man-pages/man3p/pthread_create.3p

.\" Copyright (c) 2001-2003 The Open Group, All Rights Reserved 
.TH "PTHREAD_CREATE" 3P 2003 "IEEE/The Open Group" "POSIX Programmer's Manual"
.\" pthread_create 
.SH NAME
pthread_create \- thread creation
.SH SYNOPSIS
.LP
\fB#include <pthread.h>
.br
.sp
int pthread_create(pthread_t *restrict\fP \fIthread\fP\fB,
.br
\ \ \ \ \ \  const pthread_attr_t *restrict\fP \fIattr\fP\fB,
.br
\ \ \ \ \ \  void *(*\fP\fIstart_routine\fP\fB)(void*), void *restrict\fP
\fIarg\fP\fB); \fP
\fB
.br
\fP
.SH DESCRIPTION
.LP
The \fIpthread_create\fP() function shall create a new thread, with
attributes specified by \fIattr\fP, within a process. If
\fIattr\fP is NULL, the default attributes shall be used. If the attributes
specified by \fIattr\fP are modified later, the
thread's attributes shall not be affected. Upon successful completion,
\fIpthread_create\fP() shall store the ID of the created
thread in the location referenced by \fIthread\fP.
.LP
The thread is created executing \fIstart_routine\fP with \fIarg\fP
as its sole argument. If the \fIstart_routine\fP returns,
the effect shall be as if there was an implicit call to \fIpthread_exit\fP()
using
the return value of \fIstart_routine\fP as the exit status. Note that
the thread in which \fImain\fP() was originally invoked
differs from this. When it returns from \fImain\fP(), the effect shall
be as if there was an implicit call to \fIexit\fP() using the return
value of \fImain\fP() as the exit status.
.LP
The signal state of the new thread shall be initialized as follows:
.IP " *" 3
The signal mask shall be inherited from the creating thread.
.LP
.IP " *" 3
The set of signals pending for the new thread shall be empty.
.LP
.LP
The alternate stack shall not be inherited. 
.LP
The floating-point environment shall be inherited from the creating
thread.
.LP
If \fIpthread_create\fP() fails, no new thread is created and the
contents of the location referenced by \fIthread\fP are
undefined.
.LP
If _POSIX_THREAD_CPUTIME is defined, the new thread shall have a CPU-time
clock accessible, and the initial value of this clock
shall be set to zero. 
.SH RETURN VALUE
.LP
If successful, the \fIpthread_create\fP() function shall return zero;
otherwise, an error number shall be returned to indicate
the error.
.SH ERRORS
.LP
The \fIpthread_create\fP() function shall fail if:
.TP 7
.B EAGAIN
The system lacked the necessary resources to create another thread,
or the system-imposed limit on the total number of threads
in a process {PTHREAD_THREADS_MAX} would be exceeded.
.TP 7
.B EINVAL
The value specified by \fIattr\fP is invalid.
.TP 7
.B EPERM
The caller does not have appropriate permission to set the required
scheduling parameters or scheduling policy.
.sp
.LP
The \fIpthread_create\fP() function shall not return an error code
of [EINTR].
.LP
\fIThe following sections are informative.\fP
.SH EXAMPLES
.LP
None.
.SH APPLICATION USAGE
.LP
None.
.SH RATIONALE
.LP
A suggested alternative to \fIpthread_create\fP() would be to define
two separate operations: create and start. Some
applications would find such behavior more natural. Ada, in particular,
separates the "creation" of a task from its
"activation".
.LP
Splitting the operation was rejected by the standard developers for
many reasons:
.IP " *" 3
The number of calls required to start a thread would increase from
one to two and thus place an additional burden on
applications that do not require the additional synchronization. The
second call, however, could be avoided by the additional
complication of a start-up state attribute.
.LP
.IP " *" 3
An extra state would be introduced: "created but not started". This
would require the standard to specify the behavior of the
thread operations when the target has not yet started executing.
.LP
.IP " *" 3
For those applications that require such behavior, it is possible
to simulate the two separate steps with the facilities that
are currently provided. The \fIstart_routine\fP() can synchronize
by waiting on a condition variable that is signaled by the start
operation.
.LP
.LP
An Ada implementor can choose to create the thread at either of two
points in the Ada program: when the task object is created,
or when the task is activated (generally at a "begin"). If the first
approach is adopted, the \fIstart_routine\fP() needs to
wait on a condition variable to receive the order to begin "activation".
The second approach requires no such condition variable
or extra synchronization. In either approach, a separate Ada task
control block would need to be created when the task object is
created to hold rendezvous queues, and so on.
.LP
An extension of the preceding model would be to allow the state of
the thread to be modified between the create and start. This
would allow the thread attributes object to be eliminated. This has
been rejected because:
.IP " *" 3
All state in the thread attributes object has to be able to be set
for the thread. This would require the definition of
functions to modify thread attributes. There would be no reduction
in the number of function calls required to set up the thread.
In fact, for an application that creates all threads using identical
attributes, the number of function calls required to set up
the threads would be dramatically increased. Use of a thread attributes
object permits the application to make one set of attribute
setting function calls. Otherwise, the set of attribute setting function
calls needs to be made for each thread creation.
.LP
.IP " *" 3
Depending on the implementation architecture, functions to set thread
state would require kernel calls, or for other
implementation reasons would not be able to be implemented as macros,
thereby increasing the cost of thread creation.
.LP
.IP " *" 3
The ability for applications to segregate threads by class would be
lost.
.LP
.LP
Another suggested alternative uses a model similar to that for process
creation, such as "thread fork". The fork semantics
would provide more flexibility and the "create" function can be implemented
simply by doing a thread fork followed immediately by
a call to the desired "start routine" for the thread. This alternative
has these problems:
.IP " *" 3
For many implementations, the entire stack of the calling thread would
need to be duplicated, since in many architectures there
is no way to determine the size of the calling frame.
.LP
.IP " *" 3
Efficiency is reduced since at least some part of the stack has to
be copied, even though in most cases the thread never needs
the copied context, since it merely calls the desired start routine.
.LP
.SH FUTURE DIRECTIONS
.LP
None.
.SH SEE ALSO
.LP
\fIfork\fP() , \fIpthread_exit\fP() , \fIpthread_join\fP() , the Base
Definitions volume of IEEE\ Std\ 1003.1-2001, \fI<pthread.h>\fP
.SH COPYRIGHT
Portions of this text are reprinted and reproduced in electronic form
from IEEE Std 1003.1, 2003 Edition, Standard for Information Technology
-- Portable Operating System Interface (POSIX), The Open Group Base
Specifications Issue 6, Copyright (C) 2001-2003 by the Institute of
Electrical and Electronics Engineers, Inc and The Open Group. In the
event of any discrepancy between this version and the original IEEE and
The Open Group Standard, the original IEEE and The Open Group Standard
is the referee document. The original Standard can be obtained online at
http://www.opengroup.org/unix/online.html .