[WHY]
To process AUX transactions with DMUB using inbox1 and outbox1 mail boxes.
[HOW]
1) Added inbox1 command DMUB_CMD__DP_AUX_ACCESS to issue AUX commands
to DMUB in dc_process_dmub_aux_transfer_async(). DMUB processes AUX cmd
with DCN and sends reply back in an outbox1 message triggering an
outbox1 interrupt to driver.
2) In existing driver implementation, AUX commands are processed
synchronously by configuring DCN reg. But in DMUB AUX, driver sends an
inbox1 message and waits for a conditional variable (CV) which will be
signaled by outbox1 ISR.
3) As the driver holds dal and dc locks while waiting for CV, the outbox1
ISR is registered with noMutexWait set to true, which allows ISR to run
and signal CV. This sets a constraint on ISR to not modify variables
such as dc, dmub, etc.
4) Created dmub_outbox.c with dmub_enable_outbox_notification() to enable
outbox1 mailbox.
5) New mailbox address ranges allocated for outbox1 of size DMUB_RB_SIZE.
Created dmub functions for Outbox1: dmub_dcn20_setup_out_mailbox(),
dmub_dcn20_get_outbox1_wptr() and dmub_dcn20_set_outbox1_rptr().
6) Added functions dc_stat_get_dmub_notification() and
dmub_srv_stat_get_notification() to retrieve Outbox1 message.
7) Currently, DMUB doesn't opens DDC in AUX mode before issuing AUX
transaction. A workaround is added in dce_aux_transfer_dmub_raw() to
open in DDC in AUX mode for every AUX transaction.
8) Added dc debug option enable_dmub_aux_for_legacy_ddc enable/disable
DMUB AUX. This debug option is checked dce_aux_transfer_with_retries()
to select the method to process AUX transactions.
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Meenakshikumar Somasundaram <meenakshikumar.somasundaram@amd.com>
Reviewed-by: Jun Lei <Jun.Lei@amd.com>
Acked-by: Rodrigo Siqueira <Rodrigo.Siqueira@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
[why]
Vendor based fan noise improvement
[how]
Report timing synchronizable when DP streams time frame
difference is less than 0.05 percent. Adjust DP DTOs and
sync displays using MASTER_UPDATE_LOCK_DB_X_Y
Signed-off-by: Vladimir Stempen <vladimir.stempen@amd.com>
Acked-by: Bindu Ramamurthy <bindu.r@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
[Why]
In some cases, this function is called when DIG BE is not
connected to DIG FE, in which case a value of zero isn't
invalid and assert should not be hit.
[How]
Remove assert and handle ENGINE_ID_UNKNOWN result in calling
function.
Signed-off-by: Eric Bernstein <eric.bernstein@amd.com>
Acked-by: Bindu Ramamurthy <bindu.r@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
New proposed sequence for HUBP blanking causes regressions where the
hardware would fail to enter blank which triggers an assert in the new
sequence. This change brings back the old sequence.
Fixes: 985faf2c4e ("drm/amd/display: New sequence for HUBP blank")
Signed-off-by: Aurabindo Pillai <aurabindo.pillai@amd.com>
Signed-off-by: Bhawanpreet Lakha <bhawanpreet.lakha@amd.com>
Reviewed-by: Rodrigo Siqueira <Rodrigo.Siqueira@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
It is possible for ret_vsnprintf to be assigned negative value in
error cases. As an unsigned variable, negative values which
are stored in their 2's complement form gets treated as a positive
number. This will led to treating bad cases as good ones.
eg: -1 gets stored as 0xFFFFFFFF on a 32 bit system
Signed-off-by: Dale Zhao <dale.zhao@amd.com>
Reviewed-by: Yongqiang Sun <yongqiang.sun@amd.com>
Acked-by: Anson Jacob <Anson.Jacob@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
[Why and How]
VBIOS program DIG_CLK_PATTERN using engine ID instead of PHY ID.
Workaround by writing value for 0x1f (for HDMI) after calling vbios.
Signed-off-by: Derek Lai <Derek.Lai@amd.com>
Reviewed-by: Tony Cheng <Tony.Cheng@amd.com>
Acked-by: Anson Jacob <Anson.Jacob@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
[Why]
In seamless boot without a flip case, the flag power_gated didn't
get cleared when resetting path mode because the plane_state is null.
The following sequence will cause this issue:
1. OS call set mode to clone/extended
2. Reset path mode to remove edp
[How]
Set power gated default to true in seamless boot pipe
Signed-off-by: Lewis Huang <Lewis.Huang@amd.com>
Reviewed-by: Martin Leung <Martin.Leung@amd.com>
Acked-by: Anson Jacob <Anson.Jacob@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
[Why]
The initial purpose of dcn10 pipe split is to support some high
bandwidth mode which requires dispclk greater than max dispclk. By
initial bring up power measurement data, it showed power consumption is
less with pipe split for dcn block. This could be reason for enable pipe
split by default. By battery life measurement of some Chromebooks,
result shows battery life is longer with pipe split disabled.
[How]
Disable pipe split by default. Pipe split could be still enabled when
required dispclk is greater than max dispclk.
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Hersen Wu <hersenxs.wu@amd.com>
Signed-off-by: Roman Li <Roman.Li@amd.com>
Reviewed-by: Roman Li <Roman.Li@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
- Reverts "drm/amd/display: Revert HUBP blank behaviour for now"
- Hubp blank will fail if the pipe is locked (this is the case on
linux), so add a check to make sure pipe isn't locked, if it is then
defer the blank to post_unlock.
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Bhawanpreet Lakha <Bhawanpreet.Lakha@amd.com>
Reviewed-by: Nicholas Kazlauskas <Nicholas.Kazlauskas@amd.com>
Acked-by: Rodrigo Siqueira <Rodrigo.Siqueira@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
[WHY]
DCN30 has a bug where blanking HUBP blocks pstate allow unless
HUBP_DISABLE is toggled afterwards.
[HOW]
Create a HW sequence for blanking HUBP.
1. Wait for enter VBLANK
2. Set HUBP_BLANK
3. Make sure HUBP_IN_BLANK = 1
4. Toggle HUBP_DISABLE on and off to perform soft reset
All existing calls to hubp->funcs->set_blank should be replaced with
this new sequence.
In wait_for_mpcc_disconnect, only blank the pipe being disconnected, and
leave all other pipes unmodified.
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Wesley Chalmers <Wesley.Chalmers@amd.com>
Reviewed-by: Jun Lei <Jun.Lei@amd.com>
Acked-by: Rodrigo Siqueira <Rodrigo.Siqueira@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
[why]
When OS reboots, and panel is turned off, T12 may not be maintained.
T12 is defined as the interval between VDDC off (occurs at shutdown) and
the next VDDC on (occurs when eDP is POST-ed)
[how]
DC already tracks panel power off time. Add a DC interface which DM can
call during shutdown. Ideally this should be as late as possible during
the shutdown sequence so the extra delay is minimal.
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Jun Lei <jun.lei@amd.com>
Reviewed-by: Aric Cyr <Aric.Cyr@amd.com>
Acked-by: Rodrigo Siqueira <Rodrigo.Siqueira@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
From Ard:
"Simply disabling -mgeneral-regs-only left and right is risky, given that
the standard AArch64 ABI permits the use of FP/SIMD registers anywhere,
and GCC is known to use SIMD registers for spilling, and may invent
other uses of the FP/SIMD register file that have nothing to do with the
floating point code in question. Note that putting kernel_neon_begin()
and kernel_neon_end() around the code that does use FP is not sufficient
here, the problem is in all the other code that may be emitted with
references to SIMD registers in it.
So the only way to do this properly is to put all floating point code in
a separate compilation unit, and only compile that unit with
-mgeneral-regs-only."
Disable support until the code can be properly refactored to support this
properly on aarch64.
Acked-by: Will Deacon <will@kernel.org>
Reported-by: Ard Biesheuvel <ardb@kernel.org>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
[WHY]
HUBP blanking sequence on DCN30 requires us to check if HUBP is in blank
and also toggle HUBP_DISABLE, which should instead be called
HUBP_SOFT_RESET for what it does in HW.
Signed-off-by: Wesley Chalmers <Wesley.Chalmers@amd.com>
Acked-by: Bindu Ramamurthy <bindu.r@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
[Why]
race condition of programming FP2 wrt pipe locking
and vactive state can cause underflow/black screen
[How]
Enforce the FP2 is only programmed during vactive,
and unlock pipe soon afterwards.
Signed-off-by: Martin Leung <martin.leung@amd.com>
Reviewed-by: Jun Lei <Jun.Lei@amd.com>
Acked-by: Qingqing Zhuo <qingqing.zhuo@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
[Why]
There is a warning that triggers when pstate takes too long.
Pstate can take up to ~200us on Linux without hanging but
it is currently set to 40us.
[How]
Change the timeout for the warning to be 180us on Linux.
Signed-off-by: Victor Lu <victorchengchi.lu@amd.com>
Reviewed-by: Roman Li <Roman.Li@amd.com>
Acked-by: Qingqing Zhuo <qingqing.zhuo@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
[Why]
power_down_on_boot is designed to power down HW when set mode is not
called before timeout. It can happen in headless system or booting with
the display is output by non-AMD GPU only.
The function will be executed only if it's not seamless boot. So in
seamless boot, the HW is still on.
It's not necessary to check this since there's no display data in both
cases.
[How]
Remove seamless boot checking in power_down_on_boot.
Signed-off-by: John Wu <john.wu@amd.com>
Reviewed-by: Anthony Koo <Anthony.Koo@amd.com>
Acked-by: Eryk Brol <eryk.brol@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
[Why]
The DSCL memory blocks should be powered down when they're not in use.
This will reduce power consumption.
[How]
1. Write to DSCL_MEM_PWR_FORCE to put memory to shutdown when DSCL is not
used.
2. Added a debug option to allow this behaviour to be turned off
Signed-off-by: Jacky Liao <ziyu.liao@amd.com>
Reviewed-by: Eric Yang <eric.yang2@amd.com>
Acked-by: Aurabindo Pillai <aurabindo.pillai@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
[Why]
Cursor is missing on displays 2 and up when doing rotated
display groups in 1xnY setup. Calculation puts cursor
out of bounds so it is not enabled.
[How]
In dcn10_set_cursor_position(), add in viewport.y into cursor
calculations for 1xnY rotated display groups.
For pipe split, check viewport.y for both pipes and use lower one
to normalize cursor position before calculations.
Add odm 2:1 support ( using same calculations as pipe split ).
Signed-off-by: Samson Tam <Samson.Tam@amd.com>
Acked-by: Bindu Ramamurthy <bindu.r@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
[Why]
When mpo enabled, video comes is 709. Desktop use sRGB.
So color change easily noticeable especially when switch between
mpo/non-mpo.
[How]
Force use sRGB for video TF is sRGB or BT709.
DCN1/DCN2 use predefined type with YUV.
DCN3 use distributed points type with YUV.
Signed-off-by: Jing Zhou <Jing.Zhou@amd.com>
Acked-by: Bindu Ramamurthy <bindu.r@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Avoids confusion in configurations.
v2: fix build when CONFIG_DRM_AMD_DC_DCN is disabled
v3: rebase on latest code
Reviewed-by: Luben Tuikov <luben.tuikov@amd.com> (v1)
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
This commit introduces a trace mechanism for struct pipe_ctx by adding a
middle layer struct in the amdgpu_dm_trace.h for capturing the most
important data from struct pipe_ctx and showing its data via tracepoint.
This tracepoint was added to dc.c and dcn10_hw_sequencer, however, it
can be added to other DCN architecture.
Signed-off-by: Rodrigo Siqueira <Rodrigo.Siqueira@amd.com>
Acked-by: Aurabindo Pillai <aurabindo.pillai@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
[Why]
We should leave GSL if we're not doing immediate flip no matter if
we're doing pipe split or not
[How]
Check for updating GSL state whenever we're not doing
immediate flip
v2: Squash in build fix (Alex)
Signed-off-by: Aric Cyr <aric.cyr@amd.com>
Signed-off-by: Alvin Lee <alvin.lee2@amd.com>
Acked-by: Aurabindo Pillai <aurabindo.pillai@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
[Why]
right side visual confirm is too thick due to it is 4 times of
left side (16 lines).
[How]
Change factor from 4 to 2.
Signed-off-by: Yongqiang Sun <yongqiang.sun@amd.com>
Acked-by: Aurabindo Pillai <aurabindo.pillai@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
[Why]
Underflow counter increases in AGM when performing some mode switches due
to timing sync, which is a known hardware issue.
[How]
Temporarily raise DPG height during timing sync so that underflow is not
reported.
Signed-off-by: Taimur Hassan <syed.hassan@amd.com>
Acked-by: Aurabindo Pillai <aurabindo.pillai@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
[Why]
Some sink devices wish to have access to the minimum
HBlank supported by the ASIC.
[How]
Make the ASIC minimum HBlank available in Source
Device information address 0x340.
Signed-off-by: Ashley Thomas <Ashley.Thomas2@amd.com>
Acked-by: Aurabindo Pillai <aurabindo.pillai@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
[Why]
When forcing 3D mode in DAL, we set the right address to be the same as the
left address. We need to do the same for the meta addresses.
[How]
Program right meta to be same as left meta.
Signed-off-by: Alvin Lee <alvin.lee2@amd.com>
Reviewed-by: Aric Cyr <Aric.Cyr@amd.com>
Acked-by: Eryk Brol <eryk.brol@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
[why]
HDCP 1.4 failed on SL8800 SW w/a test driver use.
[how]
Slow down the HW i2c speed when used by HW i2c.
This request: each acquired_i2c_engine setup the i2c speed needed
and sets the I2c engine for HDCP use at release_engine.
This covers SW using HW i2c engine and HDCP using HW i2c engine.
for dmcu using HW i2c engine, needs add similar logic in dmcu fw.
Signed-off-by: Charlene Liu <Charlene.Liu@amd.com>
Reviewed-by: Chris Park <Chris.Park@amd.com>
Acked-by: Eryk Brol <eryk.brol@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
[Why]
During driver unload, it is expected that p-state switching is supported.
If it's not supported, PMFW will hang due to a forced p-state switch. Even
if the current timing does not support p-state normally, we still want to
force allow because the worst that can happen is underflow. This will
match Navi10 behaviour.
[How]
- new hubbub func to control the force pstate register
- force allow when releasing display ownership
- registers are inaccessible after due to m_cgs.hwNotAvailable
- explicitly disable force signal during hw_init
- if driver is disabled and re-enabled, register not cleared otherwise
Also, remove DCN3 part of dcn10_init_hw, we will not be going back to it.
Signed-off-by: Joshua Aberback <joshua.aberback@amd.com>
Reviewed-by: Jun Lei <Jun.Lei@amd.com>
Acked-by: Eryk Brol <eryk.brol@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
[Why]
When running games or benchmarking with v-sync disabled, disabling
a plane (which is v-sync) can cause underflow. This is caused by
flips pending before pipe locking being applied after locks are
released and pipes could have been re-arranged or disconnected. This
could potentially apply a flip on incorrect pipe. Also, previous logic
of always locking pipes was unnecessary.
[How]
Only lock the pipes when there is a pipe being disabled to increase
efficiency. Before the pipes are locked, check that any pending flips
are cleared to ensure the flips are applied to the correct pipe.
Signed-off-by: Aric Cyr <aric.cyr@amd.com>
Acked-by: Eryk Brol <eryk.brol@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Update dc for vangogh support.
v2: fix compilation without DCN 301 set.
Signed-off-by: Roman Li <Roman.Li@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
[Why]
When running a game/benchmark with v-sync disabled, disabling a plane
(which is v-sync) can cause an underflow. This is due to flips that are
pending before pipe locking being applied after locks are released and
pipes have been re-arranged or disconnected. This can potentially apply
a flip on the incorrect pipe.
[How]
Check that any pending flips are cleared before locking any pipes to
ensure flips are applied on the correct pipes.
Signed-off-by: Taimur Hassan <syed.hassan@amd.com>
Reviewed-by: Aric Cyr <Aric.Cyr@amd.com>
Acked-by: Qingqing Zhuo <qingqing.zhuo@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
[WHY]
In DCN10 when a panel with YCbCr420 capability is connected via
USB-C to HDMI active dongle, no YCbCr420 option is listed in
Radeon settings.
[HOW]
Enable DP YCbCr420 mode support for DCN10
Signed-off-by: Gary Li <gary.li@amd.com>
Reviewed-by: Eric Yang <eric.yang2@amd.com>
Acked-by: Qingqing Zhuo <qingqing.zhuo@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
[WHY]
When disabling DP video, the current REG_WAIT timeout
of 50ms is too low for certain cases with very high
VSYNC intervals.
[HOW]
Increase the timeout to 102ms, so that
refresh rates as low as 10Hz can be handled properly.
Signed-off-by: Wesley Chalmers <Wesley.Chalmers@amd.com>
Reviewed-by: Aric Cyr <Aric.Cyr@amd.com>
Acked-by: Qingqing Zhuo <qingqing.zhuo@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Cc: <stable@vger.kernel.org>
[Why]
We dont's turn off backlight before power off eDP (VDD),
which is a violation of eDP specs.
[How]
Power off eDP backlight before power off eDP
Signed-off-by: Peikang Zhang <peikang.zhang@amd.com>
Reviewed-by: Anthony Koo <Anthony.Koo@amd.com>
Acked-by: Qingqing Zhuo <qingqing.zhuo@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
[Why]
The ctx field of dc_transfer_func is not always populated and therefore
isn't reliable.
[How]
Remove dc context from dc_transfer_func
Signed-off-by: Josip Pavic <Josip.Pavic@amd.com>
Acked-by: Aurabindo Pillai <aurabindo.pillai@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
[Why]
There are situations where we go from 2 pipe to 1 pipe in MPO, but this
isn't a pipe split being lost -- it's a plane disappearing in (i.e. video overlay
goes away) so we lose one pipe. In these situations we don't want to
disable the pipe in a separate operation from the rest of the pipe
programming sequence. We only want to disable a pipe in a
separate operation when we're actually disabling pipe split.
[How]
Make sure the pipe being lost has the same stream AND plane
as the old top pipe to ensure.
Signed-off-by: Alvin Lee <alvin.lee2@amd.com>
Acked-by: Aurabindo Pillai <aurabindo.pillai@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
[Why]
The function set_mst_bandwidth is poorly name since it isn't clear what
it does, and it also does not reflect any part of the allocation sequence
described in the DP spec.
[How]
Rename the function set_mst_bandwidth to set_throttled_vcp_size.
Signed-off-by: George Shen <george.shen@amd.com>
Acked-by: Aurabindo Pillai <aurabindo.pillai@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
[Why]
When pipe locks are being taken we wait for flip pending to clear first.
In some cases the pipe mapping is changed and the pending we're checking
for will never clear.
[How]
Don't check disabled pipes for flip pending.
Signed-off-by: Aric Cyr <aric.cyr@amd.com>
Acked-by: Aurabindo Pillai <aurabindo.pillai@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
[WHY]
update_clocks might not be called on headless adapters. This means
DISPLAY_OFF may not be sent in headless cases.
[HOW]
If hardware is powered down on boot because it is headless (mode set
does not happen on that adapter) also send DISPLAY_OFF notification.
Signed-off-by: Sung Lee <sung.lee@amd.com>
Reviewed-by: Yongqiang Sun <yongqiang.sun@amd.com>
Acked-by: Eryk Brol <eryk.brol@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
