On Mon, Aug 01, 2016 at 07:55:10AM -0700, Joshua Clayton wrote:

On 07/28/2016 09:41 AM, Fabio Estevam wrote:

quoted

Hi Joshua,

On Thu, Jul 28, 2016 at 12:56 PM, Joshua Clayton
[off-list ref] wrote:

quoted

I assume there is a v4 coming due to rmk's comments on patch 5.
I couldn't figure out where to null the of_node in error paths, but in testing
we did put a line of code to null the of_node on release.

but...
As an aside,
I was hoping this series would magically fix a problem
with the imx6q-evi which has forced us to disable
runtime power management. But it did not. :(

I also see a similar problem on a mx28 board with a hub.

Does it help in your case if you pass 'usbcore.autosuspend=-1' in the
kernel command line?

Regards,

Fabio Estevam

Thanks a million, Fabio!

'usbcore.autosuspend=-1' quiets the errors.

~joshua

If you are using imx6q, there are some patches which is not upstreamed
to workaround one SoC problem.

http://www.spinics.net/lists/linux-usb/msg98682.html

Try to implement the design for notify_suspend/resume at
drivers/usb/phy/phy-mxs-usb.c and drivers/usb/chipidea/host.c

You can refer to the code at:
http://git.freescale.com/git/cgit.cgi/imx/linux-2.6-imx.git/tree/?h=imx_4.1.15_1.0.0_ga

-- 

Best Regards,
Peter Chen