kernel panic - 2.6
-pbh-
using scsi instead of ide can cause so much trouble, but here goes:
hda: cd/rom
sda: scsi-disk (using aic7xxx)
- 1st partition: /boot (256MB - ext2)
- 2nd partition: swap
- 3rd partition: / (10 gig - reiserfs)
Ok - I've got more, but that's not interesting
Installed slackware-10.2 with all updates, using grub and 'my own'
kernel (2.4.32) mostly based on slackware-2.4.31 but omitting some
stuff. Also using grub as loader - it's on the MBR (I think, might be
sda1 ...). I am using an initrd which basically only contains 'aic7xxx'
which is the driver for the onboard scsi - it all works a treat!!
Enter 2.6.13 - which I install and dont really expect to make it work -
too many 'dependencies' in the module department!! Download 2.6.15.4
kernel and doctor it the same way as I did with the 2.4.32. Of
particular interest here is 'reiserfs' in the kernel and also
scsi_transport, scsi disk, scsi cd and scsi-generic. Most other stuff
are modules ie. I only really keep what I need for booting other than
the driver itself, then use initrd for that.
My two grub entries are exactly similar (within reason):
title slackware-10.2 (2.4.32-A)
root (hd0,0)
kernel /vmlinuz-2.4.32-A ro root=/dev/sda3 vga=771
initrd /initrd-2.4.32-A.img
title slackware-10.2 (2.6.15-A)
root (hd0,0)
kernel /vmlinuz-2.6.15-A ro root=/dev/sda3 vga=771
initrd /initrd-2.6.15-A.img
When I boot this sucker, it comes up allright, loads the aic7xxx from
initrd and then gives me the following (please note that _all_
scsi-tstuff and reiserfs and ext2 are _in_ the kernel):
scsi1:A:0:0: Tagged Queuing enabled. Depth 4
target1:0:0: Beginning Domain Validation
target1:0:0: wide asynchronous
target1:0:0: FAST-20 WIDE SCSI 40.0 MB/s ST (50ms, offset 63)
target1:0:0: Domain Validation skipping write tests
target1:0:0: Ending Domain Validation
SCSI device sda: 286749488 512-byte hdwr sectors (146816 MB)
sda: sda1 sda2 sda3 sda4 < sda5 sda6 sda7 >
sd 1:0:0:0: Attached scsi disk sda
sd 1:0:0:0: Attached scsi generic sg0 type 0
VFS: Cannot open root device in "sda3" or unknown-block(0,0)
Please append a correct "root=" boot option
Kernel panic - not syncing: VFS: Unable to mount root fs on
unknown-block(0,0)
Yeah - I can read what it's telling me - it can't mount /dev/sda3 even
though the 2.4 kernel has no problems doing exactly the same - and - the
kernel _should_ contain everything but the adaptec driver ...
What am I missing here???
Thanks,
-pbh-
Douglas Mayne
WAG: modprobe exportfs (add to initrd, the module and the instuction to
load it).
If that doesn't fix it, then try to boot with P.V. 2.6.15.3 kernel to show
it should work.
--
Douglas Mayne
Loki Harfagr
<snipped for clarity>
> Yeah - I can read what it's telling me - it can't mount /dev/sda3 even
> though the 2.4 kernel has no problems doing exactly the same - and - the
> kernel _should_ contain everything but the adaptec driver ...
> What am I missing here???
The reiserfs module (or in kernel), did you go initrd and missed the
mkinitrd, and/or just didn't include it in the kernel ?
As your / is in reiserfs you need it at the start :-)
Loki Harfagr
Well, that could be it too, still I have the feeling the mkinitrd
with reiserfs would be better, I've had trouble with it in the kernel
and the scsi in initrd, don't ask me why !
> If that doesn't fix it, then try to boot with P.V. 2.6.15.3 kernel to
> show it should work.
That'd be a good preliminary test !
-pbh-
want kernels and modules that I can use on more than one distribution.
It still smarts that the initrd's are soooo different - but I guess I
can live with that. Ideally, what I would like would be to have the same
vmlinuz and modules _and_ initrd irrespective of the distribution I am
running ...
So all those extra partitions on my scsi-disk - hehe, 10 gigs for each
of archlinux, redhat-7.3 (yeah sorry, but I need it!) and whatever else
I may fancy.
So yes - I did a
mkinitrd -c -k 2.6.15-A -m aic7xxx -o /boot/initrd-2.6.15-A.img
Thanks for your input - I guess trying out the 'current' kernel would be
a way ... though ... why does this not work???
-pbh-
-pbh-
Oh and btw - just to make sure there were no hiccups, I also made
reiserfs a module and did a:
mkinitrd -c -k 2.6.15-A -m aic7xxx:reiserfs -o /boot/initrd-2.6.15-A.img
Still the same problem.
I guess I _could_ put aic7xxx into the kernel as well, but I just plain
hate doing that - next week I may be on a completely different machine -
maybe different scsi-adapter, maybe ide, maybe ... whatever.
To me - the problem with slackware's mkinitrd is that it loads the
modules 'verbatim', using insmod rather than modprobe - if you miss out
just _one_ module in the whole sheebang - you're screwed!! Redhat (bless
their cotton-drawers) really went to town with their mkinitrd - there
was not much that it missed. (but that was yesteryear - I dont touch RH
anymore)
Thanks for listening to me ranting ...
-pbh-
Douglas Mayne
> -pbh- wrote:
<snip>
>
Did you not see my response:
http://groups.google.com/group/alt.os.linux.slackware/msg/88589a2f392252b7
--
Douglas Mayne
-pbh-
Unfortuneately I am now physically removed from that particular
computer, so has to wait untill Monday to try out what you suggested.
Your input _is_ appreciated!
-pbh-
Loki Harfagr
That's quite odd for sure (I read your other posts too :-)
Now, we'll see again when you'll have tested with the exportfs
and tried the PV current 2.6 kernel.
My best best would be that you used your 2.4.* .config for a
start on 'make oldconfig' and didn't check all the possible turmoils
before the 'make "et all"' and I can swear there are plenty of !
I'm not saying that you don't know what you"re doing while menuconfig
but I really had hard times on some points in stuff like
SCSI,USB,IEE1394,network,memory,Audio,HID,realtime ...
though I was pretty confident on what I was doing I really had some
funny things happening, then I started back to the point of not using my
previous tests .config but from the PV source/testing/K/.config and
make it then slowly reintroduce every idiosyncrasy I needed.
It cost me a half day but it went finally fine. And when I made a
'diff' between my previous .config and the one working there were a
few differences, of course, but not many, and not in areas that I would
have believe to be involved in my troubles ...
Now, I'm just a system and languages engineer, that doesn't turn me on
to be a king of the kernel, especially while building for specific
hardware a kernel that's still in the unstable branch.
Keep on trying, and please tell us what was the problem, any old time
you find out !-)
-pbh-
When the 2.6-kernels first came out - I tried to migrate to them from a
2.4-config ... bad idea!!
So - my template for 2.4.32 was PV's 2.4.31-config and my template for
2.6.15 was PV's 2.6.13-config. Basically I want to keep most of the
stuff as modules - but some things I want in the kernel - especially the
scsi-stuff which can be awkward at times. What I have removed are things
like radio/hamradio, broadcasting, tv-stuff ... things I know that I
will never need.
The dang thing is - I have a 2.6.15-kernel running beautifully on
slackware 10.2 - but - no scsi-disks. Hence my initial 'sigh' over the
troubles you can get into as soon as you stray from the straight and
narrow (ide in this case).
Again - thanks for your comments!
-pbh-
Douglas Mayne
My WAG is wrong. exportfs is not required for reiserfs. I was testing
something else, and I noticed this after mounting a reiserfs partition.
From dmesg:
ReiserFS: hda7: found reiserfs format "3.6" with standard journal
ReiserFS: hda7: using ordered data mode
ReiserFS: hda7: journal params: device hda7, size 8192,
journal first block 18, max trans len 1024, max batch 900,
max commit age 30, max trans age 30
ReiserFS: hda7: checking transaction log (hda7)
ReiserFS: hda7: Using r5 hash to sort names
From lsmod:
Module Size Used by
reiserfs 239088 1
ipv6 230528 8
pcmcia 30756 0
firmware_class 7552 1 pcmcia
yenta_socket 24076 0
rsrc_nonstatic 11008 1 yenta_socket
pcmcia_core 34192 3 pcmcia,yenta_socket,rsrc_nonstatic
agpgart 28360 0
psmouse 32644 0
This is a 2.6.15.3 kernel compiled with P.V.'s configuration. Maybe,
run a diff comparision of your .config against that.
--
Douglas Mayne
-pbh-
I really could not see what exportfs ever had to do with reiserfs - and
I know that I have previously just had to add in reiserfs ...
I believe the problem is in the scsi-modules somewhere .... there must
be something missing ...
The first time around I had scsi_transport as modules and had forgotten
to include them ... got a lot of spi_xxx symbols not found.
Then I put them into the kernel and got the result initially reported.
There is more to scsi than meets the eye ....
But thanks again - I do appreciate that yall are trying to help me out!
-pbh-
-pbh-
> I'm no nubee, but this has me stymied. It never cease to amaze me that
> using scsi instead of ide can cause so much trouble, but here goes:
If anyone's interested - the dang thing has me beat!!
I recompile the kernel - and only make _one_ change - putting aic7xxx
into the kernel rather than a module - and it all works great!! When I
move it back as a module and use mkinitrd with aic7xxx as a sole module
(its all I should need) - the symptoms are as initially described!
I give up - into the kernel it goes!
P
-pbh-
> ---clip
and by the way - Douglas Mayne - when I try Patricks 2.6.15.6 and load
the neccessary modules (through initrd) - I get the same error as
initially described!!
-pbh-
Douglas Mayne
This may just be academic at this point, because you have fixed it with a
"bigger hammer," but I have another WAG.
Does it change depending on which module is loaded first. It could be that
the ReiserFS modules attaches to hardware devices available when
the ReiserFS modules is loaded. Try loading aic7xxx first, followed by
ReiserFS.
--
Douglas Mayne
Loki Harfagr
Okay, so it's like we both said always as b0rken and versatile
as always ;-)
Remember I told you to try the "full kernel-inside way" and also the
"full initrd way" ? My last settings from 2.6.11 made me move to this
initrd :
mkinitrd -c -k 2.6.15.6 -m 8139too:appletalk:eth1394:evdev:exportfs:ieee1394:ip_conntrack:iptable_filter:ip_tables:ipt_LOG:ipt_state:ipv6:ipx:joydev:loop:loop_blowfish:lp:mii:nfsd:nls_cp850:ntfs:ohci1394:parport:parport_pc:pcmcia:pcmcia_core:sbp2:snd:soundcore:yenta_socket
Up to 2.6.10, all the reiser, scuzzy et al, stuff was initrd too but
when ...11 arrived I had to twist my fingers down to the bone to make it
work without thrashing my partitions any old time it "booted" too fast.
Now, as I've got my usb-audio UA-20 card silent though the system sees
it allright, I'll go up putting back the 'soundcore' into the kernel.
Makes me wonder ... Ooooh oh ooooh ...
Keep on truckin' !
-pbh-
This makes some kind of logic - I shall try it out next (well, sorry,
but it will be Monday before I have a chance again).
I'll keep you guys informed!!
And again - thanks for your input!!
-pbh
-pbh-
mainframes and punched cards ... (yup - I've been in the business _that_
long!) If you found a deck with JCL-commands that actually _worked_ -
you made 10 copies of it and hid them in every conceivable (and
unconceivable) place!
Thanks guys - I love some humour in all the despair!!
-pbh-
Ronald Matthews
There is no reason to use modules when ram is so cheap. Just
compile the drivers you need into the kernel and be done with it.
Less muss and less fuss.
cordially, as always,
rm
--
How come so many stat fans, and pseudo stat fans, are Jewish?
Ronald Matthews
> *chuckles* somehow it reminds me of the 'good ole days' with JCL
> and IBM mainframes and punched cards ... (yup - I've been in the
> business _that_ long!) If you found a deck with JCL-commands that
> actually _worked_ - you made 10 copies of it and hid them in every
> conceivable (and unconceivable) place!
Remember those huge stacks of cobol programs? And PL/1? I remember
bumping into Chinese kids at U of T...
cordially, as always,
rm