linux-2.6-microblaze.git
2 years agodrm/amd/display: Fix prefetch bandwidth calculation for DCN3.1
Nicholas Kazlauskas [Wed, 29 Sep 2021 15:37:33 +0000 (11:37 -0400)]
drm/amd/display: Fix prefetch bandwidth calculation for DCN3.1

[Why]
Prefetch BW calculated is lower than the DML reference because of a
porting error that's excluding cursor and row bandwidth from the
pixel data bandwidth.

[How]
Change the dml_max4 to dml_max3 and include cursor and row bandwidth
in the same calculation as the rest of the pixel data during vactive.

Reviewed-by: Dmytro Laktyushkin <Dmytro.Laktyushkin@amd.com>
Acked-by: Agustin Gutierrez Sanchez <agustin.gutierrez@amd.com>
Signed-off-by: Nicholas Kazlauskas <nicholas.kazlauskas@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2 years agodrm/amd/display: Fix DP2 SE and LE SYMCLK selection for B0 PHY
Hansen [Tue, 28 Sep 2021 20:09:47 +0000 (16:09 -0400)]
drm/amd/display: Fix DP2 SE and LE SYMCLK selection for B0 PHY

Remap phyd32clk to PHYF and PHYG for B0, PHYC and PHYD are unused

Reviewed-by: Nicholas Kazlauskas <Nicholas.Kazlauskas@amd.com>
Acked-by: Agustin Gutierrez Sanchez <agustin.gutierrez@amd.com>
Signed-off-by: Hansen <Hansen.Dsouza@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2 years agodrm/amd/display: Limit display scaling to up to true 4k for DCN 3.1
Nikola Cornij [Wed, 29 Sep 2021 02:43:52 +0000 (22:43 -0400)]
drm/amd/display: Limit display scaling to up to true 4k for DCN 3.1

[why]
The requirement is that image width up to 4096 shall be supported

Reviewed-by: Aric Cyr <Aric.Cyr@amd.com>
Acked-by: Agustin Gutierrez Sanchez <agustin.gutierrez@amd.com>
Signed-off-by: Nikola Cornij <nikola.cornij@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2 years agodrm/amd/display: Removed power down on boot from DCN31
Lai, Derek [Tue, 28 Sep 2021 05:58:56 +0000 (13:58 +0800)]
drm/amd/display: Removed power down on boot from DCN31

[Why]
Error message on Linux when booting.

[How]
Removed power down on boot from DCN31 HW init
to match DCN10 HW init.

Reviewed-by: Anthony Koo <Anthony.Koo@amd.com>
Acked-by: Agustin Gutierrez Sanchez <agustin.gutierrez@amd.com>
Signed-off-by: Derek Lai <derek.lai@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2 years agodrm/amd/display: Validate plane rects before use
Aric Cyr [Tue, 28 Sep 2021 20:05:27 +0000 (16:05 -0400)]
drm/amd/display: Validate plane rects before use

[Why]
Calculation of scaling ratio can result in a crash due to zero'd src or
dst plane rects.

[How]
Validate that src and dst rects are valid before using for scaling
calculations.

Reviewed-by: Josip Pavic <Josip.Pavic@amd.com>
Acked-by: Agustin Gutierrez Sanchez <agustin.gutierrez@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>
2 years agodrm/amd/display: correct apg audio channel enable golden value
Wenjing Liu [Tue, 28 Sep 2021 17:09:14 +0000 (13:09 -0400)]
drm/amd/display: correct apg audio channel enable golden value

Hardware team has recommended to generically hard code this register to
0xFF as part of the effort to eventually remove this control.  However
we set it to 0xF instead.
This causes 4 of audio 8ch to be muted.

Reviewed-by: Ariel Bernstein <eric.bernstein@amd.com>
Acked-by: Agustin Gutierrez Sanchez <agustin.gutierrez@amd.com>
Signed-off-by: Wenjing Liu <wenjing.liu@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2 years agodrm/amd/display: do not compare integers of different widths
Josip Pavic [Mon, 27 Sep 2021 21:38:10 +0000 (17:38 -0400)]
drm/amd/display: do not compare integers of different widths

[Why & How]
Increase width of some variables to avoid comparing integers of
different widths

Reviewed-by: Aric Cyr <Aric.Cyr@amd.com>
Acked-by: Agustin Gutierrez Sanchez <agustin.gutierrez@amd.com>
Signed-off-by: Josip Pavic <Josip.Pavic@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2 years agodrm/amd/display: Clean Up VPG Low Mem Power
Michael Strauss [Tue, 21 Sep 2021 17:35:19 +0000 (13:35 -0400)]
drm/amd/display: Clean Up VPG Low Mem Power

[WHAT]
One of the current VPG power on calls is unnecessary

Reviewed-by: Eric Yang <eric.yang2@amd.com>
Acked-by: Agustin Gutierrez Sanchez <agustin.gutierrez@amd.com>
Signed-off-by: Michael Strauss <michael.strauss@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2 years agodrm/amd/display: add DP2.0 debug option to set MST_EN for SST stream
Wenjing Liu [Fri, 24 Sep 2021 20:12:34 +0000 (16:12 -0400)]
drm/amd/display: add DP2.0 debug option to set MST_EN for SST stream

[why]
Some DP2.0 RX requires us to set MST_EN even for SST configuration.
We added this debug option so we can configure this temporary workaround
for the RX.

Reviewed-by: George Shen <george.shen@amd.com>
Acked-by: Agustin Gutierrez Sanchez <agustin.gutierrez@amd.com>
Signed-off-by: Wenjing Liu <wenjing.liu@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2 years agodrm/amd/display: Do not skip link training on DP quick hot plug
Jimmy Kizito [Wed, 22 Sep 2021 13:50:38 +0000 (09:50 -0400)]
drm/amd/display: Do not skip link training on DP quick hot plug

[Why]
When rapidly plugging and unplugging a DP sink, detection link
training can be mistakenly skipped.

This is due to the hotplug processing occurring before the
encoder assignment logic has had a chance to process the removal
of a stream. The encoder that would be used for detection link
training is then erroneously reported as already in use and
detection link training is skipped.

[How]
During hot plug processing, only determine a link encoder to be
unavailable for a particular link if it has been assigned to a
different link.

Reviewed-by: Meenakshikumar Somasundaram <Meenakshikumar.Somasundaram@amd.com>
Reviewed-by: Jun Lei <Jun.Lei@amd.com>
Acked-by: Agustin Gutierrez Sanchez <agustin.gutierrez@amd.com>
Signed-off-by: Jimmy Kizito <Jimmy.Kizito@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2 years agodrm/amd/display: Clear encoder assignment for copied streams
Jimmy Kizito [Tue, 21 Sep 2021 14:03:05 +0000 (10:03 -0400)]
drm/amd/display: Clear encoder assignment for copied streams

[Why]
When copying a stream, the encoder assigned to it is copied too.
Encoder assignment should only happen when executing the encoder
assignment function link_encs_assign().

[How]
Clear the link encoder pointer for copied stream.

Reviewed-by: Meenakshikumar Somasundaram <Meenakshikumar.Somasundaram@amd.com>
Reviewed-by: Jun Lei <Jun.Lei@amd.com>
Acked-by: Agustin Gutierrez Sanchez <agustin.gutierrez@amd.com>
Signed-off-by: Jimmy Kizito <Jimmy.Kizito@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2 years agodrm/amd/display: Disable dpp root clock when not being used
Jake Wang [Wed, 22 Sep 2021 18:22:59 +0000 (14:22 -0400)]
drm/amd/display: Disable dpp root clock when not being used

[Why & How]
Disable root clock for dpp when not being used.

Reviewed-by: Eric Yang <eric.yang2@amd.com>
Acked-by: Agustin Gutierrez Sanchez <agustin.gutierrez@amd.com>
Signed-off-by: Jake Wang <haonan.wang2@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2 years agodrm/amd/display: Fully switch to dmub for all dcn21 asics
Roman Li [Fri, 15 Oct 2021 17:16:31 +0000 (13:16 -0400)]
drm/amd/display: Fully switch to dmub for all dcn21 asics

[Why]
On renoir usb-c port stops functioning on resume after f/w update.
New dmub firmware caused regression due to conflict with dmcu.
With new dmub f/w dmcu is superseded and should be disabled.

[How]
- Disable dmcu for all dcn21.

Check dmesg for dmub f/w version.
The old firmware (before regression):
[drm] DMUB hardware initialized: version=0x00000001
All other versions require this patch for renoir.

Bug: https://gitlab.freedesktop.org/drm/amd/-/issues/1735
Acked-by: Alex Deucher <alexander.deucher@amd.com>
Reviewed-by: Harry Wentland <harry.wentland@amd.com>
Tested-by: Mario Limonciello <mario.limonciello@amd.com>
Signed-off-by: Roman Li <Roman.Li@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2 years agodrm/amd/amdgpu: Do irq_fini_hw after ip_fini_early
YuBiao Wang [Tue, 19 Oct 2021 21:16:23 +0000 (17:16 -0400)]
drm/amd/amdgpu: Do irq_fini_hw after ip_fini_early

[Why]
drm_irq_uninstall is called in irq_fini_hw so that irq is disabled in sw
stage. SMU (and maybe other IP blocks) fini_hw will call irq_put for
cleanup and the whole cleanup process will be skipped because of
drm->irq_enable = false.

[How]
Move ip_fini_early before irq_fini_hw.

Signed-off-by: YuBiao Wang <YuBiao.Wang@amd.com>
Reviewed-by: Andrey Grodzovsky <andrey.grodzovsky@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2 years agodrm/amdkfd: map gpu hive id to xgmi connected cpu
Jonathan Kim [Thu, 14 Oct 2021 16:34:27 +0000 (12:34 -0400)]
drm/amdkfd: map gpu hive id to xgmi connected cpu

ROCr needs to be able to identify all devices that have direct access to
fine grain memory, which should include CPUs that are connected to GPUs
over xGMI. The GPU hive ID can be mapped onto the CPU hive ID since the
CPU is part of the hive.

Signed-off-by: Jonathan Kim <jonathan.kim@amd.com>
Reviewed-by: Felix Kuehling <felix.kuehling@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2 years agodrm/amdgpu: load PSP RL in resume path
Tao Zhou [Fri, 15 Oct 2021 06:29:38 +0000 (14:29 +0800)]
drm/amdgpu: load PSP RL in resume path

Some registers' access will fail without PSP RL after resume.

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>
2 years agodrm/amdkfd: Fix an inappropriate error handling in allloc memory of gpu
Lang Yu [Mon, 11 Oct 2021 05:57:25 +0000 (13:57 +0800)]
drm/amdkfd: Fix an inappropriate error handling in allloc memory of gpu

We should unreference a gem object instead of an amdgpu bo here.

Fixes: fd9a9f8801de ("drm/amdgpu: Use GEM obj reference for KFD BOs")

Signed-off-by: Lang Yu <lang.yu@amd.com>
Reviewed-by: Felix Kuehling <Felix.Kuehling@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2 years agodrm/amdgpu/psp: add some missing cases to psp_check_pmfw_centralized_cstate_management
Alex Deucher [Tue, 12 Oct 2021 15:47:56 +0000 (11:47 -0400)]
drm/amdgpu/psp: add some missing cases to psp_check_pmfw_centralized_cstate_management

Missed a few asics.

v2: update comment

Fixes: 82d05736c47b19 ("drm/amdgpu/amdgpu_psp: convert to IP version checking")
Reviewed-by: Evan Quan <evan.quan@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2 years agodrm/amdgpu/swsmu: fix is_support_sw_smu() for VEGA20
Alex Deucher [Tue, 12 Oct 2021 15:32:49 +0000 (11:32 -0400)]
drm/amdgpu/swsmu: fix is_support_sw_smu() for VEGA20

VEGA20 is 11.0.2, but it's handled by powerplay, not
swsmu.

Fixes: a8967967f6a554 ("drm/amdgpu/amdgpu_smu: convert to IP version checking")
Reviewed-by: Evan Quan <evan.quan@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2 years agodrm/amdkfd: unregistered svm range not overlap with TTM range
Philip Yang [Tue, 12 Oct 2021 03:12:25 +0000 (23:12 -0400)]
drm/amdkfd: unregistered svm range not overlap with TTM range

When creating unregistered new svm range to recover retry fault, avoid
new svm range to overlap with ranges or userptr ranges managed by TTM,
otherwise svm migration will trigger TTM or userptr eviction, to evict
user queues unexpectedly.

Change helper amdgpu_ttm_tt_affect_userptr to return userptr which is
inside the range. Add helper svm_range_check_vm_userptr to scan all
userptr of the vm, and return overlap userptr bo start, last.

Signed-off-by: Philip Yang <Philip.Yang@amd.com>
Reviewed-by: Felix Kuehling <Felix.Kuehling@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2 years agodrm/amdgpu: fix out of bounds write
Thelford Williams [Wed, 13 Oct 2021 20:04:13 +0000 (16:04 -0400)]
drm/amdgpu: fix out of bounds write

Size can be any value and is user controlled resulting in overwriting the
40 byte array wr_buf with an arbitrary length of data from buf.

Signed-off-by: Thelford Williams <tdwilliamsiv@gmail.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2 years agodrm/amdgpu/smu11: fix firmware version check for vangogh
Alex Deucher [Tue, 12 Oct 2021 15:21:24 +0000 (11:21 -0400)]
drm/amdgpu/smu11: fix firmware version check for vangogh

Was missed in the conversion to IP version checking.

Fixes: af3b89d3a639d5 ("drm/amdgpu/smu11.0: convert to IP version checking")
Reviewed-by: Evan Quan <evan.quan@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2 years agoMAINTAINERS: Add Siqueira for AMD DC
Harry Wentland [Fri, 8 Oct 2021 21:21:22 +0000 (17:21 -0400)]
MAINTAINERS: Add Siqueira for AMD DC

He's been helping maintain it for quite a while now. Make
it official.

Signed-off-by: Harry Wentland <harry.wentland@amd.com>
Acked-by: Christian König <christian.koenig@amd.com>
Acked-by: Alex Deucher <alexander.deucher@amd.com>
Acked-by: Rodrigo Siqueira <Rodrigo.Siqueira@amd.com>
Signed-off-by: Rodrigo Siqueira <Rodrigo.Siqueira@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2 years agodrm/amdkfd: fix resume error when iommu disabled in Picasso
Yifan Zhang [Mon, 11 Oct 2021 12:42:31 +0000 (20:42 +0800)]
drm/amdkfd: fix resume error when iommu disabled in Picasso

When IOMMU disabled in sbios and kfd in iommuv2 path,
IOMMU resume failure blocks system resume. Don't allow kfd to
use iommu v2 when iommu is disabled.

Reported-by: youling <youling257@gmail.com>
Tested-by: youling <youling257@gmail.com>
Signed-off-by: Yifan Zhang <yifan1.zhang@amd.com>
Reviewed-by: James Zhu <James.Zhu@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2 years agodrm/amdkfd: fix boot failure when iommu is disabled in Picasso.
Yifan Zhang [Mon, 11 Oct 2021 12:37:01 +0000 (20:37 +0800)]
drm/amdkfd: fix boot failure when iommu is disabled in Picasso.

When IOMMU disabled in sbios and kfd in iommuv2 path, iommuv2
init will fail. But this failure should not block amdgpu driver init.

Reported-by: youling <youling257@gmail.com>
Tested-by: youling <youling257@gmail.com>
Signed-off-by: Yifan Zhang <yifan1.zhang@amd.com>
Reviewed-by: James Zhu <James.Zhu@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2 years agodrm/amdkfd: handle svm partial migration cpages 0
Philip Yang [Tue, 5 Oct 2021 15:13:21 +0000 (11:13 -0400)]
drm/amdkfd: handle svm partial migration cpages 0

migrate_vma_setup may return cpages 0, means 0 page can be migrated,
treat this as error case to skip the rest of vma migration steps.

Change svm_migrate_vma_to_vram and svm_migrate_vma_to_ram to return the
number of pages migrated successfully or error code. The caller add up
all the successful migration pages and update prange->actual_loc only if
the total migrated pages is not 0.

This also removes the warning message "VRAM BO missing during
validation" if migration cpages is 0.

Signed-off-by: Philip Yang <Philip.Yang@amd.com>
Reviewed-by: Felix Kuehling <Felix.Kuehling@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2 years agodrm/amdkfd: ratelimited svm debug messages
Philip Yang [Tue, 5 Oct 2021 15:12:18 +0000 (11:12 -0400)]
drm/amdkfd: ratelimited svm debug messages

No function change, use pr_debug_ratelimited to avoid per page debug
message overflowing dmesg buf and console log.

use dev_err to show error message from unexpected situation, to provide
clue to help debug without enabling dynamic debug log. Define dev_fmt to
output function name in error message.

Signed-off-by: Philip Yang <Philip.Yang@amd.com>
Reviewed-by: Felix Kuehling <Felix.Kuehling@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2 years agodrm/amd/pm: Fix incorrect power limit readback in smu11 if POWER_SOURCE_DC
Darren Powell [Sat, 21 Aug 2021 02:08:07 +0000 (22:08 -0400)]
drm/amd/pm: Fix incorrect power limit readback in smu11 if POWER_SOURCE_DC

 when smu->adev->pm.ac_power == 0, message parameter with bit 16 set is saved
 to smu->current_power_limit.

Fixes: 0cb4c62125a9 ("drm/amd/pm: correct power limit setting for SMU V11)"

Signed-off-by: Darren Powell <darren.powell@amd.com>
Reviewed-by: Lijo Lazar <lijo.lazar@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2 years agoamdgpu/pm: (v2) add limit_type to (pptable_funcs)->set_power_limit signature
Darren Powell [Thu, 19 Aug 2021 03:31:36 +0000 (23:31 -0400)]
amdgpu/pm: (v2) add limit_type to (pptable_funcs)->set_power_limit signature

 v2
 add check for SMU_DEFAULT_PPT_LIMIT

 v1
 modify (pptable_funcs)->set_power_limit signature
 modify smu11 set_power_limit signature (arcturus, navi10, sienna_cichlid)
 modify smu13 set_power_limit signature (aldabaran)
 modify vangogh_set_power_limit signature (vangogh)

=== Test ===
sudo bash

AMDGPU_PCI_ADDR=`lspci -nn | grep "VGA\|Display" | cut -d " " -f 1`
AMDGPU_HWMON=`ls -la /sys/class/hwmon | grep $AMDGPU_PCI_ADDR | awk '{print $9}'`
HWMON_DIR=/sys/class/hwmon/${AMDGPU_HWMON}
LOGFILE=pp_show_power_cap.log

cp $LOGFILE{,.old}
lspci -nn | grep "VGA\|Display" > $LOGFILE
FILES="
power1_cap
power2_cap"

for f in $FILES
do
  if test -f "$HWMON_DIR/$f"; then
    echo === $f === >> $LOGFILE
    cat $HWMON_DIR/$f >> $LOGFILE
    RESTORE_VALUE=`cat $HWMON_DIR/$f` 2>&1  >> $LOGFILE
    echo RESTORE_VALUE $RESTORE_VALUE >> $LOGFILE
    echo 120000000 > $HWMON_DIR/$f
    sleep 3
    cat $HWMON_DIR/$f >> $LOGFILE
    echo $RESTORE_VALUE > $HWMON_DIR/$f
    sleep 3
    cat $HWMON_DIR/$f >> $LOGFILE
  else
    echo === $f === >> $LOGFILE
    echo File Not Found >> $LOGFILE
  fi
done
cat $LOGFILE

Signed-off-by: Darren Powell <darren.powell@amd.com>
Reviewed-by: Lijo Lazar <lijo.lazar@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2 years agodrm/amdgpu: Fix RAS page retirement with mode2 reset on Aldebaran
Mukul Joshi [Tue, 21 Sep 2021 00:48:23 +0000 (20:48 -0400)]
drm/amdgpu: Fix RAS page retirement with mode2 reset on Aldebaran

During mode2 reset, the GPU is temporarily removed from the
mgpu_info list. As a result, page retirement fails because it
cannot find the GPU in the GPU list.
To fix this, create our own list of GPUs that support MCE notifier
based page retirement and use that list to check if the UMC error
occurred on a GPU that supports MCE notifier based page retirement.

Signed-off-by: Mukul Joshi <mukul.joshi@amd.com>
Reviewed-by: Tao Zhou <tao.zhou1@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2 years agodrm/amdgpu: Enable RAS error injection after mode2 reset on Aldebaran
Mukul Joshi [Mon, 20 Sep 2021 19:29:23 +0000 (15:29 -0400)]
drm/amdgpu: Enable RAS error injection after mode2 reset on Aldebaran

Add the missing call to re-enable RAS error injections on the Aldebaran
mode2 reset code path.

Signed-off-by: Mukul Joshi <mukul.joshi@amd.com>
Reviewed-by: Tao Zhou <tao.zhou1@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2 years agodrm/amd/display: fix null pointer deref when plugging in display
Aurabindo Pillai [Fri, 8 Oct 2021 20:07:45 +0000 (16:07 -0400)]
drm/amd/display: fix null pointer deref when plugging in display

[Why&How]
When system boots in headless mode, connecting a 4k display creates a
null pointer dereference due to hubp for a certain plane being null.
Add a condition to check for null hubp before dereferencing it.

Signed-off-by: Aurabindo Pillai <aurabindo.pillai@amd.com>
Reviewed-by: Harry Wentland <harry.wentland@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2 years agodrm/amd/display: Fix surface optimization regression on Carrizo
Nicholas Kazlauskas [Tue, 12 Oct 2021 14:04:03 +0000 (10:04 -0400)]
drm/amd/display: Fix surface optimization regression on Carrizo

[Why]
DCE legacy optimization path isn't well tested under new DC optimization
flow which can result in underflow occuring when initializing X11 on
Carrizo.

[How]
Retain the legacy optimization flow for DCE and keep the new one for DCN
to satisfy optimizations being correctly applied for ASIC that can
support it.

Fixes: 34316c1e561db0 ("drm/amd/display: Optimize bandwidth on following fast update")
Reported-by: Tom St Denis <tom.stdenis@amd.com>
Tested-by: Tom St Denis <tom.stdenis@amd.com>
Reviewed-by: Harry Wentland <harry.wentland@amd.com>
Signed-off-by: Nicholas Kazlauskas <nicholas.kazlauskas@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2 years agodrm/amd/display: Enable PSR by default on newer DCN
Nicholas Kazlauskas [Tue, 5 Oct 2021 14:55:57 +0000 (10:55 -0400)]
drm/amd/display: Enable PSR by default on newer DCN

[Why]
For optimal power savings on panels that can support it.

This was previously left disabled by default because of issues with
compositors that do not pageflip and scan out directly to the
frontbuffer.

For these compositors we now have detection methods that wait for x
number of pageflips after a full update - triggered by a buffer or
format change typically.

This may introduce bugs or new cases not tested by users so this is
only currently targeting newer DCN.

[How]
Add code in DM to set PSR state by default for newer DCN while falling
back to the feature mask for older.

Add a global debug flag that can be set to disable it for either.

Signed-off-by: Nicholas Kazlauskas <nicholas.kazlauskas@amd.com>
Reviewed-by: Harry Wentland <harry.wentland@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2 years agodrm/amdgpu: enable display for cyan skillfish
Lang Yu [Tue, 12 Oct 2021 06:10:46 +0000 (14:10 +0800)]
drm/amdgpu: enable display for cyan skillfish

Display support for cyan skillfish is ready now. Enable it!

Signed-off-by: Lang Yu <lang.yu@amd.com>
Reviewed-by: Huang Rui <ray.huang@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2 years agoamd/display: check cursor plane matches underlying plane
Simon Ser [Wed, 6 Oct 2021 14:06:05 +0000 (14:06 +0000)]
amd/display: check cursor plane matches underlying plane

The current logic checks whether the cursor plane blending
properties match the primary plane's. However that's wrong,
because the cursor is painted on all planes underneath. If
the cursor is over the primary plane and the overlay plane,
it's painted on both pipes.

Iterate over the CRTC planes and check their scaling match
the cursor's.

v4: fix typo in commit message (Harry)

Signed-off-by: Simon Ser <contact@emersion.fr>
Cc: Alex Deucher <alexander.deucher@amd.com>
Cc: Harry Wentland <hwentlan@amd.com>
Cc: Nicholas Kazlauskas <nicholas.kazlauskas@amd.com>
Cc: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Cc: Rodrigo Siqueira <Rodrigo.Siqueira@amd.com>
Cc: Sean Paul <seanpaul@chromium.org>
Reviewed-by: Harry Wentland <harry.wentland@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2 years agoamd/amdkfd: remove svms declaration to avoid werror
Alex Sierra [Thu, 30 Sep 2021 15:38:08 +0000 (10:38 -0500)]
amd/amdkfd: remove svms declaration to avoid werror

svm_range_list svms declaration removed to avoid werror when
CONFIG_HSA_AMD_SVM is not enabled.

Signed-off-by: Alex Sierra <alex.sierra@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2 years agodrm/amdkfd: fix KFDSVMRangeTest.PartialUnmapSysMemTest fails
Yifan Zhang [Sat, 14 Aug 2021 09:53:14 +0000 (17:53 +0800)]
drm/amdkfd: fix KFDSVMRangeTest.PartialUnmapSysMemTest fails

[ RUN      ] KFDSVMRangeTest.PartialUnmapSysMemTest
/home/yifan/brahma/libhsakmt/tests/kfdtest/src/KFDTestUtil.cpp:245: Failure
Value of: (hsaKmtAllocMemory(m_Node, m_Size, m_Flags, &m_pBuf))
  Actual: 1
Expected: HSAKMT_STATUS_SUCCESS
Which is: 0
/home/yifan/brahma/libhsakmt/tests/kfdtest/src/KFDTestUtil.cpp:248: Failure
Value of: (hsaKmtMapMemoryToGPUNodes(m_pBuf, m_Size, __null, mapFlags, 1, &m_Node))
  Actual: 1
Expected: HSAKMT_STATUS_SUCCESS
Which is: 0
/home/yifan/brahma/libhsakmt/tests/kfdtest/src/KFDTestUtil.cpp:306: Failure
Expected: ((void *)__null) != (ptr), actual: NULL vs NULL
Segmentation fault (core dumped)
[          ] Profile: Full Test
[          ] HW capabilities: 0x9

kernel log:

[  102.029150]  ret_from_fork+0x22/0x30
[  102.029158] ---[ end trace 15c34e782714f9a3 ]---
[ 3613.603598] amdgpu: Address: 0x7f7149ccc000 already allocated by SVM
[ 3613.610620] show_signal_msg: 27 callbacks suppressed

These is race with deferred actions from previous memory map
changes (e.g. munmap).Flush pending deffered work to avoid such case.

Signed-off-by: Yifan Zhang <yifan1.zhang@amd.com>
Reviewed-by: Felix Kuehling <Felix.Kuehling@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2 years agodrm/amdkfd: export svm_range_list_lock_and_flush_work
Yifan Zhang [Tue, 17 Aug 2021 08:47:26 +0000 (16:47 +0800)]
drm/amdkfd: export svm_range_list_lock_and_flush_work

export svm_range_list_lock_and_flush_work to make other kfd parts be
able to sync svm_range_list.

Signed-off-by: Yifan Zhang <yifan1.zhang@amd.com>
Reviewed-by: Felix Kuehling <Felix.Kuehling@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2 years agodrm/amdkfd: avoid conflicting address mappings
Alex Sierra [Mon, 7 Jun 2021 18:14:50 +0000 (13:14 -0500)]
drm/amdkfd: avoid conflicting address mappings

[Why]
Avoid conflict with address ranges mapped by SVM
mechanism that try to be allocated again through
ioctl_alloc in the same process. And viceversa.

[How]
For ioctl_alloc_memory_of_gpu allocations
Check if the address range passed into ioctl memory
alloc does not exist already in the kfd_process
svms->objects interval tree.

For SVM allocations
Look for the address range into the interval tree VA from
the VM inside of each pdds used in a kfd_process.

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>
2 years agodrm/amdgpu/nbio2.3: don't use GPU_HDP_FLUSH bit 12
Alex Deucher [Fri, 8 Oct 2021 14:23:38 +0000 (10:23 -0400)]
drm/amdgpu/nbio2.3: don't use GPU_HDP_FLUSH bit 12

It's used internally by firmware.  Using it in the driver
could conflict with firmware.

v2: squash in fix for navi1x (Alex)

Reviewed-by: James Zhu <James.Zhu@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2 years agodrm/amdgpu/nbio7.4: don't use GPU_HDP_FLUSH bit 12
Alex Deucher [Fri, 8 Oct 2021 14:21:42 +0000 (10:21 -0400)]
drm/amdgpu/nbio7.4: don't use GPU_HDP_FLUSH bit 12

It's used internally by firmware.  Using it in the driver
could conflict with firmware.

Reviewed-by: James Zhu <James.Zhu@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2 years agodrm/amdgpu: query default sclk from smu for cyan_skillfish
Lang Yu [Mon, 11 Oct 2021 08:27:04 +0000 (16:27 +0800)]
drm/amdgpu: query default sclk from smu for cyan_skillfish

Query default sclk instead of hard code.

Signed-off-by: Lang Yu <lang.yu@amd.com>
Acked-by: Huang Rui <ray.huang@amd.com>
Reviewed-by: Lijo Lazar <lijo.lazar@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2 years agoMerge tag 'drm-intel-gt-next-2021-10-08' of git://anongit.freedesktop.org/drm/drm...
Dave Airlie [Mon, 11 Oct 2021 08:09:39 +0000 (18:09 +1000)]
Merge tag 'drm-intel-gt-next-2021-10-08' of git://anongit.freedesktop.org/drm/drm-intel into drm-next

UAPI Changes:

- Add uAPI for using PXP protected objects

  Mesa changes: https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8064

- Add PCI IDs and LMEM discovery/placement uAPI for DG1

  Mesa changes: https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/11584

- Disable engine bonding on Gen12+ except TGL, RKL and ADL-S

Cross-subsystem Changes:

- Merges 'tip/locking/wwmutex' branch (core kernel tip)
- "mei: pxp: export pavp client to me client bus"

Core Changes:

- Update ttm_move_memcpy for async use (Thomas)

Driver Changes:

- Enable GuC submission by default on DG1 (Matt B)
- Add PXP (Protected Xe Path) support for Gen12 integrated (Daniele,
  Sean, Anshuman)
  See "drm/i915/pxp: add PXP documentation" for details!
- Remove force_probe protection for ADL-S (Raviteja)
- Add base support for XeHP/XeHP SDV (Matt R, Stuart, Lucas)
- Handle DRI_PRIME=1 on Intel igfx + Intel dgfx hybrid graphics setup (Tvrtko)
- Use Transparent Hugepages when IOMMU is enabled (Tvrtko, Chris)
- Implement LMEM backup and restore for suspend / resume (Thomas)
- Report INSTDONE_GEOM values in error state for DG2 (Matt R)
- Add DG2-specific shadow register table (Matt R)
- Update Gen11/Gen12/XeHP shadow register tables (Matt R)
- Maintain backward-compatible nested batch behavior on TGL+ (Matt R)
- Add new LRI reg offsets for DG2 (Akeem)
- Initialize unused MOCS entries to device specific values (Ayaz)
- Track and use the correct UC MOCS index on Gen12 (Ayaz)
- Add separate MOCS table for Gen12 devices other than TGL/RKL (Ayaz)
- Simplify the locking and eliminate some RCU usage (Daniel)
- Add some flushing for the 64K GTT path (Matt A)
- Mark GPU wedging on driver unregister unrecoverable (Janusz)

- Major rework in the GuC codebase, simplify locking and add docs (Matt B)
- Add DG1 GuC/HuC firmwares (Daniele, Matt B)
- Remember to call i915_sw_fence_fini on guc_state.blocked (Matt A)
- Use "gt" forcewake domain name for error messages instead of "blitter" (Matt R)
- Drop now duplicate LMEM uAPI RFC kerneldoc section (Daniel)
- Fix early tracepoints for requests (Matt A)
- Use locked access to ctx->engines in set_priority (Daniel)
- Convert gen6/gen7/gen8 read operations to fwtable (Matt R)
- Drop gen11/gen12 specific mmio write handlers (Matt R)
- Drop gen11 specific mmio read handlers (Matt R)
- Use designated initializers for init/exit table (Kees)
- Fix syncmap memory leak (Matt B)
- Add pretty printing for buddy allocator state debug (Matt A)
- Fix potential error pointer dereference in pinned_context() (Dan)
- Remove IS_ACTIVE macro (Lucas)
- Static code checker fixes (Nathan)
- Clean up disabled warnings (Nathan)
- Increase timeout in i915_gem_contexts selftests 5x for GuC submission (Matt B)
- Ensure wa_init_finish() is called for ctx workaround list (Matt R)
- Initialize L3CC table in mocs init (Sreedhar, Ayaz, Ram)
- Get PM ref before accessing HW register (Vinay)
- Move __i915_gem_free_object to ttm_bo_destroy (Maarten)
- Deduplicate frequency dump on debugfs (Lucas)
- Make wa list per-gt (Venkata)
- Do not define dummy vma in stack (Venkata)
- Take pinning into account in __i915_gem_object_is_lmem (Matt B, Thomas)
- Do not report currently active engine when describing objects (Tvrtko)
- Fix pdfdocs build error by removing nested grid from GuC docs (Akira)
- Remove false warning from the rps worker (Tejas)
- Flush buffer pools on driver remove (Janusz)
- Fix runtime pm handling in i915_gem_shrink (Maarten)
- Rework TTM object initialization slightly (Thomas)
- Use fixed offset for PTEs location (Michal Wa)
- Verify result from CTB (de)register action and improve error messages (Michal Wa)
- Fix bug in user proto-context creation that leaked contexts (Matt B)

- Re-use Gen11 forcewake read functions on Gen12 (Matt R)
- Make shadow tables range-based (Matt R)
- Ditch the i915_gem_ww_ctx loop member (Thomas, Maarten)
- Use NULL instead of 0 where appropriate (Ville)
- Rename pci/debugfs functions to respect file prefix (Jani, Lucas)
- Drop guc_communication_enabled (Daniele)
- Selftest fixes (Thomas, Daniel, Matt A, Maarten)
- Clean up inconsistent indenting (Colin)
- Use direction definition DMA_BIDIRECTIONAL instead of
  PCI_DMA_BIDIRECTIONAL (Cai)
- Add "intel_" as prefix in set_mocs_index() (Ayaz)

From: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/YWAO80MB2eyToYoy@jlahtine-mobl.ger.corp.intel.com
Signed-off-by: Dave Airlie <airlied@redhat.com>
2 years agoMerge tag 'drm-intel-next-2021-10-04' of git://anongit.freedesktop.org/drm/drm-intel...
Dave Airlie [Mon, 11 Oct 2021 06:53:56 +0000 (16:53 +1000)]
Merge tag 'drm-intel-next-2021-10-04' of git://anongit.freedesktop.org/drm/drm-intel into drm-next

Cross-subsystem Changes:
- fbdev/efifb: Release PCI device's runtime PM ref during FB destr\
oy (Imre)

i915 Core Driver Changes:
- Only access SFC_DONE in media when not fused off for graphics 12 and newer.
- Double Memory latency values from pcode for DG2 (Matt Roper)
- ADL-S PCI ID update (Tejas)
- New DG1 PCI ID (Jose)
- Fix regression with uncore refactoring (Dave)

i915 Display Changes:
- ADL-P display (XE_LPD) fixes and updates (Ankit, Jani, Matt Roper, Anusham, Jose, Imre, Vandita)
- DG2 display fixes (Ankit, Jani)
- Expand PCH_CNP tweaked display workaround to all newer displays (Anshuman)
- General display simplifications and clean-ups (Jani, Swati, Jose, Ville)
- PSR Clean-ups, dropping support for BDW/HSD and enable PSR2 selective fetch by default (Jose, Gwan-gyeong)
- Nuke ORIGIN_GTT (Jose)
- Return proper DPRX link training result (Lee)
- FBC related refactor and fixes (Ville)
- Yet another attempt to solve the fast+narrow vs slow+wide eDP link training (Kai-Heng)
- DP 2.0 preparation work (Jani)
- Silence __iomem sparse warn (Ville)
- Clean up DPLL stuff (Ville)
- Fix various dp/edp max rates (Matt Atwood, Animesh, Jani)
- Remove VBT ddi_port_info caching (Jani)
- DSI driver improvements (Lee)
- HDCP fixes (Juston)
- Associate ACPI connector nodes with connector entries (Heikki)
- Add support for out-of-bound hotplug events (Hans)
- VESA vendor block and drm/i915 MSO use of it (Jani)
- Fixes for bigjoiner (Ville)
- Update memory bandwidth parameters (RK)
- DMC related fixes (Chris, Jose)
- HDR related fixes and improvements (Tejas)
- g4x/vlv/chv CxSR/wm fixes/cleanups (Ville)
- Use BIOS provided value for RKL Audio's HDA link (Kai-Heng)
- Fix the dsc check while selecting min_cdclk (Vandita)
- Split and constify vtable (Dave)
- Add ww context to intel_dpt_pin (Maarten)
- Fix bdb version check (Lukasz)
- DP per-lane drive settings prep work and other DP fixes (Ville)

Signed-off-by: Dave Airlie <airlied@redhat.com>
# gpg: Signature made Tue 05 Oct 2021 04:58:16 AEST
# gpg:                using RSA key 6D207068EEDD65091C2CE2A3FA625F640EEB13CA
# gpg: Good signature from "Rodrigo Vivi <rodrigo.vivi@intel.com>" [unknown]
# gpg:                 aka "Rodrigo Vivi <rodrigo.vivi@gmail.com>" [unknown]
# gpg: WARNING: This key is not certified with a trusted signature!
# gpg:          There is no indication that the signature belongs to the owner.
# Primary key fingerprint: 6D20 7068 EEDD 6509 1C2C  E2A3 FA62 5F64 0EEB 13CA
From: Rodrigo Vivi <rodrigo.vivi@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/YVtPk6llsxBFiw7W@intel.com
2 years agoMerge tag 'drm-misc-next-2021-10-06' of git://anongit.freedesktop.org/drm/drm-misc...
Dave Airlie [Mon, 11 Oct 2021 02:39:14 +0000 (12:39 +1000)]
Merge tag 'drm-misc-next-2021-10-06' of git://anongit.freedesktop.org/drm/drm-misc into drm-next

drm-misc-next for v5.16:

UAPI Changes:
- Allow empty drm leases for creating separate GEM namespaces.

Cross-subsystem Changes:
- Slightly rework dma_buf_poll.
- Add dma_resv_for_each_fence_unlocked to iterate, and use it inside
  the lockless dma-resv functions.

Core Changes:
- Allow devm_drm_of_get_bridge to build without CONFIG_OF for compile testing.
- Add more DP2 headers.
- fix CONFIG_FB dependency in fb_helper.
- Add DRM_FORMAT_R8 to drm_format_info, and helpers for RGB332 and RGB888.
- Fix crash on a 0 or invalid EDID.

Driver Changes:
- Apply and revert DRM_MODESET_LOCK_ALL_BEGIN.
- Add mode_valid to ti-sn65dsi86 bridge.
- Support multiple syncobjs in v3d.
- Add R8, RGB332 and RGB888 pixel formats to GUD.
- Use devm_add_action_or_reset in dw-hdmi-cec.

Signed-off-by: Dave Airlie <airlied@redhat.com>
# gpg: Signature made Wed 06 Oct 2021 20:48:12 AEST
# gpg:                using RSA key B97BD6A80CAC4981091AE547FE558C72A67013C3
# gpg: Good signature from "Maarten Lankhorst <maarten.lankhorst@linux.intel.com>" [expired]
# gpg:                 aka "Maarten Lankhorst <maarten@debian.org>" [expired]
# gpg:                 aka "Maarten Lankhorst <maarten.lankhorst@canonical.com>" [expired]
# gpg: Note: This key has expired!
# Primary key fingerprint: B97B D6A8 0CAC 4981 091A  E547 FE55 8C72 A670 13C3
From: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/2602f4e9-a8ac-83f8-6c2a-39fd9ca2e1ba@linux.intel.com
2 years agoMerge tag 'amd-drm-next-5.16-2021-10-08' of https://gitlab.freedesktop.org/agd5f...
Dave Airlie [Mon, 11 Oct 2021 01:05:18 +0000 (11:05 +1000)]
Merge tag 'amd-drm-next-5.16-2021-10-08' of https://gitlab.freedesktop.org/agd5f/linux into drm-next

amd-drm-next-5.16-2021-10-08:

amdgpu:
- gart BO pin count fix
- RAS fixes
- Misc display fixes
- Misc code cleanups
- Validate IP discovery table
- IOMMU handling fixes for hotplug
- Cyan Skillfish display support
- DP 2.0 fixes
- Covert vega and navi to IP discovery based asic enumeration
- JPEG fixes
- More FP cleanup for display
- DCC fixes for DCN3.x
- Initial USB4 DP tunnelling support
- Aldebaran MCE support

amdkfd:
- Misc bug fixes
- Misc code cleanups
- RAS fixes

x86/MCE:
- Export symbol for use by GPU driver

Signed-off-by: Dave Airlie <airlied@redhat.com>
From: Alex Deucher <alexander.deucher@amd.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20211008200345.4689-1-alexander.deucher@amd.com
2 years agodrm/amd/display: fix duplicated inclusion
kernel test robot [Fri, 8 Oct 2021 14:40:34 +0000 (10:40 -0400)]
drm/amd/display: fix duplicated inclusion

Generated by: scripts/checkincludes.pl

Reported-by: kernel test robot <lkp@intel.com>
Signed-off-by: kernel test robot <lkp@intel.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2 years agodrm/amd/display: remove duplicate include in dcn201_clk_mgr.c
Lv Ruyi [Fri, 8 Oct 2021 08:40:19 +0000 (08:40 +0000)]
drm/amd/display: remove duplicate include in dcn201_clk_mgr.c

Remove all but the first include of reg_helper.h and core_types.h
from dcn201_clk_mgr.c.

Reported-by: Zeal Robot <zealci@zte.com.cn>
Signed-off-by: Lv Ruyi <lv.ruyi@zte.com.cn>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2 years agodrm/amdgpu: use adev_to_drm for consistency when accessing drm_device
Guchun Chen [Fri, 8 Oct 2021 17:21:45 +0000 (13:21 -0400)]
drm/amdgpu: use adev_to_drm for consistency when accessing drm_device

adev_to_drm is used everywhere, so improve recent changes
when accessing drm_device pointer from amdgpu_device.

Signed-off-by: Guchun Chen <guchun.chen@amd.com>
Acked-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2 years agodrm/amdkfd: rm BO resv on validation to avoid deadlock
Alex Sierra [Thu, 7 Oct 2021 17:04:09 +0000 (12:04 -0500)]
drm/amdkfd: rm BO resv on validation to avoid deadlock

This fix the deadlock with the BO reservations during SVM_BO evictions
while allocations in VRAM are concurrently performed. More specific,
while the ttm waits for the fence to be signaled (ttm_bo_wait), it
already has the BO reserved. In parallel, the restore worker might be
running, prefetching memory to VRAM. This also requires to reserve the
BO, but blocks the mmap semaphore first. The deadlock happens when the
SVM_BO eviction worker kicks in and waits for the mmap semaphore held
in restore worker. Preventing signal the fence back, causing the
deadlock until the ttm times out.

We don't need to hold the BO reservation anymore during validation
and mapping. Now the physical addresses are taken from hmm_range_fault.
We also take migrate_mutex to prevent range migration while
validate_and_map update GPU page table.

Signed-off-by: Alex Sierra <alex.sierra@amd.com>
Signed-off-by: Felix Kuehling <Felix.Kuehling@amd.com>
Reviewed-by: Philip Yang <philip.yang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2 years agodrm/amd/display: Fix Werror when building
Wayne Lin [Fri, 8 Oct 2021 07:08:37 +0000 (15:08 +0800)]
drm/amd/display: Fix Werror when building

[Why & How]
Got Werror when building with Clang-13:

drivers/gpu/drm/amd/amdgpu/../display/dc/core/dc_link_dpia.c:195:2:
error: variable 'ts' is used uninitialized whenever switch default is
taken [-Werror,-Wsometimes-uninitialized]

        default:

        ^~~~~~~

drivers/gpu/drm/amd/amdgpu/../display/dc/core/dc_link_dpia.c:200:9:
note: uninitialized use occurs here

        return ts;

               ^~

drivers/gpu/drm/amd/amdgpu/../display/dc/core/dc_link_dpia.c:180:2:
note: variable 'ts' is declared here

        enum dpia_set_config_ts ts;

        ^

1 error generated.

Fix it.

Reported-by: Mike Lothian <mike@fireburn.co.uk>
Signed-off-by: Wayne Lin <Wayne.Lin@amd.com>
Acked-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2 years agodrm/amdgpu: add missing case for HDP for renoir
Alex Deucher [Fri, 1 Oct 2021 17:20:34 +0000 (13:20 -0400)]
drm/amdgpu: add missing case for HDP for renoir

Missing 4.1.2.

Reviewed-by: Rodrigo Siqueira <Rodrigo.Siqueira@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2 years agodrm/amd/display: Remove redundant initialization of variable result
Colin Ian King [Thu, 7 Oct 2021 12:06:37 +0000 (13:06 +0100)]
drm/amd/display: Remove redundant initialization of variable result

The variable result is being initialized with a value that is never
read, it is being updated immediately afterwards in both branches
of an if statement. The assignment is redundant and can be removed.

Addresses-Coverity: ("Unused value")
Signed-off-by: Colin Ian King <colin.king@canonical.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2 years agodrm/amdgpu/discovery: add missing case for SMU 11.0.5
Alex Deucher [Thu, 7 Oct 2021 14:02:13 +0000 (10:02 -0400)]
drm/amdgpu/discovery: add missing case for SMU 11.0.5

Was missed when converting the driver over to IP based
initialization.

Tested-by: Harry Wentland <harry.wentland@amd.com>
Reviewed-by: Guchun Chen <guchun.chen@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2 years agoMerge tag 'du-next-20211007' of git://linuxtv.org/pinchartl/media into drm-next
Dave Airlie [Fri, 8 Oct 2021 02:26:45 +0000 (12:26 +1000)]
Merge tag 'du-next-20211007' of git://linuxtv.org/pinchartl/media into drm-next

R-Car DU improvements & enhancements to misc drivers

Most notably,

- Non-contiguous buffer import support for rcar-du
- r8a779a0 support preparation for rcar-du
- COMPILE_TEST fixes for omapdrm and sti

Signed-off-by: Dave Airlie <airlied@redhat.com>
From: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Link: https://patchwork.freedesktop.org/patch/msgid/YV5jfi+/qjTJKeb3@pendragon.ideasonboard.com
2 years agodrm/i915: remove IS_ACTIVE
Lucas De Marchi [Tue, 5 Oct 2021 17:17:28 +0000 (10:17 -0700)]
drm/i915: remove IS_ACTIVE

When trying to bring IS_ACTIVE to linux/kconfig.h I thought it wouldn't
provide much value just encapsulating it in a boolean context. So I also
added the support for handling undefined macros as the IS_ENABLED()
counterpart. However the feedback received from Masahiro Yamada was that
it is too ugly, not providing much value. And just wrapping in a boolean
context is too dumb - we could simply open code it.

As detailed in commit babaab2f4738 ("drm/i915: Encapsulate kconfig
constant values inside boolean predicates"), the IS_ACTIVE macro was
added to workaround a compilation warning. However after checking again
our current uses of IS_ACTIVE it turned out there is only
1 case in which it triggers a warning in clang (due
-Wconstant-logical-operand) and 2 in smatch. All the others
can simply use the shorter version, without wrapping it in any macro.

So here I'm dialing all the way back to simply removing the macro. That
single case hit by clang can be changed to make the constant come first,
so it doesn't think it's mask:

-       if (context && CONFIG_DRM_I915_FENCE_TIMEOUT)
+       if (CONFIG_DRM_I915_FENCE_TIMEOUT && context)

As talked with Dan Carpenter, that logic will be added in smatch as
well, so it will also stop warning about it.

Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
Reviewed-by: Jani Nikula <jani.nikula@intel.com>
Reviewed-by: Masahiro Yamada <masahiroy@kernel.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20211005171728.3147094-1-lucas.demarchi@intel.com
2 years agodrm/amdgpu: unify BO evicting method in amdgpu_ttm
Nirmoy Das [Wed, 6 Oct 2021 15:55:00 +0000 (17:55 +0200)]
drm/amdgpu: unify BO evicting method in amdgpu_ttm

Unify BO evicting functionality for possible memory
types in amdgpu_ttm.c.

Signed-off-by: Nirmoy Das <nirmoy.das@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2 years agodrm: rcar-du: Add r8a779a0 device support
Kieran Bingham [Wed, 22 Sep 2021 23:47:26 +0000 (00:47 +0100)]
drm: rcar-du: Add r8a779a0 device support

Extend the rcar_du_device_info structure and rcar_du_output enum to
support DSI outputs and utilise these additions to provide support for
the R8A779A0 V3U platform.

Signed-off-by: Kieran Bingham <kieran.bingham+renesas@ideasonboard.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: Laurent Pinchart <laurent.pinchart+renesas@ideasonboard.com>
2 years agodrm: rcar-du: Split CRTC IRQ and Clock features
Kieran Bingham [Wed, 22 Sep 2021 23:47:25 +0000 (00:47 +0100)]
drm: rcar-du: Split CRTC IRQ and Clock features

Not all platforms require both per-crtc IRQ and per-crtc clock
management. In preparation for suppporting such platforms, split the
feature macro to be able to specify both features independently.

The other features are incremented accordingly, to keep the two crtc
features adjacent.

Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
Signed-off-by: Laurent Pinchart <laurent.pinchart+renesas@ideasonboard.com>
2 years agodrm: rcar-du: Fix DIDSR field name
Kieran Bingham [Wed, 22 Sep 2021 23:47:24 +0000 (00:47 +0100)]
drm: rcar-du: Fix DIDSR field name

The DIDSR fields named LDCS were incorrectly defined as LCDS.
Both the Gen2 and Gen3 documentation refer to the fields as the "LVDS
Dot Clock Select".

Correct the definitions.

Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
Signed-off-by: Laurent Pinchart <laurent.pinchart+renesas@ideasonboard.com>
2 years agodrm: rcar-du: Only initialise TVM_TVSYNC mode when supported
Kieran Bingham [Wed, 22 Sep 2021 23:47:23 +0000 (00:47 +0100)]
drm: rcar-du: Only initialise TVM_TVSYNC mode when supported

The R-Car DU as found on the D3, E3, and V3U do not have support
for an external synchronisation method.

In these cases, the dsysr cached register should not be initialised
in DSYSR_TVM_TVSYNC, but instead should be left clear to configure as
DSYSR_TVM_MASTER by default.

Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
Signed-off-by: Laurent Pinchart <laurent.pinchart+renesas@ideasonboard.com>
2 years agodrm: rcar-du: Sort the DU outputs
Kieran Bingham [Wed, 22 Sep 2021 23:47:22 +0000 (00:47 +0100)]
drm: rcar-du: Sort the DU outputs

Sort the DU outputs alphabetically, with the exception of the final
entry which is there as a sentinal.

Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: Kieran Bingham <kieran.bingham+renesas@ideasonboard.com>
Signed-off-by: Laurent Pinchart <laurent.pinchart+renesas@ideasonboard.com>
2 years agodt-bindings: display: renesas,du: Provide bindings for r8a779a0
Kieran Bingham [Thu, 23 Sep 2021 13:01:38 +0000 (14:01 +0100)]
dt-bindings: display: renesas,du: Provide bindings for r8a779a0

Extend the Renesas DU display bindings to support the r8a779a0 V3U.

Signed-off-by: Kieran Bingham <kieran.bingham+renesas@ideasonboard.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Geert Uytterhoeven <geert+renesas@glider.be>
Reviewed-by: Rob Herring <robh@kernel.org>
Signed-off-by: Laurent Pinchart <laurent.pinchart+renesas@ideasonboard.com>
2 years agodrm: rcar-du: Make use of the helper function devm_platform_ioremap_resource()
Cai Huoqing [Tue, 31 Aug 2021 07:54:42 +0000 (15:54 +0800)]
drm: rcar-du: Make use of the helper function devm_platform_ioremap_resource()

Use the devm_platform_ioremap_resource() helper instead of
calling platform_get_resource() and devm_ioremap_resource()
separately

Signed-off-by: Cai Huoqing <caihuoqing@baidu.com>
Reviewed-by: Kieran Bingham <kieran.bingham+renesas@ideasonboard.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: Laurent Pinchart <laurent.pinchart+renesas@ideasonboard.com>
2 years agodrm/shmobile: Make use of the helper function devm_platform_ioremap_resource()
Cai Huoqing [Tue, 31 Aug 2021 13:57:30 +0000 (21:57 +0800)]
drm/shmobile: Make use of the helper function devm_platform_ioremap_resource()

Use the devm_platform_ioremap_resource() helper instead of
calling platform_get_resource() and devm_ioremap_resource()
separately

Signed-off-by: Cai Huoqing <caihuoqing@baidu.com>
Reviewed-by: Kieran Bingham <kieran.bingham+renesas@ideasonboard.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: Laurent Pinchart <laurent.pinchart+renesas@ideasonboard.com>
2 years agodrm/sti: Use correct printk format specifiers for size_t
Laurent Pinchart [Wed, 28 Jul 2021 15:16:15 +0000 (18:16 +0300)]
drm/sti: Use correct printk format specifiers for size_t

The correct format specifier for size_t is %zu. Using %d (or %u)
generates a warning on 64-bit platforms. Fix it.

Signed-off-by: Laurent Pinchart <laurent.pinchart+renesas@ideasonboard.com>
Reviewed-by: Philippe Cornu <philippe.cornu@foss.st.com>
2 years agodrm/omap: Depend on CONFIG_OF
Laurent Pinchart [Sat, 31 Jul 2021 16:24:02 +0000 (19:24 +0300)]
drm/omap: Depend on CONFIG_OF

The driver accesses the drm_bridge.of_node field, which is present only
if CONFIG_OF is enabled. As all platforms using omapdrm are OF-based, we
can simply depend on CONFIG_OF.

Signed-off-by: Laurent Pinchart <laurent.pinchart+renesas@ideasonboard.com>
Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>
2 years agodrm/omap: Cast pointer to integer without generating warning
Laurent Pinchart [Wed, 28 Jul 2021 15:19:34 +0000 (18:19 +0300)]
drm/omap: Cast pointer to integer without generating warning

On 64-bit platforms, the compiler complains that casting a void pointer
to an unsigned int loses data. Cast the pointer to a uintptr_t to fix
this.

Signed-off-by: Laurent Pinchart <laurent.pinchart+renesas@ideasonboard.com>
Reviewed-by: Tomi Valkeinen <tomi.valkeinen@ideasonboard.com>
2 years agodrm/omap: Use correct printk format specifiers for size_t
Laurent Pinchart [Wed, 28 Jul 2021 15:16:15 +0000 (18:16 +0300)]
drm/omap: Use correct printk format specifiers for size_t

The correct format specifier for size_t is %zu. Using %d (or %u)
generates a warning on 64-bit platforms. Fix it.

Signed-off-by: Laurent Pinchart <laurent.pinchart+renesas@ideasonboard.com>
Reviewed-by: Tomi Valkeinen <tomi.valkeinen@ideasonboard.com>
2 years agodrm: property: Replace strncpy() with strscpy_pad()
Laurent Pinchart [Fri, 30 Jul 2021 23:29:43 +0000 (02:29 +0300)]
drm: property: Replace strncpy() with strscpy_pad()

strncpy() is widely regarded as unsafe due to the fact that it may leave
the destination string without a nul-termination when the source string
size is too large. When compiling the kernel with W=1, the gcc warns
about this:

drivers/gpu/drm/drm_property.c: In function ‘drm_property_create’:
drivers/gpu/drm/drm_property.c:130:2: warning: ‘strncpy’ specified bound 32 equals destination size [-Wstringop-truncation]
  130 |  strncpy(property->name, name, DRM_PROP_NAME_LEN);
      |  ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

There are three occurrences of strncpy() in drm_property.c. None of them
are actually unsafe, as the very next line forces nul-termination of the
destination buffer. The warning is thus a false positive, but adds noise
to the kernel log. It can easily be silenced by using strscpy_pad()
instead. Do so.

One of the three occurrences, in drm_property_add_enum(), fills a char
array that is later copied to userspace with copy_to_user() in
drm_mode_getproperty_ioctl(). To avoid leaking kernel data,
strscpy_pad() is required. Similarly, a second occurrence, in
drm_mode_getproperty_ioctl(), copies the string to an ioctl data buffer
that isn't previously zero'ed, to strscpy_pad() is also required. The
last occurrence, in drm_property_create(), would be safe to replace with
strscpy(), as the destination buffer is copied to userspace with
strscpy_pad(). However, given that this isn't in a hot path, let's avoid
future data leaks in case someone copies the whole char array blindly.

Signed-off-by: Laurent Pinchart <laurent.pinchart+renesas@ideasonboard.com>
Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>
2 years agodrm: rcar-du: Allow importing non-contiguous dma-buf with VSP
Laurent Pinchart [Mon, 13 Nov 2017 02:40:30 +0000 (04:40 +0200)]
drm: rcar-du: Allow importing non-contiguous dma-buf with VSP

On R-Car Gen3, the DU uses a separate IP core named VSP to perform DMA
from memory and composition of planes. The DU hardware then only handles
the video timings and the interface with the encoders. This differs from
Gen2, where the DU included a composer with DMA engines.

When sourcing from the VSP, the DU hardware performs no memory access,
and thus has no requirements on imported dma-buf memory types. The GEM
CMA helpers however still create a DMA mapping to the DU device, which
isn't used. The mapping to the VSP is done when processing the atomic
commits, in the plane .prepare_fb() handler.

When the system uses an IOMMU, the VSP device is attached to it, which
enables the VSP to use non physically contiguous memory. The DU, as it
performs no memory access, isn't connected to the IOMMU. The GEM CMA
drm_gem_cma_prime_import_sg_table() helper will in that case fail to map
non-contiguous imported dma-bufs, as the DMA mapping to the DU device
will have multiple entries in its sgtable. The prevents using non
physically contiguous memory for display.

The DRM PRIME and GEM CMA helpers are designed to create the sgtable
when the dma-buf is imported. By default, the device referenced by the
drm_device is used to create the dma-buf attachment. Drivers can use a
different device by using the drm_gem_prime_import_dev() function. While
the DU has access to the VSP device, this won't help here, as different
CRTCs use different VSP instances, connected to different IOMMU
channels. The driver doesn't know at import time which CRTC a GEM object
will be used, and thus can't select the right VSP device to pass to
drm_gem_prime_import_dev().

To support non-contiguous memory, implement a custom
.gem_prime_import_sg_table() operation that accepts all imported dma-buf
regardless of the number of scatterlist entries. The sgtable will be
mapped to the VSP at .prepare_fb() time, which will reject the
framebuffer if the VSP isn't connected to an IOMMU.

Signed-off-by: Laurent Pinchart <laurent.pinchart+renesas@ideasonboard.com>
Reviewed-by: Kieran Bingham <kieran.bingham+renesas@ideasonboard.com>
2 years agodrm: rcar-du: Set the DMA coherent mask for the DU device
Laurent Pinchart [Sun, 12 Nov 2017 12:12:10 +0000 (14:12 +0200)]
drm: rcar-du: Set the DMA coherent mask for the DU device

The DU DMA address space is limited to 32 bits, so the DMA coherent mask
should be set accordingly. The DMA mapping implementation will
transparently map high memory buffers to 32-bit addresses through an
IOMMU when present (or through bounce buffers otherwise, which isn't a
supported use case as performances would be terrible).

However, when sourcing frames from a VSP, the situation is more
complicated. The DU delegates all memory accesses to the VSP and doesn't
perform any DMA access by itself. Due to how the GEM CMA helpers are
structured buffers are still mapped to the DU device. They are later
mapped to the VSP as well to perform DMA access, through the IOMMU
connected to the VSP.

Setting the DMA coherent mask to 32 bits for the DU when using a VSP can
cause issues when importing a dma_buf. If the buffer is located above
the 32-bit address space, the DMA mapping implementation will try to map
it to the DU's DMA address space. As the DU has no IOMMU a bounce buffer
will be allocated, which in the best case will waste memory and in the
worst case will just fail.

To work around this issue, set the DMA coherent mask to the full 40-bit
address space for the DU. All dma-buf instances will be imported without
any restriction, and will be mapped to the VSP when preparing the
associated framebuffer.

Signed-off-by: Laurent Pinchart <laurent.pinchart+renesas@ideasonboard.com>
Reviewed-by: Liviu Dudau <liviu.dudau@arm.com>
2 years agodrm: rcar-du: Improve kernel log messages when initializing encoders
Laurent Pinchart [Sat, 21 Aug 2021 23:44:27 +0000 (02:44 +0300)]
drm: rcar-du: Improve kernel log messages when initializing encoders

Improve the debugging and error messages printing when initializing
encoders by replacing the output number by the output name, printing the
bridge OF node name, and the error code of failed operations.

While at it, move the related rcar_du_output enumeration from
rcar_du_crtc.h to rcar_du_drv.h as it's not specific to the CRTC.

Signed-off-by: Laurent Pinchart <laurent.pinchart+renesas@ideasonboard.com>
2 years agodrm: rcar-du: Don't create encoder for unconnected LVDS outputs
Laurent Pinchart [Sat, 21 Aug 2021 23:44:27 +0000 (02:44 +0300)]
drm: rcar-du: Don't create encoder for unconnected LVDS outputs

On R-Car D3 and E3, the LVDS encoders provide the pixel clock to the DU,
even when LVDS outputs are not used. For this reason, the rcar-lvds
driver probes successfully on those platforms even if no further bridge
or panel is connected to the LVDS output, in order to provide the
rcar_lvds_clk_enable() and rcar_lvds_clk_disable() functions to the DU
driver.

If an LVDS output isn't connected, trying to create a DRM connector for
the output will fail. Fix this by skipping connector creation in that
case, and also skip creation of the DRM encoder as there's no point in
an encoder without a connector.

Fixes: e9e056949c92 ("drm: rcar-du: lvds: Convert to DRM panel bridge helper")
Reported-by: Geert Uytterhoeven <geert+renesas@glider.be>
Signed-off-by: Laurent Pinchart <laurent.pinchart+renesas@ideasonboard.com>
Tested-by: Geert Uytterhoeven <geert+renesas@glider.be>
2 years agodrm/amd/display: Add 120Hz support for freesync video mode
Solomon Chiu [Tue, 5 Oct 2021 12:52:42 +0000 (20:52 +0800)]
drm/amd/display: Add 120Hz support for freesync video mode

[Why]
For those video format with 60 fps, the user space player could
ask for 120Hz for playback.

[How]
Add 120 in the table of common rates.

Signed-off-by: Solomon Chiu <solomon.chiu@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2 years agodrm/amdgpu: Register MCE notifier for Aldebaran RAS
Mukul Joshi [Wed, 22 Sep 2021 18:49:43 +0000 (14:49 -0400)]
drm/amdgpu: Register MCE notifier for Aldebaran RAS

On Aldebaran, GPU driver will handle bad page retirement
for GPU memory even though UMC is host managed. As a result,
register a bad page retirement handler on the mce notifier
chain to retire bad pages on Aldebaran.

Signed-off-by: Mukul Joshi <mukul.joshi@amd.com>
Reviewed-by: Yazen Ghannam <yazen.ghannam@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2 years agox86/MCE/AMD: Export smca_get_bank_type symbol
Mukul Joshi [Sun, 28 Mar 2021 02:54:04 +0000 (22:54 -0400)]
x86/MCE/AMD: Export smca_get_bank_type symbol

Export smca_get_bank_type for use in the AMD GPU
driver to determine MCA bank while handling correctable
and uncorrectable errors in GPU UMC.

Signed-off-by: Mukul Joshi <mukul.joshi@amd.com>
Acked-by: Borislav Petkov <bp@suse.de>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2 years agodrm/amdgpu: return early if debugfs is not initialized
Nirmoy Das [Wed, 6 Oct 2021 09:06:33 +0000 (11:06 +0200)]
drm/amdgpu: return early if debugfs is not initialized

Check first if debugfs is initialized before creating
amdgpu debugfs files.

References: https://gitlab.freedesktop.org/drm/amd/-/issues/1686
Signed-off-by: Nirmoy Das <nirmoy.das@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Lijo Lazar <lijo.lazar@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2 years agodrm/amd/display: USB4 bring up set correct address
Jude Shih [Wed, 29 Sep 2021 03:16:51 +0000 (11:16 +0800)]
drm/amd/display: USB4 bring up set correct address

[Why]
YELLOW_CARP_B0 address was not correct

[How]
Set YELLOW_CARP_B0 to 0x1A.

Reviewed-by: Nicholas Kazlauskas <nicholas.kazlauskas@amd.com>
Acked-by: Wayne Lin <Wayne.Lin@amd.com>
Acked-by: Harry Wentland <harry.wentland@amd.com>
Signed-off-by: Jude Shih <shenshih@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2 years agodrm/amd/display: Fix USB4 Aux via DMUB terminate unexpectedly
Jude Shih [Wed, 29 Sep 2021 03:13:56 +0000 (11:13 +0800)]
drm/amd/display: Fix USB4 Aux via DMUB terminate unexpectedly

[Why]
Condition variable sometimes terminated unexpectedly

[How]
Use wait_for_completion_timeout to avoid unexpected termination of CV

Reviewed-by: Nicholas Kazlauskas <nicholas.kazlauskas@amd.com>
Acked-by: Wayne Lin <Wayne.Lin@amd.com>
Acked-by: Harry Wentland <harry.wentland@amd.com>
Signed-off-by: Jude Shih <shenshih@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2 years agodrm/amd/display: Deadlock/HPD Status/Crash Bug Fix
Jude Shih [Mon, 13 Sep 2021 06:41:34 +0000 (14:41 +0800)]
drm/amd/display: Deadlock/HPD Status/Crash Bug Fix

[why]
1. HPD callback function has deadlock problem
2. HPD status is not assigned
3. There is crash due to null pointer
4. link_enc is NULL in DPIA case

[How]
1. Fix deadlock problem by moving it out of the
   drm_modeset_lock
2. Assign HPD status from the notify of outbox
   from dmub FW
3. Fix the crash by checking if pin or enc exists
4. Use link_enc_cfg_get_link_enc_used_by_link to
   dynamically assign

Reviewed-by: Nicholas Kazlauskas <nicholas.kazlauskas@amd.com>
Acked-by: Wayne Lin <Wayne.Lin@amd.com>
Acked-by: Harry Wentland <harry.wentland@amd.com>
Signed-off-by: Jude Shih <shenshih@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2 years agodrm/amd/display: Fix for access for ddc pin and aux engine.
Meenakshikumar Somasundaram [Fri, 13 Aug 2021 17:09:25 +0000 (13:09 -0400)]
drm/amd/display: Fix for access for ddc pin and aux engine.

[Why & How]
1. Remove unnecessary dummy interrupt source for USB4 HPD & HPD RX
2. Adjust parameter for DPCD writing of link training process of DPIA link
3. Adjust specific AUX defer delay for DPIA link

Reviewed-by: Jimmy Kizito <Jimmy.Kizito@amd.com>
Acked-by: Wayne Lin <Wayne.Lin@amd.com>
Acked-by: Nicholas Kazlauskas <nicholas.kazlauskas@amd.com>
Acked-by: Harry Wentland <harry.wentland@amd.com>
Signed-off-by: Meenakshikumar Somasundaram <meenakshikumar.somasundaram@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2 years agodrm/amd/display: Add debug flags for USB4 DP link training.
Jimmy Kizito [Thu, 26 Aug 2021 01:26:33 +0000 (21:26 -0400)]
drm/amd/display: Add debug flags for USB4 DP link training.

[Why & How]
Additional debug flags that can be useful for testing USB4 DP
link training.

Add flags:
- 0x2 : Forces USB4 DP link to non-LTTPR mode
- 0x4 : Extends status read intervals to about 60s.

Reviewed-by: Meenakshikumar Somasundaram <meenakshikumar.somasundaram@amd.com>
Reviewed-by: Jun Lei <Jun.Lei@amd.com>
Acked-by: Wayne Lin <Wayne.Lin@amd.com>
Acked-by: Nicholas Kazlauskas <nicholas.kazlauskas@amd.com>
Acked-by: Harry Wentland <harry.wentland@amd.com>
Signed-off-by: Jimmy Kizito <Jimmy.Kizito@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2 years agodrm/amd/display: Fix DIG_HPD_SELECT for USB4 display endpoints.
Jimmy Kizito [Mon, 23 Aug 2021 21:23:24 +0000 (17:23 -0400)]
drm/amd/display: Fix DIG_HPD_SELECT for USB4 display endpoints.

[Why]
DIB_BE_CNTL<i>.DIG_HPD_SELECT selects the HPD block being used
by the display endpoint assigned to DIG<i>. In the case of USB4
display endpoints, no physical HPD block is assigned.

[How]
Setting DIB_BE_CNTL<i>.DIG_HPD_SELECT to 5 indicates that no HPD
is assigned to a display endpoint. Firmware decrements the
HPD_SELECT value by 1 before writing it to the register.

Reviewed-by: Meenakshikumar Somasundaram <meenakshikumar.somasundaram@amd.com>
Acked-by: Wayne Lin <Wayne.Lin@amd.com>
Acked-by: Nicholas Kazlauskas <nicholas.kazlauskas@amd.com>
Acked-by: Harry Wentland <harry.wentland@amd.com>
Signed-off-by: Jimmy Kizito <Jimmy.Kizito@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2 years agodrm/amd/display: Support for SET_CONFIG processing with DMUB
Jude Shih [Wed, 11 Aug 2021 06:41:34 +0000 (14:41 +0800)]
drm/amd/display: Support for SET_CONFIG processing with DMUB

[Why]
To process SET_CONFIG transactions with DMUB using inbox1 and
outbox1 mail boxes.

[How]
1) DMUB posts SET_CONFIG reply as an Outbox1 message of type
   DMUB_OUT_CMD__SET_CONFIG_REPLY.
2) The dmub async to sync mechanism for AUX is modified to accommodate
   SET_CONFIG commands for both command issue and reply code paths.

Reviewed-by: Nicholas Kazlauskas <Nicholas.Kazlauskas@amd.com>
Acked-by: Wayne Lin <Wayne.Lin@amd.com>
Acked-by: Harry Wentland <harry.wentland@amd.com>
Signed-off-by: Jude Shih <shenshih@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2 years agodrm/amd/display: Add dpia debug options
Meenakshikumar Somasundaram [Wed, 23 Jun 2021 21:33:19 +0000 (17:33 -0400)]
drm/amd/display: Add dpia debug options

[Why & How]
To add support for dpia debug options.

Reviewed-by: Jimmy Kizito <Jimmy.Kizito@amd.com>
Acked-by: Wayne Lin <Wayne.Lin@amd.com>
Acked-by: Nicholas Kazlauskas <nicholas.kazlauskas@amd.com>
Acked-by: Harry Wentland <harry.wentland@amd.com>
Signed-off-by: Meenakshikumar Somasundaram <meenakshikumar.somasundaram@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2 years agodrm/amd/display: Read USB4 DP tunneling data from DPCD.
Jimmy Kizito [Fri, 14 May 2021 23:56:19 +0000 (19:56 -0400)]
drm/amd/display: Read USB4 DP tunneling data from DPCD.

[Why]
We requires information from DPCD in order to identify USB4 DP
tunneling targets.

[How]
Add USB4 DP tunneling fields to DPCD struct and populate these fields
during sink detection.

Reviewed-by: Jun Lei <Jun.Lei@amd.com>
Acked-by: Wayne Lin <Wayne.Lin@amd.com>
Acked-by: Nicholas Kazlauskas <nicholas.kazlauskas@amd.com>
Acked-by: Harry Wentland <harry.wentland@amd.com>
Signed-off-by: Jimmy Kizito <Jimmy.Kizito@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2 years agodrm/amd/display: Support for SET_CONFIG processing with DMUB
Meenakshikumar Somasundaram [Fri, 30 Apr 2021 23:02:33 +0000 (19:02 -0400)]
drm/amd/display: Support for SET_CONFIG processing with DMUB

[Why]
To process SET_CONFIG transactions with DMUB using inbox1 and
outbox1 mail boxes.

[How]
1) Added inbox1 DPIA command subtype DMUB_CMD__DPIA_SET_CONFIG_ACCESS to
   issue SET_CONFIG command to DMUB in dc_process_dmub_set_config_async().
   DMUB processes the command with DPIA sends reply back immediately or
   in an outbox1 message triggering an outbox1 interrupt to driver.
2) DMUB posts SET_CONFIG reply as an Outbox1 message of type
   DMUB_OUT_CMD__SET_CONFIG_REPLY.
3) The dmub async to sync mechanism for AUX is modified to accommodate
   SET_CONFIG commands for both command issue and reply code paths.

Reviewed-by: Jun Lei <Jun.Lei@amd.com>
Acked-by: Wayne Lin <Wayne.Lin@amd.com>
Acked-by: Nicholas Kazlauskas <nicholas.kazlauskas@amd.com>
Acked-by: Harry Wentland <harry.wentland@amd.com>
Signed-off-by: Meenakshikumar Somasundaram <meenakshikumar.somasundaram@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2 years agodrm/amd/display: Implement end of training for hop in DPIA display path
Jimmy Kizito [Wed, 21 Apr 2021 14:34:34 +0000 (10:34 -0400)]
drm/amd/display: Implement end of training for hop in DPIA display path

[Why & How]
Clear training pattern sequence for hop in display path once clock
recovery and equalization phases of DP tunnel link training completed.

Reviewed-by: Jun Lei <Jun.Lei@amd.com>
Acked-by: Wayne Lin <Wayne.Lin@amd.com>
Acked-by: Nicholas Kazlauskas <nicholas.kazlauskas@amd.com>
Acked-by: Harry Wentland <harry.wentland@amd.com>
Signed-off-by: Jimmy Kizito <Jimmy.Kizito@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2 years agodrm/amd/display: Implement DPIA equalisation phase
Jimmy Kizito [Mon, 19 Apr 2021 19:12:01 +0000 (15:12 -0400)]
drm/amd/display: Implement DPIA equalisation phase

[Why]
Equalisation is the mandatory second phase of DisplayPort link training
over a USB4 DP tunnel.

[How]
Implement equalisation phase for DP tunneled over USB4 in DPIA
training module.

Reviewed-by: Jun Lei <Jun.Lei@amd.com>
Acked-by: Wayne Lin <Wayne.Lin@amd.com>
Acked-by: Nicholas Kazlauskas <nicholas.kazlauskas@amd.com>
Acked-by: Harry Wentland <harry.wentland@amd.com>
Signed-off-by: Jimmy Kizito <Jimmy.Kizito@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2 years agodrm/amd/display: Implement DPIA clock recovery phase
Jimmy Kizito [Tue, 27 Apr 2021 22:46:11 +0000 (18:46 -0400)]
drm/amd/display: Implement DPIA clock recovery phase

[Why]
Clock recovery is the mandatory first phase of DP link training.

[How]
- Implement clock recovery phase in DPIA training module.
- Add helper functions for building SET_CONFIG messages.

Reviewed-by: Jun Lei <Jun.Lei@amd.com>
Acked-by: Wayne Lin <Wayne.Lin@amd.com>
Acked-by: Nicholas Kazlauskas <nicholas.kazlauskas@amd.com>
Acked-by: Harry Wentland <harry.wentland@amd.com>
Signed-off-by: Jimmy Kizito <Jimmy.Kizito@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2 years agodrm/amd/display: Implement DPIA link configuration
Jimmy Kizito [Mon, 12 Apr 2021 20:03:11 +0000 (16:03 -0400)]
drm/amd/display: Implement DPIA link configuration

[Why]
Training settings need to be applied to DPIA link at start of each
training loop. Note: FEC readiness should be configured before link
training while FEC enablement should be configured once training is
complete.

[How]
- Implement DPIA link configuration function.
- Account for dynamically assigned link encoders during link
configuration.

Reviewed-by: Jun Lei <Jun.Lei@amd.com>
Acked-by: Wayne Lin <Wayne.Lin@amd.com>
Acked-by: Nicholas Kazlauskas <nicholas.kazlauskas@amd.com>
Acked-by: Harry Wentland <harry.wentland@amd.com>
Signed-off-by: Jimmy Kizito <Jimmy.Kizito@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2 years agodrm/amd/display: Implement DPIA training loop
Jimmy Kizito [Sat, 10 Apr 2021 02:23:53 +0000 (22:23 -0400)]
drm/amd/display: Implement DPIA training loop

[Why]
Training of DPIA link differs enough from that of conventional
DP link to warrant a separate implementation.

[How]
- Implement top-level of DPIA training loop.
- Make functions shared between DP and DPIA link training "public".

Reviewed-by: Jun Lei <Jun.Lei@amd.com>
Acked-by: Wayne Lin <Wayne.Lin@amd.com>
Acked-by: Nicholas Kazlauskas <nicholas.kazlauskas@amd.com>
Acked-by: Harry Wentland <harry.wentland@amd.com>
Signed-off-by: Jimmy Kizito <Jimmy.Kizito@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2 years agodrm/amd/display: Train DPIA links with fallback
Jimmy Kizito [Wed, 7 Apr 2021 22:56:19 +0000 (18:56 -0400)]
drm/amd/display: Train DPIA links with fallback

[Why & How]
Conventional links are trained with fallback during sink detection.
Have DPIA links trained with fallback too.

Reviewed-by: Jun Lei <Jun.Lei@amd.com>
Acked-by: Wayne Lin <Wayne.Lin@amd.com>
Acked-by: Nicholas Kazlauskas <nicholas.kazlauskas@amd.com>
Acked-by: Harry Wentland <harry.wentland@amd.com>
Signed-off-by: Jimmy Kizito <Jimmy.Kizito@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2 years agodrm/amd/display: Skip DPCD read for DPTX-to-DPIA hop
Jimmy Kizito [Mon, 5 Apr 2021 22:05:09 +0000 (18:05 -0400)]
drm/amd/display: Skip DPCD read for DPTX-to-DPIA hop

[why & how]
Driver does not need to train the first hop.

Reviewed-by: Jun Lei <Jun.Lei@amd.com>
Acked-by: Wayne Lin <Wayne.Lin@amd.com>
Acked-by: Nicholas Kazlauskas <nicholas.kazlauskas@amd.com>
Acked-by: Harry Wentland <harry.wentland@amd.com>
Signed-off-by: Jimmy Kizito <Jimmy.Kizito@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2 years agodrm/amd/display: Add stub to get DPIA tunneling device data
Jimmy Kizito [Thu, 1 Apr 2021 16:59:54 +0000 (12:59 -0400)]
drm/amd/display: Add stub to get DPIA tunneling device data

[why & how]
1. Add stub for getting tunneling device data
2. Add check for phy_repeater_cnt < 0xff to LTTPR check
3. Add two more bits of information to DPIA links

Reviewed-by: Jun Lei <Jun.Lei@amd.com>
Acked-by: Wayne Lin <Wayne.Lin@amd.com>
Acked-by: Nicholas Kazlauskas <nicholas.kazlauskas@amd.com>
Acked-by: Harry Wentland <harry.wentland@amd.com>
Signed-off-by: Jimmy Kizito <Jimmy.Kizito@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2 years agodrm/amd/display: Stub out DPIA link training call
Jimmy Kizito [Wed, 6 Jan 2021 20:21:11 +0000 (15:21 -0500)]
drm/amd/display: Stub out DPIA link training call

[why & how]
Add stub for DPIA link training and define new DPIA DMUB commands
to support it.

Reviewed-by: Jun Lei <Jun.Lei@amd.com>
Acked-by: Wayne Lin <Wayne.Lin@amd.com>
Acked-by: Nicholas Kazlauskas <nicholas.kazlauskas@amd.com>
Acked-by: Harry Wentland <harry.wentland@amd.com>
Signed-off-by: Jimmy Kizito <Jimmy.Kizito@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2 years agodrm/amd/display: Set DPIA link endpoint type
Jimmy Kizito [Tue, 5 Jan 2021 19:25:23 +0000 (14:25 -0500)]
drm/amd/display: Set DPIA link endpoint type

[why & how]
We will need a way to distinguish physically connected
links and DPIA endpoints.

Reviewed-by: Jun Lei <Jun.Lei@amd.com>
Acked-by: Wayne Lin <Wayne.Lin@amd.com>
Acked-by: Nicholas Kazlauskas <nicholas.kazlauskas@amd.com>
Acked-by: Harry Wentland <harry.wentland@amd.com>
Signed-off-by: Jimmy Kizito <Jimmy.Kizito@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2 years agodrm/amd/display: Support for DMUB HPD and HPD RX interrupt handling
Meenakshikumar Somasundaram [Wed, 27 Jan 2021 06:05:28 +0000 (01:05 -0500)]
drm/amd/display: Support for DMUB HPD and HPD RX interrupt handling

[WHY]
To add support for HPD & HPD RX interrupt handling for USB4 DPIA in
YELLOW_CARP_B0. USB4 DPIA HPD & HPD RX interrupts are issued from DMUB to
driver as a outbox1 message.

[HOW]
1) Created get_link_index_from_dpia_port_index() to retrieve link index
   from dpia port index for HPD & HPD RX dmub notifications.
2) Added DMUB HPD & HPD RX handling in dmub_srv_stat_get_notification().

Reviewed-by: Jun Lei <Jun.Lei@amd.com>
Acked-by: Wayne Lin <Wayne.Lin@amd.com>
Acked-by: Nicholas Kazlauskas <nicholas.kazlauskas@amd.com>
Acked-by: Harry Wentland <harry.wentland@amd.com>
Signed-off-by: Meenakshikumar Somasundaram <meenakshikumar.somasundaram@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>