summaryrefslogtreecommitdiff
path: root/drivers/net/ethernet/netronome/nfp/nfdk/dp.c
blob: 85dd376a6adc25a8adffe6325d2dee98b08c116e (plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
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
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
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
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
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
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
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
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
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
1001
1002
1003
1004
1005
1006
1007
1008
1009
1010
1011
1012
1013
1014
1015
1016
1017
1018
1019
1020
1021
1022
1023
1024
1025
1026
1027
1028
1029
1030
1031
1032
1033
1034
1035
1036
1037
1038
1039
1040
1041
1042
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
1112
1113
1114
1115
1116
1117
1118
1119
1120
1121
1122
1123
1124
1125
1126
1127
1128
1129
1130
1131
1132
1133
1134
1135
1136
1137
1138
1139
1140
1141
1142
1143
1144
1145
1146
1147
1148
1149
1150
1151
1152
1153
1154
1155
1156
1157
1158
1159
1160
1161
1162
1163
1164
1165
1166
1167
1168
1169
1170
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
1219
1220
1221
1222
1223
1224
1225
1226
1227
1228
1229
1230
1231
1232
1233
1234
1235
1236
1237
1238
1239
1240
1241
1242
1243
1244
1245
1246
1247
1248
1249
1250
1251
1252
1253
1254
1255
1256
1257
1258
1259
1260
1261
1262
1263
1264
1265
1266
1267
1268
1269
1270
1271
1272
1273
1274
1275
1276
1277
1278
1279
1280
1281
1282
1283
1284
1285
1286
1287
1288
1289
1290
1291
1292
1293
1294
1295
1296
1297
1298
1299
1300
1301
1302
1303
1304
1305
1306
1307
1308
1309
1310
1311
1312
1313
1314
1315
1316
1317
1318
1319
1320
1321
1322
1323
1324
1325
1326
1327
1328
1329
1330
1331
1332
1333
1334
1335
1336
1337
1338
1339
1340
1341
1342
1343
1344
1345
1346
1347
1348
1349
1350
1351
1352
1353
1354
1355
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
1385
1386
1387
1388
1389
1390
1391
1392
1393
1394
1395
1396
1397
1398
1399
1400
1401
1402
1403
1404
1405
1406
1407
1408
1409
1410
1411
1412
1413
1414
1415
1416
1417
1418
1419
1420
1421
1422
1423
1424
1425
1426
1427
1428
1429
1430
1431
1432
1433
1434
1435
1436
1437
1438
1439
1440
1441
1442
1443
1444
1445
1446
1447
1448
1449
1450
1451
1452
1453
1454
1455
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
1488
1489
1490
1491
1492
1493
1494
1495
1496
1497
1498
1499
1500
1501
1502
1503
1504
1505
1506
1507
1508
1509
1510
1511
1512
1513
1514
1515
1516
1517
1518
1519
1520
1521
1522
1523
// SPDX-License-Identifier: (GPL-2.0-only OR BSD-2-Clause)
/* Copyright (C) 2015-2019 Netronome Systems, Inc. */

#include <linux/bpf_trace.h>
#include <linux/netdevice.h>
#include <linux/overflow.h>
#include <linux/sizes.h>
#include <linux/bitfield.h>

#include "../nfp_app.h"
#include "../nfp_net.h"
#include "../nfp_net_dp.h"
#include "../crypto/crypto.h"
#include "../crypto/fw.h"
#include "nfdk.h"

static int nfp_nfdk_tx_ring_should_wake(struct nfp_net_tx_ring *tx_ring)
{
	return !nfp_net_tx_full(tx_ring, NFDK_TX_DESC_STOP_CNT * 2);
}

static int nfp_nfdk_tx_ring_should_stop(struct nfp_net_tx_ring *tx_ring)
{
	return nfp_net_tx_full(tx_ring, NFDK_TX_DESC_STOP_CNT);
}

static void nfp_nfdk_tx_ring_stop(struct netdev_queue *nd_q,
				  struct nfp_net_tx_ring *tx_ring)
{
	netif_tx_stop_queue(nd_q);

	/* We can race with the TX completion out of NAPI so recheck */
	smp_mb();
	if (unlikely(nfp_nfdk_tx_ring_should_wake(tx_ring)))
		netif_tx_start_queue(nd_q);
}

static __le64
nfp_nfdk_tx_tso(struct nfp_net_r_vector *r_vec, struct nfp_nfdk_tx_buf *txbuf,
		struct sk_buff *skb)
{
	u32 segs, hdrlen, l3_offset, l4_offset;
	struct nfp_nfdk_tx_desc txd;
	u16 mss;

	if (!skb->encapsulation) {
		l3_offset = skb_network_offset(skb);
		l4_offset = skb_transport_offset(skb);
		hdrlen = skb_tcp_all_headers(skb);
	} else {
		l3_offset = skb_inner_network_offset(skb);
		l4_offset = skb_inner_transport_offset(skb);
		hdrlen = skb_inner_tcp_all_headers(skb);
	}

	segs = skb_shinfo(skb)->gso_segs;
	mss = skb_shinfo(skb)->gso_size & NFDK_DESC_TX_MSS_MASK;

	/* Note: TSO of the packet with metadata prepended to skb is not
	 * supported yet, in which case l3/l4_offset and lso_hdrlen need
	 * be correctly handled here.
	 * Concern:
	 * The driver doesn't have md_bytes easily available at this point.
	 * The PCI.IN PD ME won't have md_bytes bytes to add to lso_hdrlen,
	 * so it needs the full length there.  The app MEs might prefer
	 * l3_offset and l4_offset relative to the start of packet data,
	 * but could probably cope with it being relative to the CTM buf
	 * data offset.
	 */
	txd.l3_offset = l3_offset;
	txd.l4_offset = l4_offset;
	txd.lso_meta_res = 0;
	txd.mss = cpu_to_le16(mss);
	txd.lso_hdrlen = hdrlen;
	txd.lso_totsegs = segs;

	txbuf->pkt_cnt = segs;
	txbuf->real_len = skb->len + hdrlen * (txbuf->pkt_cnt - 1);

	u64_stats_update_begin(&r_vec->tx_sync);
	r_vec->tx_lso++;
	u64_stats_update_end(&r_vec->tx_sync);

	return txd.raw;
}

static u8
nfp_nfdk_tx_csum(struct nfp_net_dp *dp, struct nfp_net_r_vector *r_vec,
		 unsigned int pkt_cnt, struct sk_buff *skb, u64 flags)
{
	struct ipv6hdr *ipv6h;
	struct iphdr *iph;

	if (!(dp->ctrl & NFP_NET_CFG_CTRL_TXCSUM))
		return flags;

	if (skb->ip_summed != CHECKSUM_PARTIAL)
		return flags;

	flags |= NFDK_DESC_TX_L4_CSUM;

	iph = skb->encapsulation ? inner_ip_hdr(skb) : ip_hdr(skb);
	ipv6h = skb->encapsulation ? inner_ipv6_hdr(skb) : ipv6_hdr(skb);

	/* L3 checksum offloading flag is not required for ipv6 */
	if (iph->version == 4) {
		flags |= NFDK_DESC_TX_L3_CSUM;
	} else if (ipv6h->version != 6) {
		nn_dp_warn(dp, "partial checksum but ipv=%x!\n", iph->version);
		return flags;
	}

	u64_stats_update_begin(&r_vec->tx_sync);
	if (!skb->encapsulation) {
		r_vec->hw_csum_tx += pkt_cnt;
	} else {
		flags |= NFDK_DESC_TX_ENCAP;
		r_vec->hw_csum_tx_inner += pkt_cnt;
	}
	u64_stats_update_end(&r_vec->tx_sync);

	return flags;
}

static int
nfp_nfdk_tx_maybe_close_block(struct nfp_net_tx_ring *tx_ring,
			      unsigned int nr_frags, struct sk_buff *skb)
{
	unsigned int n_descs, wr_p, nop_slots;
	const skb_frag_t *frag, *fend;
	struct nfp_nfdk_tx_desc *txd;
	unsigned int wr_idx;
	int err;

recount_descs:
	n_descs = nfp_nfdk_headlen_to_segs(skb_headlen(skb));

	frag = skb_shinfo(skb)->frags;
	fend = frag + nr_frags;
	for (; frag < fend; frag++)
		n_descs += DIV_ROUND_UP(skb_frag_size(frag),
					NFDK_TX_MAX_DATA_PER_DESC);

	if (unlikely(n_descs > NFDK_TX_DESC_GATHER_MAX)) {
		if (skb_is_nonlinear(skb)) {
			err = skb_linearize(skb);
			if (err)
				return err;
			goto recount_descs;
		}
		return -EINVAL;
	}

	/* Under count by 1 (don't count meta) for the round down to work out */
	n_descs += !!skb_is_gso(skb);

	if (round_down(tx_ring->wr_p, NFDK_TX_DESC_BLOCK_CNT) !=
	    round_down(tx_ring->wr_p + n_descs, NFDK_TX_DESC_BLOCK_CNT))
		goto close_block;

	if ((u32)tx_ring->data_pending + skb->len > NFDK_TX_MAX_DATA_PER_BLOCK)
		goto close_block;

	return 0;

close_block:
	wr_p = tx_ring->wr_p;
	nop_slots = D_BLOCK_CPL(wr_p);

	wr_idx = D_IDX(tx_ring, wr_p);
	tx_ring->ktxbufs[wr_idx].skb = NULL;
	txd = &tx_ring->ktxds[wr_idx];

	memset(txd, 0, array_size(nop_slots, sizeof(struct nfp_nfdk_tx_desc)));

	tx_ring->data_pending = 0;
	tx_ring->wr_p += nop_slots;
	tx_ring->wr_ptr_add += nop_slots;

	return 0;
}

static int nfp_nfdk_prep_port_id(struct sk_buff *skb)
{
	struct metadata_dst *md_dst = skb_metadata_dst(skb);
	unsigned char *data;

	if (likely(!md_dst))
		return 0;
	if (unlikely(md_dst->type != METADATA_HW_PORT_MUX))
		return 0;

	/* Note: Unsupported case when TSO a skb with metedata prepended.
	 * See the comments in `nfp_nfdk_tx_tso` for details.
	 */
	if (unlikely(md_dst && skb_is_gso(skb)))
		return -EOPNOTSUPP;

	if (unlikely(skb_cow_head(skb, sizeof(md_dst->u.port_info.port_id))))
		return -ENOMEM;

	data = skb_push(skb, sizeof(md_dst->u.port_info.port_id));
	put_unaligned_be32(md_dst->u.port_info.port_id, data);

	return sizeof(md_dst->u.port_info.port_id);
}

static int
nfp_nfdk_prep_tx_meta(struct nfp_app *app, struct sk_buff *skb,
		      struct nfp_net_r_vector *r_vec)
{
	unsigned char *data;
	int res, md_bytes;
	u32 meta_id = 0;

	res = nfp_nfdk_prep_port_id(skb);
	if (unlikely(res <= 0))
		return res;

	md_bytes = res;
	meta_id = NFP_NET_META_PORTID;

	if (unlikely(skb_cow_head(skb, sizeof(meta_id))))
		return -ENOMEM;

	md_bytes += sizeof(meta_id);

	meta_id = FIELD_PREP(NFDK_META_LEN, md_bytes) |
		  FIELD_PREP(NFDK_META_FIELDS, meta_id);

	data = skb_push(skb, sizeof(meta_id));
	put_unaligned_be32(meta_id, data);

	return NFDK_DESC_TX_CHAIN_META;
}

/**
 * nfp_nfdk_tx() - Main transmit entry point
 * @skb:    SKB to transmit
 * @netdev: netdev structure
 *
 * Return: NETDEV_TX_OK on success.
 */
netdev_tx_t nfp_nfdk_tx(struct sk_buff *skb, struct net_device *netdev)
{
	struct nfp_net *nn = netdev_priv(netdev);
	struct nfp_nfdk_tx_buf *txbuf, *etxbuf;
	u32 cnt, tmp_dlen, dlen_type = 0;
	struct nfp_net_tx_ring *tx_ring;
	struct nfp_net_r_vector *r_vec;
	const skb_frag_t *frag, *fend;
	struct nfp_nfdk_tx_desc *txd;
	unsigned int real_len, qidx;
	unsigned int dma_len, type;
	struct netdev_queue *nd_q;
	struct nfp_net_dp *dp;
	int nr_frags, wr_idx;
	dma_addr_t dma_addr;
	u64 metadata;

	dp = &nn->dp;
	qidx = skb_get_queue_mapping(skb);
	tx_ring = &dp->tx_rings[qidx];
	r_vec = tx_ring->r_vec;
	nd_q = netdev_get_tx_queue(dp->netdev, qidx);

	/* Don't bother counting frags, assume the worst */
	if (unlikely(nfp_net_tx_full(tx_ring, NFDK_TX_DESC_STOP_CNT))) {
		nn_dp_warn(dp, "TX ring %d busy. wrp=%u rdp=%u\n",
			   qidx, tx_ring->wr_p, tx_ring->rd_p);
		netif_tx_stop_queue(nd_q);
		nfp_net_tx_xmit_more_flush(tx_ring);
		u64_stats_update_begin(&r_vec->tx_sync);
		r_vec->tx_busy++;
		u64_stats_update_end(&r_vec->tx_sync);
		return NETDEV_TX_BUSY;
	}

	metadata = nfp_nfdk_prep_tx_meta(nn->app, skb, r_vec);
	if (unlikely((int)metadata < 0))
		goto err_flush;

	nr_frags = skb_shinfo(skb)->nr_frags;
	if (nfp_nfdk_tx_maybe_close_block(tx_ring, nr_frags, skb))
		goto err_flush;

	/* DMA map all */
	wr_idx = D_IDX(tx_ring, tx_ring->wr_p);
	txd = &tx_ring->ktxds[wr_idx];
	txbuf = &tx_ring->ktxbufs[wr_idx];

	dma_len = skb_headlen(skb);
	if (skb_is_gso(skb))
		type = NFDK_DESC_TX_TYPE_TSO;
	else if (!nr_frags && dma_len < NFDK_TX_MAX_DATA_PER_HEAD)
		type = NFDK_DESC_TX_TYPE_SIMPLE;
	else
		type = NFDK_DESC_TX_TYPE_GATHER;

	dma_addr = dma_map_single(dp->dev, skb->data, dma_len, DMA_TO_DEVICE);
	if (dma_mapping_error(dp->dev, dma_addr))
		goto err_warn_dma;

	txbuf->skb = skb;
	txbuf++;

	txbuf->dma_addr = dma_addr;
	txbuf++;

	/* FIELD_PREP() implicitly truncates to chunk */
	dma_len -= 1;
	dlen_type = FIELD_PREP(NFDK_DESC_TX_DMA_LEN_HEAD, dma_len) |
		    FIELD_PREP(NFDK_DESC_TX_TYPE_HEAD, type);

	txd->dma_len_type = cpu_to_le16(dlen_type);
	nfp_desc_set_dma_addr_48b(txd, dma_addr);

	/* starts at bit 0 */
	BUILD_BUG_ON(!(NFDK_DESC_TX_DMA_LEN_HEAD & 1));

	/* Preserve the original dlen_type, this way below the EOP logic
	 * can use dlen_type.
	 */
	tmp_dlen = dlen_type & NFDK_DESC_TX_DMA_LEN_HEAD;
	dma_len -= tmp_dlen;
	dma_addr += tmp_dlen + 1;
	txd++;

	/* The rest of the data (if any) will be in larger dma descritors
	 * and is handled with the fragment loop.
	 */
	frag = skb_shinfo(skb)->frags;
	fend = frag + nr_frags;

	while (true) {
		while (dma_len > 0) {
			dma_len -= 1;
			dlen_type = FIELD_PREP(NFDK_DESC_TX_DMA_LEN, dma_len);

			txd->dma_len_type = cpu_to_le16(dlen_type);
			nfp_desc_set_dma_addr_48b(txd, dma_addr);

			dma_len -= dlen_type;
			dma_addr += dlen_type + 1;
			txd++;
		}

		if (frag >= fend)
			break;

		dma_len = skb_frag_size(frag);
		dma_addr = skb_frag_dma_map(dp->dev, frag, 0, dma_len,
					    DMA_TO_DEVICE);
		if (dma_mapping_error(dp->dev, dma_addr))
			goto err_unmap;

		txbuf->dma_addr = dma_addr;
		txbuf++;

		frag++;
	}

	(txd - 1)->dma_len_type = cpu_to_le16(dlen_type | NFDK_DESC_TX_EOP);

	if (!skb_is_gso(skb)) {
		real_len = skb->len;
		/* Metadata desc */
		metadata = nfp_nfdk_tx_csum(dp, r_vec, 1, skb, metadata);
		txd->raw = cpu_to_le64(metadata);
		txd++;
	} else {
		/* lso desc should be placed after metadata desc */
		(txd + 1)->raw = nfp_nfdk_tx_tso(r_vec, txbuf, skb);
		real_len = txbuf->real_len;
		/* Metadata desc */
		metadata = nfp_nfdk_tx_csum(dp, r_vec, txbuf->pkt_cnt, skb, metadata);
		txd->raw = cpu_to_le64(metadata);
		txd += 2;
		txbuf++;
	}

	cnt = txd - tx_ring->ktxds - wr_idx;
	if (unlikely(round_down(wr_idx, NFDK_TX_DESC_BLOCK_CNT) !=
		     round_down(wr_idx + cnt - 1, NFDK_TX_DESC_BLOCK_CNT)))
		goto err_warn_overflow;

	skb_tx_timestamp(skb);

	tx_ring->wr_p += cnt;
	if (tx_ring->wr_p % NFDK_TX_DESC_BLOCK_CNT)
		tx_ring->data_pending += skb->len;
	else
		tx_ring->data_pending = 0;

	if (nfp_nfdk_tx_ring_should_stop(tx_ring))
		nfp_nfdk_tx_ring_stop(nd_q, tx_ring);

	tx_ring->wr_ptr_add += cnt;
	if (__netdev_tx_sent_queue(nd_q, real_len, netdev_xmit_more()))
		nfp_net_tx_xmit_more_flush(tx_ring);

	return NETDEV_TX_OK;

err_warn_overflow:
	WARN_ONCE(1, "unable to fit packet into a descriptor wr_idx:%d head:%d frags:%d cnt:%d",
		  wr_idx, skb_headlen(skb), nr_frags, cnt);
	if (skb_is_gso(skb))
		txbuf--;
err_unmap:
	/* txbuf pointed to the next-to-use */
	etxbuf = txbuf;
	/* first txbuf holds the skb */
	txbuf = &tx_ring->ktxbufs[wr_idx + 1];
	if (txbuf < etxbuf) {
		dma_unmap_single(dp->dev, txbuf->dma_addr,
				 skb_headlen(skb), DMA_TO_DEVICE);
		txbuf->raw = 0;
		txbuf++;
	}
	frag = skb_shinfo(skb)->frags;
	while (etxbuf < txbuf) {
		dma_unmap_page(dp->dev, txbuf->dma_addr,
			       skb_frag_size(frag), DMA_TO_DEVICE);
		txbuf->raw = 0;
		frag++;
		txbuf++;
	}
err_warn_dma:
	nn_dp_warn(dp, "Failed to map DMA TX buffer\n");
err_flush:
	nfp_net_tx_xmit_more_flush(tx_ring);
	u64_stats_update_begin(&r_vec->tx_sync);
	r_vec->tx_errors++;
	u64_stats_update_end(&r_vec->tx_sync);
	dev_kfree_skb_any(skb);
	return NETDEV_TX_OK;
}

/**
 * nfp_nfdk_tx_complete() - Handled completed TX packets
 * @tx_ring:	TX ring structure
 * @budget:	NAPI budget (only used as bool to determine if in NAPI context)
 */
static void nfp_nfdk_tx_complete(struct nfp_net_tx_ring *tx_ring, int budget)
{
	struct nfp_net_r_vector *r_vec = tx_ring->r_vec;
	struct nfp_net_dp *dp = &r_vec->nfp_net->dp;
	u32 done_pkts = 0, done_bytes = 0;
	struct nfp_nfdk_tx_buf *ktxbufs;
	struct device *dev = dp->dev;
	struct netdev_queue *nd_q;
	u32 rd_p, qcp_rd_p;
	int todo;

	rd_p = tx_ring->rd_p;
	if (tx_ring->wr_p == rd_p)
		return;

	/* Work out how many descriptors have been transmitted */
	qcp_rd_p = nfp_net_read_tx_cmpl(tx_ring, dp);

	if (qcp_rd_p == tx_ring->qcp_rd_p)
		return;

	todo = D_IDX(tx_ring, qcp_rd_p - tx_ring->qcp_rd_p);
	ktxbufs = tx_ring->ktxbufs;

	while (todo > 0) {
		const skb_frag_t *frag, *fend;
		unsigned int size, n_descs = 1;
		struct nfp_nfdk_tx_buf *txbuf;
		struct sk_buff *skb;

		txbuf = &ktxbufs[D_IDX(tx_ring, rd_p)];
		skb = txbuf->skb;
		txbuf++;

		/* Closed block */
		if (!skb) {
			n_descs = D_BLOCK_CPL(rd_p);
			goto next;
		}

		/* Unmap head */
		size = skb_headlen(skb);
		n_descs += nfp_nfdk_headlen_to_segs(size);
		dma_unmap_single(dev, txbuf->dma_addr, size, DMA_TO_DEVICE);
		txbuf++;

		/* Unmap frags */
		frag = skb_shinfo(skb)->frags;
		fend = frag + skb_shinfo(skb)->nr_frags;
		for (; frag < fend; frag++) {
			size = skb_frag_size(frag);
			n_descs += DIV_ROUND_UP(size,
						NFDK_TX_MAX_DATA_PER_DESC);
			dma_unmap_page(dev, txbuf->dma_addr,
				       skb_frag_size(frag), DMA_TO_DEVICE);
			txbuf++;
		}

		if (!skb_is_gso(skb)) {
			done_bytes += skb->len;
			done_pkts++;
		} else {
			done_bytes += txbuf->real_len;
			done_pkts += txbuf->pkt_cnt;
			n_descs++;
		}

		napi_consume_skb(skb, budget);
next:
		rd_p += n_descs;
		todo -= n_descs;
	}

	tx_ring->rd_p = rd_p;
	tx_ring->qcp_rd_p = qcp_rd_p;

	u64_stats_update_begin(&r_vec->tx_sync);
	r_vec->tx_bytes += done_bytes;
	r_vec->tx_pkts += done_pkts;
	u64_stats_update_end(&r_vec->tx_sync);

	if (!dp->netdev)
		return;

	nd_q = netdev_get_tx_queue(dp->netdev, tx_ring->idx);
	netdev_tx_completed_queue(nd_q, done_pkts, done_bytes);
	if (nfp_nfdk_tx_ring_should_wake(tx_ring)) {
		/* Make sure TX thread will see updated tx_ring->rd_p */
		smp_mb();

		if (unlikely(netif_tx_queue_stopped(nd_q)))
			netif_tx_wake_queue(nd_q);
	}

	WARN_ONCE(tx_ring->wr_p - tx_ring->rd_p > tx_ring->cnt,
		  "TX ring corruption rd_p=%u wr_p=%u cnt=%u\n",
		  tx_ring->rd_p, tx_ring->wr_p, tx_ring->cnt);
}

/* Receive processing */
static void *
nfp_nfdk_napi_alloc_one(struct nfp_net_dp *dp, dma_addr_t *dma_addr)
{
	void *frag;

	if (!dp->xdp_prog) {
		frag = napi_alloc_frag(dp->fl_bufsz);
		if (unlikely(!frag))
			return NULL;
	} else {
		struct page *page;

		page = dev_alloc_page();
		if (unlikely(!page))
			return NULL;
		frag = page_address(page);
	}

	*dma_addr = nfp_net_dma_map_rx(dp, frag);
	if (dma_mapping_error(dp->dev, *dma_addr)) {
		nfp_net_free_frag(frag, dp->xdp_prog);
		nn_dp_warn(dp, "Failed to map DMA RX buffer\n");
		return NULL;
	}

	return frag;
}

/**
 * nfp_nfdk_rx_give_one() - Put mapped skb on the software and hardware rings
 * @dp:		NFP Net data path struct
 * @rx_ring:	RX ring structure
 * @frag:	page fragment buffer
 * @dma_addr:	DMA address of skb mapping
 */
static void
nfp_nfdk_rx_give_one(const struct nfp_net_dp *dp,
		     struct nfp_net_rx_ring *rx_ring,
		     void *frag, dma_addr_t dma_addr)
{
	unsigned int wr_idx;

	wr_idx = D_IDX(rx_ring, rx_ring->wr_p);

	nfp_net_dma_sync_dev_rx(dp, dma_addr);

	/* Stash SKB and DMA address away */
	rx_ring->rxbufs[wr_idx].frag = frag;
	rx_ring->rxbufs[wr_idx].dma_addr = dma_addr;

	/* Fill freelist descriptor */
	rx_ring->rxds[wr_idx].fld.reserved = 0;
	rx_ring->rxds[wr_idx].fld.meta_len_dd = 0;
	nfp_desc_set_dma_addr_48b(&rx_ring->rxds[wr_idx].fld,
				  dma_addr + dp->rx_dma_off);

	rx_ring->wr_p++;
	if (!(rx_ring->wr_p % NFP_NET_FL_BATCH)) {
		/* Update write pointer of the freelist queue. Make
		 * sure all writes are flushed before telling the hardware.
		 */
		wmb();
		nfp_qcp_wr_ptr_add(rx_ring->qcp_fl, NFP_NET_FL_BATCH);
	}
}

/**
 * nfp_nfdk_rx_ring_fill_freelist() - Give buffers from the ring to FW
 * @dp:	     NFP Net data path struct
 * @rx_ring: RX ring to fill
 */
void nfp_nfdk_rx_ring_fill_freelist(struct nfp_net_dp *dp,
				    struct nfp_net_rx_ring *rx_ring)
{
	unsigned int i;

	for (i = 0; i < rx_ring->cnt - 1; i++)
		nfp_nfdk_rx_give_one(dp, rx_ring, rx_ring->rxbufs[i].frag,
				     rx_ring->rxbufs[i].dma_addr);
}

/**
 * nfp_nfdk_rx_csum_has_errors() - group check if rxd has any csum errors
 * @flags: RX descriptor flags field in CPU byte order
 */
static int nfp_nfdk_rx_csum_has_errors(u16 flags)
{
	u16 csum_all_checked, csum_all_ok;

	csum_all_checked = flags & __PCIE_DESC_RX_CSUM_ALL;
	csum_all_ok = flags & __PCIE_DESC_RX_CSUM_ALL_OK;

	return csum_all_checked != (csum_all_ok << PCIE_DESC_RX_CSUM_OK_SHIFT);
}

/**
 * nfp_nfdk_rx_csum() - set SKB checksum field based on RX descriptor flags
 * @dp:  NFP Net data path struct
 * @r_vec: per-ring structure
 * @rxd: Pointer to RX descriptor
 * @meta: Parsed metadata prepend
 * @skb: Pointer to SKB
 */
static void
nfp_nfdk_rx_csum(struct nfp_net_dp *dp, struct nfp_net_r_vector *r_vec,
		 struct nfp_net_rx_desc *rxd, struct nfp_meta_parsed *meta,
		 struct sk_buff *skb)
{
	skb_checksum_none_assert(skb);

	if (!(dp->netdev->features & NETIF_F_RXCSUM))
		return;

	if (meta->csum_type) {
		skb->ip_summed = meta->csum_type;
		skb->csum = meta->csum;
		u64_stats_update_begin(&r_vec->rx_sync);
		r_vec->hw_csum_rx_complete++;
		u64_stats_update_end(&r_vec->rx_sync);
		return;
	}

	if (nfp_nfdk_rx_csum_has_errors(le16_to_cpu(rxd->rxd.flags))) {
		u64_stats_update_begin(&r_vec->rx_sync);
		r_vec->hw_csum_rx_error++;
		u64_stats_update_end(&r_vec->rx_sync);
		return;
	}

	/* Assume that the firmware will never report inner CSUM_OK unless outer
	 * L4 headers were successfully parsed. FW will always report zero UDP
	 * checksum as CSUM_OK.
	 */
	if (rxd->rxd.flags & PCIE_DESC_RX_TCP_CSUM_OK ||
	    rxd->rxd.flags & PCIE_DESC_RX_UDP_CSUM_OK) {
		__skb_incr_checksum_unnecessary(skb);
		u64_stats_update_begin(&r_vec->rx_sync);
		r_vec->hw_csum_rx_ok++;
		u64_stats_update_end(&r_vec->rx_sync);
	}

	if (rxd->rxd.flags & PCIE_DESC_RX_I_TCP_CSUM_OK ||
	    rxd->rxd.flags & PCIE_DESC_RX_I_UDP_CSUM_OK) {
		__skb_incr_checksum_unnecessary(skb);
		u64_stats_update_begin(&r_vec->rx_sync);
		r_vec->hw_csum_rx_inner_ok++;
		u64_stats_update_end(&r_vec->rx_sync);
	}
}

static void
nfp_nfdk_set_hash(struct net_device *netdev, struct nfp_meta_parsed *meta,
		  unsigned int type, __be32 *hash)
{
	if (!(netdev->features & NETIF_F_RXHASH))
		return;

	switch (type) {
	case NFP_NET_RSS_IPV4:
	case NFP_NET_RSS_IPV6:
	case NFP_NET_RSS_IPV6_EX:
		meta->hash_type = PKT_HASH_TYPE_L3;
		break;
	default:
		meta->hash_type = PKT_HASH_TYPE_L4;
		break;
	}

	meta->hash = get_unaligned_be32(hash);
}

static bool
nfp_nfdk_parse_meta(struct net_device *netdev, struct nfp_meta_parsed *meta,
		    void *data, void *pkt, unsigned int pkt_len, int meta_len)
{
	u32 meta_info;

	meta_info = get_unaligned_be32(data);
	data += 4;

	while (meta_info) {
		switch (meta_info & NFP_NET_META_FIELD_MASK) {
		case NFP_NET_META_HASH:
			meta_info >>= NFP_NET_META_FIELD_SIZE;
			nfp_nfdk_set_hash(netdev, meta,
					  meta_info & NFP_NET_META_FIELD_MASK,
					  (__be32 *)data);
			data += 4;
			break;
		case NFP_NET_META_MARK:
			meta->mark = get_unaligned_be32(data);
			data += 4;
			break;
		case NFP_NET_META_PORTID:
			meta->portid = get_unaligned_be32(data);
			data += 4;
			break;
		case NFP_NET_META_CSUM:
			meta->csum_type = CHECKSUM_COMPLETE;
			meta->csum =
				(__force __wsum)__get_unaligned_cpu32(data);
			data += 4;
			break;
		case NFP_NET_META_RESYNC_INFO:
			if (nfp_net_tls_rx_resync_req(netdev, data, pkt,
						      pkt_len))
				return false;
			data += sizeof(struct nfp_net_tls_resync_req);
			break;
		default:
			return true;
		}

		meta_info >>= NFP_NET_META_FIELD_SIZE;
	}

	return data != pkt;
}

static void
nfp_nfdk_rx_drop(const struct nfp_net_dp *dp, struct nfp_net_r_vector *r_vec,
		 struct nfp_net_rx_ring *rx_ring, struct nfp_net_rx_buf *rxbuf,
		 struct sk_buff *skb)
{
	u64_stats_update_begin(&r_vec->rx_sync);
	r_vec->rx_drops++;
	/* If we have both skb and rxbuf the replacement buffer allocation
	 * must have failed, count this as an alloc failure.
	 */
	if (skb && rxbuf)
		r_vec->rx_replace_buf_alloc_fail++;
	u64_stats_update_end(&r_vec->rx_sync);

	/* skb is build based on the frag, free_skb() would free the frag
	 * so to be able to reuse it we need an extra ref.
	 */
	if (skb && rxbuf && skb->head == rxbuf->frag)
		page_ref_inc(virt_to_head_page(rxbuf->frag));
	if (rxbuf)
		nfp_nfdk_rx_give_one(dp, rx_ring, rxbuf->frag, rxbuf->dma_addr);
	if (skb)
		dev_kfree_skb_any(skb);
}

static bool nfp_nfdk_xdp_complete(struct nfp_net_tx_ring *tx_ring)
{
	struct nfp_net_r_vector *r_vec = tx_ring->r_vec;
	struct nfp_net_dp *dp = &r_vec->nfp_net->dp;
	struct nfp_net_rx_ring *rx_ring;
	u32 qcp_rd_p, done = 0;
	bool done_all;
	int todo;

	/* Work out how many descriptors have been transmitted */
	qcp_rd_p = nfp_net_read_tx_cmpl(tx_ring, dp);
	if (qcp_rd_p == tx_ring->qcp_rd_p)
		return true;

	todo = D_IDX(tx_ring, qcp_rd_p - tx_ring->qcp_rd_p);

	done_all = todo <= NFP_NET_XDP_MAX_COMPLETE;
	todo = min(todo, NFP_NET_XDP_MAX_COMPLETE);

	rx_ring = r_vec->rx_ring;
	while (todo > 0) {
		int idx = D_IDX(tx_ring, tx_ring->rd_p + done);
		struct nfp_nfdk_tx_buf *txbuf;
		unsigned int step = 1;

		txbuf = &tx_ring->ktxbufs[idx];
		if (!txbuf->raw)
			goto next;

		if (NFDK_TX_BUF_INFO(txbuf->val) != NFDK_TX_BUF_INFO_SOP) {
			WARN_ONCE(1, "Unexpected TX buffer in XDP TX ring\n");
			goto next;
		}

		/* Two successive txbufs are used to stash virtual and dma
		 * address respectively, recycle and clean them here.
		 */
		nfp_nfdk_rx_give_one(dp, rx_ring,
				     (void *)NFDK_TX_BUF_PTR(txbuf[0].val),
				     txbuf[1].dma_addr);
		txbuf[0].raw = 0;
		txbuf[1].raw = 0;
		step = 2;

		u64_stats_update_begin(&r_vec->tx_sync);
		/* Note: tx_bytes not accumulated. */
		r_vec->tx_pkts++;
		u64_stats_update_end(&r_vec->tx_sync);
next:
		todo -= step;
		done += step;
	}

	tx_ring->qcp_rd_p = D_IDX(tx_ring, tx_ring->qcp_rd_p + done);
	tx_ring->rd_p += done;

	WARN_ONCE(tx_ring->wr_p - tx_ring->rd_p > tx_ring->cnt,
		  "XDP TX ring corruption rd_p=%u wr_p=%u cnt=%u\n",
		  tx_ring->rd_p, tx_ring->wr_p, tx_ring->cnt);

	return done_all;
}

static bool
nfp_nfdk_tx_xdp_buf(struct nfp_net_dp *dp, struct nfp_net_rx_ring *rx_ring,
		    struct nfp_net_tx_ring *tx_ring,
		    struct nfp_net_rx_buf *rxbuf, unsigned int dma_off,
		    unsigned int pkt_len, bool *completed)
{
	unsigned int dma_map_sz = dp->fl_bufsz - NFP_NET_RX_BUF_NON_DATA;
	unsigned int dma_len, type, cnt, dlen_type, tmp_dlen;
	struct nfp_nfdk_tx_buf *txbuf;
	struct nfp_nfdk_tx_desc *txd;
	unsigned int n_descs;
	dma_addr_t dma_addr;
	int wr_idx;

	/* Reject if xdp_adjust_tail grow packet beyond DMA area */
	if (pkt_len + dma_off > dma_map_sz)
		return false;

	/* Make sure there's still at least one block available after
	 * aligning to block boundary, so that the txds used below
	 * won't wrap around the tx_ring.
	 */
	if (unlikely(nfp_net_tx_full(tx_ring, NFDK_TX_DESC_STOP_CNT))) {
		if (!*completed) {
			nfp_nfdk_xdp_complete(tx_ring);
			*completed = true;
		}

		if (unlikely(nfp_net_tx_full(tx_ring, NFDK_TX_DESC_STOP_CNT))) {
			nfp_nfdk_rx_drop(dp, rx_ring->r_vec, rx_ring, rxbuf,
					 NULL);
			return false;
		}
	}

	/* Check if cross block boundary */
	n_descs = nfp_nfdk_headlen_to_segs(pkt_len);
	if ((round_down(tx_ring->wr_p, NFDK_TX_DESC_BLOCK_CNT) !=
	     round_down(tx_ring->wr_p + n_descs, NFDK_TX_DESC_BLOCK_CNT)) ||
	    ((u32)tx_ring->data_pending + pkt_len >
	     NFDK_TX_MAX_DATA_PER_BLOCK)) {
		unsigned int nop_slots = D_BLOCK_CPL(tx_ring->wr_p);

		wr_idx = D_IDX(tx_ring, tx_ring->wr_p);
		txd = &tx_ring->ktxds[wr_idx];
		memset(txd, 0,
		       array_size(nop_slots, sizeof(struct nfp_nfdk_tx_desc)));

		tx_ring->data_pending = 0;
		tx_ring->wr_p += nop_slots;
		tx_ring->wr_ptr_add += nop_slots;
	}

	wr_idx = D_IDX(tx_ring, tx_ring->wr_p);

	txbuf = &tx_ring->ktxbufs[wr_idx];

	txbuf[0].val = (unsigned long)rxbuf->frag | NFDK_TX_BUF_INFO_SOP;
	txbuf[1].dma_addr = rxbuf->dma_addr;
	/* Note: pkt len not stored */

	dma_sync_single_for_device(dp->dev, rxbuf->dma_addr + dma_off,
				   pkt_len, DMA_BIDIRECTIONAL);

	/* Build TX descriptor */
	txd = &tx_ring->ktxds[wr_idx];
	dma_len = pkt_len;
	dma_addr = rxbuf->dma_addr + dma_off;

	if (dma_len < NFDK_TX_MAX_DATA_PER_HEAD)
		type = NFDK_DESC_TX_TYPE_SIMPLE;
	else
		type = NFDK_DESC_TX_TYPE_GATHER;

	/* FIELD_PREP() implicitly truncates to chunk */
	dma_len -= 1;
	dlen_type = FIELD_PREP(NFDK_DESC_TX_DMA_LEN_HEAD, dma_len) |
		    FIELD_PREP(NFDK_DESC_TX_TYPE_HEAD, type);

	txd->dma_len_type = cpu_to_le16(dlen_type);
	nfp_desc_set_dma_addr_48b(txd, dma_addr);

	tmp_dlen = dlen_type & NFDK_DESC_TX_DMA_LEN_HEAD;
	dma_len -= tmp_dlen;
	dma_addr += tmp_dlen + 1;
	txd++;

	while (dma_len > 0) {
		dma_len -= 1;
		dlen_type = FIELD_PREP(NFDK_DESC_TX_DMA_LEN, dma_len);
		txd->dma_len_type = cpu_to_le16(dlen_type);
		nfp_desc_set_dma_addr_48b(txd, dma_addr);

		dlen_type &= NFDK_DESC_TX_DMA_LEN;
		dma_len -= dlen_type;
		dma_addr += dlen_type + 1;
		txd++;
	}

	(txd - 1)->dma_len_type = cpu_to_le16(dlen_type | NFDK_DESC_TX_EOP);

	/* Metadata desc */
	txd->raw = 0;
	txd++;

	cnt = txd - tx_ring->ktxds - wr_idx;
	tx_ring->wr_p += cnt;
	if (tx_ring->wr_p % NFDK_TX_DESC_BLOCK_CNT)
		tx_ring->data_pending += pkt_len;
	else
		tx_ring->data_pending = 0;

	tx_ring->wr_ptr_add += cnt;
	return true;
}

/**
 * nfp_nfdk_rx() - receive up to @budget packets on @rx_ring
 * @rx_ring:   RX ring to receive from
 * @budget:    NAPI budget
 *
 * Note, this function is separated out from the napi poll function to
 * more cleanly separate packet receive code from other bookkeeping
 * functions performed in the napi poll function.
 *
 * Return: Number of packets received.
 */
static int nfp_nfdk_rx(struct nfp_net_rx_ring *rx_ring, int budget)
{
	struct nfp_net_r_vector *r_vec = rx_ring->r_vec;
	struct nfp_net_dp *dp = &r_vec->nfp_net->dp;
	struct nfp_net_tx_ring *tx_ring;
	struct bpf_prog *xdp_prog;
	bool xdp_tx_cmpl = false;
	unsigned int true_bufsz;
	struct sk_buff *skb;
	int pkts_polled = 0;
	struct xdp_buff xdp;
	int idx;

	xdp_prog = READ_ONCE(dp->xdp_prog);
	true_bufsz = xdp_prog ? PAGE_SIZE : dp->fl_bufsz;
	xdp_init_buff(&xdp, PAGE_SIZE - NFP_NET_RX_BUF_HEADROOM,
		      &rx_ring->xdp_rxq);
	tx_ring = r_vec->xdp_ring;

	while (pkts_polled < budget) {
		unsigned int meta_len, data_len, meta_off, pkt_len, pkt_off;
		struct nfp_net_rx_buf *rxbuf;
		struct nfp_net_rx_desc *rxd;
		struct nfp_meta_parsed meta;
		bool redir_egress = false;
		struct net_device *netdev;
		dma_addr_t new_dma_addr;
		u32 meta_len_xdp = 0;
		void *new_frag;

		idx = D_IDX(rx_ring, rx_ring->rd_p);

		rxd = &rx_ring->rxds[idx];
		if (!(rxd->rxd.meta_len_dd & PCIE_DESC_RX_DD))
			break;

		/* Memory barrier to ensure that we won't do other reads
		 * before the DD bit.
		 */
		dma_rmb();

		memset(&meta, 0, sizeof(meta));

		rx_ring->rd_p++;
		pkts_polled++;

		rxbuf =	&rx_ring->rxbufs[idx];
		/*         < meta_len >
		 *  <-- [rx_offset] -->
		 *  ---------------------------------------------------------
		 * | [XX] |  metadata  |             packet           | XXXX |
		 *  ---------------------------------------------------------
		 *         <---------------- data_len --------------->
		 *
		 * The rx_offset is fixed for all packets, the meta_len can vary
		 * on a packet by packet basis. If rx_offset is set to zero
		 * (_RX_OFFSET_DYNAMIC) metadata starts at the beginning of the
		 * buffer and is immediately followed by the packet (no [XX]).
		 */
		meta_len = rxd->rxd.meta_len_dd & PCIE_DESC_RX_META_LEN_MASK;
		data_len = le16_to_cpu(rxd->rxd.data_len);
		pkt_len = data_len - meta_len;

		pkt_off = NFP_NET_RX_BUF_HEADROOM + dp->rx_dma_off;
		if (dp->rx_offset == NFP_NET_CFG_RX_OFFSET_DYNAMIC)
			pkt_off += meta_len;
		else
			pkt_off += dp->rx_offset;
		meta_off = pkt_off - meta_len;

		/* Stats update */
		u64_stats_update_begin(&r_vec->rx_sync);
		r_vec->rx_pkts++;
		r_vec->rx_bytes += pkt_len;
		u64_stats_update_end(&r_vec->rx_sync);

		if (unlikely(meta_len > NFP_NET_MAX_PREPEND ||
			     (dp->rx_offset && meta_len > dp->rx_offset))) {
			nn_dp_warn(dp, "oversized RX packet metadata %u\n",
				   meta_len);
			nfp_nfdk_rx_drop(dp, r_vec, rx_ring, rxbuf, NULL);
			continue;
		}

		nfp_net_dma_sync_cpu_rx(dp, rxbuf->dma_addr + meta_off,
					data_len);

		if (meta_len) {
			if (unlikely(nfp_nfdk_parse_meta(dp->netdev, &meta,
							 rxbuf->frag + meta_off,
							 rxbuf->frag + pkt_off,
							 pkt_len, meta_len))) {
				nn_dp_warn(dp, "invalid RX packet metadata\n");
				nfp_nfdk_rx_drop(dp, r_vec, rx_ring, rxbuf,
						 NULL);
				continue;
			}
		}

		if (xdp_prog && !meta.portid) {
			void *orig_data = rxbuf->frag + pkt_off;
			unsigned int dma_off;
			int act;

			xdp_prepare_buff(&xdp,
					 rxbuf->frag + NFP_NET_RX_BUF_HEADROOM,
					 pkt_off - NFP_NET_RX_BUF_HEADROOM,
					 pkt_len, true);

			act = bpf_prog_run_xdp(xdp_prog, &xdp);

			pkt_len = xdp.data_end - xdp.data;
			pkt_off += xdp.data - orig_data;

			switch (act) {
			case XDP_PASS:
				meta_len_xdp = xdp.data - xdp.data_meta;
				break;
			case XDP_TX:
				dma_off = pkt_off - NFP_NET_RX_BUF_HEADROOM;
				if (unlikely(!nfp_nfdk_tx_xdp_buf(dp, rx_ring,
								  tx_ring,
								  rxbuf,
								  dma_off,
								  pkt_len,
								  &xdp_tx_cmpl)))
					trace_xdp_exception(dp->netdev,
							    xdp_prog, act);
				continue;
			default:
				bpf_warn_invalid_xdp_action(dp->netdev, xdp_prog, act);
				fallthrough;
			case XDP_ABORTED:
				trace_xdp_exception(dp->netdev, xdp_prog, act);
				fallthrough;
			case XDP_DROP:
				nfp_nfdk_rx_give_one(dp, rx_ring, rxbuf->frag,
						     rxbuf->dma_addr);
				continue;
			}
		}

		if (likely(!meta.portid)) {
			netdev = dp->netdev;
		} else if (meta.portid == NFP_META_PORT_ID_CTRL) {
			struct nfp_net *nn = netdev_priv(dp->netdev);

			nfp_app_ctrl_rx_raw(nn->app, rxbuf->frag + pkt_off,
					    pkt_len);
			nfp_nfdk_rx_give_one(dp, rx_ring, rxbuf->frag,
					     rxbuf->dma_addr);
			continue;
		} else {
			struct nfp_net *nn;

			nn = netdev_priv(dp->netdev);
			netdev = nfp_app_dev_get(nn->app, meta.portid,
						 &redir_egress);
			if (unlikely(!netdev)) {
				nfp_nfdk_rx_drop(dp, r_vec, rx_ring, rxbuf,
						 NULL);
				continue;
			}

			if (nfp_netdev_is_nfp_repr(netdev))
				nfp_repr_inc_rx_stats(netdev, pkt_len);
		}

		skb = build_skb(rxbuf->frag, true_bufsz);
		if (unlikely(!skb)) {
			nfp_nfdk_rx_drop(dp, r_vec, rx_ring, rxbuf, NULL);
			continue;
		}
		new_frag = nfp_nfdk_napi_alloc_one(dp, &new_dma_addr);
		if (unlikely(!new_frag)) {
			nfp_nfdk_rx_drop(dp, r_vec, rx_ring, rxbuf, skb);
			continue;
		}

		nfp_net_dma_unmap_rx(dp, rxbuf->dma_addr);

		nfp_nfdk_rx_give_one(dp, rx_ring, new_frag, new_dma_addr);

		skb_reserve(skb, pkt_off);
		skb_put(skb, pkt_len);

		skb->mark = meta.mark;
		skb_set_hash(skb, meta.hash, meta.hash_type);

		skb_record_rx_queue(skb, rx_ring->idx);
		skb->protocol = eth_type_trans(skb, netdev);

		nfp_nfdk_rx_csum(dp, r_vec, rxd, &meta, skb);

		if (rxd->rxd.flags & PCIE_DESC_RX_VLAN)
			__vlan_hwaccel_put_tag(skb, htons(ETH_P_8021Q),
					       le16_to_cpu(rxd->rxd.vlan));
		if (meta_len_xdp)
			skb_metadata_set(skb, meta_len_xdp);

		if (likely(!redir_egress)) {
			napi_gro_receive(&rx_ring->r_vec->napi, skb);
		} else {
			skb->dev = netdev;
			skb_reset_network_header(skb);
			__skb_push(skb, ETH_HLEN);
			dev_queue_xmit(skb);
		}
	}

	if (xdp_prog) {
		if (tx_ring->wr_ptr_add)
			nfp_net_tx_xmit_more_flush(tx_ring);
		else if (unlikely(tx_ring->wr_p != tx_ring->rd_p) &&
			 !xdp_tx_cmpl)
			if (!nfp_nfdk_xdp_complete(tx_ring))
				pkts_polled = budget;
	}

	return pkts_polled;
}

/**
 * nfp_nfdk_poll() - napi poll function
 * @napi:    NAPI structure
 * @budget:  NAPI budget
 *
 * Return: number of packets polled.
 */
int nfp_nfdk_poll(struct napi_struct *napi, int budget)
{
	struct nfp_net_r_vector *r_vec =
		container_of(napi, struct nfp_net_r_vector, napi);
	unsigned int pkts_polled = 0;

	if (r_vec->tx_ring)
		nfp_nfdk_tx_complete(r_vec->tx_ring, budget);
	if (r_vec->rx_ring)
		pkts_polled = nfp_nfdk_rx(r_vec->rx_ring, budget);

	if (pkts_polled < budget)
		if (napi_complete_done(napi, pkts_polled))
			nfp_net_irq_unmask(r_vec->nfp_net, r_vec->irq_entry);

	if (r_vec->nfp_net->rx_coalesce_adapt_on && r_vec->rx_ring) {
		struct dim_sample dim_sample = {};
		unsigned int start;
		u64 pkts, bytes;

		do {
			start = u64_stats_fetch_begin(&r_vec->rx_sync);
			pkts = r_vec->rx_pkts;
			bytes = r_vec->rx_bytes;
		} while (u64_stats_fetch_retry(&r_vec->rx_sync, start));

		dim_update_sample(r_vec->event_ctr, pkts, bytes, &dim_sample);
		net_dim(&r_vec->rx_dim, dim_sample);
	}

	if (r_vec->nfp_net->tx_coalesce_adapt_on && r_vec->tx_ring) {
		struct dim_sample dim_sample = {};
		unsigned int start;
		u64 pkts, bytes;

		do {
			start = u64_stats_fetch_begin(&r_vec->tx_sync);
			pkts = r_vec->tx_pkts;
			bytes = r_vec->tx_bytes;
		} while (u64_stats_fetch_retry(&r_vec->tx_sync, start));

		dim_update_sample(r_vec->event_ctr, pkts, bytes, &dim_sample);
		net_dim(&r_vec->tx_dim, dim_sample);
	}

	return pkts_polled;
}

/* Control device data path
 */

bool
nfp_nfdk_ctrl_tx_one(struct nfp_net *nn, struct nfp_net_r_vector *r_vec,
		     struct sk_buff *skb, bool old)
{
	u32 cnt, tmp_dlen, dlen_type = 0;
	struct nfp_net_tx_ring *tx_ring;
	struct nfp_nfdk_tx_buf *txbuf;
	struct nfp_nfdk_tx_desc *txd;
	unsigned int dma_len, type;
	struct nfp_net_dp *dp;
	dma_addr_t dma_addr;
	u64 metadata = 0;
	int wr_idx;

	dp = &r_vec->nfp_net->dp;
	tx_ring = r_vec->tx_ring;

	if (WARN_ON_ONCE(skb_shinfo(skb)->nr_frags)) {
		nn_dp_warn(dp, "Driver's CTRL TX does not implement gather\n");
		goto err_free;
	}

	/* Don't bother counting frags, assume the worst */
	if (unlikely(nfp_net_tx_full(tx_ring, NFDK_TX_DESC_STOP_CNT))) {
		u64_stats_update_begin(&r_vec->tx_sync);
		r_vec->tx_busy++;
		u64_stats_update_end(&r_vec->tx_sync);
		if (!old)
			__skb_queue_tail(&r_vec->queue, skb);
		else
			__skb_queue_head(&r_vec->queue, skb);
		return NETDEV_TX_BUSY;
	}

	if (nfp_app_ctrl_has_meta(nn->app)) {
		if (unlikely(skb_headroom(skb) < 8)) {
			nn_dp_warn(dp, "CTRL TX on skb without headroom\n");
			goto err_free;
		}
		metadata = NFDK_DESC_TX_CHAIN_META;
		put_unaligned_be32(NFP_META_PORT_ID_CTRL, skb_push(skb, 4));
		put_unaligned_be32(FIELD_PREP(NFDK_META_LEN, 8) |
				   FIELD_PREP(NFDK_META_FIELDS,
					      NFP_NET_META_PORTID),
				   skb_push(skb, 4));
	}

	if (nfp_nfdk_tx_maybe_close_block(tx_ring, 0, skb))
		goto err_free;

	/* DMA map all */
	wr_idx = D_IDX(tx_ring, tx_ring->wr_p);
	txd = &tx_ring->ktxds[wr_idx];
	txbuf = &tx_ring->ktxbufs[wr_idx];

	dma_len = skb_headlen(skb);
	if (dma_len < NFDK_TX_MAX_DATA_PER_HEAD)
		type = NFDK_DESC_TX_TYPE_SIMPLE;
	else
		type = NFDK_DESC_TX_TYPE_GATHER;

	dma_addr = dma_map_single(dp->dev, skb->data, dma_len, DMA_TO_DEVICE);
	if (dma_mapping_error(dp->dev, dma_addr))
		goto err_warn_dma;

	txbuf->skb = skb;
	txbuf++;

	txbuf->dma_addr = dma_addr;
	txbuf++;

	dma_len -= 1;
	dlen_type = FIELD_PREP(NFDK_DESC_TX_DMA_LEN_HEAD, dma_len) |
		    FIELD_PREP(NFDK_DESC_TX_TYPE_HEAD, type);

	txd->dma_len_type = cpu_to_le16(dlen_type);
	nfp_desc_set_dma_addr_48b(txd, dma_addr);

	tmp_dlen = dlen_type & NFDK_DESC_TX_DMA_LEN_HEAD;
	dma_len -= tmp_dlen;
	dma_addr += tmp_dlen + 1;
	txd++;

	while (dma_len > 0) {
		dma_len -= 1;
		dlen_type = FIELD_PREP(NFDK_DESC_TX_DMA_LEN, dma_len);
		txd->dma_len_type = cpu_to_le16(dlen_type);
		nfp_desc_set_dma_addr_48b(txd, dma_addr);

		dlen_type &= NFDK_DESC_TX_DMA_LEN;
		dma_len -= dlen_type;
		dma_addr += dlen_type + 1;
		txd++;
	}

	(txd - 1)->dma_len_type = cpu_to_le16(dlen_type | NFDK_DESC_TX_EOP);

	/* Metadata desc */
	txd->raw = cpu_to_le64(metadata);
	txd++;

	cnt = txd - tx_ring->ktxds - wr_idx;
	if (unlikely(round_down(wr_idx, NFDK_TX_DESC_BLOCK_CNT) !=
		     round_down(wr_idx + cnt - 1, NFDK_TX_DESC_BLOCK_CNT)))
		goto err_warn_overflow;

	tx_ring->wr_p += cnt;
	if (tx_ring->wr_p % NFDK_TX_DESC_BLOCK_CNT)
		tx_ring->data_pending += skb->len;
	else
		tx_ring->data_pending = 0;

	tx_ring->wr_ptr_add += cnt;
	nfp_net_tx_xmit_more_flush(tx_ring);

	return NETDEV_TX_OK;

err_warn_overflow:
	WARN_ONCE(1, "unable to fit packet into a descriptor wr_idx:%d head:%d frags:%d cnt:%d",
		  wr_idx, skb_headlen(skb), 0, cnt);
	txbuf--;
	dma_unmap_single(dp->dev, txbuf->dma_addr,
			 skb_headlen(skb), DMA_TO_DEVICE);
	txbuf->raw = 0;
err_warn_dma:
	nn_dp_warn(dp, "Failed to map DMA TX buffer\n");
err_free:
	u64_stats_update_begin(&r_vec->tx_sync);
	r_vec->tx_errors++;
	u64_stats_update_end(&r_vec->tx_sync);
	dev_kfree_skb_any(skb);
	return NETDEV_TX_OK;
}

static void __nfp_ctrl_tx_queued(struct nfp_net_r_vector *r_vec)
{
	struct sk_buff *skb;

	while ((skb = __skb_dequeue(&r_vec->queue)))
		if (nfp_nfdk_ctrl_tx_one(r_vec->nfp_net, r_vec, skb, true))
			return;
}

static bool
nfp_ctrl_meta_ok(struct nfp_net *nn, void *data, unsigned int meta_len)
{
	u32 meta_type, meta_tag;

	if (!nfp_app_ctrl_has_meta(nn->app))
		return !meta_len;

	if (meta_len != 8)
		return false;

	meta_type = get_unaligned_be32(data);
	meta_tag = get_unaligned_be32(data + 4);

	return (meta_type == NFP_NET_META_PORTID &&
		meta_tag == NFP_META_PORT_ID_CTRL);
}

static bool
nfp_ctrl_rx_one(struct nfp_net *nn, struct nfp_net_dp *dp,
		struct nfp_net_r_vector *r_vec, struct nfp_net_rx_ring *rx_ring)
{
	unsigned int meta_len, data_len, meta_off, pkt_len, pkt_off;
	struct nfp_net_rx_buf *rxbuf;
	struct nfp_net_rx_desc *rxd;
	dma_addr_t new_dma_addr;
	struct sk_buff *skb;
	void *new_frag;
	int idx;

	idx = D_IDX(rx_ring, rx_ring->rd_p);

	rxd = &rx_ring->rxds[idx];
	if (!(rxd->rxd.meta_len_dd & PCIE_DESC_RX_DD))
		return false;

	/* Memory barrier to ensure that we won't do other reads
	 * before the DD bit.
	 */
	dma_rmb();

	rx_ring->rd_p++;

	rxbuf =	&rx_ring->rxbufs[idx];
	meta_len = rxd->rxd.meta_len_dd & PCIE_DESC_RX_META_LEN_MASK;
	data_len = le16_to_cpu(rxd->rxd.data_len);
	pkt_len = data_len - meta_len;

	pkt_off = NFP_NET_RX_BUF_HEADROOM + dp->rx_dma_off;
	if (dp->rx_offset == NFP_NET_CFG_RX_OFFSET_DYNAMIC)
		pkt_off += meta_len;
	else
		pkt_off += dp->rx_offset;
	meta_off = pkt_off - meta_len;

	/* Stats update */
	u64_stats_update_begin(&r_vec->rx_sync);
	r_vec->rx_pkts++;
	r_vec->rx_bytes += pkt_len;
	u64_stats_update_end(&r_vec->rx_sync);

	nfp_net_dma_sync_cpu_rx(dp, rxbuf->dma_addr + meta_off,	data_len);

	if (unlikely(!nfp_ctrl_meta_ok(nn, rxbuf->frag + meta_off, meta_len))) {
		nn_dp_warn(dp, "incorrect metadata for ctrl packet (%d)\n",
			   meta_len);
		nfp_nfdk_rx_drop(dp, r_vec, rx_ring, rxbuf, NULL);
		return true;
	}

	skb = build_skb(rxbuf->frag, dp->fl_bufsz);
	if (unlikely(!skb)) {
		nfp_nfdk_rx_drop(dp, r_vec, rx_ring, rxbuf, NULL);
		return true;
	}
	new_frag = nfp_nfdk_napi_alloc_one(dp, &new_dma_addr);
	if (unlikely(!new_frag)) {
		nfp_nfdk_rx_drop(dp, r_vec, rx_ring, rxbuf, skb);
		return true;
	}

	nfp_net_dma_unmap_rx(dp, rxbuf->dma_addr);

	nfp_nfdk_rx_give_one(dp, rx_ring, new_frag, new_dma_addr);

	skb_reserve(skb, pkt_off);
	skb_put(skb, pkt_len);

	nfp_app_ctrl_rx(nn->app, skb);

	return true;
}

static bool nfp_ctrl_rx(struct nfp_net_r_vector *r_vec)
{
	struct nfp_net_rx_ring *rx_ring = r_vec->rx_ring;
	struct nfp_net *nn = r_vec->nfp_net;
	struct nfp_net_dp *dp = &nn->dp;
	unsigned int budget = 512;

	while (nfp_ctrl_rx_one(nn, dp, r_vec, rx_ring) && budget--)
		continue;

	return budget;
}

void nfp_nfdk_ctrl_poll(struct tasklet_struct *t)
{
	struct nfp_net_r_vector *r_vec = from_tasklet(r_vec, t, tasklet);

	spin_lock(&r_vec->lock);
	nfp_nfdk_tx_complete(r_vec->tx_ring, 0);
	__nfp_ctrl_tx_queued(r_vec);
	spin_unlock(&r_vec->lock);

	if (nfp_ctrl_rx(r_vec)) {
		nfp_net_irq_unmask(r_vec->nfp_net, r_vec->irq_entry);
	} else {
		tasklet_schedule(&r_vec->tasklet);
		nn_dp_warn(&r_vec->nfp_net->dp,
			   "control message budget exceeded!\n");
	}
}