Quoting Peter Chen (2016-09-14 01:03:21)

On Tue, Sep 13, 2016 at 10:58:26PM -0700, Stephen Boyd wrote:

quoted

Quoting Peter Chen (2016-09-13 20:32:00)

quoted

On Tue, Sep 13, 2016 at 06:42:46PM -0700, Stephen Boyd wrote:

quoted

On the db410c 96boards platform we have a TC7USB40MU[1] on the
board to mux the D+/D- lines from the SoC between a micro usb
"device" port and a USB hub for "host" roles. Upon a role switch,
we need to change this mux to forward the D+/D- lines to either
the port or the hub. Therefore, introduce a driver for this
device that intercepts extcon USB_HOST events and logically
asserts a gpio to mux the "host" D+/D- lines when a host cable is
attached. When the cable goes away, it will logically deassert
the gpio and mux the "device" lines.

Would you please draw the design? It can also help me review your
chipidea patch well.

1. How many ports on the board?
2. How the lines are connected on the board?

The schematic for the db410c is publically available here[2][3].

There's also the 96boards spec[4] which talks about this switch based
design a little bit. See the section titled "Single USB port Example".

[2] https://github.com/96boards/documentation/blob/master/ConsumerEdition/DragonBoard-410c/HardwareDocs/Schematics_DragonBoard.pdf
[3] https://github.com/96boards/documentation/raw/master/ConsumerEdition/DragonBoard-410c/HardwareDocs/Schematics_DragonBoard.pdf
[4] https://www.96boards.org/wp-content/uploads/2015/02/96BoardsCESpecificationv1.0-EA1.pdf

Ok, I see several use cases for this role switch

1. Using the hardware switch (218-4LPST)
In this case, you can set USB_SW_SEL as input gpio, and use
extcon-usb-gpio.c like before, just set this gpio as active
low at dts.

Nice! I didn't think of this case but it's good that we can support
that with some work.

2. Using USB_HS_ID as vbus-gpio (input), and USB_SW_SEL as id-gpio (output)

This is pretty much what has been implemented. USB_HS_ID is an
extcon-usb-gpio.c device.

I can't find hardware relationship between each other, maybe I miss
something.

I believe USB_SW_SEL is the physical switch (218-4LPST) while
USB_SW_SEL_PM is the software controllable part using a GPIO. USB_HS_ID
is just the vbus line from the uB connector and that line goes straight
into the SoC via a GPIO.

This use case (design) seems strange, usually, we use ID pin controls
vbus, but seldom use vbus pin control ID.
How you would like to implement it? When the USB cable is connected
(between PC), it receives vbus-gpio interrupt, then you set USB_SW_SEL
as low? If disconnected, you set USB_SW_SEL as high?

Right. The documented behavior is to detect the micro-usb cable and
drive USB_SW_SEL low. When the cable is unplugged we drive USB_SW_SEL
high. Maybe that should be changed though? If we always sampled
USB_SW_SEL as a USB_HOST extcon and the vbus line as a USB extcon then
we could allow the user to decide either with the physical switch or
with some sort of software control to toggle that gpio.

When the USB controller works at Host mode, what will happen if the user
connects USB cable at device port?

The devices on the two type-A connectors will be disconnected and we'll
switch from host mode to device mode.

3. Using sysfs to switch the role
Set USB_SW_SEL according to "role" at debugfs

sysfs isn't debugfs, but yes I wonder if we need to worry about the
debugfs role switching support here and toggle the gpio manually without
involving extcon. That would mean we need to have a way for the chipidea
controller to toggle this gpio/mux itself.

Which one you would like to implement? Or anything else I miss?

Mostly #2, but I'm concerned that the DT binding is going to force that
decision on others who may have the same switch and want to do #1 or #3.
So how to design it in a way that makes it work in all cases? Also, what
to do if the USB_SW_SEL switch is driven high by the physical switch?
That will "override" any software control we might be able to use, so
hopefully we can detect this somehow and prevent the role switch from
happening.