Re: x86_64 kernel update error

Hi,


On Sat, Dec 25, 2004 at 02:01:33PM +0100, Remco Barendse wrote:
> Hi list!
> 
> Sorry for the number of new threads
> 
> When running yum update I get this message :
> ypserv 100 % done 86/372
> kernel-smp 100 % done 87/372
> No module 3w-9xxx found for kernel 2.4.21-27.TLsmp, aborting.
> 
> The kernels did seem to be installed however:
> 
> [root@server1 root]# rpm -qa | sort |grep kernel-
> kernel-2.4.21-20.3.EC
> kernel-2.4.21-27.TL
> kernel-doc-2.4.21-27.TL
> kernel-smp-2.4.21-20.3.EC
> kernel-smp-2.4.21-27.TL
> kernel-source-2.4.21-20.3.EC
> kernel-source-2.4.21-27.TL
> kernel-utils-2.4-8.37.7
> 
> But no grub records were created (fortunately).
> 
> Where can I find the EC kernels?

http://core.upi.iki.fi/out/kernel/

The longer answer is that this is very unfortunate efect due to the
fact i've not been able to release those U3-level iso images
officially. That's why the updates-tree doesn't have needed stuff for
the U3-level installation - updates only have stuff for U2-level
installation (those few packages modified on U3 are left out).

Sadly it's going to be so until i'll be able to release those U3-level
images and can make U4-images (no U4-images before U3 is released).

One should wait a little as i am just now building 2.4.21-27.300.EC
images which are latest level (-27) + enhancements. There were a nasty
incompatibility on XFS-merge which was made to my attention via OOPS by
a user. kernel was calling a function with struct file as a param, but
the funtion assumed to receive struct inode (2.4-vanilla does call with
struct inode). I am amaxed that it has not popped up earlier as i've
had thise XFS in my tree long time and the incompatibility is kind of
fatal in nature.

I've been testing this fix since thursday, so begin to be pretty sure
it's working now (it's only two lines tho :), but there is also a
resync for XFS from 2.4.29-pre2 level.

The another thing brought to my attention was that ACLs was not enabled
on XFS code. I've enabled those too and it has been tested even by a
user with a data that does have 'real production amount of ACLs' and
not only my 'i just make so arbitary to see if it's working' amount.

So thank you both (the OOPS and ACLs) making me to fix things that had
slipped my tests.




-- 
Pasi Pirhonen - [email protected] - http://iki.fi/upi/