Skip to contentSkip to conversation

Security Working Group meeting - Wednesday April 27

DISCUSSION Joseph Reynolds 2 participants 3 messages Apr 27, 2022

lore
Share & tools
lore
#0 lore raw

This is a reminder of the OpenBMC Security Working Group meeting scheduled for this Wednesday April 27 at 10:00am PDT.

We'll discuss the following items on the agenda <https://docs.google.com/document/d/1b7x9BaxsfcukQDqbvZsU2ehMq4xoJRQvLxxsDUWmAOI>, and anything else that comes up:

1.

Access, agenda and notes are in the wiki:
https://github.com/openbmc/openbmc/wiki/Security-working-group <https://github.com/openbmc/openbmc/wiki/Security-working-group>

- Joseph

#1 lore raw

Re: Security Working Group meeting - Wednesday April 27 - results

Quoted 4 lines · Joseph Reynolds

This is a reminder of the OpenBMC Security Working Group meeting scheduled for this Wednesday April 27 at 10:00am PDT.

We'll discuss the following items on the agenda <https://docs.google.com/document/d/1b7x9BaxsfcukQDqbvZsU2ehMq4xoJRQvLxxsDUWmAOI>, and anything else that comes up:

Attendees: Joseph Reynolds, Ruud Haring, Dhananjay, Jiang Ziang, Daniil, Nirav Shah, Mark McCawley, Terry Duncan.

1 Followup to SELinux discussion from last time.

TODO Joseph: post the session recording and the presentation.

Note design in gerrit review https://gerrit.openbmc-project.xyz/c/openbmc/docs/+/53205 <https://gerrit.openbmc-project.xyz/c/openbmc/docs/+/53205>

We clarified the goal of the design is to make it easy for a system integrator to add SELinux to their BMC firmware image, and to set some basic SELinux policies which do not create “too many” denial event log entries when SELinux is switched to permissive mode.  The usefulness of this design is to collect data for policies needed to switch SELinux to enforcing mode.  It remains an open question what policies are generally useful to the OpenBMC community.

Dhananjay mentioned a SELinux policy analysis tool:

https://ossna2020.sched.com/event/ckpF/selint-an-selinux-policy-static-analysis-tool-daniel-burgener-microsoft <https://ossna2020.sched.com/event/ckpF/selint-an-selinux-policy-static-analysis-tool-daniel-burgener-microsoft>

https://www.youtube.com/watch?v=Gx5bxwvzN_Y <https://www.youtube.com/watch?v=Gx5bxwvzN_Y>

2 Is there a tie-in between Penetration testing and SELinux?

Note: Pen testing is performed by individual platforms, and the testing effort is not shared: only  vulnerabilities and fixes are shared.  Help wanted at the community level.

The idea is that the same kind of analysis is needed for both Pen testing and to make SELinux policy.  Can we share that analysis or develop it in the OpenBMC community?

3 Nirav Shah - Alternate idea: Use D-Bus session buses (vs the system bus).

Note that all OpenBMC applications use the D-Bus system bus, which only the root user is allowed to access.

Nirav presented an idea to change some applications to use a session bus (and away from the system bus).  Doing so allows BMC applications to run as non-root and makes it easier for different applications to communicate via D-bus APIs.

We believe this work is relatively independent of SELinux policy configuration.

Joseph

Quoted 5 lines

Access, agenda and notes are in the wiki:
https://github.com/openbmc/openbmc/wiki/Security-working-group <https://github.com/openbmc/openbmc/wiki/Security-working-group>

- Joseph

#2 lore raw

Re: Security Working Group meeting - Wednesday April 27 - results

Quoted 7 lines · Joseph Reynolds

On 4/27/22 7:31 AM, Joseph Reynolds wrote:

3 Nirav Shah - Alternate idea: Use D-Bus session buses (vs the system bus).

Note that all OpenBMC applications use the D-Bus system bus, which only the root user is allowed to access.

Quoted 7 lines

Nirav presented an idea to change some applications to use a session bus (and away from the system bus).  Doing so allows BMC applications to run as non-root and makes it easier for different applications to communicate via D-bus APIs.

We believe this work is relatively independent of SELinux policy configuration.

In my opinion, architectural changes like this fall pretty far outside of "Security" and need pretty wide consensus. It's fine if you want to use the Security Workgroup to refine the idea but I want to be clear that acceptance by the Security Workgroup isn't sufficient to move forward.

Signature · 1 line
Patrick Williams