Re: [PATCH] mfd: omap-usb-host: AM3715 OHCI needs 120m functional clock

[Roger Quadros <rogerq-l0cyMroinI0@public.gmane.org>]

Ben,

On 13/10/15 11:23, Ben Dooks wrote:
> On 12/10/15 20:19, Tony Lindgren wrote:
>> * Ben Dooks <ben.dooks-4yDnlxn2s6sWdaTGBSpHTA@public.gmane.org> [151012 11:22]:
>>> On 12/10/15 18:45, Tony Lindgren wrote:
>>>> * Ben Dooks <ben.dooks-4yDnlxn2s6sWdaTGBSpHTA@public.gmane.org> [151012 10:38]:
>>>>> The AM3715 OHCI controller will not function without the EHCI
>>>>> unit's 120m fclk being enabled. If all the ports in the system
>>>>> are set to OHCI then the 120m_fclk will not get enabled and no
>>>>> devices are detected.
>>>>>
>>>>> Add a new (optional) property to signal the system must enable
>>>>> the 120m_fck for OHCI so that if no EHCI ports are signalled
>>>>> then the 120m_fclk should be enabled.
>>>>>
>>>>> We have found no information about why this is necessary, but
>>>>> it is suspected the EHCI controller does not complete the initial
>>>>> reset sequence and therefore does not hand control of the USB
>>>>> port back.
>>>>>
>>>>> Signed-off-by: Ben Dooks <ben.dooks-4yDnlxn2s6sWdaTGBSpHTA@public.gmane.org>
>>>>> ---
>>>>>  Documentation/devicetree/bindings/usb/omap-usb.txt | 3 +++
>>>>>  drivers/mfd/omap-usb-host.c                        | 4 ++++
>>>>>  2 files changed, 7 insertions(+)
>>>>>
>>>>> diff --git a/Documentation/devicetree/bindings/usb/omap-usb.txt b/Documentation/devicetree/bindings/usb/omap-usb.txt
>>>>> index 38d9bb8..fb5fea5 100644
>>>>> --- a/Documentation/devicetree/bindings/usb/omap-usb.txt
>>>>> +++ b/Documentation/devicetree/bindings/usb/omap-usb.txt
>>>>> @@ -23,6 +23,9 @@ OMAP MUSB GLUE
>>>>>  Optional properties:
>>>>>   - ctrl-module : phandle of the control module this glue uses to write to
>>>>>     mailbox
>>>>> + - ti,ohci-needs-120m-fck : bool, enable the 120m ehci clock even if just
>>>>> +   using ohci. Needed for AM3517 in OHCI only mode.
>>>>> +
>>>>>  
>>>>>  SOC specific device node entry
>>>>>  usb_otg_hs: usb_otg_hs@4a0ab000 {
>>>>> diff --git a/drivers/mfd/omap-usb-host.c b/drivers/mfd/omap-usb-host.c
>>>>> index 1d924d1..13880cf 100644
>>>>> --- a/drivers/mfd/omap-usb-host.c
>>>>> +++ b/drivers/mfd/omap-usb-host.c
>>>>> @@ -680,6 +680,10 @@ static int usbhs_omap_probe(struct platform_device *pdev)
>>>>>  				need_logic_fck |= true;
>>>>>  		}
>>>>>  
>>>>> +		/* The AM3517 requries the 120m-fck active to allow the OHCI to work */
>>>>> +		if (of_property_read_bool(dev->of_node, "ti,ohci-needs-120m-fck"))
>>>>> +			need_logic_fck |= true;
>>>>> +
>>>>>  		if (need_logic_fck) {
>>>>>  			omap->ehci_logic_fck = devm_clk_get(dev,
>>>>>  							    "usbhost_120m_fck");
>>>>
>>>> Hmm why not just use the standard device tree clocks property and then do
>>>> clk_get_rate() on the clock?
>>>
>>> I don't see that helps enabling the clock. The code decideds if
>>> no EHCI ports in use that it doesn't need to enable the EHCI fclk.
>>
>> Right, you need to do clk_prepare_enable() in it first? :)
> 
> No, if that was the case the driver would never work for the EHCI case.
> 
> The issue is:
> 
> 1) All ports on the system are set to OHCI
> 2) The omap-usb-host.c does not touch usbhost_120m_fck if no EHCI ports
> 3) The OHCI fails to detect any devices due to point 2.
> 

Instead of your existing approach why not just modify the preceeding
if condition that sets need_logic_fck to suit the OHCI case.

That way you don't need to add a new DT binding.

The old assumption was that 120m_fck logic clock is only needed for
EHCI mode but it looks like OHCI mode needs it as well.

You should also rename omap->ehci_logic_fck to omap->hci_logic_fck
as it is no longer ehci specific.

cheers,
-roger
--
To unsubscribe from this list: send the line "unsubscribe devicetree" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html