The USB OTG port of the RK3308 exibits a bug when:

 - configured as peripheral, and
 - used in gadget mode, and
 - the USB cable is connected since before booting

The symptom is: about 6 seconds after configuring gadget mode the device is
disconnected and then re-enumerated. This happens only once per boot.

Investigation showed that in this configuration the charger detection code
turns off the PHY after 6 seconds. Patch 1 avoids this when a cable is
connected (VBUS present).

After patch 1 the connection is stable but communication stops after 6
seconds. this is addressed by patch 2.

The topic had been discussed in [0]. Thanks Alan and Minas for the
discussion and Louis for having found the 1st issue, leading to patch 1.

[0] https://lore.kernel.org/lkml/20250414185458.7767aabc@booty/

Luca

Signed-off-by: Luca Ceresoli <luca.ceresoli@bootlin.com>
---
Louis Chauvet (1):
      phy: rockchip: inno-usb2: fix disconnection in gadget mode

Luca Ceresoli (1):
      phy: rockchip: inno-usb2: fix communication disruption in gadget mode

 drivers/phy/rockchip/phy-rockchip-inno-usb2.c | 12 ++++++++----
 1 file changed, 8 insertions(+), 4 deletions(-)
---
base-commit: 15240745d6cd14183d112249a9fff87fb968b859
change-id: 20250718-rk3308-fix-usb-gadget-phy-disconnect-d7de71fb28b4

Best regards,
-- 
Luca Ceresoli <luca.ceresoli@bootlin.com>

Hello,

On Tue Jul 22, 2025 at 10:43 AM CEST, Luca Ceresoli wrote:
> The USB OTG port of the RK3308 exibits a bug when:
>
>  - configured as peripheral, and
>  - used in gadget mode, and
>  - the USB cable is connected since before booting
>
> The symptom is: about 6 seconds after configuring gadget mode the device is
> disconnected and then re-enumerated. This happens only once per boot.
>
> Investigation showed that in this configuration the charger detection code
> turns off the PHY after 6 seconds. Patch 1 avoids this when a cable is
> connected (VBUS present).
>
> After patch 1 the connection is stable but communication stops after 6
> seconds. this is addressed by patch 2.
>
> The topic had been discussed in [0]. Thanks Alan and Minas for the
> discussion and Louis for having found the 1st issue, leading to patch 1.
>
> [0] https://lore.kernel.org/lkml/20250414185458.7767aabc@booty/
>
> Luca
>
> Signed-off-by: Luca Ceresoli <luca.ceresoli@bootlin.com>

Anything I should do to move forward with this work?

These two few-liner patches solve real-world bugs, there is no complaint
about them, and both the bug and the solution is explained as clearly as
the public documentation allows.

FWIW, this series applied cleany on current master
(v6.18-rc2-8-g6548d364a3e8).

Best regards,
Luca

--
Luca Ceresoli, Bootlin
Embedded Linux and Kernel engineering
https://bootlin.com