I normally compile my own kernels but I've not had so much trouble as I'm having with 2.6.2 and 2.6.0 before it.
I'm running Debian Sarge (Testing).
When starting to boot I see 'bios data check successful' then there is a pause of two minutes before the boot continues.
In the log I see 'No module symbols loaded - kernel modules not enabled'.
In my config I have:
# Loadable module support # CONFIG_MODULES=y CONFIG_MODULE_UNLOAD=y # CONFIG_MODULE_FORCE_UNLOAD is not set CONFIG_OBSOLETE_MODPARM=y CONFIG_MODVERSIONS=y CONFIG_KMOD=y
Anyone else experiencing problems with this kernel?
Barry Samuels http://www.beenthere-donethat.org.uk The Unofficial Guide to Great Britain
On Fri, Feb 27, 2004 at 05:22:43PM +0000, Barry Samuels wrote:
I normally compile my own kernels but I've not had so much trouble as I'm having with 2.6.2 and 2.6.0 before it.
I'm running Debian Sarge (Testing).
Anyone else experiencing problems with this kernel?
<quick random guess>
I have been/am using kernel 2.6.3 with Debian, have you installed the package module-init-tools ?
Adam
adam@thebowery.co.uk wrote:
On Fri, Feb 27, 2004 at 05:22:43PM +0000, Barry Samuels wrote:
I normally compile my own kernels but I've not had so much trouble as I'm having with 2.6.2 and 2.6.0 before it.
I'm running Debian Sarge (Testing).
Anyone else experiencing problems with this kernel?
<quick random guess>
I have been/am using kernel 2.6.3 with Debian, have you installed the package module-init-tools ?
Yes it is installed.
Barry Samuels http://www.beenthere-donethat.org.uk The Unofficial Guide to Great Britain
On 27 Feb 2004, at 17:22, Barry Samuels wrote:
When starting to boot I see 'bios data check successful' then there is a pause of two minutes before the boot continues.
In the log I see 'No module symbols loaded - kernel modules not enabled'.
In my config I have:
# Loadable module support # CONFIG_MODULES=y CONFIG_MODULE_UNLOAD=y # CONFIG_MODULE_FORCE_UNLOAD is not set CONFIG_OBSOLETE_MODPARM=y CONFIG_MODVERSIONS=y CONFIG_KMOD=y
Anyone else experiencing problems with this kernel?
Well mine is different. I never select mod versions in the kernel (Due to the webcam I own and kernel testing)
# # Loadable module support # CONFIG_MODULES=y CONFIG_MODULE_UNLOAD=y CONFIG_MODULE_FORCE_UNLOAD=y CONFIG_OBSOLETE_MODPARM=y # CONFIG_MODVERSIONS is not set CONFIG_KMOD=y
Though, did you copy your kernel config from 2.4 to 2.6 and ran make oldconfig just in case?
Should always install module-init-tools, devfsd (should be udev but don't see it anywhere in Debian yet?) and that should be it. Tried 2.6.3?
It's 10.30 on a Friday night and I'm offering technical support?! Something wrong with me!
C
On 2004.02.27 22:33, Craig wrote:
On 27 Feb 2004, at 17:22, Barry Samuels wrote:
When starting to boot I see 'bios data check successful' then there is a pause of two minutes before the boot continues.
Interesting. The 'bios data check successful' message actually comes from LILO rather than from the kernel.
In the log I see 'No module symbols loaded - kernel modules not enabled'.
I am sure I have seen the 'No module symbols loaded' part but I don't remember the 'kernel modules not enabled' part and in my case kernel modules are definitely working because my network card driver is a module and that seems to be loading OK.
I am not sure if this is relevent to the problem at hand anyway - as the pause in booting comes right after that message from LILO it must be happenning either in LILO or in the very first stages of kernel initialisation long before modules get loaded.
Well mine is different. I never select mod versions in the kernel (Due to the webcam I own and kernel testing)
I have mod versions enabled on three 2.6 kernel machines and all seem to work perfectly. All are running Debian testing/unstable and using gcc version 3.3.3 20040125. The only one I have access to at the moment is using module-init-tools 3.0-pre9-1.
On two of the machines I built the kernel on that machine and installed it directly, i.e. doing:
make oldconfig make su cp arch/i386/boot/bzImage /boot/vmlinuz-2.6.3 cp System.map /boot/System.map-2.6.3 cd / rm vmlinuz ln -s boot/vmlinux-2.6.3 vmlinuz lilo cd - make modules_install
The third machine has a slow CPU so I built the kernel on one of the other 2.6 machiines with debian make-kpkg and transferred the .deb file to the slower machine and just installed it like any other package.
I did have some problems with early 2.6 kernels and ACPI on my laptop which would happily lock up shortly after logging in to it. The solution at the time was to tell the kernel not to trust APCI for PCI interrupt routing with append="pci=noacpi" in the lilo.conf file.
HTH, Steve.
On 28 Feb 2004, at 13:57, Steve Fosdick wrote:
Well mine is different. I never select mod versions in the kernel (Due to the webcam I own and kernel testing)
I have mod versions enabled on three 2.6 kernel machines and all seem to work perfectly. All are running Debian testing/unstable and using gcc version 3.3.3 20040125. The only one I have access to at the moment is using module-init-tools 3.0-pre9-1.
Well the term 'if it works, it works' comes to mind ;)
On two of the machines I built the kernel on that machine and installed it directly, i.e. doing:
<snip about building kernels>
Seeing that you use lilo there.. ever tried GRUB? I like that more than lilo ;)
I did have some problems with early 2.6 kernels and ACPI on my laptop which would happily lock up shortly after logging in to it. The solution at the time was to tell the kernel not to trust APCI for PCI interrupt routing with append="pci=noacpi" in the lilo.conf file.
I was using 2.5.60+ onwards on my old x86 laptop back then. It was nice but kept freezing when I used the cd burner (Which was fixed after 2 hours researching on google on my painful dialup back then).
C
On 2004.02.28 19:57, Craig wrote:
I was using 2.5.60+ onwards on my old x86 laptop back then. It was nice but kept freezing when I used the cd burner (Which was fixed after 2 hours researching on google on my painful dialup back then).
I remeber having trouble with locups using the CD when using SCSI emulation under 2.6.0 and I worked around it by dropping back to the ide-cd driver. That isn't a permenant solution though because, AFAIK, the ide driver doesn't support writing to the cd.
Do you know if this is fixed in 2.6.3, or is the solution you researched still relevant?
Steve.
On 29 Feb 2004, at 17:40, Steve Fosdick wrote:
On 2004.02.28 19:57, Craig wrote:
I was using 2.5.60+ onwards on my old x86 laptop back then. It was nice but kept freezing when I used the cd burner (Which was fixed after 2 hours researching on google on my painful dialup back then).
I remeber having trouble with locups using the CD when using SCSI emulation under 2.6.0 and I worked around it by dropping back to the ide-cd driver. That isn't a permenant solution though because, AFAIK, the ide driver doesn't support writing to the cd.
Yes, I remember that. Dropped the SCSI emulation, used the IDE-CD driver + latest version of cdrecord (beta back then) that supports IDE.
Should be much the same in 2.6.3... I don't own a x86 laptop so I can't test this out :( However, I presume this works now :)
C
On 2004-02-29 17:40:25 +0000 Steve Fosdick lists@pelvoux.nildram.co.uk wrote:
Do you know if this is fixed in 2.6.3, or is the solution you researched still relevant?
For me, SCSI stability improved much between 2.6.0 and 2.6.2, although I can still contrive to crash the SCSI subsystem quite easily.
On 1 Mar 2004, at 11:10, MJ Ray wrote:
On 2004-02-29 17:40:25 +0000 Steve Fosdick lists@pelvoux.nildram.co.uk wrote:
Do you know if this is fixed in 2.6.3, or is the solution you researched still relevant?
For me, SCSI stability improved much between 2.6.0 and 2.6.2, although I can still contrive to crash the SCSI subsystem quite easily.
Well 2.6.4 is out soon....
I thought they were going to 'calm' it down?
C
Steve Fosdick lists@pelvoux.nildram.co.uk wrote:
On 2004.02.27 22:33, Craig wrote:
On 27 Feb 2004, at 17:22, Barry Samuels wrote:
When starting to boot I see 'bios data check successful' then there is a pause of two minutes before the boot continues.
Interesting. The 'bios data check successful' message actually comes from LILO rather than from the kernel.
In the log I see 'No module symbols loaded - kernel modules not enabled'.
I am sure I have seen the 'No module symbols loaded' part but I don't remember the 'kernel modules not enabled' part and in my case kernel modules are definitely working because my network card driver is a module and that seems to be loading OK.
I am not sure if this is relevent to the problem at hand anyway - as the pause in booting comes right after that message from LILO it must be happenning either in LILO or in the very first stages of kernel initialisation long before modules get loaded.
Well mine is different. I never select mod versions in the kernel (Due to the webcam I own and kernel testing)
I have mod versions enabled on three 2.6 kernel machines and all seem to work perfectly. All are running Debian testing/unstable and using gcc version 3.3.3 20040125. The only one I have access to at the moment is using module-init-tools 3.0-pre9-1.
There does seem to be something wrong with the two versions I've tried i.e. 2.6.0 and 2.6.2.
When running make xconfig and looking under 'Input device support' there is a 'Mouse interface' option. In my current kernel 2.4.24 this option is compiled as a module and in the help window for this option in 2.6.2 it says enter 'm' to compile it as a module but there is no way to do this. It's either on or off.
If I change the option in the .config file directly to read CONFIG_INPUT_MOUSEDEV=m it still isn't compiled as a module.
Strange withall!
Barry Samuels http://www.beenthere-donethat.org.uk The Unofficial Guide to Great Britain
On Mon, Mar 01, 2004 at 09:20:21PM +0000, Barry Samuels wrote:
There does seem to be something wrong with the two versions I've tried i.e. 2.6.0 and 2.6.2.
When running make xconfig and looking under 'Input device support' there is a 'Mouse interface' option. In my current kernel 2.4.24 this option is compiled as a module and in the help window for this option in 2.6.2 it says enter 'm' to compile it as a module but there is no way to do this. It's either on or off.
If I change the option in the .config file directly to read CONFIG_INPUT_MOUSEDEV=m it still isn't compiled as a module.
Hmmmn, in my kernel i don't actually have that as an option (I think it is built in because of another option I have selected) anyhow, would you like a copy of my .config? it is optimised for my machibe but I am fairly sure it will boot any x86 machine with ide (once you have changed the cpu type from athlon to whatever you have, and also the acpi could cause a problem i guess, but anyhow)
Anyhow if you want a copy of the .config then drop me a mail offlist.
Adam
On 2004-03-01 21:53:43 +0000 adam@thebowery.co.uk wrote:
On Mon, Mar 01, 2004 at 09:20:21PM +0000, Barry Samuels wrote:
There does seem to be something wrong with the two versions I've tried i.e. 2.6.0 and 2.6.2.
When running make xconfig and looking under 'Input device support' there is a 'Mouse interface' option. In my current kernel 2.4.24 this option is compiled as a module and in the help window for this option in 2.6.2 it says enter 'm' to compile it as a module but there is no way to do this. It's either on or off.
If I change the option in the .config file directly to read CONFIG_INPUT_MOUSEDEV=m it still isn't compiled as a module.
Hmmmn, in my kernel i don't actually have that as an option (I think it is built in because of another option I have selected) anyhow, would you like a copy of my .config? it is optimised for my machibe but I am fairly sure it will boot any x86 machine with ide (once you have changed the cpu type from athlon to whatever you have, and also the acpi could cause a problem i guess, but anyhow)
Anyhow if you want a copy of the .config then drop me a mail offlist.
An interesting development. The source for kernel 2.6.n has disappeared from the Testing/Sarge package lists!
Thanks for the offer Adam but I'll wait to see what happens next.
-
adam＠thebowery.co.uk -
Barry Samuels -
bjsamuels＠beenthere-donethat.org.uk
-
Craig -
MJ Ray -
Steve Fosdick