Hey guys
Yesterday I completed my install of LFS systemd from the 20150723SVN systemd book. Installation was fairly easy, only a few bumps in the road which I managed to get around.
But, when I reboot into it, kernel modules are not loading. So the first thing I realized is that I had forgotten to "make install" the kmod package (whoops), so I went back and did that, and now I can modprobe the modules I need, such as nouveau. But they won't load on boot.

So far, I've discerned two things. As an Arch Linux user, I'm familiar with systemd. It usually has a systemd-load-modules.service and executable, which runs as a job at startup. Neither the startup job nor the executable are present on my LFS system. I don't know if that is specific to distributions that use systemd or what...

The second thing is that my Ethernet is not detected, which, if I'm correct, should be detected by udev which should load the correct kernel modules. So I'm thinking it could either be the missing systemd-load-modules, or a problem with udev, or both.

Any help would be appreciated! This is the last problem in getting this system to fully work!

Thank you!
drsn0w

I recompiled systemd and it now loads what I have configured for load-modules.d... still having issues with ethernet.

I had issues with it when I tried out the systemd book over a year ago, and gave up.

Networkd should auto-config dhcp over ipv4 but half the time systemctl gets retarded and deletes the networkd service startup file from the loading directory and it screws up everything entirely killing networking completely. I tried using dhclient without any luck. The kernel modules should be loaded by udevd automatically rather than selected.

The book is somewhat experimental in design still unfortunately, and is known to have issues that vary user to user.

We used to have a small script collection to revive the network if needed, but I think those were expunged a while ago.

You replied that you had forgotten to run make install when building kmod. kmod had to be present at systemd build time for modules-load service to be built and installed and for udev to have capability to load modules. I also see you rebuilt systemd and now it should be fine.

Are you sure the module is there? If so, does it require any firmware? If it does, is the firmware installed? Maybe it uses some weird interface naming as systemd-udev does for some time now. Check "ip l" output to see if its there under a weird name.

I ran into this tonight while going through the mailing list too look for an error solution. It looks like systemd development on LFS may end up sidelined. The curious bit is what he mentions about systemd upstream.

http://lists.linuxfromscratch.org/pi...ly/070283.html

systemd is kind of an "if it works..." package compared to the standard sysvinit package in LFS, but I think that's due to the design framework of LFS.

I've left the development due to lack of time and motivation, nothing else. What I did say is that recent upstream changes added a lot more work to my already tight time schedule, not that it was a reason. The branch is still maintained by other developer, so the book isn't going anywhere.

Plus the systemd books test a lot of extra packages not normally found in the books as well, so using them can prove insightful, even for the standard sysvinit book.

I have hard time parsing this. Care to explain a bit more?

I just wanted to say thank you for your contributions, I have received great benefit from the work you and the rest of the LFS team have put into their books. I just did the systemd edition myself a few days ago and found it as easy to follow along and do as the standard edition.

It has extra packages such as Wayland, for example.

Well, you said "testing packages not found in the books". All of the wayland stack is there in the book, because GNOME requires it and soon KDE Plasma 5. I presume you meant non-systemd books, and they'll follow soon if they intend to use KDE Plasma 5.

Yes, often I speak in cryptiglish for some reason. LOL