From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from mga07.intel.com (mga07.intel.com [134.134.136.100]) by dpdk.org (Postfix) with ESMTP id 5B699374F for ; Mon, 3 Jul 2017 07:56:26 +0200 (CEST) Received: from fmsmga004.fm.intel.com ([10.253.24.48]) by orsmga105.jf.intel.com with ESMTP; 02 Jul 2017 22:56:25 -0700 X-ExtLoop1: 1 X-IronPort-AV: E=Sophos;i="5.40,301,1496127600"; d="scan'208";a="281454846" Received: from fmsmsx105.amr.corp.intel.com ([10.18.124.203]) by fmsmga004.fm.intel.com with ESMTP; 02 Jul 2017 22:56:25 -0700 Received: from fmsmsx155.amr.corp.intel.com (10.18.116.71) by FMSMSX105.amr.corp.intel.com (10.18.124.203) with Microsoft SMTP Server (TLS) id 14.3.319.2; Sun, 2 Jul 2017 22:56:25 -0700 Received: from shsmsx151.ccr.corp.intel.com (10.239.6.50) by FMSMSX155.amr.corp.intel.com (10.18.116.71) with Microsoft SMTP Server (TLS) id 14.3.319.2; Sun, 2 Jul 2017 22:56:24 -0700 Received: from shsmsx102.ccr.corp.intel.com ([169.254.2.146]) by SHSMSX151.ccr.corp.intel.com ([169.254.3.56]) with mapi id 14.03.0319.002; Mon, 3 Jul 2017 13:56:21 +0800 From: "Hu, Jiayu" To: "Tan, Jianfeng" , "dev@dpdk.org" CC: "Ananyev, Konstantin" , "stephen@networkplumber.org" , "yliu@fridaylinux.org" , "Wu, Jingjing" , "Yao, Lei A" , "Bie, Tiwei" Thread-Topic: [PATCH v10 1/3] lib: add Generic Receive Offload API framework Thread-Index: AQHS8lo++XjK4V/ikEaKrenKsbqne6I/zqgAgAHKMJA= Date: Mon, 3 Jul 2017 05:56:20 +0000 Message-ID: References: <1498805618-63649-1-git-send-email-jiayu.hu@intel.com> <1498907323-17563-1-git-send-email-jiayu.hu@intel.com> <1498907323-17563-2-git-send-email-jiayu.hu@intel.com> <9ccb015f-c635-3080-f732-6412998aacca@intel.com> In-Reply-To: <9ccb015f-c635-3080-f732-6412998aacca@intel.com> Accept-Language: en-US Content-Language: en-US X-MS-Has-Attach: X-MS-TNEF-Correlator: dlp-product: dlpe-windows dlp-version: 10.0.102.7 dlp-reaction: no-action x-originating-ip: [10.239.127.40] Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: quoted-printable MIME-Version: 1.0 Subject: Re: [dpdk-dev] [PATCH v10 1/3] lib: add Generic Receive Offload API framework X-BeenThere: dev@dpdk.org X-Mailman-Version: 2.1.15 Precedence: list List-Id: DPDK patches and discussions List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Mon, 03 Jul 2017 05:56:28 -0000 Hi Jianfeng, > -----Original Message----- > From: Tan, Jianfeng > Sent: Sunday, July 2, 2017 6:20 PM > To: Hu, Jiayu ; dev@dpdk.org > Cc: Ananyev, Konstantin ; > stephen@networkplumber.org; yliu@fridaylinux.org; Wu, Jingjing > ; Yao, Lei A ; Bie, Tiwei > > Subject: Re: [PATCH v10 1/3] lib: add Generic Receive Offload API framewo= rk >=20 >=20 >=20 > On 7/1/2017 7:08 PM, Jiayu Hu wrote: > > Generic Receive Offload (GRO) is a widely used SW-based offloading > > technique to reduce per-packet processing overhead. It gains > > performance by reassembling small packets into large ones. This > > patchset is to support GRO in DPDK. To support GRO, this patch > > implements a GRO API framework. > > > > To enable more flexibility to applications, DPDK GRO is implemented as > > a user library. Applications explicitly use the GRO library to merge > > small packets into large ones. DPDK GRO provides two reassembly modes. > > One is called lightweight mode, the other is called heavyweight mode. > > If applications want to merge packets in a simple way and the number > > of packets is relatively small, they can use the lightweight mode. > > If applications need more fine-grained controls, they can choose the > > heavyweight mode. > > > > rte_gro_reassemble_burst is the main reassembly API which is used in > > lightweight mode and processes N packets at a time. For applications, > > performing GRO in lightweight mode is simple. They just need to invoke > > rte_gro_reassemble_burst. Applications can get GROed packets as soon as > > rte_gro_reassemble_burst returns. > > > > rte_gro_reassemble is the main reassembly API which is used in > > heavyweight mode and tries to merge N inputted packets with the packets > > in a givn GRO table. For applications, performing GRO in heavyweight > > mode is relatively complicated. Before performing GRO, applications nee= d > > to create a GRO table by rte_gro_tbl_create. Then they can use > > rte_gro_reassemble to merge packets. The GROed packets are in the GRO > > table. If applications want to get them, applications need to manually > > flush them by flush API. >=20 > > Signed-off-by: Jiayu Hu > > --- > > config/common_base | 5 ++ > > lib/Makefile | 2 + > > lib/librte_gro/Makefile | 50 +++++++++++ > > lib/librte_gro/rte_gro.c | 176 > +++++++++++++++++++++++++++++++++++++ > > lib/librte_gro/rte_gro.h | 176 > +++++++++++++++++++++++++++++++++++++ > > lib/librte_gro/rte_gro_version.map | 12 +++ > > mk/rte.app.mk | 1 + > > 7 files changed, 422 insertions(+) > > create mode 100644 lib/librte_gro/Makefile > > create mode 100644 lib/librte_gro/rte_gro.c > > create mode 100644 lib/librte_gro/rte_gro.h > > create mode 100644 lib/librte_gro/rte_gro_version.map > > > > diff --git a/config/common_base b/config/common_base > > index f6aafd1..167f5ef 100644 > > --- a/config/common_base > > +++ b/config/common_base > > @@ -712,6 +712,11 @@ CONFIG_RTE_LIBRTE_VHOST_DEBUG=3Dn > > CONFIG_RTE_LIBRTE_PMD_VHOST=3Dn > > > > # > > +# Compile GRO library > > +# > > +CONFIG_RTE_LIBRTE_GRO=3Dy > > + > > +# > > #Compile Xen domain0 support > > # > > CONFIG_RTE_LIBRTE_XEN_DOM0=3Dn > > diff --git a/lib/Makefile b/lib/Makefile > > index 07e1fd0..ac1c2f6 100644 > > --- a/lib/Makefile > > +++ b/lib/Makefile > > @@ -106,6 +106,8 @@ DIRS-$(CONFIG_RTE_LIBRTE_REORDER) +=3D > librte_reorder > > DEPDIRS-librte_reorder :=3D librte_eal librte_mempool librte_mbuf > > DIRS-$(CONFIG_RTE_LIBRTE_PDUMP) +=3D librte_pdump > > DEPDIRS-librte_pdump :=3D librte_eal librte_mempool librte_mbuf > librte_ether > > +DIRS-$(CONFIG_RTE_LIBRTE_GRO) +=3D librte_gro > > +DEPDIRS-librte_gro :=3D librte_eal librte_mbuf librte_ether librte_net > > > > ifeq ($(CONFIG_RTE_EXEC_ENV_LINUXAPP),y) > > DIRS-$(CONFIG_RTE_LIBRTE_KNI) +=3D librte_kni > > diff --git a/lib/librte_gro/Makefile b/lib/librte_gro/Makefile > > new file mode 100644 > > index 0000000..7e0f128 > > --- /dev/null > > +++ b/lib/librte_gro/Makefile > > @@ -0,0 +1,50 @@ > > +# 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 copyri= ght > > +# notice, this list of conditions and the following disclaimer i= n > > +# 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 derive= d > > +# 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_SDK)/mk/rte.vars.mk > > + > > +# library name > > +LIB =3D librte_gro.a > > + > > +CFLAGS +=3D -O3 > > +CFLAGS +=3D $(WERROR_FLAGS) -I$(SRCDIR) > > + > > +EXPORT_MAP :=3D rte_gro_version.map > > + > > +LIBABIVER :=3D 1 > > + > > +# source files > > +SRCS-$(CONFIG_RTE_LIBRTE_GRO) +=3D rte_gro.c > > + > > +# install this header file > > +SYMLINK-$(CONFIG_RTE_LIBRTE_GRO)-include +=3D rte_gro.h > > + > > +include $(RTE_SDK)/mk/rte.lib.mk > > diff --git a/lib/librte_gro/rte_gro.c b/lib/librte_gro/rte_gro.c > > new file mode 100644 > > index 0000000..648835b > > --- /dev/null > > +++ b/lib/librte_gro/rte_gro.c > > @@ -0,0 +1,176 @@ > > +/*- > > + * BSD LICENSE > > + * > > + * Copyright(c) 2017 Intel Corporation. All rights reserved. > > + * > > + * Redistribution and use in source and binary forms, with or withou= t > > + * modification, are permitted provided that the following condition= s > > + * are met: > > + * > > + * * Redistributions of source code must retain the above copyrigh= t > > + * notice, this list of conditions and the following disclaimer. > > + * * Redistributions in binary form must reproduce the above copyr= ight > > + * 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 deriv= ed > > + * 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 > > +#include > > + > > +#include "rte_gro.h" > > + > > +typedef void *(*gro_tbl_create_fn)(uint16_t socket_id, > > + uint16_t max_flow_num, > > + uint16_t max_item_per_flow); > > +typedef void (*gro_tbl_destroy_fn)(void *tbl); > > +typedef uint32_t (*gro_tbl_item_num_fn)(void *tbl); > > + > > +static gro_tbl_create_fn > tbl_create_functions[RTE_GRO_TYPE_MAX_NUM]; > > +static gro_tbl_destroy_fn > tbl_destroy_functions[RTE_GRO_TYPE_MAX_NUM]; > > +static gro_tbl_item_num_fn > tbl_item_num_functions[RTE_GRO_TYPE_MAX_NUM]; > > + > > +/** > > + * GRO table, which is used to merge packets. It keeps many reassembly > > + * tables of desired GRO types. Applications need to create GRO tables > > + * before using rte_gro_reassemble to perform GRO. > > + */ > > +struct gro_tbl { > > + uint64_t desired_gro_types; /**< GRO types to perform */ > > + /* max TTL measured in nanosecond */ > > + uint64_t max_timeout_cycles; > > + /* max length of merged packet measured in byte */ > > + uint32_t max_packet_size; > > + /* reassebly tables of desired GRO types */ > > + void *tbls[RTE_GRO_TYPE_MAX_NUM]; > > +}; > > + > > +void *rte_gro_tbl_create(const > > + const struct rte_gro_param *param) >=20 > The name of this API and the definition of struct gro_tbl involve some > confusion. A gro table contains gro tables? I suppose a better name is > needed, for example, struct gro_ctl. Actually, a GRO table includes N reassembly tables. But gro_tbl is not a go= od name. I will change the name. Thanks. >=20 > > +{ > > + gro_tbl_create_fn create_tbl_fn; > > + gro_tbl_destroy_fn destroy_tbl_fn; > > + struct gro_tbl *gro_tbl; > > + uint64_t gro_type_flag =3D 0; > > + uint8_t i, j; > > + > > + gro_tbl =3D rte_zmalloc_socket(__func__, > > + sizeof(struct gro_tbl), > > + RTE_CACHE_LINE_SIZE, > > + param->socket_id); > > + if (gro_tbl =3D=3D NULL) > > + return NULL; > > + gro_tbl->max_packet_size =3D param->max_packet_size; > > + gro_tbl->max_timeout_cycles =3D param->max_timeout_cycles; > > + gro_tbl->desired_gro_types =3D param->desired_gro_types; > > + > > + for (i =3D 0; i < RTE_GRO_TYPE_MAX_NUM; i++) { > > + gro_type_flag =3D 1 << i; > > + > > + if ((param->desired_gro_types & gro_type_flag) =3D=3D 0) > > + continue; > > + create_tbl_fn =3D tbl_create_functions[i]; > > + if (create_tbl_fn =3D=3D NULL) > > + continue; > > + > > + gro_tbl->tbls[i] =3D create_tbl_fn( > > + param->socket_id, > > + param->max_flow_num, > > + param->max_item_per_flow); >=20 > Here and somewhere else: the alignment seems not correct. > gro_tbl->tbls[i] =3D create_tbl_fn(param->socket_id, > /* keep > all parameters aligned like this */ > param->max_flow_num, > param->max_item_per_flow); Thanks, I will modify it. > > + if (gro_tbl->tbls[i] =3D=3D NULL) { > > + /* destroy all allocated tables */ > > + for (j =3D 0; j < i; j++) { > > + gro_type_flag =3D 1 << j; > > + if ((param->desired_gro_types & > gro_type_flag) =3D=3D 0) > > + continue; > > + destroy_tbl_fn =3D tbl_destroy_functions[j]; > > + if (destroy_tbl_fn) > > + destroy_tbl_fn(gro_tbl->tbls[j]); > > + } > > + rte_free(gro_tbl); > > + return NULL; > > + } > > + } > > + return gro_tbl; > > +} > > + > > +void rte_gro_tbl_destroy(void *tbl) > > +{ > > + gro_tbl_destroy_fn destroy_tbl_fn; > > + struct gro_tbl *gro_tbl =3D (struct gro_tbl *)tbl; > > + uint64_t gro_type_flag; > > + uint8_t i; > > + > > + if (gro_tbl =3D=3D NULL) > > + return; > > + for (i =3D 0; i < RTE_GRO_TYPE_MAX_NUM; i++) { > > + gro_type_flag =3D 1 << i; > > + if ((gro_tbl->desired_gro_types & gro_type_flag) =3D=3D 0) > > + continue; > > + destroy_tbl_fn =3D tbl_destroy_functions[i]; > > + if (destroy_tbl_fn) > > + destroy_tbl_fn(gro_tbl->tbls[i]); > > + } > > + rte_free(gro_tbl); > > +} > > + > > +uint16_t > > +rte_gro_reassemble_burst(struct rte_mbuf **pkts __rte_unused, > > + uint16_t nb_pkts, > > + const struct rte_gro_param *param __rte_unused) > > +{ > > + return nb_pkts; > > +} > > + > > +uint16_t > > +rte_gro_reassemble(struct rte_mbuf **pkts __rte_unused, > > + uint16_t nb_pkts, > > + void *tbl __rte_unused) > > +{ > > + return nb_pkts; > > +} > > + > > +uint16_t > > +rte_gro_timeout_flush(void *tbl __rte_unused, > > + uint64_t desired_gro_types __rte_unused, > > + struct rte_mbuf **out __rte_unused, > > + uint16_t max_nb_out __rte_unused) > > +{ > > + return 0; > > +} > > + > > +uint64_t rte_gro_tbl_item_num(void *tbl) >=20 > Does rte_gro_get_count() sound better? OK, I will change the name. >=20 > > +{ > > + struct gro_tbl *gro_tbl =3D (struct gro_tbl *)tbl; > > + gro_tbl_item_num_fn item_num_fn; > > + uint64_t item_num =3D 0; > > + uint64_t gro_type_flag; > > + uint8_t i; > > + > > + for (i =3D 0; i < RTE_GRO_TYPE_MAX_NUM; i++) { > > + gro_type_flag =3D 1 << i; > > + if ((gro_tbl->desired_gro_types & gro_type_flag) =3D=3D 0) > > + continue; > > + > > + item_num_fn =3D tbl_item_num_functions[i]; > > + if (item_num_fn =3D=3D NULL) > > + continue; > > + item_num +=3D item_num_fn(gro_tbl->tbls[i]); > > + } > > + return item_num; > > +} > > diff --git a/lib/librte_gro/rte_gro.h b/lib/librte_gro/rte_gro.h > > new file mode 100644 > > index 0000000..02c9113 > > --- /dev/null > > +++ b/lib/librte_gro/rte_gro.h > > @@ -0,0 +1,176 @@ > > +/*- > > + * BSD LICENSE > > + * > > + * Copyright(c) 2017 Intel Corporation. All rights reserved. > > + * > > + * Redistribution and use in source and binary forms, with or withou= t > > + * modification, are permitted provided that the following condition= s > > + * are met: > > + * > > + * * Redistributions of source code must retain the above copyrigh= t > > + * notice, this list of conditions and the following disclaimer. > > + * * Redistributions in binary form must reproduce the above copyr= ight > > + * 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 deriv= ed > > + * 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 _RTE_GRO_H_ > > +#define _RTE_GRO_H_ > > + > > +#ifdef __cplusplus > > +extern "C" { > > +#endif > > + > > +/** > > + * the max packets number that rte_gro_reassemble_burst can > > + * process in each invocation. > > + */ > > +#define RTE_GRO_MAX_BURST_ITEM_NUM 128UL > > + > > +/* max number of supported GRO types */ > > +#define RTE_GRO_TYPE_MAX_NUM 64 > > +#define RTE_GRO_TYPE_SUPPORT_NUM 0 /**< current supported > GRO num */ > > + > > + > > +struct rte_gro_param { > > + uint64_t desired_gro_types; /**< desired GRO types */ >=20 > Make it gro_types for simplicity. Thanks, I will change it. >=20 > > + uint32_t max_packet_size; /**< max length of merged packets > */ >=20 > Refer to the tcp4 gro implementation, this is the max size for tcp > payload. But in principle, the 65535-byte limitation (including TCP > header) is because IP header lenght is 2-byte long. Yes, it's a bug. The length of (IP header + TCP header + payload) should be less than 64KB. But max_packet_size is uint32_t. >=20 > What does it means for other GRO engines then? I think these should be > decided by each gro engine. And applications don't have to change them. Make sense. We shouldn't expose it applications. I will remove this paramet= er. About the above bug, I will fix it by limiting the max length of a TCP/IPv4= packet to (l2_header + 64KB). >=20 > > + uint16_t max_flow_num; /**< max flow number */ > > + uint16_t max_item_per_flow; /**< max packet number per flow > */ > > + > > + /* socket index where the Ethernet port connects to */ >=20 > The comment needs to be refined. We have different socket idx for port, > for pmd thread. We just explain how this will be used: "socket index for > allocating gro related data structure". Thanks, I will change it. >=20 > > + uint16_t socket_id; > > + /* max TTL for a packet in the GRO table, measured in nanosecond > */ > > + uint64_t max_timeout_cycles; >=20 > We don't need to set it in lightweight mode. Please add this into the > comment. Thanks, I will add this comment. >=20 > > +}; > > + > > +/** > > + * This function create a GRO table, which is used to merge packets in > > + * rte_gro_reassemble. > > + * > > + * @param param > > + * applications use it to pass needed parameters to create a GRO tabl= e. > > + * @return > > + * if create successfully, return a pointer which points to the GRO > > + * table. Otherwise, return NULL. > > + */ > > +void *rte_gro_tbl_create( > > + const struct rte_gro_param *param); >=20 > Merge above two lines into one. Thanks, I will modify it. >=20 > > +/** > > + * This function destroys a GRO table. > > + */ > > +void rte_gro_tbl_destroy(void *tbl); > > + > > +/** > > + * This is one of the main reassembly APIs, which merges numbers of > > + * packets at a time. It assumes that all inputted packets are with > > + * correct checksums. That is, applications should guarantee all > > + * inputted packets are correct. Besides, it doesn't re-calculate > > + * checksums for merged packets. If inputted packets are IP fragmented= , > > + * this function assumes them are complete (i.e. with L4 header). Afte= r > > + * finishing processing, it returns all GROed packets to applications > > + * immediately. > > + * > > + * @param pkts > > + * a pointer array which points to the packets to reassemble. Besides= , > > + * it keeps packet addresses for GROed packets. > > + * @param nb_pkts > > + * the number of packets to reassemble. > > + * @param param > > + * applications use it to tell rte_gro_reassemble_burst what rules > > + * are demanded. > > + * @return > > + * the number of packets after been GROed. > > + */ > > +uint16_t rte_gro_reassemble_burst(struct rte_mbuf **pkts, > > + uint16_t nb_pkts, > > + const struct rte_gro_param *param); >=20 > Fix the alignment. >=20 > > + > > +/** > > + * Reassembly function, which tries to merge inputted packets with > > + * the packets in a given GRO table. This function assumes all inputte= d > > + * packet is with correct checksums. And it won't update checksums if > > + * two packets are merged. Besides, if inputted packets are IP > > + * fragmented, this function assumes they are complete packets (i.e. > > + * with L4 header). > > + * > > + * If the inputted packets don't have data or are with unsupported GRO > > + * types, they won't be processed and are returned to applications. > > + * Otherwise, the inputted packets are either merged or inserted into > > + * the table. If applications want get packets in the table, they need > > + * to call flush API. > > + * > > + * @param pkts > > + * packet to reassemble. Besides, after this function finishes, it > > + * keeps the unprocessed packets (i.e. without data or unsupported > > + * GRO types). > > + * @param nb_pkts > > + * the number of packets to reassemble. > > + * @param tbl > > + * a pointer points to a GRO table. > > + * @return > > + * return the number of unprocessed packets (i.e. without data or > > + * unsupported GRO types). If all packets are processed (merged or > > + * inserted into the table), return 0. > > + */ > > +uint16_t rte_gro_reassemble(struct rte_mbuf **pkts, > > + uint16_t nb_pkts, > > + void *tbl); > > + > > +/** > > + * This function flushes the timeout packets from reassembly tables of > > + * desired GRO types. The max number of flushed timeout packets is the > > + * element number of the array which is used to keep the flushed packe= ts. > > + * > > + * Besides, this function won't re-calculate checksums for merged > > + * packets in the tables. That is, the returned packets may be with > > + * wrong checksums. > > + * > > + * @param tbl > > + * a pointer points to a GRO table object. > > + * @param desired_gro_types > > + * rte_gro_timeout_flush only processes packets which belong to the > > + * GRO types specified by desired_gro_types. > > + * @param out > > + * a pointer array that is used to keep flushed timeout packets. > > + * @param nb_out > > + * the element number of out. It's also the max number of timeout > > + * packets that can be flushed finally. > > + * @return > > + * the number of flushed packets. If no packets are flushed, return 0= . > > + */ > > +uint16_t rte_gro_timeout_flush(void *tbl, > > + uint64_t desired_gro_types, > > + struct rte_mbuf **out, > > + uint16_t max_nb_out); > > + > > +/** > > + * This function returns the number of packets in a given GRO table. > > + * @param tbl > > + * pointer points to a GRO table. > > + * @return > > + * the number of packets in the table. > > + */ > > +uint64_t rte_gro_tbl_item_num(void *tbl); > > + > > +#ifdef __cplusplus > > +} > > +#endif > > + > > +#endif > > diff --git a/lib/librte_gro/rte_gro_version.map > b/lib/librte_gro/rte_gro_version.map > > new file mode 100644 > > index 0000000..358fb9d > > --- /dev/null > > +++ b/lib/librte_gro/rte_gro_version.map > > @@ -0,0 +1,12 @@ > > +DPDK_17.08 { > > + global: > > + > > + rte_gro_tbl_create; > > + rte_gro_tbl_destroy; >=20 > As stated earlier, here are the API names I suggested: > rte_gro_ctl_create()/rte_gro_ctl_destroy()/rte_gro_get_count(). Thanks, I will change the names. >=20 > > + rte_gro_reassemble_burst; > > + rte_gro_reassemble; > > + rte_gro_timeout_flush; > > + rte_gro_tbl_item_num; > > + > > + local: *; > > +}; > > diff --git a/mk/rte.app.mk b/mk/rte.app.mk > > index bcaf1b3..fc3776d 100644 > > --- a/mk/rte.app.mk > > +++ b/mk/rte.app.mk > > @@ -98,6 +98,7 @@ _LDLIBS-$(CONFIG_RTE_LIBRTE_RING) +=3D - > lrte_ring > > _LDLIBS-$(CONFIG_RTE_LIBRTE_EAL) +=3D -lrte_eal > > _LDLIBS-$(CONFIG_RTE_LIBRTE_CMDLINE) +=3D -lrte_cmdline > > _LDLIBS-$(CONFIG_RTE_LIBRTE_REORDER) +=3D -lrte_reorder > > +_LDLIBS-$(CONFIG_RTE_LIBRTE_GRO) +=3D -lrte_gro > > > > ifeq ($(CONFIG_RTE_EXEC_ENV_LINUXAPP),y) > > _LDLIBS-$(CONFIG_RTE_LIBRTE_KNI) +=3D -lrte_kni