--- /dev/null
+_EXIT(2)
+## NAME
+_exit - exit a process freeing its resources
+## SYNOPSIS
+*#include <unistd.h>*
+
+*int _exit*(*int* status);
+## DESCRIPTION
+Upon calling this function the calling process immediately terminates and frees
+up resources. The status value is returned back to the waiting parent via the
+*waitpid* call.
+
+The _exit call does not perform any cleanup specified by *atexit*(3) nor do any
+standard I/O cleanup. Processes would normally exit via the *exit*(3) library
+function.
+
+When a process exits it frees its resources and then enters zombie state until
+the parent waits for it, unless the parent is ignoring notifications about child
+processes, in which case it exits completely and the reported status is lost.
+## RETURN VALUE
+This function does not return.
+## CONFORMING TO
+UZI, POSIX.1-2001
+##NOTES
+While the exit status argument is an *int* only the low 8bits of the status
+passed in exit will be provided to the parent.
+## SEE ALSO
+*exit*(3),*fork*(2),*waitpid*(2).
+
--- /dev/null
+CLOSE(2)
+## NAME
+close - close a file descriptor
+## SYNOPSIS
+*#include <unistd.h>*
+
+*int close*(*int* fd);
+## DESCRIPTION
+Close a file descriptor that was created with *open*(2), *pipe*(2), *socket*(2),
+or other calls that return a file descriptor. If this is the last reference to
+this file descriptor then any file locks created with *flock*(2) are closed.
+
+If *fd* is the last file descriptor referring to the underlying file table entry
+then the file table entry is freed. If this is the last reference to the file
+object then any resources are freed, devices will go inactive, and any file
+that has been removed using *unlink*(2) will be freed on disc.
+## RETURN VALUE
+On success, zero is returned. On error -1 is returned and errno is set.
+## ERRORS
+:*EBADF*
+ An invalid file descriptor was passed.
+:*EIO*
+ An I/O error occurred.
+## CONFORMING TO
+V7, UZI, POSIX.1-2001, POSIX.1-2008
+## NOTES
+The return of a *close*() call does not guarantee that the data is on the
+media. If this is required then a call to *sync*(2) is needed.
+## BUGS
+Many systems provide an *fsync*(2) and *fdatasync*(2) call to write back the
+data on just one file. In a small system like Fuzix this is not useful as
+*sync*(2) is just as good, but ought to be provided by wrappers in the C
+library.
+## SEE ALSO
+*flock*(2), *open*(2), *pipe*(2), *socket*(2), *sync*(2), *unlink*(2).
--- /dev/null
+LINK(2)
+## NAME
+link - add another path to a file
+## SYNOPSIS
+*#include <unistd.h>*
+
+*int link*(*const char \**from, *const char \**to);
+## DESCRIPTION
+*link*() adds a new path to an existing file object. The new path must be on
+the same file system as the file. Both paths reference the same object
+and all the permissions and properties of the file belong to the file not the
+name so are shared.
+
+A file is deleted when the final reference is removed with *unlink*(2) and
+nobody holds an open file handle to the file itself.
+## RETURN VALUE
+On success, zero is returned. On error -1 is returned and errno is set.
+## ERRORS
+:*EACCES*
+ Insufficient permission is available to create the new name.
+:*EFAULT*
+ The address passed for the path is invalid.
+:*EIO*
+ An I/O error occurred.
+:*EISDIR*
+ An attempt was made to add a link to a directory.
+:*EMLINK*
+ The new link would give the file more links than the filesystem permits.
+:*ENOENT*
+ The source does not exist, or the target directory does not exist.
+:*ENOMEM*
+ No memory was available.
+:*ENOSPC*
+ No disk space was available for the new name.
+:*ENOTDIR*
+ A path element before the final one was not a directory.
+:*EPERM*
+ There are insufficient permissions to perform the removal or the creation of
+ the paths.
+:*EROFS*
+ The file system is read-only.
+:*EXDEV*
+ The two paths are not on the same file system.
+## CONFORMING TO
+V7, UZI, POSIX.1-2001, POSIX.1-2008
+## SEE ALSO
+*rename*(2), *unlink*(2).
--- /dev/null
+\e[H\e[J
\ No newline at end of file
--- /dev/null
+OPEN(2)
+## NAME
+open - open an existing file or create a new one
+## SYNOPSIS
+*#include <fcntl.h>*
+
+*int open*(*const char \**path, *int* flags);
+
+*int open*(*const char \**path, *int* flags, *mode_t* mode);
+## DESCRIPTION
+The open call accesses the object given by the path, and returns a file descriptor
+referencing this object, which remains valid until the last *close*(2). This
+file descriptor is then used for other system calls while accessing the object.
+
+The file descriptor is a reference to a file table entry. The file table entry
+references the underlying object. Permissions are checked at open time only.
+An open descriptor also ensures the underlying object remains until the final
+*close*(2) even if removed from the filesystem by *unlink*(2). The file
+table entry is associated with an offset and status flags. Thus the offset
+is shared across descriptors duplicated via *fork*(2) or *dup*(2) and *dup2*(2)
+calls.
+
+The flags must indicate the access mode desired. The permitted modes are
+*O\_RDONLY*, *O\_WRONLY*, *O\_RDWR*. These request read only, write only, or
+read/write access to the object.
+
+The following additional flags can be included in the open
+:*O\_APPEND*
+All writes to the file extend the file. Each time a write is made the data
+is written to the end of the file and the file offset is set to the end of
+file.
+:*O\_CLOEXEC*
+ The file descriptor should be closed when *execve*(2) is invoked and the process
+ memory image is replaced by something new.
+:*O\_CREAT*
+ If the file does not exist then create it as a new empty file with the mode
+ calculated from the passed mode and the setting of *umask*(2). Without
+ this flag an open of a non-existing file reports an error.
+:*O\_DIRECT*
+ Hint that the file is best accessed directly without caching. Cache coherency
+ with other accessors will be maintained. Useful when large amounts of disk
+ I/O must be done for data that will be read or written once.
+:*O\_EXCL*
+ If specified with *O\_CREAT* this requires that a new file is created, and will
+ error if the file already exists.
+:*O\_NOCTTY*
+ If the pathname refers to a *tty*(4) device then do not make this the
+ controlling terminal even if it would normally do so.
+:*O\_NDELAY*
+ Indicate that neither the open nor any subsequent operation should wait but
+ instead return an error or partially complete.
+:*O\_SYNC*
+ Hint that the file should be written synchronously, so that the data is on
+ the media when I/O calls return.
+:*O\_TRUNC*
+ If the file already exists and is being opened for writing then truncate it
+ back to being empty.
+
+Many of these flags can be changed at runtime via the *fcntl*(2) call.
+## RETURN VALUE
+On success, a file descriptor is returned. On error -1 is returned and errno is set.
+## ERRORS
+:*EACCES*
+ Insufficient permission is available to move the file. This may be to remove
+ it from the old location or to create the new name.
+:*EAGAIN*
+ The open request would block and *O\_NDELAY* is specified.
+:*EEXIST*
+ The file already exists and *O\_CREAT* and *O\_EXCL* are set.
+:*EFAULT*
+ One of the addresses passed for the paths is invalid.
+:*EINTR*
+ An open of a device was interrupted by a *signal*(7).
+:*EIO*
+ An I/O error occurred.
+:*EISDIR*
+ An attempt was made to open a directory for writing.
+:*EMFILE*
+ The process already has it's maximum number of files open.
+:*ENFILE*
+ The system wide file descriptor table is full.
+:*ENODEV*
+ The pathname refers to a device node for a device that does not exist.
+:*ENOENT*
+ The source does not exist, or the target directory does not exist.
+:*ENOMEM*
+ No memory was available.
+:*ENOTDIR*
+ A path element before the final one was not a directory.
+:*ENXIO*
+ The pathname refers to a device node for a non existant class of device, or
+ an attempt wa smade to open a *fifo*(7) with *O\_NDELAY|O\_WRONLY* but there is no
+ reader.
+:*EPERM*
+ A device or similar special file is imposing it's own restrictions beyond the
+ filesystem permissions, and may not be opened.
+:*EROFS*
+ The file system is read-only.
+## CONFORMING TO
+4.3BSD, C89, C99, POSIX.1-2001, POSIX.1-2008
+## NOTES
+UZI and early Unix have a separate creat() system call which is semantically
+the same as open with the flags O\_CREAT|O\_WRONLY|O\_TRUNC. This is provided by
+the C library in FUZIX and modern systems.
+## BUGS
+Fuzix currently ignores *O\_SYNC*
+
+FUZIX does not yet implement the *ENXIO* error code for fifo devices.
+## SEE ALSO
+*close*(2), *creat*(3), *dup*(2), *dup2*(2), *execve*(2), *fcntl*(2), *umask*(2),
+*tty*(4), *fifo*(7)
--- /dev/null
+RENAME(2)
+## NAME
+rename - change the name or location of a file
+## SYNOPSIS
+*#include <stdio.h>*
+
+*int rename*(*const char \**from, *const char \**to);
+## DESCRIPTION
+*rename*() moves a filesystem object from one path to another within a single
+filesystem. If the target path already exists then it will be replaced if
+possible. As only the name is changed other links and open file handles are not
+affected.
+
+*rename*() is atomic in the sense that the transaction either completes or
+fails and the file has either the old or new name but never both barring
+media failure. It is not guaranteed that other processes cannot observe both
+the old or new names at the same time.
+## RETURN VALUE
+On success, zero is returned. On error -1 is returned and errno is set.
+## ERRORS
+:*EACCES*
+ Insufficient permission is available to move the file. This may be to remove
+ it from the old location or to create the new name.
+:*EFAULT*
+ One of the addresses passed for the paths is invalid.
+:*EINVAL*
+ An attempt was made to place a directory within itself and create a loop
+ in the file system.
+:*EIO*
+ An I/O error occurred.
+:*EISDIR*
+ The target is an existing directory.
+:*ENOENT*
+ The source does not exist, or the target directory does not exist.
+:*ENOMEM*
+ No memory was available.
+:*ENOSPC*
+ No disk space was available for the rename.
+:*ENOTDIR*
+ A path element before the final one was not a directory.
+:*EPERM*
+ There are insufficient permissions to perform the removal or the creation of
+ the paths.
+:*EROFS*
+ The file system is read-only.
+:*EXDEV*
+ The two paths are not on the same file system.
+## CONFORMING TO
+4.3BSD, C89, C99, POSIX.1-2001, POSIX.1-2008
+## NOTES
+Some operating systems permit a rename over an empty directory to succeed. Fuzix
+does not.
+
+This functionality is not present in UZI.
+## BUGS
+Fuzix does not currently permit the use of *rename*() to move a directory.
+## SEE ALSO
+*link*(2), *unlink*(2).
--- /dev/null
+UNLINK(2)
+## NAME
+unlink - remove a path to a file
+## SYNOPSIS
+*#include <unistd.h>*
+
+*int unlink*(*const char \**path);
+## DESCRIPTION
+*unlink*() removes a path to an existing file object. A file is deleted when
+the final reference is removed with *unlink*(2) and nobody holds an open
+file handle to the file itself.
+## RETURN VALUE
+On success, zero is returned. On error -1 is returned and errno is set.
+## ERRORS
+:*EACCES*
+ Insufficient permission is available to remove the name.
+:*EFAULT*
+ The address passed for the path is invalid.
+:*EINVAL*
+ An attempt was made to place a directory within itself and create a loop
+ in the file system.
+:*EIO*
+ An I/O error occurred.
+:*EISDIR*
+ An attempt was made to unlink a directory.
+:*ENOENT*
+ The path does not exist.
+:*ENOMEM*
+ No memory was available.
+:*ENOTDIR*
+ A path element before the final one was not a directory.
+:*EPERM*
+ There are insufficient permissions to perform the removal or the creation of
+ the paths.
+:*EROFS*
+ The file system is read-only.
+## CONFORMING TO
+V7, UZI, POSIX.1-2001, POSIX.1-2008
+## SEE ALSO
+*link*(2), *rename*(2), *rmdir*(2).
public / FUZIX.git / commitdiff