[Martin-Éric Racine] > gpm fails to install inside an LTSP chroot, because it tries to access static > devices during invoke-rc.d.
Not quite... > Setting up gpm (1.19.6-25) ... > Stopping mouse interface server: gpm failed! > Starting mouse interface server: gpm/etc/init.d/gpm: line 65: > /dev/input/mice: Tiedostoa tai hakemistoa ei ole > /etc/init.d/gpm: line 65: /dev/input/mice: Tiedostoa tai hakemistoa ei ole > /etc/init.d/gpm: line 65: /dev/input/mice: Tiedostoa tai hakemistoa ei ole Those errors (someone translated this to me as "no such file or directory") are actually harmless warnings. You're looking at a sort of poor man's hotplug handler for the 'mousedev' module. It gives up after 3 seconds, figuring that if mousedev hasn't initialised itself in 3 seconds, it's not going to. > /sbin/start-stop-daemon.real: nothing in /proc - not mounted? (Success) > failed! This is the real problem. start-stop-daemon is failing, because /proc is not mounted. The question is, why is start-stop-daemon even trying to run? Doesn't debootstrap disable start-stop-daemon? Certainly it doesn't make much sense to run gpm inside an installation chroot. I suggest making sure start-stop-daemon really is disabled in installation chroots. That seems like a more general solution than for gpm to try and detect the situation on its own. Peter
signature.asc
Description: Digital signature