drm/amd/pm: avoid unintentional shutdown due to temperature momentary fluctuation

An intentional delay is added on soft ctf triggered. Then there will
be a double check for the GPU temperature before taking further
action. This can avoid unintended shutdown due to temperature
momentary fluctuation.

Signed-off-by: Evan Quan <evan.quan@amd.com>
Reviewed-by: Lijo Lazar <lijo.lazar@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/pm: expose swctf threshold setting for legacy powerplay

Preparation for coming optimization which eliminates the influence of
GPU temperature momentary fluctuation.

Signed-off-by: Evan Quan <evan.quan@amd.com>
Reviewed-by: Lijo Lazar <lijo.lazar@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: 3.2.241

This version brings along the following:

- Improve debugging mechanism for Gaming FAMS
- Add monitor specific edid quirks
- Fixes for Phantom pipe
- Fixes for Shapper LUT
- Clean up asserts

Acked-by: Rodrigo Siqueira <rodrigo.siqueira@amd.com>
Signed-off-by: Aric Cyr <aric.cyr@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Take full update path if number of planes changed

[Description]
- A full update is required if the number of planes for a given
  stream changes
- The new fast update path only checked for stream and plane updates,
  but there could be a plane addition or removal without one of the
  stream and plane updates triggering a full update
- Add an explicit check for number of planes changing for a full update

Reviewed-by: Samson Tam <samson.tam@amd.com>
Acked-by: Rodrigo Siqueira <rodrigo.siqueira@amd.com>
Signed-off-by: Alvin Lee <alvin.lee2@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Create debugging mechanism for Gaming FAMS

[WHY]
To enable FAMS even during gaming sessions.

[HOW]
By leveraging a new dc.debug parameter.

Reviewed-by: Felipe Clark <felipe.clark@amd.com>
Acked-by: Rodrigo Siqueira <rodrigo.siqueira@amd.com>
Signed-off-by: Gianna Binder <gianna.binder@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Add monitor specific edid quirk

Disable FAMS on a Samsung Odyssey G9 monitor. Experiments show that this
monitor does not work well under some use cases, and is likely
implementation specific bug on the monitor's firmware.

Cc: stable@vger.kernel.org
Reviewed-by: Rodrigo Siqueira <rodrigo.siqueira@amd.com>
Signed-off-by: Aurabindo Pillai <aurabindo.pillai@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: For new fast update path, loop through each surface

[Description]
- Previous implementation didn't consider multiple surfaces in a flip
- Loop through each surface in each flip to ensure the update path is
  correct

Reviewed-by: Samson Tam <samson.tam@amd.com>
Acked-by: Rodrigo Siqueira <rodrigo.siqueira@amd.com>
Signed-off-by: Alvin Lee <alvin.lee2@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Remove Phantom Pipe Check When Calculating K1 and K2

[Why]
K1 and K2 not being setting properly when subVP is active.

[How]
Have phantom pipes use the same programing as the main pipes without
checking the paired stream

Cc: stable@vger.kernel.org
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Reviewed-by: Alvin Lee <alvin.lee2@amd.com>
Acked-by: Rodrigo Siqueira <rodrigo.siqueira@amd.com>
Signed-off-by: Austin Zheng <austin.zheng@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Limit new fast update path to addr and gamma / color

[Description]
- We want to limit the new fast update path to address and gamma updates
  only.
- Add a check in dc_update_planes_and_stream to only take the new fast
  update path if we only have the specific fast updates defined.

Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Reviewed-by: Jun Lei <jun.lei@amd.com>
Acked-by: Rodrigo Siqueira <rodrigo.siqueira@amd.com>
Signed-off-by: Alvin Lee <alvin.lee2@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Fix the delta clamping for shaper LUT

The shaper LUT requires a 10-bit value of the delta between segments. We
were using dc_fixpt_clamp_u0d10() to do that but it doesn't do what we
want it to do. It will preserve 10-bit precision after the decimal
point, but that's not quite what we want. We want 14-bit precision and
discard the 4 most-significant bytes.

To do that we'll do dc_fixpt_clamp_u0d14() & 0x3ff instead.

Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Reviewed-by: Krunoslav Kovac <krunoslav.kovac@amd.com>
Acked-by: Rodrigo Siqueira <rodrigo.siqueira@amd.com>
Signed-off-by: Harry Wentland <harry.wentland@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: Keep non-psp path for partition switch

When PSP block is not present, use direct programming.

Signed-off-by: Lijo Lazar <lijo.lazar@amd.com>
Reviewed-by: Hawking Zhang <Hawking.Zhang@amd.com>
Acked-by: Mangesh Gadre <Mangesh.Gadre@amd.com>
Tested-by: Mangesh Gadre <Mangesh.Gadre@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: program DPP shaper and 3D LUT if updated

If shaper and 3D LUT data updates, lut_3d bit in update_flag is updated
and we need to call set_input_transfer_func to program DPP shaper and 3D
LUTs. Small cleanup of code style in the related if-condition.

Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Reviewed-by: Krunoslav Kovac <krunoslav.kovac@amd.com>
Acked-by: Rodrigo Siqueira <rodrigo.siqueira@amd.com>
Signed-off-by: Melissa Wen <mwen@igalia.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

Revert "drm/amd/display: edp do not add non-edid timings"

This change causes regression when eDP and external display in mirror
mode. When external display supports low resolution than eDP, use eDP
timing to driver external display may cause corruption on external
display.

This reverts commit e749dd10e5f292061ad63d2b030194bf7d7d452c.

Cc: stable@vger.kernel.org
Link: https://gitlab.freedesktop.org/drm/amd/-/issues/2655
Signed-off-by: Hersen Wu <hersenxs.wu@amd.com>
Reviewed-by: Mario Limonciello <mario.limonciello@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: share drm device for pci amdgpu device with 1st partition device

To save render node resoure, share drm device setting for pci amdgpu
device with 1st XCP partition device.

Signed-off-by: James Zhu <James.Zhu@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/pm: Add GFX v9.4.3 unique id to sysfs

Expose unique id of GFX v9.4.3 ASICs as device attribute.

Signed-off-by: Lijo Lazar <lijo.lazar@amd.com>
Reviewed-by: Hawking Zhang <Hawking.Zhang@amd.com>
Reviewed-by: Yang Wang <kevinyang.wang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/pm: Enable pp_feature attribute

on APUs with GFX v9.4.3

Signed-off-by: Lijo Lazar <lijo.lazar@amd.com>
Acked-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu/vcn: Need to unpause dpg before stop dpg

Need to unpause dpg first, or it will hit follow error during stop dpg:
"[drm] Register(1) [regUVD_POWER_STATUS] failed to reach value 0x00000001 != 0x00000000n"

Signed-off-by: Emily Deng <Emily.Deng@amd.com>
Reviewed-by: Leo Liu <leo.liu@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: remove duplicated doorbell range init for sdma v4.4.2

Handled in earlier phase

Signed-off-by: Le Ma <le.ma@amd.com>
Reviewed-by: Hawking Zhang <Hawking.Zhang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: gpu recovers from fatal error in poison mode

Fatal error occurs in ras poison mode, mode1 reset
is used to recover gpu.

Signed-off-by: YiPeng Chai <YiPeng.Chai@amd.com>
Reviewed-by: Hawking Zhang <Hawking.Zhang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: enable mcbp by default on gfx9

It's required for high priority queues.

Link: https://gitlab.freedesktop.org/drm/amd/-/issues/2535
Reviewed-and-tested-by: Jiadong Zhu <Jiadong.Zhu@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: make mcbp a per device setting

So we can selectively enable it on certain devices.  No
intended functional change.

Reviewed-and-tested-by: Jiadong Zhu <Jiadong.Zhu@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd: Don't initialize PSP twice for Navi3x

PSP functions are already set by psp_early_init() so initializing
them a second time is unnecessary.
No intended functional changes.

Signed-off-by: Mario Limonciello <mario.limonciello@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

Revert "drm/amd: Disable PSR-SU on Parade 0803 TCON"

This reverts commit 072030b1783056b5de8b0fac5303a5e9dbc6cfde.
This is no longer necessary when using newer DMUB F/W.

Cc: stable@vger.kernel.org
Cc: Sean Wang <sean.ns.wang@amd.com>
Cc: Marc Rossi <Marc.Rossi@amd.com>
Cc: Hamza Mahfooz <Hamza.Mahfooz@amd.com>
Cc: Tsung-hua (Ryan) Lin <Tsung-hua.Lin@amd.com>
Reviewed-by: Leo Li <sunpeng.li@amd.com>
Signed-off-by: Mario Limonciello <mario.limonciello@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Set minimum requirement for using PSR-SU on Phoenix

The same parade TCON issue can potentially happen on Phoenix, and the same
PSR resilience changes have been ported into the DMUB firmware.

Don't allow running PSR-SU unless on the newer firmware.

Cc: stable@vger.kernel.org
Cc: Sean Wang <sean.ns.wang@amd.com>
Cc: Marc Rossi <Marc.Rossi@amd.com>
Cc: Hamza Mahfooz <Hamza.Mahfooz@amd.com>
Cc: Tsung-hua (Ryan) Lin <Tsung-hua.Lin@amd.com>
Signed-off-by: Mario Limonciello <mario.limonciello@amd.com>
Reviewed-by: Leo Li <sunpeng.li@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Set minimum requirement for using PSR-SU on Rembrandt

A number of parade TCONs are causing system hangs when utilized with
older DMUB firmware and PSR-SU. Some changes have been introduced into
DMUB firmware to add resilience against these failures.

Don't allow running PSR-SU unless on the newer firmware.

Cc: stable@vger.kernel.org
Cc: Sean Wang <sean.ns.wang@amd.com>
Cc: Marc Rossi <Marc.Rossi@amd.com>
Cc: Hamza Mahfooz <Hamza.Mahfooz@amd.com>
Cc: Tsung-hua (Ryan) Lin <Tsung-hua.Lin@amd.com>
Link: https://gitlab.freedesktop.org/drm/amd/-/issues/2443
Signed-off-by: Mario Limonciello <mario.limonciello@amd.com>
Reviewed-by: Leo Li <sunpeng.li@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Correct `DMUB_FW_VERSION` macro

The `DMUB_FW_VERSION` macro has a mistake in that the revision field
is off by one byte. The last byte is typically used for other purposes
and not a revision.

Cc: stable@vger.kernel.org
Cc: Sean Wang <sean.ns.wang@amd.com>
Cc: Marc Rossi <Marc.Rossi@amd.com>
Cc: Hamza Mahfooz <Hamza.Mahfooz@amd.com>
Cc: Tsung-hua (Ryan) Lin <Tsung-hua.Lin@amd.com>
Reviewed-by: Leo Li <sunpeng.li@amd.com>
Signed-off-by: Mario Limonciello <mario.limonciello@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/pm: fulfill the missing enablement for vega12/vega20 L2H and H2L interrupts

The feature mask bit was not correctly cleared. Without that, the L2H
and H2L interrupts cannot be enabled.

Signed-off-by: Evan Quan <evan.quan@amd.com>
Acked-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Remove asserts

[why]
Endless assert caused by LinesInDETChroma=0.

[how]
Don't floor for LinesInDETChroma=0

Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Reviewed-by: Jun Lei <jun.lei@amd.com>
Acked-by: Rodrigo Siqueira <rodrigo.siqueira@amd.com>
Signed-off-by: Hong-lu Cheng <hong-lu.cheng@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Work around bad DPCD state on link loss

[Why]
This display doesn't properly indicate link loss through DPCD bits such
as CR_DONE / CHANNEL_EQ_DONE / SYMBOL_LOCKED / INTERLANE_ALIGN_DONE,
which all remain set.

In addition, DPCD200Eh doesn't match the value of DPCD204h in all cases.

For these reasons, we can miss re-training the link, since we don't
properly detect link loss with this display.

[Why]
Add display-specific workaround to read DPCD204h, so that we can detect
link loss based on 128b132b-specific status bits in this register.

Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Reviewed-by: Wenjing Liu <wenjing.liu@amd.com>
Acked-by: Rodrigo Siqueira <rodrigo.siqueira@amd.com>
Signed-off-by: Ilya Bakoulin <ilya.bakoulin@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: port SRIOV VF missed changes

port SRIOV VF missed changes from gfx_v9_0 to gfx_v9_4_3.

Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Zhigang Luo <Zhigang.Luo@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd: Don't try to enable secure display TA multiple times

If the securedisplay TA failed to load the first time, it's unlikely
to work again after a suspend/resume cycle or reset cycle and it appears
to be causing problems in futher attempts.

Fixes: e42dfa66d592 ("drm/amdgpu: Add secure display TA load for Renoir")
Reported-by: Filip Hejsek <filip.hejsek@gmail.com>
Closes: https://gitlab.freedesktop.org/drm/amd/-/issues/2633
Signed-off-by: Mario Limonciello <mario.limonciello@amd.com>
Acked-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdkfd: Update interrupt handling for GFX 9.4.3

For GFX 9.4.3, interrupt handling needs to be updated for:
- Interrupt cookie will have a NodeId field. Each KFD
  node needs to check the NodeId before processing the
  interrupt.
- For CPX mode, there are additional checks of client ID
  needed to process the interrupt.
- Add NodeId to the process drain interrupt.

Signed-off-by: Mukul Joshi <mukul.joshi@amd.com>
Reviewed-by: Felix Kuehling <Felix.Kuehling@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: fix number of fence calculations

Since adding gang submit we need to take the gang size into account
while reserving fences.

Signed-off-by: Christian König <christian.koenig@amd.com>
Fixes: 4624459c84d7 ("drm/amdgpu: add gang submit frontend v6")
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: perform a bounds check before filling dirty rectangles

Currently, it is possible for us to access memory that we shouldn't.
Since, we acquire (possibly dangling) pointers to dirty rectangles
before doing a bounds check to make sure we can actually accommodate the
number of dirty rectangles userspace has requested to fill. This issue
is especially evident if a compositor requests both MPO and damage clips
at the same time, in which case I have observed a soft-hang. So, to
avoid this issue, perform the bounds check before filling a single dirty
rectangle and WARN() about it, if it is ever attempted in
fill_dc_dirty_rect().

Cc: stable@vger.kernel.org # 6.1+
Fixes: 30ebe41582d1 ("drm/amd/display: add FB_DAMAGE_CLIPS support")
Reviewed-by: Leo Li <sunpeng.li@amd.com>
Signed-off-by: Hamza Mahfooz <hamza.mahfooz@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/pm: Provide energy data in 15.625mJ units

Publish energy data in 15.625mJ unit for SMU v13.0.6. The same unit is
used in Aldebaran also.

Signed-off-by: Lijo Lazar <lijo.lazar@amd.com>
Reviewed-by: Hawking Zhang <Hawking.Zhang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: Modify for_each_inst macro

Modify it such that it doesn't change the instance mask parameter.

Signed-off-by: Lijo Lazar <lijo.lazar@amd.com>
Acked-by: Victor Skvortsov <victor.skvortsov@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu:Remove sdma halt/unhalt during frontdoor load

sdma halt/unhalt is performed by psp when frontdoor
loading used,so this can be skipped.

v2: Instead of removing halt/unhalt completely,
    driver will do it only during backdoor load.

Signed-off-by: Mangesh Gadre <Mangesh.Gadre@amd.com>
Reviewed-by: Lijo Lazar <lijo.lazar@amd.com>
Reviewed-by: Asad Kamal <asad.kamal@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: check RAS irq existence for VCN/JPEG

No RAS irq is allowed.

Signed-off-by: Tao Zhou <tao.zhou1@amd.com>
Reviewed-by: Hawking Zhang <Hawking.Zhang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/pm: add abnormal fan detection for smu 13.0.0

add abnormal fan detection for smu 13.0.0

Signed-off-by: Kenneth Feng <kenneth.feng@amd.com>
Reviewed-by: Evan Quan <evan.quan@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: remove vm sanity check from amdgpu_vm_make_compute

Since we allow kfd and graphic operate on same GPU VM to have interoperation
between them GPU VM may have been used by graphic vm operations before kfd turns
a GPU VM into a compute VM. Remove vm clean checking at amdgpu_vm_make_compute.

Signed-off-by: Xiaogang Chen <xiaogang.chen@amd.com>
Reviewed-by: Felix Kuehling <Felix.Kuehling@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd: Disable PSR-SU on Parade 0803 TCON

A number of users have reported that there are random hangs occurring
caused by PSR-SU specifically on panels that contain the parade 0803
TCON.  Users have been able to work around the issue by disabling PSR
entirely.

To avoid these hangs, disable PSR-SU when this TCON is found.

Cc: stable@vger.kernel.org
Cc: Sean Wang <sean.ns.wang@amd.com>
Cc: Marc Rossi <Marc.Rossi@amd.com>
Cc: Hamza Mahfooz <Hamza.Mahfooz@amd.com>
Suggested-by: Tsung-hua (Ryan) Lin <Tsung-hua.Lin@amd.com>
Link: https://gitlab.freedesktop.org/drm/amd/-/issues/2443
Signed-off-by: Mario Limonciello <mario.limonciello@amd.com>
Reviewed-by: Hamza Mahfooz <hamza.mahfooz@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdkfd: Enable GWS on GFX9.4.3

Enable GWS capable queue creation for forward
progress gaurantee on GFX 9.4.3.

Signed-off-by: Mukul Joshi <mukul.joshi@amd.com>
Reviewed-by: Felix Kuehling <Felix.Kuehling@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/pm: update the LC_L1_INACTIVITY setting to address possible noise issue

It is proved that insufficient LC_L1_INACTIVITY setting can cause audio
noise on some platform. With the LC_L1_INACTIVITY increased to 4ms, the
issue can be resolved.

Signed-off-by: Evan Quan <evan.quan@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: 3.2.240

This version brings along the following:
- DCN314 fixes
- DCN32x fixes
- New fast update sequence enablement
- DC mode clock switching enablement for DCN32x
- DP link loss fix
- New debugfs entry to set MST link settings

Acked-by: Hamza Mahfooz <hamza.mahfooz@amd.com>
Signed-off-by: Aric Cyr <aric.cyr@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Enable dc mode clock switching for DCN32x

- DC mode clock switch interface was previously only executed
  for DCN303. Enable it for DCN32x so that the interface is called
  correctly
- Assign function pointers for DCN32x that are used in the dc mode
  interface
- Update the dc mode interface to work generically for each ASIC
- In update_clocks, make sure to consider softmax if we're in DC mode

Reviewed-by: Jun Lei <jun.lei@amd.com>
Acked-by: Hamza Mahfooz <hamza.mahfooz@amd.com>
Signed-off-by: Alvin Lee <alvin.lee2@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Fix 128b132b link loss handling

[Why]
We don't check 128b132b-specific bits in LANE_ALIGN_STATUS_UPDATED DPCD
registers when parsing link loss status, which can cause us to miss a
link loss notification from some sinks.

[How]
Add a 128b132b-specific status bit check.

Cc: stable@vger.kernel.org # 6.3+
Reviewed-by: Wenjing Liu <wenjing.liu@amd.com>
Acked-by: Hamza Mahfooz <hamza.mahfooz@amd.com>
Signed-off-by: Ilya Bakoulin <ilya.bakoulin@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: fix odm k2 div calculation

Correct setting is div by 2 for odm. Seamless odm transitions
are enabled with enable_dp_dig_pixel_rate_div_policy debug flag.

Fixes: a2c7356f526d ("drm/amd/display: fix pixel rate update sequence")
Reviewed-by: Nicholas Kazlauskas <nicholas.kazlauskas@amd.com>
Acked-by: Hamza Mahfooz <hamza.mahfooz@amd.com>
Signed-off-by: Dmytro Laktyushkin <dmytro.laktyushkin@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Add MST Preferred Link Setting Entry

When using debugfs to change MST link settings, we need to wait until
the next stream update to apply the preferred link setting. So, trigger
a hotplug event right after the preferred link setting is applied.

Reviewed-by: Wayne Lin <wayne.lin@amd.com>
Acked-by: Hamza Mahfooz <hamza.mahfooz@amd.com>
Signed-off-by: Fangzhi Zuo <jerry.zuo@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: disable seamless boot if force_odm_combine is enabled

[Why & How]
Having seamless boot on while forcing debug option ODM combine 2 to 1
will cause some corruptions because of some missing programmings.

Cc: stable@vger.kernel.org # 6.1+
Reviewed-by: Nicholas Kazlauskas <nicholas.kazlauskas@amd.com>
Acked-by: Hamza Mahfooz <hamza.mahfooz@amd.com>
Signed-off-by: Leo Chen <sancchen@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Disable DC Mode Capping On DCN321

Why:
Limiting clocks to DC mode max results in some
display modes to no longer be supported

How:
Disable the path that limits the clock values

Fixes: 3b718dcaf163 ("drm/amd/display: Filter out AC mode frequencies on DC mode systems")
Reviewed-by: Martin Leung <martin.leung@amd.com>
Acked-by: Hamza Mahfooz <hamza.mahfooz@amd.com>
Signed-off-by: Austin Zheng <austin.zheng@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: enable the new fast update path for supported ASICs

The new fast update sequence is now supported on some ASICs. So, enable
it by default for all applicable ASICs.

Reviewed-by: Samson Tam <samson.tam@amd.com>
Acked-by: Hamza Mahfooz <hamza.mahfooz@amd.com>
Signed-off-by: Alvin Lee <alvin.lee2@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Clear update flags at end of flip

Clear update flags so the next flip does not have any redundant
programming (if a subsequent flip does not have a stream or plane
update, the update flags are not cleared).

Fixes: 0baae6246307 ("drm/amd/display: Refactor fast update to use new HWSS build sequence")
Reviewed-by: Jun Lei <jun.lei@amd.com>
Acked-by: Hamza Mahfooz <hamza.mahfooz@amd.com>
Signed-off-by: Alvin Lee <alvin.lee2@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Fix pipe check condition for manual trigger

Condition for programming manually trigger used the wrong pipe (always
used top pipe instead of the one we are iterating through).

Fixes: 0baae6246307 ("drm/amd/display: Refactor fast update to use new HWSS build sequence")
Reviewed-by: Samson Tam <samson.tam@amd.com>
Acked-by: Hamza Mahfooz <hamza.mahfooz@amd.com>
Signed-off-by: Alvin Lee <alvin.lee2@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: add missing ABM registers

[Why]
We are currently missing some ABM registers.

[How]
Add the missing registers to dce_abm.h.

Reviewed-by: Jun Lei <jun.lei@amd.com>
Reviewed-by: Chris Park <chris.park@amd.com>
Acked-by: Hamza Mahfooz <hamza.mahfooz@amd.com>
Signed-off-by: Sridevi Arvindekar <sridevi.arvindekar@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Add Clock Table Entry With Max DC Values

Why:
Certain display configs resulted in underflow

How:
Add an entry containing all max DC clock timings

Reviewed-by: Alvin Lee <alvin.lee2@amd.com>
Acked-by: Hamza Mahfooz <hamza.mahfooz@amd.com>
Signed-off-by: Austin Zheng <austin.zheng@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: add a NULL pointer check

[Why & How]

We have to check if stream is properly initialized before calling
find_matching_pll(), otherwise we might end up trying to deferecence a
NULL pointer.

Cc: stable@vger.kernel.org # 6.1+
Reviewed-by: Nicholas Kazlauskas <nicholas.kazlauskas@amd.com>
Acked-by: Hamza Mahfooz <hamza.mahfooz@amd.com>
Signed-off-by: Sung-huai Wang <danny.wang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

Partially revert "drm/amd/display: Fix possible underflow for displays with large vblank"

This partially reverts commit de231189e7bf ("drm/amd/display: Fix
possible underflow for displays with large vblank").

[Why]
The increased value of VBlankNomDefaultUS causes underflow at the
desktop of an IP KVM setup

[How]
Change the value from 800 back to 668

Reviewed-by: Nicholas Kazlauskas <nicholas.kazlauskas@amd.com>
Reviewed-by: Jun Lei <jun.lei@amd.com>
Acked-by: Hamza Mahfooz <hamza.mahfooz@amd.com>
Signed-off-by: Daniel Miess <daniel.miess@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

Revert "drm/amd/display: Move DCN314 DOMAIN power control to DMCUB"

This reverts commit e383b12709e32d6494c948422070c2464b637e44.

Controling hubp power gating using the DMCUB isn't stable so we
are reverting this change to move control back into the driver.

Cc: stable@vger.kernel.org # 6.3+
Reviewed-by: Nicholas Kazlauskas <nicholas.kazlauskas@amd.com>
Acked-by: Hamza Mahfooz <hamza.mahfooz@amd.com>
Signed-off-by: Daniel Miess <daniel.miess@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: disable RCO for DCN314

[Why]
RCO is causing error messages on some DCN314 systems

[How]
Force disable RCO for DCN314

Fixes: 17fbdbda9cc8 ("drm/amd/display: Enable dcn314 DPP RCO")
Reviewed-by: Nicholas Kazlauskas <nicholas.kazlauskas@amd.com>
Acked-by: Hamza Mahfooz <hamza.mahfooz@amd.com>
Signed-off-by: Daniel Miess <daniel.miess@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: disable power gating for DCN314

[Why]
Power gating is causing error messages on some DCN314 systems

[How]
Force disable power gating for DCN314

Fixes: 4cc1cebe08bf ("drm/amd/display: Re-enable DPP/HUBP Power Gating")
Reviewed-by: Nicholas Kazlauskas <nicholas.kazlauskas@amd.com>
Acked-by: Hamza Mahfooz <hamza.mahfooz@amd.com>
Signed-off-by: Daniel Miess <daniel.miess@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: Skip TMR for MP0_HWIP 13.0.6

For SRIOV VF, no TMR needed.

Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Zhigang Luo <Zhigang.Luo@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: Move clocks closer to its only usage in amdgpu_parse_cg_state()

After commit 8020f0f9316b ("drm/amd/amdgpu: enable W=1 for amdgpu"),
there is an instance of -Wunused-const-variable when CONFIG_DEBUG_FS is
disabled:

  drivers/gpu/drm/amd/amdgpu/../pm/amdgpu_pm.c:38:34: error: unused variable 'clocks' [-Werror,-Wunused-const-variable]
     38 | static const struct cg_flag_name clocks[] = {
        |                                  ^
  1 error generated.

clocks is only used when CONFIG_DEBUG_FS is set, so move the definition
into the CONFIG_DEBUG_FS block right above its only usage to clear up
the warning.

Signed-off-by: Nathan Chancellor <nathan@kernel.org>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/pm: revise the ASPM settings for thunderbolt attached scenario

Also, correct the comment for NAVI10_PCIE__LC_L1_INACTIVITY_TBT_DEFAULT
as 0x0000000E stands for 400ms instead of 4ms.

Signed-off-by: Evan Quan <evan.quan@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: fix clearing mappings for BOs that are always valid in VM

Per VM BOs must be marked as moved or otherwise their ranges are not
updated on use which might be necessary when the replace operation
splits mappings.

This fixes random GPU hangs when replacing sparse mappings from the
userspace, while OP_MAP/OP_UNMAP works fine because always valid BOs
are correctly handled there.

Cc: stable@vger.kernel.org
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdkfd: set coherent host access capability flag

This flag determines whether the host possesses coherent access to
the memory of the device.

Signed-off-by: Alex Sierra <alex.sierra@amd.com>
Reviewed-by: Felix Kuehling <Felix.Kuehling@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: Add vbios attribute only if supported

Not all devices carry VBIOS version information. Add the device
attribute only if supported.

Signed-off-by: Lijo Lazar <lijo.lazar@amd.com>
Reviewed-by: Le Ma <le.ma@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu/atomfirmware: fix LPDDR5 width reporting

LPDDR5 channels are 32 bit rather than 64, report the width properly
in the log.

v2: Only LPDDR5 are 32 bits per channel.  DDR5 is 64 bits per channel

Link: https://gitlab.freedesktop.org/drm/amd/-/issues/2468
Acked-by: Hawking Zhang <Hawking.Zhang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: Remove CONFIG_DEBUG_FS guard around body of amdgpu_rap_debugfs_init()

After commit 8020f0f9316b ("drm/amd/amdgpu: enable W=1 for amdgpu"),
there is an instance of -Wunused-const-variable when CONFIG_DEBUG_FS is
disabled:

  drivers/gpu/drm/amd/amdgpu/amdgpu_rap.c:110:37: error: unused variable 'amdgpu_rap_debugfs_ops' [-Werror,-Wunused-const-variable]
    110 | static const struct file_operations amdgpu_rap_debugfs_ops = {
        |                                     ^
  1 error generated.

There is no reason for the body of this function to be guarded when
CONFIG_DEBUG_FS is disabled, as debugfs_create_file() is a stub that
just returns an error pointer in that situation. Remove the preprocessor
guards so that the variable never appears unused, while not changing
anything at run time.

Signed-off-by: Nathan Chancellor <nathan@kernel.org>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/pm: remove unneeded variable

fix the following coccicheck warning:

drivers/gpu/drm/amd/pm/swsmu/smu11/navi10_ppt.c:1657:14-18: Unneeded variable: "size".

Signed-off-by: Mingtong Bao <baomingtong001@208suo.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: Move calculation of xcp per memory node

Its value is required for finding the memory id of xcp.

Fixes: d26ea1b346e7 ("drm/amdgpu: Add xcp manager num_xcp_per_mem_partition")
Signed-off-by: Lijo Lazar <lijo.lazar@amd.com>
Reviewed-by: Hawking Zhang <Hawking.Zhang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: Skip mark offset for high priority rings

Only low priority rings are using chunks to save the offset.
Bypass the mark offset callings from high priority rings.

Signed-off-by: Jiadong Zhu <Jiadong.Zhu@amd.com>
Acked-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

amd/display/dc: remove repeating expression

Identify issues that arise by using the tests/doubletest.cocci
semantic patch. Need to remove duplicate expression in if statement.

Signed-off-by: Wang Ming <machel@vivo.com>
Reviewed-by: Ammar Faizi <ammarfaizi2@gnuweeb.org>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

Merge tag 'drm-msm-next-2023-06-18' of https://gitlab.freedesktop.org/drm/msm into drm-next

Updates for v6.5.. this includes a backmerg of drm-next tree to be able
to use new DRM DSC helpers.

Core:
+ Add Marijn Suijten as drm/msm reviewer
+ Adreno A660 bindings
+ SM8350 MDSS bindings fix
+ Fix adreno_is_a690() warnings
+ More generic (DRM) and MSM-specific DSC helpers

DP:
+ Removed obsolete USB-PD remains
+ Documented DP compatible string for sm8550 platform

DPU:
+ Enable missing features (DSPP, DSC, split display) on sc8180x,
  sc8280xp, sm8450
+ Enabled writeback on sc7280
+ Implemented tearcheck support to support vsync on SM150 and
  newer platforms
+ Native HDMI output support
+ Dropped unused features: regdma, GC, IGC
+ Fixed the DSC flush operations
+ Simplified QoS handling, removing obsolete and unused features
  and merging SSPP and WB code paths
+ Reworked dpu_encoder initialisation path
+ Enabled DSPP support on sdm845
+ Disabled color-management if DSPP blocks are not available
+ Added support for DSC 1.2 blocks found on sm8350 and later
+ Added .fb_dirty to fix CMD panels

DSI:
+ Drop powerup quirks in favour of using pre_enable_prev_first for
  downstream bridges
+ Fixed 14nm DSI PHY programming
+ Added support for DSI and 28nm DSI PHY on MSM8226 platform
+ Make use of DRM and MSM DSC helpers

MDP5:
+ Added support for display controller on MSM8226 platform

GPU:
+ A690 support
+ Don't set IO_PGTABLE_QUIRK_ARM_OUTER_WBWA on devices with coherent SMMU
  (like A690)
+ Move cmdstream dumping out of fence signaling path
+ Cleanups
+ Support for a6xx devices without GMU (aka "GMU wrapper"
+ a610 support
+ a619_holi support (a619 variant without GMU)

Signed-off-by: Dave Airlie <airlied@redhat.com>
From: Rob Clark <robdclark@gmail.com>
Link: https://patchwork.freedesktop.org/patch/msgid/CAF6AEGsUB=tRB4nR6ZCJMuLhro5zN3BQWUSywVYbaipqqDZ_cQ@mail.gmail.com

Backmerge tag 'v6.4-rc7' of git://git./linux/kernel/git/torvalds/linux into drm-next

Linux 6.4-rc7

Need this to pull in the msm work.

Signed-off-by: Dave Airlie <airlied@redhat.com>

Merge tag 'amd-drm-next-6.5-2023-06-16' of https://gitlab.freedesktop.org/agd5f/linux into drm-next

amd-drm-next-6.5-2023-06-16:

amdgpu:
- Misc display fixes
- W=1 fixes
- Improve scheduler naming
- DCN 3.1.4 fixes
- kdoc fixes
- Enable W=1
- VCN 4.0 fix
- xgmi fixes
- TOPDOWN fix for large BAR systems
- eDP fix
- PSR fixes
- SubVP fixes
- Freesync fix
- DPIA fix
- SMU 13.0.5 fixes
- vblflash fix
- RAS fixes
- SDMA 4 fix
- BO locking fix
- BO backing store fix
- NBIO 7.9 fixes
- GC 9.4.3 fixes
- GPU reset recovery fixes
- HMM fix

amdkfd:
- Fix NULL check
- Trap fixes
- Queue count fix
- Add event age tracking

radeon:
- fbdev client fix

scheduler:
- Avoid an infinite loop

UAPI:
- Add KFD event age tracking:
  Proposed ROCT-Thunk-Interface:
  https://github.com/RadeonOpenCompute/ROCT-Thunk-Interface/commit/efdbf6cfbc026bd68ac3c35d00dacf84370eb81e
  https://github.com/RadeonOpenCompute/ROCT-Thunk-Interface/commit/1820ae0a2db85b6f584611dc0cde1a00e7c22915
  Proposed ROCR-Runtime:
  https://github.com/RadeonOpenCompute/ROCR-Runtime/compare/master...zhums:ROCR-Runtime:new_event_wait_review
  https://github.com/RadeonOpenCompute/ROCR-Runtime/commit/e1f5bdb88eb882ac798aeca2c00ea3fbb2dba459
  https://github.com/RadeonOpenCompute/ROCR-Runtime/commit/7d26afd14107b5c2a754c1a3f415d89f3aabb503

drm:
- DP MST fix

Signed-off-by: Dave Airlie <airlied@redhat.com>
From: Alex Deucher <alexander.deucher@amd.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20230616163548.7706-1-alexander.deucher@amd.com

Merge tag 'drm-misc-next-fixes-2023-06-15' of git://anongit.freedesktop.org/drm/drm-misc into drm-next

Short summary of fixes pull:

 * Fix fbdev initializer macros

Signed-off-by: Dave Airlie <airlied@redhat.com>
From: Thomas Zimmermann <tzimmermann@suse.de>
Link: https://patchwork.freedesktop.org/patch/msgid/20230615114009.GA27261@linux-uq9g

Linux 6.4-rc7

drm/msm/a6xx: Add A610 speedbin support

A610 is implemented on at least three SoCs: SM6115 (bengal), SM6125
(trinket) and SM6225 (khaje). Trinket does not support speed binning
(only a single SKU exists) and we don't yet support khaje upstream.
Hence, add a fuse mapping table for bengal to allow for per-chip
frequency limiting.

Reviewed-by: Dmitry Baryshkov <dmitry.baryshkov@linaro.org>
Reviewed-by: Akhil P Oommen <quic_akhilpo@quicinc.com>
Signed-off-by: Konrad Dybcio <konrad.dybcio@linaro.org>
Patchwork: https://patchwork.freedesktop.org/patch/542780/
Signed-off-by: Rob Clark <robdclark@chromium.org>

drm/msm/a6xx: Add A619_holi speedbin support

A619_holi is implemented on at least two SoCs: SM4350 (holi) and SM6375
(blair). This is what seems to be a first occurrence of this happening,
but it's easy to overcome by guarding the SoC-specific fuse values with
of_machine_is_compatible(). Do just that to enable frequency limiting
on these SoCs.

Reviewed-by: Dmitry Baryshkov <dmitry.baryshkov@linaro.org>
Reviewed-by: Akhil P Oommen <quic_akhilpo@quicinc.com>
Signed-off-by: Konrad Dybcio <konrad.dybcio@linaro.org>
Patchwork: https://patchwork.freedesktop.org/patch/542772/
Signed-off-by: Rob Clark <robdclark@chromium.org>

drm/msm/a6xx: Use adreno_is_aXYZ macros in speedbin matching

Before transitioning to using per-SoC and not per-Adreno speedbin
fuse values (need another patchset to land elsewhere), a good
improvement/stopgap solution is to use adreno_is_aXYZ macros in
place of explicit revision matching. Do so to allow differentiating
between A619 and A619_holi.

Reviewed-by: Dmitry Baryshkov <dmitry.baryshkov@linaro.org>
Reviewed-by: Akhil P Oommen <quic_akhilpo@quicinc.com>
Signed-off-by: Konrad Dybcio <konrad.dybcio@linaro.org>
Patchwork: https://patchwork.freedesktop.org/patch/542777/
Signed-off-by: Rob Clark <robdclark@chromium.org>

drm/msm/a6xx: Use "else if" in GPU speedbin rev matching

The GPU can only be one at a time. Turn a series of ifs into if +
elseifs to save some CPU cycles.

Reviewed-by: Dmitry Baryshkov <dmitry.baryshkov@linaro.org>
Reviewed-by: Akhil P Oommen <quic_akhilpo@quicinc.com>
Signed-off-by: Konrad Dybcio <konrad.dybcio@linaro.org>
Patchwork: https://patchwork.freedesktop.org/patch/542770/
Signed-off-by: Rob Clark <robdclark@chromium.org>

drm/msm/a6xx: Fix some A619 tunables

Adreno 619 expects some tunables to be set differently. Make up for it.

Fixes: b7616b5c69e6 ("drm/msm/adreno: Add A619 support")
Reviewed-by: Dmitry Baryshkov <dmitry.baryshkov@linaro.org>
Reviewed-by: Akhil P Oommen <quic_akhilpo@quicinc.com>
Signed-off-by: Konrad Dybcio <konrad.dybcio@linaro.org>
Patchwork: https://patchwork.freedesktop.org/patch/542782/
Signed-off-by: Rob Clark <robdclark@chromium.org>

drm/msm/a6xx: Add A610 support

A610 is one of (if not the) lowest-tier SKUs in the A6XX family. It
features no GMU, as it's implemented solely on SoCs with SMD_RPM.
What's more interesting is that it does not feature a VDDGX line
either, being powered solely by VDDCX and has an unfortunate hardware
quirk that makes its reset line broken - after a couple of assert/
deassert cycles, it will hang for good and will not wake up again.

This GPU requires mesa changes for proper rendering, and lots of them
at that. The command streams are quite far away from any other A6XX
GPU and hence it needs special care. This patch was validated both
by running an (incomplete) downstream mesa with some hacks (frames
rendered correctly, though some instructions made the GPU hangcheck
which is expected - garbage in, garbage out) and by replaying RD
traces captured with the downstream KGSL driver - no crashes there,
ever.

Add support for this GPU on the kernel side, which comes down to
pretty simply adding A612 HWCG tables, altering a few values and
adding a special case for handling the reset line.

Reviewed-by: Dmitry Baryshkov <dmitry.baryshkov@linaro.org>
Signed-off-by: Konrad Dybcio <konrad.dybcio@linaro.org>
Patchwork: https://patchwork.freedesktop.org/patch/542779/
Signed-off-by: Rob Clark <robdclark@chromium.org>

drm/msm/a6xx: Add support for A619_holi

A619_holi is a GMU-less variant of the already-supported A619 GPU.
It's present on at least SM4350 (holi) and SM6375 (blair). No mesa
changes are required. Add the required kernel-side support for it.

Signed-off-by: Konrad Dybcio <konrad.dybcio@linaro.org>
Patchwork: https://patchwork.freedesktop.org/patch/542775/
Signed-off-by: Rob Clark <robdclark@chromium.org>

drm/msm/adreno: Disable has_cached_coherent in GMU wrapper configurations

A610 and A619_holi don't support the feature. Disable it to make the GPU stop
crashing after almost each and every submission - the received data on
the GPU end was simply incomplete in garbled, resulting in almost nothing
being executed properly. Extend the disablement to adreno_has_gmu_wrapper,
as none of the GMU wrapper Adrenos that don't support yet seem to feature it.

Reviewed-by: Akhil P Oommen <quic_akhilpo@quicinc.com>
Signed-off-by: Konrad Dybcio <konrad.dybcio@linaro.org>
Patchwork: https://patchwork.freedesktop.org/patch/542774/
Signed-off-by: Rob Clark <robdclark@chromium.org>

drm/msm/a6xx: Introduce GMU wrapper support

Some (particularly SMD_RPM, a.k.a non-RPMh) SoCs implement A6XX GPUs
but don't implement the associated GMUs. This is due to the fact that
the GMU directly pokes at RPMh. Sadly, this means we have to take care
of enabling & scaling power rails, clocks and bandwidth ourselves.

Reuse existing Adreno-common code and modify the deeply-GMU-infused
A6XX code to facilitate these GPUs. This involves if-ing out lots
of GMU callbacks and introducing a new type of GMU - GMU wrapper (it's
the actual name that Qualcomm uses in their downstream kernels).

This is essentially a register region which is convenient to model
as a device. We'll use it for managing the GDSCs. The register
layout matches the actual GMU_CX/GX regions on the "real GMU" devices
and lets us reuse quite a bit of gmu_read/write/rmw calls.

Signed-off-by: Konrad Dybcio <konrad.dybcio@linaro.org>
Patchwork: https://patchwork.freedesktop.org/patch/542766/
Signed-off-by: Rob Clark <robdclark@chromium.org>

drm/msm/a6xx: Move CX GMU power counter enablement to hw_init

Since the introduction of A6xx support, we've been enabling the CX GMU
power counter 0 in a bit of a weird spot. Move it to hw_init so that
GMU wrapper GPUs can reuse the same code paths. As a bonus, this order
makes it easier to compare mainline and downstream register access traces.

Signed-off-by: Konrad Dybcio <konrad.dybcio@linaro.org>
Patchwork: https://patchwork.freedesktop.org/patch/542765/
Signed-off-by: Rob Clark <robdclark@chromium.org>

drm/msm/a6xx: Extend and explain UBWC config

Rename lower_bit to hbb_lo and explain what it signifies.
Add explanations (wherever possible to other tunables).

Port setting min_access_length, ubwc_mode and hbb_hi from downstream.

Reviewed-by: Rob Clark <robdclark@gmail.com>
Reviewed-by: Akhil P Oommen <quic_akhilpo@quicinc.com>
Signed-off-by: Konrad Dybcio <konrad.dybcio@linaro.org>
Patchwork: https://patchwork.freedesktop.org/patch/542764/
Signed-off-by: Rob Clark <robdclark@chromium.org>

drm/msm/a6xx: Remove both GBIF and RBBM GBIF halt on hw init

Currently we're only deasserting REG_A6XX_RBBM_GBIF_HALT, but we also
need REG_A6XX_GBIF_HALT to be set to 0.

This is typically done automatically on successful GX collapse, but in
case that fails, we should take care of it.

Also, add a memory barrier to ensure it's gone through before jumping
to further initialization.

Reviewed-by: Dmitry Baryshkov <dmitry.baryshkov@linaro.org>
Signed-off-by: Konrad Dybcio <konrad.dybcio@linaro.org>
Patchwork: https://patchwork.freedesktop.org/patch/542760/
Signed-off-by: Rob Clark <robdclark@chromium.org>

drm/msm/a6xx: Add a helper for software-resetting the GPU

Introduce a6xx_gpu_sw_reset() in preparation for adding GMU wrapper
GPUs and reuse it in a6xx_gmu_force_off().

This helper, contrary to the original usage in GMU code paths, adds
a readback+delay sequence to ensure that the reset is never deasserted
too quickly due to e.g. OoO execution going crazy.

Signed-off-by: Konrad Dybcio <konrad.dybcio@linaro.org>
Patchwork: https://patchwork.freedesktop.org/patch/542758/
Signed-off-by: Rob Clark <robdclark@chromium.org>

drm/msm/a6xx: Improve a6xx_bus_clear_pending_transactions()

Unify the indentation and explain the cryptic 0xF value.

Reviewed-by: Akhil P Oommen <quic_akhilpo@quicinc.com>
Signed-off-by: Konrad Dybcio <konrad.dybcio@linaro.org>
Patchwork: https://patchwork.freedesktop.org/patch/542756/
Signed-off-by: Rob Clark <robdclark@chromium.org>

drm/msm/a6xx: Move a6xx_bus_clear_pending_transactions to a6xx_gpu

This function is responsible for telling the GPU to halt transactions
on all of its relevant buses, drain them and leave them in a predictable
state, so that the GPU can be e.g. reset cleanly.

Move the function to a6xx_gpu.c, remove the static keyword and add a
prototype in a6xx_gpu.h to accomodate for the move.

Reviewed-by: Akhil P Oommen <quic_akhilpo@quicinc.com>
Signed-off-by: Konrad Dybcio <konrad.dybcio@linaro.org>
Patchwork: https://patchwork.freedesktop.org/patch/542762/
Signed-off-by: Rob Clark <robdclark@chromium.org>

drm/msm/a6xx: Move force keepalive vote removal to a6xx_gmu_force_off()

As pointed out by Akhil during the review process of GMU wrapper
introduction [1], it makes sense to move this write into the function
that's responsible for forcibly shutting the GMU off.

It is also very convenient to move this to GMU-specific code, so that
it does not have to be guarded by an if-condition to avoid calling it
on GMU wrapper targets.

Move the write to the aforementioned a6xx_gmu_force_off() to achieve
that. No effective functional change.

[1] https://lore.kernel.org/linux-arm-msm/20230501194022.GA18382@akhilpo-linux.qualcomm.com/

Reviewed-by: Akhil P Oommen <quic_akhilpo@quicinc.com>
Signed-off-by: Konrad Dybcio <konrad.dybcio@linaro.org>
Patchwork: https://patchwork.freedesktop.org/patch/542752/
Signed-off-by: Rob Clark <robdclark@chromium.org>

drm/msm/a6xx: Remove static keyword from sptprac en/disable functions

These two will be reused by at least A619_holi in the non-gmu
paths. Turn them non-static them to make it possible.

Reviewed-by: Dmitry Baryshkov <dmitry.baryshkov@linaro.org>
Signed-off-by: Konrad Dybcio <konrad.dybcio@linaro.org>
Patchwork: https://patchwork.freedesktop.org/patch/542751/
Signed-off-by: Rob Clark <robdclark@chromium.org>

drm/msm/adreno: Use adreno_is_revn for A690

The adreno_is_revn rework came at the same time as A690 introduction
and that resulted in it not covering all cases. Fix it.

Signed-off-by: Konrad Dybcio <konrad.dybcio@linaro.org>
Patchwork: https://patchwork.freedesktop.org/patch/542754/
Signed-off-by: Rob Clark <robdclark@chromium.org>

dt-bindings: display/msm/gmu: Add GMU wrapper

The "GMU Wrapper" is Qualcomm's name for "let's treat the GPU blocks
we'd normally assign to the GMU as if they were a part of the GMU, even
though they are not". It's a (good) software representation of the GMU_CX
and GMU_GX register spaces within the GPUSS that helps us programatically
treat these de-facto GMU-less parts in a way that's very similar to their
GMU-equipped cousins, massively saving up on code duplication.

The "wrapper" register space was specifically designed to mimic the layout
of a real GMU, though it rather obviously does not have the M3 core et al.

To sum it all up, the GMU wrapper is essentially a register space within
the GPU, which Linux sees as a dumbed-down regular GMU: there's no clocks,
interrupts, multiple reg spaces, iommus and OPP. Document it.

Reviewed-by: Krzysztof Kozlowski <krzysztof.kozlowski@linaro.org>
Signed-off-by: Konrad Dybcio <konrad.dybcio@linaro.org>
Patchwork: https://patchwork.freedesktop.org/patch/542750/
Signed-off-by: Rob Clark <robdclark@chromium.org>

dt-bindings: display/msm: gpu: Document GMU wrapper-equipped A6xx

The "GMU Wrapper" is Qualcomm's name for "let's treat the GPU blocks
we'd normally assign to the GMU as if they were a part of the GMU, even
though they are not". It's a (good) software representation of the GMU_CX
and GMU_GX register spaces within the GPUSS that helps us programatically
treat these de-facto GMU-less parts in a way that's very similar to their
GMU-equipped cousins, massively saving up on code duplication.

The "wrapper" register space was specifically designed to mimic the layout
of a real GMU, though it rather obviously does not have the M3 core et al.

GMU wrapper-equipped A6xx GPUs require clocks and clock-names to be
specified under the GPU node, just like their older cousins. Account
for that.

Acked-by: Rob Herring <robh@kernel.org>
Reviewed-by: Krzysztof Kozlowski <krzysztof.kozlowski@linaro.org>
Signed-off-by: Konrad Dybcio <konrad.dybcio@linaro.org>
Patchwork: https://patchwork.freedesktop.org/patch/542748/
Signed-off-by: Rob Clark <robdclark@chromium.org>

Merge tag 'scsi-fixes' of git://git./linux/kernel/git/jejb/scsi

Pull SCSI fixes from James Bottomley:
 "Four fixes, all in drivers: three fairly obvious small ones and a
  large one in aacraid to add block queue completion mapping and fix a
  CPU offline hang"

* tag 'scsi-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/jejb/scsi:
  scsi: lpfc: Fix incorrect big endian type assignment in bsg loopback path
  scsi: target: core: Fix error path in target_setup_session()
  scsi: storvsc: Always set no_report_opcodes
  scsi: aacraid: Reply queue mapping to CPUs based on IRQ affinity

Merge tag 'ata-6.4-rc7' of git://git./linux/kernel/git/dlemoal/libata

Pull ata fix from Damien Le Moal:

 - Avoid deadlocks on resume from sleep by delaying scsi rescan until
   the scsi device is also fully resumed.

* tag 'ata-6.4-rc7' of git://git.kernel.org/pub/scm/linux/kernel/git/dlemoal/libata:
  ata: libata-scsi: Avoid deadlock on rescan after device resume

Merge tag 'parisc-for-6.4-4' of git://git./linux/kernel/git/deller/parisc-linux

Pull parisc fix from Helge Deller:

 - Drop redundant register definitions to fix build with latest binutils

* tag 'parisc-for-6.4-4' of git://git.kernel.org/pub/scm/linux/kernel/git/deller/parisc-linux:
  parisc: Delete redundant register definitions in <asm/assembly.h>