Oded Gabbay <ogabbay@kernel.org>
21 months agoaccel/habanalabs: return tlb inv error code upon failure
Koby Elbaz [Tue, 7 Mar 2023 08:13:44 +0000 (10:13 +0200)]
accel/habanalabs: return tlb inv error code upon failure

Now that CQ-completion based jobs do not trigger a reset upon failure,
failure of such jobs (e.g., MMU cache invalidation) should be handled
by the caller itself depending on the error code returned to it.

Signed-off-by: Koby Elbaz <kelbaz@habana.ai>
Reviewed-by: Oded Gabbay <ogabbay@kernel.org>
Signed-off-by: Oded Gabbay <ogabbay@kernel.org>
21 months agoaccel/habanalabs: fix field names in hl_info_hw_ip_info
Oded Gabbay [Thu, 16 Mar 2023 10:13:34 +0000 (12:13 +0200)]
accel/habanalabs: fix field names in hl_info_hw_ip_info

Don't use padX for actual reservedX fields.

Signed-off-by: Oded Gabbay <ogabbay@kernel.org>
Reviewed-by: Ofir Bitton <obitton@habana.ai>
21 months agoaccel/habanalabs: in {e/p}dma_core events read the err cause reg
Dafna Hirschfeld [Thu, 9 Mar 2023 09:27:26 +0000 (11:27 +0200)]
accel/habanalabs: in {e/p}dma_core events read the err cause reg

Since the err_cause register is unprivileged, we should read it from
the driver instead of using the param that came from the FW.

Signed-off-by: Dafna Hirschfeld <dhirschfeld@habana.ai>
Reviewed-by: Oded Gabbay <ogabbay@kernel.org>
Signed-off-by: Oded Gabbay <ogabbay@kernel.org>
21 months agoaccel/habanalabs: fix use of var reset_sleep_ms
Dafna Hirschfeld [Wed, 8 Mar 2023 11:22:17 +0000 (13:22 +0200)]
accel/habanalabs: fix use of var reset_sleep_ms

- remove reset_sleep_ms arg from functions that don't use it.
- move the call msleep(reset_sleep_ms) from btm poll to gaudi2_hw_fini
as it is called from there already for other flow.

Signed-off-by: Dafna Hirschfeld <dhirschfeld@habana.ai>
Reviewed-by: Oded Gabbay <ogabbay@kernel.org>
Signed-off-by: Oded Gabbay <ogabbay@kernel.org>
21 months agoaccel/habanalabs: in hw_fini return error code if polling timed-out
Dafna Hirschfeld [Mon, 20 Feb 2023 08:09:25 +0000 (10:09 +0200)]
accel/habanalabs: in hw_fini return error code if polling timed-out

In hw_fini callback, we use either the cpucp packet method or polling a
register. Currently we return error only in the case of cpucp packet
failure. In this patch we also return error if polling timed out.

Signed-off-by: Dafna Hirschfeld <dhirschfeld@habana.ai>
Reviewed-by: Oded Gabbay <ogabbay@kernel.org>
Signed-off-by: Oded Gabbay <ogabbay@kernel.org>
21 months agoaccel/habanalabs: increase reset poll timeout
Ofir Bitton [Wed, 8 Mar 2023 13:04:45 +0000 (15:04 +0200)]
accel/habanalabs: increase reset poll timeout

Due to a firmware bug we need to increase reset poll timeout
or else we will timeout in secured environments.

Signed-off-by: Ofir Bitton <obitton@habana.ai>
Reviewed-by: Oded Gabbay <ogabbay@kernel.org>
Signed-off-by: Oded Gabbay <ogabbay@kernel.org>
21 months agoaccel/habanalabs: do not verify engine modes after being changed
Koby Elbaz [Mon, 6 Mar 2023 14:43:41 +0000 (16:43 +0200)]
accel/habanalabs: do not verify engine modes after being changed

Engines idle state can't always be verified between changes of
engine modes (e.g., stall/halt).
For example, if a CS is inflight when altering engine's mode,
idle state will return NOT idle, always.

Signed-off-by: Koby Elbaz <kelbaz@habana.ai>
Reviewed-by: Oded Gabbay <ogabbay@kernel.org>
Signed-off-by: Oded Gabbay <ogabbay@kernel.org>
21 months agoaccel/habanalabs: align to latest firmware specs
Oded Gabbay [Mon, 6 Mar 2023 10:34:10 +0000 (12:34 +0200)]
accel/habanalabs: align to latest firmware specs

Copy the most up-to-date interface files to the firmware.

Signed-off-by: Oded Gabbay <ogabbay@kernel.org>
Reviewed-by: Ofir Bitton <obitton@habana.ai>
21 months agoaccel/habanalabs: Fix spelling mistake "maped" -> "mapped"
Colin Ian King [Tue, 14 Mar 2023 08:18:12 +0000 (08:18 +0000)]
accel/habanalabs: Fix spelling mistake "maped" -> "mapped"

There is a spelling mistake in a dev_err message. Fix it.

Signed-off-by: Colin Ian King <colin.i.king@gmail.com>
Reviewed-by: Oded Gabbay <ogabbay@kernel.org>
Signed-off-by: Oded Gabbay <ogabbay@kernel.org>
21 months agoaccel/habanalabs: make gaudi2_is_device_idle() static
Oded Gabbay [Wed, 8 Mar 2023 12:57:32 +0000 (14:57 +0200)]
accel/habanalabs: make gaudi2_is_device_idle() static

This function is only called inside gaudi2.c file.

Reported-by: kernel test robot <lkp@intel.com>
Link: https://lore.kernel.org/oe-kbuild-all/202303071320.X5ouBlNY-lkp@intel.com/
Signed-off-by: Oded Gabbay <ogabbay@kernel.org>
Reviewed-by: Ofir Bitton <obitton@habana.ai>
21 months agodrm/vgem: Drop struct drm_vgem_gem_object
Maíra Canal [Wed, 22 Feb 2023 16:06:17 +0000 (13:06 -0300)]
drm/vgem: Drop struct drm_vgem_gem_object

Commit 45d9c8dde4cd ("drm/vgem: use shmem helpers") introduced shmem
helpers to vgem and with that, removed all uses of the struct
drm_vgem_gem_object. So, as the struct is no longer used, delete it.

Signed-off-by: Maíra Canal <mcanal@igalia.com>
Reviewed-by: Melissa Wen <mwen@igalia.com>
Signed-off-by: Maíra Canal <mairacanal@riseup.net>
Link: https://patchwork.freedesktop.org/patch/msgid/20230222160617.171429-1-mcanal@igalia.com
21 months agodrm/prime: Fix documentation of drm_gem_prime_fd_to_handle()
Petr Tesarik [Fri, 24 Feb 2023 12:09:31 +0000 (13:09 +0100)]
drm/prime: Fix documentation of drm_gem_prime_fd_to_handle()

The DRM device passed to drm_gem_prime_fd_to_handle() is where the
dma-buf is being imported, not the device where it was exported.

Also fix a trivial typo in drm_gem_prime_import_dev().

Signed-off-by: Petr Tesarik <petr.tesarik.ext@huawei.com>
Reviewed-by: Simon Ser <contact@emersion.fr>
Signed-off-by: Simon Ser <contact@emersion.fr>
Link: https://patchwork.freedesktop.org/patch/msgid/20230224120931.1024-1-petrtesarik@huaweicloud.com
21 months agodrm/bridge: ti-tfp410: Convert to platform remove callback returning void
Uwe Kleine-König [Sat, 18 Mar 2023 19:08:04 +0000 (20:08 +0100)]
drm/bridge: ti-tfp410: Convert to platform remove callback returning void

The .remove() callback for a platform driver returns an int which makes
many driver authors wrongly assume it's possible to do error handling by
returning an error code. However the value returned is (mostly) ignored
and this typically results in resource leaks. To improve here there is a
quest to make the remove callback return void. In the first step of this
quest all drivers are converted to .remove_new() which already returns
void.

Trivially convert this driver from always returning zero in the remove
callback to the void returning variant.

Signed-off-by: Uwe Kleine-König <u.kleine-koenig@pengutronix.de>
Reviewed-by: Laurent Pinchart <laurent.pinchart+renesas@ideasonboard.com>
Signed-off-by: Neil Armstrong <neil.armstrong@linaro.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20230318190804.234610-20-u.kleine-koenig@pengutronix.de
21 months agodrm/bridge: thc63lvd1024: Convert to platform remove callback returning void
Uwe Kleine-König [Sat, 18 Mar 2023 19:08:03 +0000 (20:08 +0100)]
drm/bridge: thc63lvd1024: Convert to platform remove callback returning void

The .remove() callback for a platform driver returns an int which makes
many driver authors wrongly assume it's possible to do error handling by
returning an error code. However the value returned is (mostly) ignored
and this typically results in resource leaks. To improve here there is a
quest to make the remove callback return void. In the first step of this
quest all drivers are converted to .remove_new() which already returns
void.

Trivially convert this driver from always returning zero in the remove
callback to the void returning variant.

Signed-off-by: Uwe Kleine-König <u.kleine-koenig@pengutronix.de>
Reviewed-by: Laurent Pinchart <laurent.pinchart+renesas@ideasonboard.com>
Signed-off-by: Neil Armstrong <neil.armstrong@linaro.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20230318190804.234610-19-u.kleine-koenig@pengutronix.de
21 months agodrm/bridge: dw-hdmi-i2s-audio: Convert to platform remove callback returning void
Uwe Kleine-König [Sat, 18 Mar 2023 19:08:02 +0000 (20:08 +0100)]
drm/bridge: dw-hdmi-i2s-audio: Convert to platform remove callback returning void

The .remove() callback for a platform driver returns an int which makes
many driver authors wrongly assume it's possible to do error handling by
returning an error code. However the value returned is (mostly) ignored
and this typically results in resource leaks. To improve here there is a
quest to make the remove callback return void. In the first step of this
quest all drivers are converted to .remove_new() which already returns
void.

Trivially convert this driver from always returning zero in the remove
callback to the void returning variant.

Signed-off-by: Uwe Kleine-König <u.kleine-koenig@pengutronix.de>
Reviewed-by: Laurent Pinchart <laurent.pinchart+renesas@ideasonboard.com>
Signed-off-by: Neil Armstrong <neil.armstrong@linaro.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20230318190804.234610-18-u.kleine-koenig@pengutronix.de
21 months agodrm/bridge: dw-hdmi-gp-audio: Convert to platform remove callback returning void
Uwe Kleine-König [Sat, 18 Mar 2023 19:08:01 +0000 (20:08 +0100)]
drm/bridge: dw-hdmi-gp-audio: Convert to platform remove callback returning void

The .remove() callback for a platform driver returns an int which makes
many driver authors wrongly assume it's possible to do error handling by
returning an error code. However the value returned is (mostly) ignored
and this typically results in resource leaks. To improve here there is a
quest to make the remove callback return void. In the first step of this
quest all drivers are converted to .remove_new() which already returns
void.

Trivially convert this driver from always returning zero in the remove
callback to the void returning variant.

Signed-off-by: Uwe Kleine-König <u.kleine-koenig@pengutronix.de>
Reviewed-by: Laurent Pinchart <laurent.pinchart+renesas@ideasonboard.com>
Signed-off-by: Neil Armstrong <neil.armstrong@linaro.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20230318190804.234610-17-u.kleine-koenig@pengutronix.de
21 months agodrm/bridge: dw-hdmi-cec: Convert to platform remove callback returning void
Uwe Kleine-König [Sat, 18 Mar 2023 19:08:00 +0000 (20:08 +0100)]
drm/bridge: dw-hdmi-cec: Convert to platform remove callback returning void

The .remove() callback for a platform driver returns an int which makes
many driver authors wrongly assume it's possible to do error handling by
returning an error code. However the value returned is (mostly) ignored
and this typically results in resource leaks. To improve here there is a
quest to make the remove callback return void. In the first step of this
quest all drivers are converted to .remove_new() which already returns
void.

Trivially convert this driver from always returning zero in the remove
callback to the void returning variant.

Signed-off-by: Uwe Kleine-König <u.kleine-koenig@pengutronix.de>
Reviewed-by: Laurent Pinchart <laurent.pinchart+renesas@ideasonboard.com>
Signed-off-by: Neil Armstrong <neil.armstrong@linaro.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20230318190804.234610-16-u.kleine-koenig@pengutronix.de
21 months agodrm/bridge: dw-hdmi-ahb-audio: Convert to platform remove callback returning void
Uwe Kleine-König [Sat, 18 Mar 2023 19:07:59 +0000 (20:07 +0100)]
drm/bridge: dw-hdmi-ahb-audio: Convert to platform remove callback returning void

The .remove() callback for a platform driver returns an int which makes
many driver authors wrongly assume it's possible to do error handling by
returning an error code. However the value returned is (mostly) ignored
and this typically results in resource leaks. To improve here there is a
quest to make the remove callback return void. In the first step of this
quest all drivers are converted to .remove_new() which already returns
void.

Trivially convert this driver from always returning zero in the remove
callback to the void returning variant.

Signed-off-by: Uwe Kleine-König <u.kleine-koenig@pengutronix.de>
Reviewed-by: Laurent Pinchart <laurent.pinchart+renesas@ideasonboard.com>
Signed-off-by: Neil Armstrong <neil.armstrong@linaro.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20230318190804.234610-15-u.kleine-koenig@pengutronix.de
21 months agodrm/bridge: simple-bridg: Convert to platform remove callback returning void
Uwe Kleine-König [Sat, 18 Mar 2023 19:07:58 +0000 (20:07 +0100)]
drm/bridge: simple-bridg: Convert to platform remove callback returning void

The .remove() callback for a platform driver returns an int which makes
many driver authors wrongly assume it's possible to do error handling by
returning an error code. However the value returned is (mostly) ignored
and this typically results in resource leaks. To improve here there is a
quest to make the remove callback return void. In the first step of this
quest all drivers are converted to .remove_new() which already returns
void.

Trivially convert this driver from always returning zero in the remove
callback to the void returning variant.

Signed-off-by: Uwe Kleine-König <u.kleine-koenig@pengutronix.de>
Reviewed-by: Laurent Pinchart <laurent.pinchart+renesas@ideasonboard.com>
Signed-off-by: Neil Armstrong <neil.armstrong@linaro.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20230318190804.234610-14-u.kleine-koenig@pengutronix.de
21 months agodrm/bridge: nwl-dsi: Convert to platform remove callback returning void
Uwe Kleine-König [Sat, 18 Mar 2023 19:07:57 +0000 (20:07 +0100)]
drm/bridge: nwl-dsi: Convert to platform remove callback returning void

The .remove() callback for a platform driver returns an int which makes
many driver authors wrongly assume it's possible to do error handling by
returning an error code. However the value returned is (mostly) ignored
and this typically results in resource leaks. To improve here there is a
quest to make the remove callback return void. In the first step of this
quest all drivers are converted to .remove_new() which already returns
void.

Trivially convert this driver from always returning zero in the remove
callback to the void returning variant.

Signed-off-by: Uwe Kleine-König <u.kleine-koenig@pengutronix.de>
Reviewed-by: Laurent Pinchart <laurent.pinchart+renesas@ideasonboard.com>
Signed-off-by: Neil Armstrong <neil.armstrong@linaro.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20230318190804.234610-13-u.kleine-koenig@pengutronix.de
21 months agodrm/bridge: lvds-codec: Convert to platform remove callback returning void
Uwe Kleine-König [Sat, 18 Mar 2023 19:07:56 +0000 (20:07 +0100)]
drm/bridge: lvds-codec: Convert to platform remove callback returning void

The .remove() callback for a platform driver returns an int which makes
many driver authors wrongly assume it's possible to do error handling by
returning an error code. However the value returned is (mostly) ignored
and this typically results in resource leaks. To improve here there is a
quest to make the remove callback return void. In the first step of this
quest all drivers are converted to .remove_new() which already returns
void.

Trivially convert this driver from always returning zero in the remove
callback to the void returning variant.

Signed-off-by: Uwe Kleine-König <u.kleine-koenig@pengutronix.de>
Reviewed-by: Laurent Pinchart <laurent.pinchart+renesas@ideasonboard.com>
Signed-off-by: Neil Armstrong <neil.armstrong@linaro.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20230318190804.234610-12-u.kleine-koenig@pengutronix.de
21 months agodrm/bridge: imx8qxp-pxl2dpi: Convert to platform remove callback returning void
Uwe Kleine-König [Sat, 18 Mar 2023 19:07:55 +0000 (20:07 +0100)]
drm/bridge: imx8qxp-pxl2dpi: Convert to platform remove callback returning void

The .remove() callback for a platform driver returns an int which makes
many driver authors wrongly assume it's possible to do error handling by
returning an error code. However the value returned is (mostly) ignored
and this typically results in resource leaks. To improve here there is a
quest to make the remove callback return void. In the first step of this
quest all drivers are converted to .remove_new() which already returns
void.

Trivially convert this driver from always returning zero in the remove
callback to the void returning variant.

Signed-off-by: Uwe Kleine-König <u.kleine-koenig@pengutronix.de>
Reviewed-by: Laurent Pinchart <laurent.pinchart+renesas@ideasonboard.com>
Signed-off-by: Neil Armstrong <neil.armstrong@linaro.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20230318190804.234610-11-u.kleine-koenig@pengutronix.de
21 months agodrm/bridge: imx8qxp-pixel-link: Convert to platform remove callback returning void
Uwe Kleine-König [Sat, 18 Mar 2023 19:07:54 +0000 (20:07 +0100)]
drm/bridge: imx8qxp-pixel-link: Convert to platform remove callback returning void

The .remove() callback for a platform driver returns an int which makes
many driver authors wrongly assume it's possible to do error handling by
returning an error code. However the value returned is (mostly) ignored
and this typically results in resource leaks. To improve here there is a
quest to make the remove callback return void. In the first step of this
quest all drivers are converted to .remove_new() which already returns
void.

Trivially convert this driver from always returning zero in the remove
callback to the void returning variant.

Signed-off-by: Uwe Kleine-König <u.kleine-koenig@pengutronix.de>
Reviewed-by: Laurent Pinchart <laurent.pinchart+renesas@ideasonboard.com>
Signed-off-by: Neil Armstrong <neil.armstrong@linaro.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20230318190804.234610-10-u.kleine-koenig@pengutronix.de
21 months agodrm/bridge: imx8qxp-pixel-combiner: Convert to platform remove callback returning...
Uwe Kleine-König [Sat, 18 Mar 2023 19:07:53 +0000 (20:07 +0100)]
drm/bridge: imx8qxp-pixel-combiner: Convert to platform remove callback returning void

The .remove() callback for a platform driver returns an int which makes
many driver authors wrongly assume it's possible to do error handling by
returning an error code. However the value returned is (mostly) ignored
and this typically results in resource leaks. To improve here there is a
quest to make the remove callback return void. In the first step of this
quest all drivers are converted to .remove_new() which already returns
void.

Trivially convert this driver from always returning zero in the remove
callback to the void returning variant.

Signed-off-by: Uwe Kleine-König <u.kleine-koenig@pengutronix.de>
Reviewed-by: Laurent Pinchart <laurent.pinchart+renesas@ideasonboard.com>
Signed-off-by: Neil Armstrong <neil.armstrong@linaro.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20230318190804.234610-9-u.kleine-koenig@pengutronix.de
21 months agodrm/bridge: imx8qxp-ldb: Convert to platform remove callback returning void
Uwe Kleine-König [Sat, 18 Mar 2023 19:07:52 +0000 (20:07 +0100)]
drm/bridge: imx8qxp-ldb: Convert to platform remove callback returning void

The .remove() callback for a platform driver returns an int which makes
many driver authors wrongly assume it's possible to do error handling by
returning an error code. However the value returned is (mostly) ignored
and this typically results in resource leaks. To improve here there is a
quest to make the remove callback return void. In the first step of this
quest all drivers are converted to .remove_new() which already returns
void.

Trivially convert this driver from always returning zero in the remove
callback to the void returning variant.

Signed-off-by: Uwe Kleine-König <u.kleine-koenig@pengutronix.de>
Reviewed-by: Laurent Pinchart <laurent.pinchart+renesas@ideasonboard.com>
Signed-off-by: Neil Armstrong <neil.armstrong@linaro.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20230318190804.234610-8-u.kleine-koenig@pengutronix.de
21 months agodrm/bridge: imx8qm-ldb: Convert to platform remove callback returning void
Uwe Kleine-König [Sat, 18 Mar 2023 19:07:51 +0000 (20:07 +0100)]
drm/bridge: imx8qm-ldb: Convert to platform remove callback returning void

The .remove() callback for a platform driver returns an int which makes
many driver authors wrongly assume it's possible to do error handling by
returning an error code. However the value returned is (mostly) ignored
and this typically results in resource leaks. To improve here there is a
quest to make the remove callback return void. In the first step of this
quest all drivers are converted to .remove_new() which already returns
void.

Trivially convert this driver from always returning zero in the remove
callback to the void returning variant.

Signed-off-by: Uwe Kleine-König <u.kleine-koenig@pengutronix.de>
Reviewed-by: Laurent Pinchart <laurent.pinchart+renesas@ideasonboard.com>
Signed-off-by: Neil Armstrong <neil.armstrong@linaro.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20230318190804.234610-7-u.kleine-koenig@pengutronix.de
21 months agodrm/bridge: fsl-ldb: Convert to platform remove callback returning void
Uwe Kleine-König [Sat, 18 Mar 2023 19:07:50 +0000 (20:07 +0100)]
drm/bridge: fsl-ldb: Convert to platform remove callback returning void

The .remove() callback for a platform driver returns an int which makes
many driver authors wrongly assume it's possible to do error handling by
returning an error code. However the value returned is (mostly) ignored
and this typically results in resource leaks. To improve here there is a
quest to make the remove callback return void. In the first step of this
quest all drivers are converted to .remove_new() which already returns
void.

Trivially convert this driver from always returning zero in the remove
callback to the void returning variant.

Signed-off-by: Uwe Kleine-König <u.kleine-koenig@pengutronix.de>
Reviewed-by: Laurent Pinchart <laurent.pinchart+renesas@ideasonboard.com>
Signed-off-by: Neil Armstrong <neil.armstrong@linaro.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20230318190804.234610-6-u.kleine-koenig@pengutronix.de