Kernel Panic

[K.C. Chan]

All, I've compile kernel 2.4.17 and created initrd for it. Here's what is under lib from initrd.img file:
aic7xxx.o
ext3.o
freevxfs.o
jbd.o
lvm-mod.o
md.o
multipath.o
scsi_mod.o
sd_mod.o
sym53c8xx.o

Root file system is /dev/sda3, but during boot the kernel panic because it is unable to mount /dev/dsa3, here's the err: "
request-module [block-major-8]: Root fs not mounted, please append a correct "root=" boot option, kernel panic: VFS: unable to mount root on 08:03.". I know that lilo is pointing to the correct root fs "root=/dev/sda3". Can anyone shed some lights on this? Thanks.

[Krishna Prasad]

I got the same error when installing SuSe on a LH3 Netserver. However, RedHat worked fine. RedHat was able to detect that it needed the megaraid card loaded.

Are you using Suse or Redhat?

Perhaps the newer Kernel needs to load the megaraid driver. ( or what ever controller driver you are using?)

[K.C. Chan]

Ron,
I am using RedHat. I think I've looded all neccessary driver via initrd. But the error points to root fs /dev/sda1 which is scsi. I've loaded everything I could thing of related to scsi. Any idea?

[Steve Darnell]

I always avoid using initrd whenever possible, by statically linking into the kernel all of the drivers needed to boot (in your case, scsi, aic7xxx, etc.).
I use initrd only when an essential driver is available only as a module.

[Charles Slivkoff]

You might want to try my suggestions from this post:

http://forums.itrc.hp.com/cm/QuestionAnswer/1,,0x1f2818276953d61190040090279cd0f9,00.html

[K.C. Chan]

All,
it still doesn't work, still getting the same error; I've tried compiling the scsi drivers into the kernel and having it in initrd. Even tried a fresh kernel(v2.4.18) from kernel.org.

Any idea? Thanks.

[K.C. Chan]

Does anyone know what is VFS, is it Veritas File System? This Kernel Panic is driving me up the wall.

[chloong]

hi, I had the same problem too...!
I compiling the kernel 2.4.18 and it give me kernel panic!!! the same error!
Is there any solution from HP?

[K.C. Chan]

on system trying to build kernel 2.4.18, I am getting:
VFS: Cannot open root device "803" or 008:03
Please append a correct "root=" boot option
Kernel Panic: VFS: unable to mount root fs on 08:03

On current kernel 2.4-2, I'm ok, here's the output from dmesg:
sym53c876-0: restart (scsi reset).
sym53c876-1: restart (scsi reset).
scsi0 : sym53c8xx - version 1.6b
scsi1 : sym53c8xx - version 1.6b
Attached scsi disk sda at scsi0, channel 0, id 0, lun 0
Attached scsi disk sdb at scsi0, channel 0, id 1, lun 0
Attached scsi disk sdc at scsi0, channel 0, id 2, lun 0
Attached scsi disk sdd at scsi1, channel 0, id 0, lun 0
Attached scsi disk sde at scsi1, channel 0, id 1, lun 0
Attached scsi disk sdf at scsi1, channel 0, id 2, lun 0
Attached scsi disk sdg at scsi1, channel 0, id 3, lun 0
Attached scsi disk sdh at scsi1, channel 0, id 4, lun 0
Attached scsi disk sdi at scsi1, channel 0, id 5, lun 0
Attached scsi disk sdj at scsi1, channel 0, id 6, lun 0
scsi2 : Adaptec AIC7XXX EISA/VLB/PCI SCSI HBA DRIVER, Rev 6.1.7
(scsi2:A:0): 40.000MB/s transfers (20.000MHz, offset 31, 16bit)
(scsi2:A:1): 40.000MB/s transfers (20.000MHz, offset 31, 16bit)
(scsi2:A:2): 40.000MB/s transfers (20.000MHz, offset 31, 16bit)
(scsi2:A:3): 40.000MB/s transfers (20.000MHz, offset 31, 16bit)
(scsi2:A:4): 40.000MB/s transfers (20.000MHz, offset 31, 16bit)
(scsi2:A:5): 40.000MB/s transfers (20.000MHz, offset 31, 16bit)
(scsi2:A:6): 40.000MB/s transfers (20.000MHz, offset 31, 16bit)
scsi2:0:0:0: Tagged Queuing enabled. Depth 253
scsi2:0:1:0: Tagged Queuing enabled. Depth 253
scsi2:0:2:0: Tagged Queuing enabled. Depth 253
scsi2:0:3:0: Tagged Queuing enabled. Depth 253
scsi2:0:4:0: Tagged Queuing enabled. Depth 253
scsi2:0:5:0: Tagged Queuing enabled. Depth 253
scsi2:0:6:0: Tagged Queuing enabled. Depth 253
Attached scsi disk sdk at scsi2, channel 0, id 0, lun 0
Attached scsi disk sdl at scsi2, channel 0, id 1, lun 0
Attached scsi disk sdm at scsi2, channel 0, id 2, lun 0
Attached scsi disk sdn at scsi2, channel 0, id 3, lun 0
Attached scsi disk sdo at scsi2, channel 0, id 4, lun 0
Attached scsi disk sdp at scsi2, channel 0, id 5, lun 0
Attached scsi disk sdq at scsi2, channel 0, id 6, lun 0
LVM version 0.9.1_beta2 by Heinz Mauelshagen (18/01/2001)
lvm -- Driver successfully initialized
VFS: Mounted root (ext2 filesystem) readonly.

on other system with kernel 2.4.3-12, I get this from dmesg:
Kernel command line: auto BOOT_IMAGE=linux ro root=302 BOOT_FILE=/boot/vmlinuz-2.4.3-12
VFS: Mounted root (ext2 filesystem) readonly.

Is it the LVM that's puking or is it that newer kernel specifically requires you to specify "root=". If so, what's is "root=302", what kind of value is 302, is it hex? where do I input this value into lilo.conf? Thanks.

[Jeffrey Wong]

First, on the 2.4.3-12 kernel system, is it booting off of a ide drive? If I remember correctly, the root=302 means to use device 03:02 which is /dev/hda2. VFS stands for Virtual File System which is the generic file system interface that linux uses as a middle layer over the actual file system implementation. Back to the original problem, the kernel looks like it is trying to boot off of 08:03, which is correct, /dev/sda3. I think the problem is that /dev/sda3 is a ext2 partition (according to the 2.4.2 boot messages). You might not have ext2 support in your new kernel, and the initrd image has ext3, but not ext2 support. Try rebuilding the kernel/initrd with ext2 support and give it a try.

[K.C. Chan]

CONFIG_EXT2_FS is set to "y", according to ".config" file for current kernel I'm trying to build. So I don't think this is the case. Do you have any other suggestions. Thanks.

[K.C. Chan]

Just in case someone wants to help and needs to look at the .config file of the current kernel I'm trying to build, attached is the ".config" file. Any help or suggestion is welcome. Thanks.

[Jeffrey Wong]

Looking over your .config I think I see the problem. It looks like you are using LVM on your disk drives and have LVM support compiled as a module which the kernel is supposed to get from initrd. The problem is you don't have initrd support turned on in the .config file so the kernel can't see it. Try turning on CONFIG_BLK_DEV_RAM and CONFIG_BLK_DEV_INITRD in .config and see what happens.

[K.C. Chan]

Wong, attached is new .config file. I've added your suggestion in. I ran make dep && make bzImage, then make modules && make modules_install. I moved the kernel and initrd-2.2.18.img into place, but still getting the same error during boot up. What gives????. Have you ran into this problem before. Thanks.

[K.C. Chan]

Sorry, here's the attachment and output of lspci:
-[00]-+-00.0 Intel Corporation 440BX/ZX - 82443BX/ZX Host bridge (AGP disabled)
+-06.0 Symbios Logic Inc. (formerly NCR) 53c875
+-06.1 Symbios Logic Inc. (formerly NCR) 53c875
+-07.0 Compaq Computer Corporation ProLiant Integrated Netelligent 10/100
+-08.0 ATI Technologies Inc 3D Rage IIC 215IIC [Mach64 GT IIC]
+-09.0 Compaq Computer Corporation Advanced System Management Controller
+-0a.0-[01]----08.0 Lite-On Communications Inc LNE100TX
+-0d.0 Adaptec 7892A
+-14.0 Intel Corporation 82371AB PIIX4 ISA
+-14.1 Intel Corporation 82371AB PIIX4 IDE
+-14.2 Intel Corporation 82371AB PIIX4 USB
\-14.3 Intel Corporation 82371AB PIIX4 ACPI

Can someone help in getting kernel 2.4.18 to boot on this proliant box. Thanks.

[K.C. Chan]

here's additional info from module.conf file:
#alias scsi_hostadapter sym53c8xx
#alias scsi_hostadapter1 aic7xxx
# alias eth1 tlan
# alias scsi_hostadapter2 aic7xxx
alias block-major-58 lvm-mod
alias char-major-109 lvm-mod
alias parport_lowlevel parport_pc
alias eth0 tulip
# alias eth0 tlan
# options tlan irq=

Notice lvm_mod is listed here, even though it's compile into the kernel. When I check to see if the modules is in /lib/modules/2.4.18/kernel/fs there isn't any module related to lvm_mod. What gives? I know there's shouldn't be any lvm_mod in module.conf bec. the current kernel also has lvm compiled into the kernel. Any suggestions?

[Jeffrey Wong]

Are you using LVM by itself, or are you using software RAID also? Right now, the only thing I can think of is you are using RAID and the kernel can't load the raid modules which are not on the initrd you are using. Other than that I'm not sure what to do.

[K.C. Chan]

I am using lvm by itself, But I've just tried booting without LVM by changing /etc/rc.d/rc.sysinit, comenting out vgchange and filesystem in /etc/fstab which are related to lvm. Still did not work. Getting something a little different this time, "VFS: mounted root (ext2 filesystem). Red Hat nash version 3.0.10 starting. VFS: cannot open root ...
Kernel panic....." Any idea?

[K.C. Chan]

All, I've managed to get it to work under kernel version 2.4.4 but not 2.4.8 and upward. The only thing I could think of is that ext3 is not yet available for 2.4.4. But it doesn't make since when I tried this under 2.4.8 and upward I've deselected ext3 and the kernel crap out on me during reboot. I am beginning to think that this is bug. Any one have any idea? Thanks.

[Jarle Bjorgeengen]

Hi Michael,

I recently had a similar problem.

I installed a FC adapter, to see som XP512 LUN's. Installed the driver, and made a initrd. After this, the machine kept panicing because it couldn't find the root device /dev/sda[x],

I found out that it was because the drivers are loaded in alfabetical order, hence the fc driver , that came alfabetically before the, megaraid driver (which the root disk used), was loaded first, and the FC-devices was named /dev/sda to /dev/sdj. The root partition then got the name /dev/sdh, which is sometihng else than what the bootloader says (/dev/sda)

The solution was to rename the .o file of the fc-driver in the /lib/modules/[kernel-version] directory, and likewise in the modules.conf file. And then finally make a new initrd.

Rgds Jarle

[K.C. Chan]

Sorry, ext3 isn't available yet in 2.4.8. So there's only one ext2 modules and it's compile in the kernel; I don't see that there's a driver conflict in my part. However, I am still having problems booting up with 2.4.8; I eventually wants to boot up with 2.4.18. Does any one knows what is causing this VFS: unable to mount root crap err!!!? Thanks.

[Jeffrey Wong · Accepted solution]

What order is the scsi drivers loading in? In the successfuly 2.4.2 boot, the kernel is using sym53c8xx as scsi0, does the 2.4.8 and 2.4.18 load the scsi drivers in the same order? If it's loading the adaptec driver first, then it will try to boot off the wrong drive.

[K.C. Chan]

Wong, Pingo!!!!! you hit the jack-pot. In the kernel 2.4.4 (current working one), the order of the scsi-driver is loaded correctly in order:
scsi0 : sym53c8xx-1.7.3a-20010304
scsi1 : sym53c8xx-1.7.3a-20010304
scsi2 : Adaptec AIC7XXX EISA/VLB/PCI SCSI ....
where as 2.4.8 and 2.4.18, the order is reverse, it's Adaptec then sym53c8xx.

Now that I know the cause of the problem, how I do I fixed it. How do I make sure that sym53c8xx gets loaded first. Thanks.

[K.C. Chan]

By the way both driver are compiled into the kernel. How do you specified the order of driver being loaded during boot up? Thanks.

[Jeffrey Wong]

The only way I know of to force the kernel to change the driver order would be to actually modify the host adapter detection code in the kernel. What I would try next in your case is to have the sym53c8xx driver compiled in the kernel so that gets loaded first. Then have the adaptec driver as a loadable module. They may be an append line you could add to lilo.conf to force the kernel to load the drivers in the order you want without using modules but I'm not sure what it would be.