[RFC] Bootscript changes

Alexander E. Patrakov patrakov at ums.usu.ru
Wed Apr 11 05:37:54 PDT 2007

I wrote:
> Dan Nicholson wrote:
>> http://wiki.linuxfromscratch.org/lfs/changeset?format=diff&new=7962&old=7931&new_path=trunk%2Fbootscripts%2Flfs&old_path=trunk%2Fbootscripts%2Flfs
>> Can you check if these are the same issue?
> I will do this tomorrow, thanks for the link.

The patch above doesn't help with today's SVN. Full testcase:

1) Install VMware Server, make two virtual machines (Client and Server) that 
boot from the LiveCD and have their network cards on the same private 
network (e.g. /dev/vmnet2)
2) Boot the Server, remove the "login" line from 
/etc/ppp/pppoe-server-options, uncomment the example "jdoe" user from 
3) On the Server,

modprobe pppoe
pppoe-server -L -R

4) Boot the Client, run net-setup, configure pppoe for eth0 with the "jdoe" 
user and "guessit" password. Note that the LiveCD doesn't use the stock BLFS 
script for pppoe, see 

5) On the Client, run this stupid command (it spawns two pppds that try to 
talk to each other, just to make sure that there is a pppd not belonging to 
the pppoe connection):

pppd noauth pty 'pppd noauth notty' nodetach debug

6) On the Client, /etc/sysconfig/network-devices/ifdown eth0

This should (and does) kill only the pppd that is responsible for the pppoe 
connection, and should (but doesn't without the patch) print a green [OK] in 
the top line of the output. My earlier comment about hibernation is wrong.

Alexander E. Patrakov

More information about the lfs-dev mailing list