Move domain interface management methods from qemu to hypervisor. This refactoring allows the domain management methods to be shared by ch driver. This commit does not introduce any functional changes. Signed-off-by: Praveen K Paladugu <prapal(a)linux.microsoft.com&gt; --- po/POTFILES | 1 + src/hypervisor/domain_interface.c | 457 ++++++++++++++++++++++++++++++ src/hypervisor/domain_interface.h | 45 +++ src/hypervisor/meson.build | 1 + src/libvirt_private.syms | 9 + src/qemu/qemu_command.c | 6 +- src/qemu/qemu_hotplug.c | 5 +- src/qemu/qemu_interface.c | 339 +--------------------- src/qemu/qemu_interface.h | 11 - src/qemu/qemu_process.c | 73 +---- 10 files changed, 532 insertions(+), 415 deletions(-) create mode 100644 src/hypervisor/domain_interface.c create mode 100644 src/hypervisor/domain_interface.h diff --git a/po/POTFILES b/po/POTFILES index 3a51aea5cb..023c041f61 100644 --- a/po/POTFILES +++ b/po/POTFILES @@ -92,6 +92,7 @@ src/hyperv/hyperv_util.c src/hyperv/hyperv_wmi.c src/hypervisor/domain_cgroup.c src/hypervisor/domain_driver.c +src/hypervisor/domain_interface.c src/hypervisor/virhostdev.c src/interface/interface_backend_netcf.c src/interface/interface_backend_udev.c diff --git a/src/hypervisor/domain_interface.c b/src/hypervisor/domain_interface.c new file mode 100644 index 0000000000..a7d5d46a8c --- /dev/null +++ b/src/hypervisor/domain_interface.c @@ -0,0 +1,457 @@ +/* + * Copyright (C) 2015-2016 Red Hat, Inc. + * Copyright IBM Corp. 2014 + * + * domain_interface.c: methods to manage guest/domain interfaces + * + * This library is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * This library is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with this library. If not, see + * <http://www.gnu.org/licenses/>;. + */ + +#include <config.h> + +#include "datatypes.h" +#include "domain_audit.h" +#include "domain_conf.h" +#include "domain_driver.h" +#include "domain_interface.h" +#include "domain_nwfilter.h" +#include "network_conf.h" +#include "viralloc.h" +#include "virconftypes.h" +#include "virebtables.h" +#include "virfile.h" +#include "virlog.h" +#include "virmacaddr.h" +#include "virnetdevbridge.h" +#include "virnetdevmidonet.h" +#include "virnetdevopenvswitch.h" +#include "virnetdevtap.h" + + +#define VIR_FROM_THIS VIR_FROM_DOMAIN + +VIR_LOG_INIT("domain.interface"); + +bool +virDomainInterfaceIsVnetCompatModel(const virDomainNetDef *net) +{ + return (virDomainNetIsVirtioModel(net) || + net->model == VIR_DOMAIN_NET_MODEL_E1000E || + net->model == VIR_DOMAIN_NET_MODEL_IGB || + net->model == VIR_DOMAIN_NET_MODEL_VMXNET3); +} + +/* virDomainInterfaceEthernetConnect: + * @def: the definition of the VM + * @net: a net definition in the VM + * @ebtables: ebtales context + * @macFilter: whether driver support mac Filtering + * @privileged: whether running as privileged user + * @tapfd: returned array of file descriptors for the domain interface + * @tapfdsize: returned number of file descriptors in @tapfd + * + * Called *only* called if actualType is VIR_DOMAIN_NET_TYPE_ETHERNET + * (i.e. if the connection is made with a tap device) + */ +int +virDomainInterfaceEthernetConnect(virDomainDef *def, + virDomainNetDef *net, + ebtablesContext *ebtables, + bool macFilter, + bool privileged, + int *tapfd, + size_t tapfdSize) +{ + virMacAddr tapmac; + int ret = -1; + unsigned int tap_create_flags = VIR_NETDEV_TAP_CREATE_IFUP; + bool template_ifname = false; + const char *tunpath = "/dev/net/tun"; + const char *auditdev = tunpath; + + if (net->backend.tap) { + tunpath = net->backend.tap; + if (!privileged) { + virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s", + _("cannot use custom tap device in session mode")); + goto cleanup; + } + } + + if (virDomainInterfaceIsVnetCompatModel(net)) + tap_create_flags |= VIR_NETDEV_TAP_CREATE_VNET_HDR; + + if (net->managed_tap == VIR_TRISTATE_BOOL_NO) { + if (!net->ifname) { + virReportError(VIR_ERR_INTERNAL_ERROR, "%s", + _("target dev must be supplied when managed='no'")); + goto cleanup; + } + if (virNetDevExists(net->ifname) != 1) { + virReportError(VIR_ERR_INTERNAL_ERROR, "%s", + _("target managed='no' but specified dev doesn't exist")); + goto cleanup; + } + + tap_create_flags |= VIR_NETDEV_TAP_CREATE_ALLOW_EXISTING; + + if (virNetDevMacVLanIsMacvtap(net->ifname)) { + auditdev = net->ifname; + if (virNetDevMacVLanTapOpen(net->ifname, tapfd, tapfdSize) < 0) + goto cleanup; + if (virNetDevMacVLanTapSetup(tapfd, tapfdSize, + virDomainInterfaceIsVnetCompatModel(net)) < 0) { + goto cleanup; + } + } else { + if (virNetDevTapCreate(&net->ifname, tunpath, tapfd, tapfdSize, + tap_create_flags) < 0) + goto cleanup; + } + } else { + + if (!net->ifname) + template_ifname = true; + + if (virNetDevTapCreate(&net->ifname, tunpath, tapfd, tapfdSize, + tap_create_flags) < 0) { + goto cleanup; + } + + /* The tap device's MAC address cannot match the MAC address + * used by the guest. This results in "received packet on + * vnetX with own address as source address" error logs from + * the kernel. + */ + virMacAddrSet(&tapmac, &net->mac); + if (tapmac.addr[0] == 0xFE) + tapmac.addr[0] = 0xFA; + else + tapmac.addr[0] = 0xFE; + + if (virNetDevSetMAC(net->ifname, &tapmac) < 0) + goto cleanup; + + if (virNetDevSetOnline(net->ifname, true) < 0) + goto cleanup; + } + + if (net->script && + virNetDevRunEthernetScript(net->ifname, net->script) < 0) + goto cleanup; + + if (macFilter && + ebtablesAddForwardAllowIn(ebtables, + net->ifname, + &net->mac) < 0) + goto cleanup; + + if (net->filter && + virDomainConfNWFilterInstantiate(def->name, def->uuid, net, false) < 0) { + goto cleanup; + } + + virDomainAuditNetDevice(def, net, auditdev, true); + + ret = 0; + + cleanup: + if (ret < 0) { + size_t i; + + virDomainAuditNetDevice(def, net, auditdev, false); + for (i = 0; i < tapfdSize && tapfd[i] >= 0; i++) + VIR_FORCE_CLOSE(tapfd[i]); + if (template_ifname) + VIR_FREE(net->ifname); + } + + return ret; +} + +/** + * virDomainInterfaceStartDevice: + * @net: net device to start + * + * Based upon the type of device provided, perform the appropriate + * work to completely activate the device and make it reachable from + * the rest of the network. + */ +int +virDomainInterfaceStartDevice(virDomainNetDef *net) +{ + virDomainNetType actualType = virDomainNetGetActualType(net); + + switch (actualType) { + case VIR_DOMAIN_NET_TYPE_BRIDGE: + case VIR_DOMAIN_NET_TYPE_NETWORK: + if (virDomainNetGetActualBridgeMACTableManager(net) + == VIR_NETWORK_BRIDGE_MAC_TABLE_MANAGER_LIBVIRT) { + /* libvirt is managing the FDB of the bridge this device + * is attaching to, so we have turned off learning and + * unicast_flood on the device to prevent the kernel from + * adding any FDB entries for it. This means we need to + * add an fdb entry ourselves, using the MAC address from + * the interface config. + */ + if (virNetDevBridgeFDBAdd(&net->mac, net->ifname, + VIR_NETDEVBRIDGE_FDB_FLAG_MASTER | + VIR_NETDEVBRIDGE_FDB_FLAG_TEMP) < 0) + return -1; + } + break; + + case VIR_DOMAIN_NET_TYPE_DIRECT: { + const char *physdev = virDomainNetGetActualDirectDev(net); + bool isOnline = true; + + /* set the physdev online if necessary. It may already be up, + * in which case we shouldn't re-up it just in case that causes + * some sort of "blip" in the physdev's status. + */ + if (physdev && virNetDevGetOnline(physdev, &isOnline) < 0) + return -1; + if (!isOnline && virNetDevSetOnline(physdev, true) < 0) + return -1; + + /* macvtap devices share their MAC address with the guest + * domain, and if they are set online prior to the domain CPUs + * being started, the host may send out traffic from this + * device that could confuse other entities on the network (in + * particular, if this new domain is the destination of a + * migration, and the source domain is still running, another + * host may mistakenly direct traffic for the guest to the + * destination domain rather than source domain). To prevent + * this, we create the macvtap device with IFF_UP false + * (i.e. "offline") then wait to bring it online until just as + * we are starting the domain CPUs. + */ + if (virNetDevSetOnline(net->ifname, true) < 0) + return -1; + break; + } + + case VIR_DOMAIN_NET_TYPE_ETHERNET: + if (virNetDevIPInfoAddToDev(net->ifname, &net->hostIP) < 0) + return -1; + + break; + + case VIR_DOMAIN_NET_TYPE_USER: + case VIR_DOMAIN_NET_TYPE_VHOSTUSER: + case VIR_DOMAIN_NET_TYPE_SERVER: + case VIR_DOMAIN_NET_TYPE_CLIENT: + case VIR_DOMAIN_NET_TYPE_MCAST: + case VIR_DOMAIN_NET_TYPE_UDP: + case VIR_DOMAIN_NET_TYPE_INTERNAL: + case VIR_DOMAIN_NET_TYPE_HOSTDEV: + case VIR_DOMAIN_NET_TYPE_VDPA: + case VIR_DOMAIN_NET_TYPE_NULL: + case VIR_DOMAIN_NET_TYPE_VDS: + case VIR_DOMAIN_NET_TYPE_LAST: + /* these types all require no action */ + break; + } + + return 0; +} + +/** + * virDomainInterfaceStartDevices: + * @def: domain definition + * + * Set all ifaces associated with this domain to the online state. + */ +int +virDomainInterfaceStartDevices(virDomainDef *def) +{ + size_t i; + + for (i = 0; i < def->nnets; i++) { + if (virDomainInterfaceStartDevice(def->nets[i]) < 0) + return -1; + } + return 0; +} + +/** + * virDomainInterfaceStopDevice: + * @net: net device to stop + * + * Based upon the type of device provided, perform the appropriate + * work to deactivate the device so that packets aren't forwarded to + * it from the rest of the network. + */ +int +virDomainInterfaceStopDevice(virDomainNetDef *net) +{ + virDomainNetType actualType = virDomainNetGetActualType(net); + + switch (actualType) { + case VIR_DOMAIN_NET_TYPE_BRIDGE: + case VIR_DOMAIN_NET_TYPE_NETWORK: + if (virDomainNetGetActualBridgeMACTableManager(net) + == VIR_NETWORK_BRIDGE_MAC_TABLE_MANAGER_LIBVIRT) { + /* remove the FDB entries that were added during + * virDomainInterfaceStartDevices() + */ + if (virNetDevBridgeFDBDel(&net->mac, net->ifname, + VIR_NETDEVBRIDGE_FDB_FLAG_MASTER | + VIR_NETDEVBRIDGE_FDB_FLAG_TEMP) < 0) + return -1; + } + break; + + case VIR_DOMAIN_NET_TYPE_DIRECT: { + const char *physdev = virDomainNetGetActualDirectDev(net); + + /* macvtap interfaces need to be marked !IFF_UP (ie "down") to + * prevent any host-generated traffic sent from this interface + * from putting bad info into the arp caches of other machines + * on this network. + */ + if (virNetDevSetOnline(net->ifname, false) < 0) + return -1; + + /* also mark the physdev down for passthrough macvtap, as the + * physdev has the same MAC address as the macvtap device. + */ + if (virDomainNetGetActualDirectMode(net) == + VIR_NETDEV_MACVLAN_MODE_PASSTHRU && + physdev && virNetDevSetOnline(physdev, false) < 0) + return -1; + break; + } + + case VIR_DOMAIN_NET_TYPE_ETHERNET: + case VIR_DOMAIN_NET_TYPE_USER: + case VIR_DOMAIN_NET_TYPE_VHOSTUSER: + case VIR_DOMAIN_NET_TYPE_SERVER: + case VIR_DOMAIN_NET_TYPE_CLIENT: + case VIR_DOMAIN_NET_TYPE_MCAST: + case VIR_DOMAIN_NET_TYPE_UDP: + case VIR_DOMAIN_NET_TYPE_INTERNAL: + case VIR_DOMAIN_NET_TYPE_HOSTDEV: + case VIR_DOMAIN_NET_TYPE_VDPA: + case VIR_DOMAIN_NET_TYPE_NULL: + case VIR_DOMAIN_NET_TYPE_VDS: + case VIR_DOMAIN_NET_TYPE_LAST: + /* these types all require no action */ + break; + } + + return 0; +} + +/** + * virDomainInterfaceStopDevices: + * @def: domain definition + * + * Make all interfaces associated with this domain inaccessible from + * the rest of the network. + */ +int +virDomainInterfaceStopDevices(virDomainDef *def) +{ + size_t i; + + for (i = 0; i < def->nnets; i++) { + if (virDomainInterfaceStopDevice(def->nets[i]) < 0) + return -1; + } + return 0; +} + +/** + * virDomainInterfaceDeleteDevice: + * @def: domain definition + * @net: a net definition in the VM + * @priv_net_created: whether private network created + * @stateDir: driver's stateDir + * + * Disconnect and delete domain interfaces. + */ + +void +virDomainInterfaceDeleteDevice(virDomainDef *def, + virDomainNetDef *net, + bool priv_net_created, + char *stateDir) +{ + const virNetDevVPortProfile *vport = NULL; + g_autoptr(virConnect) conn = NULL; + + vport = virDomainNetGetActualVirtPortProfile(net); + switch (virDomainNetGetActualType(net)) { + case VIR_DOMAIN_NET_TYPE_DIRECT: + if (priv_net_created) { + virNetDevMacVLanDeleteWithVPortProfile(net->ifname, &net->mac, + virDomainNetGetActualDirectDev(net), + virDomainNetGetActualDirectMode(net), + virDomainNetGetActualVirtPortProfile(net), + stateDir); + } + break; + case VIR_DOMAIN_NET_TYPE_ETHERNET: + if (net->managed_tap != VIR_TRISTATE_BOOL_NO && net->ifname) { + ignore_value(virNetDevTapDelete(net->ifname, net->backend.tap)); + VIR_FREE(net->ifname); + } + break; + case VIR_DOMAIN_NET_TYPE_BRIDGE: + case VIR_DOMAIN_NET_TYPE_NETWORK: +#ifdef VIR_NETDEV_TAP_REQUIRE_MANUAL_CLEANUP + if (!(vport && vport->virtPortType == VIR_NETDEV_VPORT_PROFILE_OPENVSWITCH)) + ignore_value(virNetDevTapDelete(net->ifname, net->backend.tap)); +#endif + break; + case VIR_DOMAIN_NET_TYPE_USER: + case VIR_DOMAIN_NET_TYPE_VHOSTUSER: + case VIR_DOMAIN_NET_TYPE_SERVER: + case VIR_DOMAIN_NET_TYPE_CLIENT: + case VIR_DOMAIN_NET_TYPE_MCAST: + case VIR_DOMAIN_NET_TYPE_INTERNAL: + case VIR_DOMAIN_NET_TYPE_HOSTDEV: + case VIR_DOMAIN_NET_TYPE_UDP: + case VIR_DOMAIN_NET_TYPE_VDPA: + case VIR_DOMAIN_NET_TYPE_NULL: + case VIR_DOMAIN_NET_TYPE_VDS: + case VIR_DOMAIN_NET_TYPE_LAST: + /* No special cleanup procedure for these types. */ + break; + } + /* release the physical device (or any other resources used by + * this interface in the network driver + */ + if (vport) { + if (vport->virtPortType == VIR_NETDEV_VPORT_PROFILE_MIDONET) { + ignore_value(virNetDevMidonetUnbindPort(vport)); + } else if (vport->virtPortType == VIR_NETDEV_VPORT_PROFILE_OPENVSWITCH) { + ignore_value(virNetDevOpenvswitchRemovePort( + virDomainNetGetActualBridgeName(net), + net->ifname)); + } + } + + /* kick the device out of the hostdev list too */ + virDomainNetRemoveHostdev(def, net); + if (net->type == VIR_DOMAIN_NET_TYPE_NETWORK) { + if (conn || (conn = virGetConnectNetwork())) + virDomainNetReleaseActualDevice(conn, net); + else + VIR_WARN("Unable to release network device '%s'", NULLSTR(net->ifname)); + } + +} diff --git a/src/hypervisor/domain_interface.h b/src/hypervisor/domain_interface.h new file mode 100644 index 0000000000..3d15e000cc --- /dev/null +++ b/src/hypervisor/domain_interface.h @@ -0,0 +1,45 @@ +/* + * Copyright (C) 2015-2016 Red Hat, Inc. + * Copyright IBM Corp. 2014 + * + * domain_interface.h: methods to manage guest/domain interfaces + * + * This library is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * This library is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with this library. If not, see + * <http://www.gnu.org/licenses/>;. + */ + +#pragma once + +#include "virebtables.h" + +int +virDomainInterfaceEthernetConnect(virDomainDef *def, + virDomainNetDef *net, + ebtablesContext *ebtables, + bool macFilter, + bool privileged, + int *tapfd, + size_t tapfdSize); + +bool +virDomainInterfaceIsVnetCompatModel(const virDomainNetDef *net); + +int virDomainInterfaceStartDevice(virDomainNetDef *net); +int virDomainInterfaceStartDevices(virDomainDef *def); +int virDomainInterfaceStopDevice(virDomainNetDef *net); +int virDomainInterfaceStopDevices(virDomainDef *def); +void virDomainInterfaceDeleteDevice(virDomainDef *def, + virDomainNetDef *net, + bool priv_net_created, + char *stateDir); diff --git a/src/hypervisor/meson.build b/src/hypervisor/meson.build index f35565b16b..819a9a82a2 100644 --- a/src/hypervisor/meson.build +++ b/src/hypervisor/meson.build @@ -1,6 +1,7 @@ hypervisor_sources = [ 'domain_cgroup.c', 'domain_driver.c', + 'domain_interface.c', 'virclosecallbacks.c', 'virhostdev.c', ] diff --git a/src/libvirt_private.syms b/src/libvirt_private.syms index 553b01b8c0..680a90034a 100644 --- a/src/libvirt_private.syms +++ b/src/libvirt_private.syms @@ -1630,6 +1630,15 @@ virDomainDriverNodeDeviceReset; virDomainDriverParseBlkioDeviceStr; virDomainDriverSetupPersistentDefBlkioParams; +# hypervisor/domain_interface.h +virDomainInterfaceDeleteDevice; +virDomainInterfaceEthernetConnect; +virDomainInterfaceIsVnetCompatModel; +virDomainInterfaceStartDevice; +virDomainInterfaceStartDevices; +virDomainInterfaceStopDevice; +virDomainInterfaceStopDevices; + # hypervisor/virclosecallbacks.h virCloseCallbacksDomainAdd; diff --git a/src/qemu/qemu_command.c b/src/qemu/qemu_command.c index d54149ed2d..c3dd777089 100644 --- a/src/qemu/qemu_command.c +++ b/src/qemu/qemu_command.c @@ -43,6 +43,7 @@ #include "domain_nwfilter.h" #include "domain_addr.h" #include "domain_conf.h" +#include "domain_interface.h" #include "netdev_bandwidth_conf.h" #include "virnetdevopenvswitch.h" #include "device_conf.h" @@ -8575,7 +8576,10 @@ qemuBuildInterfaceConnect(virDomainObj *vm, break; case VIR_DOMAIN_NET_TYPE_ETHERNET: - if (qemuInterfaceEthernetConnect(vm->def, priv->driver, net, + if (virDomainInterfaceEthernetConnect(vm->def, net, + priv->driver->ebtables, + priv->driver->config->macFilter, + priv->driver->privileged, tapfd, tapfdSize) < 0) return -1; vhostfd = true; diff --git a/src/qemu/qemu_hotplug.c b/src/qemu/qemu_hotplug.c index 45c055ea7f..0ce7b175c4 100644 --- a/src/qemu/qemu_hotplug.c +++ b/src/qemu/qemu_hotplug.c @@ -39,6 +39,7 @@ #include "qemu_virtiofs.h" #include "domain_audit.h" #include "domain_cgroup.h" +#include "domain_interface.h" #include "netdev_bandwidth_conf.h" #include "domain_nwfilter.h" #include "virlog.h" @@ -1278,7 +1279,7 @@ qemuDomainAttachNetDevice(virQEMUDriver *driver, } /* Set device online immediately */ - if (qemuInterfaceStartDevice(net) < 0) + if (virDomainInterfaceStartDevice(net) < 0) goto cleanup; qemuDomainInterfaceSetDefaultQDisc(driver, net); @@ -4778,7 +4779,7 @@ qemuDomainRemoveNetDevice(virQEMUDriver *driver, * affect the parent device (e.g. macvtap passthrough mode sets * the parent device offline) */ - ignore_value(qemuInterfaceStopDevice(net)); + ignore_value(virDomainInterfaceStopDevice(net)); qemuDomainObjEnterMonitor(vm); if (qemuMonitorRemoveNetdev(priv->mon, hostnet_name) < 0) { diff --git a/src/qemu/qemu_interface.c b/src/qemu/qemu_interface.c index 8856bb95a8..c2007c7043 100644 --- a/src/qemu/qemu_interface.c +++ b/src/qemu/qemu_interface.c @@ -24,6 +24,7 @@ #include "network_conf.h" #include "domain_audit.h" #include "domain_nwfilter.h" +#include "domain_interface.h" #include "qemu_interface.h" #include "viralloc.h" #include "virlog.h" @@ -41,211 +42,6 @@ VIR_LOG_INIT("qemu.qemu_interface"); -/** - * qemuInterfaceStartDevice: - * @net: net device to start - * - * Based upon the type of device provided, perform the appropriate - * work to completely activate the device and make it reachable from - * the rest of the network. - */ -int -qemuInterfaceStartDevice(virDomainNetDef *net) -{ - virDomainNetType actualType = virDomainNetGetActualType(net); - - switch (actualType) { - case VIR_DOMAIN_NET_TYPE_BRIDGE: - case VIR_DOMAIN_NET_TYPE_NETWORK: - if (virDomainNetGetActualBridgeMACTableManager(net) - == VIR_NETWORK_BRIDGE_MAC_TABLE_MANAGER_LIBVIRT) { - /* libvirt is managing the FDB of the bridge this device - * is attaching to, so we have turned off learning and - * unicast_flood on the device to prevent the kernel from - * adding any FDB entries for it. This means we need to - * add an fdb entry ourselves, using the MAC address from - * the interface config. - */ - if (virNetDevBridgeFDBAdd(&net->mac, net->ifname, - VIR_NETDEVBRIDGE_FDB_FLAG_MASTER | - VIR_NETDEVBRIDGE_FDB_FLAG_TEMP) < 0) - return -1; - } - break; - - case VIR_DOMAIN_NET_TYPE_DIRECT: { - const char *physdev = virDomainNetGetActualDirectDev(net); - bool isOnline = true; - - /* set the physdev online if necessary. It may already be up, - * in which case we shouldn't re-up it just in case that causes - * some sort of "blip" in the physdev's status. - */ - if (physdev && virNetDevGetOnline(physdev, &isOnline) < 0) - return -1; - if (!isOnline && virNetDevSetOnline(physdev, true) < 0) - return -1; - - /* macvtap devices share their MAC address with the guest - * domain, and if they are set online prior to the domain CPUs - * being started, the host may send out traffic from this - * device that could confuse other entities on the network (in - * particular, if this new domain is the destination of a - * migration, and the source domain is still running, another - * host may mistakenly direct traffic for the guest to the - * destination domain rather than source domain). To prevent - * this, we create the macvtap device with IFF_UP false - * (i.e. "offline") then wait to bring it online until just as - * we are starting the domain CPUs. - */ - if (virNetDevSetOnline(net->ifname, true) < 0) - return -1; - break; - } - - case VIR_DOMAIN_NET_TYPE_ETHERNET: - if (virNetDevIPInfoAddToDev(net->ifname, &net->hostIP) < 0) - return -1; - - break; - - case VIR_DOMAIN_NET_TYPE_USER: - case VIR_DOMAIN_NET_TYPE_VHOSTUSER: - case VIR_DOMAIN_NET_TYPE_SERVER: - case VIR_DOMAIN_NET_TYPE_CLIENT: - case VIR_DOMAIN_NET_TYPE_MCAST: - case VIR_DOMAIN_NET_TYPE_UDP: - case VIR_DOMAIN_NET_TYPE_INTERNAL: - case VIR_DOMAIN_NET_TYPE_HOSTDEV: - case VIR_DOMAIN_NET_TYPE_VDPA: - case VIR_DOMAIN_NET_TYPE_NULL: - case VIR_DOMAIN_NET_TYPE_VDS: - case VIR_DOMAIN_NET_TYPE_LAST: - /* these types all require no action */ - break; - } - - return 0; -} - -/** - * qemuInterfaceStartDevices: - * @def: domain definition - * - * Set all ifaces associated with this domain to the online state. - */ -int -qemuInterfaceStartDevices(virDomainDef *def) -{ - size_t i; - - for (i = 0; i < def->nnets; i++) { - if (qemuInterfaceStartDevice(def->nets[i]) < 0) - return -1; - } - return 0; -} - - -/** - * qemuInterfaceStopDevice: - * @net: net device to stop - * - * Based upon the type of device provided, perform the appropriate - * work to deactivate the device so that packets aren't forwarded to - * it from the rest of the network. - */ -int -qemuInterfaceStopDevice(virDomainNetDef *net) -{ - virDomainNetType actualType = virDomainNetGetActualType(net); - - switch (actualType) { - case VIR_DOMAIN_NET_TYPE_BRIDGE: - case VIR_DOMAIN_NET_TYPE_NETWORK: - if (virDomainNetGetActualBridgeMACTableManager(net) - == VIR_NETWORK_BRIDGE_MAC_TABLE_MANAGER_LIBVIRT) { - /* remove the FDB entries that were added during - * qemuInterfaceStartDevices() - */ - if (virNetDevBridgeFDBDel(&net->mac, net->ifname, - VIR_NETDEVBRIDGE_FDB_FLAG_MASTER | - VIR_NETDEVBRIDGE_FDB_FLAG_TEMP) < 0) - return -1; - } - break; - - case VIR_DOMAIN_NET_TYPE_DIRECT: { - const char *physdev = virDomainNetGetActualDirectDev(net); - - /* macvtap interfaces need to be marked !IFF_UP (ie "down") to - * prevent any host-generated traffic sent from this interface - * from putting bad info into the arp caches of other machines - * on this network. - */ - if (virNetDevSetOnline(net->ifname, false) < 0) - return -1; - - /* also mark the physdev down for passthrough macvtap, as the - * physdev has the same MAC address as the macvtap device. - */ - if (virDomainNetGetActualDirectMode(net) == - VIR_NETDEV_MACVLAN_MODE_PASSTHRU && - physdev && virNetDevSetOnline(physdev, false) < 0) - return -1; - break; - } - - case VIR_DOMAIN_NET_TYPE_ETHERNET: - case VIR_DOMAIN_NET_TYPE_USER: - case VIR_DOMAIN_NET_TYPE_VHOSTUSER: - case VIR_DOMAIN_NET_TYPE_SERVER: - case VIR_DOMAIN_NET_TYPE_CLIENT: - case VIR_DOMAIN_NET_TYPE_MCAST: - case VIR_DOMAIN_NET_TYPE_UDP: - case VIR_DOMAIN_NET_TYPE_INTERNAL: - case VIR_DOMAIN_NET_TYPE_HOSTDEV: - case VIR_DOMAIN_NET_TYPE_VDPA: - case VIR_DOMAIN_NET_TYPE_NULL: - case VIR_DOMAIN_NET_TYPE_VDS: - case VIR_DOMAIN_NET_TYPE_LAST: - /* these types all require no action */ - break; - } - - return 0; -} - -/** - * qemuInterfaceStopDevices: - * @def: domain definition - * - * Make all interfaces associated with this domain inaccessible from - * the rest of the network. - */ -int -qemuInterfaceStopDevices(virDomainDef *def) -{ - size_t i; - - for (i = 0; i < def->nnets; i++) { - if (qemuInterfaceStopDevice(def->nets[i]) < 0) - return -1; - } - return 0; -} - - -static bool -qemuInterfaceIsVnetCompatModel(const virDomainNetDef *net) -{ - return (virDomainNetIsVirtioModel(net) || - net->model == VIR_DOMAIN_NET_MODEL_E1000E || - net->model == VIR_DOMAIN_NET_MODEL_IGB || - net->model == VIR_DOMAIN_NET_MODEL_VMXNET3); -} - - /** * qemuInterfaceDirectConnect: * @def: the definition of the VM (needed by 802.1Qbh and audit) @@ -271,7 +67,7 @@ qemuInterfaceDirectConnect(virDomainDef *def, unsigned int macvlan_create_flags = VIR_NETDEV_MACVLAN_CREATE_WITH_TAP; qemuDomainNetworkPrivate *netpriv = QEMU_DOMAIN_NETWORK_PRIVATE(net); - if (qemuInterfaceIsVnetCompatModel(net)) + if (virDomainInterfaceIsVnetCompatModel(net)) macvlan_create_flags |= VIR_NETDEV_MACVLAN_VNET_HDR; if (virNetDevMacVLanCreateWithVPortProfile(net->ifname, @@ -409,133 +205,6 @@ qemuCreateInBridgePortWithHelper(virQEMUDriverConfig *cfg, return *tapfd < 0 ? -1 : 0; } - -/* qemuInterfaceEthernetConnect: - * @def: the definition of the VM - * @driver: qemu driver data - * @net: pointer to the VM's interface description - * @tapfd: array of file descriptor return value for the new device - * @tapfdsize: number of file descriptors in @tapfd - * - * Called *only* called if actualType is VIR_DOMAIN_NET_TYPE_ETHERNET - * (i.e. if the connection is made with a tap device) - */ -int -qemuInterfaceEthernetConnect(virDomainDef *def, - virQEMUDriver *driver, - virDomainNetDef *net, - int *tapfd, - size_t tapfdSize) -{ - virMacAddr tapmac; - int ret = -1; - unsigned int tap_create_flags = VIR_NETDEV_TAP_CREATE_IFUP; - bool template_ifname = false; - g_autoptr(virQEMUDriverConfig) cfg = virQEMUDriverGetConfig(driver); - const char *tunpath = "/dev/net/tun"; - const char *auditdev = tunpath; - - if (net->backend.tap) { - tunpath = net->backend.tap; - if (!driver->privileged) { - virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s", - _("cannot use custom tap device in session mode")); - goto cleanup; - } - } - - if (qemuInterfaceIsVnetCompatModel(net)) - tap_create_flags |= VIR_NETDEV_TAP_CREATE_VNET_HDR; - - if (net->managed_tap == VIR_TRISTATE_BOOL_NO) { - if (!net->ifname) { - virReportError(VIR_ERR_INTERNAL_ERROR, "%s", - _("target dev must be supplied when managed='no'")); - goto cleanup; - } - if (virNetDevExists(net->ifname) != 1) { - virReportError(VIR_ERR_INTERNAL_ERROR, "%s", - _("target managed='no' but specified dev doesn't exist")); - goto cleanup; - } - - tap_create_flags |= VIR_NETDEV_TAP_CREATE_ALLOW_EXISTING; - - if (virNetDevMacVLanIsMacvtap(net->ifname)) { - auditdev = net->ifname; - if (virNetDevMacVLanTapOpen(net->ifname, tapfd, tapfdSize) < 0) - goto cleanup; - if (virNetDevMacVLanTapSetup(tapfd, tapfdSize, - qemuInterfaceIsVnetCompatModel(net)) < 0) { - goto cleanup; - } - } else { - if (virNetDevTapCreate(&net->ifname, tunpath, tapfd, tapfdSize, - tap_create_flags) < 0) - goto cleanup; - } - } else { - - if (!net->ifname) - template_ifname = true; - - if (virNetDevTapCreate(&net->ifname, tunpath, tapfd, tapfdSize, - tap_create_flags) < 0) { - goto cleanup; - } - - /* The tap device's MAC address cannot match the MAC address - * used by the guest. This results in "received packet on - * vnetX with own address as source address" error logs from - * the kernel. - */ - virMacAddrSet(&tapmac, &net->mac); - if (tapmac.addr[0] == 0xFE) - tapmac.addr[0] = 0xFA; - else - tapmac.addr[0] = 0xFE; - - if (virNetDevSetMAC(net->ifname, &tapmac) < 0) - goto cleanup; - - if (virNetDevSetOnline(net->ifname, true) < 0) - goto cleanup; - } - - if (net->script && - virNetDevRunEthernetScript(net->ifname, net->script) < 0) - goto cleanup; - - if (cfg->macFilter && - ebtablesAddForwardAllowIn(driver->ebtables, - net->ifname, - &net->mac) < 0) - goto cleanup; - - if (net->filter && - virDomainConfNWFilterInstantiate(def->name, def->uuid, net, false) < 0) { - goto cleanup; - } - - virDomainAuditNetDevice(def, net, auditdev, true); - - ret = 0; - - cleanup: - if (ret < 0) { - size_t i; - - virDomainAuditNetDevice(def, net, auditdev, false); - for (i = 0; i < tapfdSize && tapfd[i] >= 0; i++) - VIR_FORCE_CLOSE(tapfd[i]); - if (template_ifname) - VIR_FREE(net->ifname); - } - - return ret; -} - - /* qemuInterfaceBridgeConnect: * @def: the definition of the VM * @driver: qemu driver data @@ -578,7 +247,7 @@ qemuInterfaceBridgeConnect(virDomainDef *def, if (!net->ifname) template_ifname = true; - if (qemuInterfaceIsVnetCompatModel(net)) + if (virDomainInterfaceIsVnetCompatModel(net)) tap_create_flags |= VIR_NETDEV_TAP_CREATE_VNET_HDR; if (driver->privileged) { @@ -598,7 +267,7 @@ qemuInterfaceBridgeConnect(virDomainDef *def, * is attaching to, so we need to turn off learning and * unicast_flood on the device to prevent the kernel from * adding any FDB entries for it. We will add an fdb - * entry ourselves (during qemuInterfaceStartDevices(), + * entry ourselves (during virDomainInterfaceStartDevices(), * using the MAC address from the interface config. */ if (virNetDevBridgePortSetLearning(brname, net->ifname, false) < 0) diff --git a/src/qemu/qemu_interface.h b/src/qemu/qemu_interface.h index 6eed3e6bd7..47d7ec03fd 100644 --- a/src/qemu/qemu_interface.h +++ b/src/qemu/qemu_interface.h @@ -25,11 +25,6 @@ #include "qemu_domain.h" #include "qemu_slirp.h" -int qemuInterfaceStartDevice(virDomainNetDef *net); -int qemuInterfaceStartDevices(virDomainDef *def); -int qemuInterfaceStopDevice(virDomainNetDef *net); -int qemuInterfaceStopDevices(virDomainDef *def); - int qemuInterfaceDirectConnect(virDomainDef *def, virQEMUDriver *driver, virDomainNetDef *net, @@ -37,12 +32,6 @@ int qemuInterfaceDirectConnect(virDomainDef *def, size_t tapfdSize, virNetDevVPortProfileOp vmop); -int qemuInterfaceEthernetConnect(virDomainDef *def, - virQEMUDriver *driver, - virDomainNetDef *net, - int *tapfd, - size_t tapfdSize); - int qemuInterfaceBridgeConnect(virDomainDef *def, virQEMUDriver *driver, virDomainNetDef *net, diff --git a/src/qemu/qemu_process.c b/src/qemu/qemu_process.c index b28d53f280..aad406f679 100644 --- a/src/qemu/qemu_process.c +++ b/src/qemu/qemu_process.c @@ -74,6 +74,7 @@ #include "virhostcpu.h" #include "domain_audit.h" #include "domain_cgroup.h" +#include "domain_interface.h" #include "domain_nwfilter.h" #include "domain_postparse.h" #include "domain_validate.h" @@ -3121,7 +3122,7 @@ qemuProcessStartCPUs(virQEMUDriver *driver, virDomainObj *vm, g_autoptr(virQEMUDriverConfig) cfg = virQEMUDriverGetConfig(driver); /* Bring up netdevs before starting CPUs */ - if (qemuInterfaceStartDevices(vm->def) < 0) + if (virDomainInterfaceStartDevices(vm->def) < 0) return -1; VIR_DEBUG("Using lock state '%s'", NULLSTR(priv->lockState)); @@ -3184,7 +3185,7 @@ int qemuProcessStopCPUs(virQEMUDriver *driver, goto cleanup; /* de-activate netdevs after stopping CPUs */ - ignore_value(qemuInterfaceStopDevices(vm->def)); + ignore_value(virDomainInterfaceStopDevices(vm->def)); if (vm->job->current) ignore_value(virTimeMillisNow(&vm->job->current->stopped)); @@ -8362,11 +8363,9 @@ void qemuProcessStop(virQEMUDriver *driver, qemuDomainObjPrivate *priv = vm->privateData; virErrorPtr orig_err; virDomainDef *def = vm->def; - const virNetDevVPortProfile *vport = NULL; size_t i; g_autofree char *timestamp = NULL; g_autoptr(virQEMUDriverConfig) cfg = virQEMUDriverGetConfig(driver); - g_autoptr(virConnect) conn = NULL; bool outgoingMigration; VIR_DEBUG("Shutting down vm=%p name=%s id=%d pid=%lld, " @@ -8505,71 +8504,13 @@ void qemuProcessStop(virQEMUDriver *driver, } qemuHostdevReAttachDomainDevices(driver, vm->def); - for (i = 0; i < def->nnets; i++) { virDomainNetDef *net = def->nets[i]; - vport = virDomainNetGetActualVirtPortProfile(net); - switch (virDomainNetGetActualType(net)) { - case VIR_DOMAIN_NET_TYPE_DIRECT: - if (QEMU_DOMAIN_NETWORK_PRIVATE(net)->created) { - virNetDevMacVLanDeleteWithVPortProfile(net->ifname, &net->mac, - virDomainNetGetActualDirectDev(net), - virDomainNetGetActualDirectMode(net), - virDomainNetGetActualVirtPortProfile(net), - cfg->stateDir); - } - break; - case VIR_DOMAIN_NET_TYPE_ETHERNET: - if (net->managed_tap != VIR_TRISTATE_BOOL_NO && net->ifname) { - ignore_value(virNetDevTapDelete(net->ifname, net->backend.tap)); - VIR_FREE(net->ifname); - } - break; - case VIR_DOMAIN_NET_TYPE_BRIDGE: - case VIR_DOMAIN_NET_TYPE_NETWORK: -#ifdef VIR_NETDEV_TAP_REQUIRE_MANUAL_CLEANUP - if (!(vport && vport->virtPortType == VIR_NETDEV_VPORT_PROFILE_OPENVSWITCH)) - ignore_value(virNetDevTapDelete(net->ifname, net->backend.tap)); -#endif - break; - case VIR_DOMAIN_NET_TYPE_USER: - case VIR_DOMAIN_NET_TYPE_VHOSTUSER: - case VIR_DOMAIN_NET_TYPE_SERVER: - case VIR_DOMAIN_NET_TYPE_CLIENT: - case VIR_DOMAIN_NET_TYPE_MCAST: - case VIR_DOMAIN_NET_TYPE_INTERNAL: - case VIR_DOMAIN_NET_TYPE_HOSTDEV: - case VIR_DOMAIN_NET_TYPE_UDP: - case VIR_DOMAIN_NET_TYPE_VDPA: - case VIR_DOMAIN_NET_TYPE_NULL: - case VIR_DOMAIN_NET_TYPE_VDS: - case VIR_DOMAIN_NET_TYPE_LAST: - /* No special cleanup procedure for these types. */ - break; - } - /* release the physical device (or any other resources used by - * this interface in the network driver - */ - if (vport) { - if (vport->virtPortType == VIR_NETDEV_VPORT_PROFILE_MIDONET) { - ignore_value(virNetDevMidonetUnbindPort(vport)); - } else if (vport->virtPortType == VIR_NETDEV_VPORT_PROFILE_OPENVSWITCH) { - ignore_value(virNetDevOpenvswitchRemovePort( - virDomainNetGetActualBridgeName(net), - net->ifname)); - } - } - - /* kick the device out of the hostdev list too */ - virDomainNetRemoveHostdev(def, net); - if (net->type == VIR_DOMAIN_NET_TYPE_NETWORK) { - if (conn || (conn = virGetConnectNetwork())) - virDomainNetReleaseActualDevice(conn, net); - else - VIR_WARN("Unable to release network device '%s'", NULLSTR(net->ifname)); - } + virDomainInterfaceDeleteDevice(def, + net, + QEMU_DOMAIN_NETWORK_PRIVATE(net)->created, + cfg->stateDir); } - retry: if ((ret = virDomainCgroupRemoveCgroup(vm, priv->cgroup, priv->machineName)) < 0) { if (ret == -EBUSY && (retries++ < 5)) { -- 2.41.0

On 11/30/23 00:11, Praveen K Paladugu wrote: I don't think this can ever be true. But if you want to keep this check, I suggest you ditch the VIR_ERROR() call and replace it with setting errno. We don't really use VIR_ERROR() after logging subsystem was initialized. And a function can either report no errors (and leave it up to caller) or report errors in all cases. Simirarly, because of sendmsg() retval check below, caller should use virReportSystemError() (because errno is set on sendmsg() failure), so other error paths should set some sensible errno too. I believe SCM_RIGHT is undefined on anything else than Linux. Just like virSocketSendFD() we need alternative, dummy impl. Otherwise, WIN32 compilation fails since this function is defined only in !WIN32 block. This is something that should live in CH driver code. So if we receive more than 1024 bytes, then the 'buf' is not '\0' terminated ... [1] Until here it's generic enough so that it could live under src/util/, but ... [2] 1: ... and this will cause us to read beyond array boundaries. What we usually do in this case is: ret = recv(sock, buf, sizeof(buf) - 1, 0); buf[ret] = '\0'; Alternatively, you may init the buffer with zeroes: char buf[1024] = { 0 }; ret = recv(sock, buf, sizeof(buf) - 1, 0); 2: ... this is too specific IMO. Also, for esx driver, which also uses HTTP to talk to the hypervisor, we use libcurl. But I'm not sure if it can do FD passing. I doubt it can. It's okay to construct HTTP requests ourselves (just like you're doing in 5/5) then. Just a thought. Michal

This capability checks if ch can receive multiple fds along with net-add api. This capability is required to enable multiple queues for domain/guest interfaces. Signed-off-by: Praveen K Paladugu <prapal(a)linux.microsoft.com&gt; --- src/ch/ch_capabilities.c | 9 +++++++++ src/ch/ch_capabilities.h | 1 + 2 files changed, 10 insertions(+) diff --git a/src/ch/ch_capabilities.c b/src/ch/ch_capabilities.c index b10485820c..b9e62d2d5b 100644 --- a/src/ch/ch_capabilities.c +++ b/src/ch/ch_capabilities.c @@ -50,6 +50,15 @@ virCHCapsInitCHVersionCaps(int version) if (version >= 18000000) virCHCapsSet(chCaps, CH_SERIAL_CONSOLE_IN_PARALLEL); + /* Starting Version 22, add-net api can accept multiple FDs in the request + * This is required to be able to configure queues for virtio-net devices + * from libvirt. + * This capability will be used to gate networking support for ch guests. + * https://github.com/cloud-hypervisor/cloud-hypervisor/releases/tag/v22.0 + */ + if (version >= 22000000) + virCHCapsSet(chCaps, CH_MULTIFD_IN_ADDNET); + return g_steal_pointer(&chCaps); } diff --git a/src/ch/ch_capabilities.h b/src/ch/ch_capabilities.h index 703a6dbfe2..ffb8881a11 100644 --- a/src/ch/ch_capabilities.h +++ b/src/ch/ch_capabilities.h @@ -26,6 +26,7 @@ typedef enum { /* 0 */ CH_KERNEL_API_DEPRCATED, /* Use `payload` in place of `kernel` api */ CH_SERIAL_CONSOLE_IN_PARALLEL, /* Serial and Console ports can work in parallel */ + CH_MULTIFD_IN_ADDNET, /* Cloud-hypervisor can accept multiple FDs in add-net api */ CH_CAPS_LAST /* this must always be the last item */ } virCHCapsFlags; -- 2.41.0