Module Core__Core_unix

Raised by the system calls below when an error is encountered. The first component is the error code; the second component is the function name; the third component is the string parameter to the function, if it has one, or the empty string otherwise.

raises Unix_error

with a given errno, function name and argument

handle_unix_error f runs f () and returns the result. If the exception Unix_error is raised, it prints a message describing the error and exits with code 2.

retry_until_no_eintr f returns f () unless f () fails with EINTR; in which case f () is run again until it raises a different error or returns a value.

Return the process environment, as an array of strings with the format ``variable=value''. The returned array is empty if the process has special privileges.

Unix.putenv ~key ~data sets the value associated to a variable in the process environment. key is the name of the environment variable, and data its new associated value.

unsetenv name deletes the variable name from the environment.

EINVAL name contained an ’=’ or an '\000' character.

The termination status of a process.

env is used to control the environment of a child process, and can take four forms. `Replace_raw replaces the entire environment with strings in the Unix style, like "VARIABLE_NAME=value". `Replace has the same effect as `Replace_raw, but using bindings represented as "VARIABLE_NAME", "value". `Extend adds entries to the existing environment rather than replacing the whole environment. `Override is similar to `Extend but allows unsetting variables too.

If env contains multiple bindings for the same variable, the last takes precedence. In the case of `Extend, bindings in env take precedence over the existing environment.

exec ~prog ~argv ?search_path ?env execs prog with argv. If use_path = true (the default) and prog doesn't contain a slash, then exec searches the PATH environment variable for prog. If env is supplied, it determines the environment when prog is executed.

The first element in argv should be the program itself; the correct way to call exec is:

exec ~prog ~argv:[ prog; arg1; arg2; ...] ()    

fork_exec ~prog ~argv ?use_path ?env () forks and execs prog with argv in the child process, returning the child PID to the parent. As in exec, the first element in argv should be the program itself.

fork () forks a new process. The return value indicates whether we are continuing in the child or the parent, and if the parent, includes the child's process id.

waitpid pid waits for child process pid to terminate, and returns its exit status. waitpid_exn is like waitpid, except it only returns if the child exits with status zero, and raises if the child terminates in any other way.

Execute the given command, wait until it terminates, and return its termination status. The string is interpreted by the shell /bin/sh and therefore can contain redirections, quotes, variables, etc. The result WEXITED 127 indicates that the shell couldn't be executed.

Return the pid of the process.

Return the pid of the parent process.

Return the pid of the parent process, if you're really sure you're never going to be the init process.

Get the numeric ID of the current thread, e.g. for identifying it in top(1).

Change the process priority. The integer argument is added to the ``nice'' value. (Higher values of the ``nice'' value mean lower priorities.) Return the new nice value.

File descriptor for standard input.

File descriptor for standard output.

File descriptor for standard standard error.

We can't use with sexp because pa_sexp inserts two copies of the val specs, which leads to a spurious "unused" warning.

file access rights

Open the named file with the given flags. Third argument is the permissions to give to the file if it is created. Return a file descriptor on the named file. Default permissions 0o644.

fcntl_getfl fd gets the current flags for fd from the open-file-descriptor table via the system call fcntl(fd, F_GETFL). See "man fcntl".

fcntl_setfl fd flags sets the flags for fd in the open-file-descriptor table via the system call fcntl(fd, F_SETFL, flags). See "man fcntl". As per the Linux man page, on Linux this only allows append and nonblock to be set.

Close a file descriptor.

with_file file ~mode ~perm ~f opens file, and applies f to the resulting file descriptor. When f finishes (or raises), with_file closes the descriptor and returns the result of f (or raises).

read fd buff ofs len reads len characters from descriptor fd, storing them in string buff, starting at position ofs in string buff. Return the number of characters actually read.

write fd buff ofs len writes len characters to descriptor fd, taking them from bytes buff, starting at position ofs in bytes buff. Return the number of characters actually written.

When an error is reported some characters might have already been written. Use single_write instead to ensure that this is not the case.

WARNING: write is an interruptible call and has no way to handle EINTR properly. You should most probably be using single write.

Same as write but with a string buffer.

Same as write but ensures that all errors are reported and that no character has ever been written when an error is reported.

Same as single_write but with a string buffer.

Create an input channel reading from the given descriptor. The channel is initially in binary mode; use set_binary_mode_in ic false if text mode is desired.

Create an output channel writing on the given descriptor. The channel is initially in binary mode; use set_binary_mode_out oc false if text mode is desired.

Return the descriptor corresponding to an input channel.

Return the descriptor corresponding to an output channel.

POSITIONING modes for UnixLabels.lseek.

Set the current position for a file descriptor

Truncates the named file to the given size.

Truncates the file corresponding to the given descriptor to the given size.

The informations returned by the UnixLabels.stat calls. The times are float number of seconds since the epoch; we don't use Time.t because Time depends on Unix, so the fix isn't so trivial. Same for Native_file.stats below.

Return the information for the named file.

Same as UnixLabels.stat, but in case the file is a symbolic link, return the information for the link itself.

Return the information for the file associated with the given descriptor.

This sub-module provides the normal OCaml Unix functions that deal with file size using native ints. These are here because, in general, you should be using 64bit file operations so that large files aren't an issue. If you have a real need to use potentially 31bit file operations (and you should be dubious of such a need) you can open this module

Commands for lockf.

lockf fd cmd size place a lock on a file_descr that prevents any other process from calling lockf successfully on the same file. Due to a limitation in the current implementation the length will be converted to a native int, potentially throwing an exception if it is too large.

flock fd cmd places or releases a lock on the fd as per the flock C call of the same name.

Return true if the given file descriptor refers to a terminal or console window, false otherwise.

Memory mapping of a file as a big array. map_file fd kind layout ~shared dims returns a big array of kind kind, layout layout, and dimensions as specified in dims.

The data contained in this big array are the contents of the file referred to by the file descriptor fd.

The optional pos parameter is the byte offset in the file of the data being mapped. It defaults to 0.

If shared is true, all modifications performed on the array are reflected in the file. This requires that fd be opened with write permissions. If shared is false, modifications performed on the array are done in memory only, using copy-on-write of the modified pages; the underlying file is not affected.

To adjust automatically the dimensions of the big array to the actual size of the file, the major dimension (that is, the first dimension for an array with C layout, and the last dimension for an array with Fortran layout) can be given as -1. map_file then determines the major dimension from the size of the file. The file must contain an integral number of sub-arrays as determined by the non-major dimensions, otherwise Failure is raised. If all dimensions of the big array are given, the file size is matched against the size of the big array. If the file is larger than the big array, only the initial portion of the file is mapped to the big array. If the file is smaller than the bigarray, the file is automatically grown to the size of the big array. This requires write permissions on fd.

Array accesses are bounds-checked, but the bounds are determined by the initial call to map_file. Therefore, you should make sure no other process modifies the mapped file while you're accessing it, or a SIGBUS signal may be raised. This happens, for instance, if the file is shrunk. Invalid_argument or Failure may be raised in cases where argument validation fails.

since

4.05.0

Removes the named file

Removes the named file or directory

rename old new changes the name of a file from old to new.

link ?force ~target ~link_name () creates a hard link named link_name to the file named target. If force is true, an existing entry in place of link_name will be unlinked. This unlinking may raise a Unix error, e.g. if the entry is a directory.

Change the permissions of the named file.

Change the permissions of an opened file.

Change the owner uid and owner gid of the named file.

Change the owner uid and owner gid of an opened file.

Set the process creation mask, and return the previous mask.

Check that the process has the given permissions over the named file.

Return a new file descriptor referencing the same file as the given descriptor.

dup2 ~src ~dst duplicates src to dst, closing dst if already opened.

Set the ``non-blocking'' flag on the given descriptor. When the non-blocking flag is set, reading on a descriptor on which there is temporarily no data available raises the EAGAIN or EWOULDBLOCK error instead of blocking; writing on a descriptor on which there is temporarily no room for writing also raises EAGAIN or EWOULDBLOCK.

Clear the ``non-blocking'' flag on the given descriptor. See UnixLabels.set_nonblock.

Set the ``close-on-exec'' flag on the given descriptor. A descriptor with the close-on-exec flag is automatically closed when the current process starts another program with one of the exec functions.

Clear the ``close-on-exec'' flag on the given descriptor. See UnixLabels.set_close_on_exec.

Create a directory. The permissions of the created directory are (perm & ~umask & 0777). The default perm is 0777.

Create a directory recursively. The permissions of the created directory are those granted by mkdir ~perm.

Remove an empty directory.

Change the process working directory.

Return the name of the current working directory.

Change the process root directory.

The type of descriptors over opened directories.

Open a descriptor on a directory

Return the next entry in a directory. Returns None when the end of the directory has been reached.

Same as readdir_opt except that it signals the end of the directory by raising End_of_file.

Reposition the descriptor to the beginning of the directory

Close a directory descriptor.

Create a pipe. The first component of the result is opened for reading, that's the exit to the pipe. The second component is opened for writing, that's the entrance to the pipe.

Create a named pipe with the given permissions.

create_process ~prog ~args forks a new process that executes the program prog with arguments args. The function returns the pid of the process along with file descriptors attached to stdin, stdout, and stderr of the new process. The executable file prog is searched for in the path. The new process has the same environment as the current process. Unlike in execve the program name is automatically passed as the first argument.

create_process_env ~prog ~args ~env as create_process, but takes an additional parameter that extends or replaces the current environment. No effort is made to ensure that the keys passed in as env are unique, so if an environment variable is set twice the second version will override the first. If argv0 is given, it is used (instead of prog) as the first element of the argv array passed to execve.

The exact program to execute is determined using the usual conventions. More precisely, if prog contains at least one '/' character then it is used as is (relative to working_dir if prog is a relative path, absolute otherwise). Note that working_dir defaults to the working directory of the current process, i.e. getcwd (). If prog contains no '/' character, then it is looked up in prog_search_path: for the first dir in prog_search_path such that Filename.concat dir prog exists and is executable, Filename.concat dir prog is the program that will be executed.

prog_search_path defaults to the list of directories encoded as a ':' separated list in the "PATH" environment variable of the current running process. If no such variable is defined, then

["/bin"; "/usr/bin"]

is used instead. Note that the "PATH" environment variable is looked up in the environment of the current running process, i.e. via getenv "PATH". Setting the "PATH" variable in the env argument of this function has no effect on how prog is resolved.

In a setuid or setgid program, or one which has inherited such privileges, reading of the PATH variable will return an empty result. If a search path is required then it should be provided explicitly using prog_search_path in such scenarios; alternatively, perhaps more satisfactorily, an absolute path should be given as prog.

High-level pipe and process management. These functions (with UnixLabels.open_process_out and UnixLabels.open_process) run the given command in parallel with the program, and return channels connected to the standard input and/or the standard output of the command. The command is interpreted by the shell /bin/sh (cf. system). Warning: writes on channels are buffered, hence be careful to call Caml.flush at the right times to ensure correct synchronization.

See UnixLabels.open_process_in.

See UnixLabels.open_process_in.

Similar to UnixLabels.open_process, but the second argument specifies the environment passed to the command. The result is a triple of channels connected to the standard output, standard input, and standard error of the command.

Close channels opened by UnixLabels.open_process_in, wait for the associated command to terminate, and return its termination status.

Close channels opened by UnixLabels.open_process_out, wait for the associated command to terminate, and return its termination status.

Close channels opened by UnixLabels.open_process, wait for the associated command to terminate, and return its termination status.

Close channels opened by UnixLabels.open_process_full, wait for the associated command to terminate, and return its termination status.

symlink ~target ~link_name creates the file link_name as a symbolic link to the file target. On Windows, this has the semantics using stat as described at: http://caml.inria.fr/pub/docs/manual-ocaml/libref/Unix.html

Read the contents of a link.

Wait until some input/output operations become possible on some channels. The three list arguments are a set of descriptors to check for reading, for writing, or for exceptional conditions. ~timeout is the maximal timeout. The result is composed of three sets of descriptors: those ready for reading, ready for writing, and over which an exceptional condition is pending.

Setting restart to true means that we want select to restart automatically on EINTR (instead of propagating the exception)...

Wait until a non-ignored, non-blocked signal is delivered.

The execution times (CPU times) of a process.

The type representing wallclock time and calendar date.

Return the current time since 00:00:00 GMT, Jan. 1, 1970, in seconds.

Same as time above, but with resolution better than 1 second.

Convert a time in seconds, as returned by UnixLabels.time, into a date and a time. Assumes UTC.

Convert a UTC time in a tm record to a time in seconds

Convert a time in seconds, as returned by UnixLabels.time, into a date and a time. Assumes the local time zone.

Convert a date and time, specified by the tm argument, into a time in seconds, as returned by UnixLabels.time. Also return a normalized copy of the given tm record, with the tm_wday, tm_yday, and tm_isdst fields recomputed from the other fields. The tm argument is interpreted in the local time zone.

Convert a date and time, specified by the tm argument, into a formatted string. See 'man strftime' for format options.

Given a format string, convert a corresponding string to a date and time See 'man strptime' for format options.

Schedule a SIGALRM signal after the given number of seconds.

Stop execution for the given number of seconds.

nanosleep f delays execution of the program for at least f seconds. The function can return earlier if a signal has been delivered, in which case the number of seconds left is returned. Any other failure raises an exception.

Return the execution times of the process.

Set the last access time (second arg) and last modification time (third arg) for a file. Times are expressed in seconds from 00:00:00 GMT, Jan. 1, 1970.

The three kinds of interval timers.

The type describing the status of an interval timer

Return the current status of the given interval timer.

setitimer t s sets the interval timer t and returns its previous status. The s argument is interpreted as follows: s.it_value, if nonzero, is the time to the next timer expiration; s.it_interval, if nonzero, specifies a value to be used in reloading it_value when the timer expires. Setting s.it_value to zero disable the timer. Setting s.it_interval to zero causes the timer to be disabled after its next expiration.

Return the user id of the user executing the process.

Return the effective user id under which the process runs.

Sets the real user id and effective user id for the process. Only use this when superuser. To setuid as an ordinary user, see Core_extended.Unix.seteuid.

Return the group id of the user executing the process.

Return the effective group id under which the process runs.

Set the real group id and effective group id for the process.

Structure of entries in the passwd database

Structure of entries in the groups database.

Return the login name of the user executing the process.

A representation of CIDR netmasks (e.g. "192.168.0.0/24") and functions to match if a given address is inside the range or not. Only IPv4 addresses are supported. Values are always normalized so the base address is the lowest IP address in the range, so for example to_string (of_string "192.168.1.101/24") = "192.168.1.0/24".

The type of socket domains.

The type of socket kinds, specifying the semantics of communications.

The type of socket addresses. ADDR_UNIX name is a socket address in the Unix domain; name is a file name in the file system. ADDR_INET(addr,port) is a socket address in the Internet domain; addr is the Internet address of the machine, and port is the port number.

sockaddr_blocking_sexp is like sockaddr, with of_sexp that performs DNS lookup to resolve Inet_addr.t.

Return the socket domain adequate for the given socket address.

Create a new socket in the given domain, and with the given kind. The third argument is the protocol type; 0 selects the default protocol for that kind of sockets.

Create a pair of unnamed sockets, connected together.

Accept connections on the given socket. The returned descriptor is a socket connected to the client; the returned address is the address of the connecting client.

Bind a socket to an address.

Connect a socket to an address.

Set up a socket for receiving connection requests. The integer argument is the number of pending requests that will be established and queued for accept. Depending on operating system, version, and configuration, subsequent connections may be refused actively (as with RST), ignored, or effectively established and queued anyway.

Because handling of excess connections varies, it is most robust for applications to accept and close excess connections if they can. To be sure the client receives an RST rather than an orderly shutdown, you can setsockopt_optint file_descr SO_LINGER (Some 0) before closing.

In Linux, for example, the system configuration parameters tcp_max_syn_backlog, tcp_abort_on_overflow, and syncookies can all affect connection queuing behavior.

The type of commands for shutdown.

Shutdown a socket connection. SHUTDOWN_SEND as second argument causes reads on the other end of the connection to return an end-of-file condition. SHUTDOWN_RECEIVE causes writes on the other end of the connection to return a closed pipe condition (SIGPIPE signal).

Return the address of the given socket.

Return the address of the host connected to the given socket.

The flags for UnixLabels.recv, UnixLabels.recvfrom, UnixLabels.send and UnixLabels.sendto.

Receive data from a connected socket.

Receive data from an unconnected socket.

Send data over a connected socket.

Same as send but with a string buffer.

Send data over an unconnected socket.

Same as sendto but with a string buffer.

The socket options that can be consulted with UnixLabels.getsockopt and modified with UnixLabels.setsockopt. These options have a boolean (true/false) value.

The socket options that can be consulted with UnixLabels.getsockopt_int and modified with UnixLabels.setsockopt_int. These options have an integer value.

The socket options that can be consulted with UnixLabels.getsockopt_optint and modified with UnixLabels.setsockopt_optint. These options have a value of type int option, with None meaning ``disabled''.

The socket options that can be consulted with UnixLabels.getsockopt_float and modified with UnixLabels.setsockopt_float. These options have a floating-point value representing a time in seconds. The value 0 means infinite timeout.

Return the current status of a boolean-valued option in the given socket.

Set or clear a boolean-valued option in the given socket.

Same as UnixLabels.getsockopt for an integer-valued socket option.

Same as UnixLabels.setsockopt for an integer-valued socket option.

Same as UnixLabels.getsockopt for a socket option whose value is an int option.

Same as UnixLabels.setsockopt for a socket option whose value is an int option.

Same as UnixLabels.getsockopt for a socket option whose value is a floating-point number.

Same as UnixLabels.setsockopt for a socket option whose value is a floating-point number.

Connect to a server at the given address. Return a pair of buffered channels connected to the server. Remember to call Caml.flush on the output channel at the right times to ensure correct synchronization.

``Shut down'' a connection established with UnixLabels.open_connection; that is, transmit an end-of-file condition to the server reading on the other side of the connection.

Establish a server on the given address. The function given as first argument is called for each connection with two buffered channels connected to the client. A new process is created for each connection. The function UnixLabels.establish_server never returns normally.

Return the name of the local host.

Address information returned by Unix.getaddrinfo.

addr_info_blocking_sexp is like addr_info, with of_sexp that performs DNS lookup to resolve Inet_addr.t.

Options to Unix.getaddrinfo.

getaddrinfo host service opts returns a list of Unix.addr_info records describing socket parameters and addresses suitable for communicating with the given host and service. The empty list is returned if the host or service names are unknown, or the constraints expressed in opts cannot be satisfied.

host is either a host name or the string representation of an IP address. host can be given as the empty string; in this case, the ``any'' address or the ``loopback'' address are used, depending whether opts contains AI_PASSIVE. service is either a service name or the string representation of a port number. service can be given as the empty string; in this case, the port field of the returned addresses is set to 0. opts is a possibly empty list of options that allows the caller to force a particular socket domain (e.g. IPv6 only, or IPv4 only) or a particular socket type (e.g. TCP only or UDP only).

Host and service information returned by Unix.getnameinfo.

Options to Unix.getnameinfo.

getnameinfo addr opts returns the host name and service name corresponding to the socket address addr. opts is a possibly empty list of options that governs how these names are obtained. Raise Caml.Not_found or Not_found_s if an error occurs.

Get a sockaddr from a hostname or IP, and a port

Set a timeout for a socket associated with an In_channel.t

Set a timeout for a socket associated with an Out_channel.t

exit_immediately exit_code immediately calls the exit system call with the given exit code without performing any other actions (unlike Caml.exit). Does not return.

mknod ?file_kind ?perm ?major ?minor path creates a filesystem entry. Note that only FIFO-entries are guaranteed to be supported across all platforms as required by the POSIX-standard. On Linux directories and symbolic links cannot be created with this function. Use Unix.mkdir and Unix.symlink instead there respectively.

raises Invalid_argument

if an unsupported file kind is used.

raises Unix_error

if the system call fails.

parameter file_kind

default = S_REG (= regular file)

parameter perm

default = 0o600 (= read/write for user only)

parameter major

default = 0

parameter minor

default = 0

I/O-vectors for scatter/gather-operations

Synchronize all filesystem buffers with disk.

Synchronize the kernel buffers of a given file descriptor with disk, but do not necessarily write file attributes.

readdir_ino_opt dh return the next entry in a directory (((filename, inode)). Returns None when the end of the directory has been reached.

Same as readdir_ino_opt except that it signals the end of the directory by raising End_of_file.

read_assume_fd_is_nonblocking fd ?pos ?len buf calls the system call read ASSUMING THAT IT IS NOT GOING TO BLOCK. Reads at most len bytes into buffer buf starting at position pos.

returns

the number of bytes actually read.

raises Invalid_argument

if buffer range out of bounds.

raises Unix_error

on Unix-errors.

parameter pos

= 0

parameter len

= String.length buf - pos

write_assume_fd_is_nonblocking fd ?pos ?len buf calls the system call write ASSUMING THAT IT IS NOT GOING TO BLOCK. Writes at most len bytes from buffer buf starting at position pos.

returns

the number of bytes actually written.

raises Invalid_argument

if buffer range out of bounds.

raises Unix_error

on Unix-errors.

parameter pos

= 0

parameter len

= String.length buf - pos

writev_assume_fd_is_nonblocking fd ?count iovecs calls the system call writev ASSUMING THAT IT IS NOT GOING TO BLOCK using count I/O-vectors iovecs.

returns

the number of bytes actually written.

raises Invalid_argument

if the designated ranges are invalid.

raises Unix_error

on Unix-errors.

writev fd ?count iovecs like writev_assume_fd_is_nonblocking, but does not require the descriptor to not block. If you feel you have to use this function, you should probably have chosen I/O-vectors that build on bigstrings, because this function has to internally blit the I/O-vectors (ordinary OCaml strings) to intermediate buffers on the C-heap.

returns

the number of bytes actually written.

raises Invalid_argument

if the designated ranges are invalid.

raises Unix_error

on Unix-errors.

pselect rfds wfds efds timeout sigmask like Core_unix.select but also allows one to wait for the arrival of signals.

System configuration

See 'man sysconf' for documentation.

Wrapper over sysconf function in C.

mkstemp prefix creates and opens a unique temporary file with prefix, automatically appending a suffix of .tmp. followed by six random characters to make the name unique. Unlike C's mkstemp, prefix should not include six X's at the end.

The file descriptor will have close-on-exec flag set if O_CLOEXEC flag is supported.

raises Unix_error

on errors.

mkdtemp prefix creates a temporary directory with prefix, automatically appending a suffix of .tmp. followed by six random characters to make the name unique.

raises Unix_error

on errors. k

Causes abnormal program termination unless the signal SIGABRT is caught and the signal handler does not return. If the SIGABRT signal is blocked or ignored, the abort() function will still override it.

getgrouplist user group returns the list of groups to which user belongs. See 'man getgrouplist'.

Return the list of groups to which the user executing the process belongs.

no system calls involved

See man page for wordexp.

See man page for uname.

if_indextoname ifindex If ifindex is an interface index, then the function returns the interface name. Otherwise, it raises Unix_error.

mcast_join ?ifname ?source sock addr join a multicast group at addr with socket sock, from source at source if specified, optionally using network interface ifname.

parameter ifname

default = any interface

mcast_leave ?ifname ?source sock addr leaves a multicast group at addr with socket sock, from source at source if specified, optionally using network interface ifname.

parameter ifname

default = any interface

get_mcast_ttl sock reads the time-to-live value of outgoing multicast packets for socket sock.

set_mcast_ttl sock ttl sets the time-to-live value of outgoing multicast packets for socket sock to ttl.

get_mcast_loop sock reads the boolean argument that determines whether sent multicast packets are looped back to local sockets.

set_mcast_loop sock loop sets the boolean argument that determines whether sent multicast packets are looped back to local sockets.

set_mcast_ifname sock "eth0" sets outgoing multicast traffic on IPv4 UDP socket sock to go out through interface eth0.

This uses setsockopt with IP_MULTICAST_IF and applies to multicast traffic. For non-multicast applications, see Linux_ext.bind_to_interface.

For keeping your memory in RAM, i.e. preventing it from being swapped out.

A network interface on the local machine. See man getifaddrs.