Review Request 117125: start_kdeinit: Use capabilities instead of SUID

[David Faure]

-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://git.reviewboard.kde.org/r/117125/#review54461
-----------------------------------------------------------



src/start_kdeinit/start_kdeinit.c
<https://git.reviewboard.kde.org/r/117125/#comment38104>

    I believe the way to check that your patch works is to check that this open() call works. Just add
    
    qDebug() << "procfile=" << procfile << "errno=" << errno();
    
    and check that line with and without the set capabilities thing.
    
    
    I assume the same can be done with kcheckpass at some point too?


- David Faure


On March 27, 2014, 7:57 p.m., Hrvoje Senjan wrote:
> 
> -----------------------------------------------------------
> This is an automatically generated e-mail. To reply, visit:
> https://git.reviewboard.kde.org/r/117125/
> -----------------------------------------------------------
> 
> (Updated March 27, 2014, 7:57 p.m.)
> 
> 
> Review request for KDE Frameworks and David Faure.
> 
> 
> Bugs: https://bugzilla.novell.com/show_bug.cgi?id=862953
>     https://bugs.kde.org/show_bug.cgi?id=https://bugzilla.novell.com/show_bug.cgi?id=862953
> 
> 
> Repository: kinit
> 
> 
> Description
> -------
> 
> The issue came up on security review of kinit package (yes, same is valid for kdelibs4...)
> SUSE security team is not happy with kdeinit being SUID helper, thus capabilities are utilized first (if available)
> I've just tried to integrate the suggested patch (from the report) with the CMake bits
> 
> 
> Diffs
> -----
> 
>   CMakeLists.txt 335ed08 
>   cmake/COPYING-CMAKE-SCRIPTS PRE-CREATION 
>   cmake/FindCap.cmake PRE-CREATION 
>   src/config-kdeinit.h.cmake c89c713 
>   src/start_kdeinit/CMakeLists.txt 6bfc496 
>   src/start_kdeinit/start_kdeinit.c 3c733e7 
> 
> Diff: https://git.reviewboard.kde.org/r/117125/diff/
> 
> 
> Testing
> -------
> 
> Built:
> with setcap & libcap present - installed as advertised;
> without one/both of them - the old procedure is in place (using SUID for the helper)
> 
> I am not sure how to test the OOM killer, fortunately it never kicked in kdelibs4 variant, so can't also say did it work as planned before...
> 
> 
> Thanks,
> 
> Hrvoje Senjan
> 
>

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.kde.org/pipermail/kde-frameworks-devel/attachments/20140328/561ffce4/attachment.html>