Fix the fact that we can't bump RAM quotas above 2048M

[invirt/packages/invirt-xen-config.git] / vif-invirtroute

diff --git a/vif-invirtroute b/vif-invirtroute
index 2b3a88d..b3d9207 100755 (executable)
--- a/vif-invirtroute
+++ b/vif-invirtroute
@@ -1,9 +1,5 @@
 #!/bin/bash
 
-
-
-
-
 #============================================================================
 # /etc/xen/vif-route
 #
@@ -17,18 +13,16 @@
 # vif-route (add|remove|online|offline)
 #
 # Environment vars:
-# vif         vif interface name (required).
+# vif    vif interface name (required).
 # XENBUS_PATH path to this device's details in the XenStore (required).
 # Read from the store:
-# ip      list of IP networks for the vif, space-separated (default given in
-#         this script).
-# V6PREFIX  prefix of v6 address to use
-# Note that the v6 support is kind of broken because there's not really a way to populate the v6 prefix
+# ip     list of IP networks for the vif, space-separated (default given in
+#        this script).
+#
 # This script will set up proxy arp  for any ip addresses that are being routed
 # type read to determine if the device is ioemu
 
 #============================================================================
-
 dir=$(dirname "$0")
 . "$dir/vif-common.sh"
 
@@ -36,55 +30,87 @@ main_ip=$(dom0_ip)
 dev=${dev:-${vif}}
 
 case "$command" in
-    online)
-        ifconfig ${dev} ${main_ip} netmask 255.255.255.255 up
-        echo 1 >/proc/sys/net/ipv4/conf/${dev}/proxy_arp
+    online|add)
+       ifconfig ${dev} ${main_ip} netmask 255.255.255.255 up
+       echo 1 >/proc/sys/net/ipv4/conf/${dev}/proxy_arp
        echo 1 >/proc/sys/net/ipv4/conf/${dev}/rp_filter 
-        if [ x${qemu_online} != xyes ]; then
-          ethtool -K ${dev} tx off
-        fi
-        ipcmd='add'
-        cmdprefix=''
-        ;;
-    offline)
-        do_without_error ifdown ${vif}
-       if [ -f /var/run/radvd/radvd.pid.${vif} ] ; then
-           do_without_error kill `cat /var/run/radvd/radvd.pid.${vif}`
-           fi
-        ipcmd='del'
-        cmdprefix='do_without_error'
-        ;;
+       xenstore-write "$XENBUS_PATH/feature-gso-tcpv4" 0
+       if [ x${qemu_online} != xyes ]; then
+         ethtool -K ${dev} tx off
+       fi
+       ipcmd='add'
+       ipt_action='-A'
+       cmdprefix=''
+       ;;
+    offline|remove)
+       do_without_error ifdown ${vif}
+       ipcmd='del'
+       ipt_action='-D'
+       cmdprefix='do_without_error'
+       ;;
 esac
 
-v6prefix=${v6prefix:-}
-v6prefix=$(xenstore_read_default "$XENBUS_PATH/v6prefix" "$v6prefix")
 vif_type=$(xenstore_read_default "$XENBUS_PATH/type" "viffront")
 if [  ${vif_type} != "ioemu"  -o  x${qemu_online} = xyes ] ; then
     if [ "${ip}" ] ; then
     # If we've been given a list of IP addresses, then add routes from dom0 to
     # the guest using those addresses.
        for addr in ${ip} ; do
-           ${cmdprefix} ip route ${ipcmd} ${addr} dev ${dev} src ${main_ip}
-           arpspoof -i $(invirt-getconf xen.network.iface) -t 18.181.0.1 ${addr}&
-           sleep 5
-           kill %arpspoof
-       done 
-    fi
+           # When PVHVM is enabled, Xen plugs two interfaces into
+           # HVMs - an emulated tap device and a paravirt vif device.
+           # vif-invirtroute (and vif-route, for that matter!) will
+           # fail when the second one is brought up, because the
+           # second invocation of 'ip route add' is identical to the
+           # first (same source and destination IPs) and the kernel
+           # rejects the new route.
+           #
+           # We work around this by adding the routes with different metrics.
+           # This should work because:
+           #
+           # 1) In the case of a pv-aware guest, the kernel will
+           # unplug the tap interface, which will bring down the tap
+           # interface's route, leaving only the one via the vif (and
+           # so the metric shouldn't matter, because it's the only
+           # route)
+           #
+           # 2) In the case of a non-pv-aware guest, the tap route
+           # (with metric 1) should take precedence over the vif
+           # route and carry all the traffic.
 
-    if [ x${v6prefix} != x ] ; then
-       sed -e "s/@interface@/${dev}/" -e "s+@prefix@+${v6prefix}+" /etc/xen/radvd.conf.template >/var/run/radvd.conf.${vif}
-       ${cmdprefix} ip -6 addr  ${ipcmd}  fe80::/64 scope link  dev ${dev} 
-       if [ $1 = online ] ; then
-           radvd  -u radvd -C /var/run/radvd.conf.${vif} -p /var/run/radvd/radvd.pid.${vif}
+           src=""
+           if [ $ipcmd == "add" ]; then
+               src="src ${main_ip}"
+               case $dev in
+                   vif*)
+                       metric="metric 2"
+                       ;;
+                   tap*)
+                       metric="metric 1"
+                       ;;
+               esac
+           fi
+           ${cmdprefix} ip route ${ipcmd} ${addr} dev ${dev} ${src} $metric
+           case "$command" in
+               online|add)
+                   timeout -s KILL 5 arpspoof -i $(invirt-getconf xen.iface) -t ${gateway} ${addr} || :
+                   ;;
+           esac
+       done 
+       if [ -n "$other_ip" ]; then
+           ${cmdprefix} ip route ${ipcmd} ${other_ip} dev ${dev} ${src} $metric
+           iptables -t nat ${ipt_action} PREROUTING -d ${other_ip} -j DNAT --to-destination ${addr}
+           case "$command" in
+               online|add)
+                   timeout -s KILL 5 arpspoof -i $(invirt-getconf xen.iface) -t ${other_gateway} ${other_ip} || :
+                   ;;
+           esac
        fi
-       ${cmdprefix} ip -6 route ${ipcmd} ${v6prefix} dev ${dev} 
     fi
 fi
 
-handle_iptable
-
 log debug "Successful vif-route $command for $vif."
-if [ "$command" == "online" ]
-then
-  success
-fi
+case "$command" in
+    online|add)
+       success
+       ;;
+esac