Hi Andrew, hi Ken,
Thanks for your help.
First a small errata: I did an kernel "upgrade" to kernel 2.6.13.1, not
to 2.6.13.
Having applied the things you have suggested (see below, section "OLD
ISSUES"), there is one single error message left that is displayed (or
seems to be displayed, it is very hard to read cause the kernel is
loaded to quickly, but something is display in red letters, i.e
"Failed") while loading the 2.6.13.1:
###########################################
NEW ISSUE:
First I thought the error message had to to with the population of /dev.
But according to sys.log, this error message has to do with loading the
driver for my RIVA TNT2 Model 64 graphic card. Sys.log says:
Sep 14 21:52:31 localhost kernel: [17179580.548000] nvidia: module
license 'NVIDIA' taints kernel.
This is due to an already existing nvidia driver which I compiled under
the older kernel 2.6.11.7. This driver or the nvidia libraries cannot
be loaded by the new kernel. The [Failed] error message that is quickly
displayed when loading the newer kernel does not appear, when I comment
out the module "nvidia" in the file /etc/sysconfig/modules, and I do a
reboot into the newer kernel.
So this fast [Failed] message is related to the nvidia driver that
cannot be loaded by the newer kernel. All error messages which I
mentioned in my first mail, i.e. error message (1), (2), (3) and (4)
have been solved (s. below) or the reason for the error message has
been found (the nvidia driver loading issue).
And the behaviour regarding the nvidia driver is a new issue in this
email communication.
I tried to compile the nvidia driver for the newer kernel (2.6.13.1),
without success. I tried the driver installation (version 7167) that
works fine under my older kernel (2.6.11.), and I tried the latest
driver installation (version 7676), but the driver compilation always
fails.
When I try to compile the nvidia driver, version 7676, under the newer
kernel, I get these errors in the nvidia-installer.log:
"NVIDIA: left KBUILD.
-> done.
-> Kernel module compilation complete.
ERROR: Unable to load the kernel module 'nvidia.ko'. This is most
likely because the kernel module was built using the wrong kernel
source files. Please make sure you have installed the kernel source
files for your kernel [...] If you know the correct kernel source files
are installed, you may specify the kernel source path with the
'--kernel-source-path' commandline option.
-> Kernel module load error: insmod: error inserting
'./usr/src/nv/nvidia.ko':
-1 No such device"
I tried the --kernel-source-path option, no success. I made a symlink
from /usr/src/linux to /usr/src/linux-2.6.13.1, and
from /lib/modules/2.6.13.1/build to /usr/src/linux-2.6.13.1, but no
success. I tried several "Advanced" options which the install script
NVIDIA-Linux-x86-1.0-7676-pkg1.run offers, no success.
PS: Under Fedora Core 3, plus the multimedia packages offered by Planet
CCRMA, I succesfully could install the nvidia driver, using
NVIDIA-Linux-x86-1.0-7676-pkg1.run.
Any ideas?
###########################################
OLD ISSUES:
@Andrew:
> Jens Radloff wrote:
> >
> > (3) Some seconds later, this message is displayed: "RNTLINK answers:
> > Files exists [Failed]"
> Edit /etc/rc.d/init.d/localnet and move the line `ip link set lo up'
> down one line so that it's after the line `ip addr add 127.0.0.1/8
> label lo dev lo'. The interface should be configured before it's set
> up.
Yes, that helped!
> > (4) Additionaly, Alsa cannot be loaded: "Starting ALSA ... [Failed]"
> Did you build alsa into the kernel? If you built alsa as modules
> you'll need to change the alsa bootscript so that it doesn't try to
> use the hardware before the modules have loaded. Take the easy route,
> build it into the kernel.
I did not build alsa into the kernel, but the error messages about alsa
disapeared from the next boot on after I applied an (not really
necessary:) "alsamixer", and and "alsactl store" (necessary).
@Ken:
> > So I did an upgrade from udev-030 to udev-58 [...]
> I _hope_ 058 is a typo - we're on udev-068 now.
Yes, I did an upgrade to udev-068 :)
> > But some error messages are displayed:
> >
> > (1) An error message which I am nearly unable to read because the
> > kernel boot is too fast when it is displayed. I think it is an
> > message which says that /dev cannnot be populated.
As mentioned above, this error message is not related to the population
of the /dev directory, but to failing to load the nvidia driver.
After I applied the steps which Andrew mentioned, I loaded the newer
kernel (2.6.13.1), ...
> If that's what it said, all subsequent errors are kind of irrelevant
> (no device nodes to mount filesystems on).
... I then I removed and re-created the following log files: sys.log,
auth.log, kern.log and maybe some others. Then I did another reboot
into the newer kernel, and I searched dmesg and all log files regarding
the pattern "failed", using the commands 'dmesg | grep -i failed" and
'grep -iR failed /var/log*'. I did not get any hits returned.
So guess that in fact no (relevant) error occured regarding the
population of the /dev directory.
> > I am quite sure that all settings in the kernel config file of the
> > newer kernel 2.6.13. are correct. When the newer kernel has been
> > loaded, a) the following (pseudo) file systems are mounted: proc,
> > sysfs, devpts, shm, usbfs and ramfs,
> From your error messages, I thought you hadn't been able to log in.
No, every boot into the newer kernel is always ending with a login
prompt, and I can login, as root, and as normal users.
> > b) the /dev directory contains 659 files (which i found out using
> > ls /dev | wc -l )
> That sounds reasonably full (653 on this box).
Good to know :)
> Work out what is going on with the ext fs message,
I thought the message "Ext-fs: mounted file system with ordered data
mode: [Failed]" is related to udev, but obviously it is not.
I found the reason for the message "Ext-fs: mounted file
system with ordered data mode: [Failed]": in my fstab file, I defined
for a FAT partition among others one mounting option called
"codepage=850", although I did not compile this codepage. In the kernel
config file of the newer kernel, I have changed the option
"CONFIG_FAT_DEFAULT_CODEPAGE" from "852" to "850", then I will compile
the kernel again, and I am quite sure that the error message about "
"Ext-fs: ..." will not show anymore.
> For alsa, ...this has been covered on blfs-support recently, I think
> it's just a question of setting the volumes and unmuting, then
> running a command to save the state.
The error messages about alsa when I load the newer kernel disapeared
from the next boot on after I applied an (not really necessary:)
"alsamixer", and and "alsactl store" (necessary).
Regards and thanks again,
Jens
___________________________________________________________
Gesendet von Yahoo! Mail - Jetzt mit 1GB Speicher kostenlos - Hier anmelden:
http://mail.yahoo.de
--
http://linuxfromscratch.org/mailman/listinfo/lfs-support
FAQ: http://www.linuxfromscratch.org/lfs/faq.html
Unsubscribe: See the above information page
