Bug 21694

Summary:	update 6.2 -> 7.0. No inclusion of needed disk driver in new kernel.
Product:	[Retired] Red Hat Linux	Reporter:	Need Real Name <bernard>
Component:	anaconda	Assignee:	Matt Wilson <msw>
Status:	CLOSED CURRENTRELEASE	QA Contact:	Brock Organ <borgan>
Severity:	medium	Docs Contact:
Priority:	medium
Version:	7.0
Target Milestone:	---
Target Release:	---
Hardware:	i386
OS:	Linux
Whiteboard:
Fixed In Version:		Doc Type:	Bug Fix
Doc Text:		Story Points:	---
Clone Of:		Environment:
Last Closed:	2001-04-11 18:07:03 UTC	Type:	---
Regression:	---	Mount Type:	---
Documentation:	---	CRM:
Verified Versions:		Category:	---
oVirt Team:	---	RHEL 7.3 requirements from Atomic Host:
Cloudforms Team:	---	Target Upstream Version:
Embargoed:

Description Need Real Name 2000-12-04 18:11:48 UTC

I run 6.2 with the EATA disk controler driver. I perform an update from
CD to RH7.0. The update request to use the diskdriver.img to find eata. 
Everything is fine.

I reboot on the new kernel, it does not work: the new kernel did not
make any provision to have eata included at boot time to drive the
disk controler.

Comment 1 Michael Fulbright 2000-12-04 19:02:51 UTC

Passed to QA to reproduce.

Comment 2 Need Real Name 2000-12-08 09:05:07 UTC

I upgraded a bunch of computers and reached the following conclusion:

- If lilo.conf before the upgrade, has a label 'linux' which is the
  default boot image, the upgraded kernel will include the correct disk
  driver and everything will be fine.
- however, most of my computers, have 2 kernels: a 'linux', which is
  the base kernel of the install, then a 'bzImage' which is the
  default. In such a case, upgrading such a system will make 'linux'
  the default boot kernel, but this new kernel will not include the
  disk driver. Hopefully I can still boot on my old 'bzImage', recompile
  the source to make a new 'bzImage' incorporating the drivers then reboot.
  (after the 'make mrproper'-smp bug :-))

Comment 3 Need Real Name 2000-12-08 12:55:15 UTC

Well things are not so clear: I upgraded two more computers, each one having
just a single 'linux' entry in lilo.conf. The upgrade went allright, I made
rescue disk, but at reboot time, the kernels did not include the required
disk driver! One was eata (again) but the other was AIC7xxx.

To be able to reboot, I wanted to rebuilt the rescue disk again since
the rescue did not contain also the correct disk drivers! (I wanted to
build the rescue disk again since I made some mistake when copying a
kernel made from another computer onto them).

So, to rebuild the rescue disk, I restart completely the update process.
This way the system would allow me to build new rescue disk and this time
I knew what to do to correctly overwrite the kernel on the disk...

First thing, anaconda wants to re-install packages that it just installed
a few minutes ago!

On one computer (fs), the first running of the upgrade processed about
1140 megs(!). The second running, just 118 megs. I made the rescue disk,
recovered the kernel on it with a kernel made from another computer and
was okay to get back control on the computer.

On the second computer (lexis), the first and second running of the upgrade
processed about 1140 megs also. But what is very strange is that, the 2th time,
the installed kernel had the correct disk drivers! (aic7xxx).

So there is at least 2 problemes:

- why a 'just-upgraded' system, if upgraded again, requires a variable
  amount of data to 're-upgrade' a system instead of 0 bytes ?
- why, sometimes, an upgrade will not correctly add required scsi disk
  drivers and sometimes it will ??

BTW: it would be really nice if a booting linux kernel, instead of
panicing when realizing that it has no correct disk driver, ask for
a driver disk...

(all this with CD install since bootnet does not want of eata drivers)

Comment 4 Michael Fulbright 2001-04-11 16:04:56 UTC

Matt do you have any ideas?