On Tue, Nov 08, 2011 at 11:46:10PM +0100, Kay Sievers wrote:
> On Tue, Nov 8, 2011 at 23:39, William Hubbs <w.d.hubbs@xxxxxxxxx> wrote:
> > On Tue, Nov 08, 2011 at 08:35:53PM +0100, Kay Sievers wrote:
> >> On Tue, Nov 8, 2011 at 19:38, William Hubbs <w.d.hubbs@xxxxxxxxx> wrote:
> >> > I know that having /usr on a separate file system with the latest udev
> >> > doesn't work without using an initramfs.
> >> >
> >> > Are there any other file systems that should be pre-mounted by the
> >> > initramfs, such as /var? It looks like /var has to be pre-mounted if you
> >> > have alsa installed, but I want to confirm whether folkson this list
> >> > know this.
> >>
> >> There is no need for that. Systemd can bring up the box without /var,
> >> and sort the services which need that after /var is mounted.
> >>
> >> Alsa has its own systemd service which initializes hardware at that
> >> point, in case the coldplug run did not do it already from udev.
> >
> > In that case, shouldn't we have the alsa-utils folks drop
> > /lib/udev/rules.d/90-alsa-restore.rules from their package?
>
> Udev still takes care for hardware you connect later. Udev does the
> hotplug path, the systemd service does the initial init during bootup.
> Both are needed.
>
> But the ACTION=="remove" rule in that file can surely be killed, not
> sure who expected saving the state of a device that is already removed
> to work. :)
I'm not sure we are talking about the same file, so I will include the
one I have for reference, this is from alsa-utils-1.0.24.2.
ACTION=="add", SUBSYSTEM=="sound", KERNEL=="controlC*", KERNELS=="card*", \
RUN+="/usr/sbin/alsactl restore $attr{number}"
The problem is the run+= portion. "alsactl restore" reads state
information from /var/lib/alsa by default, so it will fail if
/var is not mounted.
William
Attachment:
pgpvJBHGGeg9g.pgp
Description: PGP signature
