Re: :Loading LinuxEBDA is big" lilo error - PERMANENT fix?
On Wed, 17 Mar 2010 12:28:36 -0400 (EDT), Brett Charbeneau wrote:
> Google searches indicate that booting from a rescue CD and rerunning 
> lilo takes care of the problem, which is does, until the next kernel upgrade. 
I use lilo too.
If rerunning lilo fixes the problem, then the problem is that lilo did
not get run during the kernel upgrade.  I noticed that too on my
system.  When the kernel on my Squeeze system was upgraded from
linux-image-2.6.32-trunk-686 to linux-image-2.6.32-3-686, the kernel
image file was installed to /boot, the initial RAM file system was created,
and the symlinks were updated.  But lilo was not run.  I had to manually
run lilo.
The maintainer scripts for kernel image packages used to always do this,
but now, well, not necessarily.  There is a configuration file for kernel
image packages called /etc/kernel-img.conf.  There are some flags in there
that need to be set.
   do_initramfs = yes
   do_symlinks = yes
   do_bootloader = yes
However, the maintainer scripts do not always honor them anymore.
For stock kernels, do_initramfs is ignored.  It *always* creates an
initial RAM filesystem, because stock kernels require one.  do_symlinks
seems to still be honored.  Mine got updated.  do_bootloader doesn't
seem to be honored anymore.  do_bootloader seems to work only when
an initial RAM filesystem is *updated*, not when it is *created*.  Also,
if grub (version 1 or version 2) is installed (but not in use),
update-initramfs may get confused and think that it doesn't need to
run lilo.  After all, if grub is installed, then you're obviously
using it, right?  :-|
If in doubt, always manually run lilo after a kernel upgrade.
If you want to make *sure* that it always gets run, you can create a
hook script.  An example of this can be found in "Step 10:
Customize the Kernel Installation Process" on the following
web page: http://www.wowway.com/~zlinuxman/Kernel.htm
-- 
  .''`.     Stephen Powell    <zlinuxman@wowway.com>
 : :'  :
 `. `'`
   `-
Reply to: