Mahantesh Kumbar c032207351 gpu: nvgpu: PMU PSTATE init wait for pmu_ready ...

-On Turing, LS PMU RTOS is bootstrapped by SEC2 RTOS,
 so PMU needs some time to boot & get ready to process
 request as boot happened bit late in flow.
-So issue here is, code execution reaches function
 gk20a_init_pstate_pmu_support() to send boardobj
 commands to PMU before PMU ready to accept.
-As a result, this cause command request failure
 as PMU is not ready yet due to falcon clock is
 slower compared CPU clock.

-To fix issue, waiting for pmu_ready flag to become
 true before staring PSTATE requests to PMU.
-pmu_ready flag will be set to true as soon as
 INIT message received from PMU RTOS.

-Pre Turing-PMU RTOS will be ready much ahead of
 reaching this point.

JIRA NVGPU-1150

Change-Id: I4fb5d4ca6dabc22fac34ae32880d7a1f3164e1b7
Signed-off-by: Mahantesh Kumbar <mkumbar@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/1925240
Reviewed-by: mobile promotions <svcmobile_promotions@nvidia.com>
Tested-by: mobile promotions <svcmobile_promotions@nvidia.com>

2018-10-25 13:43:22 -07:00

..

pstate.c

gpu: nvgpu: PMU PSTATE init wait for pmu_ready

2018-10-25 13:43:22 -07:00

pstate.h

gpu: nvgpu: MISRA Rule 21.2 header guard fixes

2018-09-24 21:17:42 -07:00