[PATCH alternative] sulogin: fix FEATURE_SHADOWPASSWDS sematics

[ChenQi]

On 11/17/2013 04:31 PM, Tito wrote:
> On Saturday 16 November 2013 22:55:16 Harald Becker wrote:
>> Hi Tito !
>>
>> On 16-11-2013 22:03 Tito <farmatito at tiscali.it> wrote:
>>> usually passwd sets password field to x in /etc/passwd if
>>> password field in /etc/shadow is used so in a correctly managed
>>> system there should not be empty password fields at all.
>>>
>>> I wonder at this point if it could be acceptable for sulogin to
>>> rather act the same way as libbb's correct_passwd function does
>>> in login and su to check in /etc/passwd if the passwd field is
>>> set to 'x' and use /etc/shadow only in that case.
>> You are correct, but there is another case which is required to
>> be checked. For some reasons there are programs setting password
>> in /etc/passwd to a single exclamation sign (!), which if used
>> as first character of password signals disabled account.
>>
>> It would be best to check password to be either empty or an entry
>> of at least 3 (may be more?) characters, with short entries
>> meaning not to accept any password entry (except on empty entry,
>> which signals no password).
> Hi,
>
> if ((pwd->pw_passwd[0] == 'x' || pwd->pw_passwd[0] == '*') && !pwd->pw_passwd[1])
>
> we do it the other way but same result
> (not empty, just one char equal to 'x' or '*' then use shadowpass)
>   

Hi All,

Sorry if I haven't stated the problem clearly in my commit message.
The problem is that the system might have no /etc/shadow file even if 
busybox has configured FEATURE_SHADOWPASSWDS. After all, whether to use 
the shadow is a system adm choice, right?

I see in other parts of busybox that if the shadow file is missing, the 
app falls back to use the passwd file, so I choose to fix it the same 
way. This is actually the behaviour of the sulogin program from sysvinit.

I think the above solution proposed by Tito is better, because if we 
have 'x' or '*' in /etc/passwd, and we don't have /etc/shadow, we're 
having an invalid user account.
Tito, could you please send out the patch?

Best Regards,
Chen Qi

> The escalmatio sign case is the one i pointed out and that has still to be implemented:
>
> 	"If the root account is locked (!), no password prompt is displayed
>   	and sulogin behaves as if the correct password were entered."
>
> 	This feature is missing in busybox: is it worth to be implemented?
>
> I think this is rather easy to implement once there is agreement on how
> to resolve the main issue.
>
>> --
>> Harald
>>
> Ciao,
> Tito
> _______________________________________________
> busybox mailing list
> busybox at busybox.net
> http://lists.busybox.net/mailman/listinfo/busybox
>
>