Re: noautologin & live-getty
2009/1/31 Cyril Jaquier <cyril.jaquier@jaqpot.net>:
>>> if [ -z "${NOAUTOLOGIN}" ] && [ -n "${USERNAME}" ]
>>> then
>>> if [ ! -z "${LIVE_GETTY}" ]
>>> then
>>>
>>> ...
>>>
>>> What is the reason not to permit this? Maybe I missed something... But
>>> this is not really a problem as I will provide my own /etc/inittab
>>> because I need only one getty.
>>
>> what you are saying is 'why do we only allow serial console login if
>> login is not disabled?'? otherwise i don't get the questions/problem.
>>
>
> I meant: why do we only allow serial console login if auto-login is
> enabled? Having "live-getty" and "noautologin" at the same time does not
> give me the opportunity to login using an username/password.
>
> I do not want the "auto-login" feature because someone could connect a
> computer to the serial port and thus would be directly able to type in
> commands. When connected to the serial port, I want the user to enter
> valid credentials first.
>
> I hope it is a bit more clear now ;) Thank you.
Hi Cyril
Thanks for the clarification... That is a usage case I had not
envisaged when I wrote live-getty and auto serial port code.
I think I have a way to re-factor the code so that a getty is added
for the serial port if the kernel is using a serial console solving
your prolem... Then later we can rewrite that line to to use
live-getty if that is what is required by the user (ie solves my usage
case)... then we can all be happy.
I'll try and submit a patch tomorrow... but depends how hard I party tonight!
The code as writen is careful to check if you have a custom serial
port line and not amend it... so by editing the /etc/inittab at the
chroot stage is also a solution to your problem... as you already
know.
Maybe we should also put in some code so that users placing a custom
inittab in the chroot stage can add a magic comment to their
/etc/inittab so that 25configure_init leaves the custom inittab
alone.
Any comments on this idea welcome!
Regards
Alex Owen
Reply to: