Page 1 of 1

Problem with self-powered USB hubs

Posted: Thu Nov 16, 2017 12:21 pm
by sfd

We are having an issue with self-powered USB hubs on our custom, Sama5D4-based board.

If we plug the board to hubs which are not themselves connected to a host, we often get flooded with "ODD" log messages ( ... d45758abd3 - we are still using the 4.4 branch). Contrary to the comment, this is not exactly harmless because those messages can greatly delay the Linux boot-up sequence (journald apparently can't cope with that many messages).

This does not happen with non self-powered hubs.

If I understand the code correctly, these messages are caused by "ENDRESET" interrupts. I'd like to confirm this interrupt correspond to the section Reset signaling of the USB 1.1 specification:
A hub signals reset to a downstream port by driving an extended SE0 at the port. After the reset is
removed, the device will be in the Default state (refer to Section 9.1).
I suspect an electric problem around D+ and D- that causes the USB subsystem to see a reset condition.

Re: Problem with self-powered USB hubs

Posted: Tue Nov 28, 2017 10:16 am
by sfd

Could someone from Microchip at least confirm or deny that the ENDRESET interrupt is the result of a USB reset signaling described in the spec?

Re: Problem with self-powered USB hubs

Posted: Tue Nov 28, 2017 10:03 pm
by blue_z
For a Microchip response, try contacting Tech Support directly.