On Wed, Jul 18, 2012 at 07:27:50PM +0200, Christian Hesse wrote:

Dave Reisner <d@falconindy.com> on Wed, 2012/07/18 13:10:

...

On Wed, Jul 18, 2012 at 07:03:44PM +0200, Christian Hesse wrote:

...

Hello everybody,

I added a hook to the initramfs whose install script added 'uname' by using add_binary. This resulted in major breakage and a kernel panicing on boot.

I suppose this is because there is a symlink from uname to busybox - with adding uname I killed the busybox binary. Is this ok by design or do we want to raise an error if something like this happens?

I created a patch (see attachment). Does this make sense?

Nope, it doesn't make sense. We overwrite a busybox symlink in the shutdown hook (cp, since we need the -a flag), and this isn't a problem.

Ah, my fault...

I added my hook before base in HOOKS array. This way bin was a real directory instead of a symlink to usr/bin.

Yup, I figured this is what you did. Incidentally, this particular user-induced bug should be fixed by a commit I just pushed a few hours ago: http://projects.archlinux.org/mkinitcpio.git/commit/?id=5f81d06e8b

(Note to self: Never do silly things. ;) -- main(a){char*c=/* Schoene Gruesse */"B?IJj;MEH" "CX:;",b;for(a/* Chris get my mail address: */=0;b=c[a++];) putchar(b-1/(/* gcc -o sig sig.c && ./sig */b/42*2-3)*42);}