> -Original Message-
> From: Dmitry Baryshkov
> Sent: Wednesday, December 14, 2022 9:05 PM
> To: Vinod Polimera (QUIC) ; dri-
> de...@lists.freedesktop.org; linux-arm-...@vger.kernel.org;
> freedreno@lists.freedesktop.org; devicet...@vger.kernel.org
> Cc: linux-ker...@vger.kernel.org; ro
On Tue, 20 Dec 2022 at 08:44, Akhil P Oommen wrote:
>
> From: Ulf Hansson
>
> Some genpd providers doesn't ensure that it has turned off at hardware.
> This is fine until the consumer really requires during some special
> scenarios that the power domain collapse at hardware before it is
> turned
Fix the below kernel panic due to null pointer access:
[ 18.504431] Unable to handle kernel NULL pointer dereference at virtual
address 0048
[ 18.513464] Mem abort info:
[ 18.516346] ESR = 0x9605
[ 18.520204] EC = 0x25: DABT (current EL), IL = 32 bits
[ 18.525
Update gpu coredump for a660/a650 family of gpus with the extra
information available.
Signed-off-by: Akhil P Oommen
---
(no changes since v1)
drivers/gpu/drm/msm/adreno/a6xx.xml.h | 18 +++
drivers/gpu/drm/msm/adreno/a6xx_gpu_state.c | 50 -
drivers/g
Ensure that we do drm_dev_put() when there is an early return in
msm_drm_init().
Signed-off-by: Akhil P Oommen
---
(no changes since v1)
drivers/gpu/drm/msm/disp/msm_disp_snapshot.c | 3 +++
drivers/gpu/drm/msm/msm_drv.c| 11 +++
2 files changed, 10 insertions(+), 4 de
Since RoQ size differs between generations, calculate dynamically the
RoQ size while capturing coredump.
Signed-off-by: Akhil P Oommen
---
(no changes since v1)
drivers/gpu/drm/msm/adreno/a6xx_gpu_state.c | 11 ++-
drivers/gpu/drm/msm/adreno/a6xx_gpu_state.h | 17 ++---
2 f
This is a rework of [1] using genpd instead of 'reset' framework.
As per the recommended reset sequence of Adreno gpu, we should ensure that
gpucc-cx-gdsc has collapsed at hardware to reset gpu's internal hardware states.
Because this gdsc is implemented as 'votable', gdsc driver doesn't poll an
From: Ulf Hansson
Some genpd providers doesn't ensure that it has turned off at hardware.
This is fine until the consumer really requires during some special
scenarios that the power domain collapse at hardware before it is
turned ON again.
An example is the reset sequence of Adreno GPU which re
Add support for the newly added 'synced_poweroff' genpd flag. This allows
some clients (like adreno gpu driver) to request gdsc driver to ensure
a votable gdsc (like gpucc cx gdsc) has collapsed at hardware.
Signed-off-by: Akhil P Oommen
---
(no changes since v3)
Changes in v3:
- Rename the var
When a device has multiple power domains, dev->power_domain is left
empty during probe. That didn't cause any issue so far because we are
freeloading on smmu driver's vote on cx gdsc. Instead of that, create
a device_link between cx genpd device and gmu device to keep a vote from
gpu driver.
Befor
Remove the unused 'reset' interface which was supposed to help to ensure
that cx gdsc has collapsed during gpu recovery. This is was not enabled
so far due to missing gpucc driver support. Similar functionality using
genpd framework will be implemented in the upcoming patch.
This effectively rever
As per the recommended recovery sequence of adreno gpu, cx gdsc should
collapse at hardware before it is turned back ON. This helps to clear
out the stale states in hardware before it is reinitialized. Use the
genpd notifier along with the newly introduced
dev_pm_genpd_synced_poweroff() api to ensu
On 12/21/2022 8:13 PM, Ulf Hansson wrote:
> On Tue, 20 Dec 2022 at 08:44, Akhil P Oommen wrote:
>> From: Ulf Hansson
>>
>> Some genpd providers doesn't ensure that it has turned off at hardware.
>> This is fine until the consumer really requires during some special
>> scenarios that the power dom
On 21/12/2022 16:10, Vinod Polimera wrote:
-Original Message-
From: Dmitry Baryshkov
Sent: Wednesday, December 14, 2022 9:05 PM
To: Vinod Polimera (QUIC) ; dri-
de...@lists.freedesktop.org; linux-arm-...@vger.kernel.org;
freedreno@lists.freedesktop.org; devicet...@vger.kernel.org
Cc:
According to downstream /and the comment copied from it/ this comparison
should be the other way around. In other words, when the panel driver
requests to use more slices per packet than what could be sent over this
interface, it is bumped down to only use a single slice per packet (and
strangely
Resource allocation of DSC blocks should behave more like LMs and CTLs
where NULL resources (based on initial hw_blk creation via definitions
in the catalog) are skipped ^1. The current hardcoded mapping of DSC
blocks however means that resource allocation shouldn't succeed at all
when the DSC blo
Downstream calls this num_enc yet the DSC patches introduced a new
num_dsc struct member, leaving num_enc effectively unused.
Fixes: 7e9cc175b159 ("drm/msm/disp/dpu1: Add support for DSC in topology")
Signed-off-by: Marijn Suijten
---
drivers/gpu/drm/msm/disp/dpu1/dpu_encoder.c | 9 -
dr
These DSC blocks on CTL V1 need to set its corresponding PINGPONG block
index in a hardware register to configure where to send pixel output to,
via the newly-added DPU_DSC_OUTPUT_CTRL feature flag.
Signed-off-by: Marijn Suijten
Reviewed-by: Dmitry Baryshkov
Reviewed-by: Abhinav Kumar
---
.../
Active CTLs have to configure what DSC block(s) have to be enabled, and
what DSC block(s) have to be flushed; this value was initialized to zero
resulting in the necessary register writes to never happen (or would
write zero otherwise). This seems to have gotten lost in the DSC v4->v5
series while
This preliminary Display Stream Compression support package for
(initially tested on) sm8[12]50 is based on comparing DSC behaviour
between downstream and mainline. Some new callbacks are added (for
binding blocks on active CTLs), logic bugs are corrected, zeroed struct
members are now assigned pr
According to downstream the value to use for WORD_COUNT is
bytes_per_pkt, which denotes the number of bytes in a packet based on
how many slices have been configured by the panel driver times the
width of a slice times the number of bytes per pixel.
The DSC panels seen thus far use one byte per pi
In the event that the topology requests resources that have not been
created by the system (because they are typically not represented in
dpu_mdss_cfg ^1), the resource(s) in global_state (in this case DSC
blocks) remain NULL but will still be returned out of
dpu_rm_get_assigned_resources, where th
All V1 CTL blocks (active CTLs) explicitly bind the pixel output from a
DSC block to a PINGPONG block by setting the PINGPONG index in a DSC
hardware register.
Signed-off-by: Marijn Suijten
Reviewed-by: Dmitry Baryshkov
Reviewed-by: Abhinav Kumar
---
drivers/gpu/drm/msm/disp/dpu1/dpu_encoder.c
Hi Dmitry
Sorry for the late response.
On 10/31/2022 5:20 PM, Dmitry Baryshkov wrote:
On 28/10/2022 01:22, Abhinav Kumar wrote:
On 10/27/2022 10:35 AM, Dmitry Baryshkov wrote:
On 22/09/2022 03:49, Abhinav Kumar wrote:
Re-arrange the dsi_calc_pclk method to two helpers, one to
compute the D
24 matches
Mail list logo