On Fri, 10 Jul 2015, Raphael Hertzog wrote: > > Here's another thing you can do: Try capturing a usbmon trace while > > doing the unbind/bind test. Since a controller's usbmon trace file > > doesn't exist until after it is bound, you will have to run your trace > > on the /sys/kernel/debug/usb/usbmon/0u file, which will capture > > information on _all_ USB buses -- so try to minimize other USB activity > > while running the test. > > I just did this. Please find the trace attached (ubsmon-unbind-rebind.txt). > > I also attached a trace of what I get when I powercycle the disk > enclosure (usbmon-powercycle.txt). > > What do you learn in those traces? The powercycle trace looks absolutely normal. It even shows a disconnect event for the disk drive at the start, which is a little odd since apparently you never got a connect event. The unbind-rebind trace shows the system initializing the xHCI host controller. Part of that initialization involves turning on power to all the ports and then querying the port status to see if anything is connected. The disk drive was attached to port 3 of bus 1 (as shown in the second trace). Here are the important parts following the rebind: > ffff88040b8921c0 127270502 S Co:1:001:0 s 23 03 0008 0003 0000 0 > ffff88040b8921c0 127270506 C Co:1:001:0 0 0 This shows power being turned on for port 3 at timestamp 127.270502. > ffff88040b5f4b80 127369604 S Ci:1:001:0 s a3 00 0000 0003 0004 4 < > ffff88040b5f4b80 127369611 C Ci:1:001:0 0 4 = 00010000 This shows the status of port 3 at timestamp 127.369604 -- almost 100 ms later. The only status bit turned on is the Power bit; there is no indication that any device is connected. This suggests a new test you can run. Do the same thing as the unbind-rebind test before, but after the rebind, run "lsusb -v" (as root) while the trace is still being collected. Let's see what the lsusb output and the trace output show. Alan Stern -- To unsubscribe from this list: send the line "unsubscribe linux-usb" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
