Index · Directives systemd 262~devel

Name

sd-varlink — APIs for Varlink IPC

Synopsis

#include <systemd/sd-varlink.h>

pkg-config --cflags --libs libsystemd

Description

sd-varlink.h is part of libsystemd(3) and provides APIs for implementing Varlink IPC clients and services. See https://varlink.org/ for more information about Varlink IPC.

Varlink IPC uses JSON as marshalling format. The sd-varlink API relies on the sd-json(3) API for JSON serialization, deserialization and manipulation.

Canonical encoding rules: sd-varlink omits the ""parameters"" member on the wire in replies, errors, and notifications when there are no parameters to transmit. This reduces message size and avoids ambiguity. Receivers must be tolerant and accept any of the following encodings for the absence of parameters: an omitted ""parameters"" key (preferred), a JSON "null" value, or an empty object "{}". When decoding, sd-varlink treats JSON "null" as if the member was omitted.

The varlinkctl(1) tool makes the functionality implemented by sd-varlink available from the command line.

On kernels that support extended attributes on socket inodes (available since Linux 7.0), sd-varlink automatically tags the AF_UNIX socket inodes it creates and manages with the user.varlink extended attribute. The attribute records the role the socket plays in the Varlink communication, allowing other processes to discover and classify Varlink sockets, for example via varlinkctl list-sockets (see varlinkctl(1)).

Four distinct roles are defined:

Table 1. user.varlink extended attribute roles

Attribute valueMeaning
"client"Set on the connecting end of a connection, i.e. a socket created via socket() + connect(). Only for inodes on the anonymous socket file system ("sockfs").
"server"Set on the serving end of a connection, i.e. a socket obtained via accept() on a listening socket. Only for inodes on the anonymous socket file system ("sockfs").
"listen"Set on a listening socket, i.e. a socket created via socket() + listen(). Only for inodes on the anonymous socket file system ("sockfs").
"entrypoint"Set on the entrypoint socket inode, i.e. the socket node bound into the regular file system that clients connect to. Unlike the other three, this attribute lives on a real file-system inode (not on "sockfs") and is hence visible to any process that can resolve the socket path.

Note that for socket activated Varlink services it is recommended to set these extended attributes via XAttrEntryPoint=, XAttrListen= and XAttrAccept= settings on the socket units, see systemd.socket(5) for details. Specifically, for socket units for which Accept=no is set:

…
[Socket]
SocketMode=0644
XAttrEntryPoint=user.varlink=entrypoint
XAttrListen=user.varlink=listen
…

For socket units which have Accept=yes:

…
[Socket]
Accept=yes
SocketMode=0644
XAttrEntryPoint=user.varlink=entrypoint
XAttrListen=user.varlink=listen
XAttrAccept=user.varlink=server
…

Note that for Varlink sockets that shall not be world-accessible it is recommended to unset the relevant "w" bits in the socket access mode, but to keep the relevant "r" bits set. As the access check done by connect() looks for the "w" bits, this does not make the service unnecessarily accessible, however it has the benefit that the extended attributes that identify the socket as Varlink-related may still be read by other users (as that's restricted by the "r" bit). Or in other words, it's recommended to set "SocketMode=0644" for services that shall not be accessible to arbitrary users, rather than "SocketMode=0600".

Notes

Functions described here are available as a shared library, which can be compiled against and linked to with the libsystemd pkg-config(1) file.

The code described here uses getenv(3), which is declared to be not multi-thread-safe. This means that the code calling the functions described here must not call setenv(3) from a parallel thread. It is recommended to only do calls to setenv() from an early phase of the program when no other threads have been started.

Directories

When sd_varlink_connect_url() encounters a URL with a scheme that is not natively supported, it looks for a bridge helper binary named after the URL scheme in this directory. The binary is invoked the same way as "exec:" binaries but with the full URL passed as the first command line argument.

For example, if varlinkctl introspect https://example.com/ws/sockets/io.systemd.Hostname is called, varlinkctl will look for an executable /usr/lib/systemd/varlink-bridges/https and invoke it with "https://example.com/ws/sockets/io.systemd.Hostname" as its only argument.

See Also

systemd(1), sd-event(3), sd-json(3), varlinkctl(1), systemd.socket(5), sd-bus(3), pkg-config(1)