So, I'm still trying to create a custom installer capable of
using a kernel modified from the original 7.3 src rpms. The kernel
works just fine, even if we stick it on a 7.1 RedHat box, I'm just
having trouble with anaconda.
My suspicion is that there's something going wrong with the
bootloader (both lilo and grub fail) writing to the boot block.
At the end of the installation, I get, overlayed over the blue
curses screen, something like the following:
pmio_internal.h:235 fdGetFp: Assertion `fd && fd->magic == 0x04463138' failed.
install exited abnormally -- received signal 6
...and so on as it tries to shut down.
This leaves me with a broken system -- there doesn't seem to be
a kernel or updated bootloader. The machine boots to a grub prompt.
(if anyone has any suggestions as to where to go from there, that
would help, too -- I'm not so familiar with grub as with lilo)
I am currently doing a network installation, but this will
eventually go onto a CD and an image set.
When I sent mail asking this question about a week ago, I was
using 8.0 on my build machine. I've since rolled my build machine
back to 7.3 and recompiled all the relevent RPM packages. I've
tossed the rpms into the correct place, and run buildinstall.
One thing that I've discovered is that if I run buildinstall with
the BOOT version of my new kernel, it fails:
Wrote /tmp/makebootdisk.tree.31679 (2056k compressed, 1401k free)
cp: writing `/tmp/makebootdisk.tree.31679/vmlinuz': No space left on device
sed: Couldn't close {standard output}
sed: couldn't write 71 items to {standard output}: No space left on device
However, if I use the original 7.3 BOOT kernel, it seems to
go to completion. Does anyone have any documentation as to how
the magic within anaconda-runtime works, and where I should look
to debug this? I'm a competent python programmer.
--
Adam Haberlach | Give a man a fish, and he'll eat fish
adam@xxxxxxxxxxxxxxx | for a day. Give a Hobbit a ring, and he'll
http://mediariffic.com | eat fish for an age.
|
