[libvirt] [PATCH] Revert "qemu: propagate bridge MTU into qemu "host_mtu" option"
This reverts commit 2841e675. It turns out that adding the host_mtu field to the PCI capabilities in the guest bumps the length of PCI capabilities beyond the 32 byte boundary, so the virtio-net device gets 64 bytes of ioport space instead of 32, which offsets the address of all the other following devices. Migration doesn't work very well when the location and length of PCI capabilities of devices is changed between source and destination. This means that we need to make sure that the absence/presence of host_mtu on the qemu commandline always matches between source and destination, which means that we need to make setting of host_mtu an opt-in thing (it can't happen automatically when the bridge being used has a non-default MTU, which is what commit 2841e675 implemented). I do want to re-implement this feature with an <mtu auto='on'/> setting, but probably won't backport that to any stable branches, so I'm first reverting the original commit, and that revert can be pushed to the few releases that have been made since the original (3.1.0 - 3.3.0) Resolves: https://bugzilla.redhat.com/1449346 --- src/qemu/qemu_command.c | 32 ++++++++++---------------------- src/qemu/qemu_command.h | 3 +-- src/qemu/qemu_hotplug.c | 5 ++--- src/qemu/qemu_interface.c | 5 ++--- src/qemu/qemu_interface.h | 3 +-- 5 files changed, 16 insertions(+), 32 deletions(-) diff --git a/src/qemu/qemu_command.c b/src/qemu/qemu_command.c index 4c1a266..aa66e3d 100644 --- a/src/qemu/qemu_command.c +++ b/src/qemu/qemu_command.c @@ -3633,8 +3633,7 @@ qemuBuildNicDevStr(virDomainDefPtr def, int vlan, unsigned int bootindex, size_t vhostfdSize, - virQEMUCapsPtr qemuCaps, - unsigned int mtu) + virQEMUCapsPtr qemuCaps) { virBuffer buf = VIR_BUFFER_INITIALIZER; const char *nic = net->model; @@ -3758,23 +3757,13 @@ qemuBuildNicDevStr(virDomainDefPtr def, virBufferAsprintf(&buf, ",rx_queue_size=%u", net->driver.virtio.rx_queue_size); } - if (usingVirtio && mtu) { - if (virQEMUCapsGet(qemuCaps, QEMU_CAPS_VIRTIO_NET_HOST_MTU)) { - - virBufferAsprintf(&buf, ",host_mtu=%u", mtu); - - } else { - /* log an error if mtu was requested specifically for this - * interface, otherwise, if it's just what was reported by - * the attached network, ignore it. - */ - if (net->mtu) { - virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s", - _("setting MTU is not supported with " - "this QEMU binary")); - goto error; - } + if (usingVirtio && net->mtu) { + if (!virQEMUCapsGet(qemuCaps, QEMU_CAPS_VIRTIO_NET_HOST_MTU)) { + virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s", + _("setting MTU is not supported with this QEMU binary")); + goto error; } + virBufferAsprintf(&buf, ",host_mtu=%u", net->mtu); } if (vlan == -1) @@ -8320,7 +8309,7 @@ qemuBuildVhostuserCommandLine(virQEMUDriverPtr driver, VIR_FREE(netdev); if (!(nic = qemuBuildNicDevStr(def, net, -1, bootindex, - queues, qemuCaps, net->mtu))) { + queues, qemuCaps))) { virReportError(VIR_ERR_INTERNAL_ERROR, "%s", _("Error generating NIC -device string")); goto error; @@ -8366,7 +8355,6 @@ qemuBuildInterfaceCommandLine(virQEMUDriverPtr driver, virDomainNetType actualType = virDomainNetGetActualType(net); virNetDevBandwidthPtr actualBandwidth; size_t i; - unsigned int mtu = net->mtu; if (!bootindex) @@ -8421,7 +8409,7 @@ qemuBuildInterfaceCommandLine(virQEMUDriverPtr driver, memset(tapfd, -1, tapfdSize * sizeof(tapfd[0])); if (qemuInterfaceBridgeConnect(def, driver, net, - tapfd, &tapfdSize, &mtu) < 0) + tapfd, &tapfdSize) < 0) goto cleanup; break; @@ -8601,7 +8589,7 @@ qemuBuildInterfaceCommandLine(virQEMUDriverPtr driver, } if (qemuDomainSupportsNicdev(def, net)) { if (!(nic = qemuBuildNicDevStr(def, net, vlan, bootindex, - vhostfdSize, qemuCaps, mtu))) + vhostfdSize, qemuCaps))) goto cleanup; virCommandAddArgList(cmd, "-device", nic, NULL); } else { diff --git a/src/qemu/qemu_command.h b/src/qemu/qemu_command.h index 7da92c8..09cb00e 100644 --- a/src/qemu/qemu_command.h +++ b/src/qemu/qemu_command.h @@ -101,8 +101,7 @@ char *qemuBuildNicDevStr(virDomainDefPtr def, int vlan, unsigned int bootindex, size_t vhostfdSize, - virQEMUCapsPtr qemuCaps, - unsigned int mtu); + virQEMUCapsPtr qemuCaps); char *qemuDeviceDriveHostAlias(virDomainDiskDefPtr disk); diff --git a/src/qemu/qemu_hotplug.c b/src/qemu/qemu_hotplug.c index e8d2918..a1909e3 100644 --- a/src/qemu/qemu_hotplug.c +++ b/src/qemu/qemu_hotplug.c @@ -968,7 +968,6 @@ qemuDomainAttachNetDevice(virQEMUDriverPtr driver, bool charDevPlugged = false; bool netdevPlugged = false; bool hostPlugged = false; - unsigned int mtu = net->mtu; /* preallocate new slot for device */ if (VIR_REALLOC_N(vm->def->nets, vm->def->nnets + 1) < 0) @@ -1025,7 +1024,7 @@ qemuDomainAttachNetDevice(virQEMUDriverPtr driver, goto cleanup; memset(vhostfd, -1, sizeof(*vhostfd) * vhostfdSize); if (qemuInterfaceBridgeConnect(vm->def, driver, net, - tapfd, &tapfdSize, &mtu) < 0) + tapfd, &tapfdSize) < 0) goto cleanup; iface_connected = true; if (qemuInterfaceOpenVhostNet(vm->def, net, priv->qemuCaps, @@ -1239,7 +1238,7 @@ qemuDomainAttachNetDevice(virQEMUDriverPtr driver, VIR_FORCE_CLOSE(vhostfd[i]); if (!(nicstr = qemuBuildNicDevStr(vm->def, net, vlan, 0, - queueSize, priv->qemuCaps, mtu))) + queueSize, priv->qemuCaps))) goto try_remove; qemuDomainObjEnterMonitor(driver, vm); diff --git a/src/qemu/qemu_interface.c b/src/qemu/qemu_interface.c index d8a678b..cebb490 100644 --- a/src/qemu/qemu_interface.c +++ b/src/qemu/qemu_interface.c @@ -503,8 +503,7 @@ qemuInterfaceBridgeConnect(virDomainDefPtr def, virQEMUDriverPtr driver, virDomainNetDefPtr net, int *tapfd, - size_t *tapfdSize, - unsigned int *mtu) + size_t *tapfdSize) { const char *brname; int ret = -1; @@ -545,7 +544,7 @@ qemuInterfaceBridgeConnect(virDomainDefPtr def, def->uuid, tunpath, tapfd, *tapfdSize, virDomainNetGetActualVirtPortProfile(net), virDomainNetGetActualVlan(net), - net->coalesce, net->mtu, mtu, + net->coalesce, 0, NULL, tap_create_flags) < 0) { virDomainAuditNetDevice(def, net, tunpath, false); goto cleanup; diff --git a/src/qemu/qemu_interface.h b/src/qemu/qemu_interface.h index ba74ac2..a7faa0b 100644 --- a/src/qemu/qemu_interface.h +++ b/src/qemu/qemu_interface.h @@ -51,8 +51,7 @@ int qemuInterfaceBridgeConnect(virDomainDefPtr def, virQEMUDriverPtr driver, virDomainNetDefPtr net, int *tapfd, - size_t *tapfdSize, - unsigned int *mtu) + size_t *tapfdSize) ATTRIBUTE_NONNULL(2); int qemuInterfaceOpenVhostNet(virDomainDefPtr def, -- 2.9.4
On Mon, May 22, 2017 at 03:30:48AM -0400, Laine Stump wrote:
This reverts commit 2841e675.
It turns out that adding the host_mtu field to the PCI capabilities in the guest bumps the length of PCI capabilities beyond the 32 byte boundary, so the virtio-net device gets 64 bytes of ioport space instead of 32, which offsets the address of all the other following devices. Migration doesn't work very well when the location and length of PCI capabilities of devices is changed between source and destination.
This means that we need to make sure that the absence/presence of host_mtu on the qemu commandline always matches between source and destination, which means that we need to make setting of host_mtu an opt-in thing (it can't happen automatically when the bridge being used has a non-default MTU, which is what commit 2841e675 implemented).
I do want to re-implement this feature with an <mtu auto='on'/> setting, but probably won't backport that to any stable branches, so I'm first reverting the original commit, and that revert can be pushed to the few releases that have been made since the original (3.1.0 - 3.3.0)
Resolves: https://bugzilla.redhat.com/1449346 --- src/qemu/qemu_command.c | 32 ++++++++++---------------------- src/qemu/qemu_command.h | 3 +-- src/qemu/qemu_hotplug.c | 5 ++--- src/qemu/qemu_interface.c | 5 ++--- src/qemu/qemu_interface.h | 3 +-- 5 files changed, 16 insertions(+), 32 deletions(-)
ACK, please push to all relevant branches too Regards, Daniel -- |: https://berrange.com -o- https://www.flickr.com/photos/dberrange :| |: https://libvirt.org -o- https://fstop138.berrange.com :| |: https://entangle-photo.org -o- https://www.instagram.com/dberrange :|
participants (2)
-
Daniel P. Berrange -
Laine Stump