On 14.09.2023 04:32, Mike Tipton wrote:
> On Wed, Sep 13, 2023 at 10:31:49AM +0200, Konrad Dybcio wrote:
>>> The applicable voters should likely be defined in the target-specific
>>> headers, rather than the common qcom,icc.h. The bit range used for them
>>> could be common, but each target may only support a small subset of the
>>> total set of possible voters across all targets.
>> I'm not sure how client drivers would then choose the
>> correct path other than
>>
>> switch (soc) {
>> case 8450:
>> tag = QCOM_ICC_TAG_VOTER_8450_HLOS;
>> break;
>> case 8550:
>> tag = QCOM_ICC_TAG_VOTER_8550_HLOS;
>> break;
>> ...
>> }
>>
>> which would be unacceptable.
>
> The same general way it's handled for the endpoint bindings, which are
> already target-specific.
>
> Any client drivers hardcoding the endpoint bindings in their driver
> would have to include the appropriate, target-specific binding header
> (e.g. qcom,sm8550-rpmh.h). That would only be possible if their driver
> file is itself target-specific. Otherwise, it would have to pull the
> endpoint bindings from devicetree. Or just use the recommended
> of_icc_get() and let devicetree do everything for them. Same for the
> target-specific voter tag bindings.
>
> Clients can also specify their tags in devicetree. They don't actually
> have to call icc_set_tag() directly. For example:
>
> #include <dt-bindings/interconnect/qcom,sm8450.h>
>
> interconnects = <&mmss_noc MASTER_MDP QCOM_ICC_TAG_VOTER_DISP
> &mc_virt SLAVE_EBI1 QCOM_ICC_TAG_VOTER_DISP>;
>
> Then when they call of_icc_get() for this path it'll automatically have
> QCOM_ICC_TAG_VOTER_DISP set for them.
I think I'd skew towards the "define everything in the DT" approach.
One thing that makes me uneasy to go on with this approach is the
question whether there is a case in which we would want to switch
from e.g. voting through DISP to voting through APPS (or similar)
from within a single device.
Konrad
