Re: [PATCH v5 4/6] usb: roles: add API to get usb_role_switch by node

From: Chunfeng Yun
Date: Thu May 23 2019 - 06:19:09 EST


Hi Heikki,
On Wed, 2019-05-22 at 17:26 +0300, Heikki Krogerus wrote:
> On Wed, May 22, 2019 at 10:55:17AM +0000, Biju Das wrote:
> > Hi Chunfeng Yun,
> >
> > Thanks for the feedback.
> >
> > > Subject: RE: [PATCH v5 4/6] usb: roles: add API to get usb_role_switch by
> > > node
> > >
> > > Hi Biju,
> > > On Wed, 2019-05-22 at 08:05 +0000, Biju Das wrote:
> > > > Hi Heikki,
> > > >
> > > > Thanks for the feedback.
> > > >
> > > > > Subject: Re: [PATCH v5 4/6] usb: roles: add API to get
> > > > > usb_role_switch by node
> > > > >
> > > > > On Mon, May 20, 2019 at 09:45:46AM +0000, Biju Das wrote:
> > > > > >
> > > > > >
> > > > > > Hi Heikki,
> > > > > >
> > > > > > Thanks for the feedback.
> > > > > >
> > > > > > > Subject: Re: [PATCH v5 4/6] usb: roles: add API to get
> > > > > > > usb_role_switch by node
> > > > > > >
> > > > > > > On Mon, May 20, 2019 at 08:06:41AM +0000, Biju Das wrote:
> > > > > > > > Hi Heikki,
> > > > > > > >
> > > > > > > > > Subject: Re: [PATCH v5 4/6] usb: roles: add API to get
> > > > > > > > > usb_role_switch by node
> > > > > > > > >
> > > > > > > > > On Mon, May 20, 2019 at 10:39:11AM +0800, Chunfeng Yun wrote:
> > > > > > > > > > Hi,
> > > > > > > > > > On Fri, 2019-05-17 at 16:05 +0300, Heikki Krogerus wrote:
> > > > > > > > > > > Hi,
> > > > > > > > > > >
> > > > > > > > > > > On Fri, May 17, 2019 at 01:37:36PM +0300, Heikki Krogerus
> > > wrote:
> > > > > > > > > > > > On Tue, May 14, 2019 at 04:47:21PM +0800, Chunfeng Yun
> > > > > wrote:
> > > > > > > > > > > > > Add fwnode_usb_role_switch_get() to make easier to
> > > > > > > > > > > > > get usb_role_switch by fwnode which register it.
> > > > > > > > > > > > > It's useful when there is not device_connection
> > > > > > > > > > > > > registered between two drivers and only knows the
> > > > > > > > > > > > > fwnode which register usb_role_switch.
> > > > > > > > > > > > >
> > > > > > > > > > > > > Signed-off-by: Chunfeng Yun
> > > > > > > > > > > > > <chunfeng.yun@xxxxxxxxxxxx>
> > > > > > > > > > > > > Tested-by: Biju Das <biju.das@xxxxxxxxxxxxxx>
> > > > > > > > > > > >
> > > > > > > > > > > > Acked-by: Heikki Krogerus
> > > > > > > > > > > > <heikki.krogerus@xxxxxxxxxxxxxxx>
> > > > > > > > > > >
> > > > > > > > > > > Hold on. I just noticed Rob's comment on patch 2/6,
> > > > > > > > > > > where he points out that you don't need to use device
> > > > > > > > > > > graph since the controller is the parent of the
> > > > > > > > > > > connector. Doesn't that mean you don't really need this API?
> > > > > > > > > > No, I still need it.
> > > > > > > > > > The change is about the way how to get fwnode; when use
> > > > > > > > > > device graph, get fwnode by of_graph_get_remote_node();
> > > > > > > > > > but now will get fwnode by of_get_parent();
> > > > > > > > >
> > > > > > > > > OK, I get that, but I'm still not convinced about if
> > > > > > > > > something like this function is needed at all. I also have
> > > > > > > > > concerns regarding how you are using the function. I'll
> > > > > > > > > explain in comment to the patch 5/6 in this
> > > > > > > series...
> > > > > > > >
> > > > > > > > FYI, Currently I am also using this api in my patch series.
> > > > > > > > https://patchwork.kernel.org/patch/10944637/
> > > > > > >
> > > > > > > Yes, and I have the same question for you I jusb asked in
> > > > > > > comment I added to the patch 5/6 of this series. Why isn't
> > > > > > > usb_role_switch_get()
> > > > > enough?
> > > > > >
> > > > > > Currently no issue. It will work with this api as well, since the
> > > > > > port node is
> > > > > part of controller node.
> > > > > > For eg:-
> > > > > > https://patchwork.kernel.org/patch/10944627/
> > > > > >
> > > > > > However if any one adds port node inside the connector node, then
> > > > > > this
> > > > > api may won't work as expected.
> > > > > > Currently I get below error
> > > > > >
> > > > > > [ 2.299703] OF: graph: no port node found in
> > > > > /soc/i2c@e6500000/hd3ss3220@47
> > > > >
> > > > > We need to understand why is that happening?
> > > > >
> > > >
> > > > Form the stack trace the parent node is "parent_node=hd3ss3220@47" ,
> > > instead of the "connector" node.
> > > > That is the reason for the above error.
> > > >
> > > > [ 2.442429] of_graph_get_next_endpoint.part.0+0x28/0x168
> > > > [ 2.447889] of_fwnode_graph_get_next_endpoint+0x5c/0xb0
> > > > [ 2.453267] fwnode_graph_get_next_endpoint+0x20/0x30
> > > > [ 2.458374] device_connection_find_match+0x74/0x1a0
> > > > [ 2.463399] usb_role_switch_get+0x20/0x28
> > > > [ 2.467542] hd3ss3220_probe+0xc4/0x218
> > > >
> > > > The use case is
> > > >
> > > > &i2c0 {
> > > > hd3ss3220@47 {
> > > > compatible = "ti,hd3ss3220";
> > > >
> > > > usb_con: connector {
> > > > compatible = "usb-c-connector";
> > > > port {
> > > > hd3ss3220_ep: endpoint {
> > > > remote-endpoint =
> > > <&usb3_role_switch>;
> > > > };
> > > > };
> > > > };
> > > > };
> > > > };
> > > >
> > > > &usb3_peri0 {
> > > > companion = <&xhci0>;
> > > > usb-role-switch;
> > > >
> > > > port {
> > > > usb3_role_switch: endpoint {
> > > > remote-endpoint = <&hd3ss3220_ep>;
> > > > };
> > > > };
> > > > };
> > > >
> > > > Q1) How do we modify the usb_role_switch_get() function to search
> > > > Child(connector) and child's endpoint?
> > > How about firstly finding connector node in fwnode_graph_devcon_match(),
> > > then search each endpoint?
> >
> > I have done a quick prototyping with the changes you suggested and it works.
> >
> > - struct fwnode_handle *ep;
> > + struct fwnode_handle *ep,*child,*tmp = fwnode;
> >
> > - fwnode_graph_for_each_endpoint(fwnode, ep) {
> > + child = fwnode_get_named_child_node(fwnode, "connector");
> > + if (child)
> > + tmp = child;
> > +
> > + fwnode_graph_for_each_endpoint(tmp, ep) {
> >
> > Form the stack trace the parent node is "parent_node= connector" .
> >
> > [ 2.440922] of_graph_get_next_endpoint.part.0+0x28/0x168
> > [ 2.446381] of_fwnode_graph_get_next_endpoint+0x5c/0xb0
> > [ 2.451758] fwnode_graph_get_next_endpoint+0x20/0x30
> > [ 2.456866] device_connection_find_match+0x84/0x1c0
> > [ 2.461888] usb_role_switch_get+0x20/0x28
> >
> > Heikki,
> > Are you ok with the above changes?
>
> Doesn't that mean that if we made fwnode_usb_role_switch_get() the way
> I proposed, there is no problem? You just find the "connector" child
> node in your driver, and pass that to fwnode_usb_role_switch_get():
>
> struct fwnode_handle *connector;
> ...
> connector = device_get_named_child_node(&client->dev, "connector");
> if (IS_ERR(connector))
> <do something>
>
> hd3ss3220->role_sw = fwnode_usb_role_switch_get(connector);
> ...
>
> The difference is that instead of just converting a device node of an
> usb role switch to the usb role switch, it works just like
> usb_role_switch_get(), just taking fwnode instead of device entry as
> parameter.
>
> I prepared the patches implementing fwnode_usb_role_switch_get() the
> way I though it needs to work for my own tests. Please find the
> patches attached.
I tested these patches, it didn't work for case as following:

case 2:

&mtu3 {
usb-role-switch;

connector {
compatible = "linux,typeb-conn-gpio", "usb-b-connector";
label = "micro-USB";
type = "micro";
id-gpios = <&pio 12 GPIO_ACTIVE_HIGH>;
vbus-supply = <&usb_p0_vbus>;
};
};

so I consider this case in the function fwnode_graph_devcon_match()
(based on your patches)

diff --git a/drivers/base/devcon.c b/drivers/base/devcon.c
index 8311b70..1dae8b7 100644
--- a/drivers/base/devcon.c
+++ b/drivers/base/devcon.c
@@ -18,9 +18,11 @@
{
struct device_connection con = { .id = con_id };
struct fwnode_handle *ep;
+ int ep_count = 0;
void *ret;

fwnode_graph_for_each_endpoint(fwnode, ep) {
+ ep_count++;
con.fwnode = fwnode_graph_get_remote_port_parent(ep);
if (!fwnode_device_is_available(con.fwnode))
continue;
@@ -32,6 +34,19 @@
return ret;
}
}
+
+ /* if the connector has no remote endpoint, check its parent */
+ if (!ep_count) {
+ con.fwnode = fwnode_get_parent(fwnode);
+ if (!con.fwnode)
+ return NULL;
+
+ ret = match(&con, -1, data);
+ fwnode_handle_put(con.fwnode);
+ if (ret)
+ return ret;
+ }
+
return NULL;
}

see attached patch.

then both usb_role_switch_get(dev) or fwnode_usb_role_switch_get(fwnode)
work well;

but I don't know which way is better when consider this specific case,
put into class.c as you suggested before or put into devcon.c like
above.

Thanks a lot

>
> thanks,
>