update-initramfs trigger should run flash-kernel

[Oliver Grawert]

Binary package hint: initramfs-tools

if update-initramfs runs on the ARM architecture, namely on a system that uses the flash-kernel tool to update kernels and initramfs, there is no flash-kernel run after update-initramfs finished (i.e. if you install udev) ... a hook needs to be put in place so flash-kernel is run automatically

[Loïc Minier (lool)]

root@babbage:~# apt-get install --reinstall initramfs-tools
Reading package lists... Done
Building dependency tree
Reading state information... Done
0 upgraded, 0 newly installed, 1 reinstalled, 0 to remove and 0 not upgraded.
Need to get 81.3kB of archives.
After this operation, 0B of additional disk space will be used.
Do you want to continue [Y/n]?
Get:1 http://ports.ubuntu.com jaunty/main initramfs-tools 0.92bubuntu27 [81.3kB]
Fetched 81.3kB in 1s (70.4kB/s)
(Reading database ... 41068 files and directories currently installed.)
Preparing to replace initramfs-tools 0.92bubuntu27 (using .../initramfs-tools_0.92bubuntu27_all.deb) ...
Unpacking replacement initramfs-tools ...
Processing triggers for man-db ...
Setting up initramfs-tools (0.92bubuntu27) ...
update-initramfs: deferring update (trigger activated)

Processing triggers for initramfs-tools ...
update-initramfs: Generating /boot/initrd.img-2.6.28-11-imx51

[Oliver Grawert (ogra)]

--- /usr/sbin/update-initramfs 2009-03-13 10:57:32.000000000 +0100
+++ update-initramfs 2009-04-10 13:25:25.000000000 +0200
@@ -184,6 +184,12 @@
  lilo
 }

+# call flash-kernel
+run_flash_kernel()
+{
+ flash-kernel
+}
+
 # check if lilo is on mbr
 mbr_check()
 {
@@ -262,6 +268,9 @@
  if [ -r /etc/zipl.conf ]; then
   zipl
  fi
+ if [ -r /etc/flash-kernel.conf ]; then
+ run_flash_kernel
+ fi
 }

 compare_sha1()

[Oliver Grawert (ogra)]

something similar to teh above should suffice for a start, problematic seems to me that update-initramfs -c (creating an initramfs from scratch) doesnt call a bootloader at all, the run_bootloader() function is only executed when update-initramfs is called with -u ... though the same problem is true for lilo, elilo and zipl here.

[Loïc Minier (lool)]

This affects all arches; http://lists.debian.org/debian-arm/2008/07/msg00071.html has a more elegant fix:
if flash-kernel --supported >/dev/null 2>&1; then
 flash-kernel
fi

perhaps we need to test for the existence of flash-kernel as well (but not the .conf, that's arch specific).

[Oliver Grawert (ogra)]

confirming the debian fix works fine (tested with apt-get install --reinstall udev)

[Steve Langasek (vorlon)]

Loïc asked for a release team review/ack here, but I don't see that there's a final debdiff available for inspection - from what I see there are two different patches being considered? (maybe three, depending on what's in the debian fix?)

Either of the proposed bits of code here look reasonable/acceptable to me.

As for update-initramfs -c not calling the bootloader, it's assumed that this is handled by the kernel postinstall hooks instead. Does flash-kernel correctly integrate with the kernel packages on that score?

[Loïc Minier (lool)]

What Debian merged was this patch:
diff --git a/update-initramfs b/update-initramfs
index 6f978f3..9830a7d 100755
--- a/update-initramfs
+++ b/update-initramfs
@@ -262,6 +262,12 @@ run_bootloader()
        if [ -r /etc/zipl.conf ]; then
                zipl
        fi
+ if flash-kernel --supported >/dev/null 2>&1; then
+ flash-kernel
+ fi
+ if glantank-update-kernel --supported >/dev/null 2>&1; then
+ glantank-update-kernel
+ fi
 }

 compare_sha1()

(modified to drop glantank later on)

Concerning update-initramfs -c: we add a /etc/kernel-img.conf which calls flash-kernel on upgrades.

Will attach a debdiff.

[Loïc Minier (lool)]

I tested the patch with "sudo apt-get install --reinstall udev" on my desktop:
Réception de : 1 http://archive.ubuntu.com jaunty/main udev 141-1 [338kB]
338ko réceptionnés en 0s (1744ko/s)
(Lecture de la base de données... 247996 fichiers et répertoires déjà installés.)
Préparation du remplacement de udev 141-1 (en utilisant .../archives/udev_141-1_amd64.deb) ...
Ajout de « local diversion of /sbin/udevadm to /sbin/udevadm.upgrade »
Dépaquetage de la mise à jour de udev ...
Traitement des actions différées (« triggers ») pour « man-db »...
Paramétrage de udev (141-1) ...
 * Stopping kernel event manager... [ OK ]
 * Starting kernel event manager... [ OK ]
Suppression de « local diversion of /sbin/udevadm to /sbin/udevadm.upgrade »
update-initramfs: deferring update (trigger activated)

Traitement des actions différées (« triggers ») pour « initramfs-tools »...
update-initramfs: Generating /boot/initrd.img-2.6.28-11-generic

and on an armel imx51 babbage board:
Get:1 http://ports.ubuntu.com jaunty/main udev 141-1 [326kB]
Fetched 326kB in 4s (80.6kB/s)
(Reading database ... 45967 files and directories currently installed.)
Preparing to replace udev 141-1 (using .../archives/udev_141-1_armel.deb) ...
Adding `local diversion of /sbin/udevadm to /sbin/udevadm.upgrade'
Unpacking replacement udev ...
Processing triggers for man-db ...
Setting up udev (141-1) ...
 * Stopping kernel event manager... [ OK ]
 * Starting kernel event manager... [ OK ]
Removing `local diversion of /sbin/udevadm to /sbin/udevadm.upgrade'
update-initramfs: deferring update (trigger activated)

Processing triggers for initramfs-tools ...
update-initramfs: Generating /boot/initrd.img-2.6.28-11-imx51
Flashing kernel... done.
Generating padded initramfs... done.
Flashing initramfs... done.

(the effects were actually visible during installation of initramfs-tools itself, but I wanted a clean test.)

[Launchpad Janitor (janitor)]

This bug was fixed in the package initramfs-tools - 0.92bubuntu28

---------------
initramfs-tools (0.92bubuntu28) jaunty; urgency=low

  * update-initramfs: copy the logic added in Debian to run flash-kernel after
    the initrd has been generated (if the system is supported by
    flash-kernel); flash-kernel is run twice, but at least the system really
    runs the latest initrd; LP: #358762.

 -- Loic Minier <email address hidden> Sat, 11 Apr 2009 00:07:04 +0200