On 15 Oct 2006, at 16:26, Alan Stern wrote: > On Sat, 14 Oct 2006, Michael Thomson wrote: > >> I have usbmon logs from both 2.6.16 and 2.6.17, and dmesg output with >> verbose logging that shows plenty of: > > The log shows that about a half of the messages going between your > computer and the keyboard never get through. No wonder it's unusable!
Argh. I suppose there's no way to tell if the message got lost on the way to the hub or the response never made it back. Pretty academic really. >> There were no more recent kernels directly from Fedora, and I >> couldn't see any relevant check-ins on the linus-usb-devel list, so >> decided not to try a vanilla kernel.org build. Should I have? > > It probably wouldn't make any difference. At least I got *that* guess right! :-) >> Roll my own patches: >> -------------------- >> >> I hoped that the fixes in hid-core.c for other ATEN devices might >> work, so I patched a custom 2.6.17-1.2187 kernel as follows: > > No, that quirk is not related to your problem. > > It could be a matter of bad cabling, but it's much more likely that > the > internal hub in the ATEN device is distorting or attenuating the > signal > beyond your computer's ability to recognize. Or else maybe it's > interrupting the signal altogether. There's no way to tell exactly > what's > happening without using a USB bus analyzer. > > The older kernels just ignored these sorts of errors, whereas the > new ones > take positive action and try resetting the device when no reply is > received. That's why the KVM worked better before. > > If you want, it's possible to edit the source code for the USB HID > driver > to make it revert to the earlier behavior of ignoring these errors. > > Alan Stern Thanks very much for this analysis, Alan. I'm delighted to get such knowledgeable advice, swiftly and generously provided. This is what 'open' is all about. I don't believe I can improve the external cabling from the CPU to the KVM as this uses ATEN's custom cables. I will take a look at the internal hookups as the PC is a box I built myself. I've seen enough loose connection problems over the years that one more wouldn't be a surprise. I'll also try connecting the KVM to another port on the PC in case the problem is tied to one physical port. Can I assume that just counting the frequency of the status -110 messages will tell me if I'm making any significant difference? Is there a dummies guide to understanding the debug messages (or usbmon output)? Would the USB-guide (http://www.linux-usb.org/USB- guide/book1.html) and the usbmon document be the best place to start? Not having a bus analyser handy, if the above attempts fail it looks like I'll have to either give up in disgust or see if I can return the KVM under warranty in hope of a better signal to noise ratio. I don't think it's worth trying to un-patch my kernels regularly. My 'final solution' is likely to be an Intel iMac triple-booting (or virtualising) Mac OS X, Linux and Windows, and avoid the whole KVM issue, but that needs a tad more justification! Cheers Mike Thomson ------------------------------------------------------------------------- Using Tomcat but need to do more? Need to support web services, security? Get stuff done quickly with pre-integrated technology to make your job easier Download IBM WebSphere Application Server v.1.0.1 based on Apache Geronimo http://sel.as-us.falkag.net/sel?cmd=lnk&kid=120709&bid=263057&dat=121642 _______________________________________________ Linux-usb-users@lists.sourceforge.net To unsubscribe, use the last form field at: https://lists.sourceforge.net/lists/listinfo/linux-usb-users
