Red Hat Bugzilla – Bug 59518
At beginning of Linux install, crashed and disabled laptop from further boots
Last modified: 2007-04-18 12:40:09 EDT
Description of Problem:
I've entered my choices for a custom install on my IBM Thinkpad 390X, nothing on the harddrive.
Click "next" on point-of-no-return screen. It waits for a few minutes showing 00.0, 00.0 etc. then install CD 1 starts reading.
I had chosen 1,100 KB worth of packages to install into a /usr of only 1 Gig. Popup says, "You need more." OK.
Next popup says something like, You've got a bug.
Log says: isys.umount() can only umount by mount point
Version-Release number of selected component (if applicable):
I'll try installing Monday with different parameters.
Steps to Reproduce:
boot from CD 1
Choose graphical mode install
Generic 101-key PC closest to my laptop 84-key
Generic, 3-button PS/2 mouse
Custom rather than Laptop installation
Manually partition disk with Disk Druid
I have 6189 M (6.2 Gig)
/ 256M (message recommended) Linux Native
/usr 1G Linux Native
/home 1G Linux Native
/dos 1G DOS 16-bit >=32M
UTC-5, Daylight Saving, System clock uses UTC
two accounts, root and hh
X config: Video probe found NeoMagic 256 (laptop notebook), 256KB
video card RAM
X config: monitor, unprobed, horizontal 31.5-48.5 kHz, vertical 50-70
Take default color depth 16, sn resolution 1024x768, login text.
After I copied error logging to diskette, installation dumped out to text screen
"you may safely boot" or something like that.
A completed install.
I'll try attaching the log.
Created attachment 45040 [details]
error log captured on isys.umount failure
If you do not mount the DOS partition during the install process does it work
My goal is dual boot with Win2000 and am told by work mates that I should install Win first (which I did last year on another machine without this kind of problem). My
attempt to install again from the Red Hat CD was thwarted as follows.
It appears the isys.unmount() failure corrupted the master boot record of my virgin hard drive so that it would not return a proper "boot failure" error to POST. Before
starting with Linux an attempt to boot produced a short "No OS" message. Also, at first, the laptop booted from RedHat CD 1. After the corruption, an attempt to boot
from hard drive produced only a blank screen and the CD would not boot. My fincal solution was to reconfigure the machine's boot sequence, putting CD before
harddrive. But my second Linux install attempt ended directly after the "Welcome to Red Hat" screen with an error like "...logical partition type 0 detected...which
condition is not supported by the anaconda installer...use your other operating system to set it correctly" what ever "correctly" means. I don't know if that's a result of
the isys.umount() failure or of my using Partition Magic 6.0 before I reconfigured the boot sequence. P Magic first showed a C: of 250 something Meg, Active FAT. I
removed it. Boot still wouldn't pass hard drive. But than P Magic said it was still there. So I activated it and formatted it. Still no boot. Then I reconfigured the boot
sequence as already stated.
In a few days I will start with Win to see if he will straighten out the partitions and then install Linux second.
halh, have you made any progress with this issue?
Yesterday, after several trials, I completed my intended install of Dos, Win2000, and RH Linux 7.1 on my laptop 390X. By forgoing the Dos partition, as you
suggeseted, I had no further occurances of the original isys.umount()-umount problem. My problem with disabled boot arose from my inattention to whether Lilo wrote
his boot record to MBR or another partition. However, there may also be a problem in the RH 7.1 install. Even after choosing "boot partition" instead of MRB, your
etc/lilo.conf had "boot=/dev/hda". When I ran Lilo with "boot=/dev/hda8" and some other changes, then Linux booting worked as desired in conjunction with Boot
Magic, 6.0. I also had to repair your etc/fstab. For / and /home it had picked up the volumn labels I assigned with Partition Magic instead of /dev/hda8 and /dev/hda7.
I don't know Linux that well yet, but I'm sure the better it cooperates with Windows now, the quicker we will be free of Windows in the long run.
Although I don't remember any problems with the boot loader things in 7.1, I
have tested that it works properly in current releases.