Virtualization - Apr 2013 - [PATCH 0/2 V3] virtio-spec/net: dynamic network offloads configuration

From: Dmitry Fleytman <dfleytma at redhat.com>

V3 changes:
  1. Compat macro added
  2. Feature name beautification

V2 changes:
  1. _GUEST_ added to command and feature names
  2. Live migration logic fixed

Reported-by: Michael S. Tsirkin <mst at redhat.com>

One of recently introduced Windows features (RSC)
requires network driver to be able to enable and disable
HW LRO offload on the fly without device reinitialization.

Current Virtio specification doesn't support this requirement.
The solution proposed by following spec patch is to add
a new control command for this purpose.

The same solution may be used in Linux driver for ethtool interface
implementation.

Patch for QEMU mainline that implements this specification change
attached as well.

-- 
1.8.1.4

From: Dmitry Fleytman <dfleytma at redhat.com>

Virtio-net driver currently negotiates network offloads
on startup via features mechanism and have no ability to
change offloads state later.
This patch introduced a new control command that allows
to configure device network offloads state dynamically.
The patch also introduces a new feature flag
VIRTIO_NET_F_CTRL_GUEST_OFFLOADS.

Signed-off-by: Dmitry Fleytman <dfleytma at redhat.com>
---
 virtio-spec.lyx | 143 ++++++++++++++++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 143 insertions(+)

diff --git a/virtio-spec.lyx b/virtio-spec.lyx
index 3d2f485..fdba814 100644
--- a/virtio-spec.lyx
+++ b/virtio-spec.lyx
@@ -60,6 +60,7 @@
 \author -1930653948 "Amos Kong" 
 \author -608949062 "Rusty Russell,,," 
 \author -385801441 "Cornelia Huck" cornelia.huck at de.ibm.com
+\author 460276516 "Dmitry Fleytman" dfleytma at redhat.com
 \author 1112500848 "Rusty Russell" rusty at rustcorp.com.au
 \author 1531152142 "Paolo Bonzini,,," 
 \author 1717892615 "Alexey Zaytsev,,," 
@@ -4261,6 +4262,20 @@ VIRTIO_NET_F_GUEST_CSUM
 \end_inset
 
 (1) Guest handles packets with partial checksum
+\change_inserted 460276516 1363712169
+
+\end_layout
+
+\begin_layout Description
+
+\change_inserted 460276516 1363712334
+VIRTIO_NET_F_CTRL_GUEST_OFFLOADS
+\begin_inset space ~
+\end_inset
+
+(2) Control channel offloads reconfiguration support.
+\change_unchanged
+
 \end_layout
 
 \begin_layout Description
@@ -5675,6 +5690,134 @@ virtqueue_pairs = 1
  
 \end_layout
 
+\begin_layout Subsection*
+
+\change_inserted 460276516 1363765850
+Offloads State Configuration
+\end_layout
+
+\begin_layout Standard
+
+\change_inserted 460276516 1363765861
+If the VIRTIO_NET_F_CTRL_GUEST_OFFLOADS feature is negotiated, the driver can
+ send control commands for dynamic offloads state configuration.
+\end_layout
+
+\begin_layout Subsubsection*
+
+\change_inserted 460276516 1363765928
+Setting offloads state
+\end_layout
+
+\begin_layout Standard
+
+\change_inserted 460276516 1363713225
+\begin_inset listings
+inline false
+status open
+
+\begin_layout Plain Layout
+
+\change_inserted 460276516 1363765996
+
+u32 offloads;
+\end_layout
+
+\begin_layout Plain Layout
+
+\change_inserted 460276516 1363765997
+
+\end_layout
+
+\begin_layout Plain Layout
+
+\change_inserted 460276516 1363766044
+
+#define VIRTIO_NET_OFFLOAD_GUEST_CSUM       1
+\end_layout
+
+\begin_layout Plain Layout
+
+\change_inserted 460276516 1363766051
+
+#define VIRTIO_NET_OFFLOAD_GUEST_TSO4       2
+\end_layout
+
+\begin_layout Plain Layout
+
+\change_inserted 460276516 1363766055
+
+#define VIRTIO_NET_OFFLOAD_GUEST_TSO6       4
+\end_layout
+
+\begin_layout Plain Layout
+
+\change_inserted 460276516 1363766064
+
+#define VIRTIO_NET_OFFLOAD_GUEST_ECN        8
+\end_layout
+
+\begin_layout Plain Layout
+
+\change_inserted 460276516 1363766035
+
+#define VIRTIO_NET_OFFLOAD_GUEST_UFO       16
+\end_layout
+
+\begin_layout Plain Layout
+
+\change_inserted 460276516 1363766031
+
+\end_layout
+
+\begin_layout Plain Layout
+
+\change_inserted 460276516 1363765865
+
+#define VIRTIO_NET_CTRL_GUEST_OFFLOADS       5
+\end_layout
+
+\begin_layout Plain Layout
+
+\change_inserted 460276516 1363765867
+
+ #define VIRTIO_NET_CTRL_GUEST_OFFLOADS_SET          0
+\end_layout
+
+\end_inset
+
+
+\end_layout
+
+\begin_layout Standard
+
+\change_inserted 460276516 1363766082
+The class VIRTIO_NET_CTRL_GUEST_OFFLOADS has one command:
VIRTIO_NET_CTRL_GUEST_OFFLOADS_SET
+ applies the new offloads configuration.
+\end_layout
+
+\begin_layout Standard
+
+\change_inserted 460276516 1363766435
+u32 value passed as command data is a bitmask, bits set define offloads
+ to be enabled, bits cleared - offloads to be disabled.
+\end_layout
+
+\begin_layout Standard
+
+\change_inserted 460276516 1363766757
+There is a corresponding device feature for each offload.
+ Upon feature negotiation corresponding offload gets enabled to preserve
+ backward compartibility.
+\end_layout
+
+\begin_layout Standard
+
+\change_inserted 460276516 1363766720
+Corresponding feature must be negotiated at startup in order to allow dynamic
+ change of specific offload state.
+\end_layout
+
 \begin_layout Chapter*
 Appendix D: Block Device
 \end_layout
-- 
1.8.1.4

From: Dmitry Fleytman <dfleytma at redhat.com>

Virtio-net driver currently negotiates network offloads
on startup via features mechanism and have no ability to
change offloads state later.
This patch introduced a new control command that allows
to configure device network offloads state dynamically.
The patch also introduces a new feature flag
VIRTIO_NET_F_CTRL_GUEST_OFFLOADS.

Signed-off-by: Dmitry Fleytman <dfleytma at redhat.com>
---
 hw/pc.h         |   4 ++
 hw/virtio-net.c | 111 ++++++++++++++++++++++++++++++++++++++++++++++++--------
 hw/virtio-net.h |  19 ++++++++++
 3 files changed, 119 insertions(+), 15 deletions(-)

diff --git a/hw/pc.h b/hw/pc.h
index 8e1dd4c..7ca4698 100644
--- a/hw/pc.h
+++ b/hw/pc.h
@@ -221,6 +221,10 @@ int e820_add_entry(uint64_t, uint64_t, uint32_t);
             .property = "vectors",\
             /* DEV_NVECTORS_UNSPECIFIED as a uint32_t string */\
             .value    = stringify(0xFFFFFFFF),\
+        },{ \
+            .driver   = "virtio-net-pci", \
+            .property = "ctrl_guest_offloads", \
+            .value    = "off", \
         },{\
             .driver   = "e1000",\
             .property = "romfile",\
diff --git a/hw/virtio-net.c b/hw/virtio-net.c
index 5917740..d040f96 100644
--- a/hw/virtio-net.c
+++ b/hw/virtio-net.c
@@ -360,6 +360,48 @@ static uint32_t virtio_net_bad_features(VirtIODevice *vdev)
     return features;
 }
 
+static void virtio_net_apply_offloads(VirtIONet *n)
+{
+    tap_set_offload(qemu_get_subqueue(n->nic, 0)->peer,
+                    !!(n->curr_offloads &
VIRTIO_NET_OFFLOAD_GUEST_CSUM),
+                    !!(n->curr_offloads &
VIRTIO_NET_OFFLOAD_GUEST_TSO4),
+                    !!(n->curr_offloads &
VIRTIO_NET_OFFLOAD_GUEST_TSO6),
+                    !!(n->curr_offloads & VIRTIO_NET_OFFLOAD_GUEST_ECN),
+                    !!(n->curr_offloads &
VIRTIO_NET_OFFLOAD_GUEST_UFO));
+}
+
+static uint32_t virtio_net_offloads_by_features(uint32_t features)
+{
+    uint32_t offloads = 0;
+
+    if ((1 << VIRTIO_NET_F_GUEST_CSUM) & features) {
+        offloads |= VIRTIO_NET_OFFLOAD_GUEST_CSUM;
+    }
+
+    if ((1 << VIRTIO_NET_F_GUEST_TSO4) & features) {
+        offloads |= VIRTIO_NET_OFFLOAD_GUEST_TSO4;
+    }
+
+    if ((1 << VIRTIO_NET_F_GUEST_TSO6) & features) {
+        offloads |= VIRTIO_NET_OFFLOAD_GUEST_TSO6;
+    }
+
+    if ((1 << VIRTIO_NET_F_GUEST_ECN) & features) {
+        offloads |= VIRTIO_NET_OFFLOAD_GUEST_ECN;
+    }
+
+    if ((1 << VIRTIO_NET_F_GUEST_UFO) & features) {
+        offloads |= VIRTIO_NET_OFFLOAD_GUEST_UFO;
+    }
+
+    return offloads;
+}
+
+static inline uint32_t virtio_net_supported_offloads(VirtIONet *n)
+{
+    return virtio_net_offloads_by_features(n->vdev.guest_features);
+}
+
 static void virtio_net_set_features(VirtIODevice *vdev, uint32_t features)
 {
     VirtIONet *n = to_virtio_net(vdev);
@@ -371,12 +413,8 @@ static void virtio_net_set_features(VirtIODevice *vdev,
uint32_t features)
     virtio_net_set_mrg_rx_bufs(n, !!(features & (1 <<
VIRTIO_NET_F_MRG_RXBUF)));
 
     if (n->has_vnet_hdr) {
-        tap_set_offload(qemu_get_subqueue(n->nic, 0)->peer,
-                        (features >> VIRTIO_NET_F_GUEST_CSUM) & 1,
-                        (features >> VIRTIO_NET_F_GUEST_TSO4) & 1,
-                        (features >> VIRTIO_NET_F_GUEST_TSO6) & 1,
-                        (features >> VIRTIO_NET_F_GUEST_ECN)  & 1,
-                        (features >> VIRTIO_NET_F_GUEST_UFO)  & 1);
+        n->curr_offloads = virtio_net_offloads_by_features(features);
+        virtio_net_apply_offloads(n);
     }
 
     for (i = 0;  i < n->max_queues; i++) {
@@ -422,6 +460,42 @@ static int virtio_net_handle_rx_mode(VirtIONet *n, uint8_t
cmd,
     return VIRTIO_NET_OK;
 }
 
+static int virtio_net_handle_offloads(VirtIONet *n, uint8_t cmd,
+                                     struct iovec *iov, unsigned int iov_cnt)
+{
+    uint32_t offloads;
+    size_t s;
+
+    if (!((1 << VIRTIO_NET_F_CTRL_GUEST_OFFLOADS) &
n->vdev.guest_features)) {
+        return VIRTIO_NET_ERR;
+    }
+
+    s = iov_to_buf(iov, iov_cnt, 0, &offloads, sizeof(offloads));
+    if (s != sizeof(offloads)) {
+        return VIRTIO_NET_ERR;
+    }
+
+    if (cmd == VIRTIO_NET_CTRL_GUEST_OFFLOADS_SET) {
+        uint32_t supported_offloads;
+
+        if (!n->has_vnet_hdr) {
+            return VIRTIO_NET_ERR;
+        }
+
+        supported_offloads = virtio_net_supported_offloads(n);
+        if (offloads & ~supported_offloads) {
+            return VIRTIO_NET_ERR;
+        }
+
+        n->curr_offloads = offloads;
+        virtio_net_apply_offloads(n);
+
+        return VIRTIO_NET_OK;
+    } else {
+        return VIRTIO_NET_ERR;
+    }
+}
+
 static int virtio_net_handle_mac(VirtIONet *n, uint8_t cmd,
                                  struct iovec *iov, unsigned int iov_cnt)
 {
@@ -591,6 +665,8 @@ static void virtio_net_handle_ctrl(VirtIODevice *vdev,
VirtQueue *vq)
             status = virtio_net_handle_vlan_table(n, ctrl.cmd, iov, iov_cnt);
         } else if (ctrl.class == VIRTIO_NET_CTRL_MQ) {
             status = virtio_net_handle_mq(n, ctrl.cmd, iov, iov_cnt);
+        } else if (ctrl.class == VIRTIO_NET_CTRL_GUEST_OFFLOADS) {
+            status = virtio_net_handle_offloads(n, ctrl.cmd, iov, iov_cnt);
         }
 
         s = iov_from_buf(elem.in_sg, elem.in_num, 0, &status,
sizeof(status));
@@ -1100,6 +1176,10 @@ static void virtio_net_save(QEMUFile *f, void *opaque)
             qemu_put_be32(f, n->vqs[i].tx_waiting);
         }
     }
+
+    if ((1 << VIRTIO_NET_F_CTRL_GUEST_OFFLOADS) &
n->vdev.guest_features) {
+        qemu_put_be32(f, n->curr_offloads);
+    }
 }
 
 static int virtio_net_load(QEMUFile *f, void *opaque, int version_id)
@@ -1156,15 +1236,6 @@ static int virtio_net_load(QEMUFile *f, void *opaque, int
version_id)
             error_report("virtio-net: saved image requires
vnet_hdr=on");
             return -1;
         }
-
-        if (n->has_vnet_hdr) {
-            tap_set_offload(qemu_get_queue(n->nic)->peer,
-                    (n->vdev.guest_features >>
VIRTIO_NET_F_GUEST_CSUM) & 1,
-                    (n->vdev.guest_features >>
VIRTIO_NET_F_GUEST_TSO4) & 1,
-                    (n->vdev.guest_features >>
VIRTIO_NET_F_GUEST_TSO6) & 1,
-                    (n->vdev.guest_features >> VIRTIO_NET_F_GUEST_ECN)
& 1,
-                    (n->vdev.guest_features >> VIRTIO_NET_F_GUEST_UFO)
& 1);
-        }
     }
 
     if (version_id >= 9) {
@@ -1198,6 +1269,16 @@ static int virtio_net_load(QEMUFile *f, void *opaque, int
version_id)
         }
     }
 
+    if ((1 << VIRTIO_NET_F_CTRL_GUEST_OFFLOADS) &
n->vdev.guest_features) {
+        n->curr_offloads = qemu_get_be32(f);
+    } else {
+        n->curr_offloads = virtio_net_supported_offloads(n);
+    }
+
+    if (peer_has_vnet_hdr(n)) {
+        virtio_net_apply_offloads(n);
+    }
+
     virtio_net_set_queues(n);
 
     /* Find the first multicast entry in the saved MAC filter */
diff --git a/hw/virtio-net.h b/hw/virtio-net.h
index 4d1a8cd..7e93368 100644
--- a/hw/virtio-net.h
+++ b/hw/virtio-net.h
@@ -27,6 +27,8 @@
 /* The feature bitmap for virtio net */
 #define VIRTIO_NET_F_CSUM       0       /* Host handles pkts w/ partial csum */
 #define VIRTIO_NET_F_GUEST_CSUM 1       /* Guest handles pkts w/ partial csum
*/
+#define VIRTIO_NET_F_CTRL_GUEST_OFFLOADS 2 /* Control channel offload
+                                         * configuration support */
 #define VIRTIO_NET_F_MAC        5       /* Host has given MAC address. */
 #define VIRTIO_NET_F_GSO        6       /* Host handles pkts w/ any GSO type */
 #define VIRTIO_NET_F_GUEST_TSO4 7       /* Guest can handle TSOv4 in. */
@@ -182,6 +184,7 @@ typedef struct VirtIONet {
     uint16_t max_queues;
     uint16_t curr_queues;
     size_t config_size;
+    uint32_t curr_offloads;
 } VirtIONet;
 
 #define VIRTIO_NET_CTRL_MAC    1
@@ -221,6 +224,21 @@ struct virtio_net_ctrl_mq {
  #define VIRTIO_NET_CTRL_MQ_VQ_PAIRS_MIN        1
  #define VIRTIO_NET_CTRL_MQ_VQ_PAIRS_MAX        0x8000
 
+/*
+ * Control network offloads
+ *
+ * Dynamic offloads are available with the
+ * VIRTIO_NET_F_CTRL_GUEST_OFFLOADS feature bit.
+ */
+#define VIRTIO_NET_CTRL_GUEST_OFFLOADS   5
+ #define VIRTIO_NET_CTRL_GUEST_OFFLOADS_SET        0
+
+#define VIRTIO_NET_OFFLOAD_GUEST_CSUM       1
+#define VIRTIO_NET_OFFLOAD_GUEST_TSO4       2
+#define VIRTIO_NET_OFFLOAD_GUEST_TSO6       4
+#define VIRTIO_NET_OFFLOAD_GUEST_ECN        8
+#define VIRTIO_NET_OFFLOAD_GUEST_UFO       16
+
 #define DEFINE_VIRTIO_NET_FEATURES(_state, _field) \
         DEFINE_VIRTIO_COMMON_FEATURES(_state, _field), \
         DEFINE_PROP_BIT("csum", _state, _field, VIRTIO_NET_F_CSUM,
true), \
@@ -241,6 +259,7 @@ struct virtio_net_ctrl_mq {
         DEFINE_PROP_BIT("ctrl_vlan", _state, _field,
VIRTIO_NET_F_CTRL_VLAN, true), \
         DEFINE_PROP_BIT("ctrl_rx_extra", _state, _field,
VIRTIO_NET_F_CTRL_RX_EXTRA, true), \
         DEFINE_PROP_BIT("ctrl_mac_addr", _state, _field,
VIRTIO_NET_F_CTRL_MAC_ADDR, true), \
+        DEFINE_PROP_BIT("ctrl_guest_offloads", _state, _field,
VIRTIO_NET_F_CTRL_GUEST_OFFLOADS, true), \
         DEFINE_PROP_BIT("mq", _state, _field, VIRTIO_NET_F_MQ, false)
 
 #endif
-- 
1.8.1.4

On Tue, Apr 02, 2013 at 02:38:58PM +0300, Dmitry Fleytman
wrote:
> From: Dmitry Fleytman <dfleytma at redhat.com>
> 
> V3 changes:
>   1. Compat macro added
>   2. Feature name beautification
> 
> V2 changes:
>   1. _GUEST_ added to command and feature names
>   2. Live migration logic fixed
> 
> Reported-by: Michael S. Tsirkin <mst at redhat.com>
Acked-by: Michael S. Tsirkin <mst at redhat.com>

Rusty?
> One of recently introduced Windows features (RSC)
> requires network driver to be able to enable and disable
> HW LRO offload on the fly without device reinitialization.
> 
> Current Virtio specification doesn't support this requirement.
> The solution proposed by following spec patch is to add
> a new control command for this purpose.
> 
> The same solution may be used in Linux driver for ethtool interface
> implementation.
> 
> Patch for QEMU mainline that implements this specification change
> attached as well.
> 
> -- 
> 1.8.1.4