This version brings along following fixes:
- Port DCN30 420 logic to DCN32
- Remove some unused definitions from DCN32/321
- Remove dp dig pixle rate div policy from dcn314
- Fix dcn315 reading of memory channel count and width
- Fix SubVP and ODM relevant issues
- Fix pipe split, MPO and ODM relevant issues
- Support proper mst payload removal when link is not in mst mode in dc
- Assume an LTTPR is always present on fixed_vs links
- Rework recent update PHY state commit
- Add debug option and logs
Acked-by: Wayne Lin <wayne.lin@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>
[Why and How]
- For driver disable cases in current implementation, if P-State
is unsupported or still supported by firmware, we force it
supported by DCN.
- SubVP now needs to be included in this case along with FPO.
Reviewed-by: Jun Lei <Jun.Lei@amd.com>
Acked-by: Wayne Lin <wayne.lin@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>
[Why and How]
- Add a debug option for allocating extra way for cursor
- Remove usage of cache_cursor_addr since it's not gaurenteed
to be populated
- Include cursor size in MALL calculation if it exceeds the
DCN cursor buffer size (and don't need extra way for cursor)
Reviewed-by: Aurabindo Pillai <Aurabindo.Pillai@amd.com>
Acked-by: Wayne Lin <wayne.lin@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>
[Why & How]
ODM seamless transitions require DIV_MODE_AUTO. However,
DIV_MODE_AUTO only works when all the horizontal timing params
are divisible by the ODM combine factor. Therefore, disable the
ODM 2:1 policy when the horizontal timing params are not divisible
by 2.
Reviewed-by: Jun Lei <Jun.Lei@amd.com>
Acked-by: Wayne Lin <wayne.lin@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>
[why]
When user unplugs mst hubs, the current code will forcefully zero
entire mst payload allocation table structure stored in link before we
deallocate actual payload when disabling stream.
During the first disable stream sequence, we will use current mst
payload allocation table to determine if link should be turned off.
Because we zero out it before we are disabling stream, the payload
allocation table stored in link doesn't represent the actual allocation status,
so we turn off link at the first disable stream without waiting until all
streams' payloads have been deallocated. This avoilates the designed
deallocation sequence and caused system hang in DP2 scenario.
[how]
Remove payload during deallocation and never zero payload allocation structure
without actually deallocating payload.
Reviewed-by: Jun Lei <Jun.Lei@amd.com>
Acked-by: Wayne Lin <wayne.lin@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>
[why]
Original change 594b237b9a ("drm/amd/display: Add
interface to track PHY state") was implemented by assuming stream's
dpms off is equivalent to PHY power off.
This assumption doesn't hold in following situations:
1. MST multiple stream scenario, where multiple streams are sharing the
same PHY output. Toggle dpms off for one of the stream doesn't power
off the PHY due to the presence of other streams.
2. enable stream failure scenario, where enable stream fails due to
failure of link training. This will cause DPMS off is set to false, while
the actual PHY power state is off in certain cases.
Due to the problematic assumption, the logic will skip disabling
other streams for MST multiple stream scenario, therefore PHY is
not actually powered off.
[how]
1. Rework this refactor by moving PHY state update down to hardware
level, where we update PHY state in place when hardware sequencer
is actually changing the power state of the PHY hardware.
2. Reimplement symclk on TX off workaround in place when we are actually
calling transmitter control to power off PHY in dcn32. Note the workaround is
added due to the lack of proper software interface to set TX while keeping
symclk on. We plan to address this interface problem so we can set TX off
only without affecting symclk in future dcn versions.
Fixes: 594b237b9a ("drm/amd/display: Add interface to track PHY state")
Reviewed-by: Jun Lei <Jun.Lei@amd.com>
Acked-by: Wayne Lin <wayne.lin@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>
[WHY]
LTTPRs can in very rare instsances fail to increment DPCD LTTPR count.
This results in aux-i LTTPR requests to be sent to the wrong DPCD
address, which causes link training failure.
[HOW]
Override internal repeater count if fixed_vs flag is set for a given link
Reviewed-by: George Shen <George.Shen@amd.com>
Acked-by: Wayne Lin <wayne.lin@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>
[Why]
DC makes use of layer_index (zpos) when picking the HW plane to enable
HW cursor on. However, some compositors will not attach zpos information
to each DRM plane. Consequently, in amdgpu, we default layer_index to 0
and do not update it.
This causes said DC logic to enable HW cursor on all planes of the same
layer_index, which manifests as a double cursor issue if one of the
planes is scaled (and hence scaling the cursor as well).
[How]
Use DRM core helpers to calculate a normalized_zpos value for each
drm_plane_state under each crtc, within the atomic state.
This helper will first consider existing zpos values, and if
identical/unset, fallback to plane ID ordering.
The normalized_zpos is then passed to dc_plane_info during atomic check
for later use by the cursor logic.
Reviewed-by: Bhawanpreet Lakha <Bhawanpreet.Lakha@amd.com>
Acked-by: Wayne Lin <wayne.lin@amd.com>
Signed-off-by: Leo Li <sunpeng.li@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
[Why & How]
- Pipe split prediction previously only took into
account MPC split. We must also consider when
ODM combine is required, and when we apply ODM
combine by policy.
- Also re-work DET allocation function as it wasn't
properly splitting the DET per stream, per plane.
Reviewed-by: Jun Lei <Jun.Lei@amd.com>
Acked-by: Wayne Lin <wayne.lin@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>
[Why&How]
After reg offset initialization was switched to runtime rather than
compile time, some of the defintions are not needed anymore and can
be removed.
Acked-by: Wayne Lin <wayne.lin@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>
[Why]
420 modes are limited by FMT buffer width of 4096
which requires multi-pipe support in form of ODM
combine. If 420 modes have greater HActive than
4096, the DML logic should accomodate whether
it should be rejected, or ODM combine 2:1 or 4:1
is triggered accordingly.
[How]
FMT Buffer limit of 4096 in DCN32. Force ODM
combine depending on HActive and FMT Buffer limit.
Reject modes if TMDS 420 and above 4096.
Acked-by: Wayne Lin <wayne.lin@amd.com>
Signed-off-by: Chris Park <chris.park@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
This patch updates the PTE flags when translate further (TF) is
enabled:
- With translate_further enabled, invalid PTEs can be 0. Reading
consecutive invalid PTEs as 0 is considered a fault. To prevent
this, ensure invalid PTEs have at least 1 bit set.
- The current invalid PTE flags settings to translate a retry fault
into a no-retry fault, doesn't work with TF enabled. As a result,
update invalid PTE flags settings which works for both TF enabled
and disabled case.
Fixes: 352e683b72 ("drm/amdgpu: Enable translate_further to extend UTCL2 reach")
Reviewed-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Felix Kuehling <Felix.Kuehling@amd.com>
Signed-off-by: Mukul Joshi <mukul.joshi@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
SDMA update page table may be called from unlocked context, this
generate below warning. Use unlocked iterator to handle this case.
WARNING: CPU: 0 PID: 1475 at
drivers/dma-buf/dma-resv.c:483 dma_resv_iter_next
Call Trace:
dma_resv_iter_first+0x43/0xa0
amdgpu_vm_sdma_update+0x69/0x2d0 [amdgpu]
amdgpu_vm_ptes_update+0x29c/0x870 [amdgpu]
amdgpu_vm_update_range+0x2f6/0x6c0 [amdgpu]
svm_range_unmap_from_gpus+0x115/0x300 [amdgpu]
svm_range_cpu_invalidate_pagetables+0x510/0x5e0 [amdgpu]
__mmu_notifier_invalidate_range_start+0x1d3/0x230
unmap_vmas+0x140/0x150
unmap_region+0xa8/0x110
Signed-off-by: Philip Yang <Philip.Yang@amd.com>
Suggested-by: Felix Kuehling <felix.kuehling@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
On machines without an Intel video opregion the acpi_video driver
immediately probes the ACPI video bus and used to also immediately
register acpi_video# backlight devices when supported.
Once the drm/kms driver then loaded later and possibly registered
a native backlight device then the drivers/acpi/video_detect.c code
unregistered the acpi_video0 device to avoid there being 2 backlight
devices (when acpi_video_get_backlight_type()==native).
This means that userspace used to briefly see 2 devices and the
disappearing of acpi_video0 after a brief time confuses the systemd
backlight level save/restore code, see e.g.:
https://bbs.archlinux.org/viewtopic.php?id=269920
To fix this the ACPI video code has been modified to make backlight class
device registration a separate step, relying on the drm/kms driver to
ask for the acpi_video backlight registration after it is done setting up
its native backlight device.
Add a call to the new acpi_video_register_backlight() function after
setting up the gma500's native backlight, so that the acpi_video backlight
device gets registered on systems where the gma500's native backlight
device is not registered.
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Acked-by: Patrik Jakobsson <patrik.r.jakobsson@gmail.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220917205920.647212-6-hdegoede@redhat.com
Before this commit when we want userspace to use the acpi_video backlight
device we register both the GPU's native backlight device and acpi_video's
firmware acpi_video# backlight device. This relies on userspace preferring
firmware type backlight devices over native ones.
Registering 2 backlight devices for a single display really is
undesirable, don't register the GPU's native backlight device when
another backlight device should be used.
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Acked-by: Patrik Jakobsson <patrik.r.jakobsson@gmail.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220917205920.647212-5-hdegoede@redhat.com
Change the type for the registered backlight class device from platform
to raw/native.
The poulsbo/cedarview/oaktrail backlight support is using native GPU
backlight control and as such the type should be raw (aka native) as
is done by all the other native GPU backlight driver code.
Note this will not change much from userspace's point of view.
poulsbo/cedarview laptops typically offer both an ACPI-video
backlight interface as well as the native GPU backlight interface.
The /sys/class/backlight/acpi_video0 has a type of firmware and
userspace typically looks for firmware devices before looking
for platform devices. The typical standard lookup order is:
firmware -> platform -> raw
This means that both before and after this change typical userspace
backlight consumers (sich as e.g. GNOME) will prefer the firmware
acpi_video0 backlight device.
This has been tested on a Packard Bell Dot SC (Intel Atom N2600, cedarview)
and a Sony Vaio vpc-x11s1e (Intel N540, poulsbo) laptop.
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Acked-by: Patrik Jakobsson <patrik.r.jakobsson@gmail.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220917205920.647212-3-hdegoede@redhat.com
Refactor backlight support so that the gma_backlight_enable() /
gma_backlight_disable() / gma_backlight_set() functions used by
the Opregion handle will also work if no backlight_device gets
registered.
This is a preparation patch for not registering the gma500's own backlight
device when acpi_video should be used, since registering 2 backlight
devices for a single display really is undesirable.
Since the acpi-video interface often uses the OpRegion we need to keep
the OpRegion functional even when dev_priv->backlight_device is NULL.
As a result of this refactor the actual backlight_device_register()
call is moved to the shared backlight.c code and all #ifdefs related to
CONFIG_BACKLIGHT_CLASS_DEVICE are now also limited to backlight.c .
No functional changes intended.
This has been tested on a Packard Bell Dot SC (Intel Atom N2600, cedarview)
and a Sony Vaio vpc-x11s1e (Intel N540, poulsbo) laptop.
Changes in v2:
- Fix unused variable warnings when CONFIG_BACKLIGHT is not selected by
marking the 2 variables as __maybe_unused.
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Acked-by: Patrik Jakobsson <patrik.r.jakobsson@gmail.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220917205920.647212-2-hdegoede@redhat.com
