I'm a newbie on my first Kernal adventure. I've made many attempts to do a custom complie of Red Hat 2.4.20-20.9 (without backing up the original -- shame on me). I keep getting the following message (despite many variations in Grub config):

"VFS: Cannot open root device "root=" boot option
Kernal panic: VSF: Unable to mount root fs on 00:00"

The last thing displayed above this from the failed boot attempt is:

"Net4: Unix domain sockets 1.0/SMP for Linux Net4.0"

I still am able to boot Kernal 2.4.20-6 (at least I had enough presence of mind not to nuke that when I started to play around). When 2.4.20-6 boots, the text re. Net4 reads as follows:

"Net4: TCP/IP 1.0 for Net4.0"

I assume this difference is critical, but can't find "Net4" in either versions config. file. Grub configuration is identical for both kernals, except for version no. (of course), so doesn't appear to be there.

Any suggestions on where I went wrong. Please bear in mind that if there is someplace I need to configure beyond the .config file, it's probably all news to me.

Thanks.

umm, i'm not 100% sure what net 4 is, but frankly i don't think that's your problem: the kernel panic is being caused by the fact the kernel can't find a root filesystem, which makes me think that your grub entry is somehow wrong. i'm a lilo man myself, so i dunno what your grub entry should look like, but in lilo you have to tell each kernel that you want to boot where the root fs is, e.g. "root=/dev/hda3" or something. i see you said you tried different grub configs-- could you post some of them? (i know you said the configs are the same for those that work and those that don't, but humor me).

also, regarding that net4 thing. it mentions something about SMP. does your system have more than one processor? if not, and you compiled the kernel for symetric multiprocessors, it might cause it to panic.

Make sure that the kernel u compile is confgured to load the root fs system ..in my case/most cases its ext2 which u canot load it as a module ...
It will fail to boot the kernel...

Never mind the unix domain sockets it doenst have anythng to do wth ur kernel failures...

Not sure how to configure the Kernal to load the root file system, but I had another idea that seems to have solved the problem. Since I still had an earlier working version (v. 2.4.20-6), I decided to select the xconfig option that lets you work from an existing version's config file. I got a lot more "stuff" in my kernal than I really want, but with a quick test, I found that it now boots and I can work it back by de-selecting kernal configuration options from there. All in all, looks like a case of just needing to be more aware of what different choices mean during the config process (sorry -- not sure where I made the mistake, but I'll post again if I recreate it while paring the kernal back).

By the way, before finding this dummy solution, I did reconfigure and made sure to deselect multiprocessor. That did not eliminate the problem (although I had not intended to select SMP in the first place). I'm all the more convinced that grub was not the problem, but here's my grub.conf file for anyone that's interested:

# grub.conf generated by anaconda
#
# Note that you do not have to rerun grub after making changes to this file
# NOTICE: You have a /boot partition. This means that
# all kernel and initrd paths are relative to /boot/, eg.
# root (hd0,1)
# kernel /vmlinuz-version ro root=/dev/hda3
# initrd /initrd-version.img
#boot=/dev/hda
default=2
timeout=10
splashimage=(hd0,1)/grub/splash.xpm.gz
title Red Hat Linux (2.4.20-20.9-JH030924)
root (hd0,1)
kernel /vmlinuz-2.4.20-20.9-JH030924 ro root=LABEL=/
initrd /initrd-2.4.20-20.9-JH030924.img
title Red Hat Linux (2.4.20-6)
root (hd0,1)
kernel /vmlinuz-2.4.20-6 ro root=LABEL=/
initrd /initrd-2.4.20-6.img
title DOS (Windows 2000)
rootnoverify (hd0,0)
chainloader +1

Thanks, all, for the helpful replies.