linux-stable.git/drivers/gpu/drm/amd/display/dc, branch v5.8

drm/amd/display: handle failed allocation during stream construction

2020-07-14T19:14:15+00:00

[Why]
Failing to allocate a transfer function during stream construction leads
to a null pointer dereference

[How]
Handle the failed allocation by failing the stream construction

Cc: stable@vger.kernel.org
Signed-off-by: Josip Pavic 
Reviewed-by: Aric Cyr 
Acked-by: Rodrigo Siqueira 
Signed-off-by: Alex Deucher

drm/amd/display: Only revalidate bandwidth on medium and fast updates

2020-07-01T02:15:12+00:00

[Why]
Changes that are fast don't require updating DLG parameters making
this call unnecessary. Considering this is an expensive call it should
not be done on every flip.

DML touches clocks, p-state support, DLG params and a few other DC
internal flags and these aren't expected during fast. A hang has been
reported with this change when called on every flip which suggests that
modifying these fields is not recommended behavior on fast updates.

[How]
Guard the validation to only happen if update type isn't FAST.

Bug: https://gitlab.freedesktop.org/drm/amd/-/issues/1191
Fixes: a24eaa5c51255b ("drm/amd/display: Revalidate bandwidth before commiting DC updates")
Signed-off-by: Nicholas Kazlauskas 
Acked-by: Alex Deucher 
Reviewed-by: Roman Li 
Signed-off-by: Alex Deucher 
Cc: stable@vger.kernel.org

drm/amd/display: Rework dsc to isolate FPU operations

2020-06-11T20:05:32+00:00

When we want to use float point operation on Linux
we need to use within special kernel protection
(`kernel_fpu_{begin,end}()`.), otherwise the kernel
can clobber userspace FPU register state. For detecting
these issues we use a tool named objtool (with -Ffa
flags) to highlight the FPU problems, all warnings can
be summed up as follows:

./tools/objtool/objtool check -Ffa
drivers/gpu/drm/amd/display/dc/dml/dml_common_defs.o

[..] dc/dsc/rc_calc.o: warning: objtool: get_qp_set()+0x2f8:
FPU instruction outside of kernel_fpu_{begin,end}()
[..] dc/dsc/rc_calc.o: warning: objtool: dsc_roundf()+0x5:
FPU instruction outside of kernel_fpu_{begin,end}()
[..] dc/dsc/rc_calc.o: warning: objtool: dsc_ceil()+0x5:
FPU instruction outside of kernel_fpu_{begin,end}()
[..] dc/dsc/rc_calc.o: warning: objtool: get_ofs_set()+0x3eb:
FPU instruction outside of kernel_fpu_{begin,end}()
[..] dc/dsc/rc_calc.o: warning: objtool: calc_rc_params()+0x3c:
FPU instruction outside of kernel_fpu_{begin,end}()
[..] dc/dsc/dc_dsc.o: warning: objtool:
get_dsc_bandwidth_range.isra.0()+0x8d:
FPU instruction outside of kernel_fpu_{begin,end}()
[..] dc/dsc/dc_dsc.o: warning: objtool: setup_dsc_config()+0x2ef:
FPU instruction outside of kernel_fpu_{begin,end}()
[..] dc/dsc/rc_calc_dpi.o: warning: objtool:copy_pps_fields()+0xbb:
FPU instruction outside of kernel_fpu_{begin,end}()
[..] dc/dsc/rc_calc_dpi.o: warning: objtool:
dscc_compute_dsc_parameters()+0x7b:
FPU instruction outside of kernel_fpu_{begin,end}()

This commit fixes the above issues by rework DSC as described:

1. Isolate all FPU operations in a single file;
2. Use FPU flags only in the file that handles FPU operations;
3. Isolate all functions that require float point operation in static
   functions;
4. Add a mid-layer function that does not use any float point operation,
   and that could be safely invoked in other parts of the code.
5. Keep float point operation under DC_FP_{START/END} macro.

CC: Christian König 
CC: Alexander Deucher 
CC: Peter Zijlstra 
CC: Tony Cheng 
CC: Harry Wentland 
Signed-off-by: Rodrigo Siqueira 
Reviewed-by: Mikita Lipski 
Acked-by: Qingqing Zhuo 
Signed-off-by: Alex Deucher

drm/amd/display: Revalidate bandwidth before commiting DC updates

2020-06-03T19:33:06+00:00

[Why]
Whenever we switch between tiled formats without also switching pixel
formats or doing anything else that recreates the DC plane state we
can run into underflow or hangs since we're not updating the
DML parameters before committing to the hardware.

[How]
If the update type is FULL then call validate_bandwidth again to update
the DML parmeters before committing the state.

This is basically just a workaround and protective measure against
update types being added DC where we could run into this issue in
the future.

We can only fully validate the state in advance before applying it to
the hardware if we recreate all the plane and stream states since
we can't modify what's currently in use.

The next step is to update DM to ensure that we're creating the plane
and stream states for whatever could potentially be a full update in
DC to pre-emptively recreate the state for DC global validation.

The workaround can stay until this has been fixed in DM.

Signed-off-by: Nicholas Kazlauskas 
Reviewed-by: Hersen Wu 
Signed-off-by: Alex Deucher

drm/amdgpu/display: use blanked rather than plane state for sync groups

2020-06-03T19:32:34+00:00

We may end up with no planes set yet, depending on the ordering, but we
should have the proper blanking state which is either handled by either
DPG or TG depending on the hardware generation.  Check both to determine
the proper blanked state.

Bug: https://gitlab.freedesktop.org/drm/amd/issues/781
Fixes: 5fc0cbfad45648 ("drm/amd/display: determine if a pipe is synced by plane state")
Cc: nicholas.kazlauskas@amd.com
Reviewed-by: Nicholas Kazlauskas 
Signed-off-by: Alex Deucher 
Cc: stable@vger.kernel.org

drm/amd/display: Fix potential integer wraparound resulting in a hang

2020-05-27T22:42:10+00:00

[Why]
If VUPDATE_END is before VUPDATE_START the delay calculated can become
very large, causing a soft hang.

[How]
Take the absolute value of the difference between START and END.

Signed-off-by: Aric Cyr 
Reviewed-by: Nicholas Kazlauskas 
Acked-by: Qingqing Zhuo 
Signed-off-by: Alex Deucher

drm/amd/display: Set/Reset avmute when disable/enable stream

2020-05-21T16:48:43+00:00

[Why]
When disconnect fe from be, something such as unstable clock may cause
garbage occurs.

[How]
Send set avmute at the beginning of disable stream and send reset avmute
at the end of enable stream.

Signed-off-by: Jinze Xu 
Reviewed-by: Anthony Koo 
Acked-by: Rodrigo Siqueira 
Acked-by: Tony Cheng 
Signed-off-by: Alex Deucher

drm/amd/display: correct rn NUM_VMID

2020-05-21T16:48:43+00:00

Save the correct num vmid during resource creation and fix RN gpuvm
level from 1 to 16 vmid entries.

Signed-off-by: Dmytro Laktyushkin 
Reviewed-by: Eric Bernstein 
Acked-by: Rodrigo Siqueira 
Signed-off-by: Alex Deucher

drm/amd/display: Avoid pipe split when plane is too small

2020-05-21T16:48:43+00:00

[Why]
The minimum plane size we can support in DML is 16x16. If we try to pass
a 16x16 plane with dynamic pipe split then validation will fail since it
tries to split it into two pipes, each 8x8.

Some userspace doesn't check that the commit fails and because the
commit fails the old state is retained, resulting in corruption.

[How]
Add a workaround to avoid pipe split if any plane is 16x16 or smaller.

Signed-off-by: Nicholas Kazlauskas 
Reviewed-by: Aric Cyr 
Acked-by: Rodrigo Siqueira 
Signed-off-by: Alex Deucher

drm/amd/display: Defer cursor lock until after VUPDATE

2020-05-21T16:48:43+00:00

[Why]
We dropped the delay after changed the cursor functions locking the
entire pipe to locking just the CURSOR registers to fix page flip
stuttering - this introduced cursor stuttering instead, and an underflow
issue.

The cursor update can be delayed indefinitely if the cursor update
repeatedly happens right around VUPDATE.

The underflow issue can happen if we do a viewport update on a pipe
on the same frame where a cursor update happens around VUPDATE - the
old cursor registers are retained which can be in an invalid position.

This can cause a pipe hang and indefinite underflow.

[How]
The complex, ideal solution to the problem would be a software
triple buffering mechanism from the DM layer to program only one cursor
update per frame just before VUPDATE.

The simple workaround until we have that infrastructure in place is
this change - bring back the delay until VUPDATE before locking, but
with some corrections to the calculations.

This didn't work for all timings before because the calculation for
VUPDATE was wrong - it was using the offset from VSTARTUP instead and
didn't correctly handle the case where VUPDATE could be in the back
porch.

Add a new hardware sequencer function to use the existing helper to
calculate the real VUPDATE start and VUPDATE end - VUPDATE can last
multiple lines after all.

Change the udelay to incorporate the width of VUPDATE as well.

Signed-off-by: Nicholas Kazlauskas 
Reviewed-by: Aric Cyr 
Acked-by: Rodrigo Siqueira 
Signed-off-by: Alex Deucher