Description of problem:

Upon boot, parse-kickstart parses the kickstart and searches for the first "network" stanza only. It then creates /tmp/ifcfg/ifcfg-* files. This works fine.
Additionally it generates an excerpt of cmdline, e.g. "bootdev=bond0 ip=...:bond0:...

Example:
-------- 8< ---------------- 8< ---------------- 8< ---------------- 8< --------
network  --hostname=vm-test8 --device=bond0 --mtu=1500 --bondslaves=enp1s0,enp2s0 --bondopts=mode=active-backup --bootproto=static --noipv6 --activate --ip=192.168.122.47 --netmask=255.255.255.0 --gateway=192.168.122.1
-------- 8< ---------------- 8< ---------------- 8< ---------------- 8< --------

Generated files by parse-kickstart:
 - /tmp/ifcfg/ifcfg-bond0
 - /tmp/ifcfg/ifcfg-bond0_slave_1
 - /tmp/ifcfg/ifcfg-bond0_slave_2

Generated stdout (then pushed to /etc/cmdline.d/80-kickstart.conf):
-------- 8< ---------------- 8< ---------------- 8< ---------------- 8< --------
ip=192.168.122.47::192.168.122.1:255.255.255.0:vm-test8:bond0:none:1500 bootdev=bond0
-------- 8< ---------------- 8< ---------------- 8< ---------------- 8< --------

This is then used by nm-initrd-generator to generate NM profiles.

Apparently nm-initrd-generator doesn't rely on /tmp/ifcfg/ifcfg-* files, but only the command line:

/usr/lib/nm-lib.sh:
-------- 8< ---------------- 8< ---------------- 8< ---------------- 8< --------
[...]
    /usr/libexec/nm-initrd-generator -- $(getcmdline)
[...]
-------- 8< ---------------- 8< ---------------- 8< ---------------- 8< --------

From above, we can see that the generated stdout is broken and lacks the full definition of the bond, which causes an infinite hang to happen (well not exactly infinite, but almost):
-------- 8< ---------------- 8< ---------------- 8< ---------------- 8< --------
[    7.369943] IPv6: ADDRCONF(NETDEV_UP): enp1s0: link is not ready
[    7.370838] 8021q: adding VLAN 0 to HW filter on device enp1s0
[    7.373143] IPv6: ADDRCONF(NETDEV_UP): enp2s0: link is not ready
[    7.373987] 8021q: adding VLAN 0 to HW filter on device enp2s0
---> nothing then
-------- 8< ---------------- 8< ---------------- 8< ---------------- 8< --------

There are 2 solutions for this:
1. Generate the appropriate cmdline
or
2. Generate files in /etc/sysconfig/network-scripts (this works, but I fear this will collide with initial boot parameters, if any)


Version-Release number of selected component (if applicable):

dracut-049-95.git20200804.el8


How reproducible:

Always

Steps to Reproduce: Boot with a kickstart hosted on a minimal DVD (don't setup the network initiallyà

1. Create a minimal DVD stub

  # mkdir rhel83.minimal && cd rhel83.minimal
  # cp -r /var/www/html/rhel83/{EFI,images,isolinux} .

2. Create a minimal kickstart

  # cat > ks.cfg << EOF
  url --url http://192.168.122.1/rhel83
  network  --hostname=vm-test8 --device=bond0 --mtu=1500 --bondslaves=enp1s0,enp2s0 --bondopts=mode=active-backup --bootproto=static --noipv6 --activate --ip=192.168.122.47 --netmask=255.255.255.0 --gateway=192.168.122.1
  EOF

3. Build the minimal DVD

  # mkisofs -o /tmp/rhel83.minimal.iso -b isolinux/isolinux.bin -J -R -l -c isolinux/boot.cat -no-emul-boot -boot-load-size 4 -boot-info-table -eltorito-alt-boot -e images/efiboot.img -no-emul-boot -graft-points -V "RHEL-8-3-0-BaseOS-x86_64" .

4. Boot a VM with 2 NICs (enp1s0, enp2s0) and append "inst.ks=cdrom:/ks.cfg" to kernel command line


Actual results:

Hang before switching root, no way to get a dracut shell.


Expected results:

Network configuration and installation continues.