On Fri Mar 31, 2023 at 3:52 PM CEST, Bryan O'Donoghue wrote: > On 31/03/2023 09:48, Luca Weiss wrote: > > Hi Bryan, > > > > On Fri Mar 24, 2023 at 4:28 PM CET, Bryan O'Donoghue wrote: > >> On 24/03/2023 15:10, Luca Weiss wrote: > >>> When plugging in the device with TCPM on into my PC (peripheral mode) > >>> then the USB device registers and unregisters every couple of seconds, > >>> never stays stable on. No messages in dmesg when this happens. This only > >>> happens with the USB-C plug in one direction, in the other it > >>> works reliable. > >> > >> Sounds like we need to do some SoC specific debug on orientation > >> switching in the PHY. > > > > I also know that the phone has a AW35743 chip in the USB path, > > controlled by DP_AUX_EN and DP_AUX_SEL gpios but I think this is only > > for displayport, right? > > > > https://www.awinic.com/en/productDetail/AW35743CSR#product-details D+/D- > looks like USB 2.x .. > > Your DP should go over TX1+/1 TX2+/- depending on orientation and # of > lanes in use. There's USB0_DP_AUX_P & USB0_DP_AUX_M from the SoC connected on the D1+/- & D2+/- pins of this awinic chip, and out comes USB_SBU1 and USB_SBU2 which goes to the USB connector. So just switching polarity of those pins. But let's say whatever it does, it's not relevant to this use case now without DP? > > https://www.allaboutcircuits.com/uploads/articles/Fig1m11292018.png > > >> > >> I wonder how many lanes dp_opts->lanes says for your part ? > > > > Not sure.. Where is this configured? > > Heh - now that I look my lane count == 0, a bug to be fixed.. > > Hmm, Luca can you test this change > > - if (orientation == TYPEC_ORIENTATION_NONE) { > - if (qmp->init_count) > - ret = qmp_combo_dp_power_off(dp_phy); > - } else { > - if (!qmp->init_count) > - ret = qmp_combo_dp_power_on(dp_phy); > - } > + if (orientation == TYPEC_ORIENTATION_NONE) > + ret = qmp_combo_dp_power_off(dp_phy); > + else > + ret = qmp_combo_dp_power_on(dp_phy); I unfortunately don't really see any change in behavior with this.. > > > > But I also don't have DisplayPort over USB-C (video out) configured yet. > > Related question: does video out work on sm8250+pm8150b for you? > > Nope - WIP. I see getting TCPM upstream and working as a first step, > then we look at introduction of the redriver, DP work, dt etc. Clear, I look forward to that landing ;) But also already what the driver does now is really useful! Regards Luca > > > > [ 1722.130836] qcom,pmic-typec c440000.spmi:pmic@2:typec@1500: get_cc: misc 0x000000cb cc1 0x00000001 Ra cc2 0x00000002 Rd attached 1 cc=cc1 > > > > - unplug - > > > > [ 1867.223052] qcom,pmic-typec c440000.spmi:pmic@2:typec@1500: get_cc: misc 0x00000042 cc1 0x00000000 Open cc2 0x00000000 Open attached 0 cc=cc1 > > ================================================================================== > > USB unstable (device plug orientation 2): > > ================================================================================== > > > > [ 1894.263324] qcom,pmic-typec c440000.spmi:pmic@2:typec@1500: get_cc: misc 0x000000c9 cc1 0x00000002 Rd cc2 0x00000001 Ra attached 1 cc=cc2 > > - unplug - > > > > [ 1907.264840] qcom,pmic-typec c440000.spmi:pmic@2:typec@1500: get_cc: misc 0x00000040 cc1 0x00000000 Open cc2 0x00000000 Open attached 0 cc=cc2 > > Normal, correct. > > For reference on 8250 > > Attached orientation reverse / cc2 > > [ 77.719278] qcom,pmic-typec c440000.spmi:pmic@2:typec@1500: set_cc: > currsrc=2 Rp-3.0-330uA mode EN_SRC_ONLY debounce 1 attached 1 cc=cc2 > > Detached orientation (none) > [ 82.475667] qcom,pmic-typec c440000.spmi:pmic@2:typec@1500: get_cc: > misc 0x000000c0 cc1 0x00000000 Open cc2 0x00000000 Open attached 0 cc=cc2 > > Attached orientation normal / cc1 > [ 82.485375] qcom,pmic-typec c440000.spmi:pmic@2:typec@1500: > start_toggling: misc 0x00000040 attached 0 port_type 0 current cc 5 new 5 > [ 85.247368] qcom,pmic-typec c440000.spmi:pmic@2:typec@1500: get_cc: > misc 0x000000cb cc1 0x00000001 Ra cc2 0x00000002 Rd attached 1 cc=cc1 > > Anyway I reckon that guard I have in the PHY code is wrong, would > appreciate a test. > > --- > bod
