From: "Xia, Chenbo" <chenbo.xia@intel.com>
To: Maxime Coquelin <maxime.coquelin@redhat.com>,
"dev@dpdk.org" <dev@dpdk.org>,
"jasowang@redhat.com" <jasowang@redhat.com>,
"david.marchand@redhat.com" <david.marchand@redhat.com>,
"Matz, Olivier" <olivier.matz@6wind.com>,
"Ma, WenwuX" <wenwux.ma@intel.com>,
"Zhang, Yuying" <yuying.zhang@intel.com>,
"Singh, Aman Deep" <aman.deep.singh@intel.com>,
"Jiang, Cheng1" <cheng1.jiang@intel.com>
Subject: RE: [PATCH v2 5/6] net/vhost: perform SW checksum in Rx path
Date: Fri, 10 Jun 2022 03:49:28 +0000 [thread overview]
Message-ID: <SN6PR11MB3504A7350CE14609C9B3DFB39CA69@SN6PR11MB3504.namprd11.prod.outlook.com> (raw)
In-Reply-To: <20220608124946.102623-6-maxime.coquelin@redhat.com>
+ Cheng for review
> -----Original Message-----
> From: Maxime Coquelin <maxime.coquelin@redhat.com>
> Sent: Wednesday, June 8, 2022 8:50 PM
> To: dev@dpdk.org; jasowang@redhat.com; Xia, Chenbo <chenbo.xia@intel.com>;
> david.marchand@redhat.com; Matz, Olivier <olivier.matz@6wind.com>; Ma,
> WenwuX <wenwux.ma@intel.com>; Zhang, Yuying <yuying.zhang@intel.com>;
> Singh, Aman Deep <aman.deep.singh@intel.com>
> Cc: Maxime Coquelin <maxime.coquelin@redhat.com>
> Subject: [PATCH v2 5/6] net/vhost: perform SW checksum in Rx path
>
> Virtio specification supports host checksum offloading
> for L4, which is enabled with VIRTIO_NET_F_CSUM feature
> negotiation. However, the Vhost PMD does not advertise
> Rx checksum offload capabilities, so we can end-up with
> the VIRTIO_NET_F_CSUM feature being negotiated, implying
> the Vhost library returns packets with checksum being
> offloaded while the application did not request for it.
>
> Advertising these offload capabilities at the ethdev level
> is not enough, because we could still end-up with the
> application not enabling these offloads while the guest
> still negotiate them.
>
> This patch advertises the Rx checksum offload capabilities,
> and introduces a compatibility layer to cover the case
> VIRTIO_NET_F_CSUM has been negotiated but the application
> does not configure the Rx checksum offloads. This function
> performis the L4 Rx checksum in SW for UDP and TCP. Note
> that it is not needed to calculate the pseudo-header
> checksum, because the Virtio specification requires that
> the driver do it.
>
> This patch does not advertise SCTP checksum offloading
> capability for now, but it could be handled later if the
> need arises.
>
> Reported-by: Jason Wang <jasowang@redhat.com>
> Signed-off-by: Maxime Coquelin <maxime.coquelin@redhat.com>
> ---
> doc/guides/nics/features/vhost.ini | 1 +
> drivers/net/vhost/rte_eth_vhost.c | 83 ++++++++++++++++++++++++++++++
> 2 files changed, 84 insertions(+)
>
> diff --git a/doc/guides/nics/features/vhost.ini
> b/doc/guides/nics/features/vhost.ini
> index ef81abb439..15f4dfe5e8 100644
> --- a/doc/guides/nics/features/vhost.ini
> +++ b/doc/guides/nics/features/vhost.ini
> @@ -7,6 +7,7 @@
> Link status = Y
> Free Tx mbuf on demand = Y
> Queue status event = Y
> +L4 checksum offload = P
> Basic stats = Y
> Extended stats = Y
> x86-32 = Y
> diff --git a/drivers/net/vhost/rte_eth_vhost.c
> b/drivers/net/vhost/rte_eth_vhost.c
> index e931d59053..42f0d52ebc 100644
> --- a/drivers/net/vhost/rte_eth_vhost.c
> +++ b/drivers/net/vhost/rte_eth_vhost.c
> @@ -12,6 +12,7 @@
> #include <ethdev_vdev.h>
> #include <rte_malloc.h>
> #include <rte_memcpy.h>
> +#include <rte_net.h>
> #include <rte_bus_vdev.h>
> #include <rte_kvargs.h>
> #include <rte_vhost.h>
> @@ -85,10 +86,12 @@ struct pmd_internal {
> char *iface_name;
> uint64_t flags;
> uint64_t disable_flags;
> + uint64_t features;
> uint16_t max_queues;
> int vid;
> rte_atomic32_t started;
> bool vlan_strip;
> + bool rx_sw_csum;
> };
>
> struct internal_list {
> @@ -275,6 +278,70 @@ vhost_dev_xstats_get(struct rte_eth_dev *dev, struct
> rte_eth_xstat *xstats,
> return nstats;
> }
>
> +static void
> +vhost_dev_csum_configure(struct rte_eth_dev *eth_dev)
> +{
> + struct pmd_internal *internal = eth_dev->data->dev_private;
> + const struct rte_eth_rxmode *rxmode = ð_dev->data-
> >dev_conf.rxmode;
> +
> + internal->rx_sw_csum = false;
> +
> + /* SW checksum is not compatible with legacy mode */
> + if (!(internal->flags & RTE_VHOST_USER_NET_COMPLIANT_OL_FLAGS))
> + return;
> +
> + if (internal->features & (1ULL << VIRTIO_NET_F_CSUM)) {
> + if (!(rxmode->offloads &
> + (RTE_ETH_RX_OFFLOAD_UDP_CKSUM |
> RTE_ETH_RX_OFFLOAD_TCP_CKSUM))) {
> + VHOST_LOG(NOTICE, "Rx csum will be done in SW, may
> impact performance.");
> + internal->rx_sw_csum = true;
> + }
> + }
> +}
> +
> +static void
> +vhost_dev_rx_sw_csum(struct rte_mbuf *mbuf)
> +{
> + struct rte_net_hdr_lens hdr_lens;
> + uint32_t ptype, hdr_len;
> + uint16_t csum = 0, csum_offset;
> +
> + /* Return early if the L4 checksum was not offloaded */
> + if ((mbuf->ol_flags & RTE_MBUF_F_RX_L4_CKSUM_MASK) !=
> RTE_MBUF_F_RX_L4_CKSUM_NONE)
> + return;
> +
> + ptype = rte_net_get_ptype(mbuf, &hdr_lens, RTE_PTYPE_ALL_MASK);
> +
> + hdr_len = hdr_lens.l2_len + hdr_lens.l3_len;
> +
> + switch (ptype & RTE_PTYPE_L4_MASK) {
> + case RTE_PTYPE_L4_TCP:
> + csum_offset = offsetof(struct rte_tcp_hdr, cksum) + hdr_len;
> + break;
> + case RTE_PTYPE_L4_UDP:
> + csum_offset = offsetof(struct rte_udp_hdr, dgram_cksum) +
> hdr_len;
> + break;
> + default:
> + /* Unsupported packet type */
> + return;
> + }
> +
> + /* The pseudo-header checksum is already performed, as per Virtio
> spec */
> + if (rte_raw_cksum_mbuf(mbuf, hdr_len, rte_pktmbuf_pkt_len(mbuf) -
> hdr_len, &csum) < 0)
> + return;
> +
> + csum = ~csum;
> + /* See RFC768 */
> + if (unlikely((ptype & RTE_PTYPE_L4_UDP) && csum == 0))
> + csum = 0xffff;
> +
> + if (rte_pktmbuf_data_len(mbuf) >= csum_offset + 1)
> + *rte_pktmbuf_mtod_offset(mbuf, uint16_t *, csum_offset) = csum;
> +
> + mbuf->ol_flags &= ~RTE_MBUF_F_RX_L4_CKSUM_MASK;
> + mbuf->ol_flags |= RTE_MBUF_F_RX_L4_CKSUM_GOOD;
> +}
> +
> static uint16_t
> eth_vhost_rx(void *q, struct rte_mbuf **bufs, uint16_t nb_bufs)
> {
> @@ -315,6 +382,9 @@ eth_vhost_rx(void *q, struct rte_mbuf **bufs, uint16_t
> nb_bufs)
> if (r->internal->vlan_strip)
> rte_vlan_strip(bufs[i]);
>
> + if (r->internal->rx_sw_csum)
> + vhost_dev_rx_sw_csum(bufs[i]);
> +
> r->stats.bytes += bufs[i]->pkt_len;
> }
>
> @@ -711,6 +781,11 @@ new_device(int vid)
> eth_dev->data->numa_node = newnode;
> #endif
>
> + if (rte_vhost_get_negotiated_features(vid, &internal->features)) {
> + VHOST_LOG(ERR, "Failed to get device features\n");
> + return -1;
> + }
> +
> internal->vid = vid;
> if (rte_atomic32_read(&internal->started) == 1) {
> queue_setup(eth_dev, internal);
> @@ -733,6 +808,8 @@ new_device(int vid)
>
> eth_dev->data->dev_link.link_status = RTE_ETH_LINK_UP;
>
> + vhost_dev_csum_configure(eth_dev);
> +
> rte_atomic32_set(&internal->dev_attached, 1);
> update_queuing_status(eth_dev);
>
> @@ -1039,6 +1116,8 @@ eth_dev_configure(struct rte_eth_dev *dev)
>
> internal->vlan_strip = !!(rxmode->offloads &
> RTE_ETH_RX_OFFLOAD_VLAN_STRIP);
>
> + vhost_dev_csum_configure(dev);
> +
> return 0;
> }
>
> @@ -1189,6 +1268,10 @@ eth_dev_info(struct rte_eth_dev *dev,
> dev_info->tx_offload_capa = RTE_ETH_TX_OFFLOAD_MULTI_SEGS |
> RTE_ETH_TX_OFFLOAD_VLAN_INSERT;
> dev_info->rx_offload_capa = RTE_ETH_RX_OFFLOAD_VLAN_STRIP;
> + if (internal->flags & RTE_VHOST_USER_NET_COMPLIANT_OL_FLAGS) {
> + dev_info->rx_offload_capa |= RTE_ETH_RX_OFFLOAD_UDP_CKSUM |
> + RTE_ETH_RX_OFFLOAD_TCP_CKSUM;
> + }
>
> return 0;
> }
> --
> 2.35.3
next prev parent reply other threads:[~2022-06-10 3:50 UTC|newest]
Thread overview: 17+ messages / expand[flat|nested] mbox.gz Atom feed top
2022-06-08 12:49 [PATCH v2 0/6] Vhost checksum offload improvements Maxime Coquelin
2022-06-08 12:49 ` [PATCH v2 1/6] Revert "app/testpmd: modify mac in csum forwarding" Maxime Coquelin
2022-06-13 12:24 ` Singh, Aman Deep
2022-06-17 12:42 ` Maxime Coquelin
2022-06-08 12:49 ` [PATCH v2 2/6] vhost: fix missing enqueue pseudo-header calculation Maxime Coquelin
2022-06-08 12:49 ` [PATCH v2 3/6] net/vhost: enable compliant offloading mode Maxime Coquelin
2022-06-08 12:49 ` [PATCH v2 4/6] net/vhost: make VLAN stripping flag a boolean Maxime Coquelin
2022-06-08 12:49 ` [PATCH v2 5/6] net/vhost: perform SW checksum in Rx path Maxime Coquelin
2022-06-09 1:59 ` Xia, Chenbo
2022-06-10 3:49 ` Xia, Chenbo [this message]
2022-06-10 7:19 ` Jiang, Cheng1
2022-06-08 12:49 ` [PATCH v2 6/6] net/vhost: perform SW checksum in Tx path Maxime Coquelin
2022-06-09 2:26 ` Xia, Chenbo
2022-06-10 3:49 ` Xia, Chenbo
2022-06-10 7:31 ` Jiang, Cheng1
2022-06-10 3:50 ` Xia, Chenbo
2022-06-17 14:06 ` [PATCH v2 0/6] Vhost checksum offload improvements Maxime Coquelin
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=SN6PR11MB3504A7350CE14609C9B3DFB39CA69@SN6PR11MB3504.namprd11.prod.outlook.com \
--to=chenbo.xia@intel.com \
--cc=aman.deep.singh@intel.com \
--cc=cheng1.jiang@intel.com \
--cc=david.marchand@redhat.com \
--cc=dev@dpdk.org \
--cc=jasowang@redhat.com \
--cc=maxime.coquelin@redhat.com \
--cc=olivier.matz@6wind.com \
--cc=wenwux.ma@intel.com \
--cc=yuying.zhang@intel.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).