The branch main has been updated by emaste:
URL:
https://cgit.FreeBSD.org/src/commit/?id=c54534e60263954fac4db5ebb041ed4597df1195
commit c54534e60263954fac4db5ebb041ed4597df1195
Author: Ed Maste <ema...@freebsd.org>
AuthorDate: 2025-06-14 17:57:33 +0000
Commit: Ed Maste <ema...@freebsd.org>
CommitDate: 2025-06-17 16:09:53 +0000
capsicum.4: Add some more detail from the Capsicum paper
Adapt some language from "Capsicum: practical capabilities for UNIX"
https://www.cl.cam.ac.uk/research/security/capsicum/papers/2010usenix-security-capsicum-website.pdf
Reviewed by: markj
Discussed with: rwatson
Event: Kitchener-Waterloo Hackathon 202506
Sponsored by: The FreeBSD Foundation
Differential Revision: https://reviews.freebsd.org/D50855
---
share/man/man4/capsicum.4 | 47 ++++++++++++++++++++++++++++++++++++++++++++++-
1 file changed, 46 insertions(+), 1 deletion(-)
diff --git a/share/man/man4/capsicum.4 b/share/man/man4/capsicum.4
index 6aefae9d6df2..31e6f02b8893 100644
--- a/share/man/man4/capsicum.4
+++ b/share/man/man4/capsicum.4
@@ -24,7 +24,7 @@
.\" OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
.\" SUCH DAMAGE.
.\"
-.Dd June 5, 2025
+.Dd June 17, 2025
.Dt CAPSICUM 4
.Os
.Sh NAME
@@ -37,6 +37,12 @@
.Nm
is a lightweight OS capability and sandbox framework implementing a hybrid
capability system model.
+.Nm
+is designed to blend capabilities with UNIX.
+This approach achieves many of the benefits of least-privilege operation, while
+preserving existing UNIX APIs and performance, and presents application authors
+with an adoption path for capability-oriented design.
+.Pp
Capabilities are unforgeable tokens of authority that can be delegated and must
be presented to perform an action.
.Nm
@@ -59,6 +65,34 @@ namespaces) is restricted; only explicitly delegated rights,
referenced by
memory mappings or file descriptors, may be used.
Once set, the flag is inherited by future children processes, and may not be
cleared.
+.Pp
+Access to system calls in capability mode is restricted: some system calls
+requiring global namespace access are unavailable, while others are
+constrained.
+For instance,
+.Xr sysctl 2
+can be used to query process-local information such as address space layout,
+but also to monitor a system’s network connections.
+.Xr sysctl 2
+is constrained by explicitly marking ≈60 of over 15000 parameters as permitted
+in capability mode; all others are denied.
+.Pp
+The system calls which require constraints are
+.Xr sysctl 2 ,
+.Xr shm_open 2
+.Pq which is permitted to create anonymous memory objects but not named ones
+and the
+.Xr openat 2
+family of system calls.
+The
+.Xr openat 2
+calls already accept a file descriptor argument as the directory to perform the
+.Xr open 2 ,
+.Xr rename 2 ,
+etc. relative to; in capability mode the
+.Xr openat 2
+family of system calls are constrained so that they can only operate on
+objects “under” the provided file descriptor.
.It capabilities
Limit operations that can be called on file descriptors.
For example, a file descriptor returned by
@@ -152,3 +186,14 @@ and
.An Kris Kennaway Aq Mt k...@freebsd.org
at Google, Inc., and
.An Pawel Jakub Dawidek Aq Mt pa...@dawidek.net .
+Portions of this manual page are drawn from
+.Rs
+.%A Robert N. M. Watson
+.%A Jonathan Anderson
+.%A Ben Laurie
+.%A Kris Kennaway
+.%T Capsicum: practical capabilities for UNIX
+.%J USENIX Security Symposium
+.%D August 2010
+.%O DOI: 10.5555/1929820.1929824
+.Re
