Message-id: <552E2EBF.5090906@samsung.com>
Date: Wed, 15 Apr 2015 18:26:23 +0900
From: Chanwoo Choi <cw00.choi@samsung.com>
User-Agent: Mozilla/5.0 (X11; Linux i686; rv:17.0) Gecko/20130106
 Thunderbird/17.0.2
MIME-version: 1.0
To: Roger Quadros <rogerq@ti.com>
Cc: Peter Chen <peter.chen@freescale.com>,
        Robert Baldyga <r.baldyga@samsung.com>, myungjoo.ham@samsung.com,
        linux-usb@vger.kernel.org, devicetree@vger.kernel.org,
        linux-kernel@vger.kernel.org, m.szyprowski@samsung.com,
        "Balbi, Felipe" <balbi@ti.com>,
        "ABRAHAM, KISHON VIJAY" <kishon@ti.com>
Subject: Re: [PATCH v3 2/4] extcon: usb-gpio: add support for VBUS detection
References: <55277910.2070000@samsung.com> <55277FA2.6090700@samsung.com>
 <55278580.2090809@samsung.com> <55278DD3.3020409@samsung.com>
 <55279550.1060609@samsung.com> <552CE579.6060702@ti.com>
 <552CE5C0.3010207@ti.com> <552CEC97.1050205@samsung.com>
 <552CEE42.1050907@ti.com> <552CFA1E.5070400@samsung.com>
 <20150415032734.GA29095@shlinux2> <552E1862.6090709@ti.com>
In-reply-to: <552E1862.6090709@ti.com>
Content-type: text/plain; charset=windows-1252
Content-transfer-encoding: 7bit
DLP-Filter: Pass
Sender: linux-kernel-owner@vger.kernel.org
Content-Length: 9423
Lines: 207

Hi Roger and Peter,

On 04/15/2015 04:50 PM, Roger Quadros wrote:
> On 15/04/15 06:27, Peter Chen wrote:
>> On Tue, Apr 14, 2015 at 08:29:34PM +0900, Chanwoo Choi wrote:
>>> On 04/14/2015 07:38 PM, Roger Quadros wrote:
>>>> On 14/04/15 13:31, Chanwoo Choi wrote:
>>>>> On 04/14/2015 07:02 PM, Roger Quadros wrote:
>>>>>> Fixed Kishon's id.
>>>>>>
>>>>>> On 14/04/15 13:01, Roger Quadros wrote:
>>>>>>> On 10/04/15 12:18, Chanwoo Choi wrote:
>>>>>>>> On 04/10/2015 05:46 PM, Robert Baldyga wrote:
>>>>>>>>> On 04/10/2015 10:10 AM, Chanwoo Choi wrote:
>>>>>>>>>> On 04/10/2015 04:45 PM, Robert Baldyga wrote:
>>>>>>>>>>> On 04/10/2015 09:17 AM, Chanwoo Choi wrote:
>>>>>>>>>>>> Hi Robert,
>>>>>>>>>>>>
>>>>>>>>>>>> On 04/09/2015 06:24 PM, Robert Baldyga wrote:
>>>>>>>>>>>>> Hi Chanwoo,
>>>>>>>>>>>>>
>>>>>>>>>>>>> On 04/09/2015 11:07 AM, Chanwoo Choi wrote:
>>>>>>>>>>>>>> Hi Robert,
>>>>>>>>>>>>>>
>>>>>>>>>>>>>> On 04/09/2015 04:57 PM, Robert Baldyga wrote:
>>>>>>>>>>>>>>> Hi Chanwoo,
>>>>>>>>>>>>>>>
>>>>>>>>>>>>>>> On 04/09/2015 04:12 AM, Chanwoo Choi wrote:
>>>>>>>>>>>>>>>> Hi Robert,
>>>>>>>>>>>>>>>>
>>>>>>>>>>>>>>
>>>>>>>>>>>>>> [snip]
>>>>>>>>>>>>>>
>>>>>>>>>>>>>>>> But, I have one question about case[3]
>>>>>>>>>>>>>>>>
>>>>>>>>>>>>>>>> If id is low and vbus is high, this patch will update the state of both USB and USB-HOST cable as attached state.
>>>>>>>>>>>>>>>> Is it possible that two different cables (both USB and USB-HOST) are connected to one port simultaneously?
>>>>>>>>>>>>>>>>
>>>>>>>>>>>>>>>
>>>>>>>>>>>>>>> It's because state of single USB cable connection cannot be completely
>>>>>>>>>>>>>>> described using single extcon cable. USB cable state has two bits (VBUS
>>>>>>>>>>>>>>> and ID), so we need to use two cables for single cable connection. We
>>>>>>>>>>>>>>> use following convention:
>>>>>>>>>>>>>>> cable "USB" = VBUS
>>>>>>>>>>>>>>> cable "USB-HOST" = !ID.
>>>>>>>>>>>>>>
>>>>>>>>>>>>>> I think that extcon provider driver have to update the only one cable state
>>>>>>>>>>>>>> of either USB or USB-HOST because USB and USB-HOST feature can not be used
>>>>>>>>>>>>>> at the same time through one h/w port.
>>>>>>>>>>>>>>
>>>>>>>>>>>>>> If extcon-usb-gpio.c update two connected event of both USB and USB-HOST cable
>>>>>>>>>>>>>> at the same time, the extcon consumer driver can not decide what handle either USB or USB-HOST.
>>>>>>>>>>>>>>
>>>>>>>>>>>>>
>>>>>>>>>>>>> It can. USB OTG allows for that. Moreover device can be host even if
>>>>>>>>>>>>> ID=1 (so detected cable type is USB device), or peripheral when ID=0 (so
>>>>>>>>>>>>> detected cable type is USB host). Devices would need to have complete
>>>>>>>>>>>>> information about USB cable connection, because OTG state machine needs
>>>>>>>>>>>>
>>>>>>>>>>>> As I knew, USB OTG port don't send the attached cable of both USB and USB-HOST
>>>>>>>>>>>> at the same time. The case3 in your patch update two cable state about one h/w port.
>>>>>>>>>>>>
>>>>>>>>>>>
>>>>>>>>>>> It's because simple "USB" or "USB-HOST" means nothing for USB OTG
>>>>>>>>>>> machine. It needs to know exact VBUS and ID states, which cannot be
>>>>>>>>>>> concluded basing on cable type only. That's why I have used "USB-HOST"
>>>>>>>>>>> name together with "USB" to pass additional information about USB cable
>>>>>>>>>>> connection.
>>>>>>>>>>
>>>>>>>>>> I think this method is not proper to support this case.
>>>>>>>>>> It may cause the confusion about other case using USB/USB-HOST cable state
>>>>>>>>>> except of you commented case.
>>>>>>>>>
>>>>>>>>> That's why I finally proposed to use "USB-ID" and "USB-VBUS" in parallel
>>>>>>>>> with old names. It seems to be simpler solution than adding new
>>>>>>>>> mechanism notifying about VBUS and ID states changes.
>>>>>>>>
>>>>>>>>
>>>>>>>> As I commented on previous reply, I don't agree to use 'USB-ID' and 'USB-VBUS'.
>>>>>>>> If we add new strange 'USB-ID' and 'USB-VBUS' name, we would add non-general cable
>>>>>>>> name continuoulsy.
>>>>>>>>
>>>>>>>> I think that extcon core provide the helper API to get the value of VBUS.
>>>>>>>> But I need to consider it.
>>>>>>>
>>>>>>> Now it is starting to look like existing extcon states are not suitable for USB/PHY drivers to deliver
>>>>>>> VBUS and ID information reliably.
>>>>>>> This is because based on your comments the "USB" and "USB-HOST" states look like some fuzzy states
>>>>>>> and have no direct correspondence with "VBUS" and "ID". The fact that they can't become
>>>>>>> attached simultaneously makes me conclude that "USB" and "USB-HOST" cable states are really 
>>>>>>> capturing only the ID pin state.
>>>>>>>
>>>>>>> I can suggest the following options
>>>>>>> a) let "USB" and "USB-HOST" only indicate ID pin status. Add a new cable state for "VBUS" notification.
>>>>>>> Maybe call it "USB-POWER" or something.
>>>>>
>>>>> We must discuss it before using the new cable name
>>>>> such as "USB-POWER", "USB-ID" and "USB-VBUS".
>>>>
>>>> I didn't say to add "USB-ID" or "USB-VBUS". solution (a) was to have the following
>>>
>>> Right. Robert suggested the "USB-ID" and "USB-VBUS" cable name on previous mail in mail thread.
>>
>> From USB/USB-PHY driver point, it needs to know id and vbus value
>> for their internal logic, so as extcon users, the cable name
>> is better to reflect meaning of id and vbus, like "USB-ID" and "USB-VBUS",
>> if the power is from vbus pin at USB cable, I don't think we need another
>> cable name "USB-POWER" even the USB/USB-PHY driver don't need it.
> 
> I agree as well that this is the *best* option for USB case. Just because Chanwoo was
> objecting these names I suggested "USB-POWER".
> 
> Chanwoo, can we simply get rid of "USB" and "USB-HOST" cables and move to
> "USB-ID" and "USB-VBUS"?

I'm wondering about changing the previous cable name from 'USB'/'USB-HOST'
to 'USB-ID/USB-VBUS' because extcon framework update the state of cable by
using uevent and the user-space process would catch the changed state by
using cable name ('USB'/'USB-HOST').

The user-space process may not consider the both id and vbus of USB.
If 'USB-ID'/'USB-VBUS' cable name is used instead of 'USB'/'USB-HOST',
It may cause the confusion about what is meaning of cable name
on user-space process.

So,
I prefer to use existing 'USB' and 'USB-HOST' cable name.
and then want to add additional method to get the vbus state.

I think two following method to get the vbus state.
1) Add the extcon_{get|set}_vbus_state()
- extcon_{get|set}_vbus_state()
- the list of of return value
	#define EXTCON_USB_VBUS_OFF	0
	#define EXTCON_USB_VBUS_ON	1

When USB/USB-HOST is attached and receive the notification onextcon consumer driver
,extcon consumer driver would get the vbus state by extcon_get_vbus_state().

2) Add the notifier chain for vbus state update
- extcon_{register|unregister}_vbus_notifier()
- the list of notifier event
	#define EXTCON_USB_VBUS_OFF	0
	#define EXTCON_USB_VBUS_ON	1


3) add the new cable 'USB-POWER' by Roger suggestion .
- When 'USB-POWER' cable is attached, extcon will update the cable state
'USB-POWER' means only the vbus state. But, 'USB-POWER' is not h/w cable.
The user-space process would handle this uevent of 'USB-POWER'
such as h/w cable's uevent. I think it is not clear on the user-space process aspect.

> 
> The only reason you objected was saying that it is a strange cable name. Well this is
> only what we care about from USB PHY drivers and user space is not interested in it
> so what is the concern?

I added the reason why don't want to change the legacy cable name about USB/USB-HOST.

> 
>>
>>>
>>>> "USB" - attached means ID is high. i.e. Type-B plug attached.
>>>> "USB-HOST" - attached means ID is low. i.e. Type-A plug attached.
>>>> "USB-POWER" - attached means USB power is present. i.e. VBUS is alive.
>>>>
>>>> This way the definition of USB and USB-HOST remain true to their name and avoid further confusions.
>>>> VBUS state is got through the "USB-POWER" cable state.
>>>
>>> There is the same case for MHL cable.
>>> Also, MHL cable could be connected to VBUS line.
>>> - MHL : attached just MHL cable.
>>> - MHL-POWER : attache MHL cable which is connected with VBUS line.
>>>
>>> We must need the opinion of USB/PHY driver's maintainer.
>>>
>>>>
>>>>>
>>>>> What is the appropriate method of following two solution?
>>>>> - Fisrt, use the new cable name "USB-*".
>>>> I explained this above.
>>>>
>>>>> - Second, use the additional API to get the VBUS state.
>>>>
>>>> You keep mentioning additional API for VBUS. But I don't see any such API. Can you please
>>>> suggest what API you are talking about?
>>>
>>> I'm considering following functions for VBUS state. But it is my opinion,
>>> If USB/PHY drivers's maintainers don't agree the new cable ("USB-POWER"),
>>> We could use the following function to get VBUS state.
>>> Because new cable name will affect the USB/PHY drivers.
>>> - int extcon_{get|set}_vbus_state(struct extcon_dev *edev);
> 
> This is not suitable for us as USB drivers need VBUS notification event to come.
> They can't keep polling for VBUS state using this API.

I don't agree the polling method. You refer to upper my description.

Thanks,
Chanwoo Choi


--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/