[SOLVED] Question about upgrading new security kernel...

FTIO · 08-24-2016, 05:33 AM

Okay, I got the security upgrades for 14.2 today. The kernel upgrade has me a little worried.

It says to

Quote:

Be sure to upgrade your initrd after upgrading the kernel packages.
If you use lilo to boot your machine, be sure lilo.conf points to the correct
kernel and initrd and run lilo as root to update the bootloader.

Since all I ever do is install Slackware the most simple way possible by choosing all the stock choices (huge kernel, etc), I've never had to do anything with/to 'initrd'. Nor ever had to make sure lilo was pointing to the correct kernel.

How exactly do I go about doing this? If I simply do an 'upgradepkg' with the new kernel stuff, won't that simply overwrite/replace the old kernel and the initrd and lilo will simply see that as the kernel it should work with as always?

I mean, how/what does one change in lilo and initrd when, for example my lilo on my sda drive says:

Code:

# Linux bootable partition config begins
image=/boot/vmlinuz
  root=/dev/sda1
  label=Slack-main
  read-only  # Partitions should be mounted read-only for checking
# Linux bootable partition config ends
# Other Slackware bootable partition config begins
other=/dev/sdb1
  label=Slackware14.2
  table=/dev/sdb

and the lilo on sdb has:

Code:

# Linux bootable partition config begins
image = /boot/vmlinuz
  root = /dev/sdb1
  label = Slackware14.2
  read-only  # Partitions should be mounted read-only for checking
# Linux bootable partition config ends

Thanks for any help with this and please, keep it really simple, heh.

ponce · 08-24-2016, 05:41 AM

if you choose to use the huge kernel for your installation of slackware 14.2, when you're running slackware 14.2 on /deb/sdb, just check if /boot/vmlinuz is a symlink pointing to it

Code:

ls -la /boot/vmlinuz

if it does you shouldn't need any initrd, just run lilo again

Code:

lilo

suppy · 08-24-2016, 05:48 AM

If /boot/vmlinuz points to the kernel image you want to use, all you need to do is rerun /sbin/lilo, which will install the kernel in question for lilo's bootloader to use.

The bootloader doesn't have access to the filesystem in the same way that a running system does, so failure to do this will lead to it trying to use the old location on disk where your old kernel was, and that's very likely to be overwritten. But even if not, it won't be able to load modules properly, as those will be for your new kernel.

zk1234 · 08-24-2016, 05:53 AM

If it used to work without initrd, then it still should do (*).
You can also prepare a rescue (boot-able) USB with your old kernel (just in case).

.
(*) Unless some functions needed to access your HDD have been moved from the kernel to its loadable modules, but I doubt it.

kjhambrick · 08-24-2016, 07:00 AM

FITO --

As zk1234 said, you don't need an initrd for 4.4.19 huge if you didn't need one for 4.4.14 huge.

I'll add this: if you download patches to a local repository and you use the good-ole, manual, standard Slackware Package Management Scripts ( installpkg, upgradepkg, removepkg ).

My Slackware64 local repository is in /home/dld/slackware/ and the 64-bit patches are in /home/dld/slackware/slackware-14.2-64

Use installpkg for these kernel Packages ( I am on Slackware64 14.2. YMWV if you run 32-bit )

Code:

cd /home/dld/slackware/slackware-14.2-64    ### note that this is MY repository -- YMMV

installpkg patches/packages/kernel-generic-4.4.19-x86_64-1_slack14.2.txz
installpkg patches/packages/kernel-huge-4.4.19-x86_64-1_slack14.2.txz
installpkg patches/packages/kernel-modules-4.4.19-x86_64-1_slack14.2.txz
installpkg patches/packages/kernel-source-4.4.19-noarch-1_slack14.2.txz

Then you can upgradepkg the kernel-firmware package:

Code:

upgradepkg patches/packages/kernel-firmware-20160823git-noarch-1_slack14.2.txz

I am not sure what to recommend for the kernel-headers since there is no new glibc package for Slackware 14.2 ( as guanx said, we DO need a repackaged glibc for 14.2 to fix libm.so as was done in -Current ) but if you decide you do want the new kernel-headers:

Code:

upgradepkg patches/packages/kernel-headers-4.4.19-x86-1_slack14.2.txz

Since you're running huge ( me too ), update /etc/lilo.conf and add a 'block' for 4.4.19 and fix your 4.4.14 block.

One thing you WILL want to change is your existing block in /etc/lilo.conf for 4.4.14:

Code:

# Linux bootable partition config begins
image=/boot/vmlinuz                                ### change this
  root=/dev/sda1
  label=Slack-main                                 ### change this
  read-only  # Partitions should be mounted read-only for checking

To something like this ( labels are YOUR choice but be careful with the length of the labels -- there is a limit ):

Code:

<<snip>>
#
# change image=/boot/vmlinuz and label=Slack-main for your OLD 4.4.14 kernel
#
image   = /boot/vmlinuz-huge-4.4.19  
  root  = /dev/sda3
  label = Slack4419
  read-only
image   = /boot/vmlinuz-huge-4.4.14
  root  = /dev/sda3
  label = Slack4414
  read-only
<<snip>>

Finally, after lilo.conf is fixed, rerun lilo AND WATCH FOR ERRORS.

Code:

lilo -v

If there are any errors ( other than maybe: Warning: LBA32 addressing assumed ) !!! STOP AND FIX ERRORS !!!

By running installpkg on the new kernel files and by keeping a safe copy of 4.4.14 in lilo.conf, you'll always be able to boot your working 4.4.14 kernel if 4.4.19 fails.

HTH and Have Fun !

-- kjh

bassmadrigal · 08-24-2016, 08:18 AM

As a another addition to these fine comments, if you use slackpkg to update, I'm pretty sure it will prompt you to reinstall lilo at the end. As others have stated, if you aren't using an initrd, you shouldn't need to make one with this update.

hitest · 08-24-2016, 09:12 AM

Quote:

Originally Posted by bassmadrigal

As a another addition to these fine comments, if you use slackpkg to update, I'm pretty sure it will prompt you to reinstall lilo at the end.

Yes. I upgraded to the new kernel in Slackware 14.2 using slackpkg. The slackpkg utility will prompt you to run lilo at the end of the process.

FTIO · 08-24-2016, 09:35 AM

Thanks everyone, your help worked perfectly without a hiccup.

I simply did...

Code:

cd /home/me/upgrades
upgradepkg *.txz
<when that was all done>lilo

*BEFORE* I rebooted though, just to be on the safe side I changed init to 3 (instead of 4), changed the 'quicky' fix of the sysctl.d thing found in another threadby removing that .conf file, and thus when I booted back up I would be at the non-X login and could re-install my nVidia. Then rebooted once more and all's well.

Thanks again to all the quick help and great advice.

kjhambrick · 08-24-2016, 09:53 AM

Thanks for the reminder FITO !

I just now pounded-out the quicky fix in /etc/sysctl.conf too (

I had forgotten all about it until you reminded me

)

-- kjh

Code:

#
# B60824 - no longer necessary with 4.4.19+
# net.ipv4.tcp_challenge_ack_limit = 999999999

FTIO · 08-24-2016, 10:14 AM

Quote:

Originally Posted by kjhambrick

Thanks for the reminder FITO !

I just now pounded-out the quicky fix in /etc/sysctl.conf too (

I had forgotten all about it until you reminded me

)

-- kjh

Code:

#
# B60824 - no longer necessary with 4.4.19+
# net.ipv4.tcp_challenge_ack_limit = 999999999

I'm actually useful once in a very great while, heh.