On Sun, Apr 03, 2011 at 01:10:29PM +0200, Goswin von Brederlow wrote: > Roger Leigh <firstname.lastname@example.org> writes: > > > On Wed, Mar 30, 2011 at 11:20:45PM +0100, Roger Leigh wrote: > > 1) /etc/init.d/mountall.sh is broken for some reason. The "mount -a" > > invocation fails. Not because it fails to mount, but it returns a > > 32 exit status because / and /proc are already mounted. > > > > Possibly a result of the mtab.sh domtab() changes; but it should be > > behaving identically to the old version, so possibly unrelated. > > Possibly already broken and I've just exposed a bug? > > If it is what I think it is then I have already filed this bug. > > The problem arises because the initramfs mount for proc differs from the > one listed in /etc/fstab. The mount then decides that something else is > mounted on /proc and gives an error. The problem is that initramfs uses > "none" as device while mountkernfs / fstab use "proc". > > > 2) Despite only mounting /run once, I'm seeing /run and /run/lock mounted > > *again* looking at /proc/mounts. Unsure why this happens, or if it's > > just artifactual. /run, /run/lock and /run/init are all correctly > > visible, but this is a tad strange. > > Any bind mounts you forgot about? Both of these issues were due to /etc/mtab being inconsistent with reality. I've now corrected the code to make domtab() behave identically to domount() to ensure that they are always working in the same manner. Once /run is established, I wonder if moving /etc/mtab to /run/mtab would be feasible, since it would mean this complex and hairy setup could be dropped entirely. The patches for base-files and initscripts have now been tested on linux and freebsd and are TTBOMK good to go. GNU/Hurd might also need some special casing adding--if any Hurd developers would care to check over the changes that would be great. Regards, Roger -- .''`. Roger Leigh : :' : Debian GNU/Linux http://people.debian.org/~rleigh/ `. `' Printing on GNU/Linux? http://gutenprint.sourceforge.net/ `- GPG Public Key: 0x25BFB848 Please GPG sign your mail.
Description: Digital signature