Re: [PATCH] phy: core: Add consumer device link support

[Alexandre Torgue <alexandre.torgue@st.com>]

Hi,

On 2/14/20 7:46 PM, Andy Shevchenko wrote:
> On Mon, Feb 10, 2020 at 1:32 PM Alexandre Torgue
> <alexandre.torgue@st.com> wrote:
>> On 2/10/20 9:08 AM, Kishon Vijay Abraham I wrote:
>>> On 07/02/20 12:27 PM, youling 257 wrote:
>>>> test this diff, dwc3 work for my device, thanks.
>>>>
>>>> 2020-02-07 13:16 GMT+08:00, Kishon Vijay Abraham I <kishon@ti.com>:
>>>>> On 06/02/20 7:09 PM, youling257 wrote:
>>>>>> This patch cause "dwc3 dwc3.3.auto: failed to create device link to
>>>>>> dwc3.3.auto.ulpi" problem.
>>>>>> https://bugzilla.kernel.org/show_bug.cgi?id=206435
> 
> +1 to the report.
> Please revert for v5.6 or provide a fix ASAP!
> 

Kishon, do you plan to do the fix? If not, I'll send it quickly.

Thanks
Alex

>>>>>
>>>>> I'm suspecting there is some sort of reverse dependency with dwc3 ULPI.
>>>>> Can you try the following diff?
>>>>>
>>>>> diff --git a/drivers/phy/phy-core.c b/drivers/phy/phy-core.c
>>>>> index 2eb28cc2d2dc..397311dcb116 100644
>>>>> --- a/drivers/phy/phy-core.c
>>>>> +++ b/drivers/phy/phy-core.c
>>>>> @@ -687,7 +687,7 @@ struct phy *phy_get(struct device *dev, const char
>>>>> *string)
>>>>>
>>>>>           get_device(&phy->dev);
>>>>>
>>>>> -       link = device_link_add(dev, &phy->dev, DL_FLAG_STATELESS);
>>>>> +       link = device_link_add(dev, &phy->dev, DL_FLAG_SYNC_STATE_ONLY);
>>>>>           if (!link) {
>>>>>                   dev_err(dev, "failed to create device link to %s\n",
>>>>>                           dev_name(phy->dev.parent));
>>>>> @@ -802,7 +802,7 @@ struct phy *devm_of_phy_get(struct device *dev,
>>>>> struct device_node *np,
>>>>>                   return phy;
>>>>>           }
>>>>>
>>>>> -       link = device_link_add(dev, &phy->dev, DL_FLAG_STATELESS);
>>>>> +       link = device_link_add(dev, &phy->dev, DL_FLAG_SYNC_STATE_ONLY);
>>>>>           if (!link) {
>>>>>                   dev_err(dev, "failed to create device link to %s\n",
>>>>>                           dev_name(phy->dev.parent));
>>>>> @@ -851,7 +851,7 @@ struct phy *devm_of_phy_get_by_index(struct device
>>>>> *dev, struct device_node *np,
>>>>>           *ptr = phy;
>>>>>           devres_add(dev, ptr);
>>>>>
>>>>> -       link = device_link_add(dev, &phy->dev, DL_FLAG_STATELESS);
>>>>> +       link = device_link_add(dev, &phy->dev, DL_FLAG_SYNC_STATE_ONLY);
>>>>>           if (!link) {
>>>>>                   dev_err(dev, "failed to create device link to %s\n",
>>>>>                           dev_name(phy->dev.parent));Parent
>>>
>>> Can you check if this doesn't affect the suspend/resume ordering?
>>
>> With this fix, suspend/resume ordering is broken on my side. What do you
>> think to keep the STATELESS flag and to only display a warn if
>> "device_link_add" returns an error ? It's not "smart" but it could
>> solved our issue.
>>
>> As a lot of improvements have been recently done on device link topic by
>> Saravana, we could check with him what is the way to follow.
>>
>> Regards
>> Alex
>>
>>>
>>> Thanks
>>> Kishon
>>>
> 
> 
>