Hi Enric,

Am Mittwoch, 15. August 2018, 13:08:00 CEST schrieb Enric Balletbo i Serra:

On 15/08/18 12:29, Heiko Stuebner wrote:

quoted

Am Mittwoch, 15. August 2018, 11:59:32 CEST schrieb Enric Balletbo i Serra:

quoted

Commit 98898f3bc83c8 ("phy: rockchip-inno-usb2: support otg-port for
rk3399") introduces two new properties. The extcon property is used to
detect the cable-state, and the rockchip,utmi-avalid is used to indicate
which register should be used to detect the vbus state.

Document these properties in the documentation binding.

Fixes: 98898f3bc83c8 ("phy: rockchip-inno-usb2: support otg-port for rk3399")
Signed-off-by: Enric Balletbo i Serra <redacted>

[...]

quoted

quoted

@@ -45,6 +46,8 @@ Required properties (port (child) node):
 Optional properties:
  - phy-supply : phandle to a regulator that provides power to VBUS.
 		See ./phy-bindings.txt for details.
+ - rockchip,utmi-avalid : boolean, use the avalid register to get vbus status.
+			  Otherwise, use the bvalid register.

Not having looked to deeply into the usb2 phy, this might raise questions
on why this is a hardware-description? Is this needed when something is not
connected on the board?

I asked myself the same question and even I thought in just remove that code.

After some investigation, though, I saw that the UTMI+ specification [1] has two
signals similar to ID signal (page 11), the AValid signal is used to indicate if
the session for an A-peripheral is valid and the BValid signal that is used to
indicate if the session for a B-peripheral is valid. I suppose that use of one
or the other matters in some cases, but AFAICT this is not used and I didn't see
any binding using it.

Maybe someone else can give us more clues on the importance or not of this property?

so I've looked in mainline, chromeos-4.4 and the Rockchip vendor-kernel
and the only board using that property at all is the rk3399-evb-rev1
and -rev2 in the vendor kernel.

The existence of a further -rev3 (which also looks way better cared for
compared rev1+2) indicates that the older ones are probably some sort
of preproduction models, where some wiring (on the soc or board) may
have gone wrong.

So while I would keep all the avalid settings in the driver, we could just
drop reading that property quietly - as Rob wrote some days ago
"it's only an incompatible change if someone notices" [0] and from the
above it doesn't look like it ;-) .

Heiko

[0] https://www.spinics.net/lists/devicetree/msg243978.html