[BusyBox] "auto: applet not found" ...
Erik Andersen
andersen at codepoet.org
Fri Nov 7 21:35:49 UTC 2003
On Fri Nov 07, 2003 at 04:33:10PM -0500, Karim Yaghmour wrote:
>
> Karim Yaghmour wrote:
> >I'm kind'a stuck here ... For some reason which I cannot comprehend (yet),
> >busybox pukes "auto: applet not found" when started by the kernel and
> >then the kernel dies because init is gone. Yet, there is no symbolic
> >link called "auto" pointing to BusyBox ... What's even weirder is if
> >I pass "init=/sbin/init" to the kernel, BusyBox prints out its standard
> >"Usage: busybox [function] [arguments]..." message, and when I look
> >within the section of "Currently defined functions:" I see this:
> >"head, id, or, insmod, kill, ..." I really don't get this one, how could
> >"init" be replaced by "or" ... when I run the exact same busybox on
> >another system, though, I see "head, id, init, insmod, kill" ...
>
> I've now tried "init=/bin/sh" and I get this now:
> "head, id, 8361 [KLE133] Host Bridge, ..."
>
> Of course you won't find the "8361 [KLE133] Host Bridge" string anywhere
> in BusyBox because it comes from the PCI code in the kernel ... So how
> can some string which is usually in the kernel end up being used by
> BusyBox ... keeping in mind that this thing is MMU-protected ... ?
If this is really coming from user space, then you have
inadvertantly stumbled on a security leak. I rather suspect
your kernel and/or toolchain must be broken for this level of
wierdness though. Tried using a staticly linked busybox to
rule out all the wierd stuff that shared libs introduce?
-Erik
--
Erik B. Andersen http://codepoet-consulting.com/
--This message was written using 73% post-consumer electrons--
More information about the busybox
mailing list