Kernel upgrade problem.

[massai]

Hello everyone,

The problem is about attempting to upgrade the kernel from 2.4.42 to 2.6.1.
I followed the steps as illustrated here (full source method)
http://linux-newbie.sunsite.dk/lnag_kernel_upgrade.html

I am must have messed something during the configuration step, because when I try to reboot to the new kernel the process goes through LILO just fine but then it hungs immidiately after that; where it is supposed to be doing "check BIOS data" (?).

I booted with the old kernel and recompiled one more time using the old kernels "config" file, but the same thing happens.

Has anyone seen this before? How should I go about troubleshooting?

Thanks

[GulGarak]

I have the same problems
But one thing is that the kernel 2.6. does not support vesa vga mode, but this is chosen by lilo vga=...
You can disable it by vga=0 in lilo.conf
By this way i got the 2.4.22-K7 kernel from debian run. but i was not able to run 2.6.0 at all.

See Kernel 2.6.0 and links in this Forum or some postings in the HD-install-forum on this topic.

If you get the answer let me know too.

garak
s[/quote]

[Stephen]

Hello everyone,

The problem is about attempting to upgrade the kernel from 2.4.42 to 2.6.1.
I followed the steps as illustrated here (full source method)
http://linux-newbie.sunsite.dk/lnag_kernel_upgrade.html

I am must have messed something during the configuration step, because when I try to reboot to the new kernel the process goes through LILO just fine but then it hungs immidiately after that; where it is supposed to be doing "check BIOS data" (?).

I booted with the old kernel and recompiled one more time using the old kernels "config" file, but the same thing happens.

Has anyone seen this before? How should I go about troubleshooting?

Thanks

Firstly have you installled the module-init-tools program it is needed to boot a 2.6 kernel and you need a line like this in your /etc/fstab.


## Needed for kernel 2.6
none /sys sysfs defaults 0 0

If you want to check a working config for a 2.6 kernel then you can find my old one I used for test9 here (http://www3.ns.sympatico.ca/scormier). You will have to make changes for your setup it has support for ext2, ext3 and reisferfs support built in so there is no need for an initrd if you use any of these file systems for your / and if you want the framebuffer when you boot then do not make any change to the console framebuffer devices or for the built in fonts.

[massai]

Stephen, thank you for offering a helping hand.

I took you suggestions to heart and went back did the menuconfig (one more time) line by line.
My machine is a little different than yours (so I changed appropriately)

In regards to "module-init-tools" your hint forced me to revamp my sources.list file (previously it was giving me a trackload of 404 errors; so now it all cleaned-up and a new "update and upgrade" has been made). So the "module-init-tools" have been re-installed.

Your suggested line was added to the /etc/fstab.

Someone suggested that I should "make install" after the "Make bzImage". So I did that too (this also took care of the vmlinuz, System.map links automatically).

Then I rebooted....
but...
this time it passed the BIOS data check, but it still froze with the black screen of death.

Then I dug deaper and found out that the new lilo stanza (which I made previously) contained the "/boot/initrd.gz" line (as it was copied from the original knoppix stanza).

I do not know how to make a new initrd.2.6.1(??) thingy. On the other hand it may not be needed (I do have ext2 ext3 enabled via menuconifig)
Anyway, someone suggested to use mkinitrd but that wasn't working for me.
Someone else suggested that debian doesn't have mkinitrd and we should use the following steps within /usr/src/linux-2.6.1

*****
make-kpkg clean
make-kpkg --revision 5:myhost.1.0 kernel_image
dpkg -i ../kernel-image*.deb
*****

The above three steps compiled fine (it did a lot of automatic stuff including make modules, make modules_install, took care of links, made a boot disc, and some other stuff that I didn't know before.

I opened lilo.conf and hashed out the initrd.gz line.

rebooted.

This time it seemed to boot normaly (harrraay, but only for a minute).
It goes through the entire boot process (I see all the text scrolling etc..etc) but the window manager never loads. No Nvidia splash screens... nothing.
Instead, I am stuck in text mode.
Now, when I login it greats me with: "Welcome massai Linux kernel 2.4.22-xfs" !!

All this time I thought it was booting into 2.6.1 ?

Something is still broken. No idea what it is though.

Now I am confused beyond repair... need time to recover...
Any further suggestions would be appreciated.

Many Thanks

P.S. I can still boot to the old 2.4.22-xfs kernel via LILO and it boots fine (via Nvidia splash screen and KDE)

[Stephen]

Can we see the output of ll /vmlinuz*, ll /boot/* and the contents of the /etc/lilo.conf it almost looks like it did not run lilo after the install or you did not after changing the file. For the initrd you do not need one at all if you compile support for your /root file system into the kernel. Also I notice the commands you are using are you doing the compiling as root? If so you should start using the fakeroot package to compile your kernels as a normal user then use su to become root and install the new kernel. The commands would look like this:



make menuconfig
fakeroot make-kpkg clean
fakeroot make-kpkg --append-to-version=.version_number kernel_image kernel_headers


For your Nvidia driver you need to get the source and the patch for it to compile a new driver for the 2.6 kernel you can do that with Kano's (http://www.kano.mipooh.net/) script to install the driver after the hd install I would recommend the 4496 one. You would download the script then change to the directory downloaded too and chmod +x name_of_script then ./name_of_script to run this step best done at a console login as root. BTW I posted my config for 2.6.1 so it's there if you want to take a look oh and a couple of guides to compiling a kernel the Debian way if you have not seen them already.

http://newbiedoc.sourceforge.net/tutorials/kernel-pkg/index-kernel-pkg.html
http://www.osnews.com/story.php?news_id=2949

[massai]

-In regards to the initrd (and ext2, ext3 support). Here is what my .config looks like:
I am not sure if this means whether I have "compiled support for the /root filesystem into the kernel", but I suspect it should ? No?



# File systems
#
CONFIG_EXT2_FS=y
# CONFIG_EXT2_FS_XATTR is not set
CONFIG_EXT3_FS=y
# CONFIG_EXT3_FS_XATTR is not set
CONFIG_JBD=y
CONFIG_JBD_DEBUG=y
CONFIG_REISERFS_FS=y
# CONFIG_REISERFS_CHECK is not set
# CONFIG_REISERFS_PROC_INFO is not set
# CONFIG_JFS_FS is not set
CONFIG_XFS_FS=m
CONFIG_XFS_RT=y
CONFIG_XFS_QUOTA=y
CONFIG_XFS_POSIX_ACL=y
# CONFIG_MINIX_FS is not set
# CONFIG_ROMFS_FS is not set
# CONFIG_QUOTA is not set
CONFIG_QUOTACTL=y
# CONFIG_AUTOFS_FS is not set
CONFIG_AUTOFS4_FS=m


-In regards to lilo.conf.
This is what the file looks like:



boot=/dev/hdb1
compact
install=/boot/boot.b
map=/boot/map
vga=791
prompt
timeout=30
#append = "hda=scsi hdb=scsi hdc=scsi hdd=scsi hde=scsi hdf=scsi hdg=scsi hdh=scsi apm=power-off nomce"
append = "apm=power-off nomce"

#Stanza1
image=/boot/vmlinuz-2.6.1
label=Kernel(2.6.1)
root=/dev/hdb1
read-only
# initrd=/boot/initrd.gz

#Stanza2
image=/boot/vmlinuz-2.4.22-xfs
label=Kernel(2.4.22)
root=/dev/hdb1
read-only
initrd=/boot/initrd.gz

#Stanza3
# A Windows installation, example for partition 1 on first IDE harddisk
# other=/dev/hda1
# label=Windows2000
# optional
# table=/dev/hda


I am not sure what you mean by "output of ll /vmlinuz*, ll /boot/* ".
It is probably something basic but my mind is stuck right now.
This is the /boot directory listing, but I suspect this is not what you are asking



/boot:
used 3652 available 651528
drwxr-xr-x 2 root root 4096 Jan 21 17:17 .
drwxr-xr-x 29 root root 4096 Jan 21 16:59 ..
lrwxrwxrwx 1 root root 16 Jan 21 16:16 System.map -> System.map-2.6.1
-rw-r--r-- 1 root root 541726 Oct 3 21:01 System.map-2.4.22-xfs
-rw-r--r-- 1 root root 665947 Jan 21 16:54 System.map-2.6.1
lrwxrwxrwx 1 root root 21 Jan 21 16:08 System.map.old -> System.map-2.4.22-xfs
-rw-r--r-- 1 root root 512 Dec 19 23:57 boot.0341
lrwxrwxrwx 1 root root 12 Jan 21 16:16 config -> config-2.6.1
-rw-r--r-- 1 root root 38746 Oct 3 20:36 config-2.4.22-xfs
-rw-r--r-- 1 root root 31323 Jan 21 16:31 config-2.6.1
lrwxrwxrwx 1 root root 17 Jan 21 16:09 config.old -> config-2.4.22-xfs
lrwxrwxrwx 1 root root 34 Jan 3 18:49 debian.bmp -> /usr/share/lilo/contrib/debian.bmp
lrwxrwxrwx 1 root root 38 Jan 3 18:49 debianlilo.bmp -> /usr/share/lilo/contrib/debianlilo.bmp
-rw-r--r-- 1 root root 148091 Dec 19 23:57 initrd.gz
-rw------- 1 root root 15872 Jan 21 17:05 map
lrwxrwxrwx 1 root root 33 Jan 3 18:49 sarge.bmp -> /usr/share/lilo/contrib/sarge.bmp
lrwxrwxrwx 1 root root 31 Jan 3 18:49 sid.bmp -> /usr/share/lilo/contrib/sid.bmp
lrwxrwxrwx 1 root root 13 Jan 21 16:16 vmlinuz -> vmlinuz-2.6.1
-rw-r--r-- 1 root root 921593 Oct 3 21:01 vmlinuz-2.4.22-xfs
-rw-r--r-- 1 root root 1327269 Jan 21 16:54 vmlinuz-2.6.1
lrwxrwxrwx 1 root root 18 Jan 21 16:10 vmlinuz.old -> vmlinuz-2.4.22-xfs


I will worry about the Nvidia driver later. I thought that the new kernel will be transplanted and pick-up on the existing nvidia driver, but it looks like (as you hinted) it must be done from scratch.

Thanks for the new links, (no I didn't know about them) I will take a look. Yeesh, I wish there was an up to date centralized newbie documentation.
The one that I was going by http://linux-newbie.sunsite.dk/lnag_kernel_upgrade.html was dated back to 1999. On the same token, I had to fish all over the net to find a good sources.list....and still don't know whether I got it right (all these things should be in one place. No ?)

[massai]

On the link that Stephen posted "Osnews, Very Verbose Guide" at one point the author is using:



make-kpkg clean
make-kpkg --revision=786:MyKernel2.4.20 kernel_image


What does the number 786 mean ? Is this a processor thing ?

[Stephen]

All that looks good except when you boot you get to see the lilo screen and can select the Kernel(2.6.1) right? I ask this because you have boot=/dev/hdb1 which means lilo gets installed to the / partition instead of the MBR so if you are using a lilo that was installed in the MBR then this lilo.conf is not getting used. So if you get to boot properly with the lilo entry and then when it boots you only get the console where you get the message Welcome to .... 2.4.22-xfs this looks like the message of the day (MOTD) is set wrong try uname -a after you have logged in at the console I believe that you are at this point running the 2.6.1 kernel otherwise the Nvidia screen would come up you would get the KDE greeter screen and be able to login graphically like you do when you use the lilo screen entry for the old kernel.

The revision is similar to the --append-to-version I think but I'm not sure I never use it I always use the append this tells the kernel package program that you are compiling a different version number of the kernel so that when you install it will not overwrite an existing kernel you have installed the number used in either case it is a personal preference for instance when I compile I always use the year, month and day for the number eg. --append-to-version=.040121 for a kernel compiled today if I compile a second one to install then I would put a -2 on the end for the second kernel.

[massai]

when you boot you get to see the lilo screen and can select the Kernel(2.6.1) right?

You are on the ball !!
This is exactly what happens. I get a lilo screen (btw it is a red window) which times out in 30ms (as seen in the conf file) and I get to choose between the stanzas.

When I choose the kernel2.6.1 stanza a lot of text flies by (as it did with the 2.4.22 kernel) but in the end it stops in a login prompt (KDE does not start and off course there is no Nvidia splash screen).
Then when I login it greets me as kernel 2.4.22? And this confused the heck out of me.

If as you say it is just the "Message of the day" then it would be relatively good news.
(sigh... all these details are killing me... I would have never thought of something as innocent as MOTD)

I will dig deeper.

Right now it also bugs me that xconfig won't start (it never did since day one actually and that is why I have been using menuconfig). It complains about "unable to find QT installation... bla..bla..bla" I tried to adjust the QTDIR to /etc/lib/qt3 but didn't work.

Thanks about the clarification on the "revision" option.

P.S. Right now I am in the process of double-checking the .config file (based on your 2.6.1 listing as well as the suggestions on the OS news article) to make sure I don't setup more traps for my poor newbie soul later on.

[massai]

Right now it also bugs me that xconfig won't start (it never did since day one actually and that is why I have been using menuconfig). It complains about "unable to find QT installation... bla..bla..bla" I tried to adjust the QTDIR to /etc/lib/qt3 but didn't work.


Please disregard the above,
After reinstaling QT the xconfig seems to work ok now.

[Stephen]

Right now it also bugs me that xconfig won't start (it never did since day one actually and that is why I have been using menuconfig). It complains about "unable to find QT installation... bla..bla..bla" I tried to adjust the QTDIR to /etc/lib/qt3 but didn't work.


Please disregard the above,
After reinstaling QT the xconfig seems to work ok now.

Good to hear you got it going you should change the message of the day file ( /etc/motd ) to the proper settings to avoid anymore confusion what it should look like for the format that was being used incorrectly.


Welcome to $HOSTNAME (Kernel $(uname -r))

And you can undo the changes in your XF86Config-4 file and go back to using the "nv" driver until you compile your nvidia driver so you can login graphically also if you want to get rid of that nvidia logo when you get it going put this underneath the driver line.


Driver "nvidia"
Option "NoLogo" "on"


Note: Now that I look at the motd line in the preview it looks like there is a space between the $ and the ( there is none.

[massai]

I did "uname -a" after the login prompt and true enough it is indeed kernel 2.6.1.
AWESOME !! (cheers...turned up the music... jumped up and down in joy and then opened a can beer and kicked back for five minutes...)

Man, you are on the ball ! Thank for hanging around with this newbie (that's me) for as long as you did !

Now... after all the excitment toned down, I am finding out that there is no network conection under the new kernel (2.6.1). (Shoot, I had a nagging feeling that something would spoil the fan...and true enough here it goes). It looks like another Kernel recompile is coming and I dread it.

The original knoppix kernel .config file had all networking options enabled (and the kitchen sink). I wish there was a way to pin it down to my hardware instead of having to resort to such a "shotgun" approach.

I am on a Lynk-sys 10-100 ethernet card, connected to a Lynk-sys router, connected to a Lynksys cable modem.

Off I go to compare the .config files again....


P.S. I notice that your .config file is much more compact than mine (That is probably because I chickened out and didn't dare trim as much fat as you did)

[Stephen]

I did "uname -a" after the login prompt and true enough it is indeed kernel 2.6.1.
AWESOME !! (cheers...turned up the music... jumped up and down in joy and then opened a can beer and kicked back for five minutes...)

Man, you are on the ball ! Thank for hanging around with this newbie (that's me) for as long as you did !

Now... after all the excitment toned down, I am finding out that there is no network conection under the new kernel (2.6.1). (Shoot, I had a nagging feeling that something would spoil the fan...and true enough here it goes). It looks like another Kernel recompile is coming and I dread it.

The original knoppix kernel .config file had all networking options enabled (and the kitchen sink). I wish there was a way to pin it down to my hardware instead of having to resort to such a "shotgun" approach.

I am on a Lynk-sys 10-100 ethernet card, connected to a Lynk-sys router, connected to a Lynksys cable modem.

Off I go to compare the .config files again....


P.S. I notice that your .config file is much more compact than mine (That is probably because I chickened out and didn't dare trim as much fat as you did)

If you are getting the IP by DHCP (where the problem usually happens) you may want to hold off on the recompile and try either noapic or acpi=off when you boot at the lilo screen. You would just start typing in the name that is displayed at the lilo screen and the boot option to pass in your case Kernel(2.6.1) noapic then enter key to boot if either of them work then you would add the parameter to the append= line in the lilo.conf to have it used every time on boot. To start to track down what you need in the kernel use your favorite seach engine to check on your componets in the system with a linux on the end of the search and the hardware how-to (http://www.tldp.org/HOWTO/Hardware-HOWTO/index.html) at the Linux Documentation Project (http://www.tldp.org) is always good to check as well.

[massai]

during boot there is a lot of text that flies by.

Is that text saved anywhere for later inspection ?

I would like to compare the boot processes between the two kernels.

.....
Also, the /usr/src/linux/arch/i386/config.in file is missing from both kernels ? Is this normal ?


......
Neither noapic or acpi=off worked

......
After checking the Lynksys website it looks like my ethernet card requires "Tulip" and that is the one thing I did *not* include in my kernel config.
Recompiling the kernel again (I guess I can use the practice)

.....
SUCCESS !!
After enabling the tulip networking support and recompiling the kernel I finaly have online access.


......
SUCCESS !!
The nvidia driver is up and running (and better than before) thanks to kano's script.
I opened up the script and noticed that there is a provision for kernel 2.6. Veeeerry nice (the script is up to date)

THANK YOU Stephen for *all* your help.
Your contribution/effort is appriciated.

[Stephen]

during boot there is a lot of text that flies by.

Is that text saved anywhere for later inspection ?

I would like to compare the boot processes between the two kernels.

.....
Also, the /usr/src/linux/arch/i386/config.in file is missing from both kernels ? Is this normal ?


......
Neither noapic or acpi=off worked

......
After checking the Lynksys website it looks like my ethernet card requires "Tulip" and that is the one thing I did *not* include in my kernel config.
Recompiling the kernel again (I guess I can use the practice)

.....
SUCCESS !!
After enabling the tulip networking support and recompiling the kernel I finaly have online access.


......
SUCCESS !!
The nvidia driver is up and running (and better than before) thanks to kano's script.
I opened up the script and noticed that there is a provision for kernel 2.6. Veeeerry nice (the script is up to date)

THANK YOU Stephen for *all* your help.
Your contribution/effort is appriciated.

Your welcome, the messages can be found by using dmesg | less in a console window for the current boot or in the file /var/log/messages for the boots since the last log rotation. Your right about the config.in being missing in 2.6 however it is there in my old 2.4.24 it looks like it has been changed in 2.6.? Kconfig, you do not have the source installed for the Knoppix kernel (or do you) so you can not find the old one.


>$ locate config.in
/usr/src/linux-2.4.24/arch/alpha/config.in
/usr/src/linux-2.4.24/arch/arm/config.in
/usr/src/linux-2.4.24/arch/cris/config.in
/usr/src/linux-2.4.24/arch/i386/config.in
/usr/src/linux-2.4.24/arch/ia64/config.in
/usr/src/linux-2.4.24/arch/m68k/config.in
/usr/src/linux-2.4.24/arch/mips/config.in
/usr/src/linux-2.4.24/arch/mips64/config.in
/usr/src/linux-2.4.24/arch/parisc/config.in
/usr/src/linux-2.4.24/arch/ppc/config.in
/usr/src/linux-2.4.24/arch/ppc64/config.in
/usr/src/linux-2.4.24/arch/s390/config.in
/usr/src/linux-2.4.24/arch/s390x/config.in
/usr/src/linux-2.4.24/arch/sh/config.in
/usr/src/linux-2.4.24/arch/sh64/config.in
/usr/src/linux-2.4.24/arch/sparc/config.in
/usr/src/linux-2.4.24/arch/sparc64/config.in
/usr/src/linux-2.4.24/arch/x86_64/config.in
/usr/src/modules/alsa-drivers/toplevel.config.in

[massai]

Thanks, I tried dmesg... and discovered...

what it looks like there is a hick-up: There is an obsolete "update daemon" running and causes an error message: "process 'update' used the obsolete bdflush system call, Fix your initscripts"

So, I dug deeper, and found out that in 2.6 the old "update daemon" is replaced by a new thread "pdflush" (instead of bdflush) .

What I have to do is somehow remove the command that invokes bdflush somewhere in the boot scripts. The problem is that I don't know how to do this just yet.

[arkaine23]

Should be a file in /etc/init.d... perhaps by the name of of update or updated?

[massai]

I don't think "update" is a file, plus I can't find it under /etc/init.d

The curious thing is why did it tell me to fix the initscripts? Does it mean I have to reinstall initscripts ?

It looks like initscripts depents on "mount" and "mount" cannot be updated because it conflicts with "mount-aes".

I suspect mount-aes is a knoppix custom thing ?

Both mount and mount-aes are discribing the same functions (as seen in their descriptions in Kpacage)

........
Couldn't find "mount-aes" in the debian package lists.
I will go ahead and replace it with "mount" testing (ver 2.12-3)
(I will overide the "skull and bones" warning and see what hapens)

.......
"Mount" has now been installed (no carnage occured dispite the spooky warning)
However.... initscripts still does not get fixed:



root@ThisKnoppixBox:/# apt-get install initscripts
Reading Package Lists... Done
Building Dependency Tree... Done
The following NEW packages will be installed:
initscripts
0 upgraded, 1 newly installed, 0 to remove and 20 not upgraded.
Need to get 0B/16.6kB of archives.
After unpacking 176kB of additional disk space will be used.
(Reading database ... 134106 files and directories currently installed.)
Unpacking initscripts (from .../initscripts_2.85-9_all.deb) ...
dpkg: error processing /var/cache/apt/archives/initscripts_2.85-9_all.deb (--unpack):
trying to overwrite `/etc/init.d/halt', which is also in package sysvinit
Errors were encountered while processing:
/var/cache/apt/archives/initscripts_2.85-9_all.deb
E: Sub-process /usr/bin/dpkg returned an error code (1)

[Stephen]

You may want to check this thread here (http://www.knoppix.net/forum/viewtopic.php?t=5759#26324) it m_yates describes how to get rid of the knoppix variants and get real the Debian files.

[massai]

Yup, it worked !
Hey Steff what would I do without your help ! (you didn't have to make extremely detailed instructions for me, but just pointing me to the right direction has helped imensely. Thanks again)

I followed the instructions of "m_yates" and I was able to reinstall initscripts:



OK, I tried it again on a different computer and I wrote the order of things wrong:

download sysvinit<something>.deb package.

dpkg -x sysvinit<something>.deb /tmp

apt-get install mount

"Yes, do as I say!"

apt-get install installscripts

That brings an error, override with:

dpkg -i --force-overwrite /var/cache/apt/archives/initscripts<something>.deb

cp /tmp/usr/share/sysvinit/update-rc.d /usr/sbin

apt-get install sysv-rc

dpkg -i sysvinit<something>.deb

Everything works using the standard sysvinit with initscripts. I replaced bootmisc.sh with the standard version as well and everything works fine.


Erata: where he said: apt-get install installscripts
he ment: apt-get install initscripts

I had a problem doing "apt-get install sysv-rc"
so I forced it with "dpkg -i --force-overwrite ........"

Rebooted and did "dmesg" and the error message ("process 'update' used the obsolete bdflush system call, Fix your initscripts" ) is not longer there ! Woohooo !

..........
In retrospect, it was imperative that I completed this step after the kernel upgrade because the system call to the obsolete bdflush had caused several of the kernel modules to remain unloaded and some devices were left undetected.

Now (after inspecting the boot messages) the boot process seems to flow much more smoothly.

...........
This whole upgrade process is a bitch (forgive the expression). If I wanted Debian, I should have gotten Debian in the first place and be done with it. (However, the learning experience from this silly knoppix-to-debian exercise is the only good asset and it compensates for the effort.... or at least I hope so...)

[Stephen]

This whole upgrade process is a bitch (forgive the expression). If I wanted Debian, I should have gotten Debian in the first place and be done with it. (However, the learning experience from this silly knoppix-to-debian exercise is the only good asset and it compensates for the effort.... or at least I hope so...)

Definitely gets you knowing the system at the very least and Debian is worth the effort most things just work like they should. A couple of programs you should look at installing just in case they don't are apt-listchanges and apt-listbugs which show you a screen for each during your upgrading that lists the changes and bugs with the packges, the list bugs one can be misleading sometimes so it is always a good idea to go to http://bugs.debian.org and check the reports some people seem to add a comment to bugs that are months old and get them back into the system most times though it is just change logs that you get to see. You may want to go to http://lists.debian.org and subscribe to the debian-security-announce mailing list as well so you get notified anytime a new package comes out due to a problem you might if you not mind about 200-250 emails a day want to try debian-user although you might want to use a newsreader to check it out there's lots of good information on that list.

[massai]

Thanks Steff, I bookmarked your links.

It is probably a good idea (as you suggested) to linger around the debian forums as well.

[massai]

More bad lack.
(It seems that every single troubleshooting scenario that can happen will happen , but I will try to keep the spirits up and look at the positive side. )

Anyway, just found out that mounting and filisystems are messed up.

There used to be a swap partition under the previous kernel. Now it is nowhere to be found.
Also, floppies cannot be mounted, CD roms cannot be mounted. Everything gives an error of some sort.

After inspecting the dmesg entries, it looks like all the drives and partisions have been recognized by the kernel during boot but things have changed and the old fstab needs to be revamped.

What is the best way to do this ?

This is a grab of the relevant portion of dmesg entries:


hdb: QUANTUM FIREBALL SE4.3A, ATA DISK drive
ide0 at 0x1f0-0x1f7,0x3f6 on irq 14
hdc: Pioneer DVD-ROM ATAPIModel DVD-106S 012, ATAPI CD/DVD-ROM drive
hdd: PLEXTOR CD-R PX-W1610A, ATAPI CD/DVD-ROM drive
ide1 at 0x170-0x177,0x376 on irq 15
hdb: max request size: 128KiB
hdb: 8418816 sectors (4310 MB) w/80KiB Cache, CHS=14848/9/63
hdb: hdb1 hdb2 < hdb5 >
ide-cd: passing drive hdc to ide-scsi emulation.
ide-cd: passing drive hdd to ide-scsi emulation.
ide-scsi is deprecated for cd burning! Use ide-cd and give dev=/dev/hdX as device
scsi0 : SCSI host adapter emulation for IDE ATAPI devices
Vendor: PIONEER Model: DVD-ROM DVD-106 Rev: 1.22
Type: CD-ROM ANSI SCSI revision: 02
scsi1 : SCSI host adapter emulation for IDE ATAPI devices
Vendor: PLEXTOR Model: CD-R PX-W1610A Rev: 1.03
Type: CD-ROM ANSI SCSI revision: 02
sr0: scsi3-mmc drive: 15x/40x cd/rw xa/form2 cdda tray
Uniform CD-ROM driver Revision: 3.12
Attached scsi CD-ROM sr0 at scsi0, channel 0, id 0, lun 0
sr1: scsi3-mmc drive: 40x/40x writer cd/rw xa/form2 cdda tray
Attached scsi CD-ROM sr1 at scsi1, channel 0, id 0, lun 0


This is the current fstab file:


# /etc/fstab: filesystem table.
#
# filesystem mountpoint type options dump pass
/dev/hdb1 / ext3 defaults,errors=remount-ro 0 1
/dev/hdb2 none swap sw 0 0
/dev/fd0 /mnt/floppy ext2,ext3,vfat defaults,user,noauto,showexec,umask=022 0 0
/dev/cdrom /mnt/cdrom iso9660 defaults,ro,user,noexec,noauto 0 0
/dev/cdrom1 /mnt/cdrom1 iso9660 defaults,ro,user,noexec,noauto 0 0
/dev/dvd /mnt/dvd iso9660 defaults,ro,user,noexec,noauto 0 0
/dev/cdaudio /mnt/cdaudio iso9660 defaults,ro,user,noexec,noauto 0 0

none /sys sysfs defaults 0 0
proc /proc proc defaults 0 0
usbdevfs /proc/bus/usb usbdevfs defaults 0 0


So far the new world order looks like:
hdb1 (Main partition)
hdb2 (swap partition)
sr0 (DVD/CD)
sr1 (CD-rw)

..... I will dig deeper to see if I can figure this out...

[massai]

Update-01:
The floppy issue has been resolved. It is now possible to mount floppies.
(part of the problem was the disk I was testing seemed to have a bad superblock.... although I am not sure how that happened; that floppy was formated and written on during the kernel upgrade process by an automatic script........ I am inocent !..... I swear )

Update-02:
I can read vfat floppies, but when a floppy is formated with fdformat then it can not be mounted.
..........
Ok, got it.
Had to use "mkfs /dev/fd0" on the newly formated floppy.

Update-03:
Found relevant discussion http://www.knoppix.net/forum/viewtopic.php?t=6262&highlight=
That is also the case on my system; the user knoppix had been deleted, but the /etc/group file contained a lot of knoppix instances.

[Stephen]

I usually use the actual device that is being used for the CD/DVD Rom(s). So for the first device sr0 you would use the device it links too which is scd0 and sr1 -> scd1 in the /etc/fstab. Then you would check the sequence of the links for all the devices listed in the fstab to make sure they point to the correct one. eg. my DVD burner for instance:



[04:45 PM Sat Jan 24: stephen @ ~]
>$ ll /dev/cdrom
lrwxrwxrwx 1 root root 11 2004-01-24 16:05 /dev/cdrom -> /dev/cdrom0
[04:45 PM Sat Jan 24: stephen @ ~]
>$ ll /dev/cdrom0
lrwxrwxrwx 1 root root 9 2004-01-24 16:05 /dev/cdrom0 -> /dev/scd0
[04:45 PM Sat Jan 24: stephen @ ~]
>$ ll /dev/dvd
lrwxrwxrwx 1 root root 9 2004-01-18 20:01 /dev/dvd -> /dev/scd0
[04:46 PM Sat Jan 24: stephen @ ~]
>$ ll /dev/sr0
lrwxrwxrwx 1 root root 4 2004-01-18 20:01 /dev/sr0 -> scd0

Entries in fstab

/dev/scd0 /cdrom iso9660 ro,user,noauto 0 0
/dev/scd0 /dvd iso9660 ro,user,noauto 0 0

For a seperate burner I usually make a directory /cdrw and use that as the mount point in addition to the /dvd I had to make in this instance the /cdrom already existed. And while I'm thinking about it did you make the /sys directory for the line that is being mounted for the sysfs in the file? So for your file I would use (if you wanted all the extra entries in it only one line per device is necessary):


/dev/scd0 /cdrom iso9660 defaults,ro,user,noexec,noauto 0 0
/dev/scd0 /dvd iso9660 defaults,ro,user,noexec,noauto 0 0
/dev/scd1 /cdrw iso9660 defaults,ro,user,noexec,noauto 0 0
/dev/scd0 /dev/dvd iso9660 defaults,ro,user,noexec,noauto 0 0
/dev/scd0 /dev/cdaudio iso9660 defaults,ro,user,noexec,noauto 0 0


The last line for the /dev/cdaudio assumes that the first drive has the audio cable running from it to the sound card and you would make sure the links are pointing to the /dev/scd0 from both the /dev/dvd & /dev/cdaudio. Then you would correct the icons for the devices on the desktop by either editing or deleting and creating new ones this option I find is the best. How did you delete the user I believe deluser is the one that erases the configuration files for the user with a switch you can use ( --remove-all-files IIRC)?

[massai]

And while I'm thinking about it did you make the /sys directory for the line that is being mounted for the sysfs in the file?


Horror !
No I never did that...... ok.... done now. Thanks for catching that.


UPDATE-04: The CD-rw and DVD devices are now WORKING properly
I believe the fragments of the knoppix group name were causing internal confusion.
I am certain that the knoppix user was deleted (a few weeks ago) by using some command/script; I did not just merely trash the directory (but apparently whatever I did was not clean enough and left some garbage behind)

Yup, I will take your suggestion and use scd0 and scd1 in the /etc/fstab (although I had revised and it was working, I would rather trust you on this)


# /etc/fstab: filesystem table.
#
# filesystem mountpoint type options dump pass
/dev/hdb1 / ext3 defaults,errors=remount-ro 0 1
/dev/hdb5 none swap sw 0 0
/dev/fd0 /mnt/floppy auto user,noauto 0 0
/dev/scd0 /mnt/cdrom iso9660 user,ro,noexec,noauto 0 0
/dev/scd1 /mnt/cdrom1 iso9660 defaults,ro,user,noexec,noauto 0 0
#/dev/dvd /mnt/dvd iso9660 defaults,ro,user,noexec,noauto 0 0

none /sys sysfs defaults 0 0
proc /proc proc defaults 0 0
usbdevfs /proc/bus/usb usbdevfs defaults 0 0


I chose to keep it simple. There are only two icons on the desktop representing the physical drives. So whatever I pop a disk (may it be data or audio) in the particular drive the filesystem should be mounted to the same point (so that I do not have to make adjustments or make and delete icons every the time)
The same goes for the Floppy (I chose to use "auto" in the fstab) so with a single icon I can mount all kinds of file systems and be done with it)

UPDTATE-05: Next troubleshooting point is the swap file.
Right now the swap file exists in an alternate universe of some sort.

When I do "fdsisk -l" here is what I get:


Disk /dev/hdb: 4310 MB, 4310433792 bytes
255 heads, 63 sectors/track, 524 cylinders
Units = cylinders of 16065 * 512 = 8225280 bytes

Device Boot Start End Blocks Id System
/dev/hdb1 * 1 491 3943926 83 Linux
/dev/hdb2 492 524 265072+ 5 Extended
/dev/hdb5 492 524 265041 82 Linux swap


But when I do "free", this is what I get:


root@ThisKnoppixBox:/home/Thomas# free
total used free shared buffers cached
Mem: 710428 488840 221588 0 52208 246048
-/+ buffers/cache: 190584 519844
Swap: 0 0 0


...I will dig deeper....
..to be continued...

[Stephen]

Well it's set to the right type so have you mkswap /dev/hdb5 to format it as swap space then try mount -a to see if it gets mounted. And since you only have the two partitions on the drive you may want to delete the extended partition and just make a primary partition for the /swap then it would become the /dev/hdb2 for the space and then run the mkswap on it.

[massai]

After examining the dmesg output I found the following error:

"Unable to find swap-space signature"

In the past couple of hours I tried several combinations of mkswap, swapon, swapoff, but with no success.
I will try once more



P.S. Stephen, thanks for pointing the direction for the umpteenth time (I begin to feel so much obliged after you have invested so much time with this newbie)

[massai]

Swap partition is now detected !

I guess (being a newbie) it seems I collect all the right ingredients and have the right commands (most of the time) but I always mess-up the implementation.

In this case all cammands swapon, swapoff, mount -a, mkswap, etc.. etc were tried at some point.

This time I tried it in this order and it worked:


mkswap /dev/hdb5
swapon /dev/hdb5




root@ThisKnoppixBox:/home/Thomas# mkswap /dev/hdb5
Setting up swapspace version 1, size = 271396 kB
root@ThisKnoppixBox:/home/Thomas# swapon /dev/hdb5
root@ThisKnoppixBox:/home/Thomas# free
total used free shared buffers cached
Mem: 710428 248088 462340 0 14428 119980
-/+ buffers/cache: 113680 596748
Swap: 265032 0 265032


Now I have to understand what you said about deleting the extended partition....
to be continued...

[Stephen]

If you got the swap working like you have you should just leave it alone that is really just a cosmetic thing on the way I like my partitions set up I probably should not have suggested it to you to start with. Good to heard you got it going BTW.

[massai]

Yes, the extended partition thingy is a low priority and decided to push it back for later (but I will get to it at some point, becasue the general idea is to learn; instead of choosing the path of least resistance and be comfortable)

The are two more issues that are pressing:
a) Configuration of XawTV.
(The darn thing gives me a blue screen. I hear a faint sound in the background. It seems to be stack in one channel and it keeps playing even if I close XawTV. I have to reboot to make it stop)

The card is being detected left and right (it is probably dizzy by too much detection, and cannot perform its duties for me.... poor thing)



Linux video capture interface: v1.00
bttv: driver version 0.9.12 loaded
bttv: using 8 buffers with 2080k (520 pages) each for capture
bttv: Bt8xx card found (0).
PCI: Found IRQ 9 for device 0000:00:0b.0
PCI: Sharing IRQ 9 with 0000:00:07.3
PCI: Sharing IRQ 9 with 0000:00:07.2
PCI: Sharing IRQ 9 with 0000:00:0b.1
bttv0: Bt878 (rev 17) at 0000:00:0b.0, irq: 9, latency: 64, mmio: 0xdddfe000
bttv0: detected: Hauppauge WinTV [card=10], PCI subsystem ID is 0070:13eb
bttv0: using: Hauppauge (bt878) [card=10,autodetected]
bttv0: Hauppauge/Voodoo msp34xx: reset line init [5]
bttv0: Hauppauge eeprom: model=44811, tuner=Temic 4039FR5 (21), radio=yes
bttv0: using tuner=21
bttv0: i2c: checking for MSP34xx @ 0x80... not found
bttv0: i2c: checking for TDA9875 @ 0xb0... not found
bttv0: i2c: checking for TDA7432 @ 0x8a... not found
tvaudio: TV audio decoder + audio/video mux driver
tvaudio: known chips: tda9840,tda9873h,tda9874h/a,tda9850,tda9855,tea6300,tea6420,tda8425,pic16c54 (
PV951),ta8874z
tuner: chip found @ 0xc2
tuner: type set to 21 (Temic NTSC (4039 FR5))
registering 0-0061
bttv0: registered device video0
bttv0: registered device vbi0
bttv0: registered device radio0
bttv0: PLL: 28636363 => 35468950 .. ok
Via 686a/8233/8235 audio driver 1.9.1-ac4-2.5
PCI: Found IRQ 10 for device 0000:00:07.5
PCI: Sharing IRQ 10 with 0000:00:0a.0
PCI: Setting latency timer of device 0000:00:07.5 to 64
ac97_codec: AC97 Audio codec, id: ALG16 (ALC200/200P)


I see that "i2c" cannot be found. Do I smell a kernel "re-compile" coming up?

b) There is one last module (or two? Believe, this is an improvement) that is not being detected, but I don't know which one it is.

These are the very last lines of the dmesg output (everything else before it is CLEAN with no errors.



NET: Registered protocol family 17
ehci_hcd: block sizes: qh 128 qtd 96 itd 128 sitd 64
request_module: failed /sbin/modprobe -- char-major-109-0. error = 256
lp0: using parport0 (polling).
request_module: failed /sbin/modprobe -- char-major-4-64. error = 256
APIC error on CPU0: 00(40)
nvidia: module license 'NVIDIA' taints kernel.
0: nvidia: loading NVIDIA Linux x86 nvidia.o Kernel Module 1.0-4496 Wed Jul 16 19:03:09 PDT 2003
0: NVRM: AGPGART: unable to retrieve symbol table
atkbd.c: Unknown key released (translated set 2, code 0x7a on isa0060/serio0).
atkbd.c: Unknown key released (translated set 2, code 0x7a on isa0060/serio0).
APIC error on CPU0: 40(40)
request_module: failed /sbin/modprobe -- char-major-5-2. error = 256


I don't know if it is three different modules or if it is one module and the daemon keeps probing for it repeatedly.

Note: that I am focusing on modules right now (I do see some other glitches here and there which will have to wait for their turn)

P.S. What does it say ? ..... "NVIDIA taints kernel" ? What the heck is that? (How come I didn't catch that before ?)

[Stephen]

I get those stupid warnings as well I read a post in the Debian User list about a week ago where the poster said he updated the module-init-tools and got rid of them but I upgraded up to the latest version and it did not do anything for me so I have no idea how to get rid of them. I have no experience with a tv card so I do not think I can be of much help on that one although looking at the output it gets detected and it looks like it initializes without error a couple of things you can check, are you a member of the video group I believe that you have to be one to access the video playback of the card and you may want to check if there are any european settings left over in the config files for xawtv. This may be something to start a new thread on we are pretty far into this one and most people are only going to see the title and not be thinking about a tv card. And the taint the kernel that someone's subtle way of telling you that you are running non-gpl code in the kernel.

[massai]

Hi Stephen,

I think the instalation is stable enough at this point. The major hardles have been ironed and I have you to thank for a lot of it. Your good effort is witnessed and well appreciated !

Maybe this is a good point to wrap-up this thread and I will try to smooth out the rest of the "cosmetic" glitches over the course of time.

It is unbelievable how much information is availble about Linux. Somehow I never feel alone.
There are so many knowledgable people out there, from genious teenagers to seasoned professionals, that they know every bit and every byte like the back of their hand. There are people out there who are combing the manuals and the source code line by line.

Due to such staggering amound of knowledge from the community I feel much more confident in Linux than any propriatery system. (there is no sense of doom when something goes "south"; unlike in a proprietary system).

Despite the technical deficalties my experience so far has been very rewarding !