Regression: USB OTG port breaks after a few hours in host mode on iMX6
Peter Chen
peter.chen at freescale.com
Thu Oct 8 19:08:50 PDT 2015
On Thu, Oct 08, 2015 at 05:16:06PM +0100, Russell King - ARM Linux wrote:
> On Thu, Oct 08, 2015 at 09:52:52AM +0000, Peter Chen wrote:
> > I can't reproduce it for 5 hours, and will change pinmux (do the same
> > thing with your platform), and do the overnight test.
>
I still not reproduce it.
> There's definitely something weird going on. Over night last night,
> I left the Logitech universal receiver in the port, and this morning
> it was indicating in /proc/interrupts:
>
> 283: 50 0 0 0 GPC 43 Edge 2184000.usb
>
> I removed it, and now I have:
>
> 283: 50 0 1716 0 GPC 43 Edge 2184000.usb
>
> which is increasing at a rate of 90 per minute.
It is strange your interrupt occurs at cpu2 at that time, and
so many interrupts during the removal. In order to avoid other
interrupts occurring during the removal, would you please change
your dts like below:
&usbotg {
vbus-supply = <®_usb_otg_vbus>;
dr_mode = "host";
disable-over-current;
status = "okay";
};
>
> Nothing in the kernel message log indicating why this may be. It looks
> like runtime PM doesn't work on this port:
>
> /sys/bus/platform/devices/2184000.usb/power/runtime_active_time:109850496
> /sys/bus/platform/devices/2184000.usb/power/runtime_status:active
> /sys/bus/platform/devices/2184000.usb/power/runtime_suspended_time:0
>
This may be correct, assume your usb receiver does not support runtime
pm, so the controller is active when it is on the port, when the
removal, the interrupt occurs so frequently, it has no chance to enter
suspend status.
> whereas the other port (which has zero interrupts) it does:
>
> /sys/bus/platform/devices/2184200.usb/power/runtime_active_time:16924
> /sys/bus/platform/devices/2184200.usb/power/runtime_status:suspended
> /sys/bus/platform/devices/2184200.usb/power/runtime_suspended_time:109861760
>
> --
> FTTC broadband for 0.8mile line: currently at 9.6Mbps down 400kbps up
> according to speedtest.net.
--
Best Regards,
Peter Chen
More information about the linux-arm-kernel
mailing list