Bruce Dubbs wrote:
> So it looks like for us it should be /run/udev/tmp-rules--*, but we 
> could copy that to /run/udev/rules.d/ for a temporary location or to
>  /etc/udev/rules.d/ for a permanent location.
> 
> I'd like to use /run/udev/rules.d/ because the less we write to /etc,
> the better.  There is discussion of moving mtab and the like to
> other places.  Someday, I'd like to be able to leave / as read only.

Yeah.  The issue is that /run can't ever work for this, since the whole
point of these scripts is to persist the current configuration across
reboots.  :-)

Using /run *could* work if IFNAMSIZ was larger than 16 (but that's an
ABI change, and impossible) and if NIC aliases were supported -- then
the same setup as disks could work.  udev could create by-mac/xxx,
by-path/yyy, etc., aliases for each NIC, and people could just use
those.  Then it wouldn't require persistence either, and the whole set
of scripts could disappear.

But that's impossible.  :-(

>> I think before we drop udev_retry we need to either (a) encode
>> somewhere that the only way that multiple partitions are supported
>> is via an initramfs (which mounts *all* of them before giving
>> control to the rootfs), or (b) hack up an initramfs that does
>> exactly that, ourselves.
>> 
>> Of course, --type=failed will actually break at some point.  At
>> that time, we'll need to either re-trigger everything (which will
>> take a while), or do the initramfs thing.
> 
> I took a quick glance at your 2005 hint on building an initramfs.  I
> wonder how much has changed.

Not a ton, although I wouldn't include the configuration in the kernel
anymore.  (That was just to make the hint require fewer other tools.)  I
have moved to a mkinitramfs package (that I wrote, which is why I use
it; I don't actually trust the others ;-) ), but part of that package is
also there to support dm-crypt, LVM, and some RAID setups.  So it's a
lot more complex.

(I haven't tried to add the code to that one to mount other FSes from
the final system, either.  Actually, that will require a checkfs
equivalent as well, since I don't want to mount a potentially broken FS.
And it was looking like it might actually work.  Nuts.  :-P )

> In any case, it's a lot of work because upstream wants to drop what
> appears to be a small capability.

Yup.

> We may just have to accept the limitation of not having separate /usr
> and /var partitions for LFS.

Or at least point to the instructions for a couple initramfs
implementations that are known to work, when telling this to users.

Attachment: signature.asc
Description: OpenPGP digital signature

-- 
http://linuxfromscratch.org/mailman/listinfo/lfs-dev
FAQ: http://www.linuxfromscratch.org/faq/
Unsubscribe: See the above information page

Reply via email to