lilo issue after adding hard drive - Page 4

Erik_FL · 03-25-2009, 12:53 AM

There are a few things that have gotten me into trouble combining IDE and SATA controllers in Linux.

Make sure that the SATA BIOS and the IDE BIOS are both enabled. Some motherboards have a separate SATA BIOS that has to be enabled. Without the SATA BIOS then the BIOS may return the wrong information for the SATA drives.

Enable LBA support in the BIOS for all the drives. Without LBA support enabled the number of cylinders is limited to 1023. With LBA disabled the BIOS translates the LBA addresses into Cylinder Head Sector addresses to access the disk which can make the BIOS LBA locations differ from the Linux LBA locations.

Be careful of BIOS drive ID reassignments. Usually the BIOS assigns drive ID 0x80 to the drive that is actually booted. It may rearrange the other drive IDs depending on which drive the BIOS actually boots first. That can be a problem if you configure LILO using a different Linux OS than the one you are booting. Don't change the default boot drive in the BIOS after using LILO to write the configuration since that may change the drive ID assignments.

I started using GRUB because it has a "native" mode that you boot stand-alone and then you can install GRUB using the BIOS rather than the Linux OS. Among other things that allows you to see the drive ID assignments that will actually be in effect during booting. The "find" command in GRUB can also locate the Linux partition or GRUB partition and display the drive ID. GRUB also figures out the file-system using the BIOS at boot time rather than pre-storing absolute sector addresses for files (as does LILO). Modifying any of the kernel files mentioned in "lilo.conf" requires that you use the "lilo" command again (the file location may change on the disk). The results are unpredictable if you forget to run "lilo" to update the sector information.

Do not try to use a mixture of SATA and IDE ports on the same disk controller chip unless you have that controller chip set to the legacy mode. In the enhanced mode the SATA and IDE ports all share the same interrupt request and Linux can't sort out the interrupts correctly. Separate IDE and SATA controller chips are fine as long as they don't share the same interrupt request level.

Legacy mode assigns IRQ 14 and IRQ 15 to the Primary and Secondary IDE controllers. Then SATA ports are assigned to share a PCI interrupt request level that is steered to some other IRQ besides 14 or 15. Enhanced mode assigns the Primary, Secondary and SATA ports all to the same PCI interrupt and that ultimately goes to the same IRQ level. The IDE driver in Linux isn't designed to share the interrupts with a different driver. The place where you run into this problem is with the Intel ICH chips that have IDE and SATA ports.

Rollo69 · 03-25-2009, 01:41 AM

Thanks for the information Erik_Fl. I will double check my bios the next time I boot up.

@ Justwantin.... Here's the error that came up during boot :

Code:

Verifing DMI pool: ..........
Boot from CD:
DISK BOOT FAILURE, INSERT SYSTEM DISK AND PRESS ENTER

So I inserted the slack-12 disc. At the prompt I entered

Code:

boot: hugesmp.s root=/dev/sda1 rdinit=ro

Then came a error/warning:

Code:

ERROR: Root partition has already been mounted read-write. Modify kernel with rdev so root 
mounts read-only, if using lilo to boot change lilo to read-only

So I continued on, looked at lilo.conf and nothing looked out of the ordinary. I went ahead an ran liloconfig and used the expert mode, created new lilo, installed it and rebooted. Well low and behold it booted up into Slack-12.

I tried making changes to lilo.conf so I could choose to boot either Slack-12 or Slack-10.2 but wouldn't accept the image I gave it. I can mount the ide drive from Slack-12 though.

Also the sata still isn't coming up when I run fdisk -l as root. May have to do with what Erik_Fl was talking about.

I'm going to have to setup the network so I can get online from Slack-12 when I get the chance.

One step closer, Thank you justwantin.

justwantin · 03-25-2009, 02:07 AM

Did you run liloconfig and select to write to the MBR and did you confirm that vmlinuz was pointing to the huge kernel?

Rollo69 · 03-25-2009, 02:11 AM

Yes, I wrote it to the MBR and

Code:

# ls -l /boot/vmlinuz
lrwxrwxrwx 1 root root 29 2009-03-04 08:33 /boot/vmlinuz -> vmlinuz-huge-smp-2.6.21.5-smp

Rollo69 · 03-25-2009, 02:32 AM

Quote:

Originally Posted by Erik_FL

There are a few things that have gotten me into trouble combining IDE and SATA controllers in Linux.

Make sure that the SATA BIOS and the IDE BIOS are both enabled. Some motherboards have a separate SATA BIOS that has to be enabled. Without the SATA BIOS then the BIOS may return the wrong information for the SATA drives.

I don't have separate BIOS for sata and ide. There is a section for sata in my bios under IDE devices. The sata controller is enabled. I have a sata controller mode =(choose between ide or raid) ide selected.

Quote:

Enable LBA support in the BIOS for all the drives. Without LBA support enabled the number of cylinders is limited to 1023. With LBA disabled the BIOS translates the LBA addresses into Cylinder Head Sector addresses to access the disk which can make the BIOS LBA locations differ from the Linux LBA locations.

I didn't notice anything about LBA.

Quote:

Be careful of BIOS drive ID reassignments. Usually the BIOS assigns drive ID 0x80 to the drive that is actually booted. It may rearrange the other drive IDs depending on which drive the BIOS actually boots first. That can be a problem if you configure LILO using a different Linux OS than the one you are booting. Don't change the default boot drive in the BIOS after using LILO to write the configuration since that may change the drive ID assignments.

I do have an option to choose which drive (ide or sata) is loaded first in my BIOS (both drives show up) and ide is first. It's an MSI mobo with Phoenix bios.

Thanks again Erik_Fl..!

justwantin · 03-25-2009, 02:44 AM

You can select which drive is loaded first and have the ide loaded first but lilo is run on the sata drive and writes to the mbr on the sata drive? It would seem to me that you would want the mbr that lilo wrote to read first, i.e. the sata drive loaded first.

I think somewhere along the line I suggested just getting the sata working correctly and then adding the ide drive. I would unplug power and disconnected the cable to the ide before troubleshooting the sata. I'm a bit methodical and eliminate as many varibles as possible first.

Rollo69 · 03-25-2009, 03:01 AM

Quote:

Originally Posted by justwantin

You can select which drive is loaded first and have the ide loaded first but lilo is run on the sata drive and writes to the mbr on the sata drive? It would seem to me that you would want the mbr that lilo wrote to read first, i.e. the sata drive loaded first.

The selection for the ide in bios was default after I went to look at the bios, I didn't change it. I will have the sata loading first once you help me so that I can get Slack-12 to load the Slack-10.2.

Quote:

I think somewhere along the line I suggested just getting the sata working correctly and then adding the ide drive. I would unplug power and disconnected the cable to the ide before troubleshooting the sata. I'm a bit methodical and eliminate as many varibles as possible first.

You did suggest that and I need your help getting Slack-12's lilo.conf to load the Slack-10.2. I don't have internet working with 12 yet. I did a route -n (I think that's the option I chose) and it didn't show my gateway or static ip. I changed the rc.inet1.conf so eth0 has my static ip and the gateway is listed after gateway. It only shows the lo .

Code:

# route -n
Kernel IP routing table
Destination     Gateway         Genmask         Flags Metric Ref    Use Iface
127.0.0.0       0.0.0.0         255.0.0.0       U     0      0        0 lo

This isn't from 12 just a reference so you know what I'm talking about.
I got eth0 up/started but ifconfig doesn't show eth0. I tried using the command "route add gw default ###.###.#.#" but didn't work so I looked at the man page and tried using what it suggested for the command but it didn't work either, so I still need to get that working before I completely choose to boot up in Slack-12. Just my way of thinking. But you know more about 12 than me so I'll wait for your help with this.

Thanks again justwantin.

justwantin · 03-25-2009, 04:03 AM

If you want me to help you:

1) Disconnect the ide drive, forget about it for now and concentrate on getting 12.0 working correctly.

2) Once you have completely disconnected the ide drive make sure that the sata drive will be recognised by the bios and make sure everything else in the bios is set correctly.

3) Boot using your install disk

4) Once you have booted up and you are in a terminal run the command liloconfig again select simple, select mbr, then run lilo if you get any errors note them down. and come back with the exact wording of the error. Do not edit /etc/lilo.conf or anything, just note the error and post them here.

5) If you don't get any errors then most likely you will be able to boot your system so then configure your network connection

6) To configure the network connection all you need to do is run the command netconfig as root and answer the questions. I don't know what kind of a connection you will be using but The Slackbook covers it all very well so check out netconfig at http://www.slackbook.org

I am not certain you have not altered some configuration file(s) or copied over something from 10.2 that probably should not have been copied over, or removed something, etc. In short, I am really not all that confident that your 12.0 install is still in a vanilla state which makes it very difficult for me or anyone else to figure out what your problem might or mightn't be.

If it is in a vanilla state and the slackware installation recognised your sata drive, correctly identified the partitions and installed to them accordingly, then the chances should be very good that it would boot with lilo after the install but from what I understand you never booted your system after the install but instead booted up 10.2 and in your words started configuring 12.0. You are not even sure if it was working correctly before you started messing around with it.

I am really not sure what has gone on and if you handed me that box I'd just take out the ide drive, reinstall and start from scratch with a new install.

Sorry I can't be of more help and I have an early day tomorrow so this is my last post tonight maybe someone else will pick this up and run with it.

Rollo69 · 03-25-2009, 01:25 PM

Quote:

Originally Posted by justwantin

I am not certain you have not altered some configuration file(s) or copied over something from 10.2 that probably should not have been copied over, or removed something, etc. In short, I am really not all that confident that your 12.0 install is still in a vanilla state which makes it very difficult for me or anyone else to figure out what your problem might or mightn't be.

I have not copied anything into Slack-12 and the only changes to config files I made I have told you about, which was adding a static ip to the rc.inet1.conf for eth0. I had to add my usb to the fstab and in the /mnt directory so I could get the info you asked for in previous post. I ran the netconfig when I installed 12. BUT, maybe since I didn't reboot back into 12 ,the config files I ran when I installed 12, didn't take. I'm not sure. The only thing I did last night was what you said to do -"rerun liloconfig'. I did and there were no errors and rebooted nicely into Slack-12. But since I couldn't get online through 12 I plugged in the ide drive so I could tell you how things went.

Quote:

If it is in a vanilla state and the slackware installation recognised your sata drive, correctly identified the partitions and installed to them accordingly, then the chances should be very good that it would boot with lilo after the install but from what I understand you never booted your system after the install but instead booted up 10.2 and in your words started configuring 12.0. You are not even sure if it was working correctly before you started messing around with it.

I did not start configuring 12.0 from 10.2 . That was what I planned on doing not what I did. If I have irritated you during this process I'm sorry, not my intentions. Thank you for your time and help in this matter justwantin.

laters,

justwantin · 03-25-2009, 02:01 PM

There seems to be a communication problem here whether it is on my end or your end does not matter at this stage. If you are now booting your 12.0 install with lilo then this thread is redundant and perhaps you should start another one with regards to dual booting a second system (your 10.2) if you still want to and/or configuring your network access but in either cases state precisely what your problem is, what you have done, and what error messages there might have been.

Erik_FL · 03-25-2009, 02:08 PM

I'm sorry if you already posted this but, could you post the exact MSI motherboard model? If you do I'll be happy to download and review the manual to see if there are any "gocthas" in the BIOS settings.

One thing that you mentioned mentioned is the SATA settings. I'm used to seeing three choices, IDE, AHCI and RAID. You mentioned "IDE" and "RAID". The "IDE" setting that I'm used to seeing makes the SATA ports appear to be a normal Primary and Secondary IDE controller. That isn't usually what you want and it also prevents you from using any IDE ports that are on the motherboard. Setting IDE mode may actually cause Linux to use the IDE/ATA driver instead of the SATA driver. The AHCI mode usually assigns the SATA ports somewhere else besides the normal IDE controller addresses and enables faster DMA modes. I have also seen some BIOS software where you have to set RAID to get the AHCI mode. In that case just don't create any RAID arrays and the BIOS will run the SATA ports using AHCI mode. Unfortunately every BIOS is different.

Don't confuse the boot-loader (lilo) with the kernel. You can install multiple copies of lilo to different disks but you MUST install lilo with the BIOS hard disk boot order set the way it will be when you use that copy of lilo. When you switch the default boot disk between the SATA and IDE you will have to re-install lilo (use the lilo command). All the different copies of lilo can boot the same kernel files but they will only find the files if the disks are configured the way they were when lilo was installed.

The BIOS numbers hard disks like this.

0x80 - First hard disk (boot disk)
0x81 - Second hard disk
0x82 - Third hard disk
0x83 - Fourth hard disk

I do mean hard disks and NOT partitions or drive letters. All the partitions on the first hard disk use drive ID 0x80 because the BIOS doesn't know anything about partitions.

What the "lilo" installation command does is convert all of the Linux device names into the corresponding BIOS drive IDs. That's fine as long as those disks will be configured the same way when lilo actually boots. When the BIOS boots from a different disk it may assign different drive IDs even when that isn't necessary. It has to assign 0x80 to the boot disk but it might also rearrange the other IDs.

I recommend using grub for multi-booting especially if you will be booting from multiple hard disks. You can specify the file locations of the kernel and initrd to grub at boot time and boot from kernels that aren't even referenced in the grub configuration menu file.

If your ultimate goal is to boot from the SATA drive then I suggest that you configure it as the first hard disk boot drive and leave it configured that way. Double check the BIOS settings and verify that the kernel is detecting the device as an SATA drive and not an ATA/IDE drive. You can install lilo using any Linux OS as long as you identify the file locations using the device names that OS expects (not the OS ultimately being booted). Just make sure not to change the BIOS boot device after installing lilo.

What I usually do is boot the Slackware Setup CD.
Then I mount the new OS partition.

Code:

mount /dev/sda1 /mnt

I change the root to the new OS.

Code:

chroot /mnt

Mount the pseudo file-systems used by lilo to get system information.

Code:

mount -t proc none /proc
mount -t sysfs none /sys

Build the kernel if you want.

Code:

cd /usr/src/linux
make menuconfig
make clean all
make modules_install

Copy the kernel files to the /boot directory.

Change to the "etc" directory and edit the lilo configuration.

Code:

cd /etc
nano lilo.conf

After saving "lilo.conf" then use the lilo command.

Code:

lilo

And finally un-mount everything.

Code:

umount /sys
umount /proc
exit
umount /mnt

Use Ctrl Alt Delete to reboot.

You have to do the lilo part of this (not the kernel build) whenever you rearrange the hard disk ordering in the BIOS or update the kernel file locations or files.

Using the kernel option "rdinit=ro" causes Linux to use the "init" process from the "initrd" image. To mount the root device read only just use the "ro" option (no quotes). I think the "ro" option is what you actually want to use. Are you using an "initrd" image? The only reason to do that is if the root device driver (SATA or IDE) is not built into the kernel that you selected. To use an initrd image you have to specify the "initrd=" option in "lilo.conf". You should not have to include the "rdinit=ro" option in the boot command to lilo.

Rollo69 · 03-25-2009, 08:27 PM

Thanks guys for your replys. I'm not sure if I will be able to get back to you today with the information but will definitely post tomorrow.

thank you,

Rollo69 · 03-26-2009, 01:02 PM

Quote:

Originally Posted by justwantin

There seems to be a communication problem here whether it is on my end or your end does not matter at this stage. If you are now booting your 12.0 install with lilo then this thread is redundant and perhaps you should start another one with regards to dual booting a second system (your 10.2) if you still want to and/or configuring your network access but in either cases state precisely what your problem is, what you have done, and what error messages there might have been.

My problem is that I can't get online - ERROR - Can't Reach Server. I ran netconfig everything looked ok to me.

Code:

# route -n
Kernel IP routing table
Destination     Gateway         Genmask         Flags Metric Ref    Use Iface
192.168.0.0     0.0.0.0         255.255.255.0   U     0      0        0 eth0
127.0.0.0       0.0.0.0         255.0.0.0       U     0      0        0 lo
0.0.0.0         192.168.0.1     0.0.0.0         UG    1      0        0 eth0
# ifconfig
eth0      Link encap:Ethernet  HWaddr 00:19:DB:89:23:E1  
          inet addr:192.168.0.102  Bcast:192.168.0.255  Mask:255.255.255.0
          inet6 addr: fe80::219:dbff:fe89:23e1/64 Scope:Link
          UP BROADCAST RUNNING MULTICAST  MTU:1500  Metric:1
          RX packets:624 errors:0 dropped:0 overruns:0 frame:0
          TX packets:19 errors:0 dropped:0 overruns:0 carrier:0
          collisions:0 txqueuelen:1000 
          RX bytes:203691 (198.9 KiB)  TX bytes:2007 (1.9 KiB)
          Interrupt:21 Base address:0xc000 

lo        Link encap:Local Loopback  
          inet addr:127.0.0.1  Mask:255.0.0.0
          inet6 addr: ::1/128 Scope:Host
          UP LOOPBACK RUNNING  MTU:16436  Metric:1
          RX packets:132 errors:0 dropped:0 overruns:0 frame:0
          TX packets:132 errors:0 dropped:0 overruns:0 carrier:0
          collisions:0 txqueuelen:0 
          RX bytes:11812 (11.5 KiB)  TX bytes:11812 (11.5 KiB)

# ping -c2 192.168.0.1
PING 192.168.0.1 (192.168.0.1) 56(84) bytes of data.
64 bytes from 192.168.0.1: icmp_seq=1 ttl=127 time=2.23 ms
64 bytes from 192.168.0.1: icmp_seq=2 ttl=127 time=0.306 ms

--- 192.168.0.1 ping statistics ---
2 packets transmitted, 2 received, 0% packet loss, time 1001ms
rtt min/avg/max/mdev = 0.306/1.271/2.237/0.966 ms

I can communicate with my router but I can't get to the net.

@--> Erik_Fl : The MSI mobo book says BIOS : Phoenix Awards (LPC) Flash ROM.

Thanks guys,

Erik_FL · 03-26-2009, 01:54 PM

Quote:

Originally Posted by Rollo69

My problem is that I can't get online - ERROR - Can't Reach Server. I ran netconfig everything looked ok to me.

Code:

# route -n
Kernel IP routing table
Destination     Gateway         Genmask         Flags Metric Ref    Use Iface
192.168.0.0     0.0.0.0         255.255.255.0   U     0      0        0 eth0
127.0.0.0       0.0.0.0         255.0.0.0       U     0      0        0 lo
0.0.0.0         192.168.0.1     0.0.0.0         UG    1      0        0 eth0
# ifconfig
eth0      Link encap:Ethernet  HWaddr 00:19:DB:89:23:E1  
          inet addr:192.168.0.102  Bcast:192.168.0.255  Mask:255.255.255.0
          inet6 addr: fe80::219:dbff:fe89:23e1/64 Scope:Link
          UP BROADCAST RUNNING MULTICAST  MTU:1500  Metric:1
          RX packets:624 errors:0 dropped:0 overruns:0 frame:0
          TX packets:19 errors:0 dropped:0 overruns:0 carrier:0
          collisions:0 txqueuelen:1000 
          RX bytes:203691 (198.9 KiB)  TX bytes:2007 (1.9 KiB)
          Interrupt:21 Base address:0xc000 

lo        Link encap:Local Loopback  
          inet addr:127.0.0.1  Mask:255.0.0.0
          inet6 addr: ::1/128 Scope:Host
          UP LOOPBACK RUNNING  MTU:16436  Metric:1
          RX packets:132 errors:0 dropped:0 overruns:0 frame:0
          TX packets:132 errors:0 dropped:0 overruns:0 carrier:0
          collisions:0 txqueuelen:0 
          RX bytes:11812 (11.5 KiB)  TX bytes:11812 (11.5 KiB)

# ping -c2 192.168.0.1
PING 192.168.0.1 (192.168.0.1) 56(84) bytes of data.
64 bytes from 192.168.0.1: icmp_seq=1 ttl=127 time=2.23 ms
64 bytes from 192.168.0.1: icmp_seq=2 ttl=127 time=0.306 ms

--- 192.168.0.1 ping statistics ---
2 packets transmitted, 2 received, 0% packet loss, time 1001ms
rtt min/avg/max/mdev = 0.306/1.271/2.237/0.966 ms

I can communicate with my router but I can't get to the net.

@--> Erik_Fl : The MSI mobo book says BIOS : Phoenix Awards (LPC) Flash ROM.

Thanks guys,

RE: The BIOS. I need to know the MSI Motherboard model number in order to find the BIOS documentation on the MSI site.

Your routing table shows that you have the correct default gateway of "192.168.0.1" and your "ping" test verifies that you can get to the router. That pretty much leaves DNS (Domain Name System).

Check to see if you can look up names.

Code:

nslookup www.google.com

If that works then the problem has to do with the web browser or desktop software rather than the Linux OS.

If DNS name resolution doesn't work then check to make sure that there is a name server defined.

Code:

cat /etc/resolv.conf

There are two ways for information to get into "resolv.conf". When you use DHCP then DHCP updates the file. When you don't use DHCP then you have to manually add DNS servers to the file.

Code:

# Generated by dhcpd for interface eth0
nameserver 192.168.0.1

Most broadband routers have a feature called DNS relay or DNS proxy that forwards DNS requests to the servers on the WAN. You use the router's address as the DNS server and then the LAN PCs don't have to worry about changes to the DNS server addresses on the WAN. If your router doesn't support DNS relay/proxy then it should provide the WAN DNS server addresses in the DHCP information. When you configure the DNS server manually you can use the WAN DNS server addresses (always) or the router's address if it supports DNS relay/proxy as most do.

If name resolution works you may also want to test communication using "traceroute".

Code: