On 25/08/2021 01:01, Bjorn Andersson wrote:

On Tue 24 Aug 16:58 PDT 2021, Bryan O'Donoghue wrote:

quoted

On 25/08/2021 00:52, Bjorn Andersson wrote:

quoted

But isn't this role switching interaction (both B and C) already part of
the core/drd, so if we can just get the drd to invoke
dwc3_qcom_vbus_overrride_enable() we're done (and can remove all the
extcon code from the qcom glue as well)?

Provided we have an acceptable way of triggering when a role-switch happens
- then USB role switching itself is a NOP here, its really just a
convenience to invoke the callback.

Thanks for confirming. Then let's come up with an acceptable way, rather
than duplicating yet another feature already implemented in the core.

Regards,
Bjorn

The only other thing to say about USB role-switching is it appears to be 
very much 1:1.

Extcon allows a virtually unlimited number of consumers of an even.

Is it envisaged that a role-switch could or should be consumed by say 3 
or even 4 separate pieces of logic and if not, why not ?

What if we had a magic black box that needed to sit ontop of the QCOM 
layer and further consume a role switch ?

notifier/platform pointer + callback ?

---
bod