Hello!Eveyone! I'm a new user to CentOS. I builded a new kernel(2.6.12) yesteray with these command: #make defconfig #make menuconfig #make #make modules #make modules_install #make install #reboot ...
But,when I choose the new kernel,it stop with "kernel panic". When building the kernel,I choose ext3 with "y".My hard disk is IDE,so I choose the IDE support.
I don't know what's wrong. I met the same problem on FC4.But,I can build a new kernel on Debian successfully,using these commands. Please help me. :-(
___________________________________________________________ 雅虎免费G邮箱－中国第一绝无垃圾邮件骚扰超大邮箱 http://cn.mail.yahoo.com
First of all - what kind of a kernel panic do you get? Second of all, the problem is probably the initrd or the lack of one, what do you have in it? (do you have one?)
On Tue, 23 Aug 2005, cpp fire wrote:
Hello!Eveyone! I'm a new user to CentOS. I builded a new kernel(2.6.12) yesteray with these command: #make defconfig #make menuconfig #make #make modules #make modules_install #make install #reboot ...
But,when I choose the new kernel,it stop with "kernel panic". When building the kernel,I choose ext3 with "y".My hard disk is IDE,so I choose the IDE support.
I don't know what's wrong. I met the same problem on FC4.But,I can build a new kernel on Debian successfully,using these commands. Please help me. :-(
雅虎免费G邮箱－中国第一绝无垃圾邮件骚扰超大邮箱 http://cn.mail.yahoo.com
CentOS mailing list CentOS@centos.org http://lists.centos.org/mailman/listinfo/centos
Oh,sorry,I did not describe it clearly. When booting the system,it stops with these words: "kernel panic:VFS:Unable to mount root fs on unknown-block(0,0)"
Maybe it's what you want. Thank you!
___________________________________________________________ 雅虎免费G邮箱－中国第一绝无垃圾邮件骚扰超大邮箱 http://cn.mail.yahoo.com
Ok, so it looks like your kernel is missing one or more of the following: a) hard disk hardware device drivers b) partition drivers (very unlikely, only if non-MSDOS disklabels) c) file system drivers for the root filesystem
or possibly it's been setup for initrd operation and d) initrd support is lacking in the kernel e) the initrd file system (ramfs tmpfs or whatever) is lacking f) the initrd is screwed up...
What are the grub.conf entries for the kernel? does it reference an initrd? does it boot/how does it screw up if the initrd for this new kernel is commented out?
On Tue, 23 Aug 2005, cpp fire wrote:
Oh,sorry,I did not describe it clearly. When booting the system,it stops with these words: "kernel panic:VFS:Unable to mount root fs on unknown-block(0,0)"
Maybe it's what you want. Thank you!
雅虎免费G邮箱－中国第一绝无垃圾邮件骚扰超大邮箱 http://cn.mail.yahoo.com
CentOS mailing list CentOS@centos.org http://lists.centos.org/mailman/listinfo/centos
Oh,thank goodness...it works!!!!!!!!! But in fact,I have to thank maze@cela.pl. :-) I follow his (or her) instruction,and examine my .config,then found that I forget to set RAMFS,initrd support...what a stupid man I am! Now it works well.
And,also thanks Jim Perrin.You are right.Usually,I will not rebuild a kernel on RedHat series and etc,but on Debian(Most of time I use Debian).But as you said,I "honestly feel some burning pull "to rebuild my own kernel.(I will not run it as a production server.)And if I failed,I will think "oh,there's something I don't know,I have to hurdle it!",or else I can't fall asleep.This feeling is terrible.
Aha,tonight I will have a sweet sleep. Thanks again.
--- Maciej �enczykowski maze@cela.pl:
Ok, so it looks like your kernel is missing one or more of the following: a) hard disk hardware device drivers b) partition drivers (very unlikely, only if non-MSDOS disklabels) c) file system drivers for the root filesystem
or possibly it's been setup for initrd operation and d) initrd support is lacking in the kernel e) the initrd file system (ramfs tmpfs or whatever) is lacking f) the initrd is screwed up...
What are the grub.conf entries for the kernel? does it reference an initrd? does it boot/how does it screw up if the initrd for this new kernel is commented out?
On Tue, 23 Aug 2005, cpp fire wrote:
Oh,sorry,I did not describe it clearly. When booting the system,it stops with these words: "kernel panic:VFS:Unable to mount root fs on unknown-block(0,0)"
Maybe it's what you want. Thank you!
___________________________________________________________ �Ż����G���䣭�й���һ���������ʼ�ɧ�ų������� http://cn.mail.yahoo.com
On 8/23/05, cpp fire fire_cpp@yahoo.com.cn wrote:
Oh,sorry,I did not describe it clearly. When booting the system,it stops with these words: "kernel panic:VFS:Unable to mount root fs on unknown-block(0,0)"
First let me say that while this is directed at you, it's not directed JUST at you. Please do not take it personally.
It seems on the irc channel and in other areas that people are rebuilding kernels for centos left and right, and when queried as to why, they falter on an answer. Your steps used to be correct for rebuilding a kernel, but distributions have increased dramatically in complexity to the point that new steps are in order for rebuilding a kernel for an rpm based distro.
Redhat ( and by proxy centos ) have lilterally hundreds of patches built into their kernels and because of this it is becoming increasingly more difficult to use vanilla kernels from kernel.org. Things like nptl patching etc make running vanilla kernels a challenge. Redhat also backports many features of newer kernels to older versions. For example the RHEL/CentOS 3.x kernel (2.4) contains many 2.6 enhancements.
While rebuilding kernels for gentoo and the like is fine, for a package (rpm) based system it can lead to issues if you don't plan for upgrades, and not just for the kernel. If redhat releases a package that depends on kernel 1.2.3 (example. no bitching about numbers) which you installed from source, your rpm database doesn't know that the kernel is installed, and will tell you it's needed. If you force it, you'd introduced a conflict into your database which will mean you'll need to force more and more, which will lead to a near useless rpm setup.
If you have hardware that's not supported by default or you honestly feel some burning pull to rebuild your own kernel, you can rebuild the kernel src.rpm to get the functionality you want by following a guide Kevin Hobbs of Ohio University wrote. http://crab-lab.zool.ohiou.edu/kevin/kernel-compilation-tutorial-en/
Or if you absolutely have to rebuild from a stock kernel, you should be able to use the option "make rpm" so that your database at least knows it's installed.
</rant>
Again, not personal, and not directed at you alone.
On Tue, 2005-08-23 at 09:03 -0400, Jim Perrin wrote:
Your steps used to be correct for rebuilding a kernel, but distributions have increased dramatically in complexity to the point that new steps are in order for rebuilding a kernel for an rpm based distro.
Amen. I got berated by a Gentoo bigot (I am a Gentoo user) whose knowledge was still stuck at Red Hat Linux 6.2/7.3 on a LUG list when I tried to correct his RHEL4 recommendations.
Redhat ( and by proxy centos ) have lilterally hundreds of patches built into their kernels and because of this it is becoming increasingly more difficult to use vanilla kernels from kernel.org.
It's more than just that. You not only have to "prep" from .src.rpm to ensure the patches, but you really need to follow the steps in the .spec file _exactly_ afterwards.
One of the most _dangerous_ Makefile targets is "make mrproper", named after Mr. Clean (the occassional European branding). It will _wipe_out_ header files, and if you're not in the chroot environment, you could wipe out files under a symlink. For those of you who ran into this under Red Hat Linux 7, and had to reinstall the "kernel-headers" RPM, that's what I'm talking about. ;->
Some kernel.org kernels, when unarchived, make all sorts of symlinks into system headers and files.
-
Bryan J. Smith -
cpp fire -
Jim Perrin -
Maciej Żenczykowski