[dpdk-dev] [ovs-dev] [PATCH RFC] dpif-netdev: Add support Intel DPDK based ports.

[Pravin Shelar]

On Tue, Jan 28, 2014 at 7:48 AM, Thomas Graf <tgraf at redhat.com> wrote:
> On 01/28/2014 02:48 AM, pshelar at nicira.com wrote:
>>
>> From: Pravin B Shelar <pshelar at nicira.com>
>>
>> Following patch adds DPDK netdev-class to userspace datapath.
>> Approach taken in this patch differs from Intel® DPDK vSwitch
>> where DPDK datapath switching is done in saparate process.  This
>> patch adds support for DPDK type port and uses OVS userspace
>> datapath for switching.  Therefore all DPDK processing and flow
>> miss handling is done in single process.  This also avoids code
>> duplication by reusing OVS userspace datapath switching and
>> therefore it supports all flow matching and actions that
>> user-space datapath supports.  Refer to INSTALL.DPDK doc for
>> further info.
>>
>> With this patch I got similar performance for netperf TCP_STREAM
>> tests compared to kernel datapath.
>
>
> I'm happy to see this happen!
>
>
>
>> +static const struct rte_eth_conf port_conf = {
>> +        .rxmode = {
>> +                .mq_mode = ETH_MQ_RX_RSS,
>> +                .split_hdr_size = 0,
>> +                .header_split   = 0, /* Header Split disabled */
>> +                .hw_ip_checksum = 0, /* IP checksum offload enabled */
>> +                .hw_vlan_filter = 0, /* VLAN filtering disabled */
>> +                .jumbo_frame    = 0, /* Jumbo Frame Support disabled */
>> +                .hw_strip_crc   = 0, /* CRC stripped by hardware */
>> +        },
>> +        .rx_adv_conf = {
>> +                .rss_conf = {
>> +                        .rss_key = NULL,
>> +                        .rss_hf = ETH_RSS_IPV4_TCP | ETH_RSS_IPV4 |
>> ETH_RSS_IPV6,
>> +                },
>> +        },
>> +        .txmode = {
>> +                .mq_mode = ETH_MQ_TX_NONE,
>> +        },
>> +};
>
>
> I realize this is an RFC patch but I will ask anyway:
>
> What are the plans on managing runtime dependencies of a DPDK enabled OVS
> and DPDK itself? Will a OVS built against DPDK 1.5.2 work with
> drivers written for 1.5.3?
>
> Based on the above use of struct rte_eth_conf it would seem that once
> released, rte_eth_conf cannot be extended anymore without breaking
> ABI compatibility. The same applies to many of the other user
> structures. I see various structures changes between minor releases,
> for example dpdk.org ed2c69c3ef7 between 1.5.1 and 1.5.2.
>

Right, version mismatch will not work. API provided by DPDK are not
stable, So OVS has to be built for different releases for now.

I do not see how we can fix it from OVS side. DPDK needs to
standardize API, Actually OVS also needs more API, like DPDK
initialization, mempool destroy, etc.