[PATCH v3 0/2] Vhost-vdpa Shadow Virtqueue _F_CTRL_RX_EXTRA commands support

[PATCH v3 0/2] Vhost-vdpa Shadow Virtqueue _F_CTRL_RX_EXTRA commands support

[Hawkins Jiawei]

This series enables shadowed CVQ to intercept rx commands related to
VIRTIO_NET_F_CTRL_RX_EXTRA feature through shadowed CVQ, update the virtio
NIC device model so qemu send it in a migration, and the restore of
that rx state in the destination.

To test this patch series, one should modify the `n->parent_obj.guest_features`
value in vhost_vdpa_net_load_rx() using gdb, as the linux virtio-net
driver does not currently support the VIRTIO_NET_F_CTRL_RX_EXTRA
feature.

Note that this patch should be based on
[1] patch "Vhost-vdpa Shadow Virtqueue _F_CTRL_RX commands support"

[1]. https://lore.kernel.org/all/cover.1688743107.git.yin31149@gmail.com/

TestStep
========
1. test the patch series using vp-vdpa device

  - For L0 guest, boot QEMU with virtio-net-pci net device with
`ctrl_vq`, `ctrl_rx` and `ctrl_rx_extra` feature on, something like:
      -device virtio-net-pci,rx_queue_size=256,tx_queue_size=256,
iommu_platform=on,ctrl_vq=on,ctrl_rx=on,ctrl_rx_extra=on...

  - For L1 guest, apply the patch series and compile the code,
start QEMU with vdpa device with svq mode and enable the
`ctrl_vq`, `ctrl_rx` and `ctrl_rx_extra` feature on, something like:
      -netdev type=vhost-vdpa,x-svq=true,...
      -device virtio-net-pci,ctrl_vq=on,ctrl_rx=on,ctrl_rx_extra=on...
Use gdb to attach the VM and break at the net/vhost-vdpa.c:870.

With this series, gdb can hit the breakpoint. Enable the
VIRTIO_NET_F_CTRL_RX_EXTRA feature and enable the non-unicast mode
by entering the following gdb commands:
```gdb
set n->parent_obj.guest_features |= (1 << 20)
set n->nouni = 1
c
```
QEMU should not trigger any errors or warnings.

Without this series, QEMU should fail with
"x-svq=true: vdpa svq does not work with features 0x100000".

ChangeLog
=========
v3:
  - return early if mismatch the condition suggested by Eugenio in
patch 1 "vdpa: Restore packet receive filtering state relative with
_F_CTRL_RX_EXTRA feature"
  - remove the `on` variable suggested by Eugenio in patch 1 "vdpa:
Restore packet receive filtering state relative with
_F_CTRL_RX_EXTRA feature"

v2: https://lore.kernel.org/all/cover.1688365324.git.yin31149@gmail.com/
  - avoid sending CVQ command in default state suggested by Eugenio

v1: https://lists.nongnu.org/archive/html/qemu-devel/2023-06/msg04956.html

Hawkins Jiawei (2):
  vdpa: Restore packet receive filtering state relative with
    _F_CTRL_RX_EXTRA feature
  vdpa: Allow VIRTIO_NET_F_CTRL_RX_EXTRA in SVQ

 net/vhost-vdpa.c | 89 ++++++++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 89 insertions(+)

-- 
2.25.1

[PATCH v3 1/2] vdpa: Restore packet receive filtering state relative with _F_CTRL_RX_EXTRA feature

[Hawkins Jiawei]

This patch refactors vhost_vdpa_net_load_rx() to
restore the packet receive filtering state in relation to
VIRTIO_NET_F_CTRL_RX_EXTRA feature at device's startup.

Signed-off-by: Hawkins Jiawei <yin31149@gmail.com>
---
v3:
  - return early if mismatch the condition suggested by Eugenio
  - remove the `on` variable suggested by Eugenio

v2: https://lore.kernel.org/all/66ec4d7e3a680de645043d0331ab65940154f2b8.1688365324.git.yin31149@gmail.com/
  - avoid sending CVQ command in default state suggested by Eugenio

v1: https://lists.nongnu.org/archive/html/qemu-devel/2023-06/msg04957.html

 net/vhost-vdpa.c | 88 ++++++++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 88 insertions(+)

diff --git a/net/vhost-vdpa.c b/net/vhost-vdpa.c
index 0994836f8c..9a1905fddd 100644
--- a/net/vhost-vdpa.c
+++ b/net/vhost-vdpa.c
@@ -867,6 +867,94 @@ static int vhost_vdpa_net_load_rx(VhostVDPAState *s,
         }
     }
 
+    if (!virtio_vdev_has_feature(&n->parent_obj, VIRTIO_NET_F_CTRL_RX_EXTRA)) {
+        return 0;
+    }
+
+    /*
+     * According to virtio_net_reset(), device turns all-unicast mode
+     * off by default.
+     *
+     * Therefore, QEMU should only send this CVQ command if the driver
+     * sets all-unicast mode on, different from the device's defaults.
+     *
+     * Note that the device's defaults can mismatch the driver's
+     * configuration only at live migration.
+     */
+    if (n->alluni) {
+        dev_written = vhost_vdpa_net_load_rx_mode(s,
+                                            VIRTIO_NET_CTRL_RX_ALLUNI, 1);
+        if (dev_written < 0) {
+            return dev_written;
+        }
+        if (*s->status != VIRTIO_NET_OK) {
+            return -EIO;
+        }
+    }
+
+    /*
+     * According to virtio_net_reset(), device turns non-multicast mode
+     * off by default.
+     *
+     * Therefore, QEMU should only send this CVQ command if the driver
+     * sets non-multicast mode on, different from the device's defaults.
+     *
+     * Note that the device's defaults can mismatch the driver's
+     * configuration only at live migration.
+     */
+    if (n->nomulti) {
+        dev_written = vhost_vdpa_net_load_rx_mode(s,
+                                            VIRTIO_NET_CTRL_RX_NOMULTI, 1);
+        if (dev_written < 0) {
+            return dev_written;
+        }
+        if (*s->status != VIRTIO_NET_OK) {
+            return -EIO;
+        }
+    }
+
+    /*
+     * According to virtio_net_reset(), device turns non-unicast mode
+     * off by default.
+     *
+     * Therefore, QEMU should only send this CVQ command if the driver
+     * sets non-unicast mode on, different from the device's defaults.
+     *
+     * Note that the device's defaults can mismatch the driver's
+     * configuration only at live migration.
+     */
+    if (n->nouni) {
+        dev_written = vhost_vdpa_net_load_rx_mode(s,
+                                            VIRTIO_NET_CTRL_RX_NOUNI, 1);
+        if (dev_written < 0) {
+            return dev_written;
+        }
+        if (*s->status != VIRTIO_NET_OK) {
+            return -EIO;
+        }
+    }
+
+    /*
+     * According to virtio_net_reset(), device turns non-broadcast mode
+     * off by default.
+     *
+     * Therefore, QEMU should only send this CVQ command if the driver
+     * sets non-broadcast mode on, different from the device's defaults.
+     *
+     * Note that the device's defaults can mismatch the driver's
+     * configuration only at live migration.
+     */
+    if (n->nobcast) {
+        dev_written = vhost_vdpa_net_load_rx_mode(s,
+                                            VIRTIO_NET_CTRL_RX_NOBCAST, 1);
+        if (dev_written < 0) {
+            return dev_written;
+        }
+        if (*s->status != VIRTIO_NET_OK) {
+            return -EIO;
+        }
+    }
+
     return 0;
 }
 
-- 
2.25.1

[PATCH v3 2/2] vdpa: Allow VIRTIO_NET_F_CTRL_RX_EXTRA in SVQ

[Hawkins Jiawei]

Enable SVQ with VIRTIO_NET_F_CTRL_RX_EXTRA feature.

Signed-off-by: Hawkins Jiawei <yin31149@gmail.com>
Acked-by: Eugenio Pérez <eperezma@redhat.com>
---
 net/vhost-vdpa.c | 1 +
 1 file changed, 1 insertion(+)

diff --git a/net/vhost-vdpa.c b/net/vhost-vdpa.c
index 9a1905fddd..1df82636c9 100644
--- a/net/vhost-vdpa.c
+++ b/net/vhost-vdpa.c
@@ -105,6 +105,7 @@ static const uint64_t vdpa_svq_device_features =
     BIT_ULL(VIRTIO_NET_F_STATUS) |
     BIT_ULL(VIRTIO_NET_F_CTRL_VQ) |
     BIT_ULL(VIRTIO_NET_F_CTRL_RX) |
+    BIT_ULL(VIRTIO_NET_F_CTRL_RX_EXTRA) |
     BIT_ULL(VIRTIO_NET_F_MQ) |
     BIT_ULL(VIRTIO_F_ANY_LAYOUT) |
     BIT_ULL(VIRTIO_NET_F_CTRL_MAC_ADDR) |
-- 
2.25.1