[Why]
The DISPCLK value was previously requested to be 15% higher for all
ASICS that went through the dce110 bandwidth code path. As part of a
refactoring of dce_clocks and dce110 set_bandwidth this was removed
for power saving considerations.
This changed caused corruption under certain display configurations.
Originally thought to be Vega specific, it was also observed on Polaris.
[How]
The 15% is brought back but its placement differs from the original
patch. This boost should only be enable while DFS bypass is inactive.
This (like the Vega patch) is also a workaround that should be
removed after the root cause is identified.
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]
Due to a small pre-fetch window, the active vline timing is a couple
of lines off when compared to what it should be.
[How]
Changed the calculation for the start vline to account for this window.
Signed-off-by: Murton Liu <murton.liu@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>
[Description] interlace mode shows wrong vertical timing.
Interface timing in Edid is half vertical timing as progressive timing.
driver doubled the vertical timing in edid_paser,
no need to double in optc again.
Signed-off-by: Charlene Liu <charlene.liu@amd.com>
Reviewed-by: Chris Park <Chris.Park@amd.com>
Acked-by: Leo Li <sunpeng.li@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
[Why]
There is no reason why the common data needs to be kept separate.
[How]
Flatten the struct by moving common data into the DM IRQ struct.
Signed-off-by: Leo Li <sunpeng.li@amd.com>
Reviewed-by: David Francis <David.Francis@amd.com>
Acked-by: Leo Li <sunpeng.li@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
[why]
ddc engines were recently changed to be independently tracked
from pipe count. the change was reflected in resource constructor
but not in destructor. this manifests as a memory leak when
pipe harvesting is enabled, since not all constructed ddc engines
are freed
[how]
make destructor symmetric with constructor for all dcX_resource
Signed-off-by: Jun Lei <Jun.Lei@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]
DF keeps awake after S0i3 resume due to DRAM_STATE_CNTL
is set by bios command table during dcn init_hw.
[How]
As a work around, check STATE_CNTL status before init_hw,
if it is 0 before init_hw and set to 1 after init_hw,
change it to 0.
Signed-off-by: Yongqiang Sun <yongqiang.sun@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]
All ASICS we support has YCbCr support, so
the check is unnecessary, the currently logic
in validate output also returns true all
the time, so the unneccessary logic is removed
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]
Previously bandwidth validation was failing because swizzle mode was not
initialized during plane_state allocation. The swizzle mode was
calculated using pixed format which is how swizzle mode is initially
calculated in addrlib.
[How]
* Set default swizzle mode for validation to DC_SW_UNKNOWN
* Created new function in dcn10_assign_swizzle_mode which sets the
plane swizzle mode based on selected pixed format
* Added the call of assign_swizzle_mode into dc_validate_global_state
* Set failsafe swizzle mode back to DC_SW_LINEAR
Signed-off-by: Su Sung Chung <Su.Chung@amd.com>
Reviewed-by: Eric Yang <eric.yang2@amd.com>
Acked-by: Leo Li <sunpeng.li@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
[why]
Same functions exist for all other signal types.
[how]
Add a function that checks against virtual signal type.
Signed-off-by: Nikola Cornij <nikola.cornij@amd.com>
Reviewed-by: Leo Li <sunpeng.li@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
[why]
Sometimes there are indications that the incorrect driver is being
loaded in automated tests. This change adds the ability for builds to
be tagged with a string, and picked up by the test infrastructure.
[how]
dc.c will allocate const for build id, which is init-ed with default
value, indicating production build. For test builds, build server will
find/replace this value. The test machine will then verify this value.
Signed-off-by: Jun Lei <Jun.Lei@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>
The minRPM speed maybe equal to zero. so need to check
input RPM not equal to 0, otherwise cause divide-by-zero driver crash.
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Rex Zhu <Rex.Zhu@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Count all GPU instances from AMD(including iGPUs and
dGPUs) in the system.
V2: drop unnecessary initialization for other gpu_info
members except mutex
Signed-off-by: Evan Quan <evan.quan@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Make the profile name line match previous generations more closely.
E.g.,
0 3D_FULL_SCREEN :
vs:
0(3D_FULL_SCREEN )
Reviewed-by: Evan Quan <evan.quan@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Export the correct gfx voltage by hwmon interface.
V2: update the register naming for consistency
Signed-off-by: Evan Quan <evan.quan@amd.com>
Reviewed-by: Rex Zhu <Rex.Zhu@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
This patch uses the proper do_div() macro to perform u64 division and
guards against overflow if the result is too large for the unsigned long
return type
Fixes: a2c3c0a54d drm/msm/a6xx: Add devfreq support for a6xx
Cc: Sharat Masetty <smasetty@codeaurora.org>
Signed-off-by: Sean Paul <seanpaul@chromium.org>
Signed-off-by: Rob Clark <robdclark@gmail.com>
Previously only cancelled dma map of a ggtt page when the ggtt entry was
cleared. This patch will cancel dma map of an old ggtt page as well when
the ggtt entry is updated with new page address.
Fixes: 7598e8700e9a(drm/i915/gvt: Missed to cancel dma map for ggtt entries)
Signed-off-by: Hang Yuan <hang.yuan@linux.intel.com>
Signed-off-by: Zhenyu Wang <zhenyuw@linux.intel.com>
The CP performance counter selects were accidentally marked as protected
so they couldn't be written from PM4 streams. Remove the protection
because user space does have an interest in setting up their own
counters.
Signed-off-by: Jordan Crouse <jcrouse@codeaurora.org>
Signed-off-by: Rob Clark <robdclark@gmail.com>
We are currently leaking a drm_crtc_commit struct for every atomic
commit containing plane state. The dpu plane destroy function cleans up
the fb reference manually, but fails to release the commit ref. As a
result, we just keep allocating drm_crtc_commits without ever freeing
them. Fortunately there's a helper function which will clean up all of
our mess at once, so use that.
Thanks to Doug Anderson for reporting the memory leak (and leaving
breadcrumbs from kmemleak!).
Reported-by: Doug Anderson <dianders@chromium.org>
Signed-off-by: Sean Paul <seanpaul@chromium.org>
Signed-off-by: Rob Clark <robdclark@gmail.com>
This patch uses the proper do_div() macro to perform u64 division and
guards against overflow if the result is too large for the unsigned long
return type
Fixes: de0a3d094d drm/msm: re-factor devfreq code
Cc: Sharat Masetty <smasetty@codeaurora.org>
Signed-off-by: Sean Paul <seanpaul@chromium.org>
Signed-off-by: Rob Clark <robdclark@gmail.com>
As done treewide earlier, this catches several more open-coded
allocation size calculations that were added to the kernel during the
merge window. This performs the following mechanical transformations
using Coccinelle:
kvmalloc(a * b, ...) -> kvmalloc_array(a, b, ...)
kvzalloc(a * b, ...) -> kvcalloc(a, b, ...)
devm_kzalloc(..., a * b, ...) -> devm_kcalloc(..., a, b, ...)
Signed-off-by: Kees Cook <keescook@chromium.org>
size for AFBC buffers
The size of the superblocks being added to the total AFBC buffer size
got lost in the upstreaming process. Add it back.
Reviewed-by: Ayan Kumar Halder <ayan.halder@arm.com>
Signed-off-by: Liviu Dudau <liviu.dudau@arm.com>
drm_mode_setcrtc() retries modesetting in case one of the functions it
calls returns -EDEADLK. connector_set, mode and fb are freed before
retrying, but they are not set to NULL. This can cause
drm_mode_setcrtc() to use those variables.
For example: On the first try __drm_mode_set_config_internal() returns
-EDEADLK. connector_set, mode and fb are freed. Next retry starts, and
drm_modeset_lock_all_ctx() returns -EDEADLK, and we jump to 'out'. The
code will happily try to release all three again.
This leads to crashes of different kinds, depending on the sequence the
EDEADLKs happen.
Fix this by setting the three variables to NULL at the start of the
retry loop.
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Link: https://patchwork.freedesktop.org/patch/msgid/20180917110054.4053-1-tomi.valkeinen@ti.com
The variable is declared in an #ifdef section, but the user is
now unconditional, which leads to a build failure:
drivers/gpu/drm/imx/imx-drm-core.c: In function 'imx_drm_bind':
drivers/gpu/drm/imx/imx-drm-core.c:264:6: error: 'legacyfb_depth' undeclared (first use in this function); did you mean 'lockdep_depth'?
Remove the remaining #ifdef as well.
Fixes: f53705fd98 ("drm/imx: Use drm_fbdev_generic_setup()")
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: Noralf Trønnes <noralf@tronnes.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20180926193846.2490574-1-arnd@arndb.de
While we currently grab a runtime PM ref in nouveau's normal connector
detection code, we apparently don't do this for MST. This means if we're
in a scenario where the GPU is suspended and userspace attempts to do a
connector probe on an MSTC connector, the probe will fail entirely due
to the DP aux channel and GPU not being woken up:
[ 316.633489] nouveau 0000:01:00.0: i2c: aux 000a: begin idle timeout ffffffff
[ 316.635713] nouveau 0000:01:00.0: i2c: aux 000a: begin idle timeout ffffffff
[ 316.637785] nouveau 0000:01:00.0: i2c: aux 000a: begin idle timeout ffffffff
...
So, grab a runtime PM ref here.
Signed-off-by: Lyude Paul <lyude@redhat.com>
Cc: stable@vger.kernel.org
Reviewed-by: Karol Herbst <kherbst@redhat.com>
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
This mm_struct pointer should never be dereferenced. If running in
a user thread, just use current->mm. If running in a kernel worker
use get_task_mm to get a safe reference to the mm_struct.
Reviewed-by: Oded Gabbay <oded.gabbay@gmail.com>
Acked-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Felix Kuehling <Felix.Kuehling@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
In amdgpu_dm_commit_tail(), wait until flip_done() is signaled before
we signal hw_done().
[Why]
This is to temporarily address a paging error that occurs when a
nonblocking commit contends with another commit, particularly in a
mirrored display configuration where at least 2 CRTCs are updated.
The error occurs in drm_atomic_helper_wait_for_flip_done(), when we
attempt to access the contents of new_crtc_state->commit.
Here's the sequence for a mirrored 2 display setup (irrelevant steps
left out for clarity):
**THREAD 1** | **THREAD 2**
|
Initialize atomic state for flip |
|
Queue worker |
...
| Do work for flip
|
| Signal hw_done() on CRTC 1
| Signal hw_done() on CRTC 2
|
| Wait for flip_done() on CRTC 1
<---- **PREEMPTED BY THREAD 1**
Initialize atomic state for cursor |
update (1) |
|
Do cursor update work on both CRTCs |
|
Clear atomic state (2) |
**DONE** |
...
|
| Wait for flip_done() on CRTC 2
| *ERROR*
|
The issue starts with (1). When the atomic state is initialized, the
current CRTC states are duplicated to be the new_crtc_states, and
referenced to be the old_crtc_states. (The new_crtc_states are to be
filled with update data.)
Some things to note:
* Due to the mirrored configuration, the cursor updates on both CRTCs.
* At this point, the pflip IRQ has already been handled, and flip_done
signaled on all CRTCs. The cursor commit can therefore continue.
* The old_crtc_states used by the cursor update are the **same states**
as the new_crtc_states used by the flip worker.
At (2), the old_crtc_state is freed (*), and the cursor commit
completes. We then context switch back to the flip worker, where we
attempt to access the new_crtc_state->commit object. This is
problematic, as this state has already been freed.
(*) Technically, 'state->crtcs[i].state' is freed, which was made to
reference old_crtc_state in drm_atomic_helper_swap_state()
[How]
By moving hw_done() after wait_for_flip_done(), we're guaranteed that
the new_crtc_state (from the flip worker's perspective) still exists.
This is because any other commit will be blocked, waiting for the
hw_done() signal.
Note that both the i915 and imx drivers have this sequence flipped
already, masking this problem.
Signed-off-by: Shirish S <shirish.s@amd.com>
Signed-off-by: Leo Li <sunpeng.li@amd.com>
Reviewed-by: Harry Wentland <harry.wentland@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Implement routines to estimate GPU busy time and fetching the
current frequency for the polling interval. This is required by
the devfreq framework which recommends a frequency change if needed.
The driver code then tries to set this new frequency on the GPU by
sending an Out Of Band(OOB) request to the GMU.
Signed-off-by: Sharat Masetty <smasetty@codeaurora.org>
Signed-off-by: Rob Clark <robdclark@gmail.com>
The devfreq framework requires the drivers to provide busy time estimations.
The GPU driver relies on the hardware performance counteres for the busy time
estimations, but different hardware revisions have counters which can be
sourced from different clocks. So the busy time estimation will be target
dependent. Additionally on targets where the clocks are completely controlled
by the on chip microcontroller, fetching and setting the current GPU frequency
will be different. This patch aims to embrace these differences by re-factoring
the devfreq code a bit.
Signed-off-by: Sharat Masetty <smasetty@codeaurora.org>
Signed-off-by: Rob Clark <robdclark@gmail.com>
