Bug#317096: A possible solution ...
Hi,
I have also found this problem ... there is firstly a description of the
reason why it happens, which shouldn't be new to anyone ... and then
follows a discussion of the solution ...
It appears that it (this bug) affects the changeover of kernel stream.
For example, this happens when installing to hard disk from say KNOPPIX
or even the netinst iso .... you install and all is good ... then you
want to change from a regular kernel-image-2.6-686 P4 kernel to the
kernel-image-2.6-em64t-p4-smp ... which can happen sometimes !
Especially if you don't receive a distribution of debian which uses this
as the kernel, yet you want to speed at least the device drivers ...
even if the software is still 32bit !
Anyways ... back to the point ... the solution is to fix the
/boot/initrd.img-*
file relating to the kernel image and the grub or lilo (bootloader)
initrd item.
This firstly K-Os init (upon reboot), which is serious, although your
boot-loader may still be able to access other previously installed
kernels ... unless you installed and removed at the same time ... VERY
SERIOUS MISTAKE ! In this case, your system is completely unusable and
it is thus critical.
OK - not really sure where to put the bug fix, but I would imagine that
it would be handy if the kernel post-inst checked for this.
If it already does ... didn't check ...
solution :
regenerate ramdisk :
mkinitrd -o /boot/initrd.img-VERSIONNAMEHERE VERSIONNAMEHERE
Also you must inform your boot loader if there is an initrd.img change
of name ...
Matt
--
Limited time infomercial link ...
http://60.225.23.48/debian.2004.turnTheTable.Take1.ogg
Reply to: