I have had a considerable hassle getting a printer that is 3 hubs away from my machine to be reliably recognized at boot time, often having to unplug and replug the data cable to the parent hub of that branch of my usb tree, followed by a restart of the cups system when it is finally recognized. Its good then till the next reboot, or power failure since the hub on the far end of an extension hub cable is not ups protected.
Posing the question on linux-usb's mailing list, I was asked to set CONFIG_USB_DEBUG, rebuild and re-install the kernel, which I did by retrieving the config for 184.108.40.206-pclos.pae.bfs from the /boot partition, doing a make xconfig in the installed src tree after renaming the src tree to that exact version because my build/install script cd's in and out of it, which in turn requires that the Makefile, my script $VER variable and the src directory all match.
Making sure all the links in /boot were set correctly, I rebooted. But on the reboot, dkms-nvidia-current when executed, claims the nvida driver is already installed. But my script moves the old /lib/modules/$ver out of the way so I can recover (if everything else is ok that is) by renaming it, the vmlinuz and inintrd back to the original names.
But by then the damage is done as X then reverts to trying to run the newvoo(sp?) driver, which because other libs are nvidia's, fails, leaving me with a single high res text screen.
I was able to extract the logs of the miss-behaving usb stuff and post them, and am now rebooted to the previous 220.127.116.11-pclos2.pae.bfs kernel, fixed /etc/X11/xorg.conf, and X is running but poorly, the cursor disappears in kmails composer for 5 to 10 seconds every time I hit the return key.
I have removed 'completely' the dkms-nvidia-current, and the kernel-18.104.22.168-pclos3.pae.bfs packages, then re-installed them both, but the reboot gets me a messsage 'already installed on this kernel' from dkms-nvida-current, and X then plays hell with my xorg.conf trying to make it run newvoo or whatever the heck its called.
What do I need to do to revert to the latest kernel package AND use the nvidia-current driver? I checked the dkms man page and there does not seem to be a 'force' option.
Thanks, & Cheers, Gene