On Thu, 14 Sep 2017, Bruce Korb wrote: > On Thu, Sep 14, 2017 at 7:35 AM, Alan Stern <stern@xxxxxxxxxxxxxxxxxxx> wrote: > > On Wed, 13 Sep 2017, Bruce Korb wrote: > >> WRT "No way to do that without special hardware" perhaps I can > >> do the tracing on OS/X. Remember, that is still working. I could also > >> re-install openSUSE 42.1, but I won't. That would be too many > >> hours of work. > > > > No, you can't do it using only the computer, no matter what operating > > system or software you run. The computer can only tell you what data > > it gets over the wire; it can't tell you what's happening in the > > stretch of wire between the hub and the mouse. For that you need a USB > > bus analyzer. > > Though I cannot sniff the port wires in and out of the hub, it > should be possible to detect what the drivers are saying to the > device and what it is getting back. The question would be, > "How are they different?" since they _have_ to be different. I doubt there is any difference in what the drivers are saying to the mouse. Most likely the problem lies in the xhci-hcd driver. But I could be wrong -- you can compare usbmon traces collected with the mouse plugged into the hub on the USB-3 vs. the mouse plugged into the hub on the USB-2 port. > > The fact that it works under OS/X means that Linux configures either > > the xHCI host controller or the hub incorrectly. We need to know which > > and what it is doing wrong. > > It may be because my MacBook is so old that it only has USB2. > Everything works correctly under Linux if I plug the hub into a > USB2 port. It just defeats the purpose of getting the USB3 vs USB2 hub. I misunderstood -- I thought you were running Linux on the same MacBook as OS/X. > So, if I could trace everything the driver sends down and gets back > both when connected to a USB3 and USB2 ports, would that not > come close to what would be needed to figure out how the driver > for the USB3 port is going south? No, I don't think so. A different approach that is much more likely to find the answer would be to do a bisection between the 4.1 and 4.4 kernels. That should be able to find the particular change which caused things to stop working. 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
