server/admin.c (acmd_addr): Don't assert about the address family.

[tripe] / server / tripe-admin.5.in

diff --git a/server/tripe-admin.5.in b/server/tripe-admin.5.in
index 9f64133..43f7725 100644 (file)
--- a/server/tripe-admin.5.in
+++ b/server/tripe-admin.5.in
@@ -663,12 +663,18 @@ given, seconds are assumed.
 .RE
 .SP
 .B "PORT"
+.RI [ family ]
 Emits an
 .B INFO
 line containing just the number of the UDP port used by the
 .B tripe
-server.  If you've allowed your server to allocate a port dynamically,
-this is how to find out which one it chose.
+server, for the given address
+.I family
+(or one chosen arbitrarily if omitted -- though
+.B tripe
+tries to use the same port number consistently so this is not a likely
+problem in practice).  If you've allowed your server to allocate a port
+dynamically, this is how to find out which one it chose.
 .SP
 .B "RELOAD"
 Instructs the server to recheck its keyring files.  The server checks
@@ -1079,6 +1085,13 @@ is available, which does not meet the stated requirements.
 .I tag
 is already the tag of an outstanding job.
 .SP
+.BI "unknown-address-family " afam
+(For
+.BR PORT .)
+The address family
+.I afam
+is unrecognized.
+.SP
 .BI "unknown-command " token
 The command
 .I token
@@ -1538,6 +1551,51 @@ The peer (apparently) sent a transport ping response whose id doesn't
 match any outstanding ping.  Maybe it was delayed for longer than the
 server was willing to wait, or maybe the peer has gone mad; or maybe
 there are bad people trying to confuse you.
+.SS "PRIVSEP warnings"
+These indicate problems with the privilege-separation helper process.
+(The server tries to drop its privileges when it starts up, leaving a
+privileged helper process behind which will create and hand over tunnel
+descriptors on request, but hopefully not do anything else especially
+dangerous.  Tunnel descriptors are not completely safe, but this is
+probably better than nothing.)
+.SP
+.BI "PRIVSEP child-exited " rc
+The helper process exited normally with status
+.IR rc .
+Status 0 means that it thought the server didn't want it any more; 1
+means that it was invoked incorrectly; 127 means that some system call
+failed.
+.SP
+.BI "PRIVSEP child-killed " sig
+The helper process was killed by signal number
+.IR sig .
+.SP
+.BI "PRIVSEP child-died " status
+The helper process died in some unexpected way;
+.I status is the raw status code returned by
+.BR waitpid (2),
+because the server didn't understand how to decode it.
+.SP
+.BI "PRIVSEP helper-died"
+A tunnel driver requires a tunnel descriptor from the helper, but the
+helper isn't running so this won't work.
+.SP
+.BI "PRIVSEP helper-read-error " ecode " " message
+The server failed to read a response from the helper process.
+.SP
+.BI "PRIVSEP helper-short-read"
+The helper process didn't send back enough data, and has likely crashed.
+.SP
+.BI "PRIVSEP helper-write-error " ecode " " message
+The server failed to send a message to the helper process.
+.SP
+.BI "PRIVSEP no-fd-from-helper"
+The helper process sent back a positive response, but didn't include the
+requested tunnel descriptor.
+.SP
+.BI "PRIVSEP unknown-response-code"
+The helper process sent back an incomprehensible reply.  It's probably
+very confused and may crash.
 .SS "SERVER warnings"
 These indicate problems concerning the server process as a whole.
 .SP
@@ -1569,6 +1627,10 @@ option), and encountered end-of-file on standard input.
 .BI "SERVER select-error " ecode " " message
 An error occurred in the server's main event loop.  This is bad: if it
 happens too many times, the server will abort.
+.SP
+.BI "SERVER waitpid-error " ecode " " message
+The server was informed that one of its child processes had exited, but
+couldn't retrieve the child's status.
 .SS "SYMM warnings"
 These are concerned with the symmetric encryption and decryption
 process.