[OE-core] [PATCH 1/1] base-passwd: disable problematic login.defs options

Scott Garman scott.a.garman at intel.com
Mon Jun 20 03:33:42 UTC 2011


On 06/19/2011 07:41 PM, Khem Raj wrote:
> On Sun, Jun 19, 2011 at 7:13 PM, Mark Hatle<mark.hatle at windriver.com>  wrote:
>> On 6/17/11 1:10 PM, Scott Garman wrote:
>>> On 06/17/2011 10:22 AM, Otavio Salvador wrote:
>>>> On Fri, Jun 17, 2011 at 17:19, Scott Garman<scott.a.garman at intel.com>    wrote:
>>>>> Sorry, I forgot to mention that shadow-utils-native is what is used to
>>>>> modify the passwd/group files in the target sysroot. It seems that having a
>>>>> -native recipe install files into a target sysroot would be worse than
>>>>> including an optional file with base-passwd that may or may not be used in
>>>>> target systems.
>>>>
>>>> Why not make an shadow-target package with this?
>>>
>>> To just install a login.defs file? I'm open to it if a few more people
>>> think this is a better idea.
>>
>> The file is needed in order for the utilities that add, remove and modify
>> users/groups to function properly.  The full version from shadow utils is used
>> so we are sure we can dead with both shadow-less and shadowed filesystem images.
>>   (It's also more full featured then busybox, yet busybox is still compatible
>> with it.)
>>
> Will shadow be able to override this file ?
> one thing I see is that it wont get any updates that shadow might
> do to this file in future.

Now that I think of it, the reason Koen ran into the error messages was 
that the login.defs I shipped with base-passwd had various variables 
uncommented that the shadow recipe comments out (there's a sed script 
included with shadow which does this). Which means that his image, which 
had shadow installed, was *not* overriding the login.defs from base-passwd.

I'm now convinced that creating a shadow-cross package which just ships 
a login.defs file is the right thing to do, and to remove it from 
base-passwd. Thanks everyone for the feedback thus far.

I will be away at a conference for most of this coming week, but will 
try to squeeze this in on Monday.

Scott

-- 
Scott Garman
Embedded Linux Engineer - Yocto Project
Intel Open Source Technology Center




More information about the Openembedded-core mailing list