[{"id":1772672,"web_url":"http://patchwork.ozlabs.org/comment/1772672/","msgid":"","list_archive_url":null,"date":"2017-09-21T10:55:26","subject":"Re: [PATCH net-next 2/4] cxgb4: add basic tc flower offload support","submitter":{"id":71804,"url":"http://patchwork.ozlabs.org/api/people/71804/","name":"Yunsheng Lin","email":"linyunsheng@huawei.com"},"content":"Hi, Kumar\n\nOn 2017/9/21 15:33, Rahul Lakkireddy wrote:\n> From: Kumar Sanghvi \n> \n> Add support to add/remove flows for offload. Following match\n> and action are supported for offloading a flow:\n> \n> Match: ether-protocol, IPv4/IPv6 addresses, L4 ports (TCP/UDP)\n> Action: drop, redirect to another port on the device.\n> \n> The qualifying flows can have accompanying mask information.\n> \n> Signed-off-by: Kumar Sanghvi \n> Signed-off-by: Rahul Lakkireddy \n> Signed-off-by: Ganesh Goudar \n> ---\n> drivers/net/ethernet/chelsio/cxgb4/cxgb4.h | 3 +\n> drivers/net/ethernet/chelsio/cxgb4/cxgb4_filter.c | 26 ++\n> drivers/net/ethernet/chelsio/cxgb4/cxgb4_main.c | 2 +\n> .../net/ethernet/chelsio/cxgb4/cxgb4_tc_flower.c | 285 ++++++++++++++++++++-\n> .../net/ethernet/chelsio/cxgb4/cxgb4_tc_flower.h | 17 ++\n> drivers/net/ethernet/chelsio/cxgb4/cxgb4_uld.h | 1 +\n> 6 files changed, 332 insertions(+), 2 deletions(-)\n> \n> diff --git a/drivers/net/ethernet/chelsio/cxgb4/cxgb4.h b/drivers/net/ethernet/chelsio/cxgb4/cxgb4.h\n> index ea72d2d2e1b4..26eac599ab2c 100644\n> --- a/drivers/net/ethernet/chelsio/cxgb4/cxgb4.h\n> +++ b/drivers/net/ethernet/chelsio/cxgb4/cxgb4.h\n> @@ -904,6 +904,9 @@ struct adapter {\n> \t/* TC u32 offload */\n> \tstruct cxgb4_tc_u32_table *tc_u32;\n> \tstruct chcr_stats_debug chcr_stats;\n> +\n> +\t/* TC flower offload */\n> +\tDECLARE_HASHTABLE(flower_anymatch_tbl, 9);\n> };\n> \n> /* Support for \"sched-class\" command to allow a TX Scheduling Class to be\n> diff --git a/drivers/net/ethernet/chelsio/cxgb4/cxgb4_filter.c b/drivers/net/ethernet/chelsio/cxgb4/cxgb4_filter.c\n> index 45b5853ca2f1..07a4619e2164 100644\n> --- a/drivers/net/ethernet/chelsio/cxgb4/cxgb4_filter.c\n> +++ b/drivers/net/ethernet/chelsio/cxgb4/cxgb4_filter.c\n> @@ -148,6 +148,32 @@ static int get_filter_steerq(struct net_device *dev,\n> \treturn iq;\n> }\n> \n> +int cxgb4_get_free_ftid(struct net_device *dev, int family)\n> +{\n> +\tstruct adapter *adap = netdev2adap(dev);\n> +\tstruct tid_info *t = &adap->tids;\n> +\tint ftid;\n> +\n> +\tspin_lock_bh(&t->ftid_lock);\n> +\tif (family == PF_INET) {\n> +\t\tftid = find_first_zero_bit(t->ftid_bmap, t->nftids);\n> +\t\tif (ftid >= t->nftids)\n> +\t\t\tftid = -1;\n> +\t} else {\n> +\t\tftid = bitmap_find_free_region(t->ftid_bmap, t->nftids, 2);\n> +\t\tif (ftid < 0) {\n> +\t\t\tftid = -1;\n\nftid = -1 is not needed?\n\n> +\t\t\tgoto out_unlock;\n> +\t\t}\n> +\n> +\t\t/* this is only a lookup, keep the found region unallocated */\n> +\t\tbitmap_release_region(t->ftid_bmap, ftid, 2);\n> +\t}\n> +out_unlock:\n> +\tspin_unlock_bh(&t->ftid_lock);\n> +\treturn ftid;\n> +}\n> +\n> static int cxgb4_set_ftid(struct tid_info *t, int fidx, int family)\n> {\n> \tspin_lock_bh(&t->ftid_lock);\n> diff --git a/drivers/net/ethernet/chelsio/cxgb4/cxgb4_main.c b/drivers/net/ethernet/chelsio/cxgb4/cxgb4_main.c\n> index 8923affbdaf8..3ba4e1ff8486 100644\n> --- a/drivers/net/ethernet/chelsio/cxgb4/cxgb4_main.c\n> +++ b/drivers/net/ethernet/chelsio/cxgb4/cxgb4_main.c\n> @@ -5105,6 +5105,8 @@ static int init_one(struct pci_dev *pdev, const struct pci_device_id *ent)\n> \t\tif (!adapter->tc_u32)\n> \t\t\tdev_warn(&pdev->dev,\n> \t\t\t\t \"could not offload tc u32, continuing\\n\");\n> +\n> +\t\tcxgb4_init_tc_flower(adapter);\n> \t}\n> \n> \tif (is_offload(adapter)) {\n> diff --git a/drivers/net/ethernet/chelsio/cxgb4/cxgb4_tc_flower.c b/drivers/net/ethernet/chelsio/cxgb4/cxgb4_tc_flower.c\n> index 16dff71e4d02..1af01101faaf 100644\n> --- a/drivers/net/ethernet/chelsio/cxgb4/cxgb4_tc_flower.c\n> +++ b/drivers/net/ethernet/chelsio/cxgb4/cxgb4_tc_flower.c\n> @@ -38,16 +38,292 @@\n> #include \"cxgb4.h\"\n> #include \"cxgb4_tc_flower.h\"\n> \n> +static struct ch_tc_flower_entry *allocate_flower_entry(void)\n> +{\n> +\tstruct ch_tc_flower_entry *new = kzalloc(sizeof(*new), GFP_KERNEL);\n> +\treturn new;\n> +}\n> +\n> +/* Must be called with either RTNL or rcu_read_lock */\n> +static struct ch_tc_flower_entry *ch_flower_lookup(struct adapter *adap,\n> +\t\t\t\t\t\t unsigned long flower_cookie)\n> +{\n> +\tstruct ch_tc_flower_entry *flower_entry;\n> +\n> +\thash_for_each_possible_rcu(adap->flower_anymatch_tbl, flower_entry,\n> +\t\t\t\t link, flower_cookie)\n> +\t\tif (flower_entry->tc_flower_cookie == flower_cookie)\n> +\t\t\treturn flower_entry;\n> +\treturn NULL;\n> +}\n> +\n> +static void cxgb4_process_flow_match(struct net_device *dev,\n> +\t\t\t\t struct tc_cls_flower_offload *cls,\n> +\t\t\t\t struct ch_filter_specification *fs)\n> +{\n> +\tu16 addr_type = 0;\n> +\n> +\tif (dissector_uses_key(cls->dissector, FLOW_DISSECTOR_KEY_CONTROL)) {\n> +\t\tstruct flow_dissector_key_control *key =\n> +\t\t\tskb_flow_dissector_target(cls->dissector,\n> +\t\t\t\t\t\t FLOW_DISSECTOR_KEY_CONTROL,\n> +\t\t\t\t\t\t cls->key);\n> +\n> +\t\taddr_type = key->addr_type;\n> +\t}\n> +\n> +\tif (dissector_uses_key(cls->dissector, FLOW_DISSECTOR_KEY_BASIC)) {\n> +\t\tstruct flow_dissector_key_basic *key =\n> +\t\t\tskb_flow_dissector_target(cls->dissector,\n> +\t\t\t\t\t\t FLOW_DISSECTOR_KEY_BASIC,\n> +\t\t\t\t\t\t cls->key);\n> +\t\tstruct flow_dissector_key_basic *mask =\n> +\t\t\tskb_flow_dissector_target(cls->dissector,\n> +\t\t\t\t\t\t FLOW_DISSECTOR_KEY_BASIC,\n> +\t\t\t\t\t\t cls->mask);\n> +\t\tu16 ethtype_key = ntohs(key->n_proto);\n> +\t\tu16 ethtype_mask = ntohs(mask->n_proto);\n> +\n> +\t\tif (ethtype_key == ETH_P_ALL) {\n> +\t\t\tethtype_key = 0;\n> +\t\t\tethtype_mask = 0;\n> +\t\t}\n> +\n> +\t\tfs->val.ethtype = ethtype_key;\n> +\t\tfs->mask.ethtype = ethtype_mask;\n> +\t\tfs->val.proto = key->ip_proto;\n> +\t\tfs->mask.proto = mask->ip_proto;\n> +\t}\n> +\n> +\tif (addr_type == FLOW_DISSECTOR_KEY_IPV4_ADDRS) {\n> +\t\tstruct flow_dissector_key_ipv4_addrs *key =\n> +\t\t\tskb_flow_dissector_target(cls->dissector,\n> +\t\t\t\t\t\t FLOW_DISSECTOR_KEY_IPV4_ADDRS,\n> +\t\t\t\t\t\t cls->key);\n> +\t\tstruct flow_dissector_key_ipv4_addrs *mask =\n> +\t\t\tskb_flow_dissector_target(cls->dissector,\n> +\t\t\t\t\t\t FLOW_DISSECTOR_KEY_IPV4_ADDRS,\n> +\t\t\t\t\t\t cls->mask);\n> +\t\tfs->type = 0;\n> +\t\tmemcpy(&fs->val.lip[0], &key->dst, sizeof(key->dst));\n> +\t\tmemcpy(&fs->val.fip[0], &key->src, sizeof(key->src));\n> +\t\tmemcpy(&fs->mask.lip[0], &mask->dst, sizeof(mask->dst));\n> +\t\tmemcpy(&fs->mask.fip[0], &mask->src, sizeof(mask->src));\n> +\t}\n> +\n> +\tif (addr_type == FLOW_DISSECTOR_KEY_IPV6_ADDRS) {\n> +\t\tstruct flow_dissector_key_ipv6_addrs *key =\n> +\t\t\tskb_flow_dissector_target(cls->dissector,\n> +\t\t\t\t\t\t FLOW_DISSECTOR_KEY_IPV6_ADDRS,\n> +\t\t\t\t\t\t cls->key);\n> +\t\tstruct flow_dissector_key_ipv6_addrs *mask =\n> +\t\t\tskb_flow_dissector_target(cls->dissector,\n> +\t\t\t\t\t\t FLOW_DISSECTOR_KEY_IPV6_ADDRS,\n> +\t\t\t\t\t\t cls->mask);\n> +\n> +\t\tfs->type = 1;\n> +\t\tmemcpy(&fs->val.lip[0], key->dst.s6_addr, sizeof(key->dst));\n> +\t\tmemcpy(&fs->val.fip[0], key->src.s6_addr, sizeof(key->src));\n> +\t\tmemcpy(&fs->mask.lip[0], mask->dst.s6_addr, sizeof(mask->dst));\n> +\t\tmemcpy(&fs->mask.fip[0], mask->src.s6_addr, sizeof(mask->src));\n> +\t}\n> +\n> +\tif (dissector_uses_key(cls->dissector, FLOW_DISSECTOR_KEY_PORTS)) {\n> +\t\tstruct flow_dissector_key_ports *key, *mask;\n> +\n> +\t\tkey = skb_flow_dissector_target(cls->dissector,\n> +\t\t\t\t\t\tFLOW_DISSECTOR_KEY_PORTS,\n> +\t\t\t\t\t\tcls->key);\n> +\t\tmask = skb_flow_dissector_target(cls->dissector,\n> +\t\t\t\t\t\t FLOW_DISSECTOR_KEY_PORTS,\n> +\t\t\t\t\t\t cls->mask);\n> +\t\tfs->val.lport = cpu_to_be16(key->dst);\n> +\t\tfs->mask.lport = cpu_to_be16(mask->dst);\n> +\t\tfs->val.fport = cpu_to_be16(key->src);\n> +\t\tfs->mask.fport = cpu_to_be16(mask->src);\n> +\t}\n> +\n> +\t/* Match only packets coming from the ingress port where this\n> +\t * filter will be created.\n> +\t */\n> +\tfs->val.iport = netdev2pinfo(dev)->port_id;\n> +\tfs->mask.iport = ~0;\n> +}\n> +\n> +static int cxgb4_validate_flow_match(struct net_device *dev,\n> +\t\t\t\t struct tc_cls_flower_offload *cls)\n> +{\n> +\tif (cls->dissector->used_keys &\n> +\t ~(BIT(FLOW_DISSECTOR_KEY_CONTROL) |\n> +\t BIT(FLOW_DISSECTOR_KEY_BASIC) |\n> +\t BIT(FLOW_DISSECTOR_KEY_IPV4_ADDRS) |\n> +\t BIT(FLOW_DISSECTOR_KEY_IPV6_ADDRS) |\n> +\t BIT(FLOW_DISSECTOR_KEY_PORTS))) {\n> +\t\tnetdev_warn(dev, \"Unsupported key used: 0x%x\\n\",\n> +\t\t\t cls->dissector->used_keys);\n> +\t\treturn -EOPNOTSUPP;\n> +\t}\n> +\treturn 0;\n> +}\n> +\n> +static void cxgb4_process_flow_actions(struct net_device *in,\n> +\t\t\t\t struct tc_cls_flower_offload *cls,\n> +\t\t\t\t struct ch_filter_specification *fs)\n> +{\n> +\tconst struct tc_action *a;\n> +\tLIST_HEAD(actions);\n> +\n> +\ttcf_exts_to_list(cls->exts, &actions);\n> +\tlist_for_each_entry(a, &actions, list) {\n> +\t\tif (is_tcf_gact_shot(a)) {\n> +\t\t\tfs->action = FILTER_DROP;\n> +\t\t} else if (is_tcf_mirred_egress_redirect(a)) {\n> +\t\t\tint ifindex = tcf_mirred_ifindex(a);\n> +\t\t\tstruct net_device *out = __dev_get_by_index(dev_net(in),\n> +\t\t\t\t\t\t\t\t ifindex);\n> +\t\t\tstruct port_info *pi = netdev_priv(out);\n> +\n> +\t\t\tfs->action = FILTER_SWITCH;\n> +\t\t\tfs->eport = pi->port_id;\n> +\t\t}\n> +\t}\n> +}\n> +\n> +static int cxgb4_validate_flow_actions(struct net_device *dev,\n> +\t\t\t\t struct tc_cls_flower_offload *cls)\n> +{\n> +\tconst struct tc_action *a;\n> +\tLIST_HEAD(actions);\n> +\n> +\ttcf_exts_to_list(cls->exts, &actions);\n> +\tlist_for_each_entry(a, &actions, list) {\n> +\t\tif (is_tcf_gact_shot(a)) {\n> +\t\t\t/* Do nothing */\n> +\t\t} else if (is_tcf_mirred_egress_redirect(a)) {\n> +\t\t\tstruct adapter *adap = netdev2adap(dev);\n> +\t\t\tstruct net_device *n_dev;\n> +\t\t\tunsigned int i, ifindex;\n> +\t\t\tbool found = false;\n> +\n> +\t\t\tifindex = tcf_mirred_ifindex(a);\n> +\t\t\tfor_each_port(adap, i) {\n> +\t\t\t\tn_dev = adap->port[i];\n> +\t\t\t\tif (ifindex == n_dev->ifindex) {\n> +\t\t\t\t\tfound = true;\n> +\t\t\t\t\tbreak;\n> +\t\t\t\t}\n> +\t\t\t}\n> +\n> +\t\t\t/* If interface doesn't belong to our hw, then\n> +\t\t\t * the provided output port is not valid\n> +\t\t\t */\n> +\t\t\tif (!found) {\n> +\t\t\t\tnetdev_err(dev, \"%s: Out port invalid\\n\",\n> +\t\t\t\t\t __func__);\n> +\t\t\t\treturn -EINVAL;\n> +\t\t\t}\n> +\t\t} else {\n> +\t\t\tnetdev_err(dev, \"%s: Unsupported action\\n\", __func__);\n> +\t\t\treturn -EOPNOTSUPP;\n> +\t\t}\n> +\t}\n> +\treturn 0;\n> +}\n> +\n> int cxgb4_tc_flower_replace(struct net_device *dev,\n> \t\t\t struct tc_cls_flower_offload *cls)\n> {\n> -\treturn -EOPNOTSUPP;\n> +\tstruct adapter *adap = netdev2adap(dev);\n> +\tstruct ch_tc_flower_entry *ch_flower;\n> +\tstruct ch_filter_specification *fs;\n> +\tstruct filter_ctx ctx;\n> +\tint fidx;\n> +\tint ret;\n> +\n> +\tif (cxgb4_validate_flow_actions(dev, cls))\n> +\t\treturn -EOPNOTSUPP;\n> +\n> +\tif (cxgb4_validate_flow_match(dev, cls))\n> +\t\treturn -EOPNOTSUPP;\n> +\n> +\tch_flower = allocate_flower_entry();\n> +\tif (!ch_flower) {\n> +\t\tnetdev_err(dev, \"%s: ch_flower alloc failed.\\n\", __func__);\n> +\t\tret = -ENOMEM;\n> +\t\tgoto err;\n\nJust return, err label is needed?\n\n> +\t}\n> +\n> +\tfs = &ch_flower->fs;\n> +\tfs->hitcnts = 1;\n> +\tcxgb4_process_flow_actions(dev, cls, fs);\n> +\tcxgb4_process_flow_match(dev, cls, fs);\n> +\n> +\tfidx = cxgb4_get_free_ftid(dev, fs->type ? PF_INET6 : PF_INET);\n> +\tif (fidx < 0) {\n> +\t\tnetdev_err(dev, \"%s: No fidx for offload.\\n\", __func__);\n> +\t\tret = -ENOMEM;\n> +\t\tgoto free_entry;\n> +\t}\n> +\n> +\tinit_completion(&ctx.completion);\n> +\tret = __cxgb4_set_filter(dev, fidx, fs, &ctx);\n> +\tif (ret) {\n> +\t\tnetdev_err(dev, \"%s: filter creation err %d\\n\",\n> +\t\t\t __func__, ret);\n> +\t\tgoto free_entry;\n> +\t}\n> +\n> +\t/* Wait for reply */\n> +\tret = wait_for_completion_timeout(&ctx.completion, 10 * HZ);\n> +\tif (!ret) {\n> +\t\tret = -ETIMEDOUT;\n> +\t\tgoto free_entry;\n> +\t}\n> +\n> +\tret = ctx.result;\n> +\t/* Check if hw returned error for filter creation */\n> +\tif (ret) {\n> +\t\tnetdev_err(dev, \"%s: filter creation err %d\\n\",\n> +\t\t\t __func__, ret);\n> +\t\tgoto free_entry;\n> +\t}\n> +\n> +\tINIT_HLIST_NODE(&ch_flower->link);\n> +\tch_flower->tc_flower_cookie = cls->cookie;\n> +\tch_flower->filter_id = ctx.tid;\n> +\thash_add_rcu(adap->flower_anymatch_tbl, &ch_flower->link, cls->cookie);\n> +\n> +\treturn ret;\n> +\n> +free_entry:\n> +\tkfree(ch_flower);\n> +err:\n> +\treturn ret;\n> }\n> \n> int cxgb4_tc_flower_destroy(struct net_device *dev,\n> \t\t\t struct tc_cls_flower_offload *cls)\n> {\n> -\treturn -EOPNOTSUPP;\n> +\tstruct adapter *adap = netdev2adap(dev);\n> +\tstruct ch_tc_flower_entry *ch_flower;\n> +\tint ret;\n> +\n> +\tch_flower = ch_flower_lookup(adap, cls->cookie);\n> +\tif (!ch_flower) {\n> +\t\tret = -ENOENT;\n> +\t\tgoto err;\n\nSame as above\n\n> +\t}\n> +\n> +\tret = cxgb4_del_filter(dev, ch_flower->filter_id);\n> +\tif (ret)\n> +\t\tgoto err;\n> +\n> +\thash_del_rcu(&ch_flower->link);\n> +\tkfree_rcu(ch_flower, rcu);\n> +\treturn ret;\n> +\n> +err:\n> +\treturn ret;\n> }\n> \n> int cxgb4_tc_flower_stats(struct net_device *dev,\n> @@ -55,3 +331,8 @@ int cxgb4_tc_flower_stats(struct net_device *dev,\n> {\n> \treturn -EOPNOTSUPP;\n> }\n> +\n> +void cxgb4_init_tc_flower(struct adapter *adap)\n> +{\n> +\thash_init(adap->flower_anymatch_tbl);\n> +}\n> diff --git a/drivers/net/ethernet/chelsio/cxgb4/cxgb4_tc_flower.h b/drivers/net/ethernet/chelsio/cxgb4/cxgb4_tc_flower.h\n> index b321fc205b5a..6145a9e056eb 100644\n> --- a/drivers/net/ethernet/chelsio/cxgb4/cxgb4_tc_flower.h\n> +++ b/drivers/net/ethernet/chelsio/cxgb4/cxgb4_tc_flower.h\n> @@ -37,10 +37,27 @@\n> \n> #include \n> \n> +struct ch_tc_flower_stats {\n> +\tu64 packet_count;\n> +\tu64 byte_count;\n> +\tu64 last_used;\n> +};\n> +\n> +struct ch_tc_flower_entry {\n> +\tstruct ch_filter_specification fs;\n> +\tstruct ch_tc_flower_stats stats;\n> +\tunsigned long tc_flower_cookie;\n> +\tstruct hlist_node link;\n> +\tstruct rcu_head rcu;\n> +\tu32 filter_id;\n> +};\n> +\n> int cxgb4_tc_flower_replace(struct net_device *dev,\n> \t\t\t struct tc_cls_flower_offload *cls);\n> int cxgb4_tc_flower_destroy(struct net_device *dev,\n> \t\t\t struct tc_cls_flower_offload *cls);\n> int cxgb4_tc_flower_stats(struct net_device *dev,\n> \t\t\t struct tc_cls_flower_offload *cls);\n> +\n> +void cxgb4_init_tc_flower(struct adapter *adap);\n> #endif /* __CXGB4_TC_FLOWER_H */\n> diff --git a/drivers/net/ethernet/chelsio/cxgb4/cxgb4_uld.h b/drivers/net/ethernet/chelsio/cxgb4/cxgb4_uld.h\n> index 84541fce94c5..88487095d14f 100644\n> --- a/drivers/net/ethernet/chelsio/cxgb4/cxgb4_uld.h\n> +++ b/drivers/net/ethernet/chelsio/cxgb4/cxgb4_uld.h\n> @@ -212,6 +212,7 @@ struct filter_ctx {\n> \n> struct ch_filter_specification;\n> \n> +int cxgb4_get_free_ftid(struct net_device *dev, int family);\n> int __cxgb4_set_filter(struct net_device *dev, int filter_id,\n> \t\t struct ch_filter_specification *fs,\n> \t\t struct filter_ctx *ctx);\n>","headers":{"Return-Path":"","X-Original-To":"patchwork-incoming@ozlabs.org","Delivered-To":"patchwork-incoming@ozlabs.org","Authentication-Results":"ozlabs.org;\n\tspf=none (mailfrom) smtp.mailfrom=vger.kernel.org\n\t(client-ip=209.132.180.67; helo=vger.kernel.org;\n\tenvelope-from=netdev-owner@vger.kernel.org;\n\treceiver=)","Received":["from vger.kernel.org (vger.kernel.org [209.132.180.67])\n\tby ozlabs.org (Postfix) with ESMTP id 3xyYSM64tdz9t3Z\n\tfor ;\n\tThu, 21 Sep 2017 20:56:03 +1000 (AEST)","(majordomo@vger.kernel.org) by vger.kernel.org via listexpand\n\tid S1751675AbdIUK4B (ORCPT );\n\tThu, 21 Sep 2017 06:56:01 -0400","from szxga05-in.huawei.com ([45.249.212.191]:6534 \"EHLO\n\tszxga05-in.huawei.com\" rhost-flags-OK-OK-OK-OK) by vger.kernel.org\n\twith ESMTP id S1751387AbdIUK4A (ORCPT\n\t); Thu, 21 Sep 2017 06:56:00 -0400","from 172.30.72.59 (EHLO DGGEMS403-HUB.china.huawei.com)\n\t([172.30.72.59])\n\tby dggrg05-dlp.huawei.com (MOS 4.4.6-GA FastPath queued)\n\twith ESMTP id DHT45297; Thu, 21 Sep 2017 18:55:57 +0800 (CST)","from [127.0.0.1] (10.74.191.121) by DGGEMS403-HUB.china.huawei.com\n\t(10.3.19.203) with Microsoft SMTP Server id 14.3.301.0;\n\tThu, 21 Sep 2017 18:55:50 +0800"],"Subject":"Re: [PATCH net-next 2/4] cxgb4: add basic tc flower offload support","To":"Rahul Lakkireddy , ","CC":", ,\n\t, , ","References":"\n\t","From":"Yunsheng Lin ","Message-ID":"","Date":"Thu, 21 Sep 2017 18:55:26 +0800","User-Agent":"Mozilla/5.0 (Windows NT 6.1; WOW64; rv:52.0) Gecko/20100101\n\tThunderbird/52.2.0","MIME-Version":"1.0","In-Reply-To":"","Content-Type":"text/plain; charset=\"utf-8\"","Content-Language":"en-US","Content-Transfer-Encoding":"7bit","X-Originating-IP":"[10.74.191.121]","X-CFilter-Loop":"Reflected","X-Mirapoint-Virus-RAPID-Raw":"score=unknown(0),\n\trefid=str=0001.0A0B0206.59C39ABD.0179, ss=1, re=0.000, recu=0.000,\n\treip=0.000, cl=1, cld=1, fgs=0, ip=0.0.0.0,\n\tso=2014-11-16 11:51:01, dmn=2013-03-21 17:37:32","X-Mirapoint-Loop-Id":"62d93ca7b483030157da33f75d4934e8","Sender":"netdev-owner@vger.kernel.org","Precedence":"bulk","List-ID":"","X-Mailing-List":"netdev@vger.kernel.org"}},{"id":1772932,"web_url":"http://patchwork.ozlabs.org/comment/1772932/","msgid":"<20170921153153.GB14071@banjo.asicdesigners.com>","list_archive_url":null,"date":"2017-09-21T15:31:54","subject":"Re: [PATCH net-next 2/4] cxgb4: add basic tc flower offload support","submitter":{"id":9904,"url":"http://patchwork.ozlabs.org/api/people/9904/","name":"Kumar Sanghvi","email":"kumaras@chelsio.com"},"content":"Hi Yunsheng,\n\nOn Thursday, September 09/21/17, 2017 at 18:55:26 +0800, Yunsheng Lin wrote:\n> Hi, Kumar\n> \n> On 2017/9/21 15:33, Rahul Lakkireddy wrote:\n> > From: Kumar Sanghvi \n> > \n> > Add support to add/remove flows for offload. Following match\n> > and action are supported for offloading a flow:\n> > \n> > Match: ether-protocol, IPv4/IPv6 addresses, L4 ports (TCP/UDP)\n> > Action: drop, redirect to another port on the device.\n> > \n> > The qualifying flows can have accompanying mask information.\n> > \n> > Signed-off-by: Kumar Sanghvi \n> > Signed-off-by: Rahul Lakkireddy \n> > Signed-off-by: Ganesh Goudar \n> > ---\n\n[...]\n\n> > diff --git a/drivers/net/ethernet/chelsio/cxgb4/cxgb4_filter.c b/drivers/net/ethernet/chelsio/cxgb4/cxgb4_filter.c\n> > index 45b5853ca2f1..07a4619e2164 100644\n> > --- a/drivers/net/ethernet/chelsio/cxgb4/cxgb4_filter.c\n> > +++ b/drivers/net/ethernet/chelsio/cxgb4/cxgb4_filter.c\n> > @@ -148,6 +148,32 @@ static int get_filter_steerq(struct net_device *dev,\n> > \treturn iq;\n> > }\n> > \n> > +int cxgb4_get_free_ftid(struct net_device *dev, int family)\n> > +{\n> > +\tstruct adapter *adap = netdev2adap(dev);\n> > +\tstruct tid_info *t = &adap->tids;\n> > +\tint ftid;\n> > +\n> > +\tspin_lock_bh(&t->ftid_lock);\n> > +\tif (family == PF_INET) {\n> > +\t\tftid = find_first_zero_bit(t->ftid_bmap, t->nftids);\n> > +\t\tif (ftid >= t->nftids)\n> > +\t\t\tftid = -1;\n> > +\t} else {\n> > +\t\tftid = bitmap_find_free_region(t->ftid_bmap, t->nftids, 2);\n> > +\t\tif (ftid < 0) {\n> > +\t\t\tftid = -1;\n> \n> ftid = -1 is not needed?\n\nYou are right, its not needed. Thank you for pointing this.\nI will take care of this in V2.\n\n\n> \n> > +\t\t\tgoto out_unlock;\n> > +\t\t}\n> > +\n> > +\t\t/* this is only a lookup, keep the found region unallocated */\n> > +\t\tbitmap_release_region(t->ftid_bmap, ftid, 2);\n> > +\t}\n> > +out_unlock:\n> > +\tspin_unlock_bh(&t->ftid_lock);\n> > +\treturn ftid;\n> > +}\n\n[...]\n\n> > diff --git a/drivers/net/ethernet/chelsio/cxgb4/cxgb4_tc_flower.c b/drivers/net/ethernet/chelsio/cxgb4/cxgb4_tc_flower.c\n> > index 16dff71e4d02..1af01101faaf 100644\n> > --- a/drivers/net/ethernet/chelsio/cxgb4/cxgb4_tc_flower.c\n> > +++ b/drivers/net/ethernet/chelsio/cxgb4/cxgb4_tc_flower.c\n> > @@ -38,16 +38,292 @@\n> > #include \"cxgb4.h\"\n> > #include \"cxgb4_tc_flower.h\"\n> > \n\n[...]\n\n> > int cxgb4_tc_flower_replace(struct net_device *dev,\n> > \t\t\t struct tc_cls_flower_offload *cls)\n> > {\n> > -\treturn -EOPNOTSUPP;\n> > +\tstruct adapter *adap = netdev2adap(dev);\n> > +\tstruct ch_tc_flower_entry *ch_flower;\n> > +\tstruct ch_filter_specification *fs;\n> > +\tstruct filter_ctx ctx;\n> > +\tint fidx;\n> > +\tint ret;\n> > +\n> > +\tif (cxgb4_validate_flow_actions(dev, cls))\n> > +\t\treturn -EOPNOTSUPP;\n> > +\n> > +\tif (cxgb4_validate_flow_match(dev, cls))\n> > +\t\treturn -EOPNOTSUPP;\n> > +\n> > +\tch_flower = allocate_flower_entry();\n> > +\tif (!ch_flower) {\n> > +\t\tnetdev_err(dev, \"%s: ch_flower alloc failed.\\n\", __func__);\n> > +\t\tret = -ENOMEM;\n> > +\t\tgoto err;\n> \n> Just return, err label is needed?\n\nYes, err label is not needed. I will take care of this in V2.\n\n> \n> > +\t}\n> > +\n> > +\tfs = &ch_flower->fs;\n> > +\tfs->hitcnts = 1;\n> > +\tcxgb4_process_flow_actions(dev, cls, fs);\n> > +\tcxgb4_process_flow_match(dev, cls, fs);\n\n[...]\n\n\n> > int cxgb4_tc_flower_destroy(struct net_device *dev,\n> > \t\t\t struct tc_cls_flower_offload *cls)\n> > {\n> > -\treturn -EOPNOTSUPP;\n> > +\tstruct adapter *adap = netdev2adap(dev);\n> > +\tstruct ch_tc_flower_entry *ch_flower;\n> > +\tint ret;\n> > +\n> > +\tch_flower = ch_flower_lookup(adap, cls->cookie);\n> > +\tif (!ch_flower) {\n> > +\t\tret = -ENOENT;\n> > +\t\tgoto err;\n> \n> Same as above\n\nI will take care of this in V2.\nThank you.","headers":{"Return-Path":"","X-Original-To":"patchwork-incoming@ozlabs.org","Delivered-To":"patchwork-incoming@ozlabs.org","Authentication-Results":"ozlabs.org;\n\tspf=none (mailfrom) smtp.mailfrom=vger.kernel.org\n\t(client-ip=209.132.180.67; helo=vger.kernel.org;\n\tenvelope-from=netdev-owner@vger.kernel.org;\n\treceiver=)","Received":["from vger.kernel.org (vger.kernel.org [209.132.180.67])\n\tby ozlabs.org (Postfix) with ESMTP id 3xygb138tLz9t49\n\tfor ;\n\tFri, 22 Sep 2017 01:32:13 +1000 (AEST)","(majordomo@vger.kernel.org) by vger.kernel.org via listexpand\n\tid S1752007AbdIUPcK (ORCPT );\n\tThu, 21 Sep 2017 11:32:10 -0400","from stargate.chelsio.com ([12.32.117.8]:50532 \"EHLO\n\tstargate.chelsio.com\" rhost-flags-OK-OK-OK-OK) by vger.kernel.org\n\twith ESMTP id S1751808AbdIUPcI (ORCPT\n\t); Thu, 21 Sep 2017 11:32:08 -0400","from localhost (eureka2.asicdesigners.com [10.192.163.201] (may be\n\tforged))\n\tby stargate.chelsio.com (8.13.8/8.13.8) with ESMTP id v8LFVsbZ006554; \n\tThu, 21 Sep 2017 08:32:00 -0700"],"Date":"Thu, 21 Sep 2017 08:31:54 -0700","From":"Kumar Sanghvi ","To":"Yunsheng Lin ","Cc":"Rahul Lakkireddy ,\n\tnetdev@vger.kernel.org, davem@davemloft.net, ganeshgr@chelsio.com,\n\tnirranjan@chelsio.com, indranil@chelsio.com","Subject":"Re: [PATCH net-next 2/4] cxgb4: add basic tc flower offload support","Message-ID":"<20170921153153.GB14071@banjo.asicdesigners.com>","References":"\n\t\n\t","MIME-Version":"1.0","Content-Type":"text/plain; charset=us-ascii","Content-Disposition":"inline","In-Reply-To":"","User-Agent":"Mutt/1.8.3 (2017-05-23)","Sender":"netdev-owner@vger.kernel.org","Precedence":"bulk","List-ID":"","X-Mailing-List":"netdev@vger.kernel.org"}}]