On 10/29/19 5:20 AM, Cezary Rojewski wrote:
On 2019-10-26 00:40, Pierre-Louis Bossart wrote:
+/* DSP D0ix sub-state */ +enum sof_d0_substate { + SOF_DSP_D0I0 = 0, /* DSP default D0 substate */ + SOF_DSP_D0I3, /* DSP D0i3(low power) substate*/ +};
Name of the type states: "d0 substate" while description "D0ix sub-state". Why was not this named D0ix from the get-go? Goes into the same the same naming bucket as S0ix.
The definition is correct, from the pm_runtime perspective the device is 'active' i.e. D0. D0ix is a substate of D0.
On the further note, adding D0ix patch within "enable S0ix support for Intel platforms" is misleading. S-states != D-states. D0ix is especially orthogonal. It is these to further reduce power consumption when system and device are in S0 and D0 respectively and idle time between IPC communication is long enough for DSP to be power gated.
there are well-defined requirements and dependencies between S and D states:
S0: device can be in D0, D0ix, D3 S0ix: device can be in D0ix, D3 S3: device can be in D3
That's hardly orthogonal.