[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Bug#491692: initramfs-tools: udev not copied to ramdisk as hookscript is skipped



On Mon, Jul 21, 2008 at 02:03:23PM +0200, Philipp Sternberg wrote:
> 
> Hi folks,
> 
> i ran into this after trying to install a 2.6.25.10-vanilla using make-kpkg. When I tried to boot the new kernel, 
> the system hung. It said /scripts/init-premount/udev : 17 : udevd : not found. Well the reason for that was very 
> simple: udevd was not copied into the kernel image:
> 
> Allright here is how one might reprduce it (at least that was how I ran into it):
> 1. unpack the kernel sources and configure the kernel.
> 	--> After unpacking I did
> 	cd /usr/src/linux-2.6.25.10
> 	zcat /proc/config.gz > .config
> 	make oldeconfig
> 	make menuconfig
> 2. Create the kernelimage
> 	make-kpkg clean && make-kpkg --initrd --append-to-version=-custom --rootcmd fakeroot \
> 	kernel_image kernel_headers
> 3. Install the Kernel
> 	dpkg -i /usr/src/linux-image-2.6.25.10-customy_2.6.25.10-customy-10.00.Custom_i386.deb
> 	--> dpkg told me that the ramdisk was created
> 4.If I reboot now the above described problem happens.
> 5. To check if udevd was included in the image i unpacked the kernelimage into a directory and tried to find the 
> udevd binary:
> 	mkdir /home/philipp/Desktop/initrd
> 	cd /home/philipp/Desktop/initrd
> 	zcat /boot/initrd.img-2.6.25.10-customy | cpio -i
> 	find -name '*udev*'
> 	--> which gave me just
> 		./scripts/init-premount/udev
> 		./scripts/init-bottom/udev
> 
> Unfortunately i was not able to see what really happend during the creation of the image so i ran:
> 	update-initramfs -utv
> which revealed that all the hook scripts were skipped with the message
> 	/usr/share/initramfs-tools/hooks/whatever ignored: not alphanumeric or '_' file
> 	
> Thus the udev hook script was skipped as well and no udevd-binary was copied.
> 
> I worked around this by temporarily changing line 77 of /usr/share/initramfs-tools/scripts/functions
> from
> *[![:alnum:]_]*
> to ___
> very very ugly, but however the line, which is in a section which figures out which scripts to exclude from 
> participation in the ramdisk-image creation process, more or less tells to exclude everything as far as i 
> interpreted the 
> glob expression (this *[![:alnum:]_]* string) correctly. Why is that so?
> I'd guess it is a bug. Anyhow even if it isn't I would expect the freshly installed initramfs-tools scripts to 
> create a consistent ramdisk image.
> 
> Well that's it basically
> 
> Regards
> 
> Phil
> root=/dev/hda1 ro 

please sendin the output of
sh -x mkinitramfs -o /tmp/o

thanks

-- 
maks



Reply to: