Jul 29th & 30th, 2013

Horrible mess. After numerous attempts with chroot's, mkinitramfs, perceus twinking, etc, turns out that the broadcom ethernet cards need a firmware file for the kernels I was testing. To make matters worse, the loss of memory mapping persists even with 2.6.32-9. In summary, after ~10 hours we'll have to go back and start with a more recent kernel.

Can't continue without documenting things (what follows is using fc15's 2.6.38.6-26 kernel rpm files) :

• perceus vnfs clone … → create a copy of the original caos-nsa capsule
• perceus mount … → to mount the new copy
• copy files from rpm to : /mnt/…/boot/ /mnt/…/lib/modules/ /mnt/…/lib/firmware/
• chroot /mnt/… depmod 2.6.38….
• edit modprobe.dep and give full paths
• edit /mnt/…/etc/modprobe.conf and add 'alias eth0 bnx2' 'alias eth1 bnx2' =⇒ later removed
• edit /mnt/…/etc/shadow and add a password so that we can login from a terminal
• perceus umount …
• edit /var/lib/perceus/vnfs/…/config and change the kernel name
• test it

⇒ Memory issues fixed with 2.6.38.6. But still can't get a functional enternet (can't load firmware for bnx2 driver although the file is there) …

⇒ Tried changing the placement of the firmware files (/lib/firmware /lib/firmware/(uname -r) /usr/lib/firmware). Still no luck.

⇒ It appears that firmware loading resides in user space and needs additional (user space) programs to function. Which starts killing the idea of keeping the VNFS capsule intact and change only the kernel+modules. As predicted, hate every minute of it.

⇒ Reached desperation : The stage-one perceus boot works fine with the bloody cards because it a non-firmware depended driver (1.8.1). So, here comes the horror : I modified the perceus stage-one scripts in such a way as to reload the driver (with a modprobe) just before calling kexec. The hope is that not only the old driver will happily hang-on to the device for life but also that the old driver won't break the system calls (and that my days will return back to science instead of fighting with the kernel). Will test it tomorrow …

→ Failed again. Tried fixing a missing firmware.sh file, failed as well.