''

author: rpj <rpj> 2005-05-05 14:18:27 +0000
committer: rpj <rpj> 2005-05-05 14:18:27 +0000
commit: 41f88a82b33cdb357c83b582381232733ed2d039 (patch)
tree: b3690e999b4d934fce7aab79b2693ab086a4ad46 /manual/pthread_kill.html
parent: b75686484d99c90317b64184ce5080093ee7d2a9 (diff)
1 files changed, 150 insertions, 0 deletions
diff --git a/manual/pthread_kill.html b/manual/pthread_kill.html
new file mode 100644
index 0000000..6933c8d
--- /dev/null
+++ b/manual/pthread_kill.html
@@ -0,0 +1,150 @@
+<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.0 Transitional//EN">
+<HTML>
+<HEAD>
+	<META HTTP-EQUIV="CONTENT-TYPE" CONTENT="text/html; charset=utf-8">
+	<TITLE>PTHREAD_SIGNAL(3) manual page</TITLE>
+	<META NAME="GENERATOR" CONTENT="OpenOffice.org 1.1.3  (Linux)">
+	<META NAME="CREATED" CONTENT="20050504;18500100">
+	<META NAME="CHANGED" CONTENT="20050504;23005800">
+	<!-- manual page source format generated by PolyglotMan v3.2, -->
+	<!-- available at http://polyglotman.sourceforge.net/ -->
+</HEAD>
+<BODY LANG="en-GB" BGCOLOR="#ffffff" DIR="LTR">
+<P><A HREF="#toc">Table of Contents</A></P>
+<H2><A HREF="#toc0" NAME="sect0">Name</A></H2>
+<P>pthread_sigmask, pthread_kill, sigwait - handling of signals in
+threads 
+</P>
+<H2><A HREF="#toc1" NAME="sect1">Synopsis</A></H2>
+<P><B>#include &lt;pthread.h&gt;</B> <BR><B>#include &lt;signal.h&gt;</B>
+</P>
+<P><B>int pthread_sigmask(int </B><I>how</I><B>, const sigset_t
+*</B><I>newmask</I><B>, sigset_t *</B><I>oldmask</I><B>);</B> 
+</P>
+<P><B>int pthread_kill(pthread_t </B><I>thread</I><B>, int </B><I>signo</I><B>);</B>
+</P>
+<P><B>int sigwait(const sigset_t *</B>set, <B>int</B> *sig);</P>
+<H2><A HREF="#toc2" NAME="sect2">Description</A></H2>
+<P><B>pthread_sigmask</B> changes the signal mask for the calling
+thread as described by the <I>how</I> and <I>newmask</I> arguments.
+If <I>oldmask</I> is not <B>NULL</B>, the previous signal mask is
+stored in the location pointed to by <I>oldmask</I>. <B>Pthreads-w32</B>
+implements this function but no other function uses the signal mask
+yet.</P>
+<P>The meaning of the <I>how</I> and <I>newmask</I> arguments is the
+same as for <B><SPAN STYLE="font-style: normal">sigprocmask</SPAN></B>(2).
+If <I>how</I> is <B>SIG_SETMASK</B>, the signal mask is set to
+<I>newmask</I>. If <I>how</I> is <B>SIG_BLOCK</B>, the signals
+specified to <I>newmask</I> are added to the current signal mask. If
+<I>how</I> is <B>SIG_UNBLOCK</B>, the signals specified to <I>newmask</I>
+are removed from the current signal mask. 
+</P>
+<P>Recall that signal masks are set on a per-thread basis, but signal
+actions and signal handlers, as set with <B>sigaction</B>(2), are
+shared between all threads. 
+</P>
+<P><B>pthread_kill</B> send signal number <I>signo</I> to the thread
+<I>thread</I>. <B>Pthreads-w32</B> only supports signal number 0,
+which does not send any signal but causes <B>pthread_kill</B> to
+return an error if <I>thread</I> is not valid.</P>
+<P><B>sigwait</B> suspends the calling thread until one of the
+signals in <I>set</I> is delivered to the calling thread. It then
+stores the number of the signal received in the location pointed to
+by <I>sig</I> and returns. The signals in <I>set</I> must be blocked
+and not ignored on entrance to <B>sigwait</B>. If the delivered
+signal has a signal handler function attached, that function is <I>not</I>
+called. <B>Pthreads-w32</B> implements this function as a
+cancellation point only - it does not wait for any signals and does
+not change the location pointed to by <I>sig</I>.</P>
+<H2><A HREF="#toc3" NAME="sect3">Cancellation</A></H2>
+<P><B>sigwait</B> is a cancellation point. 
+</P>
+<H2><A HREF="#toc4" NAME="sect4">Return Value</A></H2>
+<P>On success, 0 is returned. On failure, a non-zero error code is
+returned. 
+</P>
+<H2><A HREF="#toc5" NAME="sect5">Errors</A></H2>
+<P>The <B>pthread_sigmask</B> function returns the following error
+codes on error: 
+</P>
+<DL>
+	<DL>
+		<DT STYLE="margin-right: 1cm; margin-bottom: 0.5cm"><B>EINVAL</B> 
+		</DT><DD STYLE="margin-right: 1cm; margin-bottom: 0.5cm">
+		<I>how</I> is not one of <B>SIG_SETMASK</B>, <B>SIG_BLOCK</B>, or
+		<B>SIG_UNBLOCK</B> 
+		</DD></DL>
+</DL>
+<P>
+The <B>pthread_kill</B> function returns the following error codes on
+error: 
+</P>
+<DL>
+	<DL>
+		<DT STYLE="margin-right: 1cm; margin-bottom: 0.5cm"><B>EINVAL</B> 
+		</DT><DD STYLE="margin-right: 1cm; margin-bottom: 0.5cm">
+		<I>signo</I> is not a valid signal number or is unsupported.</DD><DT STYLE="margin-right: 1cm; margin-bottom: 0.5cm">
+		<B>ESRCH</B> 
+		</DT><DD STYLE="margin-right: 1cm; margin-bottom: 0.5cm">
+		the thread <I>thread</I> does not exist (e.g. it has already
+		terminated) 
+		</DD></DL>
+</DL>
+<P>
+The <B>sigwait</B> function never returns an error. 
+</P>
+<H2><A HREF="#toc6" NAME="sect6">Author</A></H2>
+<P>Xavier Leroy &lt;Xavier.Leroy@inria.fr&gt; 
+</P>
+<P>Modified by Ross Johnson for use with <A HREF="http://sources.redhat.com/pthreads-win32">Pthreads-w32</A>.</P>
+<H2><A HREF="#toc7" NAME="sect7">See Also</A></H2>
+<P> 
+</P>
+<H2><A HREF="#toc8" NAME="sect8">Notes</A></H2>
+<P>In any implementation, for <B>sigwait</B> to work reliably, the
+signals being waited for must be blocked in all threads, not only in
+the calling thread, since otherwise the POSIX semantics for signal
+delivery do not guarantee that it’s the thread doing the <B>sigwait</B>
+that will receive the signal. The best way to achieve this is to
+block those signals before any threads are created, and never unblock
+them in the program other than by calling <B>sigwait</B>. This works
+because all threads inherit their initial sigmask from their creating
+thread.</P>
+<H2><A HREF="#toc9" NAME="sect9">Bugs</A></H2>
+<P><B>Pthreads-w32</B> does not implement signals yet and so these
+routines have almost no use except to prevent the compiler or linker
+from complaining. <B>pthread_kill</B> is useful in determining if the
+thread is a valid thread, but since many threads implementations
+reuse thread IDs, the valid thread may no longer be the thread you
+think it is, and so this method of determining thread validity is not
+portable, and very risky. <B>Pthreads-w32</B> from version 1.0.0
+onwards implements pseudo-unique thread IDs, so applications that use
+this technique (but really shouldn't) have some protection.</P>
+<HR>
+<P><A NAME="toc"></A><B>Table of Contents</B></P>
+<UL>
+	<LI><P STYLE="margin-bottom: 0cm"><A HREF="#sect0" NAME="toc0">Name</A>
+		</P>
+	<LI><P STYLE="margin-bottom: 0cm"><A HREF="#sect1" NAME="toc1">Synopsis</A>
+		</P>
+	<LI><P STYLE="margin-bottom: 0cm"><A HREF="#sect2" NAME="toc2">Description</A>
+		</P>
+	<LI><P STYLE="margin-bottom: 0cm"><A HREF="#sect3" NAME="toc3">Cancellation</A>
+		</P>
+	<LI><P STYLE="margin-bottom: 0cm"><A HREF="#sect4" NAME="toc4">Return
+	Value</A> 
+	</P>
+	<LI><P STYLE="margin-bottom: 0cm"><A HREF="#sect5" NAME="toc5">Errors</A>
+		</P>
+	<LI><P STYLE="margin-bottom: 0cm"><A HREF="#sect6" NAME="toc6">Author</A>
+		</P>
+	<LI><P STYLE="margin-bottom: 0cm"><A HREF="#sect7" NAME="toc7">See
+	Also</A> 
+	</P>
+	<LI><P STYLE="margin-bottom: 0cm"><A HREF="#sect8" NAME="toc8">Notes</A>
+		</P>
+	<LI><P><A HREF="#sect9" NAME="toc9">Bugs</A> 
+	</P>
+</UL>
+</BODY>
+</HTML>
author	rpj <rpj>	2005-05-05 14:18:27 +0000
committer	rpj <rpj>	2005-05-05 14:18:27 +0000
commit	41f88a82b33cdb357c83b582381232733ed2d039 (patch)
tree	b3690e999b4d934fce7aab79b2693ab086a4ad46 /manual/pthread_kill.html
parent	b75686484d99c90317b64184ce5080093ee7d2a9 (diff)