A comment on the current status:
I've received reply from Alan Stern (Linux USB developer), essentially asking us to continue the earlier experiments (see my post above). As time permits I'll play around with the suggested workaround (additional udelay), and see whether it can be tweaked (and still work!) so that it may fit into the upstream code.
Once I have a easy patch myself I 'll post this here. we should then get back to the USB developers (and the UVC guys) with our results so that they can take appropriate action.
(BTW: I'm not affiliated with Canonical in any way.)
A comment on the current status:
I've received reply from Alan Stern (Linux USB developer), essentially asking us to continue the earlier experiments (see my post above). As time permits I'll play around with the suggested workaround (additional udelay), and see whether it can be tweaked (and still work!) so that it may fit into the upstream code.
If you can compile a kernel, *you* can help by testing the workarounds in marc.info/ ?l=linux- kernel& m=1256220347135 18&w=2 marc.info/ ?l=linux- usb&m=125628206 801242& w=2
http://
and
http://
(the standard Ubuntu kernel source should be a sufficient base for patching)
and see how far we can get the udelay down to (target is <= 50 microsecs) with keeping the stack still alive.
Once I have a easy patch myself I 'll post this here. we should then get back to the USB developers (and the UVC guys) with our results so that they can take appropriate action.
(BTW: I'm not affiliated with Canonical in any way.)