Ken Moffat wrote:
This only affects clfs at the moment (lfs is still using an older
version of udev). A lot of people are reporting 'udevsend: main: error
sending message: Connection refused' messages, I've now tracked it down
and discovered that our (clfs) hotplug is not working (specifically,
when I connected my usb printer, the node /dev/usb/lp0 was not getting
created).
I eventually found a post from Kay Sievers on the hotplug list in
http://sourceforge.net/mailarchive/message.php?msg_id=13974821
saying that from udev076 the initscript has to start udevd --daemon
(this apparently used to happen through udevsend).
I've knocked up a quick patch for the installed /etc/rc.d/init.d/udev
which fixes this for me. In the CLFS book, either we can ask for new
bootscripts, at least until such time as the udev changes get into the
book, or we can just do a proper patch for the lfs-bootscripts tarball
as a short-term measure.
I think Jim's preference is for a clean new bootscripts tarball, so
I'll start by asking for that, please.
Of course, I don't know what Matt's udev plans are, and whether or not
lfs itself will also need udevd --daemon.
Yes, we will need udevd --daemon. I've committed the changes I had
planned for the big udev upgrade to the "udev_update" branch of the book
this evening. There's a couple of issues left to fix up, but it should
be pretty usable already. Here's what needs to happen before it gets
merged to trunk:
1) Fix the FIXME comment regarding not being able to run /sbin/udevstart
to populate /dev in immediately following the udev installation in
chapter 6.
2) Add the helper binaries to the list of installed files, with
explanations of what they do
3) Get plenty of testing done to see what I've broken, particularly with
regard to module handling. I have a feeling more rules are needed in
order to cope with other buses/subsystems - what I committed was what I
could easily test with my USB MP3 player.
4) Merge whatever new rules we come up with into a new version of the
rules file.
5) Release a stable version of lfs-bootscripts with the udev patch in
it, and all remains of hotplug removed.
I'd like this to happen within 4 weeks, so if anyone wants to help test,
now's the perfect time. I can't get a render done this evening, so
anyone who wants to read the HTML will have to render it themselves,
unless someone would be so kind as to make it available on the web
somewhere.
Thanks,
Matt.
--
http://linuxfromscratch.org/mailman/listinfo/lfs-dev
FAQ: http://www.linuxfromscratch.org/faq/
Unsubscribe: See the above information page