[Pkg-sysvinit-devel] Bug#620191: initscripts: [patch] Please support top-level /run

[Roger Leigh]

On Fri, Apr 01, 2011 at 05:42:53PM +0100, Roger Leigh wrote:
> On Fri, Apr 01, 2011 at 04:19:03PM +0100, Roger Leigh wrote:
> > On Fri, Apr 01, 2011 at 03:11:18PM +0100, Roger Leigh wrote:
> > > On Fri, Apr 01, 2011 at 02:18:11PM +0100, Roger Leigh wrote:
> > > > On Thu, Mar 31, 2011 at 06:55:16PM +0100, Roger Leigh wrote:
> > > > > On Wed, Mar 30, 2011 at 11:00:57PM +0100, Roger Leigh wrote:
> > > > > > The attached patch contains a detailed changelog of the changes, so
> > > > > > I won't repeat it here verbatim.  Outstanding issues are marked
> > > > > > TODO, and may require further discussion to resolve them.
> > > > > 
> > > > > Updated patch attached.  Handles upgrades and reboots; other than
> > > > > the one outstanding TODO item (transition of /lib/init/rw from
> > > > > directory to symlink, probably something we can leave until after
> > > > > squeeze).
> > > > 
> > > > Updated patch attached.  This one has been tested in a VM, and will
> > > > allow a machine to boot and transition to /run safely.
> > > > 
> > > > Minor outstanding issue:
> > > > /etc/init.d/mountall.sh fails.  This is due to (apparently) / and /proc
> > > > already being mounted, leading to a 32 exit status.  Maybe an mtab
> > > > issue?
> > > > 
> > > > In order to work properly with bind mounts on freebsd kernels, I've
> > > > synched domount and domtab logic so that they behave identically.
> > > > However, I'm seeing some odd double bind mounts in /proc/mounts
> > > > and an incomplete /etc/mtab, so there's possibly something not quite
> > > > right here.
> > > > 
> > > > Apart from this one issue, everything else is working fully.
> > > 
> > > Updated patch attached.  This fixes the above outstanding issues.
> > > It should now be safe for general testing and use.
> > > 
> > > Note: needs versioned base-files dependency adding.
> > 
> > Updated patch attached; adds kfreebsd support to initscripts postinst.
> 
> Updated postinst following feedback on #debian-devel:
> - use uname rather than dpkg --print-architecture
> - simplify conditional logic

Updated to
- not enable RAMLOCK by default (though doing so removes a DoS we
  are currently vulnerable to even with traditional /var)
- only use new paths in sendsigs
Thanks to mbiebl for this review.


-- 
  .''`.  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.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: initscripts-run-transition7.patch
Type: text/x-diff
Size: 33391 bytes
Desc: not available
URL: <http://lists.alioth.debian.org/pipermail/pkg-sysvinit-devel/attachments/20110401/c6bd73b6/attachment-0001.patch>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 198 bytes
Desc: Digital signature
URL: <http://lists.alioth.debian.org/pipermail/pkg-sysvinit-devel/attachments/20110401/c6bd73b6/attachment-0001.pgp>