github.com/prirun.p50em
1
0
Fork 0
mirror of https://github.com/prirun/p50em.git synced 2026-01-26 19:51:58 +00:00
Code Issues Releases Wiki Activity

devpnc: don't print disconnected message in error.log when we were

Browse Source 
trying to connect but failed.  Only print them when we actually were
connected.
This commit is contained in:
Jim
2011-08-17 16:28:58 -04:00
parent 1bc93f304a
commit 8200cff70d
1 changed files with 25 additions and 11 deletions
Download Patch File Download Diff File Expand all files Collapse all files

36
devpnc.h
View File

@@ -47,19 +47,32 @@
the Primenet config node-to-node password becomes redundant and
should probably be left blank for ease in configuration. Cool, eh?
Prior to rev 19.3, each node sends periodic "timer" messages to all
nodes it knows about. If the message returns ACK'd, the remote node
is up. Otherwise, the remote node must be down. This is not very
efficient, so beginning with 19.3, a broadcast timer message is sent
every 10 seconds to let all nodes know that a machine is up. On a
100-node network, instead of sending 100*99=9900 timer messages,
only 100 are required. Unfortunately, the emulated PNC does not
have the efficient broadcast mechanism of a physical token ring, so
broadcasts must be simulated by sending packets to each node. It
UNIMPLEMENTED IDEA 1: the IP address in ring.cfg could be specified
as 0.0.0.0:0 or --- to mean "this node is incoming only, don't try
to connect to it, but let it connect to me". This might be useful
for nodes that are usually offline, to keep devpnc from sending them
connect packets every 30 seconds.
UNIMPLEMENTED IDEA 2: when a connect comes in, we could/should check
that the IP address in ring.cfg matches the IP address of the
connect. Then again, that means you can't network an emulator when
away from your home location.
Prior to rev 19.3, each Primenet node sends periodic "timer"
messages to all nodes it knows about. If the message returns ACK'd,
the remote node is up. Otherwise, the remote node must be down.
This is not very efficient, so beginning with 19.3, a broadcast
timer message is sent every 10 seconds to let all nodes know that a
machine is up. On a 100-node network, instead of sending
100*99=9900 timer messages, only 100 are required. Unfortunately,
the emulated PNC does not have the efficient broadcast mechanism of
a physical token ring, so broadcasts must be simulated by sending
packets to each node, getting us back to the inefficient method. It
would be possible to store the first broadcast timer message
received from each node, and simply loop them all back within each
emulator every 10 seconds as long as the TCP/IP connection is up.
Not sure it's worth the trouble though...
Not sure it's worth the trouble though... At later revs, the timer
message has important version information.
In early version of Primos, PNC ring buffers (physical packets) are
256 words, with later versions of Primos also supporting 512, and
@@ -391,7 +404,8 @@ char * pncdumppkt(unsigned char * pkt, int len) {
unsigned short pncdisc(nodeid) {
if (ni[nodeid].cstate > PNCCSDISC) {
fprintf(stderr, "devpnc: disconnect from node %d\n", nodeid);
if (ni[nodeid].cstate > PNCCSCONN)
fprintf(stderr, "devpnc: disconnect from node %d\n", nodeid);
TRACE(T_RIO, " pncdisc: disconnect from node %d\n", nodeid);
close(ni[nodeid].fd);
ni[nodeid].cstate = PNCCSDISC;