this was introduced in MGS-6394 imx-2620, to support trace-cmd
Signed-off-by: Jiyu Yang <jiyu.yang@nxp.com>
(cherry picked from commit 05c19f3b781dc0101275a3f5707492f66f4ee8e8)
Change-Id: Ibd70ff1c8d8efad998ca21b5c88df2fc54ed4c16
Reviewed-by: jiajia Qian <jiajia.qian@nxp.com>
Acked-by: Jason Liu <jason.hui.liu@nxp.com>
Boards with gpu subsystem, such as 8MP, have gpu device and gpu2d,
gpu3d sub-devices. gpu2d and gpu3d devices have enabled runtime pm
while gpu device not. But gpu runtime pm suspend and resume callbacks
are binded to gpu device, which leads to gpu runtime pm callback
not working.
So we enable runtime pm of gpu device in gpu subsystem to
request/release high bus freq when GPU work/idle.
Signed-off-by: Guangliu Ding <guangliu.ding@nxp.com>
It's wrong to request/release high bus freq when GPU probe/remove.
Instead, we need to request/release high bus freq when GPU work/idle.
So we revert this patch to avoid ddrc freq can't enter low bus mode.
This reverts commit c413b53ce5.
Signed-off-by: Guangliu Ding <guangliu.ding@nxp.com>
Reviewed-by: Jacky Bai <jacky.bai@nxp.com>
Acked-by: Jason Liu <jason.hui.liu@nxp.com>
Boards with gpu subsystem, such as 8MP, have gpu device and gpu2d,
gpu3d sub-devices. gpu2d and gpu3d devices have enabled runtime pm
while gpu device not. But gpu runtime pm suspend and resume callbacks
are binded to gpu device, which leads to gpu runtime pm callback
not working.
So we enable runtime pm of gpu device in gpu subsystem to fix this issue.
Signed-off-by: Guangliu Ding <guangliu.ding@nxp.com>
Fix IMX-3262
GPU hang while the vertex data count is very large in 8mm through splitting
large vertex data draw call.
Signed-off-by: IPD_SCM <IPD_SCM@verisilicon.com>
Signed-off-by: Yuan Tian <yuan.tian@nxp.com>
CL864414 [KERNEL SPACE] [22Q2][IMX-3201]: Simply create a new tracepoint
gpu_work_period to pass Android-14 CTS case GpuWorkTracePointTest. [Ying.Zhang]
This pick VSI
0013-CL864414-KERNEL-SPACE-22Q2-IMX-3201-Simply-create-a-.patch
0018-Fix-license.patch
Signed-off-by: Jessie Hao <juan.hao@nxp.com>
[NXP]: Add a macro of
gcdDISABLE_RS_CLOCK_GATING. By Shiheng.Zhou
Signed-off-by: IPD_SCM <IPD_SCM@verisilicon.com>
Signed-off-by: Yuan Tian <yuan.tian@nxp.com>
[6.4.11_NXP]: Fix cl815168.
Move two function declarations out of marco gcdENABLE_TRUST_APPLICATION's
control. [Ying.Zhang]
Signed-off-by: IPD_SCM <IPD_SCM@verisilicon.com>
Signed-off-by: Yuan Tian <yuan.tian@nxp.com>
22Q2_NXP: Reallocate virtual
node when weston import capture only process node and don't fill page table
for contigourous pool when enable capture only mode.
Signed-off-by: IPD_SCM_CN <IPD_SCM_CN@verisilicon.com>
Signed-off-by: Yuan Tian <yuan.tian@nxp.com>
ktime_get_real_ts64 can jump backwards due to a leap second update,
NTP adjustment settimeofday() operation from user space.
Signed-off-by: Jiyu Yang <jiyu.yang@nxp.com>
Reviewed-by: Guangliu Ding <guangliu.ding@nxp.com>
Reviewed-by: Yuan Tian <yuan.tian@nxp.com>
Patches for imx-3234 caused vulkan sascha hang regression.
Disable USC reset to fix this regression.
The original fix for imx-3234 will be disabled accordingly.
Fixes: 39df2d1bf8 ("MGS-7524 [#imx-3234] 0004-CL747890-KERNEL-SPACE-6.4.11_22Q2_NXP-Support-USC-lo")
Signed-off-by: Yuan Tian <yuan.tian@nxp.com>
Reviewed-by: Xianzhong Li <xianzhong.li@nxp.com>
Tested-by: Yuan Tian <yuan.tian@nxp.com>
This commit fixes the issue that frequency modulation function
cannot be achieved on 8MP.
Signed-off-by: Yuan Tian <yuan.tian@nxp.com>
Reviewed by: Guangliu Ding <guangliu.ding@nxp.com>
Acked-by: Jason Liu <jason.hui.liu@nxp.com>
[NXP]: Fix the issue about the
frequency can not be turned. By Shiheng.Zhou
Signed-off-by: IPD_SCM <IPD_SCM@verisilicon.com>
Signed-off-by: Yuan Tian <yuan.tian@nxp.com>
[NXP]: Fix the issue that Scale
could not be modulated. By Shiheng.Zhou
Signed-off-by: IPD_SCM <IPD_SCM@verisilicon.com>
Signed-off-by: Yuan Tian <yuan.tian@nxp.com>
Use blit engine and secuirty feature to determine i.MX8MQ/QXP GPU type,
Both chips have CVE-2023-4969 security issue as they support OpenCL local memory with on-chip USC memory.
Fixed this security issue by toggling the GPU security mode to trigger a USC reset during GPU context switch,
VSI validated the second OpenCL process can only read 0s from the USC memory after context switch.
Signed-off-by: Xianzhong Li <xianzhong.li@nxp.com>
VIP doesn't support switch security mode, so remove the switch when it is VIP core.
Signed-off-by: Zhe Pan <zhe.pan@vivantecorp.com>
Signed-off-by: Yuan Tian <yuan.tian@nxp.com>
Meet kernel NULL pointer dereference in _PutPower.
It is caused by releasing gpu_cooling_device (null pointer) on
7ULP/8ULP while gpu_cooling_device not supported on the two boards.
So we add a null pointer check to gpu_cooling_device.
Signed-off-by: Guangliu Ding <guangliu.ding@nxp.com>
Signed-off-by: Dandan Sun <dandan.sun@nxp.com>
for ls1028a, considering the unsupported GPU frequency adjustment
feature in the current "devfreq_cooling.c" driver, replacing it with
an improved solution within the "gc_hal_kernel_platform_imx.c" GPU
driver became necessary.
in this patch, GPU register as a cooling devices, thermal framework
will auto bind it.
Signed-off-by: Dandan Sun <dandan.sun@nxp.com>
Reviewed by: Guangliu Ding <guangliu.ding@nxp.com>
