jef e wrote: > > We have a batch of HP DL320 G4's hanging around that have some funky USB > hardware that give the kernel fits. Basically, it appears to be HP's > virtual keyboard that keeps reregistering itself over and over again. > > However, we are still faced with the USB issue - apparently, the HP ILO > virtual Keyboard keeps re-registering itself. Yup, the iLO interface on the server causing the UHCI module to freak out. It took me nearly a full day to install Debian on 2 of these crappy servers. :-( I keep meaning to submit a bug report against the kernel, but haven't got round to it. You get first dibs on it now, if you want to report it. The way I managed to install Debian (from an Etch installer, because Sarge could not see the optional SATA controller card) was to tell the installer not to detect USB; and then, once the language selection screen was up, switch to a console and do "rmmod uhci_hcd" and keep doing this now and again during the rest of the install. Once the install is finished, you nay need to boot with the safe mode option (otherwise it may just hang during booting. Once in, I tried: # echo 'alias uhci-hcd off' > /etc/modutils/local.aliases # update-modules # reboot Despite this, I still had problems. In the end I had to compile a custom kernel without the UHCI module. The downside is that without the UHCI module iLO keyboard does not work. :-( > These messages don't happen when we install Redhat on these systems, so > it's something that seems to have been patched already and probably > backported by them. From what I read on the single informative thread I could find on the subject, Redhat patched their kernels against this bug, by basically restricting the number of times the virtual keyboard could re-register. Hope this helps, -- George Borisov DXSolutions Ltd
Attachment:
signature.asc
Description: OpenPGP digital signature