Re: [PATCH 14/16] usb: musb: dsps: add MUSB_DEVCTL_SESSION back after removal

From: Bin Liu
Date: Mon Jul 29 2013 - 14:41:45 EST


On Mon, Jul 29, 2013 at 12:51 PM, Sebastian Andrzej Siewior
<bigeasy@xxxxxxxxxxxxx> wrote:
> On 07/29/2013 07:26 PM, Bin Liu wrote:
>> Sebastian,
> Hi Bin,
>> I did not say AM335xEVM is not able to run OTG mode. The problem is in
>> OTG mode the SESSION bit will be cleared once the device is unplugged,
>> then there is no way the SESSION bit will come back in the current
>> mainline kernel.
> Yes, you did.

If I did, that was wrong statement.

>> The TI 3.2 kernel works around this OTG issue by toggling the SESSION
>> bit in b_idle handling in otg_timer(). The workaround makes most users
>> happy but it causes VBUS voltage pulsing. That is why I said I don't
>> know an ideal solution without a ID pin interrupt support.
> Now I understand more pieces of the puzzle. And the missing ID pin
> support is Am335x-evm specific since other am335x board may wire up
> that pin.

No, it is not board wiring issue. The AM335x EVM DOES have ID pin
wired to the ID pin of the micro-AB receptacle. I said AM335x does not
have IP pin INTERRUPT support, by which the kernel does not know when
a device is plugged in without setting the SESSION bit.

> But you say there is no way that the phy or anything else could help to
> avoid that pulsing.
>> I think you looked at a wrong file, maybe a wrong branch. Please check
>> [1], which defines
>> 2644 * mode[0:3] = USB0PORT's mode
>> 2645 * mode[4:7] = USB1PORT's mode
>> 2646 * AM335X beta EVM has USB0 in OTG mode and USB1 in host mode.
>> 2647 */
>> 2648 .mode = (MUSB_HOST << 4) | MUSB_OTG,
> Oh it is a long way. So looked into the wrong branch.
>>> I agree that without that try_idle part things keep working but it
>>> seems, that it is also called in the other tree.
>> Then because AM335xEVM USB1 port does not call *try_idle() in this 3.2
>> kernel, its SESSION bit is always set.
>> But I don't have a clear idea how to solve this in the mainline kernel
>> since is_otg_enabled() has been cleaned.
> So now I think this is a board problem. What about switching port 1
> from OTG to HOST only and avoiding kicking the timer in
> dsps_musb_try_idle() in such a case? It seems we only need to do this
> in OTG mode.

I have not read the MUSB driver in the mainline kernel yet. If somehow
port 1 can could be set to HOST mode then the SESSION bit would stay,
then that should solve the issue.

>> Regards,
>> -Bin.
> Sebastian
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at
Please read the FAQ at