Re: Help needed USB hub disconnected at resume

[Roger Quadros <rogerq@ti.com>]

On 03/06/2014 01:29 AM, Marc Murphy wrote:
>> -----Original Message-----
>> From: Felipe Balbi [mailto:balbi@ti.com]
>> Sent: 04 March 2014 23:43
>> To: Marc Murphy
>> Cc: 'balbi@ti.com'; 'Igor Grinberg'; Roger Quadros; linux-
>> omap@vger.kernel.org
>> Subject: Re: Help needed USB hub disconnected at resume
>>
>> On Tue, Mar 04, 2014 at 11:05:58PM +0000, Marc Murphy wrote:
>>>> -----Original Message-----
>>>> From: Felipe Balbi [mailto:balbi@ti.com]
>>>> Sent: 04 March 2014 22:44
>>>> To: Marc Murphy
>>>> Cc: 'Igor Grinberg'; Roger Quadros; linux-omap@vger.kernel.org
>>>> Subject: Re: Help needed USB hub disconnected at resume
>>>>
>>>> Hi,
>>>>
>>>> On Tue, Mar 04, 2014 at 10:34:56PM +0000, Marc Murphy wrote:
>>>>> static __init void tam3517_ehci_init(void) {
>>>>>     /* Configure GPIO for EHCI port */
>>>>>     omap_mux_init_gpio(TAM3517_EHCI_RESET, OMAP_PIN_OUTPUT);
>>>>>
>>>>>     gpio_request(TAM3517_EHCI_RESET, "USB_RESET");
>>>>>     gpio_direction_output(TAM3517_EHCI_RESET, 1);
>>>>>     gpio_export(TAM3517_EHCI_RESET, 0);
>>>>
>>>> why are you exporting this gpio ?
>>>
>>>
>>> It makes no difference whether I configure the GPIO or not.
>>
>> gpio_export() is only to expose the gpio to sysfs. Check if that pin is active
>> high or active low. Then what you need, most likely, is something like below:
>>
>> gpio_direction_output(RESET, HIGH);
>> usleep_range(50000, 2000000);
>> gpio_set_value(RESET, LOW);
>>
>> (assuming active high here)
> 
> Thanks, the export was for sysfs so that I could toggle the state myself to debug.  The reference has now been removed and the toggling left to the driver (ehci)
> 
> Igor, I have been looking at the ehci driver and the reset pin of the 3320 is toggled at initialisation but there is no resetting when resuming from sleep.  I have created a resume function;
> 

There should be no need to reset the PHY during suspend/resume.

Can you please try your board on v3.13? A lot of cleanup has been done since 3.4.

cheers,
-roger


> static void ehci_hcd_omap_resume(struct platform_device *pdev)
> {
>     struct device *dev              = &pdev->dev;
>     struct usb_hcd *hcd             = dev_get_drvdata(dev);
>     struct ehci_hcd_omap_platform_data *pdata   = dev->platform_data;
> 
> 
>     if (pdata->phy_reset) {
>         if (gpio_is_valid(pdata->reset_gpio_port[0]))
>             gpio_set_value(pdata->reset_gpio_port[0], 0);
> 
>         if (gpio_is_valid(pdata->reset_gpio_port[1]))
>             gpio_set_value(pdata->reset_gpio_port[1], 0);
> 
>         /* Hold the PHY in RESET for enough time till DIR is high */
>         udelay(100);
>     }
> 
>     if (pdata->phy_reset) {
>         /* Hold the PHY in RESET for enough time till
>          * PHY is settled and ready
>          */
>         udelay(10);
> 
>         if (gpio_is_valid(pdata->reset_gpio_port[0]))
>             gpio_set_value(pdata->reset_gpio_port[0], 1);
> 
>         if (gpio_is_valid(pdata->reset_gpio_port[1]))
>             gpio_set_value(pdata->reset_gpio_port[1], 1);
>     }
> }
> 
> And linked in the static struct platform_driver ehci_hcd_omap_driver = {
> 	.resume		= ehci_hcd_omap_resume,
> 
> It calls the function at resume and I can see the line toggle on the scope but still the same response from the driver of disconnecting the hub.
> 
> I have even stepped through ehci_bus_resume (struct usb_hcd *hcd) and it all seems to be OK with the bringup.
> 
> Is there any way to enable more debug so that I can see the path through the pm and hci code ?
> 
> I have enabled dynamic debug in the kernel but when I pass the dyndbg modules on the command line I don't seem to get any additional output... e.g.
> rootfstype=nfs ip=dhcp nohlt no_console_suspend=1 dyndbg=" module ehci_hcd +p ; module uhci_hcd +p ; module ohci_hcd +p" rw
> 
> 
> Marc
>>
>> --
>> balbi