[Info-vax] OS implementation languages

[Stephen Hoffman]

On 2023-09-05 17:56:40 +0000, Simon Clubley said:

> In this context, it simply means the ability to support MAC security, 
> _including_ the ability to help keep a successful compromise contained, 
> which is what SELinux offers.

OpenVMS SEVMS-style and MLS-style mandatory access controls are 
somewhere between exceedingly difficult to administer and to use, and 
approximately useless. Bell-LaPadula-style MAC works well for its 
target of 1990s-era US Department of Defense and related app designs 
and server installations. US DoD itself largely avoided adopting MAC 
and MLS and went heavily to System High designs, too. Bell-LaPadula 
security usefulness for newer applications and environments and 
expectations? Not so much. This having spent time working on and using 
SEVMS.

Somewhat newer than MAC and MLS are jails (and sandboxes and ilk). 
These are usually based on mandatory access controls mechanisms, though 
are vastly more usable and adoptable than are traditional MAC or MLS.

https://docs.freebsd.org/en/books/handbook/jails/

Preferably that jail-related work all tied into the app installation 
tooling, to ease the load on the system administrators. Unfortunately 
for that preference, PCSI and related tooling, and the most recent 
major security enhancements work from OpenVMS V6.0 are all from well 
before Y2K, and have... not kept up with competing expectations.

VSI seems unlikely to overhaul the existing OpenVMS security anytime 
soon. At most, I'd expect to see VSI add a mechanism akin to pledge(2). 
Maybe with unveil(2) added.

What's pledge(2)?  Presentation by Theo de Raadt on pledge(2): 
https://www.youtube.com/watch?v=F_7S1eqKsFk

I don't foresee speedy developer adoption of pledge(2) on OpenVMS, 
assuming that API was even added.




-- 
Pure Personal Opinion | HoffmanLabs LLC