After I removed the 2 "#pragma GCC visibility" lines from xmlconf.c everything built fine. Mind you I haven't tested to see if it works yet - I should get the usb device to test next week. Cheers John On Thu, 30 Mar 2006 21:18:11 +0100 Dr J A Gow <J.A.Gow@xxxxxxxxxxxxxxxxx> wrote: > I'll modify this in the source for the next release - it's true we don't > really > ise the .mode element in the module. How did you work around the gcc bug > _and_ > get everything to build properly? If you expose the xparse symbols in the > shared > library, builds further down the tree can fail. > > John. > > John Dally wrote: > > OK it seems that the mode field from the kernel header > > include/linux/usb.h has been removed in later 2.6 kernels:- > > http://www.ussg.iu.edu/hypermail/linux/kernel/0506.1/0822.html > > > > And from the docs.. > > http://mirror.linux.org.au/linux-mandocs/2.6.10/usb_class_driver.html > > it looks like its just needed for devfs. > > > > So I've removed the mode from the > > > > USBCLASSDRIVER AVC2210Class = { > > > > .name = "usb/avc2x10k%d", > > .fops = &Strategy, > > /* .mode = S_IFCHR|S_IRUSR|S_IWUSR|S_IRGRP|S_IWGRP, JDJD */ > > .minor_base = 0, > > }; > > > > in src/avc2210k/strategy.c > > Now everything compiles and the module loads. > > > > All I need now is the capture device to test it (from a mate who has one!) > > > > BTW have you considered creating a project page somewhere, so > > there's a static place for the releases and docs? > > > > Cheers > > > > John > > > > > > > >