On 13/04/2023 15:19, Luca Weiss wrote:
Hi Bryan, On Thu Apr 13, 2023 at 1:34 PM CEST, Bryan O'Donoghue wrote:V5: - Amagamates into once device, Heikki, Rob - Takes feedback on usage form Luka and Jianhua on VSafeV state transition detection dev_err() -> dev_warn() - Orientation graph example and general expected bindings I discussed offline with Bjorn the conclusions of the glink/sbu model. The expected orientation-switch path is connector/port@0 <-> phy/port@X <-> dp/port@0 This can then be expanded to connector/port@0 <-> redriver/port@0 <-> phy/port@X <-> dp/port@0 - Rob, Bjorn, Krzysztof - Data role The data-role path is connector/port@0 <-> dwc3/port@YI believe I have adjusted my dts correctly for v5 compared to v4 but now the usb doesn't seem to work anymore in most cases. Only when having the phone already plugged in during boot in one orientation does USB come up, but also disappears once you replug the cable. I still see the same (or at least visually similar) messages when plugging in the USB cable or the USB stick but nothing more than that happens. Not that v4 worked perfectly on pm7250b+sm7225(/sm6350) but at least it worked in most cases as described in the emails there. Since the driver structure changed quite a bit, git diff isn't helpful here unfortunately. Don't think it matters but worth mentioning that sm6350 uses the new qmpphy bindings as described in qcom,sc8280xp-qmp-usb43dp-phy.yaml (this was also the case when testing v4 of this). Any idea?
Can you confirm the output of /sys/class/typec/port0/orientation in host mode with the USB key / peripheral in both orientations ?
If that's still OK, then perhaps we can figure out the gap in the PHY code for v3
@caleb is working on this code for sdm845 which is a v3 PHY --- bod
