I just downloaded and installed linux 2.6 and I decided today was the day I was going to learn kernel building.
I had no problem with the actual build of it (besides the time it took).
The problems came afterward. I moved the new image into /boot and left it as bzImage. I updated lilo and rebooted.
LILO loaded up and I picked the new kernel. So far so good, wrong!
It went through setting up hardware and then it hits me with an error.

please append "root= ".

So I thought that i could use rdev and find out the problem. well it said that "/" was at /dev/hdb (thats my CD burner). So I changed it to the correct location /dev/hdc6.
I checked to see if it went through, it did. After another reboot it still smacks me with the same error. I have tried everything I know and i cant fix it, I do have my old kernel 2.4.22 and that works. I just wanted to test out the new 2.6 version.

Can someone tell me how to fix this problem, thanks.

A Slacker

Possibly? :

Kernel can't find the Root partition.
Kernel can't find the drivers for the Root filesystem.

Have you definately got the correct

root=......... directive in lilo.conf.

Did you build support for the Root filesystem?

What's the device name of your Root partition? - can you post lilo.conf.

my drives are : cd burner /dev/hdb , cdrom /dev/hda, hard drive /dev/hdc. Partitions are: Windows /dev/hdc1, Linux root /hdc6 , all others are swap or extended partions that are small (<1MB).
My Lilo.conf is:

# LILO configuration file
# generated by 'liloconfig'
#
# Start LILO global section
append="hdc=ide-scsi"
boot = /dev/hdc
#compact # faster, but won't work on all systems.
prompt
timeout = 50
# VESA framebuffer console @ 1024x768x256
vga = 773 // I removed commented out parts
# End LILO global section
# Linux bootable partition config begins
image = /boot/vmlinuz // Slack default kernel
root = /dev/hdc6
label = Slack
read-only
image = /boot/bzImage // 2.6 kernel
root = /dev/hdc6
label = 2.6.1
read-only

# Non-UMSDOS filesystems should be mounted read-only for checking
# Linux bootable partition config ends
# DOS bootable partition config begins
other = /dev/hdc1
label = WindowsME
# map-drive = 0x80
# to = 0x81
# map-drive = 0x81
# to = 0x80
table = /dev/hdc
# DOS bootable partition config ends



I hope this helps.
Also it seems that lilo is finding the kernel since it tells me that it's loading it, so I think that its a kernel problem (kernel cant find /).

I just wanted to know if it was possible to fix it without recompiling again.


Thanks
A Slacker

From the infos of ur current hardware configuration above I'd suggest:

append=" hda=ide-cd hdb=ide-scsi "

Next time u screw up ur casing, I'd suggest these:

Put the first Harddisk always as master on ide-channel1/dev/hda, if u have another then on channel2 (/dev/hdc). From ur existing hardware i'd suggest:

HD on /dev/hda, burner in /dev/hdc, cdrom in /dev/hdb.

Note:
Ide-channel1 is shared among /dev/hda(master) n /dev/hdb(slave)
Ide-channel2 is shared among /dev/hdc(master) n /dev/hdd(slave)

Thumb rules:

Sharing a channel means that the channel can be used by one of them (master/slave) at one time. So If u consider u use ur CDROM more often then CDRW then u might put cdrom on hdc rather than hdb, bcoz hdb shared channel1 with hda.

Always put the faster device as master on each channel, bcoz the channel is as fast as the master device. So if u put the faster device as slave, the that device is only as fast as the master device (if there is any) on the channel.

I've currently only 2 devices on my system:
I put HD as /dev/hda n CDRW as /dev/hdd, so if I need to install another HD, I ll put it on /dev/hdc, so that I don't need to change the parameter I gave to lilo, it is actually not that problematic with lilo, but somehow by installation slackware installer noticed where the CDRW is installed, so if the cdrw is later moved to another /dev/hdx it confused slack a little bit (am not quite sure, but it is cleaner to install this way for me).

Back to your: "append root= ... " error during boottime:

It could be that the problem is not related to lilo at all. If u have compiled a new kernel it'd highly possible that u have not the right config. Just copy the .config from 2.6.0-test9 somethign from Pat, the use make menuconfig to adjust to ur requirement is easier. Many thing u think ur system don't need, but if u commented them out it can cause that problem like above.

One of them is SCSI-EMULATION. Even u don't have scsi devices, u ll need it or ur HD(ide ones) won't be accessible.. (that from my experience ..), many other thing could contributed to this problem, notice the IDE-chip (or see lspci) of ur mobo into paper while u boot with other working kernel (common are PIIX4 or VIA etc.. ) make the right configuration ..

That my 2 cents of rat smell ..

Common problem: read http://www.linuxquestions.org/questi...391#post735391
Not so.

R u really really sure snacky, have u tried to comment SCSI-emulation ?

I don't know my mobo is asus tx97-e n I've no scsi devices at all on the system. But I can confirm that without "SCSI emulation" it caused that "VFS panic .." on my system, that apply to kernel 2.6.1, 2.6.0, 2.4 for sure also 2.2, but I can confirm that only on 2 or 3 systems I've tested ..

A short flash onto the reference post mentioned those popular failures being made to lose the VFS panic proiblem, such ext3, raiser, etc. In fact I don't have any ext3 or raiser fs at all, I used ONLY ext2. That config with adding ext3, raiserFs have cost me some hours (almost 3 hours each compilation on my system) wihout losing the problem ..

But in general u can not deny the unformal experience u got during long time of ur kernel config, that is usefull also for future config although not at once.

Doon't be messed from the common popular failure, that is guilty on special aspect, can not be be used to deny the experience u made with ur mobo/system during ur previous config, the memory of them come out on time u make similar experience again ..

Possibly that's due to append="hda=ide-scsi" options in lilo.conf or something like that. They're not needed.

I never used ide-scsi and it is DEFINITELY not necessary. The main reason people use it is because the cdrtools people refuse to do things right... or, at least, that's more-or-less what Linus says.

Thanks for all the posts
I was getting the VFS panic problem, so I will try and go back to building 2.6 again and include the things mentioned.

Thanks Again

A Slacker