Re: [PATCH v2] arm64: dts: rockchip: add usb typec host support to rk3588-jaguar

From: "Heiko Stübner" <heiko@sntech.de>
To: Quentin Schulz <quentin.schulz@cherry.de>
Cc: linux-arm-kernel@lists.infradead.org,
	linux-rockchip@lists.infradead.org, linux-kernel@vger.kernel.org,
	Heiko Stuebner <heiko.stuebner@cherry.de>,
	Chen-Yu Tsai <wens@csie.org>
Subject: Re: [PATCH v2] arm64: dts: rockchip: add usb typec host support to rk3588-jaguar
Date: Wed, 26 Feb 2025 09:07:11 +0100	[thread overview]
Message-ID: <1952248.GKX7oQKdZx@diego> (raw)
In-Reply-To: <18716872.sWSEgdgrri@diego>

Am Freitag, 21. Februar 2025, 22:53:20 MEZ schrieb Heiko Stübner:
> Am Freitag, 21. Februar 2025, 17:43:08 MEZ schrieb Quentin Schulz:
> > Hi Heiko,
> > 
> > On 2/18/25 10:10 PM, Heiko Stuebner wrote:
> > > From: Heiko Stuebner <heiko.stuebner@cherry.de>
> > > 
> > > Jaguar has two type-c ports connected to fusb302 controllers that can
> > > work both in host and device mode and can also run in display-port
> > > altmode.
> > > 
> > > While these ports can work in dual-role data mode, they do not support
> > > powering the device itself as power-sink. This causes issues because
> > > the current infrastructure does not cope well with dual-role data
> > > without dual-role power.
> > > 
> > > So add the necessary nodes for the type-c controllers as well
> > > as enable the relevant core usb nodes, but limit the mode to host-mode
> > > for now until we figure out device mode.
> > > 
> > > Signed-off-by: Heiko Stuebner <heiko.stuebner@cherry.de>
> > 
> > While the USB functionality does work for both ports, and the 
> > orientation is properly reported, it seems like there may be some issue 
> > with how the PHY or controller interacts with that info because I do not 
> > get USB3 speeds when the device is inserted in reverse orientation, 
> > while I do when it's in normal orientation.
> 
> I've tested both ports and saw the issue too.
> Interestingly on the usbdp-phy side, orientation detection seems correct
> and also the sbu pins are set accordingly.
> 
> 
> > I assume that's the case for the Rock 5 ITX and Orange Pi 5+ as well and 
> > probably has nothing to do with the DT?
> > 
> > Should we go still go on with trying to merge this patch knowing that? I 
> > mean USB2 is still better than no USB at all :)
> > 
> > +Cc Chen-Yu, owner of an Orange Pi 5+, who may be able to confirm the 
> > issue is widespread.

For reference, after digging around I found out that the usbdp-phy does not
handle the orientation correctly but relies on specific behaviour of the dwc3
controller (suspending between unplug and plugin).

usbdp-phy series fixing that is in
https://lore.kernel.org/linux-rockchip/20250225184519.3586926-1-heiko@sntech.de/