Hi Bryan,
On Mon Apr 17, 2023 at 12:04 PM CEST, Bryan O'Donoghue wrote:
> On 17/04/2023 08:35, Luca Weiss wrote:
> > Do you have an idea in which part of the code to start debugging this?
> > Since orientation detection is working is it maybe in the phy code and
> > not in the tcpm driver? Or does that also touch crucial stuff for USB
> > apart from telling phy which direction to use?
>
> PHY - I'd almost just do the following
>
> diff --git a/drivers/phy/qualcomm/phy-qcom-qmp-combo.c
> b/drivers/phy/qualcomm/phy-qcom-qmp-combo.c
> index edb788a71edeb..bbac82bd093f8 100644
> --- a/drivers/phy/qualcomm/phy-qcom-qmp-combo.c
> +++ b/drivers/phy/qualcomm/phy-qcom-qmp-combo.c
> @@ -3369,7 +3369,7 @@ static int qmp_combo_typec_switch_set(struct
> typec_switch_dev *sw,
>
> dev_dbg(qmp->dev, "Toggling orientation current %d requested %d\n",
> qmp->orientation, orientation);
> -
> +return 0;
>
> In that case the PHY should "just work" for host or device in one
> orientation.
>
> The other possibility is that the data role message is not hitting dwc3
> drd on your platform.
>
> If you take the last commit on this branch - plus the updated PHY commit
>
> Commit: 171d7f507511 ("usb: dwc3: drd: Enable user-space triggered
> role-switching")
>
> Commit: eb0daa19f3ad ("phy: qcom-qmp: Register as a typec switch for
> orientation detection")
>
> https://git.codelinaro.org/bryan.odonoghue/kernel/-/tree/linux-next-23-04-17-pm8150b-tcpm-qcom-wrapper-typec-mux
>
> cat /sys/class/usb_role/a600000.usb-role-switch/role
>
> On SM8250 it looks like this
>
> - Attach TypeC accessory with USB key plugged in [1]
> Mount USB key, read/write some data
> Unmount USB key
>
> cat /sys/class/usb_role/a600000.usb-role-switch/role
> host
It feels like I spent way too much time now trying to understand the
current behavior across the different patch versions, it's a bit messy,
but in short:
With the "user-space triggered role-switching" patch I can see that
whatever scenario the USB-C port is in, the role is stuck on "device".
Nothing =
Role: device, Orientation: unknown
USB(-A) cable to laptop (either direction) =
Role: device, Orientation: unknown
USB stick up =
Role: device, Orientation: reverse
USB stick down =
Role: device, Orientation: normal
Sometimes/mostly when the USB cable is attached during boot I get USB
connection to the laptop until I unplug, then it won't reenable itself.
Also the early return in qmp_combo_typec_switch_set doesn't seem to
change much I believe? But for sure normally qmp_combo_dp_power_off/on
does not get called so I wouldn't be suprised if this reinit breaks
something in the phy.
> <snip>
>
> Yep its worth checking out that the data-role switch is working, we
> might be looking at the wrong thing for you on the PHY.
>
So this seems to be the case? If that's useful, I can also go back to
the previous (v4?) TCPM revision where the switching mostly worked fine.
(btw the subject has a typo, TPCM instead of TCPM :) )
Regards
Luca
