From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from EUR04-VI1-obe.outbound.protection.outlook.com (mail-eopbgr80089.outbound.protection.outlook.com [40.107.8.89]) by dpdk.org (Postfix) with ESMTP id 6988F1B156 for ; Sat, 29 Sep 2018 01:03:35 +0200 (CEST) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=Mellanox.com; s=selector1; h=From:Date:Subject:Message-ID:Content-Type:MIME-Version:X-MS-Exchange-SenderADCheck; bh=dOsYm0NKFdg77bSjEKyBSMFNF7qkB0TGKhAXMOQP5Yw=; b=NJh36kB+gB3Ae/dAYmkayi4mbsXn9sOl274mW3l5ri69iHkeDYyPldzOptgswIQ6Ufz1sfEd08AvBOUGibhjSYiAj89F5Jrp+VZtJJLMWQ+zQF+DBaLqtbfhhECwbHuRSiyudp6CFkSWrNb9a+J/ValpQLzJeKMVMBT/MVaGO3I= Received: from DB3PR0502MB3980.eurprd05.prod.outlook.com (52.134.72.27) by DB3PR0502MB4025.eurprd05.prod.outlook.com (52.134.72.142) with Microsoft SMTP Server (version=TLS1_2, cipher=TLS_ECDHE_RSA_WITH_AES_256_GCM_SHA384) id 15.20.1143.18; Fri, 28 Sep 2018 23:03:33 +0000 Received: from DB3PR0502MB3980.eurprd05.prod.outlook.com ([fe80::1cb0:661b:ecab:6045]) by DB3PR0502MB3980.eurprd05.prod.outlook.com ([fe80::1cb0:661b:ecab:6045%2]) with mapi id 15.20.1164.024; Fri, 28 Sep 2018 23:03:33 +0000 From: Yongseok Koh To: Jack Min CC: Shahaf Shuler , "dev@dpdk.org" Thread-Topic: [PATCH] net/mlx5: eswitch-IP address UDP/TCP port rewrite Thread-Index: AQHUVMYzsnQsXgDklk+QshXlrRt6gqUGVV4A Date: Fri, 28 Sep 2018 23:03:33 +0000 Message-ID: <20180928230318.GA90324@minint-98vp2qg> References: <20180925115107.12242-1-jackmin@mellanox.com> In-Reply-To: <20180925115107.12242-1-jackmin@mellanox.com> Accept-Language: en-US Content-Language: en-US X-MS-Has-Attach: X-MS-TNEF-Correlator: x-clientproxiedby: DM5PR2001CA0003.namprd20.prod.outlook.com (2603:10b6:4:16::13) To DB3PR0502MB3980.eurprd05.prod.outlook.com (2603:10a6:8:10::27) authentication-results: spf=none (sender IP is ) smtp.mailfrom=yskoh@mellanox.com; x-ms-exchange-messagesentrepresentingtype: 1 x-originating-ip: [69.181.245.183] x-ms-publictraffictype: Email x-microsoft-exchange-diagnostics: 1; DB3PR0502MB4025; 6:YpnGDnIdIitMXAjw5L7/IMf0A6GvJoPAsNeIL8Ya80N2LNyEolyv1x36WaPjVd0TuqTr4gdT0nKYujza2KyljuNCQn9DaMt4WZiaQvwppqhsJQikP3xAz5h29rCtgK7u7Bx2sMiDtdJTpcpLgAGDLodeF+zevwbmRflhZ/2YVGyniWwfp+aXInxRet+hXlSWqlRVZ6VRGvFn1/RqdddO84TkM0g1SzUu1dArVPU7zQN3y/r6JLjJmBX9gP0BIijctYiVBBXaIlJ4C6iFvgLDBZIW5WkCCi+qw+u9NicUHSShxhna2aaCfBP+sgCnf4j0U82kz2BbRZmAyg/w0psuC5iOCgLoOD7gfojx/8ThQxAPAUhnBaUC6+GWm4y9JaGsqpqCtf/+Aan97GTuTioRhFGC7/bQRjAXj9NNWpFo3cMFXQc+Zeb3N6uZzI3aEdCNrEA+AG89t0CcYWxs7VYyrA==; 5:KueQgEhDCTvveb5nwQLPtdjfsK5nI+qifWAAOvr1eB/rqoeglYoixrsoP3zsRUer3EkCgKa17DZi2JqHGomLZHW0gIJnGmSQWJm6qDB18haLNM7I86FNntq6oNw/jjZeh5ptEqF989MuvP8CsL53wOfPvpqWzGQT0Vntjp4Idmw=; 7:2w1O91WxWVsie61gWN4xKPuKPdPb165hBp08V9txkEGEKoOkNRFrHgZtc3pgHekChcXz+1vd//M003KBUufvhFO7rFJ34Awr23/X/0DSXxabuYr6SB3NDy1xAlBX8JbePH6f2cKiFft+7ueOv7d5bzfMrjLaObjulAjoe1ao7a6M7Ek9V+P6T8aM5NImDwQjcCb3DcGKgLHxPXu5FlIijhfGbs0vcnVz0MEYvYrDIlREyO8d5QDBrjZaO4xr/BHB x-ms-office365-filtering-correlation-id: 968c8335-55eb-4178-84a7-08d625969cdf x-ms-office365-filtering-ht: Tenant x-microsoft-antispam: BCL:0; PCL:0; RULEID:(7020095)(4652040)(8989299)(5600074)(711020)(4618075)(2017052603328)(7153060)(7193020); SRVR:DB3PR0502MB4025; x-ms-traffictypediagnostic: DB3PR0502MB4025: x-microsoft-antispam-prvs: x-exchange-antispam-report-test: UriScan:; x-ms-exchange-senderadcheck: 1 x-exchange-antispam-report-cfa-test: BCL:0; PCL:0; RULEID:(8211001083)(6040522)(2401047)(8121501046)(5005006)(3231355)(944501410)(52105095)(3002001)(10201501046)(93006095)(93001095)(6055026)(149066)(150057)(6041310)(20161123560045)(201703131423095)(201702281528075)(20161123555045)(201703061421075)(201703061406153)(20161123564045)(20161123562045)(20161123558120)(201708071742011)(7699051); SRVR:DB3PR0502MB4025; BCL:0; PCL:0; RULEID:; SRVR:DB3PR0502MB4025; x-forefront-prvs: 0809C12563 x-forefront-antispam-report: SFV:NSPM; SFS:(10009020)(7916004)(136003)(39860400002)(396003)(346002)(366004)(376002)(189003)(199004)(6862004)(33896004)(8936002)(2906002)(14454004)(6636002)(81166006)(6512007)(6116002)(33656002)(102836004)(316002)(6436002)(1076002)(53936002)(53946003)(9686003)(229853002)(6486002)(52116002)(81156014)(34290500001)(76176011)(8676002)(6306002)(386003)(14444005)(105586002)(97736004)(106356001)(66066001)(26005)(476003)(6246003)(68736007)(54906003)(99286004)(86362001)(446003)(5250100002)(11346002)(478600001)(33716001)(5660300001)(6506007)(966005)(25786009)(4326008)(2900100001)(7736002)(256004)(486006)(3846002)(305945005)(71200400001)(71190400001)(4006002); DIR:OUT; SFP:1101; SCL:1; SRVR:DB3PR0502MB4025; H:DB3PR0502MB3980.eurprd05.prod.outlook.com; FPR:; SPF:None; LANG:en; PTR:InfoNoRecords; MX:1; A:1; received-spf: None (protection.outlook.com: mellanox.com does not designate permitted sender hosts) x-microsoft-antispam-message-info: HCh1F8ncDve051XEwmKs6x847LTu/ikfb3nA/JfTfFDMqjwhYZ4DArSJmSCaqw64ZD+nUIElBAataFpq6ST8IUAMfps1jCFjzWrL9xhe+NJSiCD5t4s5zp61ZIzVRw6BIOqsEZarUc9ezemDpGrMbGf6aaKP2m/5Zu+3kY6Wsq97EeYyoUsNqIjXpKsbo6+BtZgh3IlNhwdNiTrXk5bm+KE/zDX7pvU12Zg9BbnFnDkXkBUEB7JK2V/jrqIa2azmyCm6EI+Y6MFq25T0CoWYr2ATwWxKS0+nPliLkTOhQFHMJJ1n9qdXFRT6cg3Aka6v1wnGwS5qsNAEvMXAuuI6xoxgavX3xJvBKa0THJJk5jQ= spamdiagnosticoutput: 1:99 spamdiagnosticmetadata: NSPM Content-Type: text/plain; charset="us-ascii" Content-ID: <41524EA5F50588478D0E65D265A1C74B@eurprd05.prod.outlook.com> Content-Transfer-Encoding: quoted-printable MIME-Version: 1.0 X-OriginatorOrg: Mellanox.com X-MS-Exchange-CrossTenant-Network-Message-Id: 968c8335-55eb-4178-84a7-08d625969cdf X-MS-Exchange-CrossTenant-originalarrivaltime: 28 Sep 2018 23:03:33.9063 (UTC) X-MS-Exchange-CrossTenant-fromentityheader: Hosted X-MS-Exchange-CrossTenant-id: a652971c-7d2e-4d9b-a6a4-d149256f461b X-MS-Exchange-Transport-CrossTenantHeadersStamped: DB3PR0502MB4025 Subject: Re: [dpdk-dev] [PATCH] net/mlx5: eswitch-IP address UDP/TCP port rewrite 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: Fri, 28 Sep 2018 23:03:35 -0000 On Tue, Sep 25, 2018 at 07:51:06PM +0800, Xiaoyu Min wrote: > Offload the following rte_flow actions by inserting accordingly > E-Switch rules via TC Flower driver >=20 > - RTE_FLOW_ACTION_TYPE_SET_IPV4_SRC > - RTE_FLOW_ACTION_TYPE_SET_IPV4_DST > - RTE_FLOW_ACTION_TYPE_SET_IPV6_SRC > - RTE_FLOW_ACTION_TYPE_SET_IPV6_DST > - RTE_FLOW_ACTION_TYPE_SET_TP_SRC > - RTE_FLOW_ACTION_TYPE_SET_TP_DST Can you put an example command of testpmd for the reference? > Signed-off-by: Xiaoyu Min > --- > This patch bases on Rahul Lakkireddy's patchs[1][2] and > Yongseok Koh's patchset [3] >=20 > [1] https://patches.dpdk.org/patch/45191/ > [2] https://patches.dpdk.org/patch/45192/ > [3] https://patches.dpdk.org/project/dpdk/list/?series=3D1474 >=20 >=20 > drivers/net/mlx5/Makefile | 5 + > drivers/net/mlx5/meson.build | 2 + > drivers/net/mlx5/mlx5_flow.h | 6 + > drivers/net/mlx5/mlx5_flow_tcf.c | 356 +++++++++++++++++++++++++++++++ > 4 files changed, 369 insertions(+) >=20 > diff --git a/drivers/net/mlx5/Makefile b/drivers/net/mlx5/Makefile > index ca1de9f21..49b95e78e 100644 > --- a/drivers/net/mlx5/Makefile > +++ b/drivers/net/mlx5/Makefile > @@ -346,6 +346,11 @@ mlx5_autoconf.h.new: $(RTE_SDK)/buildtools/auto-conf= ig-h.sh > linux/tc_act/tc_vlan.h \ > enum TCA_VLAN_PUSH_VLAN_PRIORITY \ > $(AUTOCONF_OUTPUT) > + $Q sh -- '$<' '$@' \ > + HAVE_TC_ACT_PEDIT \ > + linux/tc_act/tc_pedit.h \ > + enum TCA_PEDIT_KEY_EX_HDR_TYPE_UDP \ > + $(AUTOCONF_OUTPUT) > $Q sh -- '$<' '$@' \ > HAVE_SUPPORTED_40000baseKR4_Full \ > /usr/include/linux/ethtool.h \ > diff --git a/drivers/net/mlx5/meson.build b/drivers/net/mlx5/meson.build > index fd93ac162..ef6a85101 100644 > --- a/drivers/net/mlx5/meson.build > +++ b/drivers/net/mlx5/meson.build > @@ -182,6 +182,8 @@ if build > 'TCA_FLOWER_KEY_VLAN_ETH_TYPE' ], > [ 'HAVE_TC_ACT_VLAN', 'linux/tc_act/tc_vlan.h', > 'TCA_VLAN_PUSH_VLAN_PRIORITY' ], > + [ 'HAVE_TC_ACT_PEDIT', 'linux/tc_act/tc_pedit.h', > + 'TCA_PEDIT_KEY_EX_HDR_TYPE_UDP' ], > [ 'HAVE_RDMA_NL_NLDEV', 'rdma/rdma_netlink.h', > 'RDMA_NL_NLDEV' ], > [ 'HAVE_RDMA_NLDEV_CMD_GET', 'rdma/rdma_netlink.h', > diff --git a/drivers/net/mlx5/mlx5_flow.h b/drivers/net/mlx5/mlx5_flow.h > index 10d700a7f..be182a643 100644 > --- a/drivers/net/mlx5/mlx5_flow.h > +++ b/drivers/net/mlx5/mlx5_flow.h > @@ -87,6 +87,12 @@ > #define MLX5_ACTION_OF_PUSH_VLAN (1u << 8) > #define MLX5_ACTION_OF_SET_VLAN_VID (1u << 9) > #define MLX5_ACTION_OF_SET_VLAN_PCP (1u << 10) > +#define MLX5_ACTION_SET_IPV4_SRC (1u << 11) > +#define MLX5_ACTION_SET_IPV4_DST (1u << 12) > +#define MLX5_ACTION_SET_IPV6_SRC (1u << 13) > +#define MLX5_ACTION_SET_IPV6_DST (1u << 14) > +#define MLX5_ACTION_SET_TP_SRC (1u << 15) > +#define MLX5_ACTION_SET_TP_DST (1u << 16) > =20 > /* possible L3 layers protocols filtering. */ > #define MLX5_IP_PROTOCOL_TCP 6 > diff --git a/drivers/net/mlx5/mlx5_flow_tcf.c b/drivers/net/mlx5/mlx5_flo= w_tcf.c > index 14376188e..85c92f369 100644 > --- a/drivers/net/mlx5/mlx5_flow_tcf.c > +++ b/drivers/net/mlx5/mlx5_flow_tcf.c > @@ -53,6 +53,62 @@ struct tc_vlan { > =20 > #endif /* HAVE_TC_ACT_VLAN */ > =20 > +#ifdef HAVE_TC_ACT_PEDIT > + > +#include > + > +#else /* HAVE_TC_ACT_VLAN */ > +enum { > + TCA_PEDIT_UNSPEC, > + TCA_PEDIT_TM, > + TCA_PEDIT_PARMS, > + TCA_PEDIT_PAD, > + TCA_PEDIT_PARMS_EX, > + TCA_PEDIT_KEYS_EX, > + TCA_PEDIT_KEY_EX, > + __TCA_PEDIT_MAX > +}; > + > +enum { > + TCA_PEDIT_KEY_EX_HTYPE =3D 1, > + TCA_PEDIT_KEY_EX_CMD =3D 2, > + __TCA_PEDIT_KEY_EX_MAX > +}; > + > +enum pedit_header_type { > + TCA_PEDIT_KEY_EX_HDR_TYPE_NETWORK =3D 0, > + TCA_PEDIT_KEY_EX_HDR_TYPE_ETH =3D 1, > + TCA_PEDIT_KEY_EX_HDR_TYPE_IP4 =3D 2, > + TCA_PEDIT_KEY_EX_HDR_TYPE_IP6 =3D 3, > + TCA_PEDIT_KEY_EX_HDR_TYPE_TCP =3D 4, > + TCA_PEDIT_KEY_EX_HDR_TYPE_UDP =3D 5, > + __PEDIT_HDR_TYPE_MAX, > +}; > + > +enum pedit_cmd { > + TCA_PEDIT_KEY_EX_CMD_SET =3D 0, > + TCA_PEDIT_KEY_EX_CMD_ADD =3D 1, > + __PEDIT_CMD_MAX, > +}; > + > +struct tc_pedit_key { > + __u32 mask; /* AND */ > + __u32 val; /*XOR */ > + __u32 off; /*offset */ > + __u32 at; > + __u32 offmask; > + __u32 shift; > +}; > + > +struct tc_pedit_sel { > + tc_gen; > + unsigned char nkeys; > + unsigned char flags; > + struct tc_pedit_key keys[0]; > +}; > + > +#endif /* HAVE_TC_ACT_VLAN */ > + > /* Normally found in linux/netlink.h. */ > #ifndef NETLINK_CAP_ACK > #define NETLINK_CAP_ACK 10 > @@ -153,6 +209,14 @@ struct tc_vlan { > #define IPV6_ADDR_LEN 16 > #endif > =20 > +#ifndef IPV4_ADDR_LEN > +#define IPV4_ADDR_LEN 4 > +#endif > + > +#ifndef TP_PORT_LEN > +#define TP_PORT_LEN 2 /* Transport Port (UDP/TCP) Length */ > +#endif > + > /** Empty masks for known item types. */ > static const union { > struct rte_flow_item_port_id port_id; > @@ -227,6 +291,220 @@ struct flow_tcf_ptoi { > =20 > #define MLX5_TCF_FATE_ACTIONS (MLX5_ACTION_DROP | MLX5_ACTION_PORT_ID) > =20 > +#define IS_MODIFY_ACTION(act_) ({typeof(act_) act =3D (act_); \ > + ((act) =3D=3D RTE_FLOW_ACTION_TYPE_SET_IPV4_SRC || \ > + (act) =3D=3D RTE_FLOW_ACTION_TYPE_SET_IPV4_DST || \ > + (act) =3D=3D RTE_FLOW_ACTION_TYPE_SET_IPV6_SRC || \ > + (act) =3D=3D RTE_FLOW_ACTION_TYPE_SET_IPV6_DST || \ > + (act) =3D=3D RTE_FLOW_ACTION_TYPE_SET_TP_SRC || \ > + (act) =3D=3D RTE_FLOW_ACTION_TYPE_SET_TP_DST) ? \ > + 1 : 0; }) The reason why you need this complex multi-conditional macro is that RTE_FLOW_ACTION_TYPE_* isn't a bitmask. But, as this actions will be conver= ted to MLX5_ACTION_* which is a bitmask, you can use that instead. Then, this would be enough to be: #define MLX5_TCF_SET_ACTIONS \ (MLX5_ACTION_SET_IPV4_SRC | ...) And in the flow_tcf_validate() below, if (action_flags & MLX5_TCF_SET_ACTIONS) { ... } And please note that I'm going to rename MLX5_ACTION_* to MLX5_FLOW_ACTION_= *. Please refer to "net/mlx5: rename flow macros" in PR #885. You might need t= o rebase it again. > +#define MAX_PEDIT_KEYS (128) > +#define SZ_PEDIT_KEY_VAL (4) > + > +struct pedit_key_ex { > + enum pedit_header_type htype; > + enum pedit_cmd cmd; > +}; > + > +struct pedit_parser { > + struct tc_pedit_sel sel; > + struct tc_pedit_key keys[MAX_PEDIT_KEYS]; > + struct pedit_key_ex keys_ex[MAX_PEDIT_KEYS]; > +}; > + > +static int > +flow_tcf_calc_pedit_keys(const uint64_t size) Please add documentation by comment for every funcs you add. Refer to the other existing ones for formality. > +{ > + int keys =3D (size / SZ_PEDIT_KEY_VAL) + > + ((size % SZ_PEDIT_KEY_VAL) ? 1 : 0); Indentation. > + return keys; > +} > + > +static void > +flow_tcf_pedit_key_set_tp_port(const struct rte_flow_action *actions, > + struct pedit_parser *p_parser, > + uint64_t item_flags) > +{ > + int idx =3D p_parser->sel.nkeys; > + > + if (item_flags & MLX5_FLOW_LAYER_OUTER_L4_UDP) > + p_parser->keys_ex[idx].htype =3D TCA_PEDIT_KEY_EX_HDR_TYPE_UDP; > + if (item_flags & MLX5_FLOW_LAYER_OUTER_L4_TCP) > + p_parser->keys_ex[idx].htype =3D TCA_PEDIT_KEY_EX_HDR_TYPE_TCP; > + p_parser->keys_ex[idx].cmd =3D TCA_PEDIT_KEY_EX_CMD_SET; > + p_parser->keys[idx].off =3D > + actions->type =3D=3D RTE_FLOW_ACTION_TYPE_SET_TP_DST ? 2 : 0; assert(offsetof(struct tcp_hdr, src_port) =3D=3D offsetof(struct udp_hdr, src_port)); assert(offsetof(struct tcp_hdr, dst_port) =3D=3D offsetof(struct udp_hdr, dst_port)); p_parser->keys[idx].off =3D actions->type =3D=3D RTE_FLOW_ACTION_TYPE_SET_TP_SRC ? offsetof(struct tcp_hdr, src_port) : offsetof(struct tcp_hdr, dst_port); assert() is just to be informative. And how about src first like others below? > + p_parser->keys[idx].mask =3D 0xFFFF0000; > + p_parser->keys[idx].val =3D ((const struct rte_flow_action_set_tp *) > + actions->conf)->port; Assigning 2B to 4B big-endian stroage? Doesn't look consistent with the mas= k above - 0xffff0000. > + p_parser->sel.nkeys =3D (++idx); > +} > + > +static void > +flow_tcf_pedit_key_set_ipv6_addr(const struct rte_flow_action *actions, > + struct pedit_parser *p_parser) > +{ > + int idx =3D p_parser->sel.nkeys; > + int keys =3D flow_tcf_calc_pedit_keys(IPV6_ADDR_LEN); > + int off_base =3D > + actions->type =3D=3D RTE_FLOW_ACTION_TYPE_SET_IPV6_SRC ? 8 : 24; offsetof(struct ipv6_hdr, src_addr) : offsetof(struct ipv6_hdr, dst_addr); > + const struct rte_flow_action_set_ipv6 *conf =3D > + (const struct rte_flow_action_set_ipv6 *)actions->conf; > + > + for (int i =3D 0; i < keys; i++, idx++) { > + p_parser->keys_ex[idx].htype =3D TCA_PEDIT_KEY_EX_HDR_TYPE_IP6; > + p_parser->keys_ex[idx].cmd =3D TCA_PEDIT_KEY_EX_CMD_SET; > + p_parser->keys[idx].off =3D off_base + i * SZ_PEDIT_KEY_VAL; > + p_parser->keys[idx].mask =3D ~UINT32_MAX; > + memcpy(&p_parser->keys[idx].val, > + conf->ipv6_addr + i * SZ_PEDIT_KEY_VAL, > + SZ_PEDIT_KEY_VAL); > + } > + p_parser->sel.nkeys +=3D keys; > +} > + > +static void > +flow_tcf_pedit_key_set_ipv4_addr(const struct rte_flow_action *actions, How about getting rte_flow_action_set_ipv4 instead of rte_flow_action? Same comment for ipv6 and tp_port. > + struct pedit_parser *p_parser) > +{ > + int idx =3D p_parser->sel.nkeys; > + > + p_parser->keys_ex[idx].htype =3D TCA_PEDIT_KEY_EX_HDR_TYPE_IP4; > + p_parser->keys_ex[idx].cmd =3D TCA_PEDIT_KEY_EX_CMD_SET; > + p_parser->keys[idx].off =3D > + (actions->type =3D=3D RTE_FLOW_ACTION_TYPE_SET_IPV4_SRC ? 12 : 16); offsetof(struct ipv4_hdr, src_addr) : offsetof(struct ipv4_hdr, dst_addr); > + p_parser->keys[idx].mask =3D ~UINT32_MAX; > + p_parser->keys[idx].val =3D > + ((const struct rte_flow_action_set_ipv4 *) > + actions->conf)->ipv4_addr; > + p_parser->sel.nkeys =3D (++idx); > +} > + > +static int > +flow_tcf_create_pedit_mnl_msg(struct nlmsghdr *nl, > + const struct rte_flow_action **actions, > + uint64_t item_flags) > +{ > + struct pedit_parser p_parser; > + > + memset(&p_parser, 0, sizeof(p_parser)); > + mnl_attr_put_strz(nl, TCA_ACT_KIND, "pedit"); > + struct nlattr *na_act_options =3D mnl_attr_nest_start(nl, > + TCA_ACT_OPTIONS); > + /* all modify header actions should be in one tc-pedit action */ > + for (; (*actions)->type !=3D RTE_FLOW_ACTION_TYPE_END; (*actions)++) { It seems that you want to aggregate all the pedit actions and form a single na attr. But what if rte_flow_action_set_* are not contiguous? E.g. flow create ... actions set1 / set2 / port_id / set3 / end Then, it would have two pedit na attrs. Is that okay? Or, need to think about another way? Same will happen in flow_tcf_get_pedit_actions_size(). > + switch ((*actions)->type) { > + case RTE_FLOW_ACTION_TYPE_SET_IPV4_SRC: > + case RTE_FLOW_ACTION_TYPE_SET_IPV4_DST: > + flow_tcf_pedit_key_set_ipv4_addr(*actions, &p_parser); > + break; > + case RTE_FLOW_ACTION_TYPE_SET_IPV6_SRC: > + case RTE_FLOW_ACTION_TYPE_SET_IPV6_DST: > + flow_tcf_pedit_key_set_ipv6_addr(*actions, &p_parser); > + break; > + case RTE_FLOW_ACTION_TYPE_SET_TP_SRC: > + case RTE_FLOW_ACTION_TYPE_SET_TP_DST: > + flow_tcf_pedit_key_set_tp_port(*actions, > + &p_parser, item_flags); > + break; > + default: > + goto pedit_mnl_msg_done; > + } > + } > +pedit_mnl_msg_done: > + p_parser.sel.action =3D TC_ACT_PIPE; > + mnl_attr_put(nl, TCA_PEDIT_PARMS_EX, > + sizeof(p_parser.sel) + > + p_parser.sel.nkeys * sizeof(struct tc_pedit_key), > + &p_parser); > + struct nlattr *na_pedit_keys =3D mnl_attr_nest_start(nl, > + TCA_PEDIT_KEYS_EX | NLA_F_NESTED); > + for (int i =3D 0; i < p_parser.sel.nkeys; i++) { > + struct nlattr *na_pedit_key =3D mnl_attr_nest_start(nl, > + TCA_PEDIT_KEY_EX | NLA_F_NESTED); > + mnl_attr_put_u16(nl, TCA_PEDIT_KEY_EX_HTYPE, > + p_parser.keys_ex[i].htype); > + mnl_attr_put_u16(nl, TCA_PEDIT_KEY_EX_CMD, > + p_parser.keys_ex[i].cmd); > + mnl_attr_nest_end(nl, na_pedit_key); > + } > + mnl_attr_nest_end(nl, na_pedit_keys); > + mnl_attr_nest_end(nl, na_act_options); > + (*actions)--; > + return 0; > +} > + > +/** > + * Calculate max memory size of one TC-pedit actions. > + * One TC-pedit action can contain set of keys each defining > + * a rewrite element (rte_flow action) > + * > + * @param[in] actions > + * actions specification. > + * @param[inout] action_flags > + * actions flags > + * @param[inout] size > + * accumulated size > + * @return > + * Max memory size of one TC-pedit action > + */ > +static int > +flow_tcf_get_pedit_actions_size(const struct rte_flow_action **actions, > + uint64_t *action_flags) > +{ > + int pedit_size =3D 0; > + int keys =3D 0; > + uint64_t flags =3D 0; > + > + pedit_size +=3D SZ_NLATTR_NEST + /* na_act_index. */ > + SZ_NLATTR_STRZ_OF("pedit") + > + SZ_NLATTR_NEST; /* TCA_ACT_OPTIONS. */ > + for (; (*actions)->type !=3D RTE_FLOW_ACTION_TYPE_END; (*actions)++) { > + switch ((*actions)->type) { > + case RTE_FLOW_ACTION_TYPE_SET_IPV4_SRC: > + keys +=3D flow_tcf_calc_pedit_keys(IPV4_ADDR_LEN); > + flags |=3D MLX5_ACTION_SET_IPV4_SRC; > + break; > + case RTE_FLOW_ACTION_TYPE_SET_IPV4_DST: > + keys +=3D flow_tcf_calc_pedit_keys(IPV4_ADDR_LEN); > + flags |=3D MLX5_ACTION_SET_IPV4_DST; > + break; > + case RTE_FLOW_ACTION_TYPE_SET_IPV6_SRC: > + keys +=3D flow_tcf_calc_pedit_keys(IPV6_ADDR_LEN); > + flags |=3D MLX5_ACTION_SET_IPV6_SRC; > + break; > + case RTE_FLOW_ACTION_TYPE_SET_IPV6_DST: > + keys +=3D flow_tcf_calc_pedit_keys(IPV6_ADDR_LEN); > + flags |=3D MLX5_ACTION_SET_IPV6_DST; > + break; > + case RTE_FLOW_ACTION_TYPE_SET_TP_SRC: > + /* TCP is as same as UDP */ > + keys +=3D flow_tcf_calc_pedit_keys(TP_PORT_LEN); > + flags |=3D MLX5_ACTION_SET_TP_SRC; > + break; > + case RTE_FLOW_ACTION_TYPE_SET_TP_DST: > + /* TCP is as same as UDP */ > + keys +=3D flow_tcf_calc_pedit_keys(TP_PORT_LEN); > + flags |=3D MLX5_ACTION_SET_TP_DST; > + break; > + default: > + goto get_pedit_action_size_done; > + } > + } > +get_pedit_action_size_done: > + /* TCA_PEDIT_PARAMS_EX */ > + pedit_size +=3D SZ_NLATTR_DATA_OF(sizeof(struct tc_pedit_sel) + > + keys * sizeof(struct tc_pedit_key)); > + pedit_size +=3D SZ_NLATTR_NEST; /* TCA_PEDIT_KEYS */ > + pedit_size +=3D keys * > + /* TCA_PEDIT_KEY_EX + HTYPE + CMD */ > + (SZ_NLATTR_NEST + SZ_NLATTR_DATA_OF(2) + SZ_NLATTR_DATA_OF(2)); > + (*action_flags) |=3D flags; > + (*actions)--; > + return pedit_size; > +} > + > /** > * Retrieve mask for pattern item. > * > @@ -430,6 +708,8 @@ flow_tcf_validate(struct rte_eth_dev *dev, > of_set_vlan_vid; > const struct rte_flow_action_of_set_vlan_pcp * > of_set_vlan_pcp; > + const struct rte_flow_action_set_ipv4 *set_ipv4; > + const struct rte_flow_action_set_ipv6 *set_ipv6; > } conf; > uint32_t item_flags =3D 0; > uint32_t action_flags =3D 0; > @@ -690,12 +970,64 @@ flow_tcf_validate(struct rte_eth_dev *dev, > case RTE_FLOW_ACTION_TYPE_OF_SET_VLAN_PCP: > action_flags |=3D MLX5_ACTION_OF_SET_VLAN_PCP; > break; > + case RTE_FLOW_ACTION_TYPE_SET_IPV4_SRC: > + action_flags |=3D MLX5_ACTION_SET_IPV4_SRC; > + break; > + case RTE_FLOW_ACTION_TYPE_SET_IPV4_DST: > + action_flags |=3D MLX5_ACTION_SET_IPV4_DST; > + break; > + case RTE_FLOW_ACTION_TYPE_SET_IPV6_SRC: > + action_flags |=3D MLX5_ACTION_SET_IPV6_SRC; > + break; > + case RTE_FLOW_ACTION_TYPE_SET_IPV6_DST: > + action_flags |=3D MLX5_ACTION_SET_IPV6_DST; > + break; > + case RTE_FLOW_ACTION_TYPE_SET_TP_SRC: > + action_flags |=3D MLX5_ACTION_SET_TP_SRC; > + break; > + case RTE_FLOW_ACTION_TYPE_SET_TP_DST: > + action_flags |=3D MLX5_ACTION_SET_TP_DST; > + break; > default: > return rte_flow_error_set(error, ENOTSUP, > RTE_FLOW_ERROR_TYPE_ACTION, > actions, > "action not supported"); > } > + if (IS_MODIFY_ACTION(actions->type)) { > + if (!actions->conf) > + return rte_flow_error_set(error, ENOTSUP, > + RTE_FLOW_ERROR_TYPE_ACTION_CONF, > + actions, > + "action configuration not set"); > + } > + } > + if (action_flags & > + (MLX5_ACTION_SET_IPV4_SRC | MLX5_ACTION_SET_IPV4_DST)) { > + if (!(item_flags & MLX5_FLOW_LAYER_OUTER_L3_IPV4)) > + return rte_flow_error_set(error, ENOTSUP, > + RTE_FLOW_ERROR_TYPE_ACTION, > + actions, > + "no ipv4 item found in" > + " pattern"); > + } > + if (action_flags & > + (MLX5_ACTION_SET_IPV6_SRC | MLX5_ACTION_SET_IPV6_DST)) { > + if (!(item_flags & MLX5_FLOW_LAYER_OUTER_L3_IPV6)) > + return rte_flow_error_set(error, ENOTSUP, > + RTE_FLOW_ERROR_TYPE_ACTION, > + actions, > + "no ipv6 item found in pattern"); > + } > + if (action_flags & (MLX5_ACTION_SET_TP_SRC | MLX5_ACTION_SET_TP_DST)) { > + if (!(item_flags & > + (MLX5_FLOW_LAYER_OUTER_L4_UDP | > + MLX5_FLOW_LAYER_OUTER_L4_TCP))) > + return rte_flow_error_set(error, ENOTSUP, > + RTE_FLOW_ERROR_TYPE_ACTION, > + actions, > + "no TCP/UDP item found in" > + " pattern"); Isn't this 'set' action compatible with drop action? No point of modifying packet which will be dropped, isn't it? > } > return 0; > } > @@ -840,6 +1172,15 @@ flow_tcf_get_actions_and_size(const struct rte_flow= _action actions[], > SZ_NLATTR_TYPE_OF(uint16_t) + /* VLAN ID. */ > SZ_NLATTR_TYPE_OF(uint8_t); /* VLAN prio. */ > break; > + case RTE_FLOW_ACTION_TYPE_SET_IPV4_SRC: > + case RTE_FLOW_ACTION_TYPE_SET_IPV4_DST: > + case RTE_FLOW_ACTION_TYPE_SET_IPV6_SRC: > + case RTE_FLOW_ACTION_TYPE_SET_IPV6_DST: > + case RTE_FLOW_ACTION_TYPE_SET_TP_SRC: > + case RTE_FLOW_ACTION_TYPE_SET_TP_DST: > + size +=3D flow_tcf_get_pedit_actions_size(&actions, > + &flags); > + break; > default: > DRV_LOG(WARNING, > "unsupported action %p type %d," > @@ -998,6 +1339,7 @@ flow_tcf_translate(struct rte_eth_dev *dev, struct m= lx5_flow *dev_flow, > struct nlattr *na_flower_act; > struct nlattr *na_vlan_id =3D NULL; > struct nlattr *na_vlan_priority =3D NULL; > + uint64_t item_flags =3D 0; > =20 > claim_nonzero(flow_tcf_build_ptoi_table(dev, ptoi, > PTOI_TABLE_SZ_MAX(dev))); > @@ -1189,6 +1531,7 @@ flow_tcf_translate(struct rte_eth_dev *dev, struct = mlx5_flow *dev_flow, > } > break; > case RTE_FLOW_ITEM_TYPE_UDP: > + item_flags |=3D MLX5_FLOW_LAYER_OUTER_L4_UDP; Let's add the same to the rest of items like flow_tcf_validate(). Thanks, Yongseok > mask.udp =3D flow_tcf_item_mask > (items, &rte_flow_item_udp_mask, > &flow_tcf_mask_supported.udp, > @@ -1218,6 +1561,7 @@ flow_tcf_translate(struct rte_eth_dev *dev, struct = mlx5_flow *dev_flow, > } > break; > case RTE_FLOW_ITEM_TYPE_TCP: > + item_flags |=3D MLX5_FLOW_LAYER_OUTER_L4_TCP; > mask.tcp =3D flow_tcf_item_mask > (items, &rte_flow_item_tcp_mask, > &flow_tcf_mask_supported.tcp, > @@ -1368,6 +1712,18 @@ flow_tcf_translate(struct rte_eth_dev *dev, struct= mlx5_flow *dev_flow, > conf.of_set_vlan_pcp->vlan_pcp; > } > break; > + case RTE_FLOW_ACTION_TYPE_SET_IPV4_SRC: > + case RTE_FLOW_ACTION_TYPE_SET_IPV4_DST: > + case RTE_FLOW_ACTION_TYPE_SET_IPV6_SRC: > + case RTE_FLOW_ACTION_TYPE_SET_IPV6_DST: > + case RTE_FLOW_ACTION_TYPE_SET_TP_SRC: > + case RTE_FLOW_ACTION_TYPE_SET_TP_DST: > + na_act_index =3D > + mnl_attr_nest_start(nlh, na_act_index_cur++); > + flow_tcf_create_pedit_mnl_msg(nlh, > + &actions, item_flags); > + mnl_attr_nest_end(nlh, na_act_index); > + break; > default: > return rte_flow_error_set(error, ENOTSUP, > RTE_FLOW_ERROR_TYPE_ACTION, > --=20 > 2.17.1 >=20