distlib.html

Vxworks API操作系统和驱动程序设计API。压缩的HTML文件

By default, the panic hook writes the panic message to standard output. 
The <b>panic(&nbsp;)</b> routine must not return.  The prototype of the <b>panic(&nbsp;)</b> routine 
should look like this:
<pre>   void panic (char *panicMsg);
</pre>
<dt>
<b>DIST_CTL_RETRY_TIMEOUT</b>
<dd>
This function sets the initial send retry timeout in clock ticks. If no 
ACK is received within a timeout period, the packet is resent. The default
value and granularity of <b>DIST_CTL_RETRY_TIMEOUT</b> is system dependent.
<table>
<tr valign=top>
<th align=left>vxWorks Version </th><th align=left> Default Value </th><th align=left> Granularity</tr>
<tr><td colspan="3"><hr></tr>
<tr valign=top>
<td align=left>5.4 and below </td><td align=left> 1000ms </td><td align=left> 500ms</tr>
<tr valign=top>
<td align=left>5.5 and AE </td><td align=left> 200ms </td><td align=left> 100ms</tr>
<tr valign=top>
<td align=left></tr>
</tr>

</table>
<b>DIST_CTL_RETRY_TIMEOUT</b> is designated in ticks, but rounded down to a multiple
of the system's granularity.  The timeout period for the <i>n</i>th send is: 
<p>
&nbsp;&nbsp;<i>n</i>&nbsp;*&nbsp;<b>DIST_CTL_RETRY_TIMEOUT</b>
<p>
<dt>
<b>DIST_CTL_MAX_RETRIES</b>
<dd>
This function sets a limit for the number of retries when sending fails.
The default value is system dependent, but is set to 5 for all current
versions of vxWorks.
<p>
<dt>
<b>DIST_CTL_NACK_SUPPORT</b>
<dd>
This function enables or disables the sending of negative acknowledgments 
(NACKs).  NACKs are used to request a resend of a single
missing fragment from a packet. They are sent immediately after a fragment
is found to be missing.  If <i>arg</i> is FALSE (0), the sending of negative
acknowledgments is disabled.  If <i>arg</i> is TRUE (1), the sending of NACKs
is enabled. By default, NACKs are enabled.
<p>
<dt>
<b>DIST_CTL_PGGYBAK_UNICST_SUPPORT</b>
<dd>
This function enables or disables unicast piggy-backing.  When unicast 
piggy-backing is enabled, the system waits some time until it sends 
an acknowledgment for a previously received packet. In the meantime, if a
data packet is sent to a host already awaiting an acknowledgment, the
acknowledgment is delivered (that is, piggy-backed) with the data packet.
Enabling piggy-backing is useful for reducing the number of packets sent;
however, it increases latency if no data packets are sent while the system 
waits.  When unicast piggy-backing is disabled, an acknowledgment is 
delivered immediately in its own packet.  This function turns piggy-backing 
on and off for unicast communication only.  If <i>arg</i> is FALSE (0), unicast 
piggy-backing is disabled.  If <i>arg</i> is TRUE (1), unicast piggy-backing is 
enabled.  By default, piggy-backing is disabled for unicast communication.
<p>
<dt>
<b>DIST_CTL_PGGYBAK_BRDCST_SUPPORT</b>
<dd>
This function enables or disables broadcast piggy-backing.  When broadcast
piggy-backing is enabled, the system waits some time until it 
sends an acknowledgment for a previously received packet.  In the meantime,
if a data packet is sent to a host already awaiting an acknowledgment, the
acknowledgment is delivered (that is, piggy-backed)with the data packet.
Enabling piggy-backing is useful for reducing the number of packets sent;
however, it increases latency if no broadcast data packets are sent while the
system waits.  When broadcast piggy-backing is disabled, an 
acknowledgment is delivered immediately in its own packet.  This function 
turns piggy-backing on and off for broadcast communication only.  If <i>arg</i> is 
FALSE (0), broadcast piggy-backing is disabled.  If <i>arg</i> is TRUE (1), 
broadcast piggy-backing is enabled.  By default, piggy-backing is disabled 
for broadcast communication.
<p>
<dt>
<b>DIST_CTL_OPERATIONAL_HOOK</b>
<dd>
This function adds a routine to a list of routines to be called each time a 
node shifts to the operational state.  A maximum of 8 routines can be added to
the list.  The prototype of each <b>operational(&nbsp;)</b> routine should look as 
follows:
<pre>   void operational (DIST_NODE_ID nodeStateChanged);
</pre>
<dt>
<b>DIST_CTL_CRASHED_HOOK</b>
<dd>
This function adds a routine to a list of routines to be called each time a 
node shifts to the crashed state.  A node shifts to the crashed state when 
it does not acknowledge a message within the maximum number of retries.  
The list can contain a maximum of 8 routines; however VxFusion supplies one 
routine, leaving room for only 7 user-supplied routines.  The prototype of 
each <b>crashed(&nbsp;)</b> routine should look as follows:
<pre>   void crashed (DIST_NODE_ID nodeStateChanged);
</pre>
<dt>
<b>DIST_CTL_GET_LOCAL_ID</b>
<dd>
This function returns the local node ID.
<p>
<dt>
<b>DIST_CTL_GET_LOCAL_STATE</b>
<dd>
This function returns the state of the local node.
<p>
<dt>
<b>DIST_CTL_SERVICE_HOOK</b>
<dd>
This function sets a routine to be called each time a service fails, for a 
service invoked by a remote node. The <i>argument</i> parameter is a pointer to
a <b>servError(&nbsp;)</b> routine.  The prototype of the <b>servError(&nbsp;)</b> routine should
look as follows:
<pre>   void servError (int servId, int status);
</pre>
The system is aware of the following services:
<pre>DIST_ID_MSG_Q_SERV     (0)  /* message queue service       */
DIST_ID_MSG_Q_GRP_SERV (1)  /* message queue group service */
DIST_ID_DNDB_SERV      (2)  /* distributed name database   */
DIST_ID_DGDB_SERV      (3)  /* distributed group database  */
DIST_ID_INCO_SERV      (4)  /* incorporation protocol      */
DIST_ID_GAP_SERV       (5)  /* group agreement protocol    */
</pre>
<dt>
<b>DIST_CTL_SERVICE_CONF</b>
<dd>
This function configures a specified service. The <i>argument</i> parameter is
a pointer to a <b>DIST_SERV_CONF</b> structure which holds the service ID and its
configuration to be set.  <b>DIST_SERV_CONF</b> is defined as follows:
<pre>typedef struct
    {
    int servId;     /* ID of service to configure  */
    int taskPrio;   /* priority of service task    */
    int netPrio;    /* network priority of service */
    } DIST_SERV_CONF;
</pre>
The system is aware of the following services:
<pre>DIST_ID_MSG_Q_SERV     (0)  /* message queue service       */
DIST_ID_MSG_Q_GRP_SERV (1)  /* message queue group service */
DIST_ID_DNDB_SERV      (2)  /* distributed name database   */
DIST_ID_DGDB_SERV      (3)  /* distributed group database  */
DIST_ID_INCO_SERV      (4)  /* incorporation protocol      */
DIST_ID_GAP_SERV       (5)  /* group agreement protocol    */
</pre>
If one of the configuration parameters is -1, it remains unchanged.
The parameter <i>taskPrio</i> can range from 0 to 255; <i>netPrio</i> can range from 
0 to 7.
<p>
A service's configuration can be changed at any time.
</dl>

</blockquote><h4>AVAILABILITY</h4><blockquote><p>
This routine is distributed as a component of the unbundled distributed
message queues option, VxFusion.
<p>
</blockquote><h4>RETURNS</h4><blockquote><p>
OK or the value requested if <i>function</i> is known;
ERROR if <i>function</i> is unknown or the argument is invalid.
<p>
</blockquote><h4>ERRNO</h4><blockquote><p>
<dl>
<dt>
<b>S_distLib_UNKNOWN_REQUEST</b>
<dd>
The control function is unknown.
</dl>

</blockquote><h4>SEE ALSO</h4><blockquote><p>
<b><a href="./distLib.html#top">distLib</a></b>

</body>
</html>