[What]
vce ring test fails consistently during resume in s3 cycle, due to
mismatch read & write pointers.
On debug/analysis its found that rptr to be compared is not being
correctly updated/read, which leads to this failure.
Below is the failure signature:
[drm:amdgpu_vce_ring_test_ring] *ERROR* amdgpu: ring 12 test failed
[drm:amdgpu_device_ip_resume_phase2] *ERROR* resume of IP block <vce_v3_0> failed -110
[drm:amdgpu_device_resume] *ERROR* amdgpu_device_ip_resume failed (-110).
[How]
fetch rptr appropriately, meaning move its read location further down
in the code flow.
With this patch applied the s3 failure is no more seen for >5k s3 cycles,
which otherwise is pretty consistent.
V2: remove reduntant fetch of rptr
Signed-off-by: Louis Li <Ching-shih.Li@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Always mark evicted queues with q->properties.is_evicted = true, even
queues that are inactive for other reason. This simplifies maintaining
the eviction state as it doesn't require updating is_evicted when other
queue activation conditions change.
On the other hand, we now need to check those other queue activation
conditions whenever an evicted queues is restored. To minimize code
duplication, move the queue activation check into a macro so it can be
maintained in one central place.
Signed-off-by: Felix Kuehling <Felix.Kuehling@amd.com>
Reviewed-by: Philip Cox <Philip.Cox@amd.com>
Acked-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
EDC counts are related to instance and se. They are not the same
for different type of EDC. EDC clearing are changed to base on
individual EDC's instance and SE number.
Signed-off-by: James Zhu <James.Zhu@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
[Why]
DCN_VERSION_1_01 is no longer handled in the dal_hw_translate_init
switch since it was inadvertently dropped in the patch that removed the
unnecessary DCN1_01 guards.
This caused numerous regressions on DCN1_01 when loading the driver.
[How]
Add it back.
Fixes: 97df424fe7 ("drm/amd/display: Drop DCN1_01 guards")
Signed-off-by: Nicholas Kazlauskas <nicholas.kazlauskas@amd.com>
Acked-by: Alex Deucher <alexander.deucher@amd.com>
Reviewed-by: Harry Wentland <harry.wentland@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
[Why]
When switching from vt to desktop with EDID emulation we can receive
an atomic commit such that we have a crtc where mode_changed = true.
During the dm_update_crtc_state disable pass we remove the stream from
the context and free it on the dm_new_crtc_state.
During the enable pass we compare the new provisional stream to the
dm_old_crtc_state->stream and determine that the stream is unchanged
and no scaling has been changed.
Following this, new_crtc_state->mode_changed is then set to false.
The connectors haven't changed and the CRTC active state hasn't changed
so drm_atomic_crtc_needs_modeset returns false, so we jump to
skip_modeset and we hit:
BUG_ON(dm_new_crtc_state->stream == NULL);
...since the old stream is gone from the context and the new stream is
also still NULL.
[How]
Ensure that we still a stream to reuse before checking if we can reuse
the old stream without a full modeset.
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>
The messages about amdgpu_cs_list_validate are duplicated because the
caller will complain into the logs as well and we can also get
interrupted by a signal here.
Also fix the the caller to not report -EAGAIN from validation.
Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Chunming Zhou <david1.zhou@amd.com>
Tested-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Split late_init into two functions, one (do_late_init) which
just does the hw init, and late_init which calls do_late_init
and schedules the IB test work. Call do_late_init in
the GPU reset code to run the init code, but not schedule
the IB test code. The IB test code is called directly
in the gpu reset code so no need to run the IB tests
in a separate work thread. If we do, we end up racing.
v2: Rework late_init. Pull out the mgpu fan boost and xgmi
pstate code into late_init so they get called in all cases.
rename the late_init worker thread to delayed work since it's
just the IB tests now which can happen later. Schedule the
work at init and resume time. It's not needed at reset time
because the IB tests are called directly.
Reviewed-by: Christian König <christian.koenig@amd.com>
Cc: Xinhui Pan <xinhui.pan@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
gpu reset will run late_init and schedule the late_init_work. if we
keep triggering gpu reset in a short time, there are potenial races.
Signed-off-by: xinhui pan <xinhui.pan@amd.com>
Reviewed-by: Feifei Xu <Feifei.Xu@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
drm-misc-next for v5.3:
UAPI Changes:
Cross-subsystem Changes:
- Add devicetree bindings for new panels.
- Convert allwinner's DT bindings to a schema.
- Drop video/hdmi static functions from kernel docs.
- Discard old fence when reserving space in reservation_object_get_fences_rcu.
Core Changes:
- Add missing -ENOMEM handling in edid loading.
- Fix null pointer deref in scheduler.
- Header cleanups, making them self-contained.
- Remove drmP.h inclusion from core.
- Fix make htmldocs warning in scheduler and HDR metadata.
- Fix a few warnings in the uapi header and add a doc section for it.
- Small MST sideband error handling fix.
- Clarify userspace review requirements.
- Clarify implicit/explicit fencing in docs.
- Flush output polling on shutdown.
Driver Changes:
- Small cleanups to stm.
- Add new driver for ST-Ericsson MCDE
- Kconfig fix for meson HDMI.
- Add support for Armadeus ST0700 Adapt panel.
- Add KOE tx14d24vm1bpa panel.
- Update timings for st7701.
- Fix compile error in mcde.
- Big series of tc358767 fixes, and enabling support for IRQ and HPD handling.
- Assorted fixes to sii902x, and implementing HDMI audio support.
- Enable HDR metadata support on amdgpu.
- Assorted fixes to atmel-hlcdc, and add sam9x60 LCD controller support.
Signed-off-by: Dave Airlie <airlied@redhat.com>
From: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/6c43ffa9-11ff-5354-d772-c20fd4d1e3d9@linux.intel.com
[What]
vce ring test fails consistently during resume in s3 cycle, due to
mismatch read & write pointers.
On debug/analysis its found that rptr to be compared is not being
correctly updated/read, which leads to this failure.
Below is the failure signature:
[drm:amdgpu_vce_ring_test_ring] *ERROR* amdgpu: ring 12 test failed
[drm:amdgpu_device_ip_resume_phase2] *ERROR* resume of IP block <vce_v3_0> failed -110
[drm:amdgpu_device_resume] *ERROR* amdgpu_device_ip_resume failed (-110).
[How]
fetch rptr appropriately, meaning move its read location further down
in the code flow.
With this patch applied the s3 failure is no more seen for >5k s3 cycles,
which otherwise is pretty consistent.
V2: remove reduntant fetch of rptr
Signed-off-by: Louis Li <Ching-shih.Li@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Cc: stable@vger.kernel.org
[Why]
For userspace to send static HDR metadata to the display we need to
attach the property on the connector and send it to DC.
[How]
The property is attached to HDMI and DP connectors. Since the metadata
isn't actually available when creating the connector this isn't a
property we can dynamically support based on the extension block
being available or not.
When the HDR metadata is changed a modeset will be forced for now.
We need to switch from 8bpc to 10bpc in most cases anyway, and we want
to fully exit HDR mode when userspace gives us a NULL metadata, so this
isn't completely unnecessary.
The requirement can later be reduced to just entering and exiting HDR
or switching max bpc.
Cc: Harry Wentland <harry.wentland@amd.com>
Signed-off-by: Nicholas Kazlauskas <nicholas.kazlauskas@amd.com>
Signed-off-by: Harry Wentland <harry.wentland@amd.com>
Reviewed-by: Harry Wentland <harry.wentland@amd.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20190528190836.10738-2-nicholas.kazlauskas@amd.com
We are already doing this for DMA-buf imports and also for
amdgpu VM BOs for quite a while now.
If this doesn't run into any problems we are probably going
to stop removing BOs from the LRU altogether.
v2: drop BUG_ON from ttm_bo_add_to_lru
Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Chunming Zhou <david1.zhou@amd.com>
Tested-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Issue:
Will have follow error when reload driver:
[ 3986.567739] sysfs: cannot create duplicate filename '/devices/pci0000:00/0000:00:07.0/drm_dp_aux_dev'
[ 3986.567743] CPU: 6 PID: 1767 Comm: modprobe Tainted: G OE 5.0.0-rc1-custom #1
[ 3986.567745] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS Ubuntu-1.8.2-1ubuntu1 04/01/2014
[ 3986.567746] Call Trace:
......
[ 3986.567808] drm_dp_aux_register_devnode+0xdc/0x140 [drm_kms_helper]
......
[ 3986.569081] kobject_add_internal failed for drm_dp_aux_dev with -EEXIST, don't try to register things with the same name in the same directory.
Reproduce sequences:
1.modprobe amdgpu
2.modprobe -r amdgpu
3.modprobe amdgpu
Root cause:
When unload driver, it doesn't unregister aux.
v2: Don't use has_aux
Signed-off-by: Emily Deng <Emily.Deng@amd.com>
Reviewed-by: Nicholas Kazlauskas <nicholas.kazlauskas@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
[Why]
We only want to manual trigger end-of-frame when the bottom-most
pipe is flipped to prevent overlays from ending the frame too early.
[How]
Check that the manual trigger is only firing on bottom plane.
Signed-off-by: Aric Cyr <aric.cyr@amd.com>
Reviewed-by: Jun Lei <Jun.Lei@amd.com>
Acked-by: Leo Li <sunpeng.li@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
[Why]
There's some unnecessary mem allocation for CS_TFM_ID. What's worse, it
depends on LUT size and since it's 4K for CS_TFM_1D, it is 16x bigger
than in regular case when it's actually needed. This leads to some
crashes in stress conditions.
[How]
Skip ramp combining designed for RGB256 and DXGI gamma with CS_TFM_1D.
Signed-off-by: Krunoslav Kovac <Krunoslav.Kovac@amd.com>
Reviewed-by: Aric Cyr <Aric.Cyr@amd.com>
Acked-by: Leo Li <sunpeng.li@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
[Why]
For commits with allow_modeset=false and CRTC degamma changes the planes
aren't reset. This results in incorrect rendering.
[How]
Reset the planes when color management has changed on the CRTC.
Technically this will include regamma changes as well, but it doesn't
really after legacy userspace since those commit with
allow_modeset=true.
Signed-off-by: Nicholas Kazlauskas <nicholas.kazlauskas@amd.com>
Reviewed-by: Harry Wentland <Harry.Wentland@amd.com>
Acked-by: Leo Li <sunpeng.li@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
[Why]
Some backlight tests fail due to backlight settling
taking too long. This happens because the step
size used to change backlight levels is too small.
[How]
1. Change the size of the backlight gain step size
2. Change how DMCU firmware gets the step size value
so that it is passed in by driver during DMCU initn
Signed-off-by: Eryk Brol <eryk.brol@amd.com>
Reviewed-by: Jun Lei <Jun.Lei@amd.com>
Acked-by: Leo Li <sunpeng.li@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
[Why]
Some displays take some time to power up AUX CH once they are
put into D3 state via write to DPCD 600h=2.
Interestingly enough, some display may simply NACK, but some might
also ACK with a bunch of 0s, which can cause issues with receiver
cap retrieval. Note that not all DPCD address return 0s, but in
particular it has been observed on some higher DPCD address such
as DPCD 2200h, etc.
[How]
Based on spec, receiver will monitor differential signal while in D3 and
AUX CH is in low power mode. When detected, it may allow up to
1 ms to power up AUX CH and reply.
If we read Sink power state D3, we should add 1 ms delay to satisfy
this spec requirement.
Signed-off-by: Anthony Koo <anthony.koo@amd.com>
Reviewed-by: Wenjing Liu <Wenjing.Liu@amd.com>
Acked-by: Leo Li <sunpeng.li@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
[Why]
Some HW specific implementations can be pulled out into clk_mgr.c.
[How]
* Pull get_active_display_cnt out to clk_mgr.
* Pull out shared logic in set_dispclk and set_dprefclk
Signed-off-by: Eric Yang <Eric.Yang2@amd.com>
Reviewed-by: Tony Cheng <Tony.Cheng@amd.com>
Acked-by: Leo Li <sunpeng.li@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
[Why]
Modern Standby may toggle display adapter state between D0
and D3 state unpredictably.
But events that cause transition to D0 are not always resulting
in a display light up scenario.
Modern eDP panels should be able to power on panel logic
quickly upon VDD going high. Based on spec, the T3 time
between VDD on and HPD high can be between 0 and 80 ms.
Doing any tricky sorts of optimization by powering on panel
VDD early during D0 transition on can negatively impact other
features due to unnecessary power drain and toggling when
final system state does not intend for the panel to be lit up.
We need OEMs to source higher end panels that have T3 time
close to 0 if they want quick S3/Modern Standby resume times.
[How]
Remove panel VDD power on in init_hw
Signed-off-by: Anthony Koo <Anthony.Koo@amd.com>
Reviewed-by: Tony Cheng <Tony.Cheng@amd.com>
Acked-by: Leo Li <sunpeng.li@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
