Re: [PATCH v2] virtio-net: Add an optional device control over the receive buffers length

[Vitaly Mireyno <vmireyno@marvell.com>]

>-----Original Message-----
>From: Michael S. Tsirkin <mst@redhat.com>
>Sent: Tuesday, 4 February, 2020 18:42
>To: Vitaly Mireyno <vmireyno@marvell.com>
>Cc: virtio-comment@lists.oasis-open.org; Jason Wang
><jasowang@redhat.com>; Ariel Elior <aelior@marvell.com>
>Subject: [EXT] Re: [PATCH v2] virtio-net: Add an optional device control over
>the receive buffers length
>
>----------------------------------------------------------------------
>On Tue, Feb 04, 2020 at 04:13:36PM +0000, Vitaly Mireyno wrote:
>> This patch gives devices some level of control over the receive buffers
>length.
>> The driver declares the minimum receive buffer length, and the device
>requests max/min buffer length ratio.
>>
>> v2 incorporates v1 comments.
>>
>> Changes from v1:
>>  * min_rx_buf_len must be set before FEATURES_OK
>>  * Use a single feature bit VIRTIO_NET_F_RXBUF_LEN_CTRL for both, the
>> min size and the ratio
>>  * min_rx_buf_len is defined as write-only for driver
>>
>>
>> Signed-off-by: Vitaly Mireyno <vmireyno@marvell.com>
>> ---
>>  content.tex | 32 ++++++++++++++++++++++++++++++--
>>  1 file changed, 30 insertions(+), 2 deletions(-)
>>
>> diff --git a/content.tex b/content.tex index d68cfaf..ab0d193 100644
>> --- a/content.tex
>> +++ b/content.tex
>> @@ -2815,6 +2815,10 @@ \subsection{Feature bits}\label{sec:Device
>> Types / Network Device / Feature bits
>\item[VIRTIO_NET_F_CTRL_MAC_ADDR(23)] Set MAC address through
>control
>>      channel.
>>
>> +\item[VIRTIO_NET_F_RXBUF_LEN_CTRL(58)] Device requests to know the
>minimum
>> +    receive buffers length, and requests to limit the maximum/minimum
>receive
>> +    buffers length ratio. Driver declares the minimum receive buffers length.
>> +
>>  \item[VIRTIO_NET_F_GUEST_HDRLEN(59)] Driver can provide the exact
>\field{hdr_len}
>>      value. Device benefits from knowing the exact header length.
>>
>> @@ -2861,8 +2865,8 @@ \subsubsection{Legacy Interface: Feature
>> bits}\label{sec:Device Types / Network  \subsection{Device
>> configuration layout}\label{sec:Device Types / Network Device / Device
>> configuration layout}  \label{sec:Device Types / Block Device /
>> Feature bits / Device configuration layout}
>>
>> -Three driver-read-only configuration fields are currently defined.
>> The \field{mac} address field -always exists (though is only valid if
>> VIRTIO_NET_F_MAC is set), and
>> +The driver-read-only \field{mac} address field always exists (though
>> +is only valid if VIRTIO_NET_F_MAC is set), and
>>  \field{status} only exists if VIRTIO_NET_F_STATUS is set. Two
>> read-only bits (for the driver) are currently defined for the status field:
>>  VIRTIO_NET_S_LINK_UP and VIRTIO_NET_S_ANNOUNCE.
>> @@ -2882,12 +2886,22 @@ \subsection{Device configuration
>> layout}\label{sec:Device Types / Network Device  VIRTIO_NET_F_MTU is
>> set. This field specifies the maximum MTU for the driver to  use.
>>
>> +The write-only for driver field \field{min_rx_buf_len} only exists if
>> +VIRTIO_NET_F_RXBUF_LEN_CTRL is set. This field specifies the minimum
>> +receive buffers length.
>> +
>> +The driver-read-only field \field{rx_buf_len_ratio} only exists if
>> +VIRTIO_NET_F_RXBUF_LEN_CTRL is set. This field specifies the
>> +maximum/minimum receive buffers length ratio. The value '0' indicates
>unrestricted ratio.
>> +
>>  \begin{lstlisting}
>>  struct virtio_net_config {
>>          u8 mac[6];
>>          le16 status;
>>          le16 max_virtqueue_pairs;
>>          le16 mtu;
>> +        le32 min_rx_buf_len;
>> +        le16 rx_buf_len_ratio;
>>  };
>>  \end{lstlisting}
>>
>> @@ -2916,6 +2930,13 @@ \subsection{Device configuration
>> layout}\label{sec:Device Types / Network Device  If the driver
>> negotiates the VIRTIO_NET_F_STANDBY feature, the device MAY act  as a
>standby device for a primary device with the same MAC address.
>>
>> +A driver SHOULD accept the VIRTIO_NET_F_RXBUF_LEN_CTRL feature if
>offered.
>> +
>> +If VIRTIO_NET_F_RXBUF_LEN_CTRL feature has been negotiated, the
>> +driver MUST set \field{min_rx_buf_len} before setting the FEATURES_OK
>status bit.
>> +
>> +A driver MUST NOT modify \field{min_rx_buf_len} once it has been set.
>> +
>>  \drivernormative{\subsubsection}{Device configuration layout}{Device
>> Types / Network Device / Device configuration layout}
>>
>>  A driver SHOULD negotiate VIRTIO_NET_F_MAC if the device offers it.
>
>I still think this is too restrictive. Switching between e.g.
>XDP/non-XDP workloads demands ability to change this without device reset.
>I am thinking about sending the avail index to device with a command to
>change the size, this way device can figure out where are the old and where
>are the new buffers.
>Doesn't look too hard on the hardware, does it?
>

Changing descriptors size once in a while sounds reasonable and possible.
I see a problem though with this specific method - updating the rx descriptors size is an asynchronous slow-path operation (in vDPA it would go through a SW emulation). So there is no guarantee that the ring index, the driver requests, will be delivered on time to the device.
I can suggest another method, where the driver requests to change a descriptor size through a slow-path, but fast-path synchronization will come from the device - it will set a dedicated flag in the first used rx descriptor with the new size (from the device perspective).

So the flow of changing the descriptor size could be as follows:

For enlarging descriptors size:
 1. Driver enlarges descriptors size. All new descriptors are made available with the new size. Obviously, the buffers behind the descriptors must be enlarged as well.
 2. Driver requests the device to change descriptors size (by writing to the device configuration space). At this stage the device still thinks that all descriptors are of an old (smaller) size. Although new descriptors are of a new size, the driver must treat them as if they were of an old size (i.e. the device would not fill the whole descriptor).
 3. At some point, the device receives the driver's request, and it changes internally the rx descriptors size. The device sets a new_size flag in the next descriptor used.
 4. When the driver receives a used descriptor with a new_size flag set, from that point on it can treat all descriptors as being of a new size.

For reducing descriptors size:
 1. Driver requests the device to change descriptors size.
 2. At some point, the device receives the driver's request, and it changes internally the rx descriptors size. the device sets a new_size flag in the next descriptor used.
 3. When the driver receives a used descriptor with a new_size flag set, from that point on it must treat all the descriptors as being of the new size.
 4. Now the driver can actually reduce descriptors size. All new descriptors will be made available with a new size.


>> @@ -3281,6 +3302,13 @@ \subsubsection{Setting Up Receive
>> Buffers}\label{sec:Device Types / Network Devi  If VIRTIO_NET_F_MQ is
>> negotiated, each of receiveq1\ldots receiveqN  that will be used SHOULD be
>populated with receive buffers.
>>
>> +If VIRTIO_NET_F_RXBUF_LEN_CTRL feature has been negotiated, the
>> +driver MUST initialize all receive virtqueue descriptors \field{len}
>> +field with the value greater than or equal to the value configured in
>> +the \field{min_rx_buf_len} device configuration field, and allocate
>> +receive buffers accordingly. If \field{rx_buf_len_ratio} > 0, then
>> +the \field{len} value must also be less than or equal to
>(\field{min_rx_buf_len} * \field{rx_buf_len_ratio}).
>> +
>>  \devicenormative{\paragraph}{Setting Up Receive Buffers}{Device Types
>> / Network Device / Device Operation / Setting Up Receive Buffers}
>>
>>  The device MUST set \field{num_buffers} to the number of descriptors
>> used to
>> --