virtio_net: allow hypervisor to indicate linkspeed and duplex setting

PATCH Superseded Jason Baron 3 participants 18 messages Dec 14, 2017 – Dec 21, 2017

Subsystems (3)

Network drivers virtio core virtio net driver

lore

Share & tools

lore

Jason Baron

author

2017-12-14 19:34 UTC #0 lore raw

cover virtio_net: allow hypervisor to indicate linkspeed and duplex setting

We have found it useful to be able to set the linkspeed and duplex settings from the host-side for virtio_net. This obviates the need for guest changes and settings for these fields.

The ability to set linkspeed and duplex was introduced by:

16032be virtio_net: add ethtool support for set and get of settings

And using 'ethtool -s' continues to over-write the linkspeed/duplex settings with this patch.

The 1/2 patch is against net-next, while the 2/2 patch is the associated qemu changes that would go in after as update-linux-headers.sh should be run first. So 2/2 is more meant as a demonstration of how I intend this to work.

Thanks,

-Jason

Jason Baron (2):
virtio_net: allow hypervisor to indicate linkspeed and duplex setting
qemu: add linkspeed and duplex setting to virtio-net

linux changes:

 drivers/net/virtio_net.c        | 11 ++++++++++-
 include/uapi/linux/virtio_net.h |  4 ++++
 2 files changed, 14 insertions(+), 1 deletion(-)

qemu changes:

 hw/net/virtio-net.c                         | 29 +++++++++++++++++++++++++++++                                
 include/hw/virtio/virtio-net.h              |  3 +++                                                          
 include/standard-headers/linux/virtio_net.h |  4 ++++                                                         
 3 files changed, 36 insertions(+)

Signature · 1 line

2.6.1

Jason Baron

author

2017-12-14 19:35 UTC #1 lore raw

PATCH 1/2 virtio_net: allow hypervisor to indicate linkspeed and duplex setting

If the hypervisor exports the link and duplex speed, let's use that instead of the default unknown speed. The user can still overwrite it later if desired via: 'ethtool -s'. This allows the hypervisor to set the default link speed and duplex setting without requiring guest changes and is consistent with how other network drivers operate. We ran into some cases where the guest software was failing due to a lack of linkspeed and had to fall back to a fully emulated network device that does export a linkspeed and duplex setting.

Implement by adding a new VIRTIO_NET_F_SPEED_DUPLEX feature flag, to indicate that a linkspeed and duplex setting are present.

Signed-off-by Jason Baron <[email protected]>

Cc "Michael S. Tsirkin" <[email protected]>

Cc Jason Wang <[email protected]>

---
 drivers/net/virtio_net.c        | 11 ++++++++++-
 include/uapi/linux/virtio_net.h |  4 ++++
 2 files changed, 14 insertions(+), 1 deletion(-)

drivers/net/virtio_net.c +10 -1

index 6fb7b65..e7a2ad6 100644
@@ -2671,6 +2671,14 @@ static int virtnet_probe(struct virtio_device *vdev)
 	netif_set_real_num_rx_queues(dev, vi->curr_queue_pairs);
 
 	virtnet_init_settings(dev);
+	if (virtio_has_feature(vdev, VIRTIO_NET_F_SPEED_DUPLEX)) {
+		vi->speed = virtio_cread32(vdev,
+					offsetof(struct virtio_net_config,
+					speed));
+		vi->duplex = virtio_cread8(vdev,
+					offsetof(struct virtio_net_config,
+					duplex));
+	}
 
 	err = register_netdev(dev);
 	if (err) {

@@ -2796,7 +2804,8 @@ static struct virtio_device_id id_table[] = {
 	VIRTIO_NET_F_CTRL_RX, VIRTIO_NET_F_CTRL_VLAN, \
 	VIRTIO_NET_F_GUEST_ANNOUNCE, VIRTIO_NET_F_MQ, \
 	VIRTIO_NET_F_CTRL_MAC_ADDR, \
-	VIRTIO_NET_F_MTU, VIRTIO_NET_F_CTRL_GUEST_OFFLOADS
+	VIRTIO_NET_F_MTU, VIRTIO_NET_F_CTRL_GUEST_OFFLOADS, \
+	VIRTIO_NET_F_SPEED_DUPLEX
 
 static unsigned int features[] = {
 	VIRTNET_FEATURES,

include/uapi/linux/virtio_net.h +4 -0

index fc353b5..acfcf68 100644
@@ -36,6 +36,7 @@
 #define VIRTIO_NET_F_GUEST_CSUM	1	/* Guest handles pkts w/ partial csum */
 #define VIRTIO_NET_F_CTRL_GUEST_OFFLOADS 2 /* Dynamic offload configuration. */
 #define VIRTIO_NET_F_MTU	3	/* Initial MTU advice */
+#define VIRTIO_NET_F_SPEED_DUPLEX 4	/* Host set linkspeed and duplex */
 #define VIRTIO_NET_F_MAC	5	/* Host has given MAC address. */
 #define VIRTIO_NET_F_GUEST_TSO4	7	/* Guest can handle TSOv4 in. */
 #define VIRTIO_NET_F_GUEST_TSO6	8	/* Guest can handle TSOv6 in. */

@@ -76,6 +77,9 @@ struct virtio_net_config {
 	__u16 max_virtqueue_pairs;
 	/* Default maximum transmit unit advice */
 	__u16 mtu;
+	/* Host exported linkspeed and duplex */
+	__u32 speed;
+	__u8 duplex;
 } __attribute__((packed));
 
 /*

Signature · 1 line

2.6.1

Michael S. Tsirkin

maintainer

2017-12-14 20:02 UTC #2 lore raw

Quoted 9 lines · Jason Baron

If the hypervisor exports the link and duplex speed, let's use that instead of the default unknown speed. The user can still overwrite it later if desired via: 'ethtool -s'. This allows the hypervisor to set the default link speed and duplex setting without requiring guest changes and is consistent with how other network drivers operate. We ran into some cases where the guest software was failing due to a lack of linkspeed and had to fall back to a fully emulated network device that does export a linkspeed and duplex setting.

Quoted 6 lines

Implement by adding a new VIRTIO_NET_F_SPEED_DUPLEX feature flag, to indicate that a linkspeed and duplex setting are present.
Signed-off-by Jason Baron <[email protected]>
Cc "Michael S. Tsirkin" <[email protected]>
Cc Jason Wang <[email protected]>

Sounds fine, but please register the new feature bit with the virtio TC by sending en email to the virtio mailing list (subscriber only, wish I could fix that).

We do not want conflicts there.

Quoted 58 lines

---
 drivers/net/virtio_net.c        | 11 ++++++++++-
 include/uapi/linux/virtio_net.h |  4 ++++
 2 files changed, 14 insertions(+), 1 deletion(-)

drivers/net/virtio_net.c +10 -1

index 6fb7b65..e7a2ad6 100644
@@ -2671,6 +2671,14 @@ static int virtnet_probe(struct virtio_device *vdev)
 	netif_set_real_num_rx_queues(dev, vi->curr_queue_pairs);
 
 	virtnet_init_settings(dev);
+	if (virtio_has_feature(vdev, VIRTIO_NET_F_SPEED_DUPLEX)) {
+		vi->speed = virtio_cread32(vdev,
+					offsetof(struct virtio_net_config,
+					speed));
+		vi->duplex = virtio_cread8(vdev,
+					offsetof(struct virtio_net_config,
+					duplex));
+	}
 
 	err = register_netdev(dev);
 	if (err) {

@@ -2796,7 +2804,8 @@ static struct virtio_device_id id_table[] = {
 	VIRTIO_NET_F_CTRL_RX, VIRTIO_NET_F_CTRL_VLAN, \
 	VIRTIO_NET_F_GUEST_ANNOUNCE, VIRTIO_NET_F_MQ, \
 	VIRTIO_NET_F_CTRL_MAC_ADDR, \
-	VIRTIO_NET_F_MTU, VIRTIO_NET_F_CTRL_GUEST_OFFLOADS
+	VIRTIO_NET_F_MTU, VIRTIO_NET_F_CTRL_GUEST_OFFLOADS, \
+	VIRTIO_NET_F_SPEED_DUPLEX
 
 static unsigned int features[] = {
 	VIRTNET_FEATURES,

include/uapi/linux/virtio_net.h +4 -0

index fc353b5..acfcf68 100644
@@ -36,6 +36,7 @@
 #define VIRTIO_NET_F_GUEST_CSUM	1	/* Guest handles pkts w/ partial csum */
 #define VIRTIO_NET_F_CTRL_GUEST_OFFLOADS 2 /* Dynamic offload configuration. */
 #define VIRTIO_NET_F_MTU	3	/* Initial MTU advice */
+#define VIRTIO_NET_F_SPEED_DUPLEX 4	/* Host set linkspeed and duplex */
 #define VIRTIO_NET_F_MAC	5	/* Host has given MAC address. */
 #define VIRTIO_NET_F_GUEST_TSO4	7	/* Guest can handle TSOv4 in. */
 #define VIRTIO_NET_F_GUEST_TSO6	8	/* Guest can handle TSOv6 in. */

@@ -76,6 +77,9 @@ struct virtio_net_config {
 	__u16 max_virtqueue_pairs;
 	/* Default maximum transmit unit advice */
 	__u16 mtu;
+	/* Host exported linkspeed and duplex */
+	__u32 speed;
+	__u8 duplex;
 } __attribute__((packed));
 
 /*

Signature · 1 line

2.6.1

Michael S. Tsirkin

maintainer

2017-12-20 14:57 UTC #3 lore raw

In reply to Jason Baron

Quoted 29 lines · Jason Baron

If the hypervisor exports the link and duplex speed, let's use that instead of the default unknown speed. The user can still overwrite it later if desired via: 'ethtool -s'. This allows the hypervisor to set the default link speed and duplex setting without requiring guest changes and is consistent with how other network drivers operate. We ran into some cases where the guest software was failing due to a lack of linkspeed and had to fall back to a fully emulated network device that does export a linkspeed and duplex setting.
Implement by adding a new VIRTIO_NET_F_SPEED_DUPLEX feature flag, to indicate that a linkspeed and duplex setting are present.
Signed-off-by Jason Baron <[email protected]>
Cc "Michael S. Tsirkin" <[email protected]>
Cc Jason Wang <[email protected]>
---
 drivers/net/virtio_net.c        | 11 ++++++++++-
 include/uapi/linux/virtio_net.h |  4 ++++
 2 files changed, 14 insertions(+), 1 deletion(-)
drivers/net/virtio_net.c +1 -0
index 6fb7b65..e7a2ad6 100644
@@ -2671,6 +2671,14 @@ static int virtnet_probe(struct virtio_device *vdev)
 	netif_set_real_num_rx_queues(dev, vi->curr_queue_pairs);
 
 	virtnet_init_settings(dev);
+	if (virtio_has_feature(vdev, VIRTIO_NET_F_SPEED_DUPLEX)) {

drivers/net/virtio_net.c static int virtnet_probe(struct virtio_device *vdev) · line 2684

Quoted 10 lines

drivers/net/virtio_net.c +7 -0

+		vi->speed = virtio_cread32(vdev,
+					offsetof(struct virtio_net_config,
+					speed));
+		vi->duplex = virtio_cread8(vdev,
+					offsetof(struct virtio_net_config,
+					duplex));
+	}
 
 	err = register_netdev(dev);
 	if (err) {

How are we going to validate speed values? Imagine host using a new 1000Gbit device and exposing that to guest.

Need to think what do we want guest to do.
I think that ideally we'd say it's a 100Gbit device.

For duplex, force to one of 3 valid values?

Quoted 15 lines

drivers/net/virtio_net.c +2 -1

@@ -2796,7 +2804,8 @@ static struct virtio_device_id id_table[] = {
 	VIRTIO_NET_F_CTRL_RX, VIRTIO_NET_F_CTRL_VLAN, \
 	VIRTIO_NET_F_GUEST_ANNOUNCE, VIRTIO_NET_F_MQ, \
 	VIRTIO_NET_F_CTRL_MAC_ADDR, \
-	VIRTIO_NET_F_MTU, VIRTIO_NET_F_CTRL_GUEST_OFFLOADS
+	VIRTIO_NET_F_MTU, VIRTIO_NET_F_CTRL_GUEST_OFFLOADS, \
+	VIRTIO_NET_F_SPEED_DUPLEX
 
 static unsigned int features[] = {
 	VIRTNET_FEATURES,

include/uapi/linux/virtio_net.h +0 -0

index fc353b5..acfcf68 100644

include/uapi/linux/virtio_net.h line 42

Quoted 7 lines

include/uapi/linux/virtio_net.h +1 -0

@@ -36,6 +36,7 @@
 #define VIRTIO_NET_F_GUEST_CSUM	1	/* Guest handles pkts w/ partial csum */
 #define VIRTIO_NET_F_CTRL_GUEST_OFFLOADS 2 /* Dynamic offload configuration. */
 #define VIRTIO_NET_F_MTU	3	/* Initial MTU advice */
+#define VIRTIO_NET_F_SPEED_DUPLEX 4	/* Host set linkspeed and duplex */
 #define VIRTIO_NET_F_MAC	5	/* Host has given MAC address. */
 #define VIRTIO_NET_F_GUEST_TSO4	7	/* Guest can handle TSOv4 in. */
 #define VIRTIO_NET_F_GUEST_TSO6	8	/* Guest can handle TSOv6 in. */

I think I'd prefer a high feature bit - low bits are ones that can be backported to legacy interfaces, so I think we should hang on to these for fixing issues that break communication completely (like the mtu).

Quoted 12 lines

include/uapi/linux/virtio_net.h +3 -0

@@ -76,6 +77,9 @@ struct virtio_net_config {
 	__u16 max_virtqueue_pairs;
 	/* Default maximum transmit unit advice */
 	__u16 mtu;
+	/* Host exported linkspeed and duplex */
+	__u32 speed;
+	__u8 duplex;
 } __attribute__((packed));
 
 /*

Signature · 1 line

2.6.1

Jason Baron

author

2017-12-20 17:08 UTC #4 lore raw

Quoted 44 lines · Michael S. Tsirkin

Quoted 29 lines · Jason Baron
If the hypervisor exports the link and duplex speed, let's use that instead of the default unknown speed. The user can still overwrite it later if desired via: 'ethtool -s'. This allows the hypervisor to set the default link speed and duplex setting without requiring guest changes and is consistent with how other network drivers operate. We ran into some cases where the guest software was failing due to a lack of linkspeed and had to fall back to a fully emulated network device that does export a linkspeed and duplex setting.
Implement by adding a new VIRTIO_NET_F_SPEED_DUPLEX feature flag, to indicate that a linkspeed and duplex setting are present.
Signed-off-by Jason Baron <[email protected]>
Cc "Michael S. Tsirkin" <[email protected]>
Cc Jason Wang <[email protected]>
---
 drivers/net/virtio_net.c        | 11 ++++++++++-
 include/uapi/linux/virtio_net.h |  4 ++++
 2 files changed, 14 insertions(+), 1 deletion(-)
drivers/net/virtio_net.c +1 -0
index 6fb7b65..e7a2ad6 100644
@@ -2671,6 +2671,14 @@ static int virtnet_probe(struct virtio_device *vdev)
 	netif_set_real_num_rx_queues(dev, vi->curr_queue_pairs);
 
 	virtnet_init_settings(dev);
+	if (virtio_has_feature(vdev, VIRTIO_NET_F_SPEED_DUPLEX)) {
drivers/net/virtio_net.c static int virtnet_probe(struct virtio_device *vdev) · line 2684
Quoted 10 lines
drivers/net/virtio_net.c +7 -0
+		vi->speed = virtio_cread32(vdev,
+					offsetof(struct virtio_net_config,
+					speed));
+		vi->duplex = virtio_cread8(vdev,
+					offsetof(struct virtio_net_config,
+					duplex));
+	}
 
 	err = register_netdev(dev);
 	if (err) {
How are we going to validate speed values? Imagine host using a new 1000Gbit device and exposing that to guest.

Quoted 4 lines

Need to think what do we want guest to do.
I think that ideally we'd say it's a 100Gbit device.
For duplex, force to one of 3 valid values?

So I didn't provide validation here b/c as you point out its not clear how we would validate it. I don't believe h/w drivers do any validation here either. They simply propagate the value from the the underlying device. So that seemed reasonable to me.

Why do you divide by 10 in the above example? Would you propose always dividing what the device reports by 10?

Quoted 21 lines

drivers/net/virtio_net.c +2 -1

@@ -2796,7 +2804,8 @@ static struct virtio_device_id id_table[] = {
 	VIRTIO_NET_F_CTRL_RX, VIRTIO_NET_F_CTRL_VLAN, \
 	VIRTIO_NET_F_GUEST_ANNOUNCE, VIRTIO_NET_F_MQ, \
 	VIRTIO_NET_F_CTRL_MAC_ADDR, \
-	VIRTIO_NET_F_MTU, VIRTIO_NET_F_CTRL_GUEST_OFFLOADS
+	VIRTIO_NET_F_MTU, VIRTIO_NET_F_CTRL_GUEST_OFFLOADS, \
+	VIRTIO_NET_F_SPEED_DUPLEX
 
 static unsigned int features[] = {
 	VIRTNET_FEATURES,

include/uapi/linux/virtio_net.h +1 -0

index fc353b5..acfcf68 100644
@@ -36,6 +36,7 @@
 #define VIRTIO_NET_F_GUEST_CSUM	1	/* Guest handles pkts w/ partial csum */
 #define VIRTIO_NET_F_CTRL_GUEST_OFFLOADS 2 /* Dynamic offload configuration. */
 #define VIRTIO_NET_F_MTU	3	/* Initial MTU advice */
+#define VIRTIO_NET_F_SPEED_DUPLEX 4	/* Host set linkspeed and duplex */

Quoted 9 lines

include/uapi/linux/virtio_net.h line 42
Quoted 3 lines
include/uapi/linux/virtio_net.h +0 -0
 #define VIRTIO_NET_F_MAC	5	/* Host has given MAC address. */
 #define VIRTIO_NET_F_GUEST_TSO4	7	/* Guest can handle TSOv4 in. */
 #define VIRTIO_NET_F_GUEST_TSO6	8	/* Guest can handle TSOv6 in. */
I think I'd prefer a high feature bit - low bits are ones that can be backported to legacy interfaces, so I think we should hang on to these for fixing issues that break communication completely (like the mtu).

So I went with a low bit here b/c in the virtio spec 'section 2.2 Feature Bits':

0 to 23
Feature bits for the specific device type
24 to 32
Feature bits reserved for extensions to the queue and feature
negotiation mechanisms
33 and above
Feature bits reserved for future extensions.

So virtio_net already goes up to 23 (but omits 4 and 6), and I wasn't sure if it was reasonable to use the higher bits. It looks like the code would handle the higher bits ok, so I can try that - bit 33 perhaps ?

Thanks,

-Jason

Quoted 13 lines

include/uapi/linux/virtio_net.h +3 -0

@@ -76,6 +77,9 @@ struct virtio_net_config {
 	__u16 max_virtqueue_pairs;
 	/* Default maximum transmit unit advice */
 	__u16 mtu;
+	/* Host exported linkspeed and duplex */
+	__u32 speed;
+	__u8 duplex;
 } __attribute__((packed));
 
 /*

Signature · 1 line

2.6.1

Michael S. Tsirkin

maintainer

2017-12-20 17:52 UTC #5 lore raw

Quoted 47 lines · Jason Baron

Quoted 44 lines · Michael S. Tsirkin
Quoted 29 lines · Jason Baron
If the hypervisor exports the link and duplex speed, let's use that instead of the default unknown speed. The user can still overwrite it later if desired via: 'ethtool -s'. This allows the hypervisor to set the default link speed and duplex setting without requiring guest changes and is consistent with how other network drivers operate. We ran into some cases where the guest software was failing due to a lack of linkspeed and had to fall back to a fully emulated network device that does export a linkspeed and duplex setting.
Implement by adding a new VIRTIO_NET_F_SPEED_DUPLEX feature flag, to indicate that a linkspeed and duplex setting are present.
Signed-off-by Jason Baron <[email protected]>
Cc "Michael S. Tsirkin" <[email protected]>
Cc Jason Wang <[email protected]>
---
 drivers/net/virtio_net.c        | 11 ++++++++++-
 include/uapi/linux/virtio_net.h |  4 ++++
 2 files changed, 14 insertions(+), 1 deletion(-)
drivers/net/virtio_net.c +1 -0
index 6fb7b65..e7a2ad6 100644
@@ -2671,6 +2671,14 @@ static int virtnet_probe(struct virtio_device *vdev)
 	netif_set_real_num_rx_queues(dev, vi->curr_queue_pairs);
 
 	virtnet_init_settings(dev);
+	if (virtio_has_feature(vdev, VIRTIO_NET_F_SPEED_DUPLEX)) {
drivers/net/virtio_net.c static int virtnet_probe(struct virtio_device *vdev) · line 2684
Quoted 10 lines
drivers/net/virtio_net.c +7 -0
+		vi->speed = virtio_cread32(vdev,
+					offsetof(struct virtio_net_config,
+					speed));
+		vi->duplex = virtio_cread8(vdev,
+					offsetof(struct virtio_net_config,
+					duplex));
+	}
 
 	err = register_netdev(dev);
 	if (err) {
How are we going to validate speed values? Imagine host using a new 1000Gbit device and exposing that to guest.

Quoted 8 lines

Quoted 4 lines
Need to think what do we want guest to do.
I think that ideally we'd say it's a 100Gbit device.
For duplex, force to one of 3 valid values?
So I didn't provide validation here b/c as you point out its not clear how we would validate it. I don't believe h/w drivers do any validation here either.

Right but hardware tends not to change as quickly as the hypervisors :) For virtual device drivers, we need some way to handle forward compatibility since hypervisors do change quite quickly.

Quoted 5 lines

They simply propagate the value from the the underlying device. So that seemed reasonable to me.
Why do you divide by 10 in the above example? Would you propose always dividing what the device reports by 10?

No, that was just an example. I was just suggesting rounding down to next valid known speed.

Quoted 43 lines

Quoted 21 lines
Quoted 21 lines
drivers/net/virtio_net.c +2 -1
@@ -2796,7 +2804,8 @@ static struct virtio_device_id id_table[] = {
 	VIRTIO_NET_F_CTRL_RX, VIRTIO_NET_F_CTRL_VLAN, \
 	VIRTIO_NET_F_GUEST_ANNOUNCE, VIRTIO_NET_F_MQ, \
 	VIRTIO_NET_F_CTRL_MAC_ADDR, \
-	VIRTIO_NET_F_MTU, VIRTIO_NET_F_CTRL_GUEST_OFFLOADS
+	VIRTIO_NET_F_MTU, VIRTIO_NET_F_CTRL_GUEST_OFFLOADS, \
+	VIRTIO_NET_F_SPEED_DUPLEX
 
 static unsigned int features[] = {
 	VIRTNET_FEATURES,
include/uapi/linux/virtio_net.h +1 -0
index fc353b5..acfcf68 100644
@@ -36,6 +36,7 @@
 #define VIRTIO_NET_F_GUEST_CSUM	1	/* Guest handles pkts w/ partial csum */
 #define VIRTIO_NET_F_CTRL_GUEST_OFFLOADS 2 /* Dynamic offload configuration. */
 #define VIRTIO_NET_F_MTU	3	/* Initial MTU advice */
+#define VIRTIO_NET_F_SPEED_DUPLEX 4	/* Host set linkspeed and duplex */
Quoted 9 lines
include/uapi/linux/virtio_net.h line 42
Quoted 3 lines
include/uapi/linux/virtio_net.h +0 -0
 #define VIRTIO_NET_F_MAC	5	/* Host has given MAC address. */
 #define VIRTIO_NET_F_GUEST_TSO4	7	/* Guest can handle TSOv4 in. */
 #define VIRTIO_NET_F_GUEST_TSO6	8	/* Guest can handle TSOv6 in. */
I think I'd prefer a high feature bit - low bits are ones that can be backported to legacy interfaces, so I think we should hang on to these for fixing issues that break communication completely (like the mtu).
So I went with a low bit here b/c in the virtio spec 'section 2.2 Feature Bits':
0 to 23
Feature bits for the specific device type
24 to 32
Feature bits reserved for extensions to the queue and feature
negotiation mechanisms
33 and above
Feature bits reserved for future extensions.

Quoted 7 lines

So virtio_net already goes up to 23 (but omits 4 and 6), and I wasn't sure if it was reasonable to use the higher bits. It looks like the code would handle the higher bits ok, so I can try that - bit 33 perhaps ?
Thanks,
-Jason

Transports started from bit 24 and are growing up. So I would say devices should start from bit 63 and grow down.

Quoted 14 lines

include/uapi/linux/virtio_net.h +3 -0

@@ -76,6 +77,9 @@ struct virtio_net_config {
 	__u16 max_virtqueue_pairs;
 	/* Default maximum transmit unit advice */
 	__u16 mtu;
+	/* Host exported linkspeed and duplex */
+	__u32 speed;
+	__u8 duplex;
 } __attribute__((packed));
 
 /*

Signature · 1 line

2.6.1

Jason Baron

author

2017-12-20 21:33 UTC #6 lore raw

Quoted 61 lines · Michael S. Tsirkin

Quoted 47 lines · Jason Baron
Quoted 44 lines · Michael S. Tsirkin
Quoted 29 lines · Jason Baron
If the hypervisor exports the link and duplex speed, let's use that instead of the default unknown speed. The user can still overwrite it later if desired via: 'ethtool -s'. This allows the hypervisor to set the default link speed and duplex setting without requiring guest changes and is consistent with how other network drivers operate. We ran into some cases where the guest software was failing due to a lack of linkspeed and had to fall back to a fully emulated network device that does export a linkspeed and duplex setting.
Implement by adding a new VIRTIO_NET_F_SPEED_DUPLEX feature flag, to indicate that a linkspeed and duplex setting are present.
Signed-off-by Jason Baron <[email protected]>
Cc "Michael S. Tsirkin" <[email protected]>
Cc Jason Wang <[email protected]>
---
 drivers/net/virtio_net.c        | 11 ++++++++++-
 include/uapi/linux/virtio_net.h |  4 ++++
 2 files changed, 14 insertions(+), 1 deletion(-)
drivers/net/virtio_net.c +1 -0
index 6fb7b65..e7a2ad6 100644
@@ -2671,6 +2671,14 @@ static int virtnet_probe(struct virtio_device *vdev)
 	netif_set_real_num_rx_queues(dev, vi->curr_queue_pairs);
 
 	virtnet_init_settings(dev);
+	if (virtio_has_feature(vdev, VIRTIO_NET_F_SPEED_DUPLEX)) {
drivers/net/virtio_net.c static int virtnet_probe(struct virtio_device *vdev) · line 2684
Quoted 10 lines
drivers/net/virtio_net.c +7 -0
+		vi->speed = virtio_cread32(vdev,
+					offsetof(struct virtio_net_config,
+					speed));
+		vi->duplex = virtio_cread8(vdev,
+					offsetof(struct virtio_net_config,
+					duplex));
+	}
 
 	err = register_netdev(dev);
 	if (err) {
How are we going to validate speed values? Imagine host using a new 1000Gbit device and exposing that to guest.
Quoted 8 lines
Quoted 4 lines
Need to think what do we want guest to do.
I think that ideally we'd say it's a 100Gbit device.
For duplex, force to one of 3 valid values?
So I didn't provide validation here b/c as you point out its not clear how we would validate it. I don't believe h/w drivers do any validation here either.
Right but hardware tends not to change as quickly as the hypervisors :) For virtual device drivers, we need some way to handle forward compatibility since hypervisors do change quite quickly.

Quoted 8 lines

Quoted 5 lines
They simply propagate the value from the the underlying device. So that seemed reasonable to me.
Why do you divide by 10 in the above example? Would you propose always dividing what the device reports by 10?
No, that was just an example. I was just suggesting rounding down to next valid known speed.

I see, but virtio currently uses ethtool_validate_speed() which allows arbitrary values up to INT_MAX in units of Mbps. That seems to leave plenty of headroom. So I could use that function for validation as well as well as ethtool_validate_duplex() and if they fail fall back to SPEED_UNKNOWN and DUPLEX_UNKNOWN?

Quoted 48 lines

Quoted 48 lines
Quoted 22 lines
Quoted 22 lines
drivers/net/virtio_net.c +2 -1
@@ -2796,7 +2804,8 @@ static struct virtio_device_id id_table[] = {
 	VIRTIO_NET_F_CTRL_RX, VIRTIO_NET_F_CTRL_VLAN, \
 	VIRTIO_NET_F_GUEST_ANNOUNCE, VIRTIO_NET_F_MQ, \
 	VIRTIO_NET_F_CTRL_MAC_ADDR, \
-	VIRTIO_NET_F_MTU, VIRTIO_NET_F_CTRL_GUEST_OFFLOADS
+	VIRTIO_NET_F_MTU, VIRTIO_NET_F_CTRL_GUEST_OFFLOADS, \
+	VIRTIO_NET_F_SPEED_DUPLEX
 
 static unsigned int features[] = {
 	VIRTNET_FEATURES,
include/uapi/linux/virtio_net.h +1 -0
index fc353b5..acfcf68 100644
@@ -36,6 +36,7 @@
 #define VIRTIO_NET_F_GUEST_CSUM	1	/* Guest handles pkts w/ partial csum */
 #define VIRTIO_NET_F_CTRL_GUEST_OFFLOADS 2 /* Dynamic offload configuration. */
 #define VIRTIO_NET_F_MTU	3	/* Initial MTU advice */
+#define VIRTIO_NET_F_SPEED_DUPLEX 4	/* Host set linkspeed and duplex */
Quoted 9 lines
include/uapi/linux/virtio_net.h line 42
Quoted 3 lines
include/uapi/linux/virtio_net.h +0 -0
 #define VIRTIO_NET_F_MAC	5	/* Host has given MAC address. */
 #define VIRTIO_NET_F_GUEST_TSO4	7	/* Guest can handle TSOv4 in. */
 #define VIRTIO_NET_F_GUEST_TSO6	8	/* Guest can handle TSOv6 in. */
I think I'd prefer a high feature bit - low bits are ones that can be backported to legacy interfaces, so I think we should hang on to these for fixing issues that break communication completely (like the mtu).
So I went with a low bit here b/c in the virtio spec 'section 2.2 Feature Bits':
0 to 23
Feature bits for the specific device type
24 to 32
Feature bits reserved for extensions to the queue and feature
negotiation mechanisms
33 and above
Feature bits reserved for future extensions.
So virtio_net already goes up to 23 (but omits 4 and 6), and I wasn't sure if it was reasonable to use the higher bits. It looks like the code would handle the higher bits ok, so I can try that - bit 33 perhaps ?

Quoted 8 lines

Quoted 3 lines
Thanks,
-Jason
Transports started from bit 24 and are growing up. So I would say devices should start from bit 63 and grow down.

Ok, I will use 63.

Thanks,

-Jason

Michael S. Tsirkin

maintainer

2017-12-21 00:10 UTC #7 lore raw

Quoted 70 lines · Jason Baron

Quoted 67 lines · Michael S. Tsirkin
Quoted 47 lines · Jason Baron
Quoted 44 lines · Michael S. Tsirkin
Quoted 29 lines · Jason Baron
If the hypervisor exports the link and duplex speed, let's use that instead of the default unknown speed. The user can still overwrite it later if desired via: 'ethtool -s'. This allows the hypervisor to set the default link speed and duplex setting without requiring guest changes and is consistent with how other network drivers operate. We ran into some cases where the guest software was failing due to a lack of linkspeed and had to fall back to a fully emulated network device that does export a linkspeed and duplex setting.
Implement by adding a new VIRTIO_NET_F_SPEED_DUPLEX feature flag, to indicate that a linkspeed and duplex setting are present.
Signed-off-by Jason Baron <[email protected]>
Cc "Michael S. Tsirkin" <[email protected]>
Cc Jason Wang <[email protected]>
---
 drivers/net/virtio_net.c        | 11 ++++++++++-
 include/uapi/linux/virtio_net.h |  4 ++++
 2 files changed, 14 insertions(+), 1 deletion(-)
drivers/net/virtio_net.c +1 -0
index 6fb7b65..e7a2ad6 100644
@@ -2671,6 +2671,14 @@ static int virtnet_probe(struct virtio_device *vdev)
 	netif_set_real_num_rx_queues(dev, vi->curr_queue_pairs);
 
 	virtnet_init_settings(dev);
+	if (virtio_has_feature(vdev, VIRTIO_NET_F_SPEED_DUPLEX)) {
drivers/net/virtio_net.c static int virtnet_probe(struct virtio_device *vdev) · line 2684
Quoted 10 lines
drivers/net/virtio_net.c +7 -0
+		vi->speed = virtio_cread32(vdev,
+					offsetof(struct virtio_net_config,
+					speed));
+		vi->duplex = virtio_cread8(vdev,
+					offsetof(struct virtio_net_config,
+					duplex));
+	}
 
 	err = register_netdev(dev);
 	if (err) {
How are we going to validate speed values? Imagine host using a new 1000Gbit device and exposing that to guest.
Quoted 8 lines
Quoted 4 lines
Need to think what do we want guest to do.
I think that ideally we'd say it's a 100Gbit device.
For duplex, force to one of 3 valid values?
So I didn't provide validation here b/c as you point out its not clear how we would validate it. I don't believe h/w drivers do any validation here either.
Right but hardware tends not to change as quickly as the hypervisors :) For virtual device drivers, we need some way to handle forward compatibility since hypervisors do change quite quickly.
Quoted 5 lines
They simply propagate the value from the the underlying device. So that seemed reasonable to me.
Why do you divide by 10 in the above example? Would you propose always dividing what the device reports by 10?

Quoted 8 lines

Quoted 2 lines
No, that was just an example. I was just suggesting rounding down to next valid known speed.
I see, but virtio currently uses ethtool_validate_speed() which allows arbitrary values up to INT_MAX in units of Mbps. That seems to leave plenty of headroom. So I could use that function for validation as well as well as ethtool_validate_duplex() and if they fail fall back to SPEED_UNKNOWN and DUPLEX_UNKNOWN?

Sounds good.

Quoted 63 lines

Quoted 48 lines
Quoted 48 lines
Quoted 22 lines
Quoted 22 lines
drivers/net/virtio_net.c +2 -1
@@ -2796,7 +2804,8 @@ static struct virtio_device_id id_table[] = {
 	VIRTIO_NET_F_CTRL_RX, VIRTIO_NET_F_CTRL_VLAN, \
 	VIRTIO_NET_F_GUEST_ANNOUNCE, VIRTIO_NET_F_MQ, \
 	VIRTIO_NET_F_CTRL_MAC_ADDR, \
-	VIRTIO_NET_F_MTU, VIRTIO_NET_F_CTRL_GUEST_OFFLOADS
+	VIRTIO_NET_F_MTU, VIRTIO_NET_F_CTRL_GUEST_OFFLOADS, \
+	VIRTIO_NET_F_SPEED_DUPLEX
 
 static unsigned int features[] = {
 	VIRTNET_FEATURES,
include/uapi/linux/virtio_net.h +1 -0
index fc353b5..acfcf68 100644
@@ -36,6 +36,7 @@
 #define VIRTIO_NET_F_GUEST_CSUM	1	/* Guest handles pkts w/ partial csum */
 #define VIRTIO_NET_F_CTRL_GUEST_OFFLOADS 2 /* Dynamic offload configuration. */
 #define VIRTIO_NET_F_MTU	3	/* Initial MTU advice */
+#define VIRTIO_NET_F_SPEED_DUPLEX 4	/* Host set linkspeed and duplex */
Quoted 9 lines
include/uapi/linux/virtio_net.h line 42
Quoted 3 lines
include/uapi/linux/virtio_net.h +0 -0
 #define VIRTIO_NET_F_MAC	5	/* Host has given MAC address. */
 #define VIRTIO_NET_F_GUEST_TSO4	7	/* Guest can handle TSOv4 in. */
 #define VIRTIO_NET_F_GUEST_TSO6	8	/* Guest can handle TSOv6 in. */
I think I'd prefer a high feature bit - low bits are ones that can be backported to legacy interfaces, so I think we should hang on to these for fixing issues that break communication completely (like the mtu).
So I went with a low bit here b/c in the virtio spec 'section 2.2 Feature Bits':
0 to 23
Feature bits for the specific device type
24 to 32
Feature bits reserved for extensions to the queue and feature
negotiation mechanisms
33 and above
Feature bits reserved for future extensions.
So virtio_net already goes up to 23 (but omits 4 and 6), and I wasn't sure if it was reasonable to use the higher bits. It looks like the code would handle the higher bits ok, so I can try that - bit 33 perhaps ?
Quoted 8 lines
Quoted 3 lines
Thanks,
-Jason
Transports started from bit 24 and are growing up. So I would say devices should start from bit 63 and grow down.
Ok, I will use 63.
Thanks,
-Jason

Jason Baron

author

2017-12-14 19:34 UTC #8 lore raw

PATCH 2/2 qemu: add linkspeed and duplex setting to virtio-net

In reply to Jason Baron

Although they can be currently set in linux via 'ethtool -s', this requires guest changes, and thus it would be nice to extend this functionality such that it can be configured automatically from the host (as other network do).

Linkspeed and duplex settings can be set as:
'-device virtio-net,speed=10000,duplex=full'

where speed is [-1...INT_MAX], and duplex is ["half"|"full"].

Signed-off-by Jason Baron <[email protected]>

Cc "Michael S. Tsirkin" <[email protected]>

Cc Jason Wang <[email protected]>

---
 hw/net/virtio-net.c                         | 29 +++++++++++++++++++++++++++++
 include/hw/virtio/virtio-net.h              |  3 +++
 include/standard-headers/linux/virtio_net.h |  4 ++++
 3 files changed, 36 insertions(+)

hw/net/virtio-net.c +29 -0

index 38674b0..d63e790 100644
@@ -40,6 +40,12 @@
 #define VIRTIO_NET_RX_QUEUE_MIN_SIZE VIRTIO_NET_RX_QUEUE_DEFAULT_SIZE
 #define VIRTIO_NET_TX_QUEUE_MIN_SIZE VIRTIO_NET_TX_QUEUE_DEFAULT_SIZE
 
+/* duplex and speed defines */
+#define DUPLEX_UNKNOWN          0xff
+#define DUPLEX_HALF             0x00
+#define DUPLEX_FULL             0x01
+#define SPEED_UNKNOWN           -1
+
 /*
  * Calculate the number of bytes up to and including the given 'field' of
  * 'container'.

@@ -61,6 +67,8 @@ static VirtIOFeature feature_sizes[] = {
      .end = endof(struct virtio_net_config, max_virtqueue_pairs)},
     {.flags = 1 << VIRTIO_NET_F_MTU,
      .end = endof(struct virtio_net_config, mtu)},
+    {.flags = 1 << VIRTIO_NET_F_SPEED_DUPLEX,
+     .end = endof(struct virtio_net_config, duplex)},
     {}
 };

@@ -88,6 +96,8 @@ static void virtio_net_get_config(VirtIODevice *vdev, uint8_t *config)
     virtio_stw_p(vdev, &netcfg.status, n->status);
     virtio_stw_p(vdev, &netcfg.max_virtqueue_pairs, n->max_queues);
     virtio_stw_p(vdev, &netcfg.mtu, n->net_conf.mtu);
+    virtio_stl_p(vdev, &netcfg.speed, n->net_conf.speed);
+    netcfg.duplex = n->net_conf.duplex;
     memcpy(netcfg.mac, n->mac, ETH_ALEN);
     memcpy(config, &netcfg, n->config_size);
 }

@@ -1941,6 +1951,23 @@ static void virtio_net_device_realize(DeviceState *dev, Error **errp)
         n->host_features |= (0x1 << VIRTIO_NET_F_MTU);
     }
 
+    n->host_features |= (0x1 << VIRTIO_NET_F_SPEED_DUPLEX);
+    if (n->net_conf.duplex_str) {
+        if (strncmp(n->net_conf.duplex_str, "half", 5) == 0) {
+            n->net_conf.duplex = DUPLEX_HALF;
+        } else if (strncmp(n->net_conf.duplex_str, "full", 5) == 0) {
+            n->net_conf.duplex = DUPLEX_FULL;
+        } else {
+            error_setg(errp, "'duplex' must be 'half' or 'full'");
+        }
+    } else {
+        n->net_conf.duplex = DUPLEX_UNKNOWN;
+    }
+    if (n->net_conf.speed < SPEED_UNKNOWN) {
+            error_setg(errp, "'speed' must be between -1 (SPEED_UNKOWN) and "
+                       "INT_MAX");
+    }
+
     virtio_net_set_config_size(n, n->host_features);
     virtio_init(vdev, "virtio-net", VIRTIO_ID_NET, n->config_size);

@@ -2160,6 +2187,8 @@ static Property virtio_net_properties[] = {
     DEFINE_PROP_UINT16("host_mtu", VirtIONet, net_conf.mtu, 0),
     DEFINE_PROP_BOOL("x-mtu-bypass-backend", VirtIONet, mtu_bypass_backend,
                      true),
+    DEFINE_PROP_INT32("speed", VirtIONet, net_conf.speed, SPEED_UNKNOWN),
+    DEFINE_PROP_STRING("duplex", VirtIONet, net_conf.duplex_str),
     DEFINE_PROP_END_OF_LIST(),
 };

include/hw/virtio/virtio-net.h +3 -0

index b81b6a4..af74a94 100644
@@ -38,6 +38,9 @@ typedef struct virtio_net_conf
     uint16_t rx_queue_size;
     uint16_t tx_queue_size;
     uint16_t mtu;
+    int32_t speed;
+    char *duplex_str;
+    uint8_t duplex;
 } virtio_net_conf;
 
 /* Maximum packet size we can receive from tap device: header + 64k */

include/standard-headers/linux/virtio_net.h +4 -0

index 30ff249..0ff1447 100644
@@ -36,6 +36,7 @@
 #define VIRTIO_NET_F_GUEST_CSUM	1	/* Guest handles pkts w/ partial csum */
 #define VIRTIO_NET_F_CTRL_GUEST_OFFLOADS 2 /* Dynamic offload configuration. */
 #define VIRTIO_NET_F_MTU	3	/* Initial MTU advice */
+#define VIRTIO_NET_F_SPEED_DUPLEX 4	/* Host set linkspeed and duplex */
 #define VIRTIO_NET_F_MAC	5	/* Host has given MAC address. */
 #define VIRTIO_NET_F_GUEST_TSO4	7	/* Guest can handle TSOv4 in. */
 #define VIRTIO_NET_F_GUEST_TSO6	8	/* Guest can handle TSOv6 in. */

@@ -76,6 +77,9 @@ struct virtio_net_config {
 	uint16_t max_virtqueue_pairs;
 	/* Default maximum transmit unit advice */
 	uint16_t mtu;
+	/* Host exported linkspeed and duplex */
+	uint32_t speed;
+	uint8_t duplex;
 } QEMU_PACKED;
 
 /*

Signature · 1 line

2.6.1

Yan Vugenfirer

2017-12-18 11:34 UTC #9 lore raw

Re: [Qemu-devel] [PATCH 2/2] qemu: add linkspeed and duplex setting to virtio-net

Quoted 87 lines · Jason Baron via Qemu-devel

Linkspeed and duplex settings can be set as:
'-device virtio-net,speed=10000,duplex=full'

where speed is [-1...INT_MAX], and duplex is ["half"|"full"].

Signed-off-by Jason Baron <[email protected]>

Cc "Michael S. Tsirkin" <[email protected]>

Cc Jason Wang <[email protected]>

---

hw/net/virtio-net.c                         | 29 +++++++++++++++++++++++++++++
include/hw/virtio/virtio-net.h              |  3 +++
include/standard-headers/linux/virtio_net.h |  4 ++++
3 files changed, 36 insertions(+)

hw/net/virtio-net.c +28 -0

index 38674b0..d63e790 100644
@@ -40,6 +40,12 @@
 #define VIRTIO_NET_RX_QUEUE_MIN_SIZE VIRTIO_NET_RX_QUEUE_DEFAULT_SIZE
 #define VIRTIO_NET_TX_QUEUE_MIN_SIZE VIRTIO_NET_TX_QUEUE_DEFAULT_SIZE
 
+/* duplex and speed defines */
+#define DUPLEX_UNKNOWN          0xff
+#define DUPLEX_HALF             0x00
+#define DUPLEX_FULL             0x01
+#define SPEED_UNKNOWN           -1
+
 /*
 * Calculate the number of bytes up to and including the given 'field' of
 * 'container'.

@@ -61,6 +67,8 @@ static VirtIOFeature feature_sizes[] = {
     .end = endof(struct virtio_net_config, max_virtqueue_pairs)},
    {.flags = 1 << VIRTIO_NET_F_MTU,
     .end = endof(struct virtio_net_config, mtu)},
+    {.flags = 1 << VIRTIO_NET_F_SPEED_DUPLEX,
+     .end = endof(struct virtio_net_config, duplex)},
    {}
 };

@@ -88,6 +96,8 @@ static void virtio_net_get_config(VirtIODevice *vdev, uint8_t *config)
    virtio_stw_p(vdev, &netcfg.status, n->status);
    virtio_stw_p(vdev, &netcfg.max_virtqueue_pairs, n->max_queues);
    virtio_stw_p(vdev, &netcfg.mtu, n->net_conf.mtu);
+    virtio_stl_p(vdev, &netcfg.speed, n->net_conf.speed);
+    netcfg.duplex = n->net_conf.duplex;
    memcpy(netcfg.mac, n->mac, ETH_ALEN);
    memcpy(config, &netcfg, n->config_size);
 }

@@ -1941,6 +1951,23 @@ static void virtio_net_device_realize(DeviceState *dev, Error **errp)
        n->host_features |= (0x1 << VIRTIO_NET_F_MTU);
    }
 
+    n->host_features |= (0x1 << VIRTIO_NET_F_SPEED_DUPLEX);
+    if (n->net_conf.duplex_str) {
+        if (strncmp(n->net_conf.duplex_str, "half", 5) == 0) {
+            n->net_conf.duplex = DUPLEX_HALF;
+        } else if (strncmp(n->net_conf.duplex_str, "full", 5) == 0) {
+            n->net_conf.duplex = DUPLEX_FULL;
+        } else {
+            error_setg(errp, "'duplex' must be 'half' or 'full'");
+        }
+    } else {
+        n->net_conf.duplex = DUPLEX_UNKNOWN;
+    }
+    if (n->net_conf.speed < SPEED_UNKNOWN) {
+            error_setg(errp, "'speed' must be between -1 (SPEED_UNKOWN) and "
+                       "INT_MAX");
+    }
+
    virtio_net_set_config_size(n, n->host_features);
    virtio_init(vdev, "virtio-net", VIRTIO_ID_NET, n->config_size);

@@ -2160,6 +2187,8 @@ static Property virtio_net_properties[] = {
    DEFINE_PROP_UINT16("host_mtu", VirtIONet, net_conf.mtu, 0),
    DEFINE_PROP_BOOL("x-mtu-bypass-backend", VirtIONet, mtu_bypass_backend,
                     true),
+    DEFINE_PROP_INT32("speed", VirtIONet, net_conf.speed, SPEED_UNKNOWN),

Quoted 1 line

From Windows guest perspective I prefer to have some reasonable default (10G for example).

Thanks,
Yan.

Quoted 44 lines

+ DEFINE_PROP_STRING("duplex", VirtIONet, net_conf.duplex_str),

    DEFINE_PROP_END_OF_LIST(),
};

include/hw/virtio/virtio-net.h +3 -0

index b81b6a4..af74a94 100644
@@ -38,6 +38,9 @@ typedef struct virtio_net_conf
    uint16_t rx_queue_size;
    uint16_t tx_queue_size;
    uint16_t mtu;
+    int32_t speed;
+    char *duplex_str;
+    uint8_t duplex;
 } virtio_net_conf;
 
 /* Maximum packet size we can receive from tap device: header + 64k */

include/standard-headers/linux/virtio_net.h +4 -0

index 30ff249..0ff1447 100644
@@ -36,6 +36,7 @@
 #define VIRTIO_NET_F_GUEST_CSUM	1	/* Guest handles pkts w/ partial csum */
 #define VIRTIO_NET_F_CTRL_GUEST_OFFLOADS 2 /* Dynamic offload configuration. */
 #define VIRTIO_NET_F_MTU	3	/* Initial MTU advice */
+#define VIRTIO_NET_F_SPEED_DUPLEX 4	/* Host set linkspeed and duplex */
 #define VIRTIO_NET_F_MAC	5	/* Host has given MAC address. */
 #define VIRTIO_NET_F_GUEST_TSO4	7	/* Guest can handle TSOv4 in. */
 #define VIRTIO_NET_F_GUEST_TSO6	8	/* Guest can handle TSOv6 in. */

@@ -76,6 +77,9 @@ struct virtio_net_config {
 	uint16_t max_virtqueue_pairs;
 	/* Default maximum transmit unit advice */
 	uint16_t mtu;
+	/* Host exported linkspeed and duplex */
+	uint32_t speed;
+	uint8_t duplex;
 } QEMU_PACKED;
 
 /*

Signature · 1 line

2.6.1

Jason Baron

author

2017-12-18 16:04 UTC #10 lore raw

Re: [Qemu-devel] [PATCH 2/2] qemu: add linkspeed and duplex setting to virtio-net

Quoted 81 lines · Yan Vugenfirer

Quoted 81 lines · Jason Baron via Qemu-devel

Linkspeed and duplex settings can be set as:
'-device virtio-net,speed=10000,duplex=full'

where speed is [-1...INT_MAX], and duplex is ["half"|"full"].

Signed-off-by Jason Baron <[email protected]>

Cc "Michael S. Tsirkin" <[email protected]>

Cc Jason Wang <[email protected]>

---

hw/net/virtio-net.c                         | 29 +++++++++++++++++++++++++++++
include/hw/virtio/virtio-net.h              |  3 +++
include/standard-headers/linux/virtio_net.h |  4 ++++
3 files changed, 36 insertions(+)

hw/net/virtio-net.c +27 -0

index 38674b0..d63e790 100644
@@ -40,6 +40,12 @@
 #define VIRTIO_NET_RX_QUEUE_MIN_SIZE VIRTIO_NET_RX_QUEUE_DEFAULT_SIZE
 #define VIRTIO_NET_TX_QUEUE_MIN_SIZE VIRTIO_NET_TX_QUEUE_DEFAULT_SIZE
 
+/* duplex and speed defines */
+#define DUPLEX_UNKNOWN          0xff
+#define DUPLEX_HALF             0x00
+#define DUPLEX_FULL             0x01
+#define SPEED_UNKNOWN           -1
+
 /*
 * Calculate the number of bytes up to and including the given 'field' of
 * 'container'.

@@ -61,6 +67,8 @@ static VirtIOFeature feature_sizes[] = {
     .end = endof(struct virtio_net_config, max_virtqueue_pairs)},
    {.flags = 1 << VIRTIO_NET_F_MTU,
     .end = endof(struct virtio_net_config, mtu)},
+    {.flags = 1 << VIRTIO_NET_F_SPEED_DUPLEX,
+     .end = endof(struct virtio_net_config, duplex)},
    {}
 };

@@ -88,6 +96,8 @@ static void virtio_net_get_config(VirtIODevice *vdev, uint8_t *config)
    virtio_stw_p(vdev, &netcfg.status, n->status);
    virtio_stw_p(vdev, &netcfg.max_virtqueue_pairs, n->max_queues);
    virtio_stw_p(vdev, &netcfg.mtu, n->net_conf.mtu);
+    virtio_stl_p(vdev, &netcfg.speed, n->net_conf.speed);
+    netcfg.duplex = n->net_conf.duplex;
    memcpy(netcfg.mac, n->mac, ETH_ALEN);
    memcpy(config, &netcfg, n->config_size);
 }

@@ -1941,6 +1951,23 @@ static void virtio_net_device_realize(DeviceState *dev, Error **errp)
        n->host_features |= (0x1 << VIRTIO_NET_F_MTU);
    }
 
+    n->host_features |= (0x1 << VIRTIO_NET_F_SPEED_DUPLEX);
+    if (n->net_conf.duplex_str) {
+        if (strncmp(n->net_conf.duplex_str, "half", 5) == 0) {
+            n->net_conf.duplex = DUPLEX_HALF;
+        } else if (strncmp(n->net_conf.duplex_str, "full", 5) == 0) {
+            n->net_conf.duplex = DUPLEX_FULL;
+        } else {
+            error_setg(errp, "'duplex' must be 'half' or 'full'");
+        }
+    } else {
+        n->net_conf.duplex = DUPLEX_UNKNOWN;
+    }
+    if (n->net_conf.speed < SPEED_UNKNOWN) {
+            error_setg(errp, "'speed' must be between -1 (SPEED_UNKOWN) and "
+                       "INT_MAX");
+    }
+
    virtio_net_set_config_size(n, n->host_features);
    virtio_init(vdev, "virtio-net", VIRTIO_ID_NET, n->config_size);

Quoted 7 lines

hw/net/virtio-net.c static Property virtio_net_properties[] = { · line 2190
Quoted 5 lines
hw/net/virtio-net.c +1 -0
@@ -2160,6 +2187,8 @@ static Property virtio_net_properties[] = {
    DEFINE_PROP_UINT16("host_mtu", VirtIONet, net_conf.mtu, 0),
    DEFINE_PROP_BOOL("x-mtu-bypass-backend", VirtIONet, mtu_bypass_backend,
                     true),
+    DEFINE_PROP_INT32("speed", VirtIONet, net_conf.speed, SPEED_UNKNOWN),
From Windows guest perspective I prefer to have some reasonable default (10G for example).

hmmm, I didn't want to change/set the default here in case it broke something, but I'm ok setting it to some 'reasonable' value - (10G and duplex?), if the consensus is that that would be safe.

Thanks,

-Jason

Quoted 49 lines

Thanks,
Yan.

Quoted 44 lines

hw/net/virtio-net.c +1 -0

+    DEFINE_PROP_STRING("duplex", VirtIONet, net_conf.duplex_str),
    DEFINE_PROP_END_OF_LIST(),
 };

include/hw/virtio/virtio-net.h +3 -0

index b81b6a4..af74a94 100644
@@ -38,6 +38,9 @@ typedef struct virtio_net_conf
    uint16_t rx_queue_size;
    uint16_t tx_queue_size;
    uint16_t mtu;
+    int32_t speed;
+    char *duplex_str;
+    uint8_t duplex;
 } virtio_net_conf;
 
 /* Maximum packet size we can receive from tap device: header + 64k */

include/standard-headers/linux/virtio_net.h +4 -0

index 30ff249..0ff1447 100644
@@ -36,6 +36,7 @@
 #define VIRTIO_NET_F_GUEST_CSUM	1	/* Guest handles pkts w/ partial csum */
 #define VIRTIO_NET_F_CTRL_GUEST_OFFLOADS 2 /* Dynamic offload configuration. */
 #define VIRTIO_NET_F_MTU	3	/* Initial MTU advice */
+#define VIRTIO_NET_F_SPEED_DUPLEX 4	/* Host set linkspeed and duplex */
 #define VIRTIO_NET_F_MAC	5	/* Host has given MAC address. */
 #define VIRTIO_NET_F_GUEST_TSO4	7	/* Guest can handle TSOv4 in. */
 #define VIRTIO_NET_F_GUEST_TSO6	8	/* Guest can handle TSOv6 in. */

@@ -76,6 +77,9 @@ struct virtio_net_config {
 	uint16_t max_virtqueue_pairs;
 	/* Default maximum transmit unit advice */
 	uint16_t mtu;
+	/* Host exported linkspeed and duplex */
+	uint32_t speed;
+	uint8_t duplex;
 } QEMU_PACKED;
 
 /*

Signature · 1 line

2.6.1

Yan Vugenfirer

2017-12-19 09:20 UTC #11 lore raw

Re: [Qemu-devel] [PATCH 2/2] qemu: add linkspeed and duplex setting to virtio-net

Quoted 96 lines · Jason Baron via Qemu-devel

Quoted 81 lines · Yan Vugenfirer

Quoted 81 lines · Jason Baron via Qemu-devel

Linkspeed and duplex settings can be set as:
'-device virtio-net,speed=10000,duplex=full'

where speed is [-1...INT_MAX], and duplex is ["half"|"full"].

Signed-off-by Jason Baron <[email protected]>

Cc "Michael S. Tsirkin" <[email protected]>

Cc Jason Wang <[email protected]>

---

hw/net/virtio-net.c                         | 29 +++++++++++++++++++++++++++++
include/hw/virtio/virtio-net.h              |  3 +++
include/standard-headers/linux/virtio_net.h |  4 ++++
3 files changed, 36 insertions(+)

hw/net/virtio-net.c +27 -0

index 38674b0..d63e790 100644
@@ -40,6 +40,12 @@
 #define VIRTIO_NET_RX_QUEUE_MIN_SIZE VIRTIO_NET_RX_QUEUE_DEFAULT_SIZE
 #define VIRTIO_NET_TX_QUEUE_MIN_SIZE VIRTIO_NET_TX_QUEUE_DEFAULT_SIZE
 
+/* duplex and speed defines */
+#define DUPLEX_UNKNOWN          0xff
+#define DUPLEX_HALF             0x00
+#define DUPLEX_FULL             0x01
+#define SPEED_UNKNOWN           -1
+
 /*
 * Calculate the number of bytes up to and including the given 'field' of
 * 'container'.

@@ -61,6 +67,8 @@ static VirtIOFeature feature_sizes[] = {
    .end = endof(struct virtio_net_config, max_virtqueue_pairs)},
   {.flags = 1 << VIRTIO_NET_F_MTU,
    .end = endof(struct virtio_net_config, mtu)},
+    {.flags = 1 << VIRTIO_NET_F_SPEED_DUPLEX,
+     .end = endof(struct virtio_net_config, duplex)},
   {}
 };

@@ -88,6 +96,8 @@ static void virtio_net_get_config(VirtIODevice *vdev, uint8_t *config)
   virtio_stw_p(vdev, &netcfg.status, n->status);
   virtio_stw_p(vdev, &netcfg.max_virtqueue_pairs, n->max_queues);
   virtio_stw_p(vdev, &netcfg.mtu, n->net_conf.mtu);
+    virtio_stl_p(vdev, &netcfg.speed, n->net_conf.speed);
+    netcfg.duplex = n->net_conf.duplex;
   memcpy(netcfg.mac, n->mac, ETH_ALEN);
   memcpy(config, &netcfg, n->config_size);
 }

@@ -1941,6 +1951,23 @@ static void virtio_net_device_realize(DeviceState *dev, Error **errp)
       n->host_features |= (0x1 << VIRTIO_NET_F_MTU);
   }
 
+    n->host_features |= (0x1 << VIRTIO_NET_F_SPEED_DUPLEX);
+    if (n->net_conf.duplex_str) {
+        if (strncmp(n->net_conf.duplex_str, "half", 5) == 0) {
+            n->net_conf.duplex = DUPLEX_HALF;
+        } else if (strncmp(n->net_conf.duplex_str, "full", 5) == 0) {
+            n->net_conf.duplex = DUPLEX_FULL;
+        } else {
+            error_setg(errp, "'duplex' must be 'half' or 'full'");
+        }
+    } else {
+        n->net_conf.duplex = DUPLEX_UNKNOWN;
+    }
+    if (n->net_conf.speed < SPEED_UNKNOWN) {
+            error_setg(errp, "'speed' must be between -1 (SPEED_UNKOWN) and "
+                       "INT_MAX");
+    }
+
   virtio_net_set_config_size(n, n->host_features);
   virtio_init(vdev, "virtio-net", VIRTIO_ID_NET, n->config_size);

Quoted 7 lines

hw/net/virtio-net.c static Property virtio_net_properties[] = { · line 2190
Quoted 5 lines
hw/net/virtio-net.c +1 -0
@@ -2160,6 +2187,8 @@ static Property virtio_net_properties[] = {
   DEFINE_PROP_UINT16("host_mtu", VirtIONet, net_conf.mtu, 0),
   DEFINE_PROP_BOOL("x-mtu-bypass-backend", VirtIONet, mtu_bypass_backend,
                    true),
+    DEFINE_PROP_INT32("speed", VirtIONet, net_conf.speed, SPEED_UNKNOWN),
From Windows guest perspective I prefer to have some reasonable default (10G for example).

Quoted 3 lines

hmmm, I didn't want to change/set the default here in case it broke something, but I'm ok setting it to some 'reasonable' value - (10G and duplex?), if the consensus is that that would be safe.

OK from my side.
Thanks.

Quoted 51 lines

Thanks,

-Jason

Quoted 46 lines

Thanks,
Yan.

Quoted 42 lines

hw/net/virtio-net.c +1 -0

+    DEFINE_PROP_STRING("duplex", VirtIONet, net_conf.duplex_str),
   DEFINE_PROP_END_OF_LIST(),
 };

include/hw/virtio/virtio-net.h +3 -0

index b81b6a4..af74a94 100644
@@ -38,6 +38,9 @@ typedef struct virtio_net_conf
   uint16_t rx_queue_size;
   uint16_t tx_queue_size;
   uint16_t mtu;
+    int32_t speed;
+    char *duplex_str;
+    uint8_t duplex;
 } virtio_net_conf;
 
 /* Maximum packet size we can receive from tap device: header + 64k */

include/standard-headers/linux/virtio_net.h +4 -0

index 30ff249..0ff1447 100644
@@ -36,6 +36,7 @@
 #define VIRTIO_NET_F_GUEST_CSUM	1	/* Guest handles pkts w/ partial csum */
 #define VIRTIO_NET_F_CTRL_GUEST_OFFLOADS 2 /* Dynamic offload configuration. */
 #define VIRTIO_NET_F_MTU	3	/* Initial MTU advice */
+#define VIRTIO_NET_F_SPEED_DUPLEX 4	/* Host set linkspeed and duplex */
 #define VIRTIO_NET_F_MAC	5	/* Host has given MAC address. */
 #define VIRTIO_NET_F_GUEST_TSO4	7	/* Guest can handle TSOv4 in. */
 #define VIRTIO_NET_F_GUEST_TSO6	8	/* Guest can handle TSOv6 in. */

@@ -76,6 +77,9 @@ struct virtio_net_config {
 	uint16_t max_virtqueue_pairs;
 	/* Default maximum transmit unit advice */
 	uint16_t mtu;
+	/* Host exported linkspeed and duplex */
+	uint32_t speed;
+	uint8_t duplex;
 } QEMU_PACKED;
 
 /*

Signature · 1 line

2.6.1

Jason Baron

author

2017-12-19 16:52 UTC #12 lore raw

Re: [Qemu-devel] [PATCH 2/2] qemu: add linkspeed and duplex setting to virtio-net

Quoted 108 lines · Yan Vugenfirer

Quoted 108 lines
On 18 Dec 2017, at 18:04, Jason Baron via Qemu-devel <[email protected] <mailto:[email protected]>> wrote:
Quoted 99 lines · Yan Vugenfirer
Quoted 90 lines
On 14 Dec 2017, at 21:33, Jason Baron via Qemu-devel <[email protected] <mailto:[email protected]>> wrote:
Although they can be currently set in linux via 'ethtool -s', this requires
guest changes, and thus it would be nice to extend this functionality such
that it can be configured automatically from the host (as other network do).
Linkspeed and duplex settings can be set as:
'-device virtio-net,speed=10000,duplex=full'
where speed is [-1...INT_MAX], and duplex is ["half"|"full"].
Signed-off-by Jason Baron <[email protected]
<mailto:[email protected]>>
Cc: "Michael S. Tsirkin" <[email protected] <mailto:[email protected]>>
Cc: Jason Wang <[email protected] <mailto:[email protected]>>
---
hw/net/virtio-net.c | 29
+++++++++++++++++++++++++++++
include/hw/virtio/virtio-net.h | 3 +++ include/standard-headers/linux/virtio_net.h | 4 ++++
3 files changed, 36 insertions(+)
hw/net/virtio-net.c +10 -0
index 38674b0..d63e790 100644
@@ -40,6 +40,12 @@
 #define VIRTIO_NET_RX_QUEUE_MIN_SIZE VIRTIO_NET_RX_QUEUE_DEFAULT_SIZE
 #define VIRTIO_NET_TX_QUEUE_MIN_SIZE VIRTIO_NET_TX_QUEUE_DEFAULT_SIZE
 
+/* duplex and speed defines */
+#define DUPLEX_UNKNOWN          0xff
+#define DUPLEX_HALF             0x00
+#define DUPLEX_FULL             0x01
+#define SPEED_UNKNOWN           -1
+
 /*
 * Calculate the number of bytes up to and including the given 'field' of
 * 'container'.
@@ -61,6 +67,8 @@ static VirtIOFeature feature_sizes[] = {
     .end = endof(struct virtio_net_config, max_virtqueue_pairs)},
    {.flags = 1 << VIRTIO_NET_F_MTU,
     .end = endof(struct virtio_net_config, mtu)},
+    {.flags = 1 << VIRTIO_NET_F_SPEED_DUPLEX,
+     .end = endof(struct virtio_net_config, duplex)},
    {}
 };
 
@@ -88,6 +96,8 @@ static void virtio_net_get_config(VirtIODevice
 *vdev, uint8_t *config)
    virtio_stw_p(vdev, &netcfg.status, n->status);
    virtio_stw_p(vdev, &netcfg.max_virtqueue_pairs, n->max_queues);
    virtio_stw_p(vdev, &netcfg.mtu, n->net_conf.mtu);
+    virtio_stl_p(vdev, &netcfg.speed, n->net_conf.speed);
+    netcfg.duplex = n->net_conf.duplex;
    memcpy(netcfg.mac, n->mac, ETH_ALEN);
    memcpy(config, &netcfg, n->config_size);
}
hw/net/virtio-net.c +14 -0
@@ -1941,6 +1951,23 @@ static void
 virtio_net_device_realize(DeviceState *dev, Error **errp)
        n->host_features |= (0x1 << VIRTIO_NET_F_MTU);
    }
 
+    n->host_features |= (0x1 << VIRTIO_NET_F_SPEED_DUPLEX);
+    if (n->net_conf.duplex_str) {
+        if (strncmp(n->net_conf.duplex_str, "half", 5) == 0) {
+            n->net_conf.duplex = DUPLEX_HALF;
+        } else if (strncmp(n->net_conf.duplex_str, "full", 5) == 0) {
+            n->net_conf.duplex = DUPLEX_FULL;
+        } else {
+            error_setg(errp, "'duplex' must be 'half' or 'full'");
+        }
+    } else {
+        n->net_conf.duplex = DUPLEX_UNKNOWN;
+    }
+    if (n->net_conf.speed < SPEED_UNKNOWN) {
+            error_setg(errp, "'speed' must be between -1
(SPEED_UNKOWN) and "
+                       "INT_MAX");
+    }
+
   virtio_net_set_config_size(n, n->host_features); virtio_init(vdev, "virtio-net", VIRTIO_ID_NET, n->config_size);
hw/net/virtio-net.c static Property virtio_net_properties[] = { · line 2192
Quoted 6 lines
hw/net/virtio-net.c +1 -0
@@ -2160,6 +2187,8 @@ static Property virtio_net_properties[] = {
    DEFINE_PROP_UINT16("host_mtu", VirtIONet, net_conf.mtu, 0),
    DEFINE_PROP_BOOL("x-mtu-bypass-backend", VirtIONet,
 mtu_bypass_backend,
                     true),
+    DEFINE_PROP_INT32("speed", VirtIONet, net_conf.speed,
 SPEED_UNKNOWN),
From Windows guest perspective I prefer to have some reasonable default (10G for example).

Quoted 6 lines

Quoted 3 lines
hmmm, I didn't want to change/set the default here in case it broke something, but I'm ok setting it to some 'reasonable' value - (10G and duplex?), if the consensus is that that would be safe.
OK from my side.
Thanks.

I presume your speaking for windows - i'm wondering if under linux the virtio device suddenly start showing up as duplex, 10Gbps, will that break anything? I can speak for the use-cases we have here and that would certainly be fine for us, but i'm really not sure if that is ok more generally.

Thanks,

-Jason

Quoted 56 lines

Quoted 55 lines
Thanks,
-Jason
Quoted 49 lines
Thanks,
Yan.
Quoted 45 lines
hw/net/virtio-net.c +1 -0
+    DEFINE_PROP_STRING("duplex", VirtIONet, net_conf.duplex_str),
    DEFINE_PROP_END_OF_LIST(),
};
include/hw/virtio/virtio-net.h +3 -0
index b81b6a4..af74a94 100644
@@ -38,6 +38,9 @@ typedef struct virtio_net_conf
    uint16_t rx_queue_size;
    uint16_t tx_queue_size;
    uint16_t mtu;
+    int32_t speed;
+    char *duplex_str;
+    uint8_t duplex;
 } virtio_net_conf;
 
 /* Maximum packet size we can receive from tap device: header + 64k */
include/standard-headers/linux/virtio_net.h +0 -0
index 30ff249..0ff1447 100644
@@ -36,6 +36,7 @@
 #define VIRTIO_NET_F_GUEST_CSUM1/* Guest handles pkts w/ partial csum */
 #define VIRTIO_NET_F_CTRL_GUEST_OFFLOADS 2 /* Dynamic offload
configuration. */
#define VIRTIO_NET_F_MTU3/* Initial MTU advice */
+#define VIRTIO_NET_F_SPEED_DUPLEX 4/* Host set linkspeed and duplex */ #define VIRTIO_NET_F_MAC5/* Host has given MAC address. */ #define VIRTIO_NET_F_GUEST_TSO47/* Guest can handle TSOv4 in. */ #define VIRTIO_NET_F_GUEST_TSO68/* Guest can handle TSOv6 in. */
include/standard-headers/linux/virtio_net.h +3 -0
@@ -76,6 +77,9 @@ struct virtio_net_config {
 uint16_t max_virtqueue_pairs;
 /* Default maximum transmit unit advice */
 uint16_t mtu;
+/* Host exported linkspeed and duplex */
+uint32_t speed;
+uint8_t duplex;
 } QEMU_PACKED;
 
 /*
--
2.6.1

Michael S. Tsirkin

maintainer

2017-12-20 14:31 UTC #13 lore raw

Re: [Qemu-devel] [PATCH 2/2] qemu: add linkspeed and duplex setting to virtio-net

Quoted 124 lines · Jason Baron

Quoted 108 lines · Yan Vugenfirer
Quoted 108 lines
On 18 Dec 2017, at 18:04, Jason Baron via Qemu-devel <[email protected] <mailto:[email protected]>> wrote:
Quoted 99 lines · Yan Vugenfirer
Quoted 90 lines
On 14 Dec 2017, at 21:33, Jason Baron via Qemu-devel <[email protected] <mailto:[email protected]>> wrote:
Although they can be currently set in linux via 'ethtool -s', this requires
guest changes, and thus it would be nice to extend this functionality such
that it can be configured automatically from the host (as other network do).
Linkspeed and duplex settings can be set as:
'-device virtio-net,speed=10000,duplex=full'
where speed is [-1...INT_MAX], and duplex is ["half"|"full"].
Signed-off-by Jason Baron <[email protected]
<mailto:[email protected]>>
Cc: "Michael S. Tsirkin" <[email protected] <mailto:[email protected]>>
Cc: Jason Wang <[email protected] <mailto:[email protected]>>
---
hw/net/virtio-net.c | 29
+++++++++++++++++++++++++++++
include/hw/virtio/virtio-net.h | 3 +++ include/standard-headers/linux/virtio_net.h | 4 ++++
3 files changed, 36 insertions(+)
hw/net/virtio-net.c +10 -0
index 38674b0..d63e790 100644
@@ -40,6 +40,12 @@
 #define VIRTIO_NET_RX_QUEUE_MIN_SIZE VIRTIO_NET_RX_QUEUE_DEFAULT_SIZE
 #define VIRTIO_NET_TX_QUEUE_MIN_SIZE VIRTIO_NET_TX_QUEUE_DEFAULT_SIZE
 
+/* duplex and speed defines */
+#define DUPLEX_UNKNOWN          0xff
+#define DUPLEX_HALF             0x00
+#define DUPLEX_FULL             0x01
+#define SPEED_UNKNOWN           -1
+
 /*
 * Calculate the number of bytes up to and including the given 'field' of
 * 'container'.
@@ -61,6 +67,8 @@ static VirtIOFeature feature_sizes[] = {
     .end = endof(struct virtio_net_config, max_virtqueue_pairs)},
    {.flags = 1 << VIRTIO_NET_F_MTU,
     .end = endof(struct virtio_net_config, mtu)},
+    {.flags = 1 << VIRTIO_NET_F_SPEED_DUPLEX,
+     .end = endof(struct virtio_net_config, duplex)},
    {}
 };
 
@@ -88,6 +96,8 @@ static void virtio_net_get_config(VirtIODevice
 *vdev, uint8_t *config)
    virtio_stw_p(vdev, &netcfg.status, n->status);
    virtio_stw_p(vdev, &netcfg.max_virtqueue_pairs, n->max_queues);
    virtio_stw_p(vdev, &netcfg.mtu, n->net_conf.mtu);
+    virtio_stl_p(vdev, &netcfg.speed, n->net_conf.speed);
+    netcfg.duplex = n->net_conf.duplex;
    memcpy(netcfg.mac, n->mac, ETH_ALEN);
    memcpy(config, &netcfg, n->config_size);
}
hw/net/virtio-net.c +14 -0
@@ -1941,6 +1951,23 @@ static void
 virtio_net_device_realize(DeviceState *dev, Error **errp)
        n->host_features |= (0x1 << VIRTIO_NET_F_MTU);
    }
 
+    n->host_features |= (0x1 << VIRTIO_NET_F_SPEED_DUPLEX);
+    if (n->net_conf.duplex_str) {
+        if (strncmp(n->net_conf.duplex_str, "half", 5) == 0) {
+            n->net_conf.duplex = DUPLEX_HALF;
+        } else if (strncmp(n->net_conf.duplex_str, "full", 5) == 0) {
+            n->net_conf.duplex = DUPLEX_FULL;
+        } else {
+            error_setg(errp, "'duplex' must be 'half' or 'full'");
+        }
+    } else {
+        n->net_conf.duplex = DUPLEX_UNKNOWN;
+    }
+    if (n->net_conf.speed < SPEED_UNKNOWN) {
+            error_setg(errp, "'speed' must be between -1
(SPEED_UNKOWN) and "
+                       "INT_MAX");
+    }
+
   virtio_net_set_config_size(n, n->host_features); virtio_init(vdev, "virtio-net", VIRTIO_ID_NET, n->config_size);
hw/net/virtio-net.c static Property virtio_net_properties[] = { · line 2192
Quoted 6 lines
hw/net/virtio-net.c +1 -0
@@ -2160,6 +2187,8 @@ static Property virtio_net_properties[] = {
    DEFINE_PROP_UINT16("host_mtu", VirtIONet, net_conf.mtu, 0),
    DEFINE_PROP_BOOL("x-mtu-bypass-backend", VirtIONet,
 mtu_bypass_backend,
                     true),
+    DEFINE_PROP_INT32("speed", VirtIONet, net_conf.speed,
 SPEED_UNKNOWN),
From Windows guest perspective I prefer to have some reasonable default (10G for example).
Quoted 6 lines
Quoted 3 lines
hmmm, I didn't want to change/set the default here in case it broke something, but I'm ok setting it to some 'reasonable' value - (10G and duplex?), if the consensus is that that would be safe.
OK from my side.
Thanks.
I presume your speaking for windows - i'm wondering if under linux the virtio device suddenly start showing up as duplex, 10Gbps, will that break anything? I can speak for the use-cases we have here and that would certainly be fine for us, but i'm really not sure if that is ok more generally.

Quoted 3 lines

Thanks,
-Jason

How about not enabling the flag unless the user specified the speed? This way we also do not need an "unknown" value.

Quoted 56 lines

Quoted 56 lines
Quoted 55 lines
Thanks,
-Jason
Quoted 49 lines
Thanks,
Yan.
Quoted 45 lines
hw/net/virtio-net.c +1 -0
+    DEFINE_PROP_STRING("duplex", VirtIONet, net_conf.duplex_str),
    DEFINE_PROP_END_OF_LIST(),
};
include/hw/virtio/virtio-net.h +3 -0
index b81b6a4..af74a94 100644
@@ -38,6 +38,9 @@ typedef struct virtio_net_conf
    uint16_t rx_queue_size;
    uint16_t tx_queue_size;
    uint16_t mtu;
+    int32_t speed;
+    char *duplex_str;
+    uint8_t duplex;
 } virtio_net_conf;
 
 /* Maximum packet size we can receive from tap device: header + 64k */
include/standard-headers/linux/virtio_net.h +0 -0
index 30ff249..0ff1447 100644
@@ -36,6 +36,7 @@
 #define VIRTIO_NET_F_GUEST_CSUM1/* Guest handles pkts w/ partial csum */
 #define VIRTIO_NET_F_CTRL_GUEST_OFFLOADS 2 /* Dynamic offload
configuration. */
#define VIRTIO_NET_F_MTU3/* Initial MTU advice */
+#define VIRTIO_NET_F_SPEED_DUPLEX 4/* Host set linkspeed and duplex */ #define VIRTIO_NET_F_MAC5/* Host has given MAC address. */ #define VIRTIO_NET_F_GUEST_TSO47/* Guest can handle TSOv4 in. */ #define VIRTIO_NET_F_GUEST_TSO68/* Guest can handle TSOv6 in. */
include/standard-headers/linux/virtio_net.h +3 -0
@@ -76,6 +77,9 @@ struct virtio_net_config {
 uint16_t max_virtqueue_pairs;
 /* Default maximum transmit unit advice */
 uint16_t mtu;
+/* Host exported linkspeed and duplex */
+uint32_t speed;
+uint8_t duplex;
 } QEMU_PACKED;
 
 /*
--
2.6.1

Yan Vugenfirer

2017-12-20 14:32 UTC #14 lore raw

Re: [Qemu-devel] [PATCH 2/2] qemu: add linkspeed and duplex setting to virtio-net

Quoted 128 lines · Michael S. Tsirkin

Quoted 124 lines · Jason Baron
Quoted 108 lines · Yan Vugenfirer
Quoted 108 lines
On 18 Dec 2017, at 18:04, Jason Baron via Qemu-devel <[email protected] <mailto:[email protected]>> wrote:
Quoted 99 lines · Yan Vugenfirer
Quoted 90 lines
On 14 Dec 2017, at 21:33, Jason Baron via Qemu-devel <[email protected] <mailto:[email protected]>> wrote:
Although they can be currently set in linux via 'ethtool -s', this requires
guest changes, and thus it would be nice to extend this functionality such
that it can be configured automatically from the host (as other network do).
Linkspeed and duplex settings can be set as:
'-device virtio-net,speed=10000,duplex=full'
where speed is [-1...INT_MAX], and duplex is ["half"|"full"].
Signed-off-by Jason Baron <[email protected]
<mailto:[email protected]>>
Cc: "Michael S. Tsirkin" <[email protected] <mailto:[email protected]>>
Cc: Jason Wang <[email protected] <mailto:[email protected]>>
---
hw/net/virtio-net.c | 29
+++++++++++++++++++++++++++++
include/hw/virtio/virtio-net.h              |  3 +++
include/standard-headers/linux/virtio_net.h |  4 ++++
3 files changed, 36 insertions(+)
hw/net/virtio-net.c +10 -0
index 38674b0..d63e790 100644
@@ -40,6 +40,12 @@
 #define VIRTIO_NET_RX_QUEUE_MIN_SIZE VIRTIO_NET_RX_QUEUE_DEFAULT_SIZE
 #define VIRTIO_NET_TX_QUEUE_MIN_SIZE VIRTIO_NET_TX_QUEUE_DEFAULT_SIZE
 
+/* duplex and speed defines */
+#define DUPLEX_UNKNOWN          0xff
+#define DUPLEX_HALF             0x00
+#define DUPLEX_FULL             0x01
+#define SPEED_UNKNOWN           -1
+
 /*
 * Calculate the number of bytes up to and including the given 'field' of
 * 'container'.
@@ -61,6 +67,8 @@ static VirtIOFeature feature_sizes[] = {
    .end = endof(struct virtio_net_config, max_virtqueue_pairs)},
   {.flags = 1 << VIRTIO_NET_F_MTU,
    .end = endof(struct virtio_net_config, mtu)},
+    {.flags = 1 << VIRTIO_NET_F_SPEED_DUPLEX,
+     .end = endof(struct virtio_net_config, duplex)},
   {}
 };
 
@@ -88,6 +96,8 @@ static void virtio_net_get_config(VirtIODevice
 *vdev, uint8_t *config)
   virtio_stw_p(vdev, &netcfg.status, n->status);
   virtio_stw_p(vdev, &netcfg.max_virtqueue_pairs, n->max_queues);
   virtio_stw_p(vdev, &netcfg.mtu, n->net_conf.mtu);
+    virtio_stl_p(vdev, &netcfg.speed, n->net_conf.speed);
+    netcfg.duplex = n->net_conf.duplex;
   memcpy(netcfg.mac, n->mac, ETH_ALEN);
   memcpy(config, &netcfg, n->config_size);
}
hw/net/virtio-net.c +14 -0
@@ -1941,6 +1951,23 @@ static void
 virtio_net_device_realize(DeviceState *dev, Error **errp)
       n->host_features |= (0x1 << VIRTIO_NET_F_MTU);
   }
 
+    n->host_features |= (0x1 << VIRTIO_NET_F_SPEED_DUPLEX);
+    if (n->net_conf.duplex_str) {
+        if (strncmp(n->net_conf.duplex_str, "half", 5) == 0) {
+            n->net_conf.duplex = DUPLEX_HALF;
+        } else if (strncmp(n->net_conf.duplex_str, "full", 5) == 0) {
+            n->net_conf.duplex = DUPLEX_FULL;
+        } else {
+            error_setg(errp, "'duplex' must be 'half' or 'full'");
+        }
+    } else {
+        n->net_conf.duplex = DUPLEX_UNKNOWN;
+    }
+    if (n->net_conf.speed < SPEED_UNKNOWN) {
+            error_setg(errp, "'speed' must be between -1
(SPEED_UNKOWN) and "
hw/net/virtio-net.c +3 -0
+                       "INT_MAX");
+    }
+
   virtio_net_set_config_size(n, n->host_features);
   virtio_init(vdev, "virtio-net", VIRTIO_ID_NET, n->config_size);
hw/net/virtio-net.c static Property virtio_net_properties[] = { · line 2192
Quoted 6 lines
hw/net/virtio-net.c +1 -0
@@ -2160,6 +2187,8 @@ static Property virtio_net_properties[] = {
   DEFINE_PROP_UINT16("host_mtu", VirtIONet, net_conf.mtu, 0),
   DEFINE_PROP_BOOL("x-mtu-bypass-backend", VirtIONet,
 mtu_bypass_backend,
                    true),
+    DEFINE_PROP_INT32("speed", VirtIONet, net_conf.speed,
 SPEED_UNKNOWN),
From Windows guest perspective I prefer to have some reasonable default (10G for example).
Quoted 6 lines
Quoted 3 lines
hmmm, I didn't want to change/set the default here in case it broke something, but I'm ok setting it to some 'reasonable' value - (10G and duplex?), if the consensus is that that would be safe.
OK from my side.
Thanks.
I presume your speaking for windows - i'm wondering if under linux the virtio device suddenly start showing up as duplex, 10Gbps, will that break anything? I can speak for the use-cases we have here and that would certainly be fine for us, but i'm really not sure if that is ok more generally.

Quoted 6 lines

Quoted 3 lines
Thanks,
-Jason
How about not enabling the flag unless the user specified the speed? This way we also do not need an "unknown" value.

This is even better.

Quoted 56 lines

Quoted 56 lines
Quoted 56 lines
Quoted 56 lines
Thanks,
-Jason
Quoted 49 lines
Thanks,
Yan.
Quoted 45 lines
hw/net/virtio-net.c +1 -0
+    DEFINE_PROP_STRING("duplex", VirtIONet, net_conf.duplex_str),
   DEFINE_PROP_END_OF_LIST(),
};
include/hw/virtio/virtio-net.h +3 -0
index b81b6a4..af74a94 100644
@@ -38,6 +38,9 @@ typedef struct virtio_net_conf
   uint16_t rx_queue_size;
   uint16_t tx_queue_size;
   uint16_t mtu;
+    int32_t speed;
+    char *duplex_str;
+    uint8_t duplex;
 } virtio_net_conf;
 
 /* Maximum packet size we can receive from tap device: header + 64k */
include/standard-headers/linux/virtio_net.h +0 -0
index 30ff249..0ff1447 100644
@@ -36,6 +36,7 @@
 #define VIRTIO_NET_F_GUEST_CSUM1/* Guest handles pkts w/ partial csum */
 #define VIRTIO_NET_F_CTRL_GUEST_OFFLOADS 2 /* Dynamic offload
configuration. */
#define VIRTIO_NET_F_MTU3/* Initial MTU advice */
+#define VIRTIO_NET_F_SPEED_DUPLEX 4/* Host set linkspeed and duplex */ #define VIRTIO_NET_F_MAC5/* Host has given MAC address. */ #define VIRTIO_NET_F_GUEST_TSO47/* Guest can handle TSOv4 in. */ #define VIRTIO_NET_F_GUEST_TSO68/* Guest can handle TSOv6 in. */
include/standard-headers/linux/virtio_net.h +3 -0
@@ -76,6 +77,9 @@ struct virtio_net_config {
 uint16_t max_virtqueue_pairs;
 /* Default maximum transmit unit advice */
 uint16_t mtu;
+/* Host exported linkspeed and duplex */
+uint32_t speed;
+uint8_t duplex;
 } QEMU_PACKED;
 
 /*
Signature · 1 line
2.6.1

Yan Vugenfirer

2017-12-20 14:33 UTC #15 lore raw

Re: [Qemu-devel] [PATCH 2/2] qemu: add linkspeed and duplex setting to virtio-net

In reply to Michael S. Tsirkin

Quoted 128 lines · Michael S. Tsirkin

Quoted 124 lines · Jason Baron
Quoted 108 lines · Yan Vugenfirer
Quoted 108 lines
On 18 Dec 2017, at 18:04, Jason Baron via Qemu-devel <[email protected] <mailto:[email protected]>> wrote:
Quoted 99 lines · Yan Vugenfirer
Quoted 90 lines
On 14 Dec 2017, at 21:33, Jason Baron via Qemu-devel <[email protected] <mailto:[email protected]>> wrote:
Although they can be currently set in linux via 'ethtool -s', this requires
guest changes, and thus it would be nice to extend this functionality such
that it can be configured automatically from the host (as other network do).
Linkspeed and duplex settings can be set as:
'-device virtio-net,speed=10000,duplex=full'
where speed is [-1...INT_MAX], and duplex is ["half"|"full"].
Signed-off-by Jason Baron <[email protected]
<mailto:[email protected]>>
Cc: "Michael S. Tsirkin" <[email protected] <mailto:[email protected]>>
Cc: Jason Wang <[email protected] <mailto:[email protected]>>
---
hw/net/virtio-net.c | 29
+++++++++++++++++++++++++++++
include/hw/virtio/virtio-net.h              |  3 +++
include/standard-headers/linux/virtio_net.h |  4 ++++
3 files changed, 36 insertions(+)
hw/net/virtio-net.c +10 -0
index 38674b0..d63e790 100644
@@ -40,6 +40,12 @@
 #define VIRTIO_NET_RX_QUEUE_MIN_SIZE VIRTIO_NET_RX_QUEUE_DEFAULT_SIZE
 #define VIRTIO_NET_TX_QUEUE_MIN_SIZE VIRTIO_NET_TX_QUEUE_DEFAULT_SIZE
 
+/* duplex and speed defines */
+#define DUPLEX_UNKNOWN          0xff
+#define DUPLEX_HALF             0x00
+#define DUPLEX_FULL             0x01
+#define SPEED_UNKNOWN           -1
+
 /*
 * Calculate the number of bytes up to and including the given 'field' of
 * 'container'.
@@ -61,6 +67,8 @@ static VirtIOFeature feature_sizes[] = {
    .end = endof(struct virtio_net_config, max_virtqueue_pairs)},
   {.flags = 1 << VIRTIO_NET_F_MTU,
    .end = endof(struct virtio_net_config, mtu)},
+    {.flags = 1 << VIRTIO_NET_F_SPEED_DUPLEX,
+     .end = endof(struct virtio_net_config, duplex)},
   {}
 };
 
@@ -88,6 +96,8 @@ static void virtio_net_get_config(VirtIODevice
 *vdev, uint8_t *config)
   virtio_stw_p(vdev, &netcfg.status, n->status);
   virtio_stw_p(vdev, &netcfg.max_virtqueue_pairs, n->max_queues);
   virtio_stw_p(vdev, &netcfg.mtu, n->net_conf.mtu);
+    virtio_stl_p(vdev, &netcfg.speed, n->net_conf.speed);
+    netcfg.duplex = n->net_conf.duplex;
   memcpy(netcfg.mac, n->mac, ETH_ALEN);
   memcpy(config, &netcfg, n->config_size);
}
hw/net/virtio-net.c +14 -0
@@ -1941,6 +1951,23 @@ static void
 virtio_net_device_realize(DeviceState *dev, Error **errp)
       n->host_features |= (0x1 << VIRTIO_NET_F_MTU);
   }
 
+    n->host_features |= (0x1 << VIRTIO_NET_F_SPEED_DUPLEX);
+    if (n->net_conf.duplex_str) {
+        if (strncmp(n->net_conf.duplex_str, "half", 5) == 0) {
+            n->net_conf.duplex = DUPLEX_HALF;
+        } else if (strncmp(n->net_conf.duplex_str, "full", 5) == 0) {
+            n->net_conf.duplex = DUPLEX_FULL;
+        } else {
+            error_setg(errp, "'duplex' must be 'half' or 'full'");
+        }
+    } else {
+        n->net_conf.duplex = DUPLEX_UNKNOWN;
+    }
+    if (n->net_conf.speed < SPEED_UNKNOWN) {
+            error_setg(errp, "'speed' must be between -1
(SPEED_UNKOWN) and "
hw/net/virtio-net.c +3 -0
+                       "INT_MAX");
+    }
+
   virtio_net_set_config_size(n, n->host_features);
   virtio_init(vdev, "virtio-net", VIRTIO_ID_NET, n->config_size);
hw/net/virtio-net.c static Property virtio_net_properties[] = { · line 2192
Quoted 6 lines
hw/net/virtio-net.c +1 -0
@@ -2160,6 +2187,8 @@ static Property virtio_net_properties[] = {
   DEFINE_PROP_UINT16("host_mtu", VirtIONet, net_conf.mtu, 0),
   DEFINE_PROP_BOOL("x-mtu-bypass-backend", VirtIONet,
 mtu_bypass_backend,
                    true),
+    DEFINE_PROP_INT32("speed", VirtIONet, net_conf.speed,
 SPEED_UNKNOWN),
From Windows guest perspective I prefer to have some reasonable default (10G for example).
Quoted 6 lines
Quoted 3 lines
hmmm, I didn't want to change/set the default here in case it broke something, but I'm ok setting it to some 'reasonable' value - (10G and duplex?), if the consensus is that that would be safe.
OK from my side.
Thanks.
I presume your speaking for windows - i'm wondering if under linux the virtio device suddenly start showing up as duplex, 10Gbps, will that break anything? I can speak for the use-cases we have here and that would certainly be fine for us, but i'm really not sure if that is ok more generally.

Quoted 6 lines

Quoted 3 lines
Thanks,
-Jason
How about not enabling the flag unless the user specified the speed? This way we also do not need an "unknown" value.

This is even better.

Quoted 56 lines

Quoted 56 lines
Quoted 56 lines
Quoted 56 lines
Thanks,
-Jason
Quoted 49 lines
Thanks,
Yan.
Quoted 45 lines
hw/net/virtio-net.c +1 -0
+    DEFINE_PROP_STRING("duplex", VirtIONet, net_conf.duplex_str),
   DEFINE_PROP_END_OF_LIST(),
};
include/hw/virtio/virtio-net.h +3 -0
index b81b6a4..af74a94 100644
@@ -38,6 +38,9 @@ typedef struct virtio_net_conf
   uint16_t rx_queue_size;
   uint16_t tx_queue_size;
   uint16_t mtu;
+    int32_t speed;
+    char *duplex_str;
+    uint8_t duplex;
 } virtio_net_conf;
 
 /* Maximum packet size we can receive from tap device: header + 64k */
include/standard-headers/linux/virtio_net.h +0 -0
index 30ff249..0ff1447 100644
@@ -36,6 +36,7 @@
 #define VIRTIO_NET_F_GUEST_CSUM1/* Guest handles pkts w/ partial csum */
 #define VIRTIO_NET_F_CTRL_GUEST_OFFLOADS 2 /* Dynamic offload
configuration. */
#define VIRTIO_NET_F_MTU3/* Initial MTU advice */
+#define VIRTIO_NET_F_SPEED_DUPLEX 4/* Host set linkspeed and duplex */ #define VIRTIO_NET_F_MAC5/* Host has given MAC address. */ #define VIRTIO_NET_F_GUEST_TSO47/* Guest can handle TSOv4 in. */ #define VIRTIO_NET_F_GUEST_TSO68/* Guest can handle TSOv6 in. */
include/standard-headers/linux/virtio_net.h +3 -0
@@ -76,6 +77,9 @@ struct virtio_net_config {
 uint16_t max_virtqueue_pairs;
 /* Default maximum transmit unit advice */
 uint16_t mtu;
+/* Host exported linkspeed and duplex */
+uint32_t speed;
+uint8_t duplex;
 } QEMU_PACKED;
 
 /*
Signature · 1 line
2.6.1

Jason Baron

author

2017-12-21 19:42 UTC #16 lore raw

Re: [Qemu-devel] [PATCH 2/2] qemu: add linkspeed and duplex setting to virtio-net

Quoted 128 lines · Yan Vugenfirer

Quoted 128 lines · Michael S. Tsirkin
Quoted 124 lines · Jason Baron
Quoted 108 lines · Yan Vugenfirer
Quoted 108 lines
On 18 Dec 2017, at 18:04, Jason Baron via Qemu-devel <[email protected] <mailto:[email protected]>> wrote:
Quoted 99 lines · Yan Vugenfirer
Quoted 90 lines
On 14 Dec 2017, at 21:33, Jason Baron via Qemu-devel <[email protected] <mailto:[email protected]>> wrote:
Although they can be currently set in linux via 'ethtool -s', this requires
guest changes, and thus it would be nice to extend this functionality such
that it can be configured automatically from the host (as other network do).
Linkspeed and duplex settings can be set as:
'-device virtio-net,speed=10000,duplex=full'
where speed is [-1...INT_MAX], and duplex is ["half"|"full"].
Signed-off-by Jason Baron <[email protected]
<mailto:[email protected]>>
Cc: "Michael S. Tsirkin" <[email protected] <mailto:[email protected]>>
Cc: Jason Wang <[email protected] <mailto:[email protected]>>
---
hw/net/virtio-net.c | 29
+++++++++++++++++++++++++++++
include/hw/virtio/virtio-net.h              |  3 +++
include/standard-headers/linux/virtio_net.h |  4 ++++
3 files changed, 36 insertions(+)
hw/net/virtio-net.c +10 -0
index 38674b0..d63e790 100644
@@ -40,6 +40,12 @@
 #define VIRTIO_NET_RX_QUEUE_MIN_SIZE VIRTIO_NET_RX_QUEUE_DEFAULT_SIZE
 #define VIRTIO_NET_TX_QUEUE_MIN_SIZE VIRTIO_NET_TX_QUEUE_DEFAULT_SIZE
 
+/* duplex and speed defines */
+#define DUPLEX_UNKNOWN          0xff
+#define DUPLEX_HALF             0x00
+#define DUPLEX_FULL             0x01
+#define SPEED_UNKNOWN           -1
+
 /*
 * Calculate the number of bytes up to and including the given 'field' of
 * 'container'.
@@ -61,6 +67,8 @@ static VirtIOFeature feature_sizes[] = {
    .end = endof(struct virtio_net_config, max_virtqueue_pairs)},
   {.flags = 1 << VIRTIO_NET_F_MTU,
    .end = endof(struct virtio_net_config, mtu)},
+    {.flags = 1 << VIRTIO_NET_F_SPEED_DUPLEX,
+     .end = endof(struct virtio_net_config, duplex)},
   {}
 };
 
@@ -88,6 +96,8 @@ static void virtio_net_get_config(VirtIODevice
 *vdev, uint8_t *config)
   virtio_stw_p(vdev, &netcfg.status, n->status);
   virtio_stw_p(vdev, &netcfg.max_virtqueue_pairs, n->max_queues);
   virtio_stw_p(vdev, &netcfg.mtu, n->net_conf.mtu);
+    virtio_stl_p(vdev, &netcfg.speed, n->net_conf.speed);
+    netcfg.duplex = n->net_conf.duplex;
   memcpy(netcfg.mac, n->mac, ETH_ALEN);
   memcpy(config, &netcfg, n->config_size);
}
hw/net/virtio-net.c +14 -0
@@ -1941,6 +1951,23 @@ static void
 virtio_net_device_realize(DeviceState *dev, Error **errp)
       n->host_features |= (0x1 << VIRTIO_NET_F_MTU);
   }
 
+    n->host_features |= (0x1 << VIRTIO_NET_F_SPEED_DUPLEX);
+    if (n->net_conf.duplex_str) {
+        if (strncmp(n->net_conf.duplex_str, "half", 5) == 0) {
+            n->net_conf.duplex = DUPLEX_HALF;
+        } else if (strncmp(n->net_conf.duplex_str, "full", 5) == 0) {
+            n->net_conf.duplex = DUPLEX_FULL;
+        } else {
+            error_setg(errp, "'duplex' must be 'half' or 'full'");
+        }
+    } else {
+        n->net_conf.duplex = DUPLEX_UNKNOWN;
+    }
+    if (n->net_conf.speed < SPEED_UNKNOWN) {
+            error_setg(errp, "'speed' must be between -1
(SPEED_UNKOWN) and "
hw/net/virtio-net.c +3 -0
+                       "INT_MAX");
+    }
+
   virtio_net_set_config_size(n, n->host_features);
   virtio_init(vdev, "virtio-net", VIRTIO_ID_NET, n->config_size);
hw/net/virtio-net.c static Property virtio_net_properties[] = { · line 2192
Quoted 6 lines
hw/net/virtio-net.c +1 -0
@@ -2160,6 +2187,8 @@ static Property virtio_net_properties[] = {
   DEFINE_PROP_UINT16("host_mtu", VirtIONet, net_conf.mtu, 0),
   DEFINE_PROP_BOOL("x-mtu-bypass-backend", VirtIONet,
 mtu_bypass_backend,
                    true),
+    DEFINE_PROP_INT32("speed", VirtIONet, net_conf.speed,
 SPEED_UNKNOWN),
From Windows guest perspective I prefer to have some reasonable default (10G for example).
Quoted 6 lines
Quoted 3 lines
hmmm, I didn't want to change/set the default here in case it broke something, but I'm ok setting it to some 'reasonable' value - (10G and duplex?), if the consensus is that that would be safe.
OK from my side.
Thanks.
I presume your speaking for windows - i'm wondering if under linux the virtio device suddenly start showing up as duplex, 10Gbps, will that break anything? I can speak for the use-cases we have here and that would certainly be fine for us, but i'm really not sure if that is ok more generally.

Quoted 9 lines

Quoted 6 lines
Quoted 3 lines
Thanks,
-Jason
How about not enabling the flag unless the user specified the speed? This way we also do not need an "unknown" value.
This is even better.

We may still want "unknown" here as a setting, because otherwise we can't support dynamically changing the speed later (if we added say a qmp monitor command), because the flag would not have been set.

So I am ok with the 10Gps and full duplex defaults, if at least one of 'speed=' or 'duplex=' is supplied, but I would prefer to leave the "unknown" values as a setting (to still potentially allow dynamic updates in this case).

Thanks,

-Jason

Quoted 57 lines

Quoted 56 lines
Quoted 56 lines
Quoted 56 lines
Quoted 56 lines
Thanks,
-Jason
Quoted 49 lines
Thanks,
Yan.
Quoted 45 lines
hw/net/virtio-net.c +1 -0
+    DEFINE_PROP_STRING("duplex", VirtIONet, net_conf.duplex_str),
   DEFINE_PROP_END_OF_LIST(),
};
include/hw/virtio/virtio-net.h +3 -0
index b81b6a4..af74a94 100644
@@ -38,6 +38,9 @@ typedef struct virtio_net_conf
   uint16_t rx_queue_size;
   uint16_t tx_queue_size;
   uint16_t mtu;
+    int32_t speed;
+    char *duplex_str;
+    uint8_t duplex;
 } virtio_net_conf;
 
 /* Maximum packet size we can receive from tap device: header + 64k */
include/standard-headers/linux/virtio_net.h +0 -0
index 30ff249..0ff1447 100644
@@ -36,6 +36,7 @@
 #define VIRTIO_NET_F_GUEST_CSUM1/* Guest handles pkts w/ partial csum */
 #define VIRTIO_NET_F_CTRL_GUEST_OFFLOADS 2 /* Dynamic offload
configuration. */
#define VIRTIO_NET_F_MTU3/* Initial MTU advice */
+#define VIRTIO_NET_F_SPEED_DUPLEX 4/* Host set linkspeed and duplex */ #define VIRTIO_NET_F_MAC5/* Host has given MAC address. */ #define VIRTIO_NET_F_GUEST_TSO47/* Guest can handle TSOv4 in. */ #define VIRTIO_NET_F_GUEST_TSO68/* Guest can handle TSOv6 in. */
include/standard-headers/linux/virtio_net.h +3 -0
@@ -76,6 +77,9 @@ struct virtio_net_config {
 uint16_t max_virtqueue_pairs;
 /* Default maximum transmit unit advice */
 uint16_t mtu;
+/* Host exported linkspeed and duplex */
+uint32_t speed;
+uint8_t duplex;
 } QEMU_PACKED;
 
 /*
Signature · 1 line
2.6.1

Michael S. Tsirkin

maintainer

2017-12-21 20:40 UTC #17 lore raw

Re: [Qemu-devel] [PATCH 2/2] qemu: add linkspeed and duplex setting to virtio-net

Quoted 144 lines · Jason Baron

Quoted 128 lines · Yan Vugenfirer
Quoted 128 lines · Michael S. Tsirkin
Quoted 124 lines · Jason Baron
Quoted 108 lines · Yan Vugenfirer
Quoted 108 lines
On 18 Dec 2017, at 18:04, Jason Baron via Qemu-devel <[email protected] <mailto:[email protected]>> wrote:
Quoted 99 lines · Yan Vugenfirer
Quoted 90 lines
On 14 Dec 2017, at 21:33, Jason Baron via Qemu-devel <[email protected] <mailto:[email protected]>> wrote:
Although they can be currently set in linux via 'ethtool -s', this requires
guest changes, and thus it would be nice to extend this functionality such
that it can be configured automatically from the host (as other network do).
Linkspeed and duplex settings can be set as:
'-device virtio-net,speed=10000,duplex=full'
where speed is [-1...INT_MAX], and duplex is ["half"|"full"].
Signed-off-by Jason Baron <[email protected]
<mailto:[email protected]>>
Cc: "Michael S. Tsirkin" <[email protected] <mailto:[email protected]>>
Cc: Jason Wang <[email protected] <mailto:[email protected]>>
---
hw/net/virtio-net.c | 29
+++++++++++++++++++++++++++++
include/hw/virtio/virtio-net.h              |  3 +++
include/standard-headers/linux/virtio_net.h |  4 ++++
3 files changed, 36 insertions(+)
hw/net/virtio-net.c +10 -0
index 38674b0..d63e790 100644
@@ -40,6 +40,12 @@
 #define VIRTIO_NET_RX_QUEUE_MIN_SIZE VIRTIO_NET_RX_QUEUE_DEFAULT_SIZE
 #define VIRTIO_NET_TX_QUEUE_MIN_SIZE VIRTIO_NET_TX_QUEUE_DEFAULT_SIZE
 
+/* duplex and speed defines */
+#define DUPLEX_UNKNOWN          0xff
+#define DUPLEX_HALF             0x00
+#define DUPLEX_FULL             0x01
+#define SPEED_UNKNOWN           -1
+
 /*
 * Calculate the number of bytes up to and including the given 'field' of
 * 'container'.
@@ -61,6 +67,8 @@ static VirtIOFeature feature_sizes[] = {
    .end = endof(struct virtio_net_config, max_virtqueue_pairs)},
   {.flags = 1 << VIRTIO_NET_F_MTU,
    .end = endof(struct virtio_net_config, mtu)},
+    {.flags = 1 << VIRTIO_NET_F_SPEED_DUPLEX,
+     .end = endof(struct virtio_net_config, duplex)},
   {}
 };
 
@@ -88,6 +96,8 @@ static void virtio_net_get_config(VirtIODevice
 *vdev, uint8_t *config)
   virtio_stw_p(vdev, &netcfg.status, n->status);
   virtio_stw_p(vdev, &netcfg.max_virtqueue_pairs, n->max_queues);
   virtio_stw_p(vdev, &netcfg.mtu, n->net_conf.mtu);
+    virtio_stl_p(vdev, &netcfg.speed, n->net_conf.speed);
+    netcfg.duplex = n->net_conf.duplex;
   memcpy(netcfg.mac, n->mac, ETH_ALEN);
   memcpy(config, &netcfg, n->config_size);
}
hw/net/virtio-net.c +14 -0
@@ -1941,6 +1951,23 @@ static void
 virtio_net_device_realize(DeviceState *dev, Error **errp)
       n->host_features |= (0x1 << VIRTIO_NET_F_MTU);
   }
 
+    n->host_features |= (0x1 << VIRTIO_NET_F_SPEED_DUPLEX);
+    if (n->net_conf.duplex_str) {
+        if (strncmp(n->net_conf.duplex_str, "half", 5) == 0) {
+            n->net_conf.duplex = DUPLEX_HALF;
+        } else if (strncmp(n->net_conf.duplex_str, "full", 5) == 0) {
+            n->net_conf.duplex = DUPLEX_FULL;
+        } else {
+            error_setg(errp, "'duplex' must be 'half' or 'full'");
+        }
+    } else {
+        n->net_conf.duplex = DUPLEX_UNKNOWN;
+    }
+    if (n->net_conf.speed < SPEED_UNKNOWN) {
+            error_setg(errp, "'speed' must be between -1
(SPEED_UNKOWN) and "
hw/net/virtio-net.c +3 -0
+                       "INT_MAX");
+    }
+
   virtio_net_set_config_size(n, n->host_features);
   virtio_init(vdev, "virtio-net", VIRTIO_ID_NET, n->config_size);
hw/net/virtio-net.c static Property virtio_net_properties[] = { · line 2192
Quoted 6 lines
hw/net/virtio-net.c +1 -0
@@ -2160,6 +2187,8 @@ static Property virtio_net_properties[] = {
   DEFINE_PROP_UINT16("host_mtu", VirtIONet, net_conf.mtu, 0),
   DEFINE_PROP_BOOL("x-mtu-bypass-backend", VirtIONet,
 mtu_bypass_backend,
                    true),
+    DEFINE_PROP_INT32("speed", VirtIONet, net_conf.speed,
 SPEED_UNKNOWN),
From Windows guest perspective I prefer to have some reasonable default (10G for example).
Quoted 6 lines
Quoted 3 lines
hmmm, I didn't want to change/set the default here in case it broke something, but I'm ok setting it to some 'reasonable' value - (10G and duplex?), if the consensus is that that would be safe.
OK from my side.
Thanks.
I presume your speaking for windows - i'm wondering if under linux the virtio device suddenly start showing up as duplex, 10Gbps, will that break anything? I can speak for the use-cases we have here and that would certainly be fine for us, but i'm really not sure if that is ok more generally.
Quoted 9 lines
Quoted 6 lines
Quoted 3 lines
Thanks,
-Jason
How about not enabling the flag unless the user specified the speed? This way we also do not need an "unknown" value.
This is even better.
We may still want "unknown" here as a setting, because otherwise we can't support dynamically changing the speed later (if we added say a qmp monitor command), because the flag would not have been set.

Quoted 8 lines

So I am ok with the 10Gps and full duplex defaults, if at least one of 'speed=' or 'duplex=' is supplied, but I would prefer to leave the "unknown" values as a setting (to still potentially allow dynamic updates in this case).
Thanks,
-Jason

That's fine. And I really think we should leave defaults to unknown and not 10G.

Quoted 57 lines

Quoted 57 lines
Quoted 56 lines
Quoted 56 lines
Quoted 56 lines
Quoted 56 lines
Thanks,
-Jason
Quoted 49 lines
Thanks,
Yan.
Quoted 45 lines
hw/net/virtio-net.c +1 -0
+    DEFINE_PROP_STRING("duplex", VirtIONet, net_conf.duplex_str),
   DEFINE_PROP_END_OF_LIST(),
};
include/hw/virtio/virtio-net.h +3 -0
index b81b6a4..af74a94 100644
@@ -38,6 +38,9 @@ typedef struct virtio_net_conf
   uint16_t rx_queue_size;
   uint16_t tx_queue_size;
   uint16_t mtu;
+    int32_t speed;
+    char *duplex_str;
+    uint8_t duplex;
 } virtio_net_conf;
 
 /* Maximum packet size we can receive from tap device: header + 64k */
include/standard-headers/linux/virtio_net.h +0 -0
index 30ff249..0ff1447 100644
@@ -36,6 +36,7 @@
 #define VIRTIO_NET_F_GUEST_CSUM1/* Guest handles pkts w/ partial csum */
 #define VIRTIO_NET_F_CTRL_GUEST_OFFLOADS 2 /* Dynamic offload
configuration. */
#define VIRTIO_NET_F_MTU3/* Initial MTU advice */
+#define VIRTIO_NET_F_SPEED_DUPLEX 4/* Host set linkspeed and duplex */ #define VIRTIO_NET_F_MAC5/* Host has given MAC address. */ #define VIRTIO_NET_F_GUEST_TSO47/* Guest can handle TSOv4 in. */ #define VIRTIO_NET_F_GUEST_TSO68/* Guest can handle TSOv6 in. */
include/standard-headers/linux/virtio_net.h +3 -0
@@ -76,6 +77,9 @@ struct virtio_net_config {
 uint16_t max_virtqueue_pairs;
 /* Default maximum transmit unit advice */
 uint16_t mtu;
+/* Host exported linkspeed and duplex */
+uint32_t speed;
+uint8_t duplex;
 } QEMU_PACKED;
 
 /*
Signature · 1 line
2.6.1