Re: [PATCH v5 6/8] usb: misc: onboard_dev: add support for non-hub devices

[Javier Carrasco]

On 28.02.24 21:41, Matthias Kaehlcke wrote:
> On Wed, Feb 28, 2024 at 09:21:00PM +0100, Javier Carrasco wrote:
>> On 28.02.24 19:10, Matthias Kaehlcke wrote:
>>> On Wed, Feb 28, 2024 at 02:51:33PM +0100, Javier Carrasco wrote:
>>>> Most of the functionality this driver provides can be used by non-hub
>>>> devices as well.
>>>>
>>>> To account for the hub-specific code, add a flag to the device data
>>>> structure and check its value for hub-specific code.
>>>>
>>>> The 'always_powered_in_supend' attribute is only available for hub
>>>> devices, keeping the driver's default behavior for non-hub devices (keep
>>>> on in suspend).
>>>>
>>>> Signed-off-by: Javier Carrasco <javier.carrasco@xxxxxxxxxxxxxx>
>>>> ---
>>>>  drivers/usb/misc/onboard_usb_dev.c | 25 +++++++++++++++++++++++--
>>>>  drivers/usb/misc/onboard_usb_dev.h | 10 ++++++++++
>>>>  2 files changed, 33 insertions(+), 2 deletions(-)
>>>>
>>>> diff --git a/drivers/usb/misc/onboard_usb_dev.c b/drivers/usb/misc/onboard_usb_dev.c
>>>> index e1779bd2d126..df0ed172c7ec 100644
>>>> --- a/drivers/usb/misc/onboard_usb_dev.c
>>>> +++ b/drivers/usb/misc/onboard_usb_dev.c
>>>> @@ -132,7 +132,8 @@ static int __maybe_unused onboard_dev_suspend(struct device *dev)
>>>>  	struct usbdev_node *node;
>>>>  	bool power_off = true;
>>>>  
>>>> -	if (onboard_dev->always_powered_in_suspend)
>>>> +	if (onboard_dev->always_powered_in_suspend &&
>>>> +	    !onboard_dev->pdata->is_hub)
>>>>  		return 0;
>>>
>>> With this non-hub devices would always be powered down, since
>>> 'always_powerd_in_suspend' is not set for them. This should be:
>>>
>>
>> May I ask you what you meant in v4 with this comment?
>>
>>> Even without the sysfs attribute the field 'always_powered_in_suspend'
>>> could
>>> be set to true by probe() for non-hub devices.
> 
> struct onboard_dev always has the field 'always_powered_in_suspend',
> even for non-hubs, that don't have the corresponding sysfs attribute.
> Currently it is left uninitialized (i.e. false) for non-hubs. Instead
> it could be initialized to true by probe() for non-hubs, which would
> be semantically correct. With that it wouldn't be necessary to check
> here whether a device is hub, because the field would provide the
> necessary information.
> 

That is maybe what is confusing me a bit. Should it not be false for
non-hub devices? That property is only meant for hubs, so why should
non-hub devices be always powered in suspend? I thought it should always
be false for non-hub devices, and configurable for hubs.

>>>   if (!onboard_dev->pdata->is_hub ||
>>>        onboard_dev->always_powered_in_suspend)
>>>
>>> Checking for the (non-)hub status first is clearer IMO, also it avoids
>>> an unneccessary check of 'always_powered' for non-hub devices.
>>>
>>
>> That makes sense and will be fixed.
>>
>>> Without code context: for hubs there can be multiple device tree nodes
>>> for the same physical hub chip (e.g. one for the USB2 and another for
>>> the USB3 part). I suppose this could also be the case for non-hub
>>> devices. For hubs there is the 'peer-hub' device tree property to
>>> establish a link between the two USB devices, as a result the onboard
>>> driver only creates a single platform device (which is desired,
>>> otherwise two platform devices would be in charge for power sequencing
>>> the same phyiscal device. For non-hub devices there is currently no such
>>> link. In many cases I expect there will be just one DT entry even though
>>> the device has multiple USB interfaces, but it could happen and would
>>> actually be a more accurate representation.
>>>
>>> General support is already there (the code dealing with 'peer-hub'), but
>>> we'd have to come up with a suitable name. 'peer-device' is the first
>>> thing that comes to my mind, but there might be better options. If such
>>> a generic property is added then we should deprecate 'peer-hub', but
>>> maintain backwards compatibility.
>>
>> I have nothing against that, but the first non-hub device that will be
>> added does not have multiple DT nodes, so I have nothing to test that
>> extension with real hardware.
> 
> I see, the XVF3500 is USB 2.0 only, so it isn't suitable for testing.
> 
>> That could be added in the future, though, if the need ever arises.
> 
> I expect it will, when a DT maintainer asks the hardware to be
> represented correctly for a device that is connected to more than one USB
> bus. IIRC that's how 'peer-hub' was born :)
> 
> Ok, we can leave it out for now. I might send a dedicated patch after your
> series landed. If a switch to 'peer-device' or similar is anticipated then
> it's probably best to deprecate 'peer-hub' ASAP, to avoid it from getting
> added to more bindings.

Best regards,
Javier Carrasco