> I just upgraded the fuse-utils package, and got bitten by this, although > indirectly. It went like this: > * prerm called "invoke-rc.d fuse-utils stop". The module removal > failed, since a fuse FS was still mounted. The fusectl pseudo-FS got > successfully unmounted. > * postint called "invoke-rc.d fuse-utils start". Re-mounting the fusectl > pseudo-FS while a fuse FS was still mounted triggered a bug in the > kernel. I got a nice stack trace and mount blocked forever. Drama ensued. > > This is a kernel bug so it's not directly related to this package. So > what's my point ? I don't know much about the Debian policy, but I don't > think that the stop argument of the init script should do anything. > Messing with the kernel module and the pseudo-FS after the system boot > introduces more problems than it solves. BTW, this is the approach of > the nfs-common package: its init script does not remove the nfs kernel > module when called with the stop argument.
Yeah, although normally after deinstalling the package you'd want to return to the state prior to installing it, which is no module loaded and control filesystem not mounted. But I agree that this is really unfortunate, given this kernel bug :( Thanks, Miklos -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
