| 1 |
Systemtap builds kernel modules. To insert a kernel module on a |
| 2 |
system, root access is needed. |
| 3 |
|
| 4 |
SECURITY MODEL |
| 5 |
============== |
| 6 |
|
| 7 |
Originally sudo(8) was used to grant root access. After compiling a |
| 8 |
new kernel module, stap ran "sudo staprun module_path". This worked, |
| 9 |
but required all systemtap users to have root access. Many sysadmins |
| 10 |
on enterprise systems do not have root access. |
| 11 |
|
| 12 |
So, a new security model was developed. To run the staprun program |
| 13 |
(which installs systemtap kernel modules), a user must be one of the |
| 14 |
following: |
| 15 |
|
| 16 |
* the root user; |
| 17 |
|
| 18 |
* a member of both 'stapdev' and 'stapusr' groups; or |
| 19 |
|
| 20 |
* a member of the 'stapusr' group. Members of the stapusr group can |
| 21 |
only use modules located in the /lib/modules/VERSION/systemtap |
| 22 |
directory (where VERSION is the output of "uname -r"). This |
| 23 |
directory must be owned by root and not be world writable. |
| 24 |
|
| 25 |
So, there are two classes of users: systemtap developers (the root user |
| 26 |
and members of the stapdev/stapusr groups) and systemtap users (members of |
| 27 |
only the stapusr group). Systemtap developers can compile and run any |
| 28 |
systemtap script. Systemtap users can only run "approved" |
| 29 |
pre-compiled modules located in /lib/modules/VERSION/systemtap. |
| 30 |
|
| 31 |
USAGE |
| 32 |
===== |
| 33 |
|
| 34 |
Here's the usage case. A systemtap developer hears of a problem on a |
| 35 |
production machine (which doesn't have a compiler or kernel debuginfo |
| 36 |
installed). So, he write a systemtap script to probe certain areas of |
| 37 |
the kernel that will give him a better idea of what is going on. He |
| 38 |
develops the script on a development machine (that has the compiler |
| 39 |
and kernel debuginfo installed). Once he is satisfied with the |
| 40 |
systemtap script, he creates the systemtap kernel module and copies it |
| 41 |
to /lib/modules/VERSION/systemtap on the target production machine. |
| 42 |
He then asks a systemtap user on that machine to run the module and |
| 43 |
report the results. |
| 44 |
|
| 45 |
The above usage case would look something like this: |
| 46 |
|
| 47 |
On the development machine: |
| 48 |
# vi pmod.stp |
| 49 |
(The systemtap developer writes the systemtap script.) |
| 50 |
|
| 51 |
# stap -m pmod pmod.stp |
| 52 |
(The systemtap developer compiles and runs the script. If necessary, |
| 53 |
the script may need to be edited to fix any errors.) |
| 54 |
|
| 55 |
# scp pmod.ko prod_machine:/lib/modules/`uname -r`/systemtap |
| 56 |
(The systemtap developer copies the compiled kernel module to the proper |
| 57 |
directory on the production machine. Of course other methods - ftp, |
| 58 |
nfs, etc. could be used to transfer the module.) |
| 59 |
|
| 60 |
On the production machine: |
| 61 |
$ staprun pmod |
| 62 |
(The systemtap user runs the newly developed systemtap kernel module.) |
| 63 |
|
| 64 |
There are (at least) 2 different usage scenarios for the |
| 65 |
/lib/modules/VERSION/systemtap directory. |
| 66 |
|
| 67 |
1) Most restrictive usage. If only root should be able to add |
| 68 |
"approved" systemtap modules to /lib/modules/VERSION/systemtap, the |
| 69 |
permissions should be 755, like this: |
| 70 |
|
| 71 |
drwxr-xr-x 2 root root 4096 2007-08-07 13:54 systemtap/ |
| 72 |
|
| 73 |
2) More permissive usage. If all systemtap developers should be able |
| 74 |
to add "approved" systemtap modules to /lib/modules/VERSION/systemtap, |
| 75 |
its permissions should be 775 (and be owned by root, group stapdev), |
| 76 |
like this: |
| 77 |
|
| 78 |
drwxrwxr-x 2 root stapdev 4096 2007-08-07 13:54 systemtap/ |
| 79 |
|
| 80 |
INTERNALS |
| 81 |
========= |
| 82 |
|
| 83 |
To accomplish the new security model, staprun has been split into two |
| 84 |
programs: staprun and stapio. |
| 85 |
|
| 86 |
Here is a description of a typical systemtap session. The staprun |
| 87 |
program is a setuid program that does some system setup, loads the |
| 88 |
kernel module, then runs stapio (and waits for it to finish). The |
| 89 |
stapio program runs as the invoking user and is responsible for all |
| 90 |
communication with the kernel module. After the script runs to |
| 91 |
completion, stapio fork/execs staprun -d to unload the kernel module. |
| 92 |
|
| 93 |
staprun is a setuid program. It holds on to the root privileges only |
| 94 |
for the least amount of time (as required to verify/load compiled |
| 95 |
kernel module files). It invokes only stapio, and only as the |
| 96 |
original (unprivileged) user. |
| 97 |
|