Lilo - SCSI

[msherman]

I am finishing up my first Gentoo install. I installed it on a SCSI drive, while windows resides on an IDE one.
Which option should I set for SCSI in lilo.conf? I don't really know the syntax.
Gentoo manual says the following:
"genkernel users should know that their kernels use the same boot options as is used for the LiveCD. For instance, if you have SCSI devices, you should add doscsi as kernel option. "

Also, if something goes wrong, I should be able to boot with a Live CD, chroot and update the lilo.conf (or whatever goes wrong) ?

Lastly, when running /sbin/lilo - will it install itself into the MBR?

Thanks in advance

[slycordinator]

[paulecoyote]

Make sure you've compiled everything as a * in your kernel for scsi drives... esp the /dev stuff that's called "obsolete".

The drivers you need to install depends on your scsi card, but if it does't pick up your scsi device you know you've not compiled the right one in yet. You could go failsafe and compile in all the scsi drivers, and then take them out one by one.

lspci is your friend for finding out the correct driver for your card (when you're in livecd still)

lsmod is useful to see what kernel modules are actually being used too - also helpful when picking drivers and modules for your kernel.

Also be wary about the naming of the devices in /dev .... the live cd may call the sdb1 sdb2 ... etc... while when the drive has been made primary it will infact become sda1... sda2... etc. Though as you are using a mixture of scsi and ide I doubt this will be an issue.

If you have all the right drivers in you shouldn't need the hotplugging... i suggest you compile that in as a module and enable as necessary. Some chipsets (including mine) seem to make the usb hotplugging hang while shutting down.

[msherman]

Thanks

I am getting kernel to panic, saying the root= is not pointing to the right location. Here's the ouput of fdisk -l

Disk /dev/hda: 30.7 GB, 30735581184 bytes
255 heads, 63 sectors/track, 3736 cylinders
Units = cylinders of 16065 * 512 = 8225280 bytes

Device Boot Start End Blocks Id System
/dev/hda1 * 1 522 4192933+ 7 HPFS/NTFS
/dev/hda2 523 3736 25816455 7 HPFS/NTFS

Disk /dev/sda: 73.4 GB, 73407867392 bytes
255 heads, 63 sectors/track, 8924 cylinders
Units = cylinders of 16065 * 512 = 8225280 bytes

Device Boot Start End Blocks Id System
/dev/sda1 * 1 5 40131 83 Linux
/dev/sda2 6 68 506047+ 82 Linux swap
/dev/sda3 69 8924 71135820 83 Linux

and here's lilo.conf

boot=/dev/hda
prompt
timeout=50
default=gentoo
vga=771

image=/boot/kernel-2.6.7-gentoo-r11
label=gentoo
read-only
root=/dev/sda3

other=/dev/hda1
label=Windows

What could that be?

[slycordinator]

Kernel panic has nothing to do with lilo.

Kernel panic, obviously, is part of the kernel. So we can only figure out what is causing the panic if you say what the kernel panic says (the error message).

[paulecoyote]

See, I reckon you haven't got the right modules and scsi things configured for your kernel.

When in live cd do lsmod - and note down the names of the modules and drivers being used to drive your scsi bits. manually edited your .config file and find in nano ctrl+w) the names of the modules and hard set them to be in if you like. Then run make menuconfig to get it to parse it all and make any other changes. Recompile your kernel and install it, and give it a whirl.

I think you need to select absoultely all of the scsi hard-drive stuff if you want something quick, or try and error enabling things until it does work if you want to be through.

sda SHOULD be your linux scsi drive - the fact it can't find it means the device very likely hasn't been included in your dev bit. Make sure the /dev directory stuff is enabled in the kernel configuration.

if it finds root and screws up later, your fstab is probably a little off. ReiserFs likes to have certain options on too as well sometimes.

[javabiz]

I'm slipping over into this discussion because my situation with the scsi lilo is similar but not exactly the same problem. Please see the posts at the bottom--for a review of my situation.
At present I am at a
livecd root #
prompt
Essentially I know my lilo has sda1 instead of sda in it so I can't boot correctly.
I'm trying to save my install which was near done!
I've rebooted the X86LiveCD to bail me out and I've done the following:
at the boot prompt
boot:
I entered
smp nodhcp noraid noapic
I then mounted all my devices similar to the mounting part of the
manual eg
mount /dev/sda1 /mnt/gentoo
and through sdc8
# df -v
returns what I would expect after this!
Now what?
I'm changing the lilo.conf file to sda instead of sda1.
Believe all my stuff is listed at the other posts.

https://forums.gentoo.org/viewtopic.php?t=230426
https://forums.gentoo.org/viewtopic.php?t=230219&highlight=

Any help would be greatly appreciated! And I'll try and help out where I can. Thanks

[paulecoyote]

Again this was fustrating me with my scsi only system - it's a case of having everything in your kernel enabled for your scsi stuff.

sda = hard disk
sda1 = boot partition (probably)
sda2 = swap partition (probably)
sda3> = root and everything else.


I use grub personally, this is my file (the box shares itself with a windows installation - it uses hardware raid - hd0 = windows raid array, hd1 = linux raid array - they appear as single drives to the operating systems).

Note the root=/dev/sdb3 bit - thats probably where yours is failing. And it's probably failing because you haven't got the /dev stuff (called obsolute in the kernel config menu) installed, or you've not got all your scsi stuff. Triple, quadruple check your kernel configutation, it's easy to miss.

Note mine is called sdb3 - sd(b) because it's the second physical scsi drive in the operating systems eyes - 3 because thats the / partition (or /mnt/gentoo before chrooting) and 1 is /boot (or /mnt/gentoo/boot before chrooting).


(grub.conf)
# Which listing to boot as default. 0 is the first, etc
default 0

# How many seconds to wait before the default listing is booted.
timeout 10

# Splash screen
splashimage=(hd1,0)/grub/splash.xpm.gz


# ---------------------------------
# Gentoo partition
# ---------------------------------
title=Gentoo Linux - 2.6.8-gentoo-r3
# Partition where the kernel image (or operating system) is located - boot partition very likely!
root (hd1,0)
kernel /kernel-2.6.8-gentoo-r3 root=/dev/sdb3



# ---------------------------------
# Windows partition
# ---------------------------------
title=Windows 2000 Server
rootnoverify (hd0,0)
makeactive
chainloader +1

[javabiz]

Paul
Here's where it gets a little interesting! When I used the boot command
#smp nodhcp noraid noapic
I assumed that it would carry over into the following commands:
#emerge gentoo-dev-sources
#emerge genkernel
#genkernel all --menuconfig
The --menuconfig was used at the suggestion of another post but I didn't change anything in the kernel menuconfig and exited without saving and it compiled just as a regular
#genkernel
command I assumed!
But later on when I did my lilo command and fstab--the system wouldn't reboot with the
#exit
#umount /mnt/gentoo/boot /mnt/gentoo/proc /mnt/gentoo
#reboot
sequence
It gave me a keyboard error when I cold booted
again I assume that this relates to my sda1 use instead of sda as the mbr at boot!
So, now that I've rebooted in the livecd root prompt and remounted all my mount points except for /dev/proc
--what should I do now?

[javabiz]

Here's the lilo that I propose to solve the problem-what do you think. And below is my fstab

boot=/dev/sda #Install LILO in the mbr of scsi drive
prompt #Give user chance to select another section
timeout=50 #Wait 5 seconds before booting default
default=gentoo #When timeout passed-boot gentoo section
#Only if you use framebuffer. Otherwise remove the following line:
vga 788 #Framebuffer setting. Adjust to your will.

#For genkernel users
image=/boot/kernel-2.6.8-gentoo-r3
label=smp #Name we give this section
read-only
root=/dev/ram0
#old append="init=/linux ramdisk=8192 real_root=/dev/sda5 nodhcp noraid doscsi noapic"
append="init=/linux ramdisk=8192 real_root=/dev/sda5"
initrd=/boot/initrd-2.6.8-gentoo-r

When I originally mounted my filesystems I used the following:

#mount /dev/sda5 /mnt/gentoo
#mkdir /mnt/gentoo/boot
#mount /dev/sda1/ /mnt/gentoo/boot
#mkdir /mnt/gentoo/var
#mount /dev/sdb1 /mnt/gentoo/var
#mkdir /mnt/gentoo/usr
#mount /dev/sdc1 /mnt/gentoo/usr
#mkdir /mnt/gentoo/home
#mount /dev/sdc5 /mnt/gentoo/home
#mkdir /mnt/gentoo/tmp
#mount /dev/sdc6 /mnt/gentoo/tmp
#mkdir /mnt/gentoo/var/ftp
#mount /dev/sdc7 /mnt/gentoo/var/ftp
#mkdir /mnt/gentoo/var/www
#mount /dev/sdc8 /mnt/gentoo/var/www

I used all the same mounts in fstab and remarked out /dev/ROOT
and /dev/BOOT

Here is my fstab file:

/dev/sda5 / ext3 defaults 1 1
/dev/sda1 /boot ext3 defaults 1 2
none /dev/pts devpts mode=0620 0 0
/dev/sdc5 /home ext3 defaults 1 2
/dev/hda /mnt/cdrom auto umask=0,user,iocharset=iso8859-1,codepage=850,noauto,ro.exec 0 0
none /mnt/floppy supermount dev=/dev/fd0,fs=ext2:vfat,--,umask=0,iocharset=iso8859-1,sync,codepage=850 0 0
none /proc proc defaults 0 0
/dev/sdc6 /tmp ext3 defaults 1 2
/dev/sdc1 /usr ext3 defaults 1 2
/dev/sdb1 /var ext3 defaults 1 2
/dev/sdc7 /var/ftp ext3 defaults 1 2
/dev/sdc8 /var/www ext3 defaults 1 2
/dev/sda6 swap swap defaults 0 0
none /mnt/nd supermount dev=/dev/scsi/host0/bus0/target1/lun0/part1,fs=ext2:vfat,--, umask=0,iocharset=iso8859-1,kudzu,codepage=850 0 0
none /dev/shm tmpfs defaults 0 0

Do you see any problems here--shouldn't it boot up with these devices then mounted--if I cold boot?
Your evaluations would be greatly appreciated!

[paulecoyote]

Nah, what you put in the options when you boot from the livecd is only for the livecd. All that mounting and everything when you're in that enviornment means nothing to your actual installed one. Yeah you can interact with it, even chroot to it - but you're not actually relying on it's version of /dev - you are relying on livecds version of /dev.

When you reboot off your hd, your installation is on its own - it didn't have livecd to set up all the devices... and because the kernel isn't configured quite right when you compiled it your installation can't handle the scsi stuff. That's why you can see it when you are on the livecd fine but it can't find squat when you boot up from your hd.

Follow my advice - go through your kernel menu and make sure stuff is enabled like I said. The actual LiveCD kernel probably has almost everything enabled and compiled in by default so it can detected everything... when you are installing your kernel proper they encourage you to only put in what you need to stream line things... and at the moment it's too streamlined for you - it actually doesn't have everything you need.

you seem to be able to get your other config files posted - post your .config file for your kernel and I expect someone (me or whomever) would be able to spot what you need and what you haven't done.

EDIT: But try it yourself first

[javabiz]

I thought it was supposed to take everything your CD says you have-as far as hardware and automatically include that in the
.config file.
I was trying to stay away from genkernel all --menuconfig
for that very reason.
When I booted it came to a boot: prompt and said to pick which kernel I wanted and I selected smp
which I suppose is nothing but what I had in the lilo.conf file, right!
It then complained about my root device which I had in the real_root part of the append command set to /dev/sda5 in the above lilo post.
I looked at the .config command and there seems to be some scsi lines uncommented and some network lines uncommented. It seems when you run genkernel all--that this command should include all the hardware you had in the cd mount. I don't understand why the manual says you do and then you evidently don't. Maybe what I'm trying to do with the smp and the three scsis is way too complicated for the Gentoo genkernel command. If so, can I run it a different way and have it show me in menuconfig what's actually there as opposed to what needs to be there?

[slycordinator]

[javabiz]

But, according to the manual:
Genkernel works by configuring a kernel nearly identically to the way our LiveCD kernel is configured. This means that when you use genkernel to build your kernel, your system will generally detect all your hardware at boot-time, just like our Live CD does. Because genkernel doesn't require any manual kernel configuration, it is an ideal solution for those users who may not be comfortable compiling their own kernels.

The question is-is genkernel as smart as the X86LiveCD--I believe it is because if gentoo goes to all the trouble to create a cd that can figure out your hardware--why wouldn't it be smart enough to transfer that info to your .config when you run
#emerge genkernel
#genkernel all
It may include alot of other drivers that are not needed but it should work for any hardware that it finds with the live cd as well as the genkernel process!
I think my initial problem lies with lilo.conf at this point or at least I need to get to the point where I get a kernel error instead of a lilo error. I will double check my .config file a couple of times with the info you gave me above!

[slycordinator]

[paulecoyote]

The kernel on the livecd is different to the one you compile - after all it's built with your compiler flags on your machine etc.

Just ensure all the scsi stuff is [*] and the obsolete dev stuff is [*] when you compile your kernel in the menu config and I'm sure it will all be fine.

Then when you are more comfortable with this whole Linux thing you can change your .config to something to produce a more streamlined Kernel. First get a config that works, back that up then screw about.

This Gentoo stuff can lead to a common developer aliment of pre-emptive and unnecessary "optimisation" before you even get something working - don't get caught out.

EDIT: YOUR problem is not your LILO - it's very likely the lack of the proper bloody drivers in your kernel!!! It doesn't matter how many ways you tell your LILO that sda1 exists if the Kernel doesn't understand how to mount that drive in the first place!

[paulecoyote]

[paulecoyote]

Oh and that error you got earlier? WIth it not rebooting properly? I get that when usb hotplugging is enabled. I've seen posts about other people having similar problems and solving them, but I've not managed to fix it myself.


Again from what I've read I don't think I've enabled everything the right way in the Kernel, cos the LiveCD doesn't have the same problem. But it can be tough configuring it correctly - even the handbook says so.

[javabiz]

Thanks,
That's encouraging!
The line that includes (fs=ext2:vfat) is from my Mandrake install with the same hardware. Unfortunately the fs should be ext3! It looks like the manual is assuming a plain jane install of hda instead of dual processors and scsi drives (3 each)!
Looks like I'm going to have to look at the nitty gritty in that .config file or do the menuconfig gig!