[Date Prev] | [Thread Prev] | [Thread Next] | [Date Next] -- [Date Index] | [Thread Index] | [List Home]
Subject: Re: [virtio-comment] Re: [PATCH v3] virtio_net: support inner header hash
å 2022/12/16 äå6:14, Heng Qi åé:
å 2022/12/16 äå2:33, Jason Wang åé:On Mon, Dec 5, 2022 at 2:36 PM Heng Qi <hengqi@linux.alibaba.com> wrote:When VIRTIO_NET_F_RSS is negotiated and the tunnel is used to encapsulate the packets, the hash calculated using the outer header of the receive packets is always fixed for the same flow packets, i.e. they will be steered to the same receive queue. We add a VIRTIO_NET_F_HASH_GRE_VXLAN_GENEVE_INNER feature bit andI think we can simply use VIRTIO_NET_F_HASH_TUNNEL now. And introduce VIRTIO_NET_F_HASH_TUNNEL_XYZ for future extension?Well, I very much agree.related bitmasks in \field{hash_types}, which instructs the device to calculate the hash using the inner headers of GRE, VXLAN or GENEVE-encapsulated packets. Besides, \field{hash_report_tunnel} areadded to report packet type when calculating hash over the inner header.Signed-off-by: Heng Qi <hengqi@linux.alibaba.com> Signed-off-by: Xuan Zhuo <xuanzhuo@linux.alibaba.com> --- v2: 1. Add a feature bit for GRE/VXLAN/GENEVE inner hash. @Jason Wang  2. Chang \field{hash_tunnel} to \field{hash_report_tunnel}. @Jason Wang, @Michael S. Tsirkinv1:  1. Remove the patch for the bitmask fix. @Michael S. Tsirkin  2. Clarify some paragraphs. @Jason Wang 3. Add \field{hash_tunnel} and VIRTIO_NET_HASH_REPORT_GRE. @Yuri Benditovich content.tex | 190 +++++++++++++++++++++++++++++++++++++++++++++++++--- 1 file changed, 181 insertions(+), 9 deletions(-) diff --git a/content.tex b/content.tex index e863709..675da87 100644 --- a/content.tex +++ b/content.tex@@ -3084,6 +3084,9 @@ \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_HASH_GRE_VXLAN_GENEVE_INNER(52)] Device supports inner+ header hash for GRE, VXLAN and GENEVE-encapsulated packets. + \item[VIRTIO_NET_F_NOTF_COAL(53)] Device supports notifications coalescing. \item[VIRTIO_NET_F_GUEST_USO4 (54)] Driver can receive USOv4 packets.@@ -3095,7 +3098,8 @@ \subsection{Feature bits}\label{sec:Device Types / Network Device / Feature bits  to several segments when each of these smaller packets has UDP header. \item[VIRTIO_NET_F_HASH_REPORT(57)] Device can report per-packet hash - value and a type of calculated hash. + value, a type of calculated hash and a tunnel packet type if + VIRTIO_NET_F_HASH_GRE_VXLAN_GENEVE_INNER is negotiated. \item[VIRTIO_NET_F_GUEST_HDRLEN(59)] Driver can provide the exact \field{hdr_len} value. Device benefits from knowing the exact header length.@@ -3140,6 +3144,7 @@ \subsubsection{Feature bit requirements}\label{sec:Device Types / Network Device \item[VIRTIO_NET_F_NOTF_COAL] Requires VIRTIO_NET_F_CTRL_VQ. \item[VIRTIO_NET_F_RSC_EXT] Requires VIRTIO_NET_F_HOST_TSO4 or VIRTIO_NET_F_HOST_TSO6. \item[VIRTIO_NET_F_RSS] Requires VIRTIO_NET_F_CTRL_VQ.+\item[VIRTIO_NET_F_HASH_GRE_VXLAN_GENEVE_INNER] Requires VIRTIO_NET_F_CTRL_VQ and VIRTIO_NET_F_RSS. \end{description} \subsubsection{Legacy Interface: Feature bits}\label{sec:Device Types / Network Device / Feature bits / Legacy Interface: Feature bits} @@ -3386,7 +3391,8 @@ \subsection{Device Operation}\label{sec:Device Types / Network Device / Device O le16 num_buffers; le32 hash_value; (Only if VIRTIO_NET_F_HASH_REPORT negotiated)  le16 hash_report; (Only if VIRTIO_NET_F_HASH_REPORT negotiated) - le16 padding_reserved; (Only if VIRTIO_NET_F_HASH_REPORT negotiated) + le8 hash_report_tunnel; (Only if VIRTIO_NET_F_HASH_REPORT negotiated)Only if both VIRTIO_NET_F_HASH_REPORT and VIRTIO_NET_F_HASH_GRE_VXLAN_GENEVE_INNER are negotiated?Yes, I will make it clear.+ le8 padding_reserved; (Only if VIRTIO_NET_F_HASH_REPORT negotiated) };  \end{lstlisting}@@ -3837,19 +3843,22 @@ \subsubsection{Processing of Incoming Packets}\label{sec:Device Types / Network  A device attempts to calculate a per-packet hash in the following cases: \begin{itemize} \item The feature VIRTIO_NET_F_RSS was negotiated. The device uses the hash to determine the receive virtqueue to place incoming packets. -\item The feature VIRTIO_NET_F_HASH_REPORT was negotiated. The device reports the hash value and the hash type with the packet. +\item The feature VIRTIO_NET_F_HASH_REPORT was negotiated. The device reports the hash value, the hash type and the tunnel packet type.The tunnel type can only be reported when both HASH_REPORT and HASH_GRE_VXLAN_GENEVE_INNER are negotiated?Yes, I will make it clear. \end{itemize}  If the feature VIRTIO_NET_F_RSS was negotiated:  \begin{itemize}-\item The device uses \field{hash_types} of the virtio_net_rss_config structure as 'Enabled hash types' bitmask. +\item The device uses \field{hash_types} of the virtio_net_rss_config structure as 'Enabled hash types' bitmask, +but VIRTIO_NET_HASH_TYPE_GRE_INNER, VIRTIO_NET_HASH_TYPE_VXLAN_INNER and VIRTIO_NET_HASH_TYPE_GENEVE_INNER in +\field{hash_types} can only be used when the feature VIRTIO_NET_F_HASH_GRE_VXLAN_GENEVE_INNER is negotiated.I think we can avoid this change by clarifying the feature dependency where the tunnel hash types are defined.Ok, I'll do it in the next version. \item The device uses a key as defined in \field{hash_key_data} and \field{hash_key_length} of the virtio_net_rss_config structure (see  \ref{sec:Device Types / Network Device / Device Operation / Control Virtqueue / Receive-side scaling (RSS) / Setting RSS parameters}). \end{itemize}  If the feature VIRTIO_NET_F_RSS was not negotiated:  \begin{itemize}-\item The device uses \field{hash_types} of the virtio_net_hash_config structure as 'Enabled hash types' bitmask. +\item The device uses \field{hash_types} of the virtio_net_hash_config structure as 'Enabled hash types' bitmask, +except VIRTIO_NET_HASH_TYPE_GRE_INNER, VIRTIO_NET_HASH_TYPE_VXLAN_INNER and VIRTIO_NET_HASH_TYPE_GENEVE_INNER.Why exclude those tunnel hash types from _F_MQ? Or do we want to say the automatic packet steering could be done via the inner packet when negotiated?As far as I know about spec, there are two ways to select a queue to place a packet.The first one is _F_MQ. In this scenario, if a packet arrives for the first time, it will be randomly placed (not based on hash) in a queue , subsequent packets of the samedata flow will be placed in the same queue. Here is what the spec says:âFor uni-directional protocols, or where no packets have been transmitted yet, the device MAY steer a packet to a random queue out of the specified receiveq1\ldots receiveqn.";The second is _F_RSS. In this scenario, the device calculates the hash based on the outer or inner header, and uses the hash value to select the rx queue. Therefore, when _F_RSS is not negotiated, we seem to have no initiative to actively select queues. But there is no need to actively explain the exclusion of these types here.Cc Xuan.
Hi Jason, what do you think about this? Thanks.
 \item The device uses a key as defined in \field{hash_key_data} and \field{hash_key_length} of the virtio_net_hash_config structure (see  \ref{sec:Device Types / Network Device / Device Operation / Control Virtqueue / Automatic receive steering in multiqueue mode / Hash calculation}). \end{itemize}@@ -3883,6 +3892,18 @@ \subsubsection{Processing of Incoming Packets}\label{sec:Device Types / Network #define VIRTIO_NET_HASH_TYPE_TCP_EX (1 << 7)  #define VIRTIO_NET_HASH_TYPE_UDP_EX (1 << 8)  \end{lstlisting} +Hash types applicable to inner payloads of GRE-encapsulated packets +\begin{lstlisting} +#define VIRTIO_NET_HASH_TYPE_GRE_INNER (1 << 9) +\end{lstlisting} +Hash types applicable to inner payloads of VXLAN-encapsulated packets +\begin{lstlisting} +#define VIRTIO_NET_HASH_TYPE_VXLAN_INNER (1 << 10) +\end{lstlisting} +Hash types applicable to inner payloads of GENEVE-encapsulated packets +\begin{lstlisting} +#define VIRTIO_NET_HASH_TYPE_GENEVE_INNER (1 << 11) +\end{lstlisting}  \subparagraph{IPv4 packets} \label{sec:Device Types / Network Device / Device Operation / Processing of Incoming Packets / Hash calculation for incoming packets / IPv4 packets} @@ -3975,15 +3996,148 @@ \subsubsection{Processing of Incoming Packets}\label{sec:Device Types / Network  (see \ref{sec:Device Types / Network Device / Device Operation / Processing of Incoming Packets / Hash calculation for incoming packets / IPv6 packets without extension header}). \end{itemize}+\subparagraph{Inner payloads of GRE, VXLAN or GENEVE-encapsulated packets} +\label{sec:Device Types / Network Device / Device Operation / Processing of Incoming Packets / Hash calculation for incoming packets / Inner payloads of GRE, VXLAN or GENEVE-encapsulated packets}}+VIRTIO_NET_HASH_TYPE_GRE_INNER, VIRTIO_NET_HASH_TYPE_VXLAN_INNER or+VIRTIO_NET_HASH_TYPE_GENEVE_INNER bit is set at the same time as one of +the bits between VIRTIO_NET_HASH_TYPE_IPv4 and VIRTIO_NET_HASH_TYPE_UDP_EX.++The device calculates the hash on GRE, VXLAN or GENEVE-encapsulated packets whose inner +payloads are IPv4 packets according to 'Enabled hash types' bitmasks as follows:+\begin{itemize} + \item If VIRTIO_NET_HASH_TYPE_TCPv4 is set along with one of+ VIRTIO_NET_HASH_TYPE_GRE_INNER, VIRTIO_NET_HASH_TYPE_VXLAN_INNER and + VIRTIO_NET_HASH_TYPE_GENEVE_INNER, and the GRE, VXLAN or GENEVE-encapsulated + packet has an inner TCPv4 header in its payload, the hash is calculated over+ the following fields: + \begin{itemsize} + \item inner Source IP address + \item inner Destination IP address + \item inner Source TCP port + \item inner Destination TCP port + \end{itemsize} + \item Else if VIRTIO_NET_HASH_TYPE_UDPv4 is set along with one of+ VIRTIO_NET_HASH_TYPE_GRE_INNER, VIRTIO_NET_HASH_TYPE_VXLAN_INNER and + VIRTIO_NET_HASH_TYPE_GENEVE_INNER, and the GRE, VXLAN or GENEVE-encapsulated + packet has an inner UDPv4 header in its payload, the hash is calculated over+ the following fields: + \begin{itemsize} + \item inner Source IP address + \item inner Destination IP address + \item inner Source UDP port + \item inner Destination UDP port + \end{itemize} + \item Else if VIRTIO_NET_HASH_TYPE_IPv4 is set along with one of+ VIRTIO_NET_HASH_TYPE_GRE_INNER, VIRTIO_NET_HASH_TYPE_VXLAN_INNER and + VIRTIO_NET_HASH_TYPE_GENEVE_INNER, the hash is calculated over the+ following fields: + \begin{itemsize} + \item inner Source IP address + \item inner Destination IP address + \end{itemsize} + \item Else the device does not calculate the hash +\end{itemize} ++The device calculates the hash on GRE, VXLAN or GENEVE-encapsulated packets whose inner +payloads are IPv6 packets without extension headers according to 'Enabled hash types'+bitmasks as follows: +\begin{itemize} + \item If VIRTIO_NET_HASH_TYPE_TCPv6 is set along with one of+ VIRTIO_NET_HASH_TYPE_GRE_INNER, VIRTIO_NET_HASH_TYPE_VXLAN_INNER and + VIRTIO_NET_HASH_TYPE_GENEVE_INNER, and the GRE, VXLAN or GENEVE-encapsulated + packet has an inner TCPv6 header in its payload, the hash is calculated over+ the following fields: + \begin{itemsize} + \item inner Source IPv6 address + \item inner Destination IPv6 address + \item inner Source TCP port + \item inner Destination TCP port + \end{itemsize} + \item Else if VIRTIO_NET_HASH_TYPE_UDPv6 is set along with one of+ VIRTIO_NET_HASH_TYPE_GRE_INNER, VIRTIO_NET_HASH_TYPE_VXLAN_INNER and + VIRTIO_NET_HASH_TYPE_GENEVE_INNER, and the GRE, VXLAN or GENEVE-encapsulated + packet has an inner UDPv6 header in its payload, the hash is calculated over+ the following fields: + \begin{itemsize} + \item inner Source IPv6 address + \item inner Destination IPv6 address + \item inner Source UDP port + \item inner Destination UDP port + \end{itemize} + \item Else if VIRTIO_NET_HASH_TYPE_IPv6 is set along with one of+ VIRTIO_NET_HASH_TYPE_GRE_INNER, VIRTIO_NET_HASH_TYPE_VXLAN_INNER and + VIRTIO_NET_HASH_TYPE_GENEVE_INNER, the hash is calculated over the+ following fields: + \begin{itemsize} + \item inner Source IPv6 address + \item inner Destination IPv6 address + \end{itemsize} + \item Else the device does not calculate the hash +\end{itemize} ++The device calculates the hash on GRE, VXLAN or GENEV-encapsulated packets whose inner +payloads are IPv6 packets with extension headers according to 'Enabled hash types'+bitmasks as follows: +\begin{itemsize} + \item If VIRTIO_NET_HASH_TYPE_TCP_EX is set along with one of+ VIRTIO_NET_HASH_TYPE_GRE_INNER, VIRTIO_NET_HASH_TYPE_VXLAN_INNER and + VIRTIO_NET_HASH_TYPE_GENEVE_INNER, and the GRE, VXLAN or GENEVE-encapsulated + packet has an inner TCPv6 header in its payload, the hash is calculated over+ the following fields: + \begin{itemize}+ \item Home address from the home address option in the inner IPv6 destination + options header. If the inner extension header is not present, use the+ inner Source IPv6 address.+ \item IPv6 address that is contained in the Routing-Header-Type-2 from the + associated inner extension header. If the inner extension header is not+ present, use the inner Destination IPv6 address. + \item inner Source TCP port + \item inner Destination TCP port + \end{itemize} + \item Else if VIRTIO_NET_HASH_TYPE_UDP_EX is set along with one of+ VIRTIO_NET_HASH_TYPE_GRE_INNER, VIRTIO_NET_HASH_TYPE_VXLAN_INNER and + VIRTIO_NET_HASH_TYPE_GENEVE_INNER, and the GRE, VXLAN or GENEVE-encapsulated + packet has an inner UDPv6 header in its payload, the hash is calculated over+ the following fields: + \begin{itemsize}+ \item Home address from the home address option in the inner IPv6 destination + options header. If the inner extension header is not present, use the+ inner Source IPv6 address.+ \item IPv6 address that is contained in the Routing-Header-Type-2 from the + associated inner extension header. If the inner extension header is not+ present, use the inner Destination IPv6 address. + \item inner Source UDP port + \item inner Destination UDP port + \end{itemize} + \item Else if VIRTIO_NET_HASH_TYPE_IP_EX is set along with one of+ VIRTIO_NET_HASH_TYPE_GRE_INNER, VIRTIO_NET_HASH_TYPE_VXLAN_INNER and + VIRTIO_NET_HASH_TYPE_GENEVE_INNER, the hash is calculated over the+ following fields: + \begin{itemsize}+ \item Home address from the home address option in the inner IPv6 destination + options header. If the inner extension header is not present, use the+ inner Source IPv6 address.+ \item IPv6 address that is contained in the Routing-Header-Type-2 from the + associated inner extension header. If the inner extension header is not+ present, use the inner Destination IPv6 address. + \end{itemize}+ \item Else skip inner IPv6 extension headers and calculate the hash as defined + for a GRE, VXLAN or GENEVE-encapsulated packet whose inner payload is an IPv6+ packet without extension headers +\end{itemsize} +  \paragraph{Hash reporting for incoming packets} \label{sec:Device Types / Network Device / Device Operation / Processing of Incoming Packets / Hash reporting for incoming packets}-If VIRTIO_NET_F_HASH_REPORT was negotiated and- the device has calculated the hash for the packet, the device fills \field{hash_report} with the report type of calculated hash-and \field{hash_value} with the value of calculated hash.+If VIRTIO_NET_F_HASH_REPORT was negotiated and the device has calculated the +hash for the packet, the device fills \field{hash_report} with the report type +of calculated hash, and \field{hash_value} with the value of calculated hash. +And if VIRTIO_NET_F_HASH_GRE_VXLAN_GENEVE_INNER was negotiated, the device also +needs to fills \field{hash_report_tunnel} with the type of the tunnel packet, If VIRTIO_NET_F_HASH_REPORT was negotiated but due to any reason the-hash was not calculated, the device sets \field{hash_report} to VIRTIO_NET_HASH_REPORT_NONE. +hash was not calculated, the device sets \field{hash_report} to VIRTIO_NET_HASH_REPORT_NONE,+and \field{hash_report_tunnel} is no longer valid. Possible values that the device can report in \field{hash_report} are defined below. They correspond to supported hash types defined in@@ -4005,6 +4159,24 @@ \subsubsection{Processing of Incoming Packets}\label{sec:Device Types / Network #define VIRTIO_NET_HASH_REPORT_UDPv6_EX 9  \end{lstlisting} +If \field{hash_report} differs from VIRTIO_NET_HASH_REPORT_NONE,+\field{hash_report_tunnel} can report the type of the tunnel-encapsulated+packet to the driver over the inner header hash calculation.+Possible values that the device can report in \field{hash_report_tunnel}+are defined below: + +\begin{lstlisting} +#define VIRTIO_NET_HASH_REPORT_GRE 1 +#define VIRTIO_NET_HASH_REPORT_VXLAN 2 +#define VIRTIO_NET_HASH_REPORT_GENEVE 3 +\end{lstlisting} ++The values VIRTIO_NET_HASH_REPORT_GRE, VIRTIO_NET_HASH_REPORT_VXLAN and +VIRTIO_NET_HASH_REPORT_GENEVE correspond to VIRTIO_NET_HASH_TYPE_GRE_INNER, +VIRTIO_NET_HASH_TYPE_VXLAN_INNER and VIRTIO_NET_HASH_TYPE_GENEVE_INNER bits+of supported hash types defined in respectively+\ref{sec:Device Types / Network Device / Device Operation / Processing of Incoming Packets / Hash calculation for incoming packets / Supported/enabled hash types}.+I think we can simply say "They correspond to supported hash types defined in \ref{xyz}" as what is done for non-tunnel hash types?Ok, I agree with you. Thanks.Thanks \subsubsection{Control Virtqueue}\label{sec:Device Types / Network Device / Device Operation / Control Virtqueue} The driver uses the control virtqueue (if VIRTIO_NET_F_CTRL_VQ is -- 2.19.1.6.gb485710bThis publicly archived list offers a means to provide input to the OASIS Virtual I/O Device (VIRTIO) TC. In order to verify user consent to the Feedback License terms and to minimize spam in the list archive, subscription is required before posting. Subscribe: virtio-comment-subscribe@lists.oasis-open.org Unsubscribe: virtio-comment-unsubscribe@lists.oasis-open.org List help: virtio-comment-help@lists.oasis-open.org List archive: https://lists.oasis-open.org/archives/virtio-comment/ Feedback License: https://www.oasis-open.org/who/ipr/feedback_license.pdfList Guidelines: https://www.oasis-open.org/policies-guidelines/mailing-listsCommittee: https://www.oasis-open.org/committees/virtio/ Join OASIS: https://www.oasis-open.org/join/
[Date Prev] | [Thread Prev] | [Thread Next] | [Date Next] -- [Date Index] | [Thread Index] | [List Home]