Re: How should we handle bogus bInterval descriptor values?
Alan Stern <[email protected]>
| Newsgroups | gmane.linux.usb.devel |
|---|---|
| Message-ID | <[email protected]> |
On Tue, 8 May 2007, Robert Marquardt wrote: > Alan Stern schrieb: > > In this bug report, lsusb shows an HID device which has bInterval set > to 0 > > in its two Interrupt endpoint descriptors: > > > > http://bugzilla.kernel.org/show_bug.cgi?id=8432 > > > > Of course this confuses the USB drivers, causing a lot of problems. > > > > What should we do about such things? > > The lower bounds for the polling interval are cleanly defined as 1 msec > for full-speed and 10 msecs for low-speed HID. So enforcing the rules > seems a good idea. Healing descriptors as soon as possible. I agree. Interestingly, the USB 2.0 spec lists 10 ms as the lower bound for low-speed interrupt endpoints in section 5.7.4, but Table 9-13 says that the value can go down to 1. > The most general way to do so is a database of known bad devices which > get their descriptor completely replaced by a good one. As this is > probably too much effort in code and computation time only healing > critical parts and common errors should do the trick. I'll write a patch to set out-of-range values to 32 ms, which seems like a nice default. Alan Stern ------------------------------------------------------------------------- This SF.net email is sponsored by DB2 Express Download DB2 Express C - the FREE version of DB2 express and take control of your XML. No limits. Just data. Click to get it now. http://sourceforge.net/powerbar/db2/ _______________________________________________ [email protected] To unsubscribe, use the last form field at: https://lists.sourceforge.net/lists/listinfo/linux-usb-devel