gno/usr.man/man2/fork.2

.\"
.\" Devin Reade, 1997
.\"
.\" $Id: fork.2,v 1.2 1998/01/25 18:03:31 gdr-ftp Exp $
.\"
.\" .TH FORK2 2 GNO "System Calls" "16 December 1996"
.TH FORK 2 GNO "16 January 1997" "System Calls"
.SH NAME
.BR fork ,
.BR fork2 ,
.BR vfork 
\- start a new process from inside the current application space
.SH SYNOPSIS
#include <unistd.h>
.sp 1
int
\fBfork\fR (void *\fIproc\fR);
.br
int
\fBvfork\fR (void *\fIproc\fR);
.br
int
\fBfork2\fR (void *\fIproc\fR, unsigned short \fIstack\fR,
unsigned short \fIprio\fR, char *\fIprocname\fR,
unsigned short \fInumargs\fR, ...);
.SH DESCRIPTION
.BR fork
causes the creation of a new process.
.LP
.BR fork 's
argument
.IR proc
is typically the address of a C function, although it can be any valid
address inside the IIgs RAM space.  In a successful
.BR fork
call, the parent process resumes execution by returning from 
.BR fork 
and, under GNO, the child process resumes execution at the entry into function
.IR proc .
This is different from traditional Unix implementations.
.LP
.BR fork
creates a new entry in the process table, and sets up default settings
for the new process.  The process is allocated 1k (1024 bytes) of stack
space, and the direct page is set to the beginning of this memory.  The
process is executed in 16-bit full native mode, and the registers upon
entry to the routine are set as follows:
.RS
.nf

A	the userID assigned to the new process
X	0
Y	0
.fi
.RE
.LP
The child inherits the memory shadowing and machine state parameters of
the parent, as well as signal blocking information and the ID of the 
controlling TTY. In addition, the child inherits all the open
files of its parent.
.LP
A forked process may share code with other children or the parent.
However, this is only allowed in a forward manner; any forked process
that exits by function return will be terminated.  Note that any shared
global variables will need to be moderated with some type of mutual
exclusion, either the kernel
.BR semaphore (2)
routines or custom routines.  This includes C stdio routines.
.LP
There is no way to pass parameters directly to a child with
.BR fork .
Use
.BR fork2
instead.
.LP
Under GNO,
.BR vfork
is identical to 
.BR fork ;
other than the 
.IR proc
parameter (and therefore the point at which the child process resumes
execution), 
.BR vfork
is the same as the BSD implementation in that the two processes
share an address space.
.LP
.B fork2
is similar to
.BR fork (2),
except that it allows parameters to be passed both to 
.IR proc
and the execution environment:
.RS
.IP \fIstack\fR
is the number of bytes of stack to allocate to the proess. If
.I stack
is not a multiple of 256, then it
is rounded up to the next highest multiple of 256 bytes.
.IP \fIprio\fR
is the priority to assign to the process. Priorities are not currently
implemented, and you should pass 0 for this argument.
.IP \fIprocname\fR
is a string you can have associated with the process when viewing
the process table (See
.BR ps (1)).
.IP \fInumargs\fR
is the number of (16-bit)
.B words
of arguments which follow, not the
number of arguments.
Any arguments following
.I numargs
are passed as parameters to the child's procedure.
.RE
See below for an example of the use of 
.BR fork2 .
.SH "RETURN VALUE"
On success 
.BR fork ,
.BR vfork ,
and
.BR fork2
return to the parent the process ID of the new (child) process.
On failure, -1 is returned and
.B errno
is set.
.SH CAVEATS
Care must be taken such that the child process does not call
.BR exit (3)
either directly or by function call, since
.BR exit
will flush and close standard I/O data structures.  Instead, if 
the child process does not or cannot 
.BR execve (2),
it should call
.BR _exit (3).
.LP
ORCA/C's stack checking and stack repair code should be turned off
when compiling routines containing these functions.  See 
.BR occ (1)
or the ORCA/C Reference Manual for details.
.LP
Because the
.BR fork
functions are called from the kernel (which is implemented as a user
tool set), the child function
.IR proc
must restore the data bank register if it is going to access global 
variables.  See the description of the
.BR databank
pragma in the ORCA/C Reference Manual for details.
.LP
While
.BR Splat!
is an excellent source level debugger, it is (at the time of this
writing) unable to handle the execution of 
.BR fork ,
.BR fork2 ,
or 
.BR vfork
calls.  If you attempt to do so, your machine will almost certainly
crash.  Otherwise, 
.BR Splat!
is strongly recommended as a powerful tool for GNO programmers using
C as their source language.  (For the record, the author of this manual
page receives no renumeration from the sales of 
.BR Splat! )
.SH CONFORMANCE
Most Unix
.BR fork s
take no parameters; they copy the entire address space of the calling
process and return witha different value in the parent and child.  Due
to hardware limitations, this sort of manipulation isn't possible on
the IIgs.  Unix programs utilizing
.BR fork
will have to be modified slightly to work under GNO.
.LP
The implementation of 
.BR fork
is therefore not  POSIX conforming, nor is the implementation of 
.BR vfork
conforming to traditional Unix implementations.
.SH EXAMPLE
.nf                            

int main (int argc, char **argv) {
   ...
   pid = fork2(proc1,1024,0,"sub-process",3,argc,argv);
   ...
   return 0;
}

void proc1(int argc, char *argv[])
{
   ...
}

.fi
.SH HISTORY
.B fork2
first appeared in XINU.
.LP
.B vfork
first appeared in 3.0BSD.
.SH "SEE ALSO"
.BR occ (1),
.BR exec (2),
.BR execve (2),
.BR semaphore (2),
.BR wait (2),
the
.IR "GNO Kernel Reference Manual" ,
and the
.IR "ORCA/C Reference Manual" .