commit 7a3727f385dc64773db1c144f6b15c1e9d4735bb upstream.
The SF and clipper units mishandle the provoking vertex in some cases,
which can cause misrendering with shaders that use flat shaded inputs.
There are chicken bits in 3D_CHICKEN3 (for SF) and FF_SLICE_CHICKEN
(for the clipper) that work around the issue. These registers are
unfortunately not part of the logical context (even the power context),
and so we must reload them every time we start executing in a context.
Bugzilla: https://bugs.freedesktop.org/103047
Signed-off-by: Kenneth Graunke <kenneth@whitecape.org>
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Link: https://patchwork.freedesktop.org/patch/msgid/20180615190605.16238-1-chris@chris-wilson.co.uk
Reviewed-by: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Cc: stable@vger.kernel.org
(cherry picked from commit b77422f80337d363eed60c8c48db9cb6e33085c9)
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
commit 5e9244ff585239630f15f8ad8e676bc91a94ca9e upstream.
Preparation for the following fix, no functional change intended.
Cc: stable@vger.kernel.org
Signed-off-by: Michel Dänzer <michel.daenzer@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
commit 6fa39bc1e01dab8b4f54b23e95a181a2ed5a2d38 upstream.
It can be quite big, and there's no need for it to be physically
contiguous. This is less likely to fail under memory pressure (has
actually happened while running piglit).
Cc: stable@vger.kernel.org
Signed-off-by: Michel Dänzer <michel.daenzer@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
commit 9fcf2b3c1c0276650fea537c71b513d27d929b05 upstream.
The statement always evaluates to true since the struct fields
are arrays. This has shown up as a warning when compiling with
clang:
warning: address of array 'desc->layout.xstride' will always
evaluate to 'true' [-Wpointer-bool-conversion]
Check for values in the first plane instead.
Fixes: 1a396789f65a ("drm: add Atmel HLCDC Display Controller support")
Cc: <stable@vger.kernel.org>
Signed-off-by: Stefan Agner <stefan@agner.ch>
Signed-off-by: Boris Brezillon <boris.brezillon@bootlin.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20180617084826.31885-1-stefan@agner.ch
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
commit 889ad63d41eea20184b0483e7e585e5b20fb6cfe upstream.
"qxl_bo_unref" may sleep, but calling "qxl_release_map" causes
"preempt_disable()" to be called and "preempt_enable()" isn't called
until "qxl_release_unmap" is used. Move the call to "qxl_bo_unref" out
from in between the two to avoid sleeping from an atomic context.
This issue can be demonstrated on a kernel with CONFIG_LOCKDEP=y by
creating a VM using QXL, using a desktop environment using Xorg, then
moving the cursor on or off a window.
Bugzilla: https://bugzilla.redhat.com/show_bug.cgi?id=1571128
Fixes: 9428088c90b6 ("drm/qxl: reapply cursor after resetting primary")
Cc: stable@vger.kernel.org
Signed-off-by: Jeremy Cline <jcline@redhat.com>
Link: http://patchwork.freedesktop.org/patch/msgid/20180601200532.13619-1-jcline@redhat.com
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
commit a0b2ac29415bb44d1c212184c1385a1abe68db5c upstream.
It missed vcn.fw_version setting when init vcn microcode, and it will be used to
report vcn ucode version via amdgpu_firmware_info sysfs interface.
Signed-off-by: Huang Rui <ray.huang@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Cc: stable@vger.kernel.org
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
commit 08ebb6e9f4fd7098c28e0ebbb42847cf0488ebb8 upstream.
1. fix set vce clocks failed on Cz/St
which lead 1s delay when boot up.
2. remove the workaround in vce_v3_0.c
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Reviewed-by: Huang Rui <ray.huang@amd.com>
Acked-by: Christian König <christian.koenig@amd.com>
Acked-by: Shirish S <shirish.s@amd.com>
Signed-off-by: Rex Zhu <Rex.Zhu@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Cc: stable@vger.kernel.org
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
commit 819a23f83e3b2513cffbef418458a47ca02c36b3 upstream.
fix the issue set uvd clock failed on CZ/ST
which lead 1s delay when boot up.
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Reviewed-by: Huang Rui <ray.huang@amd.com>
Acked-by: Christian König <christian.koenig@amd.com>
Acked-by: Shirish S <shirish.s@amd.com>
Signed-off-by: Rex Zhu <Rex.Zhu@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Cc: stable@vger.kernel.org
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
When irq is preinstalled, IRQ_NOAUTOEN will block irq for power event,
and if there was already a power enable event before, the next power
disable event after irq install will cause irq depth equals to 2 and
the next power enable event will not be able to enable the actual irq.
The change will check previous power event status, and skip IRQ_NOAUTOEN
if power is already enabled before irq install.
Change-Id: I20a0e65ef729e006c669e737e61fe1432090328e
Signed-off-by: Xiaowen Wu <wxiaowen@codeaurora.org>
Change frees tx_buf after sending the pps command. Without
this memory was leaking for everytime pps command was sent.
CRs-Fixed: 2211837
Change-Id: I081f958075e7fdf863ae5647fcbd142e91f6e59b
Signed-off-by: Vara Reddy <varar@codeaurora.org>
When ESD recovery is underway, there is high likelihood
of a ping pong timeout which might trigger a TE check on panel.
Avoid this check as panel is already in bad state.
Change-Id: I14aa6b70df928cc986b907ad98838ade000a89e3
Signed-off-by: Shashank Babu Chinta Venkata <sbchin@codeaurora.org>
DSI controller can trigger spurious error interrupts
during ESD attack scenarios. Add a check for such spurious
interrupts during error handling.
Change-Id: Iab86e34a7d94991de5c2ad1c62404f1dee1c81b6
Signed-off-by: Shashank Babu Chinta Venkata <sbchin@codeaurora.org>
In the current implementation dsi driver is always sending the
status check command in HS mode. This change fixes the issue by
setting the proper state based on panel device tree entry, before
sending the status check command from dsi host.
Change-Id: I9e8ed8616295de2a2f1d5fa2a8a83b706c70f033
Signed-off-by: Sandeep Panda <spanda@codeaurora.org>
Signed-off-by: Shashank Babu Chinta Venkata <sbchin@codeaurora.org>
In ESD recovery sequence, there is high likelihood of encountering
DSI FIFO overflow/underflow errors which might put controller in
bad state. This change disables FIFO error interrupts prior to
attempting panel status read.
Change-Id: Ib4245241789f74b51ef06778df63cbae572f60e0
Signed-off-by: Shashank Babu Chinta Venkata <sbchin@codeaurora.org>
During ESD recovery, we cannot ensure health of DSI
PHY link. Thus, sending ESD panel off commands may
subject to fail. Avoid sending panel off commands
when ESD recovery is underway.
Change-Id: I443c8b1044ad68ae3fc6e4b2a904134970ec8c18
Signed-off-by: Shashank Babu Chinta Venkata <sbchin@codeaurora.org>
Some display panel have the requirement of waiting for certain
duration before dsi host can read back the response from panel,
to a DCS read command. This change adds the support to store
the delay required in dsi message structure it self.
Change-Id: I82f53d811f82195eb900e4594b01c6e0f23fed53
Signed-off-by: Sandeep Panda <spanda@codeaurora.org>
Signed-off-by: Shashank Babu Chinta Venkata <sbchin@codeaurora.org>
Currently to synchronize ESD check and DSI disable, ESD
thread acquires display mutex before the check starts.
But this is causing performance issues when atomic check
also tries to acquire the display mutex and blocks on ESD
thread if the later has already acquired it.
Fix this race condition by removing display mutex lock from
ESD thread and instead synchronize ESD check and DSI disable
by always ensuring that ESD check has been stopped before
DSI disable sequence kicks in.
Change-Id: I5a0b44a7cbf01648303eb85c7a08a6a62c1ef94e
Signed-off-by: Sandeep Panda <spanda@codeaurora.org>
Signed-off-by: Shashank Babu Chinta Venkata <sbchin@codeaurora.org>
There are possibilities of deadlock if esd trigger work is
queued to display thread. There is a corner case where
during bridge disable, display thread is waiting for esd
trigger work to complete to cancel esd thread, while esd
trigger work is blocked on display thread. To fix this
queue esd trigger work to event thread.
Below is the call stack of both threads:
ESD thread
----------
__switch_to+0xb8/0xec
kthread_flush_work+0xdc/0x150
sde_encoder_display_failure_notification+0xc0/0x1e8
_sde_connector_report_panel_dead.part.8+0x6c/0xd4
sde_connector_check_status_work+0x10c/0x1d4
process_one_work+0x168/0x468
worker_thread+0x140/0x460
kthread+0xf4/0x108
ret_from_fork+0x10/0x50
0xffffffffffffffff
Commit thread
-------------
__switch_to+0xb8/0xec
flush_work+0x44/0x68
__cancel_work_timer+0x138/0x1e0
cancel_delayed_work_sync+0x24/0x30
sde_connector_schedule_status_work.part.16+0xc0/0x114
_sde_connector_update_power_locked+0x1c4/0x224
_sde_connector_update_dirty_properties+0x80/0x100
sde_connector_helper_bridge_disable+0x24/0x100
dsi_bridge_disable+0x30/0x98
drm_bridge_disable+0x128/0x144
msm_atomic_helper_commit_modeset_disables+0x148/0x578
complete_commit+0x74/0x668
_msm_drm_commit_work_cb+0x4c/0x1c4
kthread_worker_fn+0xc0/0x1e4
kthread+0xf4/0x108
ret_from_fork+0x10/0x50
0xffffffffffffffff
Change-Id: Ifd95664113857fd5c06098a3a2f87f28c1975513
Signed-off-by: Jayant Shekhar <jshekhar@codeaurora.org>
In case of ESD recovery display disable/enable happens
without the knowledge of backlight framework. So backlight
does not change during this period. This causes issues
where panel used DCS backlight update method where backlight
is lost during ESD trigger and it is not set again after
recovery is complete. This change fixes the issue by
calling backlight update API during ESD recovery display
enable/disable case.
Change-Id: I157120688603dd0558b773bbf41996b6d7373d94
Signed-off-by: Sandeep Panda <spanda@codeaurora.org>
Signed-off-by: Shashank Babu Chinta Venkata <sbchin@codeaurora.org>
There can be scenarios where display panel might not send TE signal
at a regular intervals. It may take more time than what kms driver
can wait for and will be fatal. In those use cases add recovery mechanism
to reset the panel and display subsystem.
Change-Id: I8c28f6212b7fc68073d84b274cef92e670e0352e
Signed-off-by: Sandeep Panda <spanda@codeaurora.org>
Signed-off-by: Shashank Babu Chinta Venkata <sbchin@codeaurora.org>
Parse dsc version and dsc scr version supported by panel
and send the pps command accordingly.
CRs-Fixed: 2202007
Change-Id: I680cc50cb5ae6817f600a8b3eb37f60aab741b4b
Signed-off-by: Vara Reddy <varar@codeaurora.org>
Signed-off-by: Shashank Babu Chinta Venkata <sbchin@codeaurora.org>
During state checks in dsi control, enable state machine error
logs by default to catch rarely reproducible issues.
CRs-Fixed: 2170957
Change-Id: I1dd3afeba8bb023e427a4ed15b40d3758b53e885
Signed-off-by: Vara Reddy <varar@codeaurora.org>
Signed-off-by: Shashank Babu Chinta Venkata <sbchin@codeaurora.org>
Fix the dsi_host_transfer API to accurately return the error
code in case of failures.
CRs-Fixed: 2191069
Change-Id: I54503ca2f2fc02bf2374abae5147b8c1f13eff02
Signed-off-by: Vara Reddy <varar@codeaurora.org>
Signed-off-by: Shashank Babu Chinta Venkata <sbchin@codeaurora.org>
Send 2a/2b roi commands when display resolution is changed.
This is fixing panel corruption issue if resolution switch
command list doesn't have the new roi region updated.
Change-Id: I5244424e5dff2d07b88d52290365f3ad600eeb13
CRs-Fixed: 2166948
Signed-off-by: Vara Reddy <varar@codeaurora.org>
Signed-off-by: Shashank Babu Chinta Venkata <sbchin@codeaurora.org>
Frame drops are observed when crtc_commit thread tries to hold panel
lock for updating the frame, while backlight service thread is
pushing the backlight commands holding panel_lock. Remove holding
panel lock from commit thread.
CRs-Fixed: 2179313
Change-Id: I7d6e5044db570102b7c2c7edb1c5c4b24847ed3f
Signed-off-by: Vara Reddy <varar@codeaurora.org>
Signed-off-by: Shashank Babu Chinta Venkata <sbchin@codeaurora.org>
Fix issue around checking esd register read back value with
the expected value from the dtsi entries.
Change-Id: I8aeccc3acb61f10c3141e42a15eb6c674af251e6
CRs-Fixed: 2183397
Signed-off-by: Vara Reddy <varar@codeaurora.org>
Signed-off-by: Shashank Babu Chinta Venkata <sbchin@codeaurora.org>
In the current implementation when ESD check needs
to send multiple read commands then the last command
flag is only set for first read command. So the
transfer of rest of the read commands will fail. This
changes fixes all the issues related to multiple
reg read feature.
Change-Id: If783c9e7bc15630358d2cb1f6cd5a6b3320fbf6d
Signed-off-by: Sandeep Panda <spanda@codeaurora.org>
Signed-off-by: Shashank Babu Chinta Venkata <sbchin@codeaurora.org>
Currently DSI driver is only considering command mode panel
for getting the number of data lanes enabled, while waiting
for lanes to be idle. So in video mode the wait is failing
because of mismatch in number of data lanes. Also when
coming out of ulps in suspend we should not configure the
DSI controller twice, which may put DSI HW in bad state.
This change fixes the same for ulps during suspend feature.
Change-Id: I090e7d726759a8317dee7941c1067207c986b98b
Signed-off-by: Shashank Babu Chinta Venkata <sbchin@codeaurora.org>
Signed-off-by: Sandeep Panda <spanda@codeaurora.org>
Ensure that the escape clock is turned on prior to reinitializing
the DSI PHY when exiting idle power collapse. This will ensure that
there are no spurious transitions on the lanes after the PHY clamps
are removed.
Change-Id: Ic7ce1af62581b71d2a610b18381c9c57eb80696c
Signed-off-by: Aravind Venkateswaran <aravindh@codeaurora.org>
Separate DSI high speed (HS) clocks from low power (LP) clock, so that
we can have better control in enabling and disabling these clocks
independently. Some DSI hardware versions need LP clocks to be turned on
before programming DSI PHY.
Change-Id: I05efde2bb2a7f737e2f952554b41e3ab74e990e5
Signed-off-by: Aravind Venkateswaran <aravindh@codeaurora.org>
If external DSI bridge has implemented connector functions, we
allow empty timing information in DSI panel device tree file.
In this mode, all the timing information comes from external
bridge, e.g. from EDID, and DSI controller will be configured
accordingly.
If timing nodes are defined in DSI panel, they will overwrite
the modes from external bridge. This mode is useful when no EDID
or wrong EDID are reported, or dedicated mode is selected.
CRs-Fixed: 2268280
Change-Id: I6a929d7f3f7a6bbeed04f61a317c5c07f4643aa8
Signed-off-by: Xiaowen Wu <wxiaowen@codeaurora.org>
Register power domain before smmu init to trigger extra probe attempt
for any device that depends on mdp power domain registeration.
This fixed the issue that sde_rotator is not probed during boot.
CRs-Fixed: 2270743
Change-Id: Ia802ee3bf069b8084fc61db3fcbc62dc5934d021
Signed-off-by: Xiaowen Wu <wxiaowen@codeaurora.org>
