[ovs-dev] [RFC PATCH 1/1] netdev-dpdk: vHost client mode and reconnect
Loftus, Ciara
ciara.loftus at intel.com
Wed Jul 13 08:29:52 UTC 2016
> Ciara Loftus <ciara.loftus at intel.com> writes:
>
> > A new other_config DB option has been added called 'vhost_driver_mode'.
> > By default this is set to 'server' which is the mode of operation OVS
> > with DPDK has used up until this point - whereby OVS creates and manages
> > vHost user sockets.
> >
> > If set to 'client', OVS will act as the vHost client and connect to
> > sockets created and managed by QEMU which acts as the server. This
> mode
> > allows for reconnect capability, which allows vHost ports to resume
> > normal connectivity in event of switch reset.
> >
> > QEMU v2.7.0+ is required when using OVS in client mode and QEMU in
> > server mode.
> >
> > Signed-off-by: Ciara Loftus <ciara.loftus at intel.com>
> > ---
> > INSTALL.DPDK-ADVANCED.md | 27 +++++++++++++++++++++++++++
> > NEWS | 1 +
> > lib/netdev-dpdk.c | 37 ++++++++++++++++++++++++++-----------
> > vswitchd/vswitch.xml | 13 +++++++++++++
> > 4 files changed, 67 insertions(+), 11 deletions(-)
> >
> > diff --git a/INSTALL.DPDK-ADVANCED.md b/INSTALL.DPDK-ADVANCED.md
> > index ec1de29..ad3e59e 100644
> > --- a/INSTALL.DPDK-ADVANCED.md
> > +++ b/INSTALL.DPDK-ADVANCED.md
> > @@ -489,6 +489,33 @@ DPDK 16.07 supports two types of vhost:
> > where `-L`: Changes the numbers of channels of the specified network
> device
> > and `combined`: Changes the number of multi-purpose channels.
> >
> > + 4. Enable OVS vHost client-mode & vHost reconnect (OPTIONAL)
> > +
> > + By default, OVS DPDK acts as the vHost socket server and QEMU the
> > + client. In QEMU v2.7 the option is available for QEMU to act as the
> > + server. In order for this to work, OVS DPDK must be switched to 'client'
> > + mode. This is possible by setting the 'vhost_driver_mode' DB entry to
> > + 'client' like so:
> > +
> > + ```
> > + ovs-vsctl set Open_vSwitch .
> other_config:vhost_driver_mode="client"
> > + ```
> > +
> > + This must be done before the switch is launched. It cannot sucessfully
> > + be changed after switch has launched.
> > +
> > + One must also append ',server' to the 'chardev' arguments on the
> QEMU
> > + command line, to instruct QEMU to use vHost server mode, like so:
> > +
> > + ````
> > + -chardev
> socket,id=char0,path=/usr/local/var/run/openvswitch/vhost0,server
> > + ````
> > +
> > + One benefit of using this mode is the ability for vHost ports to
> > + 'reconnect' in event of the switch crashing or being brought down.
> Once
> > + it is brought back up, the vHost ports will reconnect automatically and
> > + normal service will resume.
> > +
> > - VM Configuration with libvirt
> >
> > * change the user/group, access control policty and restart libvirtd.
> > diff --git a/NEWS b/NEWS
> > index f50b05e..08bac37 100644
> > --- a/NEWS
> > +++ b/NEWS
> > @@ -53,6 +53,7 @@ Post-v2.5.0
> > * PMD threads servicing vHost User ports can now come from the
> NUMA
> > node that device memory is located on if
> CONFIG_RTE_LIBRTE_VHOST_NUMA
> > is enabled in DPDK.
> > + * OVS client mode for vHost and vHost reconnect (Requires QEMU 2.7)
> > - ovs-benchmark: This utility has been removed due to lack of use and
> > bitrot.
> > - ovs-appctl:
> > diff --git a/lib/netdev-dpdk.c b/lib/netdev-dpdk.c
> > index 9cf0b0c..6763039 100644
> > --- a/lib/netdev-dpdk.c
> > +++ b/lib/netdev-dpdk.c
> > @@ -138,9 +138,11 @@ BUILD_ASSERT_DECL((MAX_NB_MBUF /
> ROUND_DOWN_POW2(MAX_NB_MBUF/MIN_NB_MBUF))
> > * yet mapped to another queue. */
> >
> > #ifdef VHOST_CUSE
> > -static char *cuse_dev_name = NULL; /* Character device
> cuse_dev_name. */
> > +static char *cuse_dev_name = NULL; /* Character device
> cuse_dev_name. */
> > +#else
> > +static char *vhost_sock_dir = NULL; /* Location of vhost-user sockets */
> > +static uint64_t vhost_driver_flags = 0; /* Denote whether client/server
> mode */
> > #endif
> > -static char *vhost_sock_dir = NULL; /* Location of vhost-user sockets */
> >
> > #define VHOST_ENQ_RETRY_NUM 8
> >
> > @@ -845,7 +847,6 @@ netdev_dpdk_vhost_user_construct(struct netdev
> *netdev)
> > struct netdev_dpdk *dev = netdev_dpdk_cast(netdev);
> > const char *name = netdev->name;
> > int err;
> > - uint64_t flags = 0;
> >
> > /* 'name' is appended to 'vhost_sock_dir' and used to create a socket in
> > * the file system. '/' or '\' would traverse directories, so they're not
> > @@ -868,14 +869,17 @@ netdev_dpdk_vhost_user_construct(struct
> netdev *netdev)
> > snprintf(dev->vhost_id, sizeof(dev->vhost_id), "%s/%s",
> > vhost_sock_dir, name);
> >
> > - err = rte_vhost_driver_register(dev->vhost_id, flags);
> > + err = rte_vhost_driver_register(dev->vhost_id, vhost_driver_flags);
> > if (err) {
> > VLOG_ERR("vhost-user socket device setup failure for socket %s\n",
> > dev->vhost_id);
> > } else {
> > - fatal_signal_add_file_to_unlink(dev->vhost_id);
> > - VLOG_INFO("Socket %s created for vhost-user port %s\n",
> > - dev->vhost_id, name);
> > + if (!(vhost_driver_flags & RTE_VHOST_USER_CLIENT)) {
> > + /* OVS server mode - OVS may delete the socket */
> > + fatal_signal_add_file_to_unlink(dev->vhost_id);
> > + VLOG_INFO("Socket %s created for vhost-user port %s\n",
> > + dev->vhost_id, name);
> > + }
> > err = vhost_construct_helper(netdev);
> > }
> >
> > @@ -945,7 +949,8 @@ netdev_dpdk_vhost_destruct(struct netdev
> *netdev)
> >
> > if (rte_vhost_driver_unregister(dev->vhost_id)) {
> > VLOG_ERR("Unable to remove vhost-user socket %s", dev->vhost_id);
> > - } else {
> > + } else if (!(vhost_driver_flags & RTE_VHOST_USER_CLIENT)) {
> > + /* OVS server mode - OVS may delete the socket */
> > fatal_signal_remove_file_to_unlink(dev->vhost_id);
> > }
> >
> > @@ -3208,6 +3213,7 @@ dpdk_init__(const struct smap
> *ovs_other_config)
> > cpu_set_t cpuset;
> > #ifndef VHOST_CUSE
> > char *sock_dir_subcomponent;
> > + const char *val;
> > #endif
> >
> > if (!smap_get_bool(ovs_other_config, "dpdk-init", false)) {
> > @@ -3218,8 +3224,8 @@ dpdk_init__(const struct smap
> *ovs_other_config)
> > VLOG_INFO("DPDK Enabled, initializing");
> >
> > #ifdef VHOST_CUSE
> > - if (process_vhost_flags("cuse-dev-name", xstrdup("vhost-net"),
> > - PATH_MAX, ovs_other_config, &cuse_dev_name)) {
> > + process_vhost_flags("cuse-dev-name", xstrdup("vhost-net"),
> > + PATH_MAX, ovs_other_config, &cuse_dev_name);
> > #else
> > if (process_vhost_flags("vhost-sock-dir", xstrdup(ovs_rundir()),
> > NAME_MAX, ovs_other_config,
> > @@ -3243,8 +3249,17 @@ dpdk_init__(const struct smap
> *ovs_other_config)
> > free(sock_dir_subcomponent);
> > } else {
> > vhost_sock_dir = sock_dir_subcomponent;
> > -#endif
> > }
> > + val = smap_get(ovs_other_config, "vhost_driver_mode");
> > + if (val != NULL && strncmp(val, "client", strlen(val)) == 0) {
> > + vhost_driver_flags |= RTE_VHOST_USER_CLIENT;
> > + VLOG_INFO("OVS client mode (QEMU server mode) selected for
> vHost");
> > + } else {
> > + /* default case */
> > + vhost_driver_flags &= ~RTE_VHOST_USER_CLIENT;
> > + VLOG_INFO("OVS server mode (QEMU client mode) selected for
> vHost");
> > + }
>
> Is it possible to just use the process_vhost_flags call for this section
> instead? Do you think it would make sense for consistency?
Good suggestion - I'll try make this change in the next version.
Thanks,
Ciara
>
> > +#endif
> >
> > argv = grow_argv(&argv, 0, 1);
> > argc = 1;
> > diff --git a/vswitchd/vswitch.xml b/vswitchd/vswitch.xml
> > index fed6f56..66007e5 100644
> > --- a/vswitchd/vswitch.xml
> > +++ b/vswitchd/vswitch.xml
> > @@ -311,6 +311,19 @@
> > </p>
> > </column>
> >
> > + <column name="other_config" key="vhost_driver_mode"
> > + type='{"type": "string", "enum": ["set", ["server", "client"]]}'>
> > + <p>
> > + Specifies which mode OVS will use for vHost. In 'server' mode, OVS
> > + creates and destroys the vHost User sockets. In 'client' mode, OVS
> > + attaches to sockets created by QEMU.
> > + </p>
> > + <p>
> > + Defaults to 'server' mode. Changing this value requires restarting
> > + the daemon.
> > + </p>
> > + </column>
> > +
> > <column name="other_config" key="n-handler-threads"
> > type='{"type": "integer", "minInteger": 1}'>
> > <p>
More information about the dev
mailing list