On Wed, 18 Mar 2009, Melchior FRANZ wrote: > Err ... that wasn't really meant as an ungrateful, defiant "bah, 2.6.29 > will be crap!" kind of message, although it sounds quite like it. I really > thought that these would be my options for now. I consider your efforts > to make this broken device work already far more than I could expect, > and thought that this joystick wouldn't really deserve more love. Well, the kernel developers really dislike regressions. We go out of our way to fix them, whenever possible. Besides, some other devices might turn up in the future with the same bug. > BTW: This js has worked for years, with dozens of kernels, so I didn't > assume that mine is in some way special. But I can't say the same of my > mobo's USB interface. This has made some problems about a year ago, and > one of the two inlets seems to be damaged. The other one has always worked > fine (with a 4-way hub), though. Could this cause the effect that we are > seeing here? In that case, committing the quirk to the kernel wouldn't > be justified. No, the cause is something else -- a change in the order of events. The old kernels would retrieve the joystick's configuration string after binding usbhid, but this ordering caused problems with udev (since udev might need to see that string or other attributes before the binding occurs). So now the new kernel fetches the string first. Unforunately a bug in the joystick causes it to stop responding to Control requests when asked for its configuration string. Since usbhid doesn't rely on Control requests during normal operation, this used to be okay. But it does rely on at least one Control request during binding, so now it fails. The patch adds a quirks flag to prevent the kernel from reading a device's configuration or interface strings. Apparently these things are used so infrequently that they don't get tested very well and some devices (like yours!) can't support them. > > There are other approaches. You can try this patch. > > Woohoo! This works. Thanks, a lot! Okay, I'll submit it. 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
