offload.c 41.5 KB
Newer Older
1 2
// SPDX-License-Identifier: (GPL-2.0-only OR BSD-2-Clause)
/* Copyright (C) 2017-2018 Netronome Systems, Inc. */
3 4 5 6 7 8 9 10 11 12 13 14 15 16

#include <linux/skbuff.h>
#include <net/devlink.h>
#include <net/pkt_cls.h>

#include "cmsg.h"
#include "main.h"
#include "../nfpcore/nfp_cpp.h"
#include "../nfpcore/nfp_nsp.h"
#include "../nfp_app.h"
#include "../nfp_main.h"
#include "../nfp_net.h"
#include "../nfp_port.h"

17 18 19 20
#define NFP_FLOWER_SUPPORTED_TCPFLAGS \
	(TCPHDR_FIN | TCPHDR_SYN | TCPHDR_RST | \
	 TCPHDR_PSH | TCPHDR_URG)

21 22 23 24
#define NFP_FLOWER_SUPPORTED_CTLFLAGS \
	(FLOW_DIS_IS_FRAGMENT | \
	 FLOW_DIS_FIRST_FRAG)

25 26 27 28 29
#define NFP_FLOWER_WHITELIST_DISSECTOR \
	(BIT(FLOW_DISSECTOR_KEY_CONTROL) | \
	 BIT(FLOW_DISSECTOR_KEY_BASIC) | \
	 BIT(FLOW_DISSECTOR_KEY_IPV4_ADDRS) | \
	 BIT(FLOW_DISSECTOR_KEY_IPV6_ADDRS) | \
30
	 BIT(FLOW_DISSECTOR_KEY_TCP) | \
31 32 33
	 BIT(FLOW_DISSECTOR_KEY_PORTS) | \
	 BIT(FLOW_DISSECTOR_KEY_ETH_ADDRS) | \
	 BIT(FLOW_DISSECTOR_KEY_VLAN) | \
34 35 36 37 38
	 BIT(FLOW_DISSECTOR_KEY_ENC_KEYID) | \
	 BIT(FLOW_DISSECTOR_KEY_ENC_IPV4_ADDRS) | \
	 BIT(FLOW_DISSECTOR_KEY_ENC_IPV6_ADDRS) | \
	 BIT(FLOW_DISSECTOR_KEY_ENC_CONTROL) | \
	 BIT(FLOW_DISSECTOR_KEY_ENC_PORTS) | \
39
	 BIT(FLOW_DISSECTOR_KEY_ENC_OPTS) | \
40
	 BIT(FLOW_DISSECTOR_KEY_ENC_IP) | \
41
	 BIT(FLOW_DISSECTOR_KEY_MPLS) | \
42 43
	 BIT(FLOW_DISSECTOR_KEY_IP))

44 45 46 47 48
#define NFP_FLOWER_WHITELIST_TUN_DISSECTOR \
	(BIT(FLOW_DISSECTOR_KEY_ENC_CONTROL) | \
	 BIT(FLOW_DISSECTOR_KEY_ENC_KEYID) | \
	 BIT(FLOW_DISSECTOR_KEY_ENC_IPV4_ADDRS) | \
	 BIT(FLOW_DISSECTOR_KEY_ENC_IPV6_ADDRS) | \
49
	 BIT(FLOW_DISSECTOR_KEY_ENC_OPTS) | \
50 51
	 BIT(FLOW_DISSECTOR_KEY_ENC_PORTS) | \
	 BIT(FLOW_DISSECTOR_KEY_ENC_IP))
52 53 54

#define NFP_FLOWER_WHITELIST_TUN_DISSECTOR_R \
	(BIT(FLOW_DISSECTOR_KEY_ENC_CONTROL) | \
55
	 BIT(FLOW_DISSECTOR_KEY_ENC_IPV4_ADDRS))
56

57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78
#define NFP_FLOWER_MERGE_FIELDS \
	(NFP_FLOWER_LAYER_PORT | \
	 NFP_FLOWER_LAYER_MAC | \
	 NFP_FLOWER_LAYER_TP | \
	 NFP_FLOWER_LAYER_IPV4 | \
	 NFP_FLOWER_LAYER_IPV6)

struct nfp_flower_merge_check {
	union {
		struct {
			__be16 tci;
			struct nfp_flower_mac_mpls l2;
			struct nfp_flower_tp_ports l4;
			union {
				struct nfp_flower_ipv4 ipv4;
				struct nfp_flower_ipv6 ipv6;
			};
		};
		unsigned long vals[8];
	};
};

79
static int
80 81
nfp_flower_xmit_flow(struct nfp_app *app, struct nfp_fl_payload *nfp_flow,
		     u8 mtype)
82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100
{
	u32 meta_len, key_len, mask_len, act_len, tot_len;
	struct sk_buff *skb;
	unsigned char *msg;

	meta_len =  sizeof(struct nfp_fl_rule_metadata);
	key_len = nfp_flow->meta.key_len;
	mask_len = nfp_flow->meta.mask_len;
	act_len = nfp_flow->meta.act_len;

	tot_len = meta_len + key_len + mask_len + act_len;

	/* Convert to long words as firmware expects
	 * lengths in units of NFP_FL_LW_SIZ.
	 */
	nfp_flow->meta.key_len >>= NFP_FL_LW_SIZ;
	nfp_flow->meta.mask_len >>= NFP_FL_LW_SIZ;
	nfp_flow->meta.act_len >>= NFP_FL_LW_SIZ;

101
	skb = nfp_flower_cmsg_alloc(app, tot_len, mtype, GFP_KERNEL);
102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118
	if (!skb)
		return -ENOMEM;

	msg = nfp_flower_cmsg_get_data(skb);
	memcpy(msg, &nfp_flow->meta, meta_len);
	memcpy(&msg[meta_len], nfp_flow->unmasked_data, key_len);
	memcpy(&msg[meta_len + key_len], nfp_flow->mask_data, mask_len);
	memcpy(&msg[meta_len + key_len + mask_len],
	       nfp_flow->action_data, act_len);

	/* Convert back to bytes as software expects
	 * lengths in units of bytes.
	 */
	nfp_flow->meta.key_len <<= NFP_FL_LW_SIZ;
	nfp_flow->meta.mask_len <<= NFP_FL_LW_SIZ;
	nfp_flow->meta.act_len <<= NFP_FL_LW_SIZ;

119
	nfp_ctrl_tx(app->ctrl, skb);
120 121 122 123

	return 0;
}

124
static bool nfp_flower_check_higher_than_mac(struct flow_cls_offload *f)
125
{
126
	struct flow_rule *rule = flow_cls_offload_flow_rule(f);
127 128 129 130 131

	return flow_rule_match_key(rule, FLOW_DISSECTOR_KEY_IPV4_ADDRS) ||
	       flow_rule_match_key(rule, FLOW_DISSECTOR_KEY_IPV6_ADDRS) ||
	       flow_rule_match_key(rule, FLOW_DISSECTOR_KEY_PORTS) ||
	       flow_rule_match_key(rule, FLOW_DISSECTOR_KEY_ICMP);
132 133
}

134
static bool nfp_flower_check_higher_than_l3(struct flow_cls_offload *f)
135
{
136
	struct flow_rule *rule = flow_cls_offload_flow_rule(f);
137 138 139 140 141

	return flow_rule_match_key(rule, FLOW_DISSECTOR_KEY_PORTS) ||
	       flow_rule_match_key(rule, FLOW_DISSECTOR_KEY_ICMP);
}

142
static int
143
nfp_flower_calc_opt_layer(struct flow_dissector_key_enc_opts *enc_opts,
144 145
			  u32 *key_layer_two, int *key_size,
			  struct netlink_ext_ack *extack)
146
{
147
	if (enc_opts->len > NFP_FL_MAX_GENEVE_OPT_KEY) {
148
		NL_SET_ERR_MSG_MOD(extack, "unsupported offload: geneve options exceed maximum length");
149
		return -EOPNOTSUPP;
150
	}
151

152
	if (enc_opts->len > 0) {
153 154 155 156 157 158 159
		*key_layer_two |= NFP_FLOWER_LAYER2_GENEVE_OP;
		*key_size += sizeof(struct nfp_flower_geneve_options);
	}

	return 0;
}

160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210
static int
nfp_flower_calc_udp_tun_layer(struct flow_dissector_key_ports *enc_ports,
			      struct flow_dissector_key_enc_opts *enc_op,
			      u32 *key_layer_two, u8 *key_layer, int *key_size,
			      struct nfp_flower_priv *priv,
			      enum nfp_flower_tun_type *tun_type,
			      struct netlink_ext_ack *extack)
{
	int err;

	switch (enc_ports->dst) {
	case htons(IANA_VXLAN_UDP_PORT):
		*tun_type = NFP_FL_TUNNEL_VXLAN;
		*key_layer |= NFP_FLOWER_LAYER_VXLAN;
		*key_size += sizeof(struct nfp_flower_ipv4_udp_tun);

		if (enc_op) {
			NL_SET_ERR_MSG_MOD(extack, "unsupported offload: encap options not supported on vxlan tunnels");
			return -EOPNOTSUPP;
		}
		break;
	case htons(GENEVE_UDP_PORT):
		if (!(priv->flower_ext_feats & NFP_FL_FEATS_GENEVE)) {
			NL_SET_ERR_MSG_MOD(extack, "unsupported offload: loaded firmware does not support geneve offload");
			return -EOPNOTSUPP;
		}
		*tun_type = NFP_FL_TUNNEL_GENEVE;
		*key_layer |= NFP_FLOWER_LAYER_EXT_META;
		*key_size += sizeof(struct nfp_flower_ext_meta);
		*key_layer_two |= NFP_FLOWER_LAYER2_GENEVE;
		*key_size += sizeof(struct nfp_flower_ipv4_udp_tun);

		if (!enc_op)
			break;
		if (!(priv->flower_ext_feats & NFP_FL_FEATS_GENEVE_OPT)) {
			NL_SET_ERR_MSG_MOD(extack, "unsupported offload: loaded firmware does not support geneve option offload");
			return -EOPNOTSUPP;
		}
		err = nfp_flower_calc_opt_layer(enc_op, key_layer_two,
						key_size, extack);
		if (err)
			return err;
		break;
	default:
		NL_SET_ERR_MSG_MOD(extack, "unsupported offload: tunnel type unknown");
		return -EOPNOTSUPP;
	}

	return 0;
}

211
static int
212
nfp_flower_calculate_key_layers(struct nfp_app *app,
213
				struct net_device *netdev,
214
				struct nfp_fl_key_ls *ret_key_ls,
215
				struct flow_cls_offload *flow,
216 217
				enum nfp_flower_tun_type *tun_type,
				struct netlink_ext_ack *extack)
218
{
219
	struct flow_rule *rule = flow_cls_offload_flow_rule(flow);
220 221
	struct flow_dissector *dissector = rule->match.dissector;
	struct flow_match_basic basic = { NULL, NULL};
222
	struct nfp_flower_priv *priv = app->priv;
223 224 225
	u32 key_layer_two;
	u8 key_layer;
	int key_size;
226
	int err;
227

228 229
	if (dissector->used_keys & ~NFP_FLOWER_WHITELIST_DISSECTOR) {
		NL_SET_ERR_MSG_MOD(extack, "unsupported offload: match not supported");
230
		return -EOPNOTSUPP;
231
	}
232

233
	/* If any tun dissector is used then the required set must be used. */
234 235
	if (dissector->used_keys & NFP_FLOWER_WHITELIST_TUN_DISSECTOR &&
	    (dissector->used_keys & NFP_FLOWER_WHITELIST_TUN_DISSECTOR_R)
236 237
	    != NFP_FLOWER_WHITELIST_TUN_DISSECTOR_R) {
		NL_SET_ERR_MSG_MOD(extack, "unsupported offload: tunnel match not supported");
238
		return -EOPNOTSUPP;
239
	}
240 241

	key_layer_two = 0;
242
	key_layer = NFP_FLOWER_LAYER_PORT;
J
John Hurley 已提交
243
	key_size = sizeof(struct nfp_flower_meta_tci) +
244 245
		   sizeof(struct nfp_flower_in_port);

246 247
	if (flow_rule_match_key(rule, FLOW_DISSECTOR_KEY_ETH_ADDRS) ||
	    flow_rule_match_key(rule, FLOW_DISSECTOR_KEY_MPLS)) {
248 249 250
		key_layer |= NFP_FLOWER_LAYER_MAC;
		key_size += sizeof(struct nfp_flower_mac_mpls);
	}
251

252 253
	if (flow_rule_match_key(rule, FLOW_DISSECTOR_KEY_VLAN)) {
		struct flow_match_vlan vlan;
254

255
		flow_rule_match_vlan(rule, &vlan);
256
		if (!(priv->flower_ext_feats & NFP_FL_FEATS_VLAN_PCP) &&
257 258
		    vlan.key->vlan_priority) {
			NL_SET_ERR_MSG_MOD(extack, "unsupported offload: loaded firmware does not support VLAN PCP offload");
259
			return -EOPNOTSUPP;
260
		}
261 262
	}

263 264 265 266 267 268 269 270
	if (flow_rule_match_key(rule, FLOW_DISSECTOR_KEY_ENC_CONTROL)) {
		struct flow_match_enc_opts enc_op = { NULL, NULL };
		struct flow_match_ipv4_addrs ipv4_addrs;
		struct flow_match_control enc_ctl;
		struct flow_match_ports enc_ports;

		flow_rule_match_enc_control(rule, &enc_ctl);

271 272 273 274 275 276
		if (enc_ctl.mask->addr_type != 0xffff) {
			NL_SET_ERR_MSG_MOD(extack, "unsupported offload: wildcarded protocols on tunnels are not supported");
			return -EOPNOTSUPP;
		}
		if (enc_ctl.key->addr_type != FLOW_DISSECTOR_KEY_IPV4_ADDRS) {
			NL_SET_ERR_MSG_MOD(extack, "unsupported offload: only IPv4 tunnels are supported");
277
			return -EOPNOTSUPP;
278
		}
279 280

		/* These fields are already verified as used. */
281
		flow_rule_match_enc_ipv4_addrs(rule, &ipv4_addrs);
282 283
		if (ipv4_addrs.mask->dst != cpu_to_be32(~0)) {
			NL_SET_ERR_MSG_MOD(extack, "unsupported offload: only an exact match IPv4 destination address is supported");
284
			return -EOPNOTSUPP;
285
		}
286

287 288
		if (flow_rule_match_key(rule, FLOW_DISSECTOR_KEY_ENC_OPTS))
			flow_rule_match_enc_opts(rule, &enc_op);
289 290


291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331
		if (!flow_rule_match_key(rule, FLOW_DISSECTOR_KEY_ENC_PORTS)) {
			/* check if GRE, which has no enc_ports */
			if (netif_is_gretap(netdev)) {
				*tun_type = NFP_FL_TUNNEL_GRE;
				key_layer |= NFP_FLOWER_LAYER_EXT_META;
				key_size += sizeof(struct nfp_flower_ext_meta);
				key_layer_two |= NFP_FLOWER_LAYER2_GRE;
				key_size +=
					sizeof(struct nfp_flower_ipv4_gre_tun);

				if (enc_op.key) {
					NL_SET_ERR_MSG_MOD(extack, "unsupported offload: encap options not supported on GRE tunnels");
					return -EOPNOTSUPP;
				}
			} else {
				NL_SET_ERR_MSG_MOD(extack, "unsupported offload: an exact match on L4 destination port is required for non-GRE tunnels");
				return -EOPNOTSUPP;
			}
		} else {
			flow_rule_match_enc_ports(rule, &enc_ports);
			if (enc_ports.mask->dst != cpu_to_be16(~0)) {
				NL_SET_ERR_MSG_MOD(extack, "unsupported offload: only an exact match L4 destination port is supported");
				return -EOPNOTSUPP;
			}

			err = nfp_flower_calc_udp_tun_layer(enc_ports.key,
							    enc_op.key,
							    &key_layer_two,
							    &key_layer,
							    &key_size, priv,
							    tun_type, extack);
			if (err)
				return err;

			/* Ensure the ingress netdev matches the expected
			 * tun type.
			 */
			if (!nfp_fl_netdev_is_tunnel_type(netdev, *tun_type)) {
				NL_SET_ERR_MSG_MOD(extack, "unsupported offload: ingress netdev does not match the expected tunnel type");
				return -EOPNOTSUPP;
			}
332
		}
333
	}
334

335 336
	if (flow_rule_match_key(rule, FLOW_DISSECTOR_KEY_BASIC))
		flow_rule_match_basic(rule, &basic);
337

338
	if (basic.mask && basic.mask->n_proto) {
339
		/* Ethernet type is present in the key. */
340
		switch (basic.key->n_proto) {
341 342 343 344 345 346 347 348 349 350 351 352 353 354
		case cpu_to_be16(ETH_P_IP):
			key_layer |= NFP_FLOWER_LAYER_IPV4;
			key_size += sizeof(struct nfp_flower_ipv4);
			break;

		case cpu_to_be16(ETH_P_IPV6):
			key_layer |= NFP_FLOWER_LAYER_IPV6;
			key_size += sizeof(struct nfp_flower_ipv6);
			break;

		/* Currently we do not offload ARP
		 * because we rely on it to get to the host.
		 */
		case cpu_to_be16(ETH_P_ARP):
355
			NL_SET_ERR_MSG_MOD(extack, "unsupported offload: ARP not supported");
356 357
			return -EOPNOTSUPP;

358 359 360 361 362 363 364 365
		case cpu_to_be16(ETH_P_MPLS_UC):
		case cpu_to_be16(ETH_P_MPLS_MC):
			if (!(key_layer & NFP_FLOWER_LAYER_MAC)) {
				key_layer |= NFP_FLOWER_LAYER_MAC;
				key_size += sizeof(struct nfp_flower_mac_mpls);
			}
			break;

366 367 368 369 370
		/* Will be included in layer 2. */
		case cpu_to_be16(ETH_P_8021Q):
			break;

		default:
371 372
			NL_SET_ERR_MSG_MOD(extack, "unsupported offload: match on given EtherType is not supported");
			return -EOPNOTSUPP;
373
		}
374 375 376
	} else if (nfp_flower_check_higher_than_mac(flow)) {
		NL_SET_ERR_MSG_MOD(extack, "unsupported offload: cannot match above L2 without specified EtherType");
		return -EOPNOTSUPP;
377 378
	}

379 380
	if (basic.mask && basic.mask->ip_proto) {
		switch (basic.key->ip_proto) {
381 382 383 384 385 386 387 388 389 390 391 392
		case IPPROTO_TCP:
		case IPPROTO_UDP:
		case IPPROTO_SCTP:
		case IPPROTO_ICMP:
		case IPPROTO_ICMPV6:
			key_layer |= NFP_FLOWER_LAYER_TP;
			key_size += sizeof(struct nfp_flower_tp_ports);
			break;
		default:
			/* Other ip proto - we need check the masks for the
			 * remainder of the key to ensure we can offload.
			 */
393 394
			if (nfp_flower_check_higher_than_l3(flow)) {
				NL_SET_ERR_MSG_MOD(extack, "unsupported offload: unknown IP protocol with L4 matches not supported");
395
				return -EOPNOTSUPP;
396
			}
397
			break;
398 399 400
		}
	}

401 402
	if (flow_rule_match_key(rule, FLOW_DISSECTOR_KEY_TCP)) {
		struct flow_match_tcp tcp;
403 404
		u32 tcp_flags;

405 406
		flow_rule_match_tcp(rule, &tcp);
		tcp_flags = be16_to_cpu(tcp.key->flags);
407

408 409
		if (tcp_flags & ~NFP_FLOWER_SUPPORTED_TCPFLAGS) {
			NL_SET_ERR_MSG_MOD(extack, "unsupported offload: no match support for selected TCP flags");
410
			return -EOPNOTSUPP;
411
		}
412 413 414 415 416

		/* We only support PSH and URG flags when either
		 * FIN, SYN or RST is present as well.
		 */
		if ((tcp_flags & (TCPHDR_PSH | TCPHDR_URG)) &&
417 418
		    !(tcp_flags & (TCPHDR_FIN | TCPHDR_SYN | TCPHDR_RST))) {
			NL_SET_ERR_MSG_MOD(extack, "unsupported offload: PSH and URG is only supported when used with FIN, SYN or RST");
419
			return -EOPNOTSUPP;
420
		}
421

422 423 424
		/* We need to store TCP flags in the either the IPv4 or IPv6 key
		 * space, thus we need to ensure we include a IPv4/IPv6 key
		 * layer if we have not done so already.
425
		 */
426 427
		if (!basic.key) {
			NL_SET_ERR_MSG_MOD(extack, "unsupported offload: match on TCP flags requires a match on L3 protocol");
428
			return -EOPNOTSUPP;
429
		}
430 431 432

		if (!(key_layer & NFP_FLOWER_LAYER_IPV4) &&
		    !(key_layer & NFP_FLOWER_LAYER_IPV6)) {
433
			switch (basic.key->n_proto) {
434 435 436 437 438 439
			case cpu_to_be16(ETH_P_IP):
				key_layer |= NFP_FLOWER_LAYER_IPV4;
				key_size += sizeof(struct nfp_flower_ipv4);
				break;

			case cpu_to_be16(ETH_P_IPV6):
J
John Hurley 已提交
440
					key_layer |= NFP_FLOWER_LAYER_IPV6;
441 442 443 444
				key_size += sizeof(struct nfp_flower_ipv6);
				break;

			default:
445
				NL_SET_ERR_MSG_MOD(extack, "unsupported offload: match on TCP flags requires a match on IPv4/IPv6");
446 447
				return -EOPNOTSUPP;
			}
448 449 450
		}
	}

451 452
	if (flow_rule_match_key(rule, FLOW_DISSECTOR_KEY_CONTROL)) {
		struct flow_match_control ctl;
453

454
		flow_rule_match_control(rule, &ctl);
455 456
		if (ctl.key->flags & ~NFP_FLOWER_SUPPORTED_CTLFLAGS) {
			NL_SET_ERR_MSG_MOD(extack, "unsupported offload: match on unknown control flag");
457
			return -EOPNOTSUPP;
458
		}
459 460
	}

461 462 463 464 465 466 467 468
	ret_key_ls->key_layer = key_layer;
	ret_key_ls->key_layer_two = key_layer_two;
	ret_key_ls->key_size = key_size;

	return 0;
}

static struct nfp_fl_payload *
469
nfp_flower_allocate_new(struct nfp_fl_key_ls *key_layer)
470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486
{
	struct nfp_fl_payload *flow_pay;

	flow_pay = kmalloc(sizeof(*flow_pay), GFP_KERNEL);
	if (!flow_pay)
		return NULL;

	flow_pay->meta.key_len = key_layer->key_size;
	flow_pay->unmasked_data = kmalloc(key_layer->key_size, GFP_KERNEL);
	if (!flow_pay->unmasked_data)
		goto err_free_flow;

	flow_pay->meta.mask_len = key_layer->key_size;
	flow_pay->mask_data = kmalloc(key_layer->key_size, GFP_KERNEL);
	if (!flow_pay->mask_data)
		goto err_free_unmasked;

487 488 489 490
	flow_pay->action_data = kmalloc(NFP_FL_MAX_A_SIZ, GFP_KERNEL);
	if (!flow_pay->action_data)
		goto err_free_mask;

491
	flow_pay->nfp_tun_ipv4_addr = 0;
492
	flow_pay->meta.flags = 0;
493
	INIT_LIST_HEAD(&flow_pay->linked_flows);
J
John Hurley 已提交
494
	flow_pay->in_hw = false;
495

496 497
	return flow_pay;

498 499
err_free_mask:
	kfree(flow_pay->mask_data);
500 501 502 503 504 505 506
err_free_unmasked:
	kfree(flow_pay->unmasked_data);
err_free_flow:
	kfree(flow_pay);
	return NULL;
}

507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706
static int
nfp_flower_update_merge_with_actions(struct nfp_fl_payload *flow,
				     struct nfp_flower_merge_check *merge,
				     u8 *last_act_id, int *act_out)
{
	struct nfp_fl_set_ipv6_tc_hl_fl *ipv6_tc_hl_fl;
	struct nfp_fl_set_ip4_ttl_tos *ipv4_ttl_tos;
	struct nfp_fl_set_ip4_addrs *ipv4_add;
	struct nfp_fl_set_ipv6_addr *ipv6_add;
	struct nfp_fl_push_vlan *push_vlan;
	struct nfp_fl_set_tport *tport;
	struct nfp_fl_set_eth *eth;
	struct nfp_fl_act_head *a;
	unsigned int act_off = 0;
	u8 act_id = 0;
	u8 *ports;
	int i;

	while (act_off < flow->meta.act_len) {
		a = (struct nfp_fl_act_head *)&flow->action_data[act_off];
		act_id = a->jump_id;

		switch (act_id) {
		case NFP_FL_ACTION_OPCODE_OUTPUT:
			if (act_out)
				(*act_out)++;
			break;
		case NFP_FL_ACTION_OPCODE_PUSH_VLAN:
			push_vlan = (struct nfp_fl_push_vlan *)a;
			if (push_vlan->vlan_tci)
				merge->tci = cpu_to_be16(0xffff);
			break;
		case NFP_FL_ACTION_OPCODE_POP_VLAN:
			merge->tci = cpu_to_be16(0);
			break;
		case NFP_FL_ACTION_OPCODE_SET_IPV4_TUNNEL:
			/* New tunnel header means l2 to l4 can be matched. */
			eth_broadcast_addr(&merge->l2.mac_dst[0]);
			eth_broadcast_addr(&merge->l2.mac_src[0]);
			memset(&merge->l4, 0xff,
			       sizeof(struct nfp_flower_tp_ports));
			memset(&merge->ipv4, 0xff,
			       sizeof(struct nfp_flower_ipv4));
			break;
		case NFP_FL_ACTION_OPCODE_SET_ETHERNET:
			eth = (struct nfp_fl_set_eth *)a;
			for (i = 0; i < ETH_ALEN; i++)
				merge->l2.mac_dst[i] |= eth->eth_addr_mask[i];
			for (i = 0; i < ETH_ALEN; i++)
				merge->l2.mac_src[i] |=
					eth->eth_addr_mask[ETH_ALEN + i];
			break;
		case NFP_FL_ACTION_OPCODE_SET_IPV4_ADDRS:
			ipv4_add = (struct nfp_fl_set_ip4_addrs *)a;
			merge->ipv4.ipv4_src |= ipv4_add->ipv4_src_mask;
			merge->ipv4.ipv4_dst |= ipv4_add->ipv4_dst_mask;
			break;
		case NFP_FL_ACTION_OPCODE_SET_IPV4_TTL_TOS:
			ipv4_ttl_tos = (struct nfp_fl_set_ip4_ttl_tos *)a;
			merge->ipv4.ip_ext.ttl |= ipv4_ttl_tos->ipv4_ttl_mask;
			merge->ipv4.ip_ext.tos |= ipv4_ttl_tos->ipv4_tos_mask;
			break;
		case NFP_FL_ACTION_OPCODE_SET_IPV6_SRC:
			ipv6_add = (struct nfp_fl_set_ipv6_addr *)a;
			for (i = 0; i < 4; i++)
				merge->ipv6.ipv6_src.in6_u.u6_addr32[i] |=
					ipv6_add->ipv6[i].mask;
			break;
		case NFP_FL_ACTION_OPCODE_SET_IPV6_DST:
			ipv6_add = (struct nfp_fl_set_ipv6_addr *)a;
			for (i = 0; i < 4; i++)
				merge->ipv6.ipv6_dst.in6_u.u6_addr32[i] |=
					ipv6_add->ipv6[i].mask;
			break;
		case NFP_FL_ACTION_OPCODE_SET_IPV6_TC_HL_FL:
			ipv6_tc_hl_fl = (struct nfp_fl_set_ipv6_tc_hl_fl *)a;
			merge->ipv6.ip_ext.ttl |=
				ipv6_tc_hl_fl->ipv6_hop_limit_mask;
			merge->ipv6.ip_ext.tos |= ipv6_tc_hl_fl->ipv6_tc_mask;
			merge->ipv6.ipv6_flow_label_exthdr |=
				ipv6_tc_hl_fl->ipv6_label_mask;
			break;
		case NFP_FL_ACTION_OPCODE_SET_UDP:
		case NFP_FL_ACTION_OPCODE_SET_TCP:
			tport = (struct nfp_fl_set_tport *)a;
			ports = (u8 *)&merge->l4.port_src;
			for (i = 0; i < 4; i++)
				ports[i] |= tport->tp_port_mask[i];
			break;
		case NFP_FL_ACTION_OPCODE_PRE_TUNNEL:
		case NFP_FL_ACTION_OPCODE_PRE_LAG:
		case NFP_FL_ACTION_OPCODE_PUSH_GENEVE:
			break;
		default:
			return -EOPNOTSUPP;
		}

		act_off += a->len_lw << NFP_FL_LW_SIZ;
	}

	if (last_act_id)
		*last_act_id = act_id;

	return 0;
}

static int
nfp_flower_populate_merge_match(struct nfp_fl_payload *flow,
				struct nfp_flower_merge_check *merge,
				bool extra_fields)
{
	struct nfp_flower_meta_tci *meta_tci;
	u8 *mask = flow->mask_data;
	u8 key_layer, match_size;

	memset(merge, 0, sizeof(struct nfp_flower_merge_check));

	meta_tci = (struct nfp_flower_meta_tci *)mask;
	key_layer = meta_tci->nfp_flow_key_layer;

	if (key_layer & ~NFP_FLOWER_MERGE_FIELDS && !extra_fields)
		return -EOPNOTSUPP;

	merge->tci = meta_tci->tci;
	mask += sizeof(struct nfp_flower_meta_tci);

	if (key_layer & NFP_FLOWER_LAYER_EXT_META)
		mask += sizeof(struct nfp_flower_ext_meta);

	mask += sizeof(struct nfp_flower_in_port);

	if (key_layer & NFP_FLOWER_LAYER_MAC) {
		match_size = sizeof(struct nfp_flower_mac_mpls);
		memcpy(&merge->l2, mask, match_size);
		mask += match_size;
	}

	if (key_layer & NFP_FLOWER_LAYER_TP) {
		match_size = sizeof(struct nfp_flower_tp_ports);
		memcpy(&merge->l4, mask, match_size);
		mask += match_size;
	}

	if (key_layer & NFP_FLOWER_LAYER_IPV4) {
		match_size = sizeof(struct nfp_flower_ipv4);
		memcpy(&merge->ipv4, mask, match_size);
	}

	if (key_layer & NFP_FLOWER_LAYER_IPV6) {
		match_size = sizeof(struct nfp_flower_ipv6);
		memcpy(&merge->ipv6, mask, match_size);
	}

	return 0;
}

static int
nfp_flower_can_merge(struct nfp_fl_payload *sub_flow1,
		     struct nfp_fl_payload *sub_flow2)
{
	/* Two flows can be merged if sub_flow2 only matches on bits that are
	 * either matched by sub_flow1 or set by a sub_flow1 action. This
	 * ensures that every packet that hits sub_flow1 and recirculates is
	 * guaranteed to hit sub_flow2.
	 */
	struct nfp_flower_merge_check sub_flow1_merge, sub_flow2_merge;
	int err, act_out = 0;
	u8 last_act_id = 0;

	err = nfp_flower_populate_merge_match(sub_flow1, &sub_flow1_merge,
					      true);
	if (err)
		return err;

	err = nfp_flower_populate_merge_match(sub_flow2, &sub_flow2_merge,
					      false);
	if (err)
		return err;

	err = nfp_flower_update_merge_with_actions(sub_flow1, &sub_flow1_merge,
						   &last_act_id, &act_out);
	if (err)
		return err;

	/* Must only be 1 output action and it must be the last in sequence. */
	if (act_out != 1 || last_act_id != NFP_FL_ACTION_OPCODE_OUTPUT)
		return -EOPNOTSUPP;

	/* Reject merge if sub_flow2 matches on something that is not matched
	 * on or set in an action by sub_flow1.
	 */
	err = bitmap_andnot(sub_flow2_merge.vals, sub_flow2_merge.vals,
			    sub_flow1_merge.vals,
			    sizeof(struct nfp_flower_merge_check) * 8);
	if (err)
		return -EINVAL;

	return 0;
}

707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723
static unsigned int
nfp_flower_copy_pre_actions(char *act_dst, char *act_src, int len,
			    bool *tunnel_act)
{
	unsigned int act_off = 0, act_len;
	struct nfp_fl_act_head *a;
	u8 act_id = 0;

	while (act_off < len) {
		a = (struct nfp_fl_act_head *)&act_src[act_off];
		act_len = a->len_lw << NFP_FL_LW_SIZ;
		act_id = a->jump_id;

		switch (act_id) {
		case NFP_FL_ACTION_OPCODE_PRE_TUNNEL:
			if (tunnel_act)
				*tunnel_act = true;
724
			/* fall through */
725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754 755 756 757 758 759 760 761 762 763 764 765 766 767 768 769 770 771 772 773 774 775 776 777 778 779 780 781 782 783 784 785 786 787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 802 803 804 805 806 807 808 809 810 811 812 813
		case NFP_FL_ACTION_OPCODE_PRE_LAG:
			memcpy(act_dst + act_off, act_src + act_off, act_len);
			break;
		default:
			return act_off;
		}

		act_off += act_len;
	}

	return act_off;
}

static int nfp_fl_verify_post_tun_acts(char *acts, int len)
{
	struct nfp_fl_act_head *a;
	unsigned int act_off = 0;

	while (act_off < len) {
		a = (struct nfp_fl_act_head *)&acts[act_off];
		if (a->jump_id != NFP_FL_ACTION_OPCODE_OUTPUT)
			return -EOPNOTSUPP;

		act_off += a->len_lw << NFP_FL_LW_SIZ;
	}

	return 0;
}

static int
nfp_flower_merge_action(struct nfp_fl_payload *sub_flow1,
			struct nfp_fl_payload *sub_flow2,
			struct nfp_fl_payload *merge_flow)
{
	unsigned int sub1_act_len, sub2_act_len, pre_off1, pre_off2;
	bool tunnel_act = false;
	char *merge_act;
	int err;

	/* The last action of sub_flow1 must be output - do not merge this. */
	sub1_act_len = sub_flow1->meta.act_len - sizeof(struct nfp_fl_output);
	sub2_act_len = sub_flow2->meta.act_len;

	if (!sub2_act_len)
		return -EINVAL;

	if (sub1_act_len + sub2_act_len > NFP_FL_MAX_A_SIZ)
		return -EINVAL;

	/* A shortcut can only be applied if there is a single action. */
	if (sub1_act_len)
		merge_flow->meta.shortcut = cpu_to_be32(NFP_FL_SC_ACT_NULL);
	else
		merge_flow->meta.shortcut = sub_flow2->meta.shortcut;

	merge_flow->meta.act_len = sub1_act_len + sub2_act_len;
	merge_act = merge_flow->action_data;

	/* Copy any pre-actions to the start of merge flow action list. */
	pre_off1 = nfp_flower_copy_pre_actions(merge_act,
					       sub_flow1->action_data,
					       sub1_act_len, &tunnel_act);
	merge_act += pre_off1;
	sub1_act_len -= pre_off1;
	pre_off2 = nfp_flower_copy_pre_actions(merge_act,
					       sub_flow2->action_data,
					       sub2_act_len, NULL);
	merge_act += pre_off2;
	sub2_act_len -= pre_off2;

	/* FW does a tunnel push when egressing, therefore, if sub_flow 1 pushes
	 * a tunnel, sub_flow 2 can only have output actions for a valid merge.
	 */
	if (tunnel_act) {
		char *post_tun_acts = &sub_flow2->action_data[pre_off2];

		err = nfp_fl_verify_post_tun_acts(post_tun_acts, sub2_act_len);
		if (err)
			return err;
	}

	/* Copy remaining actions from sub_flows 1 and 2. */
	memcpy(merge_act, sub_flow1->action_data + pre_off1, sub1_act_len);
	merge_act += sub1_act_len;
	memcpy(merge_act, sub_flow2->action_data + pre_off2, sub2_act_len);

	return 0;
}

814 815 816 817 818 819 820 821 822 823 824 825 826 827 828 829 830 831 832 833 834 835 836 837 838 839 840 841 842 843 844 845 846 847 848 849 850
/* Flow link code should only be accessed under RTNL. */
static void nfp_flower_unlink_flow(struct nfp_fl_payload_link *link)
{
	list_del(&link->merge_flow.list);
	list_del(&link->sub_flow.list);
	kfree(link);
}

static void nfp_flower_unlink_flows(struct nfp_fl_payload *merge_flow,
				    struct nfp_fl_payload *sub_flow)
{
	struct nfp_fl_payload_link *link;

	list_for_each_entry(link, &merge_flow->linked_flows, merge_flow.list)
		if (link->sub_flow.flow == sub_flow) {
			nfp_flower_unlink_flow(link);
			return;
		}
}

static int nfp_flower_link_flows(struct nfp_fl_payload *merge_flow,
				 struct nfp_fl_payload *sub_flow)
{
	struct nfp_fl_payload_link *link;

	link = kmalloc(sizeof(*link), GFP_KERNEL);
	if (!link)
		return -ENOMEM;

	link->merge_flow.flow = merge_flow;
	list_add_tail(&link->merge_flow.list, &merge_flow->linked_flows);
	link->sub_flow.flow = sub_flow;
	list_add_tail(&link->sub_flow.list, &sub_flow->linked_flows);

	return 0;
}

851 852 853 854 855 856 857 858 859 860 861 862 863 864 865
/**
 * nfp_flower_merge_offloaded_flows() - Merge 2 existing flows to single flow.
 * @app:	Pointer to the APP handle
 * @sub_flow1:	Initial flow matched to produce merge hint
 * @sub_flow2:	Post recirculation flow matched in merge hint
 *
 * Combines 2 flows (if valid) to a single flow, removing the initial from hw
 * and offloading the new, merged flow.
 *
 * Return: negative value on error, 0 in success.
 */
int nfp_flower_merge_offloaded_flows(struct nfp_app *app,
				     struct nfp_fl_payload *sub_flow1,
				     struct nfp_fl_payload *sub_flow2)
{
866
	struct flow_cls_offload merge_tc_off;
J
John Hurley 已提交
867
	struct nfp_flower_priv *priv = app->priv;
868
	struct netlink_ext_ack *extack = NULL;
869 870
	struct nfp_fl_payload *merge_flow;
	struct nfp_fl_key_ls merge_key_ls;
871 872
	int err;

873 874
	ASSERT_RTNL();

875
	extack = merge_tc_off.common.extack;
876 877 878 879 880
	if (sub_flow1 == sub_flow2 ||
	    nfp_flower_is_merge_flow(sub_flow1) ||
	    nfp_flower_is_merge_flow(sub_flow2))
		return -EINVAL;

881 882 883 884
	err = nfp_flower_can_merge(sub_flow1, sub_flow2);
	if (err)
		return err;

885 886 887 888 889 890 891 892 893 894 895 896 897 898 899 900 901 902
	merge_key_ls.key_size = sub_flow1->meta.key_len;

	merge_flow = nfp_flower_allocate_new(&merge_key_ls);
	if (!merge_flow)
		return -ENOMEM;

	merge_flow->tc_flower_cookie = (unsigned long)merge_flow;
	merge_flow->ingress_dev = sub_flow1->ingress_dev;

	memcpy(merge_flow->unmasked_data, sub_flow1->unmasked_data,
	       sub_flow1->meta.key_len);
	memcpy(merge_flow->mask_data, sub_flow1->mask_data,
	       sub_flow1->meta.mask_len);

	err = nfp_flower_merge_action(sub_flow1, sub_flow2, merge_flow);
	if (err)
		goto err_destroy_merge_flow;

903 904 905 906 907 908 909 910
	err = nfp_flower_link_flows(merge_flow, sub_flow1);
	if (err)
		goto err_destroy_merge_flow;

	err = nfp_flower_link_flows(merge_flow, sub_flow2);
	if (err)
		goto err_unlink_sub_flow1;

J
John Hurley 已提交
911 912
	merge_tc_off.cookie = merge_flow->tc_flower_cookie;
	err = nfp_compile_flow_metadata(app, &merge_tc_off, merge_flow,
913
					merge_flow->ingress_dev, extack);
J
John Hurley 已提交
914 915
	if (err)
		goto err_unlink_sub_flow2;
916

J
John Hurley 已提交
917 918 919 920 921 922 923 924 925 926 927 928 929 930 931 932 933 934 935 936 937 938
	err = rhashtable_insert_fast(&priv->flow_table, &merge_flow->fl_node,
				     nfp_flower_table_params);
	if (err)
		goto err_release_metadata;

	err = nfp_flower_xmit_flow(app, merge_flow,
				   NFP_FLOWER_CMSG_TYPE_FLOW_MOD);
	if (err)
		goto err_remove_rhash;

	merge_flow->in_hw = true;
	sub_flow1->in_hw = false;

	return 0;

err_remove_rhash:
	WARN_ON_ONCE(rhashtable_remove_fast(&priv->flow_table,
					    &merge_flow->fl_node,
					    nfp_flower_table_params));
err_release_metadata:
	nfp_modify_flow_metadata(app, merge_flow);
err_unlink_sub_flow2:
939 940 941
	nfp_flower_unlink_flows(merge_flow, sub_flow2);
err_unlink_sub_flow1:
	nfp_flower_unlink_flows(merge_flow, sub_flow1);
942 943 944 945 946 947
err_destroy_merge_flow:
	kfree(merge_flow->action_data);
	kfree(merge_flow->mask_data);
	kfree(merge_flow->unmasked_data);
	kfree(merge_flow);
	return err;
948 949
}

950 951 952 953 954 955 956 957 958 959 960 961
/**
 * nfp_flower_add_offload() - Adds a new flow to hardware.
 * @app:	Pointer to the APP handle
 * @netdev:	netdev structure.
 * @flow:	TC flower classifier offload structure.
 *
 * Adds a new flow to the repeated hash structure and action payload.
 *
 * Return: negative value on error, 0 if configured successfully.
 */
static int
nfp_flower_add_offload(struct nfp_app *app, struct net_device *netdev,
962
		       struct flow_cls_offload *flow)
963
{
964
	enum nfp_flower_tun_type tun_type = NFP_FL_TUNNEL_NONE;
965
	struct nfp_flower_priv *priv = app->priv;
966
	struct netlink_ext_ack *extack = NULL;
967 968
	struct nfp_fl_payload *flow_pay;
	struct nfp_fl_key_ls *key_layer;
969
	struct nfp_port *port = NULL;
970 971
	int err;

972
	extack = flow->common.extack;
973 974 975
	if (nfp_netdev_is_nfp_repr(netdev))
		port = nfp_port_from_netdev(netdev);

976 977 978 979
	key_layer = kmalloc(sizeof(*key_layer), GFP_KERNEL);
	if (!key_layer)
		return -ENOMEM;

980
	err = nfp_flower_calculate_key_layers(app, netdev, key_layer, flow,
981
					      &tun_type, extack);
982 983 984
	if (err)
		goto err_free_key_ls;

985
	flow_pay = nfp_flower_allocate_new(key_layer);
986 987 988 989 990
	if (!flow_pay) {
		err = -ENOMEM;
		goto err_free_key_ls;
	}

991
	err = nfp_flower_compile_flow_match(app, flow, key_layer, netdev,
992
					    flow_pay, tun_type, extack);
993 994 995
	if (err)
		goto err_destroy_flow;

996
	err = nfp_flower_compile_action(app, flow, netdev, flow_pay, extack);
997 998 999
	if (err)
		goto err_destroy_flow;

1000
	err = nfp_compile_flow_metadata(app, flow, flow_pay, netdev, extack);
1001 1002 1003 1004
	if (err)
		goto err_destroy_flow;

	flow_pay->tc_flower_cookie = flow->cookie;
1005 1006
	err = rhashtable_insert_fast(&priv->flow_table, &flow_pay->fl_node,
				     nfp_flower_table_params);
1007 1008
	if (err) {
		NL_SET_ERR_MSG_MOD(extack, "invalid entry: cannot insert flow into tables for offloads");
1009
		goto err_release_metadata;
1010
	}
1011

1012
	err = nfp_flower_xmit_flow(app, flow_pay,
1013 1014 1015
				   NFP_FLOWER_CMSG_TYPE_FLOW_ADD);
	if (err)
		goto err_remove_rhash;
1016

1017 1018
	if (port)
		port->tc_offload_cnt++;
1019

J
John Hurley 已提交
1020 1021
	flow_pay->in_hw = true;

1022 1023 1024 1025
	/* Deallocate flow payload when flower rule has been destroyed. */
	kfree(key_layer);

	return 0;
1026

1027 1028 1029 1030
err_remove_rhash:
	WARN_ON_ONCE(rhashtable_remove_fast(&priv->flow_table,
					    &flow_pay->fl_node,
					    nfp_flower_table_params));
1031 1032
err_release_metadata:
	nfp_modify_flow_metadata(app, flow_pay);
1033
err_destroy_flow:
1034
	kfree(flow_pay->action_data);
1035 1036 1037 1038 1039 1040
	kfree(flow_pay->mask_data);
	kfree(flow_pay->unmasked_data);
	kfree(flow_pay);
err_free_key_ls:
	kfree(key_layer);
	return err;
1041 1042
}

J
John Hurley 已提交
1043 1044 1045 1046 1047 1048 1049 1050 1051 1052 1053 1054 1055 1056 1057 1058 1059 1060 1061 1062 1063 1064 1065 1066 1067 1068 1069 1070 1071 1072 1073 1074 1075 1076 1077 1078 1079 1080 1081 1082 1083 1084 1085 1086 1087 1088 1089 1090 1091 1092 1093 1094 1095 1096 1097 1098 1099 1100 1101 1102 1103 1104 1105 1106 1107 1108 1109 1110 1111
static void
nfp_flower_remove_merge_flow(struct nfp_app *app,
			     struct nfp_fl_payload *del_sub_flow,
			     struct nfp_fl_payload *merge_flow)
{
	struct nfp_flower_priv *priv = app->priv;
	struct nfp_fl_payload_link *link, *temp;
	struct nfp_fl_payload *origin;
	bool mod = false;
	int err;

	link = list_first_entry(&merge_flow->linked_flows,
				struct nfp_fl_payload_link, merge_flow.list);
	origin = link->sub_flow.flow;

	/* Re-add rule the merge had overwritten if it has not been deleted. */
	if (origin != del_sub_flow)
		mod = true;

	err = nfp_modify_flow_metadata(app, merge_flow);
	if (err) {
		nfp_flower_cmsg_warn(app, "Metadata fail for merge flow delete.\n");
		goto err_free_links;
	}

	if (!mod) {
		err = nfp_flower_xmit_flow(app, merge_flow,
					   NFP_FLOWER_CMSG_TYPE_FLOW_DEL);
		if (err) {
			nfp_flower_cmsg_warn(app, "Failed to delete merged flow.\n");
			goto err_free_links;
		}
	} else {
		__nfp_modify_flow_metadata(priv, origin);
		err = nfp_flower_xmit_flow(app, origin,
					   NFP_FLOWER_CMSG_TYPE_FLOW_MOD);
		if (err)
			nfp_flower_cmsg_warn(app, "Failed to revert merge flow.\n");
		origin->in_hw = true;
	}

err_free_links:
	/* Clean any links connected with the merged flow. */
	list_for_each_entry_safe(link, temp, &merge_flow->linked_flows,
				 merge_flow.list)
		nfp_flower_unlink_flow(link);

	kfree(merge_flow->action_data);
	kfree(merge_flow->mask_data);
	kfree(merge_flow->unmasked_data);
	WARN_ON_ONCE(rhashtable_remove_fast(&priv->flow_table,
					    &merge_flow->fl_node,
					    nfp_flower_table_params));
	kfree_rcu(merge_flow, rcu);
}

static void
nfp_flower_del_linked_merge_flows(struct nfp_app *app,
				  struct nfp_fl_payload *sub_flow)
{
	struct nfp_fl_payload_link *link, *temp;

	/* Remove any merge flow formed from the deleted sub_flow. */
	list_for_each_entry_safe(link, temp, &sub_flow->linked_flows,
				 sub_flow.list)
		nfp_flower_remove_merge_flow(app, sub_flow,
					     link->merge_flow.flow);
}

1112 1113 1114 1115 1116 1117 1118
/**
 * nfp_flower_del_offload() - Removes a flow from hardware.
 * @app:	Pointer to the APP handle
 * @netdev:	netdev structure.
 * @flow:	TC flower classifier offload structure
 *
 * Removes a flow from the repeated hash structure and clears the
J
John Hurley 已提交
1119
 * action payload. Any flows merged from this are also deleted.
1120 1121 1122 1123 1124
 *
 * Return: negative value on error, 0 if removed successfully.
 */
static int
nfp_flower_del_offload(struct nfp_app *app, struct net_device *netdev,
1125
		       struct flow_cls_offload *flow)
1126
{
1127
	struct nfp_flower_priv *priv = app->priv;
1128
	struct netlink_ext_ack *extack = NULL;
1129
	struct nfp_fl_payload *nfp_flow;
1130
	struct nfp_port *port = NULL;
1131 1132
	int err;

1133
	extack = flow->common.extack;
1134 1135 1136
	if (nfp_netdev_is_nfp_repr(netdev))
		port = nfp_port_from_netdev(netdev);

1137
	nfp_flow = nfp_flower_search_fl_table(app, flow->cookie, netdev);
1138 1139
	if (!nfp_flow) {
		NL_SET_ERR_MSG_MOD(extack, "invalid entry: cannot remove flow that does not exist");
1140
		return -ENOENT;
1141
	}
1142 1143

	err = nfp_modify_flow_metadata(app, nfp_flow);
1144
	if (err)
J
John Hurley 已提交
1145
		goto err_free_merge_flow;
1146

1147 1148 1149
	if (nfp_flow->nfp_tun_ipv4_addr)
		nfp_tunnel_del_ipv4_off(app, nfp_flow->nfp_tun_ipv4_addr);

J
John Hurley 已提交
1150 1151 1152 1153 1154
	if (!nfp_flow->in_hw) {
		err = 0;
		goto err_free_merge_flow;
	}

1155
	err = nfp_flower_xmit_flow(app, nfp_flow,
1156
				   NFP_FLOWER_CMSG_TYPE_FLOW_DEL);
J
John Hurley 已提交
1157
	/* Fall through on error. */
1158

J
John Hurley 已提交
1159 1160
err_free_merge_flow:
	nfp_flower_del_linked_merge_flows(app, nfp_flow);
1161 1162
	if (port)
		port->tc_offload_cnt--;
1163 1164 1165
	kfree(nfp_flow->action_data);
	kfree(nfp_flow->mask_data);
	kfree(nfp_flow->unmasked_data);
1166 1167 1168
	WARN_ON_ONCE(rhashtable_remove_fast(&priv->flow_table,
					    &nfp_flow->fl_node,
					    nfp_flower_table_params));
1169 1170
	kfree_rcu(nfp_flow, rcu);
	return err;
1171 1172
}

1173 1174 1175 1176 1177 1178 1179 1180 1181 1182 1183 1184 1185 1186 1187 1188 1189 1190 1191 1192 1193 1194 1195 1196 1197 1198 1199 1200 1201 1202 1203 1204 1205 1206 1207 1208 1209 1210 1211 1212 1213 1214 1215 1216 1217 1218
static void
__nfp_flower_update_merge_stats(struct nfp_app *app,
				struct nfp_fl_payload *merge_flow)
{
	struct nfp_flower_priv *priv = app->priv;
	struct nfp_fl_payload_link *link;
	struct nfp_fl_payload *sub_flow;
	u64 pkts, bytes, used;
	u32 ctx_id;

	ctx_id = be32_to_cpu(merge_flow->meta.host_ctx_id);
	pkts = priv->stats[ctx_id].pkts;
	/* Do not cycle subflows if no stats to distribute. */
	if (!pkts)
		return;
	bytes = priv->stats[ctx_id].bytes;
	used = priv->stats[ctx_id].used;

	/* Reset stats for the merge flow. */
	priv->stats[ctx_id].pkts = 0;
	priv->stats[ctx_id].bytes = 0;

	/* The merge flow has received stats updates from firmware.
	 * Distribute these stats to all subflows that form the merge.
	 * The stats will collected from TC via the subflows.
	 */
	list_for_each_entry(link, &merge_flow->linked_flows, merge_flow.list) {
		sub_flow = link->sub_flow.flow;
		ctx_id = be32_to_cpu(sub_flow->meta.host_ctx_id);
		priv->stats[ctx_id].pkts += pkts;
		priv->stats[ctx_id].bytes += bytes;
		max_t(u64, priv->stats[ctx_id].used, used);
	}
}

static void
nfp_flower_update_merge_stats(struct nfp_app *app,
			      struct nfp_fl_payload *sub_flow)
{
	struct nfp_fl_payload_link *link;

	/* Get merge flows that the subflow forms to distribute their stats. */
	list_for_each_entry(link, &sub_flow->linked_flows, sub_flow.list)
		__nfp_flower_update_merge_stats(app, link->merge_flow.flow);
}

1219 1220 1221
/**
 * nfp_flower_get_stats() - Populates flow stats obtained from hardware.
 * @app:	Pointer to the APP handle
1222
 * @netdev:	Netdev structure.
1223 1224 1225 1226 1227 1228 1229 1230
 * @flow:	TC flower classifier offload structure
 *
 * Populates a flow statistics structure which which corresponds to a
 * specific flow.
 *
 * Return: negative value on error, 0 if stats populated successfully.
 */
static int
1231
nfp_flower_get_stats(struct nfp_app *app, struct net_device *netdev,
1232
		     struct flow_cls_offload *flow)
1233
{
1234
	struct nfp_flower_priv *priv = app->priv;
1235
	struct netlink_ext_ack *extack = NULL;
1236
	struct nfp_fl_payload *nfp_flow;
1237
	u32 ctx_id;
1238

1239
	extack = flow->common.extack;
1240
	nfp_flow = nfp_flower_search_fl_table(app, flow->cookie, netdev);
1241 1242
	if (!nfp_flow) {
		NL_SET_ERR_MSG_MOD(extack, "invalid entry: cannot dump stats for flow that does not exist");
1243
		return -EINVAL;
1244
	}
1245

1246 1247 1248
	ctx_id = be32_to_cpu(nfp_flow->meta.host_ctx_id);

	spin_lock_bh(&priv->stats_lock);
1249 1250 1251 1252
	/* If request is for a sub_flow, update stats from merged flows. */
	if (!list_empty(&nfp_flow->linked_flows))
		nfp_flower_update_merge_stats(app, nfp_flow);

1253 1254
	flow_stats_update(&flow->stats, priv->stats[ctx_id].bytes,
			  priv->stats[ctx_id].pkts, priv->stats[ctx_id].used);
1255

1256 1257 1258
	priv->stats[ctx_id].pkts = 0;
	priv->stats[ctx_id].bytes = 0;
	spin_unlock_bh(&priv->stats_lock);
1259 1260

	return 0;
1261 1262 1263 1264
}

static int
nfp_flower_repr_offload(struct nfp_app *app, struct net_device *netdev,
1265
			struct flow_cls_offload *flower)
1266
{
1267
	if (!eth_proto_is_802_3(flower->common.protocol))
1268 1269
		return -EOPNOTSUPP;

1270
	switch (flower->command) {
1271
	case FLOW_CLS_REPLACE:
1272
		return nfp_flower_add_offload(app, netdev, flower);
1273
	case FLOW_CLS_DESTROY:
1274
		return nfp_flower_del_offload(app, netdev, flower);
1275
	case FLOW_CLS_STATS:
1276
		return nfp_flower_get_stats(app, netdev, flower);
1277 1278 1279
	default:
		return -EOPNOTSUPP;
	}
1280 1281
}

1282 1283 1284
static int nfp_flower_setup_tc_block_cb(enum tc_setup_type type,
					void *type_data, void *cb_priv)
{
1285
	struct nfp_repr *repr = cb_priv;
1286

1287
	if (!tc_cls_can_offload_and_chain0(repr->netdev, type_data))
1288 1289
		return -EOPNOTSUPP;

1290 1291
	switch (type) {
	case TC_SETUP_CLSFLOWER:
1292
		return nfp_flower_repr_offload(repr->app, repr->netdev,
1293
					       type_data);
1294 1295 1296
	case TC_SETUP_CLSMATCHALL:
		return nfp_flower_setup_qos_offload(repr->app, repr->netdev,
						    type_data);
1297 1298 1299 1300 1301
	default:
		return -EOPNOTSUPP;
	}
}

1302 1303
static LIST_HEAD(nfp_block_cb_list);

1304
static int nfp_flower_setup_tc_block(struct net_device *netdev,
1305
				     struct flow_block_offload *f)
1306
{
1307
	struct nfp_repr *repr = netdev_priv(netdev);
1308
	struct nfp_flower_repr_priv *repr_priv;
1309
	struct flow_block_cb *block_cb;
1310

1311
	if (f->binder_type != FLOW_BLOCK_BINDER_TYPE_CLSACT_INGRESS)
1312 1313
		return -EOPNOTSUPP;

1314
	repr_priv = repr->app_priv;
1315 1316
	repr_priv->block_shared = f->block_shared;
	f->driver_block_list = &nfp_block_cb_list;
1317

1318
	switch (f->command) {
1319
	case FLOW_BLOCK_BIND:
1320 1321 1322 1323
		if (flow_block_cb_is_busy(nfp_flower_setup_tc_block_cb, repr,
					  &nfp_block_cb_list))
			return -EBUSY;

1324 1325 1326 1327 1328 1329 1330 1331 1332
		block_cb = flow_block_cb_alloc(f->net,
					       nfp_flower_setup_tc_block_cb,
					       repr, repr, NULL);
		if (IS_ERR(block_cb))
			return PTR_ERR(block_cb);

		flow_block_cb_add(block_cb, f);
		list_add_tail(&block_cb->driver_list, &nfp_block_cb_list);
		return 0;
1333
	case FLOW_BLOCK_UNBIND:
1334 1335 1336 1337 1338 1339 1340
		block_cb = flow_block_cb_lookup(f, nfp_flower_setup_tc_block_cb,
						repr);
		if (!block_cb)
			return -ENOENT;

		flow_block_cb_remove(block_cb, f);
		list_del(&block_cb->driver_list);
1341 1342 1343 1344 1345 1346 1347 1348 1349 1350 1351 1352 1353 1354 1355
		return 0;
	default:
		return -EOPNOTSUPP;
	}
}

int nfp_flower_setup_tc(struct nfp_app *app, struct net_device *netdev,
			enum tc_setup_type type, void *type_data)
{
	switch (type) {
	case TC_SETUP_BLOCK:
		return nfp_flower_setup_tc_block(netdev, type_data);
	default:
		return -EOPNOTSUPP;
	}
1356
}
1357 1358 1359 1360 1361 1362 1363 1364 1365 1366 1367 1368 1369 1370 1371 1372 1373 1374 1375 1376 1377 1378 1379 1380 1381 1382 1383 1384

struct nfp_flower_indr_block_cb_priv {
	struct net_device *netdev;
	struct nfp_app *app;
	struct list_head list;
};

static struct nfp_flower_indr_block_cb_priv *
nfp_flower_indr_block_cb_priv_lookup(struct nfp_app *app,
				     struct net_device *netdev)
{
	struct nfp_flower_indr_block_cb_priv *cb_priv;
	struct nfp_flower_priv *priv = app->priv;

	/* All callback list access should be protected by RTNL. */
	ASSERT_RTNL();

	list_for_each_entry(cb_priv, &priv->indr_block_cb_priv, list)
		if (cb_priv->netdev == netdev)
			return cb_priv;

	return NULL;
}

static int nfp_flower_setup_indr_block_cb(enum tc_setup_type type,
					  void *type_data, void *cb_priv)
{
	struct nfp_flower_indr_block_cb_priv *priv = cb_priv;
1385
	struct flow_cls_offload *flower = type_data;
1386 1387 1388 1389 1390 1391 1392

	if (flower->common.chain_index)
		return -EOPNOTSUPP;

	switch (type) {
	case TC_SETUP_CLSFLOWER:
		return nfp_flower_repr_offload(priv->app, priv->netdev,
1393
					       type_data);
1394 1395 1396 1397 1398
	default:
		return -EOPNOTSUPP;
	}
}

1399 1400 1401 1402 1403 1404 1405 1406
static void nfp_flower_setup_indr_tc_release(void *cb_priv)
{
	struct nfp_flower_indr_block_cb_priv *priv = cb_priv;

	list_del(&priv->list);
	kfree(priv);
}

1407 1408
static int
nfp_flower_setup_indr_tc_block(struct net_device *netdev, struct nfp_app *app,
1409
			       struct flow_block_offload *f)
1410 1411 1412
{
	struct nfp_flower_indr_block_cb_priv *cb_priv;
	struct nfp_flower_priv *priv = app->priv;
1413
	struct flow_block_cb *block_cb;
1414

1415 1416
	if (f->binder_type != FLOW_BLOCK_BINDER_TYPE_CLSACT_INGRESS &&
	    !(f->binder_type == FLOW_BLOCK_BINDER_TYPE_CLSACT_EGRESS &&
1417
	      nfp_flower_internal_port_can_offload(app, netdev)))
1418 1419 1420
		return -EOPNOTSUPP;

	switch (f->command) {
1421
	case FLOW_BLOCK_BIND:
1422 1423 1424 1425 1426 1427 1428 1429
		cb_priv = kmalloc(sizeof(*cb_priv), GFP_KERNEL);
		if (!cb_priv)
			return -ENOMEM;

		cb_priv->netdev = netdev;
		cb_priv->app = app;
		list_add(&cb_priv->list, &priv->indr_block_cb_priv);

1430 1431 1432 1433 1434
		block_cb = flow_block_cb_alloc(f->net,
					       nfp_flower_setup_indr_block_cb,
					       cb_priv, cb_priv,
					       nfp_flower_setup_indr_tc_release);
		if (IS_ERR(block_cb)) {
1435 1436
			list_del(&cb_priv->list);
			kfree(cb_priv);
1437
			return PTR_ERR(block_cb);
1438 1439
		}

1440 1441 1442
		flow_block_cb_add(block_cb, f);
		list_add_tail(&block_cb->driver_list, &nfp_block_cb_list);
		return 0;
1443
	case FLOW_BLOCK_UNBIND:
1444
		cb_priv = nfp_flower_indr_block_cb_priv_lookup(app, netdev);
1445 1446 1447
		if (!cb_priv)
			return -ENOENT;

1448 1449 1450 1451 1452
		block_cb = flow_block_cb_lookup(f,
						nfp_flower_setup_indr_block_cb,
						cb_priv);
		if (!block_cb)
			return -ENOENT;
1453

1454 1455
		flow_block_cb_remove(block_cb, f);
		list_del(&block_cb->driver_list);
1456 1457 1458 1459 1460 1461 1462 1463 1464 1465 1466 1467 1468 1469 1470 1471 1472 1473 1474 1475 1476 1477 1478 1479 1480 1481 1482 1483 1484 1485 1486 1487
		return 0;
	default:
		return -EOPNOTSUPP;
	}
	return 0;
}

static int
nfp_flower_indr_setup_tc_cb(struct net_device *netdev, void *cb_priv,
			    enum tc_setup_type type, void *type_data)
{
	switch (type) {
	case TC_SETUP_BLOCK:
		return nfp_flower_setup_indr_tc_block(netdev, cb_priv,
						      type_data);
	default:
		return -EOPNOTSUPP;
	}
}

int nfp_flower_reg_indir_block_handler(struct nfp_app *app,
				       struct net_device *netdev,
				       unsigned long event)
{
	int err;

	if (!nfp_fl_is_netdev_to_offload(netdev))
		return NOTIFY_OK;

	if (event == NETDEV_REGISTER) {
		err = __tc_indr_block_cb_register(netdev, app,
						  nfp_flower_indr_setup_tc_cb,
1488
						  app);
1489 1490 1491 1492 1493 1494
		if (err)
			nfp_flower_cmsg_warn(app,
					     "Indirect block reg failed - %s\n",
					     netdev->name);
	} else if (event == NETDEV_UNREGISTER) {
		__tc_indr_block_cb_unregister(netdev,
1495
					      nfp_flower_indr_setup_tc_cb, app);
1496 1497 1498 1499
	}

	return NOTIFY_OK;
}