Skip to main content

SELinux: How to create a CIL from a TE file

For a project, I needed to compile and install a SELinux module only when necessary. Unfortunatelly, there is no versioning on stored modules; I can't check and compare with the source module. I came up with an idea: using MD5 signatures.

Modules are stored in /var/lib/selinux/Policy/active/modules/400/ModuleName/cil
The Policy can be found with sestatus:

# sestatus
SELinux status:                 enabled
SELinuxfs mount:                /sys/fs/selinux
SELinux root directory:         /etc/selinux
Loaded policy name:             targeted
...
  
Stored modules are actually compressed CIL, which can easily been dumped with bzcat:
# bzcat /var/lib/selinux/targeted/active/modules/400/my-httpd/cil
(typeattributeset cil_gen_require httpd_t)
(typeattributeset cil_gen_require default_t)
(typeattributeset cil_gen_require sysstat_log_t)
(typeattributeset cil_gen_require usr_t)
(allow httpd_t usr_t (file (write create unlink setattr)))
(allow httpd_t default_t (file (open read)))
(allow httpd_t sysstat_log_t (dir (add_name write)))
(allow httpd_t sysstat_log_t (file (open create)))
Piping the result to md5sum provides a signature.

The trickiest part is transforming the source (a .te file) to a CIL. A TE files looks like:

module my-httpd 1.2;

require {
        type httpd_t;
        type default_t;
        type sysstat_log_t;
        type usr_t;
        class dir { add_name write };
        class file { open read write create unlink setattr };
}
#============= httpd_t ==============
allow httpd_t usr_t:file { write create unlink setattr };
allow httpd_t default_t:file { open read };
allow httpd_t sysstat_log_t:dir { add_name write };
allow httpd_t sysstat_log_t:file { create open };
Hopefully, both syntax are similar, so the transformation can be achieved via sed with this simple program:
# remove semi-columns
s/;//g

# remove comments
s/#.*$//g        

# turn multiple space chars into one
s/[ \t]\+/ /g    

# remove leading and trailing spaces
s/^ \+\| \+$//   

# Turn { to (
s/ *{ */ (/      

# Turn } to )
s/ *}/)/         

# For each declaration of new type, add a line with the typeattribute,
# another line with the typeattributeset and a third line with the type.
s/^type \(.\+_t\), *\(.\+\)/(typeattribute \2)\n(typeattributeset \2 (\1))\n(type \2)/

# Turn the declaration of a existing type to a typeattributeset
s/^type \(.\+_t\)/(typeattributeset cil_gen_require \1)/

# Rule with 3 levels
s/^\(allow\|deny\|dontaudit\) *\(.\+_t\) \(.\+\):\(.\+\) \((.\+)\)/(\1 \2 \3 (\4 \5))/

# Rule with 4 levels
s/^\(allow\|deny\|dontaudit\) *\(.\+_t\) \(.\+\):\(.\+\) \(.\+\)/(\1 \2 \3 (\4 (\5)))/

# Remove any space before )
s/ \+)/)/g

# Remove lines not starting with (
/^(/!d           
Then, to turn the TE into CIL, simply do:
# sed -f te2cil.sed source-module.te 
However, another issue arose: the order of attributes within a class.
(allow ... (file (open create)))
will not provide the same signature as
(allow ... (file (create open)))
.
Sure enough, I could check all my sources and make sure the attributes are in alphabetical order, but I always prefer automation. So I came up with this awk file which does the trick:
NF>4 {  gsub(/\(|\)/,"");
        for(i=5;i<=NF;i++) A[i-5]=$i;
        n=asort(A);
        C="";
        for(i=5;i<=NF;i++)C=C" "A[i-4];
        delete A;
        print "("$1,$2,$3" ("$4,"("substr(C,2)")))"
}
Hence, getting the signature of the CIL becomes:
# sed -f te2cil.sed source-module.te | awk -f te2cil.awk | sort | md5sum

Finally, I make sure to sort the stored module as well:

# bzcat /var/lib/selinux/targeted/active/modules/400/my-httpd/cil | sed -f te2cil.sed | awk -f te2cil.awk | sort | md5sum

Comments

Popular posts from this blog

Drive replacement for Fostex DMT8-vl

The IDE hard drive on my Fostex DMT8-vl multitrack recorder shows signs of its imminent death; when getting hot, I could not record anymore. Must be said this drive comes from an old Sun Station, and has been replaced because I/O failures were detected by Solaris. It worked at least 5 years in my recorder: not so bad. However, time is now to replace it. The DMT8-vl is not able to handle drives bigger than 8.4 GB. Well, it is able to (the current drive is 15 GB), but only 8.4 GB will be usable. My tought was to use a 8 GB CompactFlash; having no moving parts means no noise, which is quite temptating for a music recording device. I purchased a CompactFlash-IDE adapter on the internet (8$) and I had to build a male-male IDE cable adapter (4$). Unfortunately, this doesn't work. The drive is correctly discovered by the operating system, which proposes to format it ("format IDE?"). After answering "yes", the formating runs pretty fast (faster than on a real drive), ...

Samba: Clients get "system error 1223" (or 123) after a server reboot

Facts: a Linux+Samba server shares anonymously a folder. After a reboot, Win clients could not attach the share drive anymore. C:\>net use \\mylinux\folder Enter the user name for 'mylinux': System error 1223 has occurred. The operation was canceled by the user. C:\>net view \\mylinux\ System error 123 has occurred. The filename, directory name, or volume label syntax is incorrect. The process are present, and tcpdump doesn't provide much information. What's going on? After hours of headscratching, the light came: the firewall was on and no rules for the Samba protocol! Grrr!

Issue with Soundpool MO4

I have a Atari STe with a Soundpool MO4 MIDI extension. It used to work very well, but unfortunatelly doesn't anymore: Cubase still detects it, and I can output MIDI to it but nothing is coming out from any MIDI Out. It took me a while to tackle it (lack of time, lack of tool, other items to play with), but I gave a glance last week-end. The parallel port on the Atari uses only the following signals: Pin 1 : Strobe (Atari -> MO4) Pin 2 : Data 0 (Atari -> MO4) Pin 3 : Data 1 (Atari -> MO4) Pin 4 : Data 2 (Atari -> MO4) Pin 5 : Data 3 (Atari -> MO4) Pin 6 : Data 4 (Atari -> MO4) Pin 7 : Data 5 (Atari -> MO4) Pin 8 : Data 6 (Atari -> MO4) Pin 9 : Data 7 (Atari -> MO4) Pin 11: Busy (MO4 -> Atari) The MO4 also decodes few other pins, but since the Atari doesn't, my guess is the MO4 was also targeted for PC. Inside the box, the MO4 is architectured around a CPLD (IspLSI1016 from Lattice) which contains the logi...