Remove listen mode

Changing behaviour so that instead of rebinding after a successful
migration and continuing as an ordinary server, we simply quit with a
0 exit code and let our caller restart us as a server if they want to.
This means that everything in listen.c, listen.h, and anything making
reference to a rebind address is unneeded.

tests/acceptance/fakes/source/close_after_entrust.rb

@@ -3,8 +3,8 @@
 # Connect, send a migration, entrust then *immediately* disconnect.
 # This simulates a client which fails while the client is blocked.
 #
-# We attempt to reconnect immediately afterwards to prove that we can
-# retry the mirroring.
+# In this situation we expect the destination to quit with an error
+# status.
 
 require 'flexnbd/fake_source'
 include FlexNBD
@@ -28,7 +28,11 @@ system "kill -CONT #{srv_pid}"
 
 sleep(0.25)
 
-client2 = FakeSource.new( addr, port, "Timed out reconnecting" )
-client2.close
+begin
+  client2 = FakeSource.new( addr, port, "Expected timeout" )
+  fail "Unexpected reconnection"
+rescue Timeout::Error
+  # expected
+end
 
 exit(0)

tests/acceptance/fakes/source/close_after_entrust_reply.rb

@@ -1,10 +1,9 @@
 #!/usr/bin/env ruby
 
-# Connect, send a migration, entrust then *immediately* disconnect.
+# Connect, send a migration, entrust, read the reply, then disconnect.
 # This simulates a client which fails while the client is blocked.
 #
-# We attempt to reconnect immediately afterwards to prove that we can
-# retry the mirroring.
+# We expect the destination to quit with an error status.
 
 require 'flexnbd/fake_source'
 include FlexNBD
@@ -22,11 +21,12 @@ client.close
 
 sleep(0.25)
 
-client2 = FakeSource.new( addr, port, "Timed out reconnecting to mirror" )
-client2.send_mirror
 
-sleep(1)
-client3 = FakeSource.new( rebind_addr, rebind_port, "Timed out reconnecting to read" )
-client3.close
+begin
+  client2 = FakeSource.new( addr, port, "Expected timeout" )
+  fail "Unexpected reconnection"
+rescue Timeout::Error
+  # expected
+end
 
 exit(0)

tests/acceptance/fakes/source/close_after_write.rb

@@ -12,10 +12,11 @@ addr, port, srv_pid = *ARGV
 
 client = FakeSource.new( addr, port, "Timed out connecting" )
 client.read_hello
-Process.kill( "STOP", srv_pid.to_i )
+
+system "kill -STOP #{srv_pid}"
 client.write_write_request( 0, 8 )
 client.close
-Process.kill( "CONT", srv_pid.to_i )
+system "kill -CONT #{srv_pid}"
 
 # This sleep ensures that we don't return control to the test runner
 # too soon, giving the flexnbd process time to fall over if it's going

tests/acceptance/fakes/source/close_after_write_data.rb

@@ -13,13 +13,13 @@ addr, port, srv_pid = *ARGV
 client = FakeSource.new( addr, port, "Timed out connecting" )
 client.read_hello
 
-Process.kill( "STOP", srv_pid.to_i )
+system "kill -STOP #{srv_pid}"
 
 client.write_write_request( 0, 8 )
 client.write_data( "12345678" )
 client.close
 
-Process.kill( "CONT", srv_pid.to_i )
+system "kill -CONT #{srv_pid}"
 
 # This sleep ensures that we don't return control to the test runner
 # too soon, giving the flexnbd process time to fall over if it's going

tests/acceptance/fakes/source/successful_transfer.rb

@@ -1,17 +1,14 @@
 #!/usr/bin/env ruby
 
-# Successfully send a migration, but squat on the IP and port which
-# the destination wants to rebind to.  The destination should retry
-# every second, so we give it up then attempt to connect to the new
-# server.
+# Successfully send a migration.  This test just makes sure that the
+# happy path is covered.  We expect the destination to quit with a
+# success status.
 
 require 'flexnbd/fake_source'
 include FlexNBD
 
 addr, port, srv_pid, newaddr, newport = *ARGV
 
-squatter = TCPServer.open( newaddr, newport.to_i )
-
 client  = FakeSource.new( addr, port, "Timed out connecting" )
 client.send_mirror()