Currently drivers get called to move a buffer, but if they have to
move it temporarily through another space (SYSTEM->VRAM via TT)
then they can end up with a lot of ttm->driver->ttm call stacks,
if the temprorary space moves requires eviction.
Instead of letting the driver do all the placement/space for the
temporary, allow it to report back (-EMULTIHOP) and a placement (hop)
to the move code, which will then do the temporary move, and the
correct placement move afterwards.
This removes a lot of code from drivers, at the expense of
adding some midlayering. I've some further ideas on how to turn
it inside out, but I think this is a good solution to the call
stack problems.
v2: separate out the driver patches, add WARN for getting
MULTHOP in paths we shouldn't (Daniel)
v3: use memset (Christian)
Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Reviewed-by: hristian König <christian.koenig@amd.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20201109005432.861936-2-airlied@gmail.com
For sinks that do not support audio, there is no need to notify
audio subsystem of the connection event.
This will make sure that audio routes only to the primary display
when connected to such sinks.
changes in v2:
- Added fixes tag
- Removed nested if condition and removed usage of global pointer
Fixes: d13e36d7d2 ("drm/msm/dp: add audio support for Display Port on MSM")
Signed-off-by: Abhinav Kumar <abhinavk@codeaurora.org>
Signed-off-by: Rob Clark <robdclark@chromium.org>
Some dongles require link training done at irq_hpd request instead
of plugin request. This patch promote irq_hpd handler to handle link
training and setup hpd_state correctly.
Signed-off-by: Kuogee Hsieh <khsieh@codeaurora.org>
Signed-off-by: Rob Clark <robdclark@chromium.org>
Some dongle will not clear LINK_STATUS_UPDATED bit after
DPCD read which cause link training failed. This patch
just read 6 bytes of DPCD link status from sink and return
without checking LINK_STATUS_UPDATED bit.
Only 8 bits are used to represent link rate at sinker DPCD.
The really link rate is 2.7Mb times the 8 bits value.
For example, 0x0A at DPCD is equal to 2.7Gb (10 * 2.7Mb).
This patch also convert 8 bits value of DPCD to really link
rate to fix worng link rate error during phy compliance test.
Fixes: 6625e2637d ("drm/msm/dp: DisplayPort PHY compliance tests fixup")
Signed-off-by: Kuogee Hsieh <khsieh@codeaurora.org>
Signed-off-by: Rob Clark <robdclark@chromium.org>
DP compo phy have to be enable to start link training. When
link training failed phy need to be disabled so that next
link traning can be proceed smoothly at next plug in. This
patch de-initialize mainlink to disable phy if link training
failed. This prevent system crash due to
disp_cc_mdss_dp_link_intf_clk stuck at "off" state. This patch
also perform checking power_on flag at dp_display_enable() and
dp_display_disable() to avoid crashing when unplug cable while
display is off.
Signed-off-by: Kuogee Hsieh <khsieh@codeaurora.org>
Signed-off-by: Rob Clark <robdclark@chromium.org>
Connection state is not set correctly happen when either failure of link
train due to cable unplugged in the middle of aux channel reading or
cable plugged in while in suspended state. This patch fixes these problems.
This patch also replace ST_SUSPEND_PENDING with ST_DISPLAY_OFF.
Changes in V2:
-- Add more information to commit message.
Changes in V3:
-- change base
Changes in V4:
-- add Fixes tag
Signed-off-by: Kuogee Hsieh <khsieh@codeaurora.org>
Signed-off-by: Rob Clark <robdclark@chromium.org>
During suspend, dp host controller and hpd block are disabled due to
both ahb and aux clock are disabled. Therefore hpd plug/unplug interrupts
will not be generated. At dp_pm_resume(), reinitialize both dp host
controller and hpd block so that hpd plug/unplug interrupts will be
generated and handled by driver so that hpd connection state is updated
correctly. This patch will fix link training flaky issues.
Changes in v2:
-- use container_of to cast correct dp_display_private pointer
at both dp_pm_suspend() and dp_pm_resume().
Changes in v3:
-- replace hpd_state atomic_t with u32
Changes in v4
-- call dp_display_host_deinit() at dp_pm_suspend()
-- call dp_display_host_init() at msm_dp_display_enable()
-- fix phy->init_count unbalance which causes link training failed
Changes in v5
-- add Fixes tag
Fixes: 8ede2ecc3e (drm/msm/dp: Add DP compliance tests on Snapdragon Chipsets)
Tested-by: Stephen Boyd <swboyd@chromium.org>
Signed-off-by: Kuogee Hsieh <khsieh@codeaurora.org>
Signed-off-by: Rob Clark <robdclark@chromium.org>
idr_init() uses base 0 which is an invalid identifier for this driver.
The new function idr_init_base allows IDR to set the ID lookup from
base 1. This avoids all lookups that otherwise starts from 0 since
0 is always unused.
References: commit 6ce711f275 ("idr: Make 1-based IDRs more efficient")
Signed-off-by: Deepak R Varma <mh12gx2825@gmail.com>
Signed-off-by: Thierry Reding <treding@nvidia.com>
In function amdgpu_dm_connector_get_modes, drm_edid_is_valid
will check weather (!edid), no need to check again in the if
branch.
Reviewed-by: Harry Wentland <harry.wentland@amd.com>
Signed-off-by: Bernard Zhao <bernard@vivo.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
So we print the proper SMU firmware version in debugfs, ioctls, etc.
for vangogh.
Reviewed-by: Evan Quan <evan.quan@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
So we print the proper SMU firmware version in debugfs, ioctls, etc.
for Renoir.
Reviewed-by: Evan Quan <evan.quan@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Also rid some unused ones.
This patch solves 2000 warnings!
Fixes the following W=1 kernel build warning(s):
In file included from drivers/gpu/drm/amd/amdgpu/../display/dc/dc_types.h:33,
from drivers/gpu/drm/amd/amdgpu/../display/dc/dm_services_types.h:30,
from drivers/gpu/drm/amd/amdgpu/../display/dc/dm_services.h:35,
from drivers/gpu/drm/amd/amdgpu/../display/amdgpu_dm/amdgpu_dm_helpers.c:35:
At top level:
drivers/gpu/drm/amd/amdgpu/../display/include/fixed31_32.h:76:32: warning: ‘dc_fixpt_ln2_div_2’ defined but not used [-Wunused-const-variable=]
76 | static const struct fixed31_32 dc_fixpt_ln2_div_2 = { 1488522236LL };
| ^~~~~~~~~~~~~~~~~~
drivers/gpu/drm/amd/amdgpu/../display/include/fixed31_32.h:75:32: warning: ‘dc_fixpt_ln2’ defined but not used [-Wunused-const-variable=]
75 | static const struct fixed31_32 dc_fixpt_ln2 = { 2977044471LL };
| ^~~~~~~~~~~~
drivers/gpu/drm/amd/amdgpu/../display/include/fixed31_32.h:74:32: warning: ‘dc_fixpt_e’ defined but not used [-Wunused-const-variable=]
74 | static const struct fixed31_32 dc_fixpt_e = { 11674931555LL };
| ^~~~~~~~~~
drivers/gpu/drm/amd/amdgpu/../display/include/fixed31_32.h:73:32: warning: ‘dc_fixpt_two_pi’ defined but not used [-Wunused-const-variable=]
73 | static const struct fixed31_32 dc_fixpt_two_pi = { 26986075409LL };
| ^~~~~~~~~~~~~~~
drivers/gpu/drm/amd/amdgpu/../display/include/fixed31_32.h:72:32: warning: ‘dc_fixpt_pi’ defined but not used [-Wunused-const-variable=]
72 | static const struct fixed31_32 dc_fixpt_pi = { 13493037705LL };
| ^~~~~~~~~~~
Cc: Harry Wentland <harry.wentland@amd.com>
Cc: Leo Li <sunpeng.li@amd.com>
Cc: Alex Deucher <alexander.deucher@amd.com>
Cc: "Christian König" <christian.koenig@amd.com>
Cc: David Airlie <airlied@linux.ie>
Cc: Daniel Vetter <daniel@ffwll.ch>
Cc: amd-gfx@lists.freedesktop.org
Cc: dri-devel@lists.freedesktop.org
Signed-off-by: Lee Jones <lee.jones@linaro.org>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
- When we are under SRIOV setup, the rev_id cannot be read
properly. Therefore, we will return default value for it
Signed-off-by: Bokun Zhang <Bokun.Zhang@amd.com>
Reviewed-by: Monk Liu <monk.liu@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
- The original logic uses a counter based index assignment,
which is incorrect if we only assign VCN1 to this VF but no VCN0
The doorbell index is absolute, so we can calculate it by
using index variable i and j
Signed-off-by: Bokun Zhang <Bokun.Zhang@amd.com>
Reviewed-by: Monk Liu <monk.liu@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
[Why]
4K monitor shows corruption if dpp dto phase is larger than modulo.
[How]
cap phase value never larger than modulo.
Signed-off-by: Yongqiang Sun <yongqiang.sun@amd.com>
Acked-by: Bindu Ramamurthy <bindu.r@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
[Why]
Actual dpp/disp clock PMFW set are always equal or more than requested.
Need to cache the actual clock for debugging purpose.
[How]
Add actuall dpp/disp clk in clock manager and save it when set clocks.
Add an ASSERT in case of actual clock less than requested.
Signed-off-by: Yongqiang Sun <yongqiang.sun@amd.com>
Acked-by: Bindu Ramamurthy <bindu.r@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
[Why]
We don't want GSL to be enabled when only updating
plane address
[How]
Only check other pipes for immediate flip if GSL is
already enabled
Signed-off-by: Alvin Lee <alvin.lee2@amd.com>
Acked-by: Bindu Ramamurthy <bindu.r@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
[why]
On DCE, the cursor is not being disabled on commits where there is no
CRTC or FB enabled.
[how]
Explicitly disable the cursor in the commit tail only if the pending
commit disables the CRTC.
Signed-off-by: Victor Lu <victorchengchi.lu@amd.com>
Acked-by: Bindu Ramamurthy <bindu.r@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
[Why & How]
Program modulo with ref dpp clk Mhz/10.
Program phase with pipe dpp clk Mhz /10.
DMUB FW could use these value to determine optimization clk
for PSR power saving.
Signed-off-by: Yongqiang Sun <yongqiang.sun@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>
[Why]
We want to make sure all immediate flips are completed before locking
the pipes, but not pipes that are not flip immediate as they will be
locked by the OTG
[How]
Skip non flip immediate pipes when checking for flip pending.
Signed-off-by: Aric Cyr <aric.cyr@amd.com>
Acked-by: Bindu Ramamurthy <bindu.r@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
[Why & How]
The intended use is to force PSR into active state and ignore all
events until explicit EXIT.
A new event force_static is added to power module. It is then sent
to FW.
Signed-off-by: Krunoslav Kovac <Krunoslav.Kovac@amd.com>
Acked-by: Bindu Ramamurthy <bindu.r@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
[Why]
Due to the freesync power optimization increasing vtotal, the driver
believes that MPO is supported. MPO is turned on. Freesync then
exits, MPO can no longer be supported and immediate flip fails. This
causes corruption on the panel.
[How]
Avoid the freesync optimization when doing validation so in the case
freesync exits, driver can still support the configuration.
Signed-off-by: Isabel Zhang <isabel.zhang@amd.com>
Acked-by: Bindu Ramamurthy <bindu.r@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
[Why]
In some cases, prefetch mode 0 is unsupported but prefetch 1 is
supported. Due to previous change always forcing prefetch mode to 0, we
are failing bandwidth validation in cases where we should not.
[How]
By default try prefetch mode 0 but in the case validation fails, attempt
to do prefetch mode 1 to see if it is supported.
Signed-off-by: Isabel Zhang <isabel.zhang@amd.com>
Acked-by: Bindu Ramamurthy <bindu.r@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
There is a NULL pointer crash when DCN disabled on headless SKU.
On normal SKU, the variable adev->ddev.mode_config.funcs is
initialized in dm_hw_init(), and it is fine to access it in
amdgpu_device_resume(). But on headless SKU, DCN is disabled,
the funcs variable is not initialized, then crash arises.
Enable DCN to fix this issue.
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Reviewed-by: Guchun Chen <guchun.chen@amd.com>
Signed-off-by: Tianci.Yin <tianci.yin@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
[Why]
Prevent compiler warnings when building amdgpu kernel module
[How]
Add the include directive for amdgpu_amdkfd_gfx_v9 header file
Fixes: aeee2a48ec ("drm/amd/amdgpu: Enable arcturus devices to access the method kgd_gfx_v9_get_cu_occupancy that is already defined")
Reviewed-by: Harish Kasiviswanathan <Harish.Kasiviswanathan@amd.com>
Signed-off-by: Ramesh Errabolu <Ramesh.Errabolu@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Reported-by: kernel test robot <lkp@intel.com>
Currently we are using dcn3 reg headers for dcn302. The offsets are
different between the two so they need seperate headers.
Add dcn302 header files and use these instead of dcn3 header
Signed-off-by: Bhawanpreet Lakha <Bhawanpreet.Lakha@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Similar to the previous patch, clear shadow on suspend to avoid timeouts
waiting for ringbuffer space.
Fixes: 8907afb476 ("drm/msm: Allow a5xx to mark the RPTR shadow as privileged")
Signed-off-by: Rob Clark <robdclark@chromium.org>
Clear the shadow rptr on suspend. Otherwise, when we resume, we can
have a stale value until CP_WHERE_AM_I executes. If we suspend near
the ringbuffer wraparound point, this can lead to a chicken/egg
situation where we are waiting for ringbuffer space to write the
CP_WHERE_AM_I (or CP_INIT) packet, because we mistakenly believe that
the ringbuffer is full (due to stale rptr value in the shadow).
Fixes errors like:
[drm:adreno_wait_ring [msm]] *ERROR* timeout waiting for space in ringbuffer 0
in the resume path.
Fixes: d3a569fccf ("drm/msm: a6xx: Use WHERE_AM_I for eligible targets")
Signed-off-by: Rob Clark <robdclark@chromium.org>
commit f8f6ae5d07 ("mm: always have io_remap_pfn_range() set
pgprot_decrypted()") moves the pgprot_decrypted() into
io_remap_pfn_range(). Delete any, now confusing, open coded calls that
directly precede io_remap_pfn_range():
- drm_io_prot() is only in drm_mmap_locked() to call io_remap_pfn_range()
- fb_mmap() immediately calls vm_iomap_memory() which is a convenience
wrapper for io_remap_pfn_range()
Signed-off-by: Jason Gunthorpe <jgg@nvidia.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Link: https://patchwork.freedesktop.org/patch/msgid/0-v1-2e6a0db57868+166-drm_sme_clean_jgg@nvidia.com
The Intel Keem Bay display controller is only present on Intel Keem Bay
SoCs. Hence add a dependency on ARCH_KEEMBAY, to prevent asking the
user about this driver when configuring a kernel without Intel Keem Bay
platform support.
Note that:
1. The dependency on ARM is dropped, as Keem Bay SoCs are only
supported in arm64 kernel builds,
2. The dependencies on OF and COMMON_CLK can be dropped for
compile-testing, as the driver builds fine regardless.
Fixes: ed794057b0 ("drm/kmb: Build files for KeemBay Display driver")
Signed-off-by: Geert Uytterhoeven <geert+renesas@glider.be>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Link: https://patchwork.freedesktop.org/patch/msgid/20201110144350.3279147-1-geert+renesas@glider.be
If CONFIG_DRM_MIPI_DSI=n:
aarch64-linux-gnu-ld: drivers/gpu/drm/kmb/kmb_dsi.o: in function `kmb_dsi_host_unregister':
kmb_dsi.c:(.text+0xa48): undefined reference to `mipi_dsi_host_unregister'
aarch64-linux-gnu-ld: drivers/gpu/drm/kmb/kmb_dsi.o: in function `kmb_dsi_host_bridge_init':
kmb_dsi.c:(.text+0xb14): undefined reference to `mipi_dsi_host_register'
Fix this be selecting DRM_MIPI_DSI, like other drivers do.
Fixes: ed794057b0 ("drm/kmb: Build files for KeemBay Display driver")
Signed-off-by: Geert Uytterhoeven <geert+renesas@glider.be>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Link: https://patchwork.freedesktop.org/patch/msgid/20201110144219.3278831-1-geert+renesas@glider.be
We need commit f8f6ae5d07 ("mm: always have io_remap_pfn_range() set
pgprot_decrypted()") to be able to merge Jason's cleanup patch.
Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
