| 1 |
On Fri, 22 Apr 2011 08:00:35 +0200, Jörg Faschingbauer |
| 2 |
<joerg.faschingbauer@×××.at> wrote: |
| 3 |
> On Thu, 2011-04-21 at 16:18 -0500, chris@××××××××××××.com wrote: |
| 4 |
>> I've got a filesystem created with the following toolchain: (from |
| 5 |
>> crossdev) |
| 6 |
>> |
| 7 |
>> arm-unknown-linux-gnu- |
| 8 |
>> |
| 9 |
>> with |
| 10 |
>> |
| 11 |
>> arm-unknown-linux-gnu-emerge system |
| 12 |
>> arm-unknown-linux-gnu-emerge baselayout |
| 13 |
>> |
| 14 |
>> I created the kernel separately, and was able to boot it with |
| 15 |
>> another |
| 16 |
>> filesystem that was created by somebody else. |
| 17 |
>> |
| 18 |
>> During the boot process the kernel says "Attempted to kill init" |
| 19 |
>> because the /init program exited for some reason. |
| 20 |
>> The problem is that I can't get any other error from the init |
| 21 |
>> program. |
| 22 |
>> I have no idea why it's exiting. It could be a missing library, |
| 23 |
>> segfault |
| 24 |
>> or bad invocation, so I have no idea where to start. |
| 25 |
> |
| 26 |
> Provided that /init is an executable: |
| 27 |
> |
| 28 |
> If the instruction set that the root filesystem was compiled for |
| 29 |
> doesn't |
| 30 |
> match your architecture's, then it's an "illegal instruction" signal |
| 31 |
> that bites init. This usually happens quite early, before any |
| 32 |
> configuration error is seen. |
| 33 |
> |
| 34 |
> You can verify this by building your own busybox-only tree with the |
| 35 |
> same |
| 36 |
> toolchain you built the kernel with. Choose a static build |
| 37 |
> (CONFIG_STATIC, "Busybox Settings"/"Build Options"/"blah static |
| 38 |
> blah"). |
| 39 |
> Copy init and all that it needs (glibc?) over from your real rootfs |
| 40 |
> into |
| 41 |
> your small tree. Boot the kernel into the small tree, passing it |
| 42 |
> init=/bin/sh (which is busybox). Call /init and see what happens. |
| 43 |
> |
| 44 |
> Alternatively, you could augment your tree with busybox, by |
| 45 |
> installing |
| 46 |
> it into it. Boot into the tree, but with init=/bin/sh (the augmented |
| 47 |
> busybox). |
| 48 |
> |
| 49 |
> Have fun, |
| 50 |
> Joerg |
| 51 |
|
| 52 |
You were right about the "Illegal Instruction". I booted the working |
| 53 |
filesystem and chrooted into the new one where I was able to play around |
| 54 |
with the "broken" init file. |
| 55 |
|
| 56 |
I'm guessing I stuck a bad flag in make.conf, since the same compiler |
| 57 |
was able to produce a working kernel. |
| 58 |
|
| 59 |
Thanks. |
Archives