From: Jiayu Hu <jiayu.hu@intel.com>
To: "Ananyev, Konstantin" <konstantin.ananyev@intel.com>
Cc: "dev@dpdk.org" <dev@dpdk.org>,
"Kavanagh, Mark B" <mark.b.kavanagh@intel.com>,
"Tan, Jianfeng" <jianfeng.tan@intel.com>
Subject: Re: [dpdk-dev] [PATCH v3 2/5] gso: add TCP/IPv4 GSO support
Date: Wed, 13 Sep 2017 10:48:01 +0800 [thread overview]
Message-ID: <20170913024801.GB44293@dpdk15.sh.intel.com> (raw)
In-Reply-To: <2601191342CEEE43887BDE71AB9772584F249E8E@irsmsx105.ger.corp.intel.com>
Hi Konstantin,
On Tue, Sep 12, 2017 at 07:17:49PM +0800, Ananyev, Konstantin wrote:
> Hi Jayu,
>
> > -----Original Message-----
> > From: Hu, Jiayu
> > Sent: Tuesday, September 12, 2017 3:43 AM
> > To: dev@dpdk.org
> > Cc: Ananyev, Konstantin <konstantin.ananyev@intel.com>; Kavanagh, Mark B <mark.b.kavanagh@intel.com>; Tan, Jianfeng
> > <jianfeng.tan@intel.com>; Hu, Jiayu <jiayu.hu@intel.com>
> > Subject: [PATCH v3 2/5] gso: add TCP/IPv4 GSO support
> >
> > This patch adds GSO support for TCP/IPv4 packets. Supported packets
> > may include a single VLAN tag. TCP/IPv4 GSO assumes that all input
> > packets have correct checksums, and doesn't update checksums for output
> > packets (the responsibility for this lies with the application).
>
> Probably it shouldn't say that checksum have to be valid, right?
> As you don't update checksum(s) inside the lib - it probably doesn't matter.
Yes, you are right. It's better to use:
"TCP/IPv4 GSO doesn't check if checksums are correct and doesn't update
checksums for output packets".
>
> > Additionally, TCP/IPv4 GSO doesn't process IP fragmented packets.
> >
> > TCP/IPv4 GSO uses two chained MBUFs, one direct MBUF and one indrect
> > MBUF, to organize an output packet. Note that we refer to these two
> > chained MBUFs as a two-segment MBUF. The direct MBUF stores the packet
> > header, while the indirect mbuf simply points to a location within the
> > original packet's payload. Consequently, use of the GSO library requires
> > multi-segment MBUF support in the TX functions of the NIC driver.
> >
> > If a packet is GSOed, TCP/IPv4 GSO reduces its MBUF refcnt by 1. As a
> > result, when all of its GSOed segments are freed, the packet is freed
> > automatically.
> >
> > Signed-off-by: Jiayu Hu <jiayu.hu@intel.com>
> > Signed-off-by: Mark Kavanagh <mark.b.kavanagh@intel.com>
> > ---
> > lib/librte_eal/common/include/rte_log.h | 1 +
> > lib/librte_gso/Makefile | 2 +
> > lib/librte_gso/gso_common.c | 202 ++++++++++++++++++++++++++++++++
> > lib/librte_gso/gso_common.h | 113 ++++++++++++++++++
> > lib/librte_gso/gso_tcp4.c | 83 +++++++++++++
> > lib/librte_gso/gso_tcp4.h | 76 ++++++++++++
> > lib/librte_gso/rte_gso.c | 41 ++++++-
> > 7 files changed, 515 insertions(+), 3 deletions(-)
> > create mode 100644 lib/librte_gso/gso_common.c
> > create mode 100644 lib/librte_gso/gso_common.h
> > create mode 100644 lib/librte_gso/gso_tcp4.c
> > create mode 100644 lib/librte_gso/gso_tcp4.h
> >
> > diff --git a/lib/librte_eal/common/include/rte_log.h b/lib/librte_eal/common/include/rte_log.h
> > index ec8dba7..2fa1199 100644
> > --- a/lib/librte_eal/common/include/rte_log.h
> > +++ b/lib/librte_eal/common/include/rte_log.h
> > @@ -87,6 +87,7 @@ extern struct rte_logs rte_logs;
> > #define RTE_LOGTYPE_CRYPTODEV 17 /**< Log related to cryptodev. */
> > #define RTE_LOGTYPE_EFD 18 /**< Log related to EFD. */
> > #define RTE_LOGTYPE_EVENTDEV 19 /**< Log related to eventdev. */
> > +#define RTE_LOGTYPE_GSO 20 /**< Log related to GSO. */
> >
> > /* these log types can be used in an application */
> > #define RTE_LOGTYPE_USER1 24 /**< User-defined log type 1. */
> > diff --git a/lib/librte_gso/Makefile b/lib/librte_gso/Makefile
> > index aeaacbc..2be64d1 100644
> > --- a/lib/librte_gso/Makefile
> > +++ b/lib/librte_gso/Makefile
> > @@ -42,6 +42,8 @@ LIBABIVER := 1
> >
> > #source files
> > SRCS-$(CONFIG_RTE_LIBRTE_GSO) += rte_gso.c
> > +SRCS-$(CONFIG_RTE_LIBRTE_GSO) += gso_common.c
> > +SRCS-$(CONFIG_RTE_LIBRTE_GSO) += gso_tcp4.c
> >
> > # install this header file
> > SYMLINK-$(CONFIG_RTE_LIBRTE_GSO)-include += rte_gso.h
> > diff --git a/lib/librte_gso/gso_common.c b/lib/librte_gso/gso_common.c
> > new file mode 100644
> > index 0000000..7c32e03
> > --- /dev/null
> > +++ b/lib/librte_gso/gso_common.c
> > @@ -0,0 +1,202 @@
> > +/*-
> > + * BSD LICENSE
> > + *
> > + * Copyright(c) 2017 Intel Corporation. All rights reserved.
> > + * All rights reserved.
> > + *
> > + * Redistribution and use in source and binary forms, with or without
> > + * modification, are permitted provided that the following conditions
> > + * are met:
> > + *
> > + * * Redistributions of source code must retain the above copyright
> > + * notice, this list of conditions and the following disclaimer.
> > + * * Redistributions in binary form must reproduce the above copyright
> > + * notice, this list of conditions and the following disclaimer in
> > + * the documentation and/or other materials provided with the
> > + * distribution.
> > + * * Neither the name of Intel Corporation nor the names of its
> > + * contributors may be used to endorse or promote products derived
> > + * from this software without specific prior written permission.
> > + *
> > + * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
> > + * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
> > + * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
> > + * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
> > + * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
> > + * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
> > + * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
> > + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
> > + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
> > + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
> > + * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
> > + */
> > +
> > +#include <stdbool.h>
> > +#include <errno.h>
> > +
> > +#include <rte_memcpy.h>
> > +#include <rte_mempool.h>
> > +#include <rte_ether.h>
> > +#include <rte_ip.h>
> > +#include <rte_tcp.h>
> > +
> > +#include "gso_common.h"
> > +
> > +static inline void
> > +hdr_segment_init(struct rte_mbuf *hdr_segment, struct rte_mbuf *pkt,
> > + uint16_t pkt_hdr_offset)
> > +{
> > + /* Copy MBUF metadata */
> > + hdr_segment->nb_segs = 1;
> > + hdr_segment->port = pkt->port;
> > + hdr_segment->ol_flags = pkt->ol_flags;
> > + hdr_segment->packet_type = pkt->packet_type;
> > + hdr_segment->pkt_len = pkt_hdr_offset;
> > + hdr_segment->data_len = pkt_hdr_offset;
> > + hdr_segment->tx_offload = pkt->tx_offload;
> > +
> > + /* Copy the packet header */
> > + rte_memcpy(rte_pktmbuf_mtod(hdr_segment, char *),
> > + rte_pktmbuf_mtod(pkt, char *),
> > + pkt_hdr_offset);
> > +}
> > +
> > +static inline void
> > +free_gso_segment(struct rte_mbuf **pkts, uint16_t nb_pkts)
> > +{
> > + uint16_t i;
> > +
> > + for (i = 0; i < nb_pkts; i++)
> > + rte_pktmbuf_free(pkts[i]);
> > +}
> > +
> > +int
> > +gso_do_segment(struct rte_mbuf *pkt,
> > + uint16_t pkt_hdr_offset,
> > + uint16_t pyld_unit_size,
> > + struct rte_mempool *direct_pool,
> > + struct rte_mempool *indirect_pool,
> > + struct rte_mbuf **pkts_out,
> > + uint16_t nb_pkts_out)
> > +{
> > + struct rte_mbuf *pkt_in;
> > + struct rte_mbuf *hdr_segment, *pyld_segment, *prev_segment;
> > + uint16_t pkt_in_data_pos, segment_bytes_remaining;
> > + uint16_t pyld_len, nb_segs;
> > + bool more_in_pkt, more_out_segs;
> > +
> > + pkt_in = pkt;
> > + nb_segs = 0;
> > + more_in_pkt = 1;
> > + pkt_in_data_pos = pkt_hdr_offset;
> > +
> > + while (more_in_pkt) {
> > + if (unlikely(nb_segs >= nb_pkts_out)) {
> > + free_gso_segment(pkts_out, nb_segs);
> > + return -EINVAL;
> > + }
> > +
> > + /* Allocate a direct MBUF */
> > + hdr_segment = rte_pktmbuf_alloc(direct_pool);
> > + if (unlikely(hdr_segment == NULL)) {
> > + free_gso_segment(pkts_out, nb_segs);
> > + return -ENOMEM;
> > + }
> > + /* Fill the packet header */
> > + hdr_segment_init(hdr_segment, pkt, pkt_hdr_offset);
> > +
> > + prev_segment = hdr_segment;
> > + segment_bytes_remaining = pyld_unit_size;
> > + more_out_segs = 1;
> > +
> > + while (more_out_segs && more_in_pkt) {
> > + /* Allocate an indirect MBUF */
> > + pyld_segment = rte_pktmbuf_alloc(indirect_pool);
> > + if (unlikely(pyld_segment == NULL)) {
> > + rte_pktmbuf_free(hdr_segment);
> > + free_gso_segment(pkts_out, nb_segs);
> > + return -ENOMEM;
> > + }
> > + /* Attach to current MBUF segment of pkt */
> > + rte_pktmbuf_attach(pyld_segment, pkt_in);
> > +
> > + prev_segment->next = pyld_segment;
> > + prev_segment = pyld_segment;
> > +
> > + pyld_len = segment_bytes_remaining;
> > + if (pyld_len + pkt_in_data_pos > pkt_in->data_len)
> > + pyld_len = pkt_in->data_len - pkt_in_data_pos;
> > +
> > + pyld_segment->data_off = pkt_in_data_pos +
> > + pkt_in->data_off;
> > + pyld_segment->data_len = pyld_len;
> > +
> > + /* Update header segment */
> > + hdr_segment->pkt_len += pyld_len;
> > + hdr_segment->nb_segs++;
> > +
> > + pkt_in_data_pos += pyld_len;
> > + segment_bytes_remaining -= pyld_len;
> > +
> > + /* Finish processing a MBUF segment of pkt */
> > + if (pkt_in_data_pos == pkt_in->data_len) {
> > + pkt_in = pkt_in->next;
> > + pkt_in_data_pos = 0;
> > + if (pkt_in == NULL)
> > + more_in_pkt = 0;
> > + }
> > +
> > + /* Finish generating a GSO segment */
> > + if (segment_bytes_remaining == 0)
> > + more_out_segs = 0;
> > + }
> > + pkts_out[nb_segs++] = hdr_segment;
> > + }
> > + return nb_segs;
> > +}
> > +
> > +static inline void
> > +update_inner_tcp4_header(struct rte_mbuf *pkt, uint8_t ipid_delta,
> > + struct rte_mbuf **segs, uint16_t nb_segs)
> > +{
> > + struct tcp_hdr *tcp_hdr;
> > + struct ipv4_hdr *ipv4_hdr;
> > + struct rte_mbuf *seg;
> > + uint32_t sent_seq;
> > + uint16_t inner_l2_offset;
> > + uint16_t id, i;
> > +
> > + inner_l2_offset = pkt->outer_l2_len + pkt->outer_l3_len + pkt->l2_len;
>
> Shouldn't it be: pkt->l2_len here?
> Or probably even better to pass l2_len as an input parameter.
Oh, yes. Applications won't guarantee outer_l2_len and outer_l3_len are 0
for non-tunnelling packets. I will add l2_len as a parameter instead.
>
> > + ipv4_hdr = (struct ipv4_hdr *)(rte_pktmbuf_mtod(pkt, char *) +
> > + inner_l2_offset);
> > + tcp_hdr = (struct tcp_hdr *)((char *)ipv4_hdr + pkt->l3_len);
> > + id = rte_be_to_cpu_16(ipv4_hdr->packet_id);
> > + sent_seq = rte_be_to_cpu_32(tcp_hdr->sent_seq);
> > +
> > + for (i = 0; i < nb_segs; i++) {
> > + seg = segs[i];
> > + /* Update the inner IPv4 header */
> > + ipv4_hdr = (struct ipv4_hdr *)(rte_pktmbuf_mtod(seg, char *) +
> > + inner_l2_offset);
> > + ipv4_hdr->total_length = rte_cpu_to_be_16(seg->pkt_len -
> > + inner_l2_offset);
> > + ipv4_hdr->packet_id = rte_cpu_to_be_16(id);
> > + id += ipid_delta;
> > +
> > + /* Update the inner TCP header */
> > + tcp_hdr = (struct tcp_hdr *)((char *)ipv4_hdr + seg->l3_len);
> > + tcp_hdr->sent_seq = rte_cpu_to_be_32(sent_seq);
> > + if (likely(i < nb_segs - 1))
> > + tcp_hdr->tcp_flags &= (~(TCP_HDR_PSH_MASK |
> > + TCP_HDR_FIN_MASK));
> > + sent_seq += (seg->pkt_len - seg->data_len);
> > + }
> > +}
> > +
> > +void
> > +gso_update_pkt_headers(struct rte_mbuf *pkt, uint8_t ipid_delta,
> > + struct rte_mbuf **segs, uint16_t nb_segs)
> > +{
> > + if (is_ipv4_tcp(pkt->packet_type))
> > + update_inner_tcp4_header(pkt, ipid_delta, segs, nb_segs);
> > +}
> > diff --git a/lib/librte_gso/gso_common.h b/lib/librte_gso/gso_common.h
> > new file mode 100644
> > index 0000000..3c76520
> > --- /dev/null
> > +++ b/lib/librte_gso/gso_common.h
> > @@ -0,0 +1,113 @@
> > +/*-
> > + * BSD LICENSE
> > + *
> > + * Copyright(c) 2017 Intel Corporation. All rights reserved.
> > + * All rights reserved.
> > + *
> > + * Redistribution and use in source and binary forms, with or without
> > + * modification, are permitted provided that the following conditions
> > + * are met:
> > + *
> > + * * Redistributions of source code must retain the above copyright
> > + * notice, this list of conditions and the following disclaimer.
> > + * * Redistributions in binary form must reproduce the above copyright
> > + * notice, this list of conditions and the following disclaimer in
> > + * the documentation and/or other materials provided with the
> > + * distribution.
> > + * * Neither the name of Intel Corporation nor the names of its
> > + * contributors may be used to endorse or promote products derived
> > + * from this software without specific prior written permission.
> > + *
> > + * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
> > + * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
> > + * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
> > + * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
> > + * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
> > + * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
> > + * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
> > + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
> > + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
> > + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
> > + * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
> > + */
> > +
> > +#ifndef _GSO_COMMON_H_
> > +#define _GSO_COMMON_H_
> > +
> > +#include <stdint.h>
> > +#include <rte_mbuf.h>
> > +
> > +#define IPV4_HDR_DF_SHIFT 14
>
> We have that already defined in librte_net/rte_ip.h
Yes. I will remove it here.
>
> > +#define IPV4_HDR_DF_MASK (1 << IPV4_HDR_DF_SHIFT)
> > +
> > +#define TCP_HDR_PSH_MASK ((uint8_t)0x08)
> > +#define TCP_HDR_FIN_MASK ((uint8_t)0x01)
> > +
> > +#define ETHER_TCP_PKT (RTE_PTYPE_L2_ETHER | RTE_PTYPE_L4_TCP)
> > +#define ETHER_VLAN_TCP_PKT (RTE_PTYPE_L2_ETHER_VLAN | RTE_PTYPE_L4_TCP)
> > +static inline uint8_t is_ipv4_tcp(uint32_t ptype)
> > +{
> > + switch (ptype & (~RTE_PTYPE_L3_MASK)) {
> > + case ETHER_VLAN_TCP_PKT:
> > + case ETHER_TCP_PKT:
>
> Why not just:
> return RTE_ETH_IS_IPV4_HDR(ptype) && (ptype & RTE_PTYPE_L4_MASK) == RTE_PTYPE_L4_TCP;
> ?
Yes, we don't need to check if the packet is vlan encapsulated.
>
> > + return RTE_ETH_IS_IPV4_HDR(ptype);
> > + default:
> > + return 0;
> > + }
> > +}
> > +
> > +/**
> > + * Internal function which updates relevant packet headers, following
> > + * segmentation. This is required to update, for example, the IPv4
> > + * 'total_length' field, to reflect the reduced length of the now-
> > + * segmented packet.
> > + *
> > + * @param pkt
> > + * The original packet.
> > + * @param ipid_delta
> > + * The increasing uint of IP ids.
> > + * @param segs
> > + * Pointer array used for storing mbuf addresses for GSO segments.
> > + * @param nb_segs
> > + * The number of GSO segments placed in segs.
> > + */
> > +void gso_update_pkt_headers(struct rte_mbuf *pkt, uint8_t ipid_delta,
> > + struct rte_mbuf **segs, uint16_t nb_segs);
> > +
> > +/**
> > + * Internal function which divides the input packet into small segments.
> > + * Each of the newly-created segments is organized as a two-segment MBUF,
> > + * where the first segment is a standard mbuf, which stores a copy of
> > + * packet header, and the second is an indirect mbuf which points to a
> > + * section of data in the input packet.
> > + *
> > + * @param pkt
> > + * Packet to segment.
> > + * @param pkt_hdr_offset
> > + * Packet header offset, measured in bytes.
> > + * @param pyld_unit_size
> > + * The max payload length of a GSO segment.
> > + * @param direct_pool
> > + * MBUF pool used for allocating direct buffers for output segments.
> > + * @param indirect_pool
> > + * MBUF pool used for allocating indirect buffers for output segments.
> > + * @param pkts_out
> > + * Pointer array used to keep the mbuf addresses of output segments. If
> > + * the memory space in pkts_out is insufficient, gso_do_segment() fails
> > + * and returns -EINVAL.
> > + * @param nb_pkts_out
> > + * The max number of items that pkts_out can keep.
> > + *
> > + * @return
> > + * - The number of segments created in the event of success.
> > + * - Return -ENOMEM if run out of memory in MBUF pools.
> > + * - Return -EINVAL for invalid parameters.
> > + */
> > +int gso_do_segment(struct rte_mbuf *pkt,
> > + uint16_t pkt_hdr_offset,
> > + uint16_t pyld_unit_size,
> > + struct rte_mempool *direct_pool,
> > + struct rte_mempool *indirect_pool,
> > + struct rte_mbuf **pkts_out,
> > + uint16_t nb_pkts_out);
> > +#endif
> > diff --git a/lib/librte_gso/gso_tcp4.c b/lib/librte_gso/gso_tcp4.c
> > new file mode 100644
> > index 0000000..8d4bfb2
> > --- /dev/null
> > +++ b/lib/librte_gso/gso_tcp4.c
> > @@ -0,0 +1,83 @@
> > +/*-
> > + * BSD LICENSE
> > + *
> > + * Copyright(c) 2017 Intel Corporation. All rights reserved.
> > + * All rights reserved.
> > + *
> > + * Redistribution and use in source and binary forms, with or without
> > + * modification, are permitted provided that the following conditions
> > + * are met:
> > + *
> > + * * Redistributions of source code must retain the above copyright
> > + * notice, this list of conditions and the following disclaimer.
> > + * * Redistributions in binary form must reproduce the above copyright
> > + * notice, this list of conditions and the following disclaimer in
> > + * the documentation and/or other materials provided with the
> > + * distribution.
> > + * * Neither the name of Intel Corporation nor the names of its
> > + * contributors may be used to endorse or promote products derived
> > + * from this software without specific prior written permission.
> > + *
> > + * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
> > + * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
> > + * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
> > + * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
> > + * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
> > + * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
> > + * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
> > + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
> > + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
> > + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
> > + * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
> > + */
> > +
> > +
> > +#include <rte_ether.h>
> > +#include <rte_ip.h>
> > +
> > +#include "gso_common.h"
> > +#include "gso_tcp4.h"
> > +
> > +int
> > +gso_tcp4_segment(struct rte_mbuf *pkt,
> > + uint16_t gso_size,
> > + uint8_t ipid_delta,
> > + struct rte_mempool *direct_pool,
> > + struct rte_mempool *indirect_pool,
> > + struct rte_mbuf **pkts_out,
> > + uint16_t nb_pkts_out)
> > +{
> > + struct ipv4_hdr *ipv4_hdr;
> > + uint16_t tcp_dl;
> > + uint16_t pyld_unit_size;
> > + uint16_t hdr_offset;
> > + int ret = 1;
> > +
> > + ipv4_hdr = (struct ipv4_hdr *)(rte_pktmbuf_mtod(pkt, char *) +
> > + pkt->l2_len);
> > + /* Don't process the fragmented packet */
> > + if (unlikely((ipv4_hdr->fragment_offset & rte_cpu_to_be_16(
> > + IPV4_HDR_DF_MASK)) == 0)) {
>
>
> It is not a check for fragmented packet - it is a check that fragmentation is allowed for that packet.
> Should be IPV4_HDR_DF_MASK - 1, I think.
IMO, IPV4_HDR_DF_MASK whose value is (1 << 14) is used to get DF bit. It's a
little-endian value. But ipv4_hdr->fragment_offset is big-endian order.
So the value of DF bit should be "ipv4_hdr->fragment_offset & rte_cpu_to_be_16(
IPV4_HDR_DF_MASK)". If this value is 0, the input packet is fragmented.
>
> > + pkts_out[0] = pkt;
> > + return ret;
> > + }
> > +
> > + tcp_dl = rte_be_to_cpu_16(ipv4_hdr->total_length) - pkt->l3_len -
> > + pkt->l4_len;
>
> Why not use pkt->pkt_len - pkt->l2_len -pkt_l3_len - pkt_l4_len?
Yes, we can use pkt->pkt_len - pkt->l2_len -pkt_l3_len - pkt_l4_len here.
>
> > + /* Don't process the packet without data */
> > + if (unlikely(tcp_dl == 0)) {
> > + pkts_out[0] = pkt;
> > + return ret;
> > + }
> > +
> > + hdr_offset = pkt->l2_len + pkt->l3_len + pkt->l4_len;
> > + pyld_unit_size = gso_size - hdr_offset - ETHER_CRC_LEN;
>
> Hmm, why do we need to count CRC_LEN here?
Yes, we shouldn't count ETHER_CRC_LEN here. Its length should be
included in gso_size.
>
> > +
> > + /* Segment the payload */
> > + ret = gso_do_segment(pkt, hdr_offset, pyld_unit_size, direct_pool,
> > + indirect_pool, pkts_out, nb_pkts_out);
> > + if (ret > 1)
> > + gso_update_pkt_headers(pkt, ipid_delta, pkts_out, ret);
> > +
> > + return ret;
> > +}
> > diff --git a/lib/librte_gso/gso_tcp4.h b/lib/librte_gso/gso_tcp4.h
> > new file mode 100644
> > index 0000000..9c07984
> > --- /dev/null
> > +++ b/lib/librte_gso/gso_tcp4.h
> > @@ -0,0 +1,76 @@
> > +/*-
> > + * BSD LICENSE
> > + *
> > + * Copyright(c) 2017 Intel Corporation. All rights reserved.
> > + * All rights reserved.
> > + *
> > + * Redistribution and use in source and binary forms, with or without
> > + * modification, are permitted provided that the following conditions
> > + * are met:
> > + *
> > + * * Redistributions of source code must retain the above copyright
> > + * notice, this list of conditions and the following disclaimer.
> > + * * Redistributions in binary form must reproduce the above copyright
> > + * notice, this list of conditions and the following disclaimer in
> > + * the documentation and/or other materials provided with the
> > + * distribution.
> > + * * Neither the name of Intel Corporation nor the names of its
> > + * contributors may be used to endorse or promote products derived
> > + * from this software without specific prior written permission.
> > + *
> > + * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
> > + * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
> > + * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
> > + * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
> > + * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
> > + * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
> > + * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
> > + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
> > + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
> > + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
> > + * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
> > + */
> > +
> > +#ifndef _GSO_TCP4_H_
> > +#define _GSO_TCP4_H_
> > +
> > +#include <stdint.h>
> > +#include <rte_mbuf.h>
> > +
> > +/**
> > + * Segment an IPv4/TCP packet. This function assumes the input packet has
> > + * correct checksums and doesn't update checksums for GSO segment.
> > + * Furthermore, it doesn't process IP fragment packets.
> > + *
> > + * @param pkt
> > + * The packet mbuf to segment.
> > + * @param gso_size
> > + * The max length of a GSO segment, measured in bytes.
> > + * @param ipid_delta
> > + * The increasing uint of IP ids.
> > + * @param direct_pool
> > + * MBUF pool used for allocating direct buffers for output segments.
> > + * @param indirect_pool
> > + * MBUF pool used for allocating indirect buffers for output segments.
> > + * @param pkts_out
> > + * Pointer array used to store the MBUF addresses of output GSO
> > + * segments, when gso_tcp4_segment() successes. If the memory space in
> > + * pkts_out is insufficient, gso_tcp4_segment() fails and returns
> > + * -EINVAL.
> > + * @param nb_pkts_out
> > + * The max number of items that 'pkts_out' can keep.
> > + *
> > + * @return
> > + * - The number of GSO segments filled in pkts_out on success.
> > + * - Return -ENOMEM if run out of memory in MBUF pools.
> > + * - Return -EINVAL for invalid parameters.
> > + */
> > +int gso_tcp4_segment(struct rte_mbuf *pkt,
> > + uint16_t gso_size,
> > + uint8_t ip_delta,
> > + struct rte_mempool *direct_pool,
> > + struct rte_mempool *indirect_pool,
> > + struct rte_mbuf **pkts_out,
> > + uint16_t nb_pkts_out);
> > +
> > +#endif
> > diff --git a/lib/librte_gso/rte_gso.c b/lib/librte_gso/rte_gso.c
> > index dda50ee..95f6ea6 100644
> > --- a/lib/librte_gso/rte_gso.c
> > +++ b/lib/librte_gso/rte_gso.c
> > @@ -33,18 +33,53 @@
> >
> > #include <errno.h>
> >
> > +#include <rte_log.h>
> > +
> > #include "rte_gso.h"
> > +#include "gso_common.h"
> > +#include "gso_tcp4.h"
> >
> > int
> > rte_gso_segment(struct rte_mbuf *pkt,
> > - struct rte_gso_ctx gso_ctx __rte_unused,
> > + struct rte_gso_ctx gso_ctx,
> > struct rte_mbuf **pkts_out,
> > uint16_t nb_pkts_out)
> > {
> > + struct rte_mempool *direct_pool, *indirect_pool;
> > + struct rte_mbuf *pkt_seg;
> > + uint16_t gso_size;
> > + uint8_t ipid_delta;
> > + int ret = 1;
> > +
> > if (pkt == NULL || pkts_out == NULL || nb_pkts_out < 1)
> > return -EINVAL;
> >
> > - pkts_out[0] = pkt;
> > + if (gso_ctx.gso_size >= pkt->pkt_len ||
> > + (pkt->packet_type & gso_ctx.gso_types) !=
> > + pkt->packet_type) {
> > + pkts_out[0] = pkt;
> > + return ret;
> > + }
> > +
> > + direct_pool = gso_ctx.direct_pool;
> > + indirect_pool = gso_ctx.indirect_pool;
> > + gso_size = gso_ctx.gso_size;
> > + ipid_delta = gso_ctx.ipid_flag == RTE_GSO_IPID_INCREASE;
> > +
> > + if (is_ipv4_tcp(pkt->packet_type)) {
>
> Probably we need here:
> If (is_ipv4_tcp(pkt->packet_type) && (gso_ctx->gso_types & DEV_TX_OFFLOAD_TCP_TSO) != 0) {...
>
> > + ret = gso_tcp4_segment(pkt, gso_size, ipid_delta,
> > + direct_pool, indirect_pool,
> > + pkts_out, nb_pkts_out);
> > + } else
> > + RTE_LOG(WARNING, GSO, "Unsupported packet type\n");
>
> Shouldn't we do pkt_out[0] = pkt; here?
Yes, we need to add it here. Thanks for reminder.
>
> > +
> > + if (ret > 1) {
> > + pkt_seg = pkt;
> > + while (pkt_seg) {
> > + rte_mbuf_refcnt_update(pkt_seg, -1);
> > + pkt_seg = pkt_seg->next;
> > + }
> > + }
> >
> > - return 1;
> > + return ret;
> > }
> > --
> > 2.7.4
next prev parent reply other threads:[~2017-09-13 2:45 UTC|newest]
Thread overview: 157+ messages / expand[flat|nested] mbox.gz Atom feed top
2017-08-24 14:15 [dpdk-dev] [PATCH 0/5] Support TCP/IPv4, VxLAN and GRE GSO in DPDK Jiayu Hu
2017-08-24 14:15 ` [dpdk-dev] [PATCH 1/5] lib: add Generic Segmentation Offload API framework Jiayu Hu
2017-08-30 1:38 ` Ananyev, Konstantin
2017-08-30 7:57 ` Jiayu Hu
2017-08-24 14:15 ` [dpdk-dev] [PATCH 2/5] gso/lib: add TCP/IPv4 GSO support Jiayu Hu
2017-08-30 1:38 ` Ananyev, Konstantin
2017-08-30 2:55 ` Jiayu Hu
2017-08-30 9:25 ` Kavanagh, Mark B
2017-08-30 9:39 ` Ananyev, Konstantin
2017-08-30 9:59 ` Ananyev, Konstantin
2017-08-30 13:27 ` Kavanagh, Mark B
2017-08-30 9:03 ` Jiayu Hu
2017-09-04 3:31 ` Jiayu Hu
2017-09-04 9:54 ` Ananyev, Konstantin
2017-09-05 1:09 ` Hu, Jiayu
2017-09-11 13:04 ` Ananyev, Konstantin
2017-08-24 14:15 ` [dpdk-dev] [PATCH 3/5] lib/gso: add VxLAN " Jiayu Hu
2017-08-24 14:15 ` [dpdk-dev] [PATCH 4/5] lib/gso: add GRE " Jiayu Hu
2017-08-24 14:15 ` [dpdk-dev] [PATCH 5/5] app/testpmd: enable TCP/IPv4, VxLAN and GRE GSO Jiayu Hu
2017-08-30 1:37 ` [dpdk-dev] [PATCH 0/5] Support TCP/IPv4, VxLAN and GRE GSO in DPDK Ananyev, Konstantin
2017-08-30 7:36 ` Jiayu Hu
2017-08-30 10:49 ` Ananyev, Konstantin
2017-08-30 13:32 ` Kavanagh, Mark B
2017-09-05 7:57 ` [dpdk-dev] [PATCH v2 " Jiayu Hu
2017-09-05 7:57 ` [dpdk-dev] [PATCH v2 1/5] gso: add Generic Segmentation Offload API framework Jiayu Hu
2017-09-05 7:57 ` [dpdk-dev] [PATCH v2 2/5] gso: add TCP/IPv4 GSO support Jiayu Hu
2017-09-05 7:57 ` [dpdk-dev] [PATCH v2 3/5] gso: add VxLAN " Jiayu Hu
2017-09-05 7:57 ` [dpdk-dev] [PATCH v2 4/5] gso: add GRE " Jiayu Hu
2017-09-05 7:57 ` [dpdk-dev] [PATCH v2 5/5] app/testpmd: enable TCP/IPv4, VxLAN and GRE GSO Jiayu Hu
2017-09-12 2:43 ` [dpdk-dev] [PATCH v3 0/5] Support TCP/IPv4, VxLAN and GRE GSO in DPDK Jiayu Hu
2017-09-12 2:43 ` [dpdk-dev] [PATCH v3 1/5] gso: add Generic Segmentation Offload API framework Jiayu Hu
2017-09-12 10:36 ` Ananyev, Konstantin
2017-09-13 2:11 ` Jiayu Hu
2017-09-14 18:33 ` Ferruh Yigit
2017-09-15 1:12 ` Hu, Jiayu
2017-09-12 2:43 ` [dpdk-dev] [PATCH v3 2/5] gso: add TCP/IPv4 GSO support Jiayu Hu
2017-09-12 11:17 ` Ananyev, Konstantin
2017-09-13 2:48 ` Jiayu Hu [this message]
2017-09-13 9:38 ` Ananyev, Konstantin
2017-09-13 10:23 ` Hu, Jiayu
2017-09-13 14:52 ` Kavanagh, Mark B
2017-09-13 15:13 ` Ananyev, Konstantin
2017-09-14 0:59 ` Hu, Jiayu
2017-09-14 8:35 ` Kavanagh, Mark B
2017-09-14 8:39 ` Ananyev, Konstantin
2017-09-14 9:00 ` Kavanagh, Mark B
2017-09-14 9:10 ` Ananyev, Konstantin
2017-09-14 9:35 ` Kavanagh, Mark B
2017-09-12 14:17 ` Ananyev, Konstantin
2017-09-13 10:44 ` Jiayu Hu
2017-09-13 22:10 ` Ananyev, Konstantin
2017-09-14 6:07 ` Jiayu Hu
2017-09-14 8:47 ` Ananyev, Konstantin
2017-09-14 9:29 ` Hu, Jiayu
2017-09-14 9:35 ` Ananyev, Konstantin
2017-09-14 10:01 ` Hu, Jiayu
2017-09-14 15:42 ` Kavanagh, Mark B
2017-09-14 18:38 ` Ananyev, Konstantin
2017-09-15 7:54 ` Hu, Jiayu
2017-09-15 8:15 ` Ananyev, Konstantin
2017-09-15 8:17 ` Ananyev, Konstantin
2017-09-15 8:38 ` Hu, Jiayu
2017-09-14 8:51 ` Kavanagh, Mark B
2017-09-14 9:45 ` Hu, Jiayu
2017-09-12 2:43 ` [dpdk-dev] [PATCH v3 3/5] gso: add VxLAN " Jiayu Hu
2017-09-12 2:43 ` [dpdk-dev] [PATCH v3 4/5] gso: add GRE " Jiayu Hu
2017-09-12 2:43 ` [dpdk-dev] [PATCH v3 5/5] app/testpmd: enable TCP/IPv4, VxLAN and GRE GSO Jiayu Hu
2017-09-14 18:33 ` Ferruh Yigit
2017-09-15 1:13 ` Hu, Jiayu
2017-09-19 7:32 ` [dpdk-dev] [PATCH v4 0/5] Support TCP/IPv4, VxLAN and GRE GSO in DPDK Jiayu Hu
2017-09-19 7:32 ` [dpdk-dev] [PATCH v4 1/5] gso: add Generic Segmentation Offload API framework Jiayu Hu
2017-09-19 7:32 ` [dpdk-dev] [PATCH v4 2/5] gso: add TCP/IPv4 GSO support Jiayu Hu
2017-09-20 7:03 ` Yao, Lei A
2017-09-19 7:32 ` [dpdk-dev] [PATCH v4 3/5] gso: add VxLAN " Jiayu Hu
2017-09-20 3:11 ` Tan, Jianfeng
2017-09-20 3:17 ` Hu, Jiayu
2017-09-19 7:32 ` [dpdk-dev] [PATCH v4 4/5] gso: add GRE " Jiayu Hu
2017-09-20 2:53 ` Tan, Jianfeng
2017-09-20 6:01 ` Hu, Jiayu
2017-09-19 7:32 ` [dpdk-dev] [PATCH v4 5/5] app/testpmd: enable TCP/IPv4, VxLAN and GRE GSO Jiayu Hu
2017-09-28 22:13 ` [dpdk-dev] [PATCH v5 0/6] Support TCP/IPv4, VxLAN and GRE GSO in DPDK Mark Kavanagh
2017-10-02 16:45 ` [dpdk-dev] [PATCH v6 0/6] Support TCP/IPv4, VxLAN, " Mark Kavanagh
2017-10-02 16:45 ` [dpdk-dev] [PATCH v6 1/6] gso: add Generic Segmentation Offload API framework Mark Kavanagh
2017-10-04 13:11 ` Ananyev, Konstantin
2017-10-04 13:21 ` Kavanagh, Mark B
2017-10-02 16:45 ` [dpdk-dev] [PATCH v6 2/6] gso: add TCP/IPv4 GSO support Mark Kavanagh
2017-10-04 13:32 ` Ananyev, Konstantin
2017-10-04 14:30 ` Kavanagh, Mark B
2017-10-04 14:49 ` Ananyev, Konstantin
2017-10-04 14:59 ` Kavanagh, Mark B
2017-10-04 13:35 ` Ananyev, Konstantin
2017-10-04 14:22 ` Kavanagh, Mark B
2017-10-02 16:45 ` [dpdk-dev] [PATCH v6 3/6] gso: add VxLAN " Mark Kavanagh
2017-10-04 14:12 ` Ananyev, Konstantin
2017-10-04 14:35 ` Kavanagh, Mark B
2017-10-04 16:13 ` Kavanagh, Mark B
2017-10-04 16:17 ` Ananyev, Konstantin
2017-10-02 16:45 ` [dpdk-dev] [PATCH v6 4/6] gso: add GRE " Mark Kavanagh
2017-10-04 14:15 ` Ananyev, Konstantin
2017-10-04 14:36 ` Kavanagh, Mark B
2017-10-02 16:45 ` [dpdk-dev] [PATCH v6 5/6] app/testpmd: enable TCP/IPv4, VxLAN and GRE GSO Mark Kavanagh
2017-10-04 15:08 ` Ananyev, Konstantin
2017-10-04 16:23 ` Kavanagh, Mark B
2017-10-04 16:26 ` Ananyev, Konstantin
2017-10-04 16:51 ` Kavanagh, Mark B
2017-10-02 16:45 ` [dpdk-dev] [PATCH v6 6/6] doc: add GSO programmer's guide Mark Kavanagh
2017-10-04 13:51 ` Mcnamara, John
2017-10-05 11:02 ` [dpdk-dev] [PATCH v7 0/6] Support TCP/IPv4, VxLAN, and GRE GSO in DPDK Mark Kavanagh
2017-10-05 11:02 ` [dpdk-dev] [PATCH v7 1/6] gso: add Generic Segmentation Offload API framework Mark Kavanagh
2017-10-05 11:02 ` [dpdk-dev] [PATCH v7 2/6] gso: add TCP/IPv4 GSO support Mark Kavanagh
2017-10-05 11:02 ` [dpdk-dev] [PATCH v7 3/6] gso: add VxLAN " Mark Kavanagh
2017-10-05 11:02 ` [dpdk-dev] [PATCH v7 4/6] gso: add GRE " Mark Kavanagh
2017-10-05 11:02 ` [dpdk-dev] [PATCH v7 5/6] app/testpmd: enable TCP/IPv4, VxLAN and GRE GSO Mark Kavanagh
2017-10-05 11:02 ` [dpdk-dev] [PATCH v7 6/6] doc: add GSO programmer's guide Mark Kavanagh
2017-10-05 13:22 ` [dpdk-dev] [PATCH v7 0/6] Support TCP/IPv4, VxLAN, and GRE GSO in DPDK Ananyev, Konstantin
2017-10-05 14:39 ` Kavanagh, Mark B
2017-10-05 15:43 ` [dpdk-dev] [PATCH v8 " Mark Kavanagh
2017-10-05 17:12 ` Ananyev, Konstantin
2017-10-05 20:16 ` Kavanagh, Mark B
2017-10-05 20:36 ` [dpdk-dev] [PATCH v9 " Mark Kavanagh
2017-10-05 22:24 ` Ananyev, Konstantin
2017-10-06 8:24 ` [dpdk-dev] FW: " Kavanagh, Mark B
2017-10-06 10:35 ` [dpdk-dev] " Kavanagh, Mark B
2017-10-06 23:32 ` Ferruh Yigit
2017-10-06 23:34 ` Ferruh Yigit
2017-10-07 14:56 ` [dpdk-dev] [PATCH v10 " Jiayu Hu
2017-10-07 14:56 ` [dpdk-dev] [PATCH v10 1/6] gso: add Generic Segmentation Offload API framework Jiayu Hu
2017-10-07 14:56 ` [dpdk-dev] [PATCH v10 2/6] gso: add TCP/IPv4 GSO support Jiayu Hu
2017-10-07 14:56 ` [dpdk-dev] [PATCH v10 3/6] gso: add VxLAN " Jiayu Hu
2017-10-07 14:56 ` [dpdk-dev] [PATCH v10 4/6] gso: add GRE " Jiayu Hu
2017-10-07 14:56 ` [dpdk-dev] [PATCH v10 5/6] app/testpmd: enable TCP/IPv4, VxLAN and GRE GSO Jiayu Hu
2017-10-07 14:56 ` [dpdk-dev] [PATCH v10 6/6] doc: add GSO programmer's guide Jiayu Hu
2017-10-08 3:40 ` [dpdk-dev] [PATCH v10 0/6] Support TCP/IPv4, VxLAN, and GRE GSO in DPDK Ferruh Yigit
2017-10-05 20:36 ` [dpdk-dev] [PATCH v9 1/6] gso: add Generic Segmentation Offload API framework Mark Kavanagh
2017-10-05 20:36 ` [dpdk-dev] [PATCH v9 2/6] gso: add TCP/IPv4 GSO support Mark Kavanagh
2017-10-05 20:36 ` [dpdk-dev] [PATCH v9 3/6] gso: add VxLAN " Mark Kavanagh
2017-10-05 20:36 ` [dpdk-dev] [PATCH v9 4/6] gso: add GRE " Mark Kavanagh
2017-10-05 20:36 ` [dpdk-dev] [PATCH v9 5/6] app/testpmd: enable TCP/IPv4, VxLAN and GRE GSO Mark Kavanagh
2017-10-05 20:36 ` [dpdk-dev] [PATCH v9 6/6] doc: add GSO programmer's guide Mark Kavanagh
2017-10-06 13:34 ` Mcnamara, John
2017-10-06 13:41 ` Kavanagh, Mark B
2017-10-05 15:43 ` [dpdk-dev] [PATCH v8 1/6] gso: add Generic Segmentation Offload API framework Mark Kavanagh
2017-10-05 15:44 ` [dpdk-dev] [PATCH v8 2/6] gso: add TCP/IPv4 GSO support Mark Kavanagh
2017-10-05 15:44 ` [dpdk-dev] [PATCH v8 3/6] gso: add VxLAN " Mark Kavanagh
2017-10-05 15:44 ` [dpdk-dev] [PATCH v8 4/6] gso: add GRE " Mark Kavanagh
2017-10-05 15:44 ` [dpdk-dev] [PATCH v8 5/6] app/testpmd: enable TCP/IPv4, VxLAN and GRE GSO Mark Kavanagh
2017-10-05 15:44 ` [dpdk-dev] [PATCH v8 6/6] doc: add GSO programmer's guide Mark Kavanagh
2017-10-05 17:57 ` Mcnamara, John
2017-09-28 22:13 ` [dpdk-dev] [PATCH v5 1/6] gso: add Generic Segmentation Offload API framework Mark Kavanagh
2017-09-28 22:13 ` [dpdk-dev] [PATCH v5 2/6] gso: add TCP/IPv4 GSO support Mark Kavanagh
2017-09-29 3:12 ` Jiayu Hu
2017-09-29 9:05 ` Kavanagh, Mark B
2017-09-28 22:13 ` [dpdk-dev] [PATCH v5 3/6] gso: add VxLAN " Mark Kavanagh
2017-09-28 22:13 ` [dpdk-dev] [PATCH v5 4/6] gso: add GRE " Mark Kavanagh
2017-09-28 22:13 ` [dpdk-dev] [PATCH v5 5/6] app/testpmd: enable TCP/IPv4, VxLAN and GRE GSO Mark Kavanagh
2017-09-28 22:13 ` [dpdk-dev] [PATCH v5 6/6] doc: add GSO programmer's guide Mark Kavanagh
2017-09-28 22:18 ` [dpdk-dev] [PATCH v5 2/6] gso: add TCP/IPv4 GSO support Mark Kavanagh
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=20170913024801.GB44293@dpdk15.sh.intel.com \
--to=jiayu.hu@intel.com \
--cc=dev@dpdk.org \
--cc=jianfeng.tan@intel.com \
--cc=konstantin.ananyev@intel.com \
--cc=mark.b.kavanagh@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).