Re: Problems with md-crypt

[Thomas Wuensche <wuensche@ems-wuensche.com>]

Frank schrieb:
>>But it seems as if the shell script was attempted to be executed,
>>but the exec just can't start the interpreter. I once waded through the
>>code to start an executable for an embedded project, but this is long
>>ago. However this would not depend on the device underlying a file
>>system, I guess.
> 
> 
> But can you start the interpreter (/bin/sh) directly? I suppose, you have 
> checked already, but better ask.
> $ /bin/sh
> sh-3.00$ echo "Hello"
> Hello
> sh-3.00$      

Yes, in the second example you can see that I start the script
with the interpreter, which leads to the first script being
executed and the second script complain about not being able
to start the interpreter. Problem seems to be the script being
executable and the interpreter (/bin/sh) being taken from within
the script.

If the interpreter is passed the script, there seems to be no
problem.

>>Maybe something is in the way that does not understand how to run
>>64 bit executables (the shell in this case). I'm rather clueless
>>right now.
> 
> 
> I don't understand this sentence. But if it was a problem with 64bit, I would 
> be experiencing it as well, wouldn't I?

You run a newer distro (I guess testing is no longer Sarge today, but
Sid?). What kernel version do you run?

My sentence above related to my experience with some older kernel
(around 2.4.10). There was some code in the kernel that would examine
the type of program being executed and start it in an appropriate way.
This was on uCLinux (MMU-less) in an embedded device. Hopefully it is
not required to dig through the kernel to fix this, I'm a little bit
out of practice with that ;)

Thomas