2.9BSD/usr/man/man2/vfork.2

.TH VFORK 2
.UC
.SH NAME
vfork \- spawn new process in an efficient way
.SH SYNOPSIS
.B vfork()
.SH DESCRIPTION
.I Vfork
can be used to create new processes without fully copying the address
space of the old process.
It is useful when the purpose of
.IR fork (2)
would have been to create a new system context for an
.IR exec .
.I Vfork
differs from
.I fork
in that the child borrows the parent's memory and thread of
control until a call to
.IR execve (2)
or an exit (either by a call to
.IR exit (2)
or abnormally.)
The parent process is suspended while the child is using its resources.
.PP
.I Vfork
returns 0 in the child's context and (later) the pid of the child in
the parent's context.
.PP
.I Vfork
can normally be used just like
.I fork.
It does not work, however, to return
from the procedure which called
.I vfork
while running in the child's context, since the eventual return from
.I vfork
in the parent would then return to a no longer existent stack frame.
Be careful, also, to call
.I _exit
rather than
.I exit
if you can't
.IR exec ,
since
.I exit
will flush and close standard I/O channels, and thereby mess up the
parent process's standard I/O data structures.
(Even with
.I fork
it is wrong to call
.I exit
since buffered data would then be flushed twice.)
.PP
Similarly when using the new signal mechanism of
.IR sigset (3)
mechanism be sure to call
.I sigsys
rather than
.IR signal (2).
.SH DIAGNOSTICS
Same as for
.IR fork .
.SH SEE ALSO
execve(2), fork(2), sigsys(2J), wait(2)
.SH BUGS
This call is not portable to most dialects of UNIX.
Users should not depend on the memory
sharing semantics of
.I vfork
as it could, in that case, be made synonymous to
.I fork.
.PP
To avoid a possible deadlock situation,
processes which are children in the middle
of a
.I vfork
are never sent SIGTTOU or SIGTTIN signals; rather,
output or
.IR ioctl s
are allowed
and input attempts result in an end-of-file indication.