Bug 1162811

I have found a solution: to build a binary version of setroubleshootd, using cython.
I wonder how to compare its performance with the python script. Does anybody have an idea?
I am attaching the patch & files. Note that the files are mainly a proof of concept: the binary built is currently working as a replacement of the daemon, but I have only implemented one option (-f). It was written 'in haste', so there is still work to do.

Version used: setroubleshoot-3.2.20-3

Created attachment 957174 [details]
Fix compilation errors

Fix cython compilation errors:
 - replace tabs by spaces;
 - remove unreechable part;
 - remove never-defined functions & variables un uuid.py

Created attachment 957175 [details]
cython module

Hasty conversion of the scrypt into a cython module.

Created attachment 957176 [details]
c code

Hasty written c code to launch the daemon.

Created attachment 957179 [details]
Makefile

The makefile used to build the binary daemon.
To build:
 - apply the patch;
 - copy the .c, .pyx files & Makefile in the setroubleshoot directory;
 - run 'make setroubleshootd_c'

To test:
 - copy the binary setroubleshootd_c in /usr/sbin/
 - chcon -v --reference /usr/sbin/setroubleshootd{,_c} 
 - modify /usr/share/dbus-1/system-services/org.fedoraproject.Setroubleshootd.service to use the binary daemon;
 - reboot

Created attachment 957243 [details]
c code

I finally found how to pass argv to the cython module.
Now, the binary daemon should be a perfect candidate as a replacement of the script daemon.
The makefile & cython module should be revised, though.

Created attachment 957244 [details]
cython module

The corresponding module.

Actually, only the c code & cython module have to be compiled to build the binary: it loads the python modules at runtime.
I though that they could be compiled and embeded, but strace shows it is not the case. So, there are probably no gain in performance: it just fixes the current bug.

Created attachment 957804 [details]
Makefile

In the end I cannot find a way to embed the python files.
So, this bug can be solved using only the attached .c file and .pyx cython module, to build a binary version of setroubleshootd using this Makefile.
The gain in performance is probably only at startup (around twice time faster).
Here some time comparison between the script and the stripped binary version:

$ time setroubleshootd -h
real: ~0.380s

$ time setroubleshootd_c -h
real: ~0.200s

(In reply to Alphonse Steiner from comment #0)
> Description of problem:
> After disabling the 'unconfined' module, the rule 
>     $ sesearch -A -C -s system_dbusd_t -p execute -t bin_t
>     Found 1 semantic av rules:
>         allow system_dbusd_t bin_t : file { read getattr execute open } ;
> disappears.
> Setroubleshootd failed to be activated after that, generating the AVC:
> -----
> type=AVC msg=audit(1415725718.861:15433): avc:  denied  { execute } for 
> pid=31406 comm="dbus-daemon-lau" name="python2.7" dev="vda" ino=803242
> scontext=system_u:system_r:system_dbusd_t:s0-s0:c0.c1023
> tcontext=system_u:object_r:bin_t:s0 tclass=file permissive=0
> -----
> 
> The activation is retried again and again, locking the cpu and filling the
> audit log with the same denial message.
> 
> This did not occur in fedora 19, where the rule was present independantly of
> the 'unconfined' module.
> 
> selinux-policy version: 3.13.1-92

Nice catch. This is caused by


commit 13e05612c9d08eb4a7a791965b2f6733c27deb10
Author: Miroslav Grepl <mgrepl>
Date:   Thu Sep 18 16:06:07 2014 +0200

    We want to have also unconfined D-bus services running as unconfined_service_t instead of initrc_t.

This bug appears to have been reported against 'rawhide' during the Fedora 22 development cycle.
Changing version to '22'.

More information and reason for this action is here:
https://fedoraproject.org/wiki/Fedora_Program_Management/HouseKeeping/Fedora22

Fedora 22 changed to end-of-life (EOL) status on 2016-07-19. Fedora 22 is
no longer maintained, which means that it will not receive any further
security or bug fix updates. As a result we are closing this bug.

If you can reproduce this bug against a currently maintained version of
Fedora please feel free to reopen this bug against that version. If you
are unable to reopen this bug, please file a new report against the
current release. If you experience problems, please add a comment to this
bug.

Thank you for reporting this bug and we are sorry it could not be fixed.