* Re: [PATCH] vdpa: Set speed and duplex of mlx5_vnet to UNKNOWN
2024-08-28 18:16 [PATCH] vdpa: Set speed and duplex of mlx5_vnet to UNKNOWN Carlos Bilbao
@ 2024-08-29 4:49 ` Jason Wang
2024-08-29 6:08 ` Michael S. Tsirkin
2024-08-29 7:12 ` Michael S. Tsirkin
2024-08-29 10:20 ` Dragos Tatulea
2 siblings, 1 reply; 6+ messages in thread
From: Jason Wang @ 2024-08-29 4:49 UTC (permalink / raw)
To: Carlos Bilbao
Cc: dtatulea, mst, bilbao, xuanzhuo, eperezma, cratiu, lingshan.zhu,
virtualization, linux-kernel, Carlos Bilbao
On Thu, Aug 29, 2024 at 2:16 AM Carlos Bilbao
<carlos.bilbao.osdev@gmail.com> wrote:
>
> From: Carlos Bilbao <cbilbao@digitalocean.com>
>
> mlx5_vdpa vDPA devices currently don't support reporting or setting the
> speed and duplex and hence should be UNKNOWN instead of zero.
>
> Signed-off-by: Carlos Bilbao <cbilbao@digitalocean.com>
> ---
> drivers/vdpa/mlx5/net/mlx5_vnet.c | 7 +++++++
> drivers/vdpa/vdpa.c | 23 +++++++++++++++++++++++
> include/uapi/linux/vdpa.h | 2 ++
> 3 files changed, 32 insertions(+)
>
> diff --git a/drivers/vdpa/mlx5/net/mlx5_vnet.c b/drivers/vdpa/mlx5/net/mlx5_vnet.c
> index fa78e8288ebb..319f5c6121de 100644
> --- a/drivers/vdpa/mlx5/net/mlx5_vnet.c
> +++ b/drivers/vdpa/mlx5/net/mlx5_vnet.c
> @@ -3795,6 +3795,13 @@ static int mlx5_vdpa_dev_add(struct vdpa_mgmt_dev *v_mdev, const char *name,
> init_rwsem(&ndev->reslock);
> config = &ndev->config;
>
> + /*
> + * mlx5_vdpa vDPA devices currently don't support reporting or
> + * setting the speed or duplex.
> + */
> + config->speed = SPEED_UNKNOWN;
> + config->duplex = DUPLEX_UNKNOWN;
> +
> if (add_config->mask & BIT_ULL(VDPA_ATTR_DEV_NET_CFG_MTU)) {
> err = config_func_mtu(mdev, add_config->net.mtu);
> if (err)
> diff --git a/drivers/vdpa/vdpa.c b/drivers/vdpa/vdpa.c
> index 4dbd2e55a288..abde23e0041d 100644
> --- a/drivers/vdpa/vdpa.c
> +++ b/drivers/vdpa/vdpa.c
> @@ -15,6 +15,7 @@
> #include <net/genetlink.h>
> #include <linux/mod_devicetable.h>
> #include <linux/virtio_ids.h>
> +#include <uapi/linux/ethtool.h>
>
> static LIST_HEAD(mdev_head);
> /* A global mutex that protects vdpa management device and device level operations. */
> @@ -919,6 +920,22 @@ static int vdpa_dev_net_status_config_fill(struct sk_buff *msg, u64 features,
> return nla_put_u16(msg, VDPA_ATTR_DEV_NET_STATUS, val_u16);
> }
>
> +static int vdpa_dev_net_speed_config_fill(struct sk_buff *msg, u64 features,
> + struct virtio_net_config *config)
> +{
> + __le32 speed = cpu_to_le32(SPEED_UNKNOWN);
> +
> + return nla_put(msg, VDPA_ATTR_DEV_NET_CFG_SPEED, sizeof(speed), &speed);
> +}
> +
> +static int vdpa_dev_net_duplex_config_fill(struct sk_buff *msg, u64 features,
> + struct virtio_net_config *config)
> +{
> + u8 duplex = DUPLEX_UNKNOWN;
> +
> + return nla_put(msg, VDPA_ATTR_DEV_NET_CFG_DUPLEX, sizeof(duplex), &duplex);
> +}
> +
> static int vdpa_dev_net_config_fill(struct vdpa_device *vdev, struct sk_buff *msg)
> {
> struct virtio_net_config config = {};
> @@ -941,6 +958,12 @@ static int vdpa_dev_net_config_fill(struct vdpa_device *vdev, struct sk_buff *ms
> if (vdpa_dev_net_status_config_fill(msg, features_device, &config))
> return -EMSGSIZE;
>
> + if (vdpa_dev_net_speed_config_fill(msg, features_device, &config))
> + return -EMSGSIZE;
> +
> + if (vdpa_dev_net_duplex_config_fill(msg, features_device, &config))
> + return -EMSGSIZE;
> +
> return vdpa_dev_net_mq_config_fill(msg, features_device, &config);
> }
>
> diff --git a/include/uapi/linux/vdpa.h b/include/uapi/linux/vdpa.h
> index 842bf1201ac4..1c64ee0dd7b1 100644
> --- a/include/uapi/linux/vdpa.h
> +++ b/include/uapi/linux/vdpa.h
> @@ -43,6 +43,8 @@ enum vdpa_attr {
> VDPA_ATTR_DEV_NET_STATUS, /* u8 */
> VDPA_ATTR_DEV_NET_CFG_MAX_VQP, /* u16 */
> VDPA_ATTR_DEV_NET_CFG_MTU, /* u16 */
> + VDPA_ATTR_DEV_NET_CFG_SPEED, /* u32 */
> + VDPA_ATTR_DEV_NET_CFG_DUPLEX, /* u8 */
This should be an independent patch as it allows vdpa tool to
provision speed and duplex.
But a fundamental question is if we need such provisioning?
Thanks
>
> VDPA_ATTR_DEV_NEGOTIATED_FEATURES, /* u64 */
> VDPA_ATTR_DEV_MGMTDEV_MAX_VQS, /* u32 */
> --
> 2.34.1
>
^ permalink raw reply [flat|nested] 6+ messages in thread* Re: [PATCH] vdpa: Set speed and duplex of mlx5_vnet to UNKNOWN
2024-08-29 4:49 ` Jason Wang
@ 2024-08-29 6:08 ` Michael S. Tsirkin
0 siblings, 0 replies; 6+ messages in thread
From: Michael S. Tsirkin @ 2024-08-29 6:08 UTC (permalink / raw)
To: Jason Wang
Cc: Carlos Bilbao, dtatulea, bilbao, xuanzhuo, eperezma, cratiu,
lingshan.zhu, virtualization, linux-kernel, Carlos Bilbao
On Thu, Aug 29, 2024 at 12:49:24PM +0800, Jason Wang wrote:
> On Thu, Aug 29, 2024 at 2:16 AM Carlos Bilbao
> <carlos.bilbao.osdev@gmail.com> wrote:
> >
> > From: Carlos Bilbao <cbilbao@digitalocean.com>
> >
> > mlx5_vdpa vDPA devices currently don't support reporting or setting the
> > speed and duplex and hence should be UNKNOWN instead of zero.
> >
> > Signed-off-by: Carlos Bilbao <cbilbao@digitalocean.com>
> > ---
> > drivers/vdpa/mlx5/net/mlx5_vnet.c | 7 +++++++
> > drivers/vdpa/vdpa.c | 23 +++++++++++++++++++++++
> > include/uapi/linux/vdpa.h | 2 ++
> > 3 files changed, 32 insertions(+)
> >
> > diff --git a/drivers/vdpa/mlx5/net/mlx5_vnet.c b/drivers/vdpa/mlx5/net/mlx5_vnet.c
> > index fa78e8288ebb..319f5c6121de 100644
> > --- a/drivers/vdpa/mlx5/net/mlx5_vnet.c
> > +++ b/drivers/vdpa/mlx5/net/mlx5_vnet.c
> > @@ -3795,6 +3795,13 @@ static int mlx5_vdpa_dev_add(struct vdpa_mgmt_dev *v_mdev, const char *name,
> > init_rwsem(&ndev->reslock);
> > config = &ndev->config;
> >
> > + /*
> > + * mlx5_vdpa vDPA devices currently don't support reporting or
> > + * setting the speed or duplex.
> > + */
> > + config->speed = SPEED_UNKNOWN;
> > + config->duplex = DUPLEX_UNKNOWN;
> > +
> > if (add_config->mask & BIT_ULL(VDPA_ATTR_DEV_NET_CFG_MTU)) {
> > err = config_func_mtu(mdev, add_config->net.mtu);
> > if (err)
> > diff --git a/drivers/vdpa/vdpa.c b/drivers/vdpa/vdpa.c
> > index 4dbd2e55a288..abde23e0041d 100644
> > --- a/drivers/vdpa/vdpa.c
> > +++ b/drivers/vdpa/vdpa.c
> > @@ -15,6 +15,7 @@
> > #include <net/genetlink.h>
> > #include <linux/mod_devicetable.h>
> > #include <linux/virtio_ids.h>
> > +#include <uapi/linux/ethtool.h>
> >
> > static LIST_HEAD(mdev_head);
> > /* A global mutex that protects vdpa management device and device level operations. */
> > @@ -919,6 +920,22 @@ static int vdpa_dev_net_status_config_fill(struct sk_buff *msg, u64 features,
> > return nla_put_u16(msg, VDPA_ATTR_DEV_NET_STATUS, val_u16);
> > }
> >
> > +static int vdpa_dev_net_speed_config_fill(struct sk_buff *msg, u64 features,
> > + struct virtio_net_config *config)
> > +{
> > + __le32 speed = cpu_to_le32(SPEED_UNKNOWN);
> > +
> > + return nla_put(msg, VDPA_ATTR_DEV_NET_CFG_SPEED, sizeof(speed), &speed);
> > +}
> > +
> > +static int vdpa_dev_net_duplex_config_fill(struct sk_buff *msg, u64 features,
> > + struct virtio_net_config *config)
> > +{
> > + u8 duplex = DUPLEX_UNKNOWN;
> > +
> > + return nla_put(msg, VDPA_ATTR_DEV_NET_CFG_DUPLEX, sizeof(duplex), &duplex);
> > +}
> > +
> > static int vdpa_dev_net_config_fill(struct vdpa_device *vdev, struct sk_buff *msg)
> > {
> > struct virtio_net_config config = {};
> > @@ -941,6 +958,12 @@ static int vdpa_dev_net_config_fill(struct vdpa_device *vdev, struct sk_buff *ms
> > if (vdpa_dev_net_status_config_fill(msg, features_device, &config))
> > return -EMSGSIZE;
> >
> > + if (vdpa_dev_net_speed_config_fill(msg, features_device, &config))
> > + return -EMSGSIZE;
> > +
> > + if (vdpa_dev_net_duplex_config_fill(msg, features_device, &config))
> > + return -EMSGSIZE;
> > +
> > return vdpa_dev_net_mq_config_fill(msg, features_device, &config);
> > }
> >
> > diff --git a/include/uapi/linux/vdpa.h b/include/uapi/linux/vdpa.h
> > index 842bf1201ac4..1c64ee0dd7b1 100644
> > --- a/include/uapi/linux/vdpa.h
> > +++ b/include/uapi/linux/vdpa.h
> > @@ -43,6 +43,8 @@ enum vdpa_attr {
> > VDPA_ATTR_DEV_NET_STATUS, /* u8 */
> > VDPA_ATTR_DEV_NET_CFG_MAX_VQP, /* u16 */
> > VDPA_ATTR_DEV_NET_CFG_MTU, /* u16 */
> > + VDPA_ATTR_DEV_NET_CFG_SPEED, /* u32 */
> > + VDPA_ATTR_DEV_NET_CFG_DUPLEX, /* u8 */
>
> This should be an independent patch as it allows vdpa tool to
> provision speed and duplex.
or same patch, but document this more clearly:
when subj starts with mlx5 I expect it only affects mlx5.
> But a fundamental question is if we need such provisioning?
why not?
> Thanks
>
> >
> > VDPA_ATTR_DEV_NEGOTIATED_FEATURES, /* u64 */
> > VDPA_ATTR_DEV_MGMTDEV_MAX_VQS, /* u32 */
> > --
> > 2.34.1
> >
^ permalink raw reply [flat|nested] 6+ messages in thread
* Re: [PATCH] vdpa: Set speed and duplex of mlx5_vnet to UNKNOWN
2024-08-28 18:16 [PATCH] vdpa: Set speed and duplex of mlx5_vnet to UNKNOWN Carlos Bilbao
2024-08-29 4:49 ` Jason Wang
@ 2024-08-29 7:12 ` Michael S. Tsirkin
2024-08-29 10:20 ` Dragos Tatulea
2 siblings, 0 replies; 6+ messages in thread
From: Michael S. Tsirkin @ 2024-08-29 7:12 UTC (permalink / raw)
To: Carlos Bilbao
Cc: dtatulea, jasowang, bilbao, xuanzhuo, eperezma, cratiu,
lingshan.zhu, virtualization, linux-kernel, Carlos Bilbao
On Wed, Aug 28, 2024 at 01:16:25PM -0500, Carlos Bilbao wrote:
> From: Carlos Bilbao <cbilbao@digitalocean.com>
>
> mlx5_vdpa vDPA devices currently don't support reporting or setting the
> speed and duplex and hence should be UNKNOWN instead of zero.
>
> Signed-off-by: Carlos Bilbao <cbilbao@digitalocean.com>
As Jason points out, commit log and subject should says what
the patch does.
> ---
> drivers/vdpa/mlx5/net/mlx5_vnet.c | 7 +++++++
> drivers/vdpa/vdpa.c | 23 +++++++++++++++++++++++
> include/uapi/linux/vdpa.h | 2 ++
> 3 files changed, 32 insertions(+)
>
> diff --git a/drivers/vdpa/mlx5/net/mlx5_vnet.c b/drivers/vdpa/mlx5/net/mlx5_vnet.c
> index fa78e8288ebb..319f5c6121de 100644
> --- a/drivers/vdpa/mlx5/net/mlx5_vnet.c
> +++ b/drivers/vdpa/mlx5/net/mlx5_vnet.c
> @@ -3795,6 +3795,13 @@ static int mlx5_vdpa_dev_add(struct vdpa_mgmt_dev *v_mdev, const char *name,
> init_rwsem(&ndev->reslock);
> config = &ndev->config;
>
> + /*
> + * mlx5_vdpa vDPA devices currently don't support reporting or
> + * setting the speed or duplex.
> + */
> + config->speed = SPEED_UNKNOWN;
> + config->duplex = DUPLEX_UNKNOWN;
> +
> if (add_config->mask & BIT_ULL(VDPA_ATTR_DEV_NET_CFG_MTU)) {
> err = config_func_mtu(mdev, add_config->net.mtu);
> if (err)
> diff --git a/drivers/vdpa/vdpa.c b/drivers/vdpa/vdpa.c
> index 4dbd2e55a288..abde23e0041d 100644
> --- a/drivers/vdpa/vdpa.c
> +++ b/drivers/vdpa/vdpa.c
> @@ -15,6 +15,7 @@
> #include <net/genetlink.h>
> #include <linux/mod_devicetable.h>
> #include <linux/virtio_ids.h>
> +#include <uapi/linux/ethtool.h>
>
> static LIST_HEAD(mdev_head);
> /* A global mutex that protects vdpa management device and device level operations. */
> @@ -919,6 +920,22 @@ static int vdpa_dev_net_status_config_fill(struct sk_buff *msg, u64 features,
> return nla_put_u16(msg, VDPA_ATTR_DEV_NET_STATUS, val_u16);
> }
>
> +static int vdpa_dev_net_speed_config_fill(struct sk_buff *msg, u64 features,
> + struct virtio_net_config *config)
> +{
> + __le32 speed = cpu_to_le32(SPEED_UNKNOWN);
> +
> + return nla_put(msg, VDPA_ATTR_DEV_NET_CFG_SPEED, sizeof(speed), &speed);
> +}
> +
> +static int vdpa_dev_net_duplex_config_fill(struct sk_buff *msg, u64 features,
> + struct virtio_net_config *config)
> +{
> + u8 duplex = DUPLEX_UNKNOWN;
> +
> + return nla_put(msg, VDPA_ATTR_DEV_NET_CFG_DUPLEX, sizeof(duplex), &duplex);
> +}
> +
> static int vdpa_dev_net_config_fill(struct vdpa_device *vdev, struct sk_buff *msg)
> {
> struct virtio_net_config config = {};
> @@ -941,6 +958,12 @@ static int vdpa_dev_net_config_fill(struct vdpa_device *vdev, struct sk_buff *ms
> if (vdpa_dev_net_status_config_fill(msg, features_device, &config))
> return -EMSGSIZE;
>
> + if (vdpa_dev_net_speed_config_fill(msg, features_device, &config))
> + return -EMSGSIZE;
> +
> + if (vdpa_dev_net_duplex_config_fill(msg, features_device, &config))
> + return -EMSGSIZE;
> +
> return vdpa_dev_net_mq_config_fill(msg, features_device, &config);
> }
>
> diff --git a/include/uapi/linux/vdpa.h b/include/uapi/linux/vdpa.h
> index 842bf1201ac4..1c64ee0dd7b1 100644
> --- a/include/uapi/linux/vdpa.h
> +++ b/include/uapi/linux/vdpa.h
> @@ -43,6 +43,8 @@ enum vdpa_attr {
> VDPA_ATTR_DEV_NET_STATUS, /* u8 */
> VDPA_ATTR_DEV_NET_CFG_MAX_VQP, /* u16 */
> VDPA_ATTR_DEV_NET_CFG_MTU, /* u16 */
> + VDPA_ATTR_DEV_NET_CFG_SPEED, /* u32 */
> + VDPA_ATTR_DEV_NET_CFG_DUPLEX, /* u8 */
>
> VDPA_ATTR_DEV_NEGOTIATED_FEATURES, /* u64 */
> VDPA_ATTR_DEV_MGMTDEV_MAX_VQS, /* u32 */
> --
> 2.34.1
^ permalink raw reply [flat|nested] 6+ messages in thread* Re: [PATCH] vdpa: Set speed and duplex of mlx5_vnet to UNKNOWN
2024-08-28 18:16 [PATCH] vdpa: Set speed and duplex of mlx5_vnet to UNKNOWN Carlos Bilbao
2024-08-29 4:49 ` Jason Wang
2024-08-29 7:12 ` Michael S. Tsirkin
@ 2024-08-29 10:20 ` Dragos Tatulea
2024-08-29 13:36 ` Carlos Bilbao
2 siblings, 1 reply; 6+ messages in thread
From: Dragos Tatulea @ 2024-08-29 10:20 UTC (permalink / raw)
To: Carlos Bilbao, mst, jasowang
Cc: bilbao, xuanzhuo, eperezma, cratiu, lingshan.zhu, virtualization,
linux-kernel, Carlos Bilbao
On 28.08.24 20:16, Carlos Bilbao wrote:
> From: Carlos Bilbao <cbilbao@digitalocean.com>
>
> mlx5_vdpa vDPA devices currently don't support reporting or setting the
> speed and duplex and hence should be UNKNOWN instead of zero.
>
> Signed-off-by: Carlos Bilbao <cbilbao@digitalocean.com>
> ---
> drivers/vdpa/mlx5/net/mlx5_vnet.c | 7 +++++++
> drivers/vdpa/vdpa.c | 23 +++++++++++++++++++++++
> include/uapi/linux/vdpa.h | 2 ++
> 3 files changed, 32 insertions(+)
>
> diff --git a/drivers/vdpa/mlx5/net/mlx5_vnet.c b/drivers/vdpa/mlx5/net/mlx5_vnet.c
> index fa78e8288ebb..319f5c6121de 100644
> --- a/drivers/vdpa/mlx5/net/mlx5_vnet.c
> +++ b/drivers/vdpa/mlx5/net/mlx5_vnet.c
> @@ -3795,6 +3795,13 @@ static int mlx5_vdpa_dev_add(struct vdpa_mgmt_dev *v_mdev, const char *name,
> init_rwsem(&ndev->reslock);
> config = &ndev->config;
>
> + /*
> + * mlx5_vdpa vDPA devices currently don't support reporting or
> + * setting the speed or duplex.
> + */
> + config->speed = SPEED_UNKNOWN;
> + config->duplex = DUPLEX_UNKNOWN;
> +
The values in virtio_net_config are little endian so you'll need to explicitly
convert them. As speed is a u32, you'll need to add a cpu_to_mlx5vdpa32() helper.
Thanks,
Dragos
> if (add_config->mask & BIT_ULL(VDPA_ATTR_DEV_NET_CFG_MTU)) {
> err = config_func_mtu(mdev, add_config->net.mtu);
> if (err)
> diff --git a/drivers/vdpa/vdpa.c b/drivers/vdpa/vdpa.c
> index 4dbd2e55a288..abde23e0041d 100644
> --- a/drivers/vdpa/vdpa.c
> +++ b/drivers/vdpa/vdpa.c
> @@ -15,6 +15,7 @@
> #include <net/genetlink.h>
> #include <linux/mod_devicetable.h>
> #include <linux/virtio_ids.h>
> +#include <uapi/linux/ethtool.h>
>
> static LIST_HEAD(mdev_head);
> /* A global mutex that protects vdpa management device and device level operations. */
> @@ -919,6 +920,22 @@ static int vdpa_dev_net_status_config_fill(struct sk_buff *msg, u64 features,
> return nla_put_u16(msg, VDPA_ATTR_DEV_NET_STATUS, val_u16);
> }
>
> +static int vdpa_dev_net_speed_config_fill(struct sk_buff *msg, u64 features,
> + struct virtio_net_config *config)
> +{
> + __le32 speed = cpu_to_le32(SPEED_UNKNOWN);
> +
> + return nla_put(msg, VDPA_ATTR_DEV_NET_CFG_SPEED, sizeof(speed), &speed);
> +}
> +
> +static int vdpa_dev_net_duplex_config_fill(struct sk_buff *msg, u64 features,
> + struct virtio_net_config *config)
> +{
> + u8 duplex = DUPLEX_UNKNOWN;
> +
> + return nla_put(msg, VDPA_ATTR_DEV_NET_CFG_DUPLEX, sizeof(duplex), &duplex);
> +}
> +
> static int vdpa_dev_net_config_fill(struct vdpa_device *vdev, struct sk_buff *msg)
> {
> struct virtio_net_config config = {};
> @@ -941,6 +958,12 @@ static int vdpa_dev_net_config_fill(struct vdpa_device *vdev, struct sk_buff *ms
> if (vdpa_dev_net_status_config_fill(msg, features_device, &config))
> return -EMSGSIZE;
>
> + if (vdpa_dev_net_speed_config_fill(msg, features_device, &config))
> + return -EMSGSIZE;
> +
> + if (vdpa_dev_net_duplex_config_fill(msg, features_device, &config))
> + return -EMSGSIZE;
> +
> return vdpa_dev_net_mq_config_fill(msg, features_device, &config);
> }
>
> diff --git a/include/uapi/linux/vdpa.h b/include/uapi/linux/vdpa.h
> index 842bf1201ac4..1c64ee0dd7b1 100644
> --- a/include/uapi/linux/vdpa.h
> +++ b/include/uapi/linux/vdpa.h
> @@ -43,6 +43,8 @@ enum vdpa_attr {
> VDPA_ATTR_DEV_NET_STATUS, /* u8 */
> VDPA_ATTR_DEV_NET_CFG_MAX_VQP, /* u16 */
> VDPA_ATTR_DEV_NET_CFG_MTU, /* u16 */
> + VDPA_ATTR_DEV_NET_CFG_SPEED, /* u32 */
> + VDPA_ATTR_DEV_NET_CFG_DUPLEX, /* u8 */
>
> VDPA_ATTR_DEV_NEGOTIATED_FEATURES, /* u64 */
> VDPA_ATTR_DEV_MGMTDEV_MAX_VQS, /* u32 */
^ permalink raw reply [flat|nested] 6+ messages in thread* Re: [PATCH] vdpa: Set speed and duplex of mlx5_vnet to UNKNOWN
2024-08-29 10:20 ` Dragos Tatulea
@ 2024-08-29 13:36 ` Carlos Bilbao
0 siblings, 0 replies; 6+ messages in thread
From: Carlos Bilbao @ 2024-08-29 13:36 UTC (permalink / raw)
To: Dragos Tatulea, Carlos Bilbao, mst, jasowang
Cc: bilbao, xuanzhuo, eperezma, cratiu, lingshan.zhu, virtualization,
linux-kernel
Hello,
On 8/29/24 5:20 AM, Dragos Tatulea wrote:
>
> On 28.08.24 20:16, Carlos Bilbao wrote:
>> From: Carlos Bilbao <cbilbao@digitalocean.com>
>>
>> mlx5_vdpa vDPA devices currently don't support reporting or setting the
>> speed and duplex and hence should be UNKNOWN instead of zero.
>>
>> Signed-off-by: Carlos Bilbao <cbilbao@digitalocean.com>
>> ---
>> drivers/vdpa/mlx5/net/mlx5_vnet.c | 7 +++++++
>> drivers/vdpa/vdpa.c | 23 +++++++++++++++++++++++
>> include/uapi/linux/vdpa.h | 2 ++
>> 3 files changed, 32 insertions(+)
>>
>> diff --git a/drivers/vdpa/mlx5/net/mlx5_vnet.c b/drivers/vdpa/mlx5/net/mlx5_vnet.c
>> index fa78e8288ebb..319f5c6121de 100644
>> --- a/drivers/vdpa/mlx5/net/mlx5_vnet.c
>> +++ b/drivers/vdpa/mlx5/net/mlx5_vnet.c
>> @@ -3795,6 +3795,13 @@ static int mlx5_vdpa_dev_add(struct vdpa_mgmt_dev *v_mdev, const char *name,
>> init_rwsem(&ndev->reslock);
>> config = &ndev->config;
>>
>> + /*
>> + * mlx5_vdpa vDPA devices currently don't support reporting or
>> + * setting the speed or duplex.
>> + */
>> + config->speed = SPEED_UNKNOWN;
>> + config->duplex = DUPLEX_UNKNOWN;
>> +
> The values in virtio_net_config are little endian so you'll need to explicitly
> convert them. As speed is a u32, you'll need to add a cpu_to_mlx5vdpa32() helper.
>
> Thanks,
> Dragos
>
>> if (add_config->mask & BIT_ULL(VDPA_ATTR_DEV_NET_CFG_MTU)) {
>> err = config_func_mtu(mdev, add_config->net.mtu);
>> if (err)
>> diff --git a/drivers/vdpa/vdpa.c b/drivers/vdpa/vdpa.c
>> index 4dbd2e55a288..abde23e0041d 100644
>> --- a/drivers/vdpa/vdpa.c
>> +++ b/drivers/vdpa/vdpa.c
>> @@ -15,6 +15,7 @@
>> #include <net/genetlink.h>
>> #include <linux/mod_devicetable.h>
>> #include <linux/virtio_ids.h>
>> +#include <uapi/linux/ethtool.h>
>>
>> static LIST_HEAD(mdev_head);
>> /* A global mutex that protects vdpa management device and device level operations. */
>> @@ -919,6 +920,22 @@ static int vdpa_dev_net_status_config_fill(struct sk_buff *msg, u64 features,
>> return nla_put_u16(msg, VDPA_ATTR_DEV_NET_STATUS, val_u16);
>> }
>>
>> +static int vdpa_dev_net_speed_config_fill(struct sk_buff *msg, u64 features,
>> + struct virtio_net_config *config)
>> +{
>> + __le32 speed = cpu_to_le32(SPEED_UNKNOWN);
>> +
>> + return nla_put(msg, VDPA_ATTR_DEV_NET_CFG_SPEED, sizeof(speed), &speed);
>> +}
>> +
>> +static int vdpa_dev_net_duplex_config_fill(struct sk_buff *msg, u64 features,
>> + struct virtio_net_config *config)
>> +{
>> + u8 duplex = DUPLEX_UNKNOWN;
>> +
>> + return nla_put(msg, VDPA_ATTR_DEV_NET_CFG_DUPLEX, sizeof(duplex), &duplex);
>> +}
>> +
>> static int vdpa_dev_net_config_fill(struct vdpa_device *vdev, struct sk_buff *msg)
>> {
>> struct virtio_net_config config = {};
>> @@ -941,6 +958,12 @@ static int vdpa_dev_net_config_fill(struct vdpa_device *vdev, struct sk_buff *ms
>> if (vdpa_dev_net_status_config_fill(msg, features_device, &config))
>> return -EMSGSIZE;
>>
>> + if (vdpa_dev_net_speed_config_fill(msg, features_device, &config))
>> + return -EMSGSIZE;
>> +
>> + if (vdpa_dev_net_duplex_config_fill(msg, features_device, &config))
>> + return -EMSGSIZE;
>> +
>> return vdpa_dev_net_mq_config_fill(msg, features_device, &config);
>> }
>>
>> diff --git a/include/uapi/linux/vdpa.h b/include/uapi/linux/vdpa.h
>> index 842bf1201ac4..1c64ee0dd7b1 100644
>> --- a/include/uapi/linux/vdpa.h
>> +++ b/include/uapi/linux/vdpa.h
>> @@ -43,6 +43,8 @@ enum vdpa_attr {
>> VDPA_ATTR_DEV_NET_STATUS, /* u8 */
>> VDPA_ATTR_DEV_NET_CFG_MAX_VQP, /* u16 */
>> VDPA_ATTR_DEV_NET_CFG_MTU, /* u16 */
>> + VDPA_ATTR_DEV_NET_CFG_SPEED, /* u32 */
>> + VDPA_ATTR_DEV_NET_CFG_DUPLEX, /* u8 */
>>
>> VDPA_ATTR_DEV_NEGOTIATED_FEATURES, /* u64 */
>> VDPA_ATTR_DEV_MGMTDEV_MAX_VQS, /* u32 */
Thank you all for the feedback, everyone. I will prepare a patch set with
two commits:
- One to set speed and duplex to UNKNOWN (in little-endian format, thank
you, Dragos).
- Another to update the vDPA tool to allow setting the values for speed
and duplex, including support in set_config(). I'll add a comment
clarifying that this support isn't fully implemented at the hardware
level yet.
Regards,
Carlos
^ permalink raw reply [flat|nested] 6+ messages in thread