2020-12-29 11:17:45

by Arseny Krasnov

Subject: [PATCH 2/3] vhost/vsock: support for SOCK_SEQPACKET socket.

This patch simply adds transport ops and removes
ignore of non-stream type of packets.

Signed-off-by: Arseny Krasnov <[email protected]>
---
drivers/vhost/vsock.c | 6 ++++--
1 file changed, 4 insertions(+), 2 deletions(-)

diff --git a/drivers/vhost/vsock.c b/drivers/vhost/vsock.c
index a483cec31d5c..4a36ef1c52d0 100644
--- a/drivers/vhost/vsock.c
+++ b/drivers/vhost/vsock.c
@@ -346,8 +346,7 @@ vhost_vsock_alloc_pkt(struct vhost_virtqueue *vq,
return NULL;
}

- if (le16_to_cpu(pkt->hdr.type) == VIRTIO_VSOCK_TYPE_STREAM)
- pkt->len = le32_to_cpu(pkt->hdr.len);
+ pkt->len = le32_to_cpu(pkt->hdr.len);

/* No payload */
if (!pkt->len)
@@ -416,6 +415,9 @@ static struct virtio_transport vhost_transport = {
.stream_is_active = virtio_transport_stream_is_active,
.stream_allow = virtio_transport_stream_allow,

+ .seqpacket_seq_send_len = virtio_transport_seqpacket_seq_send_len,
+ .seqpacket_seq_get_len = virtio_transport_seqpacket_seq_get_len,
+
.notify_poll_in = virtio_transport_notify_poll_in,
.notify_poll_out = virtio_transport_notify_poll_out,
.notify_recv_init = virtio_transport_notify_recv_init,
--
2.25.1

2020-12-30 20:58:54

by Michael S. Tsirkin

Subject: Re: [PATCH 2/3] vhost/vsock: support for SOCK_SEQPACKET socket.

On Tue, Dec 29, 2020 at 02:06:33PM +0300, Arseny Krasnov wrote:
> This patch simply adds transport ops and removes
> ignore of non-stream type of packets.
>
> Signed-off-by: Arseny Krasnov <[email protected]>

How is this supposed to work? virtio vsock at the moment
has byte level end to end credit accounting at the
protocol level. I suspect some protocol changes involving
more than this tweak would
be needed to properly support anything that isn't a stream.

> ---
> drivers/vhost/vsock.c | 6 ++++--
> 1 file changed, 4 insertions(+), 2 deletions(-)
>
> diff --git a/drivers/vhost/vsock.c b/drivers/vhost/vsock.c
> index a483cec31d5c..4a36ef1c52d0 100644
> --- a/drivers/vhost/vsock.c
> +++ b/drivers/vhost/vsock.c
> @@ -346,8 +346,7 @@ vhost_vsock_alloc_pkt(struct vhost_virtqueue *vq,
> return NULL;
> }
>
> - if (le16_to_cpu(pkt->hdr.type) == VIRTIO_VSOCK_TYPE_STREAM)
> - pkt->len = le32_to_cpu(pkt->hdr.len);
> + pkt->len = le32_to_cpu(pkt->hdr.len);
>
> /* No payload */
> if (!pkt->len)
> @@ -416,6 +415,9 @@ static struct virtio_transport vhost_transport = {
> .stream_is_active = virtio_transport_stream_is_active,
> .stream_allow = virtio_transport_stream_allow,
>
> + .seqpacket_seq_send_len = virtio_transport_seqpacket_seq_send_len,
> + .seqpacket_seq_get_len = virtio_transport_seqpacket_seq_get_len,
> +
> .notify_poll_in = virtio_transport_notify_poll_in,
> .notify_poll_out = virtio_transport_notify_poll_out,
> .notify_recv_init = virtio_transport_notify_recv_init,
> --
> 2.25.1

2020-12-30 21:01:16

by Michael S. Tsirkin

Subject: Re: [PATCH 2/3] vhost/vsock: support for SOCK_SEQPACKET socket.

On Wed, Dec 30, 2020 at 03:56:03PM -0500, Michael S. Tsirkin wrote:
> On Tue, Dec 29, 2020 at 02:06:33PM +0300, Arseny Krasnov wrote:
> > This patch simply adds transport ops and removes
> > ignore of non-stream type of packets.
> >
> > Signed-off-by: Arseny Krasnov <[email protected]>
>
> How is this supposed to work? virtio vsock at the moment
> has byte level end to end credit accounting at the
> protocol level. I suspect some protocol changes involving
> more than this tweak would
> be needed to properly support anything that isn't a stream.


oh this was not threaded correctly so I did not see rest of
patches. ignore this pls will respond there.

> > ---
> > drivers/vhost/vsock.c | 6 ++++--
> > 1 file changed, 4 insertions(+), 2 deletions(-)
> >
> > diff --git a/drivers/vhost/vsock.c b/drivers/vhost/vsock.c
> > index a483cec31d5c..4a36ef1c52d0 100644
> > --- a/drivers/vhost/vsock.c
> > +++ b/drivers/vhost/vsock.c
> > @@ -346,8 +346,7 @@ vhost_vsock_alloc_pkt(struct vhost_virtqueue *vq,
> > return NULL;
> > }
> >
> > - if (le16_to_cpu(pkt->hdr.type) == VIRTIO_VSOCK_TYPE_STREAM)
> > - pkt->len = le32_to_cpu(pkt->hdr.len);
> > + pkt->len = le32_to_cpu(pkt->hdr.len);
> >
> > /* No payload */
> > if (!pkt->len)
> > @@ -416,6 +415,9 @@ static struct virtio_transport vhost_transport = {
> > .stream_is_active = virtio_transport_stream_is_active,
> > .stream_allow = virtio_transport_stream_allow,
> >
> > + .seqpacket_seq_send_len = virtio_transport_seqpacket_seq_send_len,
> > + .seqpacket_seq_get_len = virtio_transport_seqpacket_seq_get_len,
> > +
> > .notify_poll_in = virtio_transport_notify_poll_in,
> > .notify_poll_out = virtio_transport_notify_poll_out,
> > .notify_recv_init = virtio_transport_notify_recv_init,
> > --
> > 2.25.1

2021-01-02 10:15:45

by Arseny Krasnov

Subject: Re: [PATCH 2/3] vhost/vsock: support for SOCK_SEQPACKET socket.

> oh this was not threaded correctly so I did not see rest of
> patches. ignore this pls will respond there.

Sorry, i'll resend whole patchset ASAP in one thread.

On 30.12.2020 23:58, Michael S. Tsirkin wrote:
> On Wed, Dec 30, 2020 at 03:56:03PM -0500, Michael S. Tsirkin wrote:
>> On Tue, Dec 29, 2020 at 02:06:33PM +0300, Arseny Krasnov wrote:
>>> This patch simply adds transport ops and removes
>>> ignore of non-stream type of packets.
>>>
>>> Signed-off-by: Arseny Krasnov <[email protected]>
>> How is this supposed to work? virtio vsock at the moment
>> has byte level end to end credit accounting at the
>> protocol level. I suspect some protocol changes involving
>> more than this tweak would
>> be needed to properly support anything that isn't a stream.
>
> oh this was not threaded correctly so I did not see rest of
> patches. ignore this pls will respond there.
>
>>> ---
>>> drivers/vhost/vsock.c | 6 ++++--
>>> 1 file changed, 4 insertions(+), 2 deletions(-)
>>>
>>> diff --git a/drivers/vhost/vsock.c b/drivers/vhost/vsock.c
>>> index a483cec31d5c..4a36ef1c52d0 100644
>>> --- a/drivers/vhost/vsock.c
>>> +++ b/drivers/vhost/vsock.c
>>> @@ -346,8 +346,7 @@ vhost_vsock_alloc_pkt(struct vhost_virtqueue *vq,
>>> return NULL;
>>> }
>>>
>>> - if (le16_to_cpu(pkt->hdr.type) == VIRTIO_VSOCK_TYPE_STREAM)
>>> - pkt->len = le32_to_cpu(pkt->hdr.len);
>>> + pkt->len = le32_to_cpu(pkt->hdr.len);
>>>
>>> /* No payload */
>>> if (!pkt->len)
>>> @@ -416,6 +415,9 @@ static struct virtio_transport vhost_transport = {
>>> .stream_is_active = virtio_transport_stream_is_active,
>>> .stream_allow = virtio_transport_stream_allow,
>>>
>>> + .seqpacket_seq_send_len = virtio_transport_seqpacket_seq_send_len,
>>> + .seqpacket_seq_get_len = virtio_transport_seqpacket_seq_get_len,
>>> +
>>> .notify_poll_in = virtio_transport_notify_poll_in,
>>> .notify_poll_out = virtio_transport_notify_poll_out,
>>> .notify_recv_init = virtio_transport_notify_recv_init,
>>> --
>>> 2.25.1
>