Short SOF of 92 us on Full Speed Bus

[USBVIDPID]

I see using Ellisys USB tracing a short SOF of 92 us at about 8.5 to 8.6 seconds after every USB suspend - resume cycle.  We are doing a test of 2500 suspend - resume cycles.  Sometimes I see this short SOF causes the down steam USB device to stop responding to polls and sometimes I see the device do a watchdog reset.  I also see on Windows XP but not on Windows 7 all of the USB HS hubs no longer accept any device hot plugged to it.  This condition is all fixed after a Windows reboot.  I have not tried Linux to see if it happens on it too.  But I have seen this on at least two different PCs and restating, Windows XP and Windows 7.

What I really want to know is how to best debug this to see what in the OS or even the BIOS that might be causing this short SOF.  I am suspecting that the OS is perhaps issuing a redundant USB host controller command, but I need proof to get it fixed.  One coworker suggested NetMon.exe but I think this might be better for LAN problems. 

Thanks for your time, jmw

[Jan Axelson]

The host-controller driver is responsible for scheduling traffic on the bus.

I'm not sure what you mean by "a short SOF of 92 us".

A high-speed SOF occurs 92 microsecs after the previous SOF, instead of 125 microsecs after?

[USBVIDPID]

I am tracing on a USB Full Speed bus and its SOF rate is 1 per ms.  The USB spec. says, "in chapter 8.4.3 from USB spec 2.0, the host should issue SOF every 1.0 ms +/- 0.0005ms for a full speed bus".

See the attachment:  SOF screen shot.gif  (this is from an Ellisys full speed USB trace).

Time reference is set to 0 for Frame # 640 and Frame # 641 shows that it is 92 us later instead of 1 ms.

Thanks, jmw

[attachment deleted by admin]

[Jan Axelson]

OK, the bad SOF is way early (0.000092 sec instead of 0.001 sec), but the next one arrives as if the previous one had been on time (0.002 sec).

[Tsuneo]

USBVIDPID:
One coworker suggested NetMon.exe but I think this might be better for LAN problems.

Maybe, he refers to this method, in which NetMon filters USB event log.

USB Event Tracing For Windows
http://msdn.microsoft.com/en-us/library/windows/hardware/gg463163.aspx

Some device driver may reset the host controller.
USB event trace would catch such an activity.

Tsuneo