UNIX(4)

📦
version 15.1 · 4 versions
, in freebsd 15.1, nixpkgs unstable

💅 UNIX(4) — unix: UNIX-domain protocol family 💅

🌸 NAME 🌸

unix — the UNIX-domain protocol family 🦋

📜 SYNOPSIS 📜

#include <sys/types.h>
#include <sys/un.h>

🌈 DESCRIPTION 🌈

Okay so the Unix-domain protocol family is a collection of protocols that gives you local (on-machine) interprocess communication through the normal socket(2) mechanisms. It’s giving main-character IPC energy, no cap ✨

The Unix-domain family supports the SOCK_STREAM, SOCK_SEQPACKET, and SOCK_DGRAM socket types and uses file system pathnames for addressing.

🎀 ADDRESSING 🎀

Unix-domain addresses are variable-length file system pathnames of at most 104 characters. The include file <sys/un.h> defines this address:

struct sockaddr_un {
        u_char  sun_len;
        u_char  sun_family;
        char    sun_path[104];
};

Binding a name to a Unix-domain socket with bind(2) causes a socket file to be created in the file system. This file is not removed when the socket is closed — unlink(2) must be used to remove the file. yes bestie, it just sits there 💀

Prior to binding a socket, fchmod(2) can be used to set the permissions of the socket file. This avoids the race that would otherwise occur between creation of the file and a subsequent call to chmod(2). Once the socket is bound to a file name, the permissions of the file can not be changed this way.

The length of Unix-domain address, required by bind(2) and connect(2), can be calculated by the macro SUN_LEN() defined in <sys/un.h>. The sun_path field must be terminated by a NUL character to be used with SUN_LEN(), but the terminating NUL is not part of the address.

The Unix-domain protocol family does not support broadcast addressing or any form of “wildcard” matching on incoming messages. All addresses are absolute- or relative-pathnames of other Unix-domain sockets. Normal file system access-control mechanisms are also applied when referencing pathnames; e.g., the destination of a connect(2) or sendto(2) must be writable.

💌 CONTROL MESSAGES 💌

Unix-domain sockets support the communication of Unix file descriptors and process credentials through the use of the msg_control field in the msg argument to sendmsg(2) and recvmsg(2). The items to be passed are described using a struct cmsghdr that is defined in the include file <sys/socket.h>.

To send file descriptors, the type of the message is SCM_RIGHTS, and the data portion of the messages is an array of integers representing the file descriptors to be passed. The number of descriptors being passed is defined by the length field of the message; the length field is the sum of the size of the header plus the size of the array of file descriptors.

The received descriptor is a duplicate of the sender's descriptor, as if it were created via dup(fd) or fcntl(fd, F_DUPFD_CLOEXEC, 0) depending on whether MSG_CMSG_CLOEXEC is passed in the recvmsg(2) call. Descriptors that are awaiting delivery, or that are purposely not received, are automatically closed by the system when the destination socket is closed.

Credentials of the sending process can be transmitted explicitly using a control message of type SCM_CREDS with a data portion of type struct cmsgcred, defined in <sys/socket.h> as follows:

struct cmsgcred {
  pid_t cmcred_pid;             /* PID of sending process */
  uid_t cmcred_uid;             /* real UID of sending process */
  uid_t cmcred_euid;            /* effective UID of sending process */
  gid_t cmcred_gid;             /* real GID of sending process */
  short cmcred_ngroups;         /* number of groups */
  gid_t cmcred_groups[CMGROUP_MAX];     /* groups */
};

The sender should pass a zeroed buffer which will be filled in by the system.

The group list is truncated to at most CMGROUP_MAX GIDs.

🚨 security tea: the process ID cmcred_pid should not be looked up (such as via the KERN_PROC_PID sysctl) for making security decisions. The sending process could have exited and its process ID already been reused for a new process. fr fr do not do it 🙈

🛠️ SOCKET OPTIONS 🛠️

UNIX domain sockets support a number of socket options for the options level SOL_LOCAL, which can be set with setsockopt(2) and tested with getsockopt(2):

💖 LOCAL_CREDS 💖

This option may be enabled on SOCK_DGRAM, SOCK_SEQPACKET, or a SOCK_STREAM socket. This option provides a mechanism for the receiver to receive the credentials of the process calling write(2), send(2), sendto(2) or sendmsg(2) as a recvmsg(2) control message. The msg_control field in the msghdr structure points to a buffer that contains a cmsghdr structure followed by a variable length sockcred structure, defined in <sys/socket.h> as follows:

struct sockcred {
  uid_t sc_uid;         /* real user id */
  uid_t sc_euid;        /* effective user id */
  gid_t sc_gid;         /* real group id */
  gid_t sc_egid;        /* effective group id */
  int   sc_ngroups;     /* number of supplemental groups */
  gid_t sc_groups[1];   /* variable length */
};

The current implementation truncates the group list to at most CMGROUP_MAX groups.

The SOCKCREDSIZE() macro computes the size of the sockcred structure for a specified number of groups. The cmsghdr fields have the following values:

cmsg_len = CMSG_LEN(SOCKCREDSIZE(ngroups))
cmsg_level = SOL_SOCKET
cmsg_type = SCM_CREDS

On SOCK_STREAM and SOCK_SEQPACKET sockets credentials are passed only on the first read from a socket, then the system clears the option on the socket.

This option and the above explicit struct cmsgcred both use the same value SCM_CREDS but incompatible control messages. If this option is enabled and the sender attached a SCM_CREDS control message with a struct cmsgcred, it will be discarded and a struct sockcred will be included.

🚨 more security tea: many setuid programs will write(2) data at least partially controlled by the invoker, such as error messages. Therefore, a message accompanied by a particular sc_euid value should not be trusted as being from that user. periodt 💅

🦋 LOCAL_CREDS_PERSISTENT 🦋

This option is similar to LOCAL_CREDS, except that socket credentials are passed on every read from a SOCK_STREAM or SOCK_SEQPACKET socket, instead of just the first read. Additionally, the msg_control field in the msghdr structure points to a buffer that contains a cmsghdr structure followed by a variable length sockcred2 structure, defined in <sys/socket.h> as follows:

struct sockcred2 {
  int   sc_version;     /* version of this structure */
  pid_t sc_pid;         /* PID of sending process */
  uid_t sc_uid;         /* real user id */
  uid_t sc_euid;        /* effective user id */
  gid_t sc_gid;         /* real group id */
  gid_t sc_egid;        /* effective group id */
  int   sc_ngroups;     /* number of supplemental groups */
  gid_t sc_groups[1];   /* variable length */
};

The current version is zero.

The cmsghdr fields have the following values:

cmsg_len = CMSG_LEN(SOCKCRED2SIZE(ngroups))
cmsg_level = SOL_SOCKET
cmsg_type = SCM_CREDS2

The LOCAL_CREDS and LOCAL_CREDS_PERSISTENT options are mutually exclusive. you cannot have both, bestie 😭

👑 LOCAL_PEERCRED 👑

Requested via getsockopt(2) on a SOCK_STREAM or SOCK_SEQPACKET socket returns credentials of the remote side. These will arrive in the form of a filled in xucred structure, defined in <sys/ucred.h> as follows:

struct xucred {
  u_int cr_version;             /* structure layout version */
  uid_t cr_uid;                 /* effective user id */
  short cr_ngroups;             /* number of groups */
  gid_t cr_groups[XU_NGROUPS];  /* groups */
  pid_t cr_pid;                 /* process id of the sending process */
};

The cr_version fields should be checked against the XUCRED_VERSION define.

The credentials presented to the server (the listen(2) caller) are those of the client when it called connect(2); the credentials presented to the client (the connect(2) caller) are those of the server when it called listen(2). This mechanism is reliable; there is no way for either party to influence the credentials presented to its peer except by calling the appropriate system call (e.g., connect(2) or listen(2)) under different effective credentials.

To reliably obtain peer credentials on a SOCK_DGRAM socket refer to the LOCAL_CREDS socket option.

✨ BUFFERING ✨

Okay so here's a fun lil quirk bestie — because Unix-domain sockets are literally local, they do not implement send buffers. None. Zero. Zip. 🎀 So when you call the send(2) and write(2) families of system calls, they just go straight for the destination socket's receive buffer and try to write your data there. It's giving "no middleman" energy, honestly iconic.

The default buffer sizes for SOCK_STREAM and SOCK_SEQPACKET Unix-domain sockets can be configured with the net.local.stream and net.local.seqpacket branches of the sysctl(3) MIB respectively. Cute, right? 🦋 Just a heads up though — setting the send buffer size (sendspace) affects only the maximum write size. That's it. That's the whole job. Don't ask it for more, she's doing her best. 💅

Now for the SOCK_DGRAM girlies: those sockets are unreliable and always non-blocking for write operations. Not a bug, just her vibe, no cap. You can configure the default receive buffer with net.local.dgram.recvspace, and the maximum allowed datagram size is limited by net.local.dgram.maxdgram. 🌸

A SOCK_DGRAM socket that's been bound with bind(2) can have multiple peers connected at the same time — main character energy, she's juggling. 🙈 The modern FreeBSD implementation will allocate net.local.dgram.recvspace sized private buffers in the receive buffer of the bound socket for every connected socket, which prevents that one situation where a single writer can just exhaust all of the buffer space. Rude behavior, prevented. ✨

Messages coming from unconnected sends using sendto(2) land on the shared buffer of the receiving socket, which has the same size limit. And here's the side effect you need to know about, bestie: the implementation doesn't guarantee that writes from different senders will arrive at the receiver in the same chronological order they were sent. Yep. Timeline chaos. 💀 The order is preserved for writes coming through a particular connection though, so at least each connection keeps her own story straight. 🫶

👑 SEE ALSO 👑

Also, for the academically inclined besties, there's these two iconic texts 📚:

  • “An Introductory 4.3 BSD Interprocess Communication Tutorial”, PS1, 7.
  • “An Advanced 4.3 BSD Interprocess Communication Tutorial”, PS1, 8.

Okay that's the whole page, bestie — you ate that 💖✨ This one's from the Debian crew, dated October 31, 2024, and it's the UNIX(4) manual page. Now go forth and socket responsibly 🎀🦋💜

📦 where this page comes from