I would appreciate any donations. Wishlist or send e-mail type donations to maekawa AT daemon-systems.org.
CLONE(2) System Calls Manual CLONE(2) NAME clone, __clone - spawn new process with options LIBRARY Standard C Library (libc, -lc) SYNOPSIS #include <sched.h> pid_t clone(int (*func)(void *arg), void *stack, int flags, void *arg); pid_t __clone(int (*func)(void *arg), void *stack, int flags, void *arg); DESCRIPTION The clone system call (and associated library support code) creates a new process in a way that allows the caller to specify several options for the new process creation. Unlike fork(2) or vfork(2), in which the child process returns to the call site, clone causes the child process to begin execution at the function specified by func. The argument arg is passed to the entry point, as a means for the parent to provide context to the child. The stack pointer for the child process will be set to stack. Note that the clone interface requires that the application know the stack direction for the architecture, and that the caller initialize the stack argument as appropriate for the stack direction. The flags argument specifies several options that control how the child process is created. The lower 8 bits of flags specify the signal that is to be sent to the parent when the child exits. The following flags may also be specified by bitwise-or'ing them with the signal value: CLONE_VM Share the virtual address space with the parent. The address space is shared in the same way as vfork(2). CLONE_FS Share the "file system information" with the parent. This include the current working directory and file creation mask. CLONE_FILES Share the file descriptor table with the parent. CLONE_SIGHAND Share the signal handler set with the parent. Note that the signal mask is never shared between the parent and the child, even if CLONE_SIGHAND is set. CLONE_VFORK Preserve the synchronization semantics of vfork(2); the parent blocks until the child exits. The clone call returns the pid of the child in the parent's context. The child is provided no return value, since it begins execution at a different address. If the child process's entry point returns, the value it returns is passed to _exit(2), and the child process exits. Note that if the child process wants to exit directly, it should use _exit(2), and not exit(3), since exit(3) will flush and close standard I/O channels, and thereby corrupt the parent process's standard I/O data structures (even with fork(2) it is wrong to call exit(3) since buffered data would then be flushed twice). Note that clone is not intended to be used for new native NetBSD applications. It is provided as a means to port software originally written for the Linux operating system to NetBSD. RETURN VALUES Same as for fork(2). ERRORS Same as for fork(2). SEE ALSO chdir(2), chroot(2), fork(2), sigaction(2), sigprocmask(2), umask(2), vfork(2), wait(2) HISTORY The clone() function call appeared in NetBSD 1.6. It is compatible with the Linux function call of the same name with respect to the described options. BUGS The NetBSD implementation of clone() does not implement the following flags that are present in the Linux implementation: ⊕ CLONE_CHILD_CLEARTID ⊕ CLONE_CHILD_SETTID ⊕ CLONE_IO ⊕ CLONE_NEWIPC ⊕ CLONE_NEWNET ⊕ CLONE_NEWNS ⊕ CLONE_NEWPID ⊕ CLONE_NEWUTS ⊕ CLONE_PARENT ⊕ CLONE_PARENT_SETTID ⊕ CLONE_PID ⊕ CLONE_PTRACE ⊕ CLONE_SETTLS ⊕ CLONE_STOPPED ⊕ CLONE_SYSVSEM ⊕ CLONE_THREAD ⊕ CLONE_UNTRACED NetBSD 8.0 May 4, 2010 NetBSD 8.0