Loading...
Note: File does not exist in v4.6.
1// SPDX-License-Identifier: GPL-2.0
2/* Copyright (c) 2018, Intel Corporation. */
3
4#include "ice.h"
5#include "ice_base.h"
6#include "ice_flow.h"
7#include "ice_lib.h"
8#include "ice_fltr.h"
9#include "ice_dcb_lib.h"
10#include "ice_devlink.h"
11
12/**
13 * ice_vsi_type_str - maps VSI type enum to string equivalents
14 * @vsi_type: VSI type enum
15 */
16const char *ice_vsi_type_str(enum ice_vsi_type vsi_type)
17{
18 switch (vsi_type) {
19 case ICE_VSI_PF:
20 return "ICE_VSI_PF";
21 case ICE_VSI_VF:
22 return "ICE_VSI_VF";
23 case ICE_VSI_CTRL:
24 return "ICE_VSI_CTRL";
25 case ICE_VSI_LB:
26 return "ICE_VSI_LB";
27 default:
28 return "unknown";
29 }
30}
31
32/**
33 * ice_vsi_ctrl_all_rx_rings - Start or stop a VSI's Rx rings
34 * @vsi: the VSI being configured
35 * @ena: start or stop the Rx rings
36 *
37 * First enable/disable all of the Rx rings, flush any remaining writes, and
38 * then verify that they have all been enabled/disabled successfully. This will
39 * let all of the register writes complete when enabling/disabling the Rx rings
40 * before waiting for the change in hardware to complete.
41 */
42static int ice_vsi_ctrl_all_rx_rings(struct ice_vsi *vsi, bool ena)
43{
44 int ret = 0;
45 u16 i;
46
47 for (i = 0; i < vsi->num_rxq; i++)
48 ice_vsi_ctrl_one_rx_ring(vsi, ena, i, false);
49
50 ice_flush(&vsi->back->hw);
51
52 for (i = 0; i < vsi->num_rxq; i++) {
53 ret = ice_vsi_wait_one_rx_ring(vsi, ena, i);
54 if (ret)
55 break;
56 }
57
58 return ret;
59}
60
61/**
62 * ice_vsi_alloc_arrays - Allocate queue and vector pointer arrays for the VSI
63 * @vsi: VSI pointer
64 *
65 * On error: returns error code (negative)
66 * On success: returns 0
67 */
68static int ice_vsi_alloc_arrays(struct ice_vsi *vsi)
69{
70 struct ice_pf *pf = vsi->back;
71 struct device *dev;
72
73 dev = ice_pf_to_dev(pf);
74
75 /* allocate memory for both Tx and Rx ring pointers */
76 vsi->tx_rings = devm_kcalloc(dev, vsi->alloc_txq,
77 sizeof(*vsi->tx_rings), GFP_KERNEL);
78 if (!vsi->tx_rings)
79 return -ENOMEM;
80
81 vsi->rx_rings = devm_kcalloc(dev, vsi->alloc_rxq,
82 sizeof(*vsi->rx_rings), GFP_KERNEL);
83 if (!vsi->rx_rings)
84 goto err_rings;
85
86 /* XDP will have vsi->alloc_txq Tx queues as well, so double the size */
87 vsi->txq_map = devm_kcalloc(dev, (2 * vsi->alloc_txq),
88 sizeof(*vsi->txq_map), GFP_KERNEL);
89
90 if (!vsi->txq_map)
91 goto err_txq_map;
92
93 vsi->rxq_map = devm_kcalloc(dev, vsi->alloc_rxq,
94 sizeof(*vsi->rxq_map), GFP_KERNEL);
95 if (!vsi->rxq_map)
96 goto err_rxq_map;
97
98 /* There is no need to allocate q_vectors for a loopback VSI. */
99 if (vsi->type == ICE_VSI_LB)
100 return 0;
101
102 /* allocate memory for q_vector pointers */
103 vsi->q_vectors = devm_kcalloc(dev, vsi->num_q_vectors,
104 sizeof(*vsi->q_vectors), GFP_KERNEL);
105 if (!vsi->q_vectors)
106 goto err_vectors;
107
108 vsi->af_xdp_zc_qps = bitmap_zalloc(max_t(int, vsi->alloc_txq, vsi->alloc_rxq), GFP_KERNEL);
109 if (!vsi->af_xdp_zc_qps)
110 goto err_zc_qps;
111
112 return 0;
113
114err_zc_qps:
115 devm_kfree(dev, vsi->q_vectors);
116err_vectors:
117 devm_kfree(dev, vsi->rxq_map);
118err_rxq_map:
119 devm_kfree(dev, vsi->txq_map);
120err_txq_map:
121 devm_kfree(dev, vsi->rx_rings);
122err_rings:
123 devm_kfree(dev, vsi->tx_rings);
124 return -ENOMEM;
125}
126
127/**
128 * ice_vsi_set_num_desc - Set number of descriptors for queues on this VSI
129 * @vsi: the VSI being configured
130 */
131static void ice_vsi_set_num_desc(struct ice_vsi *vsi)
132{
133 switch (vsi->type) {
134 case ICE_VSI_PF:
135 case ICE_VSI_CTRL:
136 case ICE_VSI_LB:
137 /* a user could change the values of num_[tr]x_desc using
138 * ethtool -G so we should keep those values instead of
139 * overwriting them with the defaults.
140 */
141 if (!vsi->num_rx_desc)
142 vsi->num_rx_desc = ICE_DFLT_NUM_RX_DESC;
143 if (!vsi->num_tx_desc)
144 vsi->num_tx_desc = ICE_DFLT_NUM_TX_DESC;
145 break;
146 default:
147 dev_dbg(ice_pf_to_dev(vsi->back), "Not setting number of Tx/Rx descriptors for VSI type %d\n",
148 vsi->type);
149 break;
150 }
151}
152
153/**
154 * ice_vsi_set_num_qs - Set number of queues, descriptors and vectors for a VSI
155 * @vsi: the VSI being configured
156 * @vf_id: ID of the VF being configured
157 *
158 * Return 0 on success and a negative value on error
159 */
160static void ice_vsi_set_num_qs(struct ice_vsi *vsi, u16 vf_id)
161{
162 struct ice_pf *pf = vsi->back;
163 struct ice_vf *vf = NULL;
164
165 if (vsi->type == ICE_VSI_VF)
166 vsi->vf_id = vf_id;
167 else
168 vsi->vf_id = ICE_INVAL_VFID;
169
170 switch (vsi->type) {
171 case ICE_VSI_PF:
172 if (vsi->req_txq) {
173 vsi->alloc_txq = vsi->req_txq;
174 vsi->num_txq = vsi->req_txq;
175 } else {
176 vsi->alloc_txq = min3(pf->num_lan_msix,
177 ice_get_avail_txq_count(pf),
178 (u16)num_online_cpus());
179 }
180
181 pf->num_lan_tx = vsi->alloc_txq;
182
183 /* only 1 Rx queue unless RSS is enabled */
184 if (!test_bit(ICE_FLAG_RSS_ENA, pf->flags)) {
185 vsi->alloc_rxq = 1;
186 } else {
187 if (vsi->req_rxq) {
188 vsi->alloc_rxq = vsi->req_rxq;
189 vsi->num_rxq = vsi->req_rxq;
190 } else {
191 vsi->alloc_rxq = min3(pf->num_lan_msix,
192 ice_get_avail_rxq_count(pf),
193 (u16)num_online_cpus());
194 }
195 }
196
197 pf->num_lan_rx = vsi->alloc_rxq;
198
199 vsi->num_q_vectors = min_t(int, pf->num_lan_msix,
200 max_t(int, vsi->alloc_rxq,
201 vsi->alloc_txq));
202 break;
203 case ICE_VSI_VF:
204 vf = &pf->vf[vsi->vf_id];
205 if (vf->num_req_qs)
206 vf->num_vf_qs = vf->num_req_qs;
207 vsi->alloc_txq = vf->num_vf_qs;
208 vsi->alloc_rxq = vf->num_vf_qs;
209 /* pf->num_msix_per_vf includes (VF miscellaneous vector +
210 * data queue interrupts). Since vsi->num_q_vectors is number
211 * of queues vectors, subtract 1 (ICE_NONQ_VECS_VF) from the
212 * original vector count
213 */
214 vsi->num_q_vectors = pf->num_msix_per_vf - ICE_NONQ_VECS_VF;
215 break;
216 case ICE_VSI_CTRL:
217 vsi->alloc_txq = 1;
218 vsi->alloc_rxq = 1;
219 vsi->num_q_vectors = 1;
220 break;
221 case ICE_VSI_LB:
222 vsi->alloc_txq = 1;
223 vsi->alloc_rxq = 1;
224 break;
225 default:
226 dev_warn(ice_pf_to_dev(pf), "Unknown VSI type %d\n", vsi->type);
227 break;
228 }
229
230 ice_vsi_set_num_desc(vsi);
231}
232
233/**
234 * ice_get_free_slot - get the next non-NULL location index in array
235 * @array: array to search
236 * @size: size of the array
237 * @curr: last known occupied index to be used as a search hint
238 *
239 * void * is being used to keep the functionality generic. This lets us use this
240 * function on any array of pointers.
241 */
242static int ice_get_free_slot(void *array, int size, int curr)
243{
244 int **tmp_array = (int **)array;
245 int next;
246
247 if (curr < (size - 1) && !tmp_array[curr + 1]) {
248 next = curr + 1;
249 } else {
250 int i = 0;
251
252 while ((i < size) && (tmp_array[i]))
253 i++;
254 if (i == size)
255 next = ICE_NO_VSI;
256 else
257 next = i;
258 }
259 return next;
260}
261
262/**
263 * ice_vsi_delete - delete a VSI from the switch
264 * @vsi: pointer to VSI being removed
265 */
266static void ice_vsi_delete(struct ice_vsi *vsi)
267{
268 struct ice_pf *pf = vsi->back;
269 struct ice_vsi_ctx *ctxt;
270 enum ice_status status;
271
272 ctxt = kzalloc(sizeof(*ctxt), GFP_KERNEL);
273 if (!ctxt)
274 return;
275
276 if (vsi->type == ICE_VSI_VF)
277 ctxt->vf_num = vsi->vf_id;
278 ctxt->vsi_num = vsi->vsi_num;
279
280 memcpy(&ctxt->info, &vsi->info, sizeof(ctxt->info));
281
282 status = ice_free_vsi(&pf->hw, vsi->idx, ctxt, false, NULL);
283 if (status)
284 dev_err(ice_pf_to_dev(pf), "Failed to delete VSI %i in FW - error: %s\n",
285 vsi->vsi_num, ice_stat_str(status));
286
287 kfree(ctxt);
288}
289
290/**
291 * ice_vsi_free_arrays - De-allocate queue and vector pointer arrays for the VSI
292 * @vsi: pointer to VSI being cleared
293 */
294static void ice_vsi_free_arrays(struct ice_vsi *vsi)
295{
296 struct ice_pf *pf = vsi->back;
297 struct device *dev;
298
299 dev = ice_pf_to_dev(pf);
300
301 if (vsi->af_xdp_zc_qps) {
302 bitmap_free(vsi->af_xdp_zc_qps);
303 vsi->af_xdp_zc_qps = NULL;
304 }
305 /* free the ring and vector containers */
306 if (vsi->q_vectors) {
307 devm_kfree(dev, vsi->q_vectors);
308 vsi->q_vectors = NULL;
309 }
310 if (vsi->tx_rings) {
311 devm_kfree(dev, vsi->tx_rings);
312 vsi->tx_rings = NULL;
313 }
314 if (vsi->rx_rings) {
315 devm_kfree(dev, vsi->rx_rings);
316 vsi->rx_rings = NULL;
317 }
318 if (vsi->txq_map) {
319 devm_kfree(dev, vsi->txq_map);
320 vsi->txq_map = NULL;
321 }
322 if (vsi->rxq_map) {
323 devm_kfree(dev, vsi->rxq_map);
324 vsi->rxq_map = NULL;
325 }
326}
327
328/**
329 * ice_vsi_clear - clean up and deallocate the provided VSI
330 * @vsi: pointer to VSI being cleared
331 *
332 * This deallocates the VSI's queue resources, removes it from the PF's
333 * VSI array if necessary, and deallocates the VSI
334 *
335 * Returns 0 on success, negative on failure
336 */
337static int ice_vsi_clear(struct ice_vsi *vsi)
338{
339 struct ice_pf *pf = NULL;
340 struct device *dev;
341
342 if (!vsi)
343 return 0;
344
345 if (!vsi->back)
346 return -EINVAL;
347
348 pf = vsi->back;
349 dev = ice_pf_to_dev(pf);
350
351 if (!pf->vsi[vsi->idx] || pf->vsi[vsi->idx] != vsi) {
352 dev_dbg(dev, "vsi does not exist at pf->vsi[%d]\n", vsi->idx);
353 return -EINVAL;
354 }
355
356 mutex_lock(&pf->sw_mutex);
357 /* updates the PF for this cleared VSI */
358
359 pf->vsi[vsi->idx] = NULL;
360 if (vsi->idx < pf->next_vsi && vsi->type != ICE_VSI_CTRL)
361 pf->next_vsi = vsi->idx;
362 if (vsi->idx < pf->next_vsi && vsi->type == ICE_VSI_CTRL &&
363 vsi->vf_id != ICE_INVAL_VFID)
364 pf->next_vsi = vsi->idx;
365
366 ice_vsi_free_arrays(vsi);
367 mutex_unlock(&pf->sw_mutex);
368 devm_kfree(dev, vsi);
369
370 return 0;
371}
372
373/**
374 * ice_msix_clean_ctrl_vsi - MSIX mode interrupt handler for ctrl VSI
375 * @irq: interrupt number
376 * @data: pointer to a q_vector
377 */
378static irqreturn_t ice_msix_clean_ctrl_vsi(int __always_unused irq, void *data)
379{
380 struct ice_q_vector *q_vector = (struct ice_q_vector *)data;
381
382 if (!q_vector->tx.ring)
383 return IRQ_HANDLED;
384
385#define FDIR_RX_DESC_CLEAN_BUDGET 64
386 ice_clean_rx_irq(q_vector->rx.ring, FDIR_RX_DESC_CLEAN_BUDGET);
387 ice_clean_ctrl_tx_irq(q_vector->tx.ring);
388
389 return IRQ_HANDLED;
390}
391
392/**
393 * ice_msix_clean_rings - MSIX mode Interrupt Handler
394 * @irq: interrupt number
395 * @data: pointer to a q_vector
396 */
397static irqreturn_t ice_msix_clean_rings(int __always_unused irq, void *data)
398{
399 struct ice_q_vector *q_vector = (struct ice_q_vector *)data;
400
401 if (!q_vector->tx.ring && !q_vector->rx.ring)
402 return IRQ_HANDLED;
403
404 q_vector->total_events++;
405
406 napi_schedule(&q_vector->napi);
407
408 return IRQ_HANDLED;
409}
410
411/**
412 * ice_vsi_alloc - Allocates the next available struct VSI in the PF
413 * @pf: board private structure
414 * @vsi_type: type of VSI
415 * @vf_id: ID of the VF being configured
416 *
417 * returns a pointer to a VSI on success, NULL on failure.
418 */
419static struct ice_vsi *
420ice_vsi_alloc(struct ice_pf *pf, enum ice_vsi_type vsi_type, u16 vf_id)
421{
422 struct device *dev = ice_pf_to_dev(pf);
423 struct ice_vsi *vsi = NULL;
424
425 /* Need to protect the allocation of the VSIs at the PF level */
426 mutex_lock(&pf->sw_mutex);
427
428 /* If we have already allocated our maximum number of VSIs,
429 * pf->next_vsi will be ICE_NO_VSI. If not, pf->next_vsi index
430 * is available to be populated
431 */
432 if (pf->next_vsi == ICE_NO_VSI) {
433 dev_dbg(dev, "out of VSI slots!\n");
434 goto unlock_pf;
435 }
436
437 vsi = devm_kzalloc(dev, sizeof(*vsi), GFP_KERNEL);
438 if (!vsi)
439 goto unlock_pf;
440
441 vsi->type = vsi_type;
442 vsi->back = pf;
443 set_bit(ICE_VSI_DOWN, vsi->state);
444
445 if (vsi_type == ICE_VSI_VF)
446 ice_vsi_set_num_qs(vsi, vf_id);
447 else
448 ice_vsi_set_num_qs(vsi, ICE_INVAL_VFID);
449
450 switch (vsi->type) {
451 case ICE_VSI_PF:
452 if (ice_vsi_alloc_arrays(vsi))
453 goto err_rings;
454
455 /* Setup default MSIX irq handler for VSI */
456 vsi->irq_handler = ice_msix_clean_rings;
457 break;
458 case ICE_VSI_CTRL:
459 if (ice_vsi_alloc_arrays(vsi))
460 goto err_rings;
461
462 /* Setup ctrl VSI MSIX irq handler */
463 vsi->irq_handler = ice_msix_clean_ctrl_vsi;
464 break;
465 case ICE_VSI_VF:
466 if (ice_vsi_alloc_arrays(vsi))
467 goto err_rings;
468 break;
469 case ICE_VSI_LB:
470 if (ice_vsi_alloc_arrays(vsi))
471 goto err_rings;
472 break;
473 default:
474 dev_warn(dev, "Unknown VSI type %d\n", vsi->type);
475 goto unlock_pf;
476 }
477
478 if (vsi->type == ICE_VSI_CTRL && vf_id == ICE_INVAL_VFID) {
479 /* Use the last VSI slot as the index for PF control VSI */
480 vsi->idx = pf->num_alloc_vsi - 1;
481 pf->ctrl_vsi_idx = vsi->idx;
482 pf->vsi[vsi->idx] = vsi;
483 } else {
484 /* fill slot and make note of the index */
485 vsi->idx = pf->next_vsi;
486 pf->vsi[pf->next_vsi] = vsi;
487
488 /* prepare pf->next_vsi for next use */
489 pf->next_vsi = ice_get_free_slot(pf->vsi, pf->num_alloc_vsi,
490 pf->next_vsi);
491 }
492
493 if (vsi->type == ICE_VSI_CTRL && vf_id != ICE_INVAL_VFID)
494 pf->vf[vf_id].ctrl_vsi_idx = vsi->idx;
495 goto unlock_pf;
496
497err_rings:
498 devm_kfree(dev, vsi);
499 vsi = NULL;
500unlock_pf:
501 mutex_unlock(&pf->sw_mutex);
502 return vsi;
503}
504
505/**
506 * ice_alloc_fd_res - Allocate FD resource for a VSI
507 * @vsi: pointer to the ice_vsi
508 *
509 * This allocates the FD resources
510 *
511 * Returns 0 on success, -EPERM on no-op or -EIO on failure
512 */
513static int ice_alloc_fd_res(struct ice_vsi *vsi)
514{
515 struct ice_pf *pf = vsi->back;
516 u32 g_val, b_val;
517
518 /* Flow Director filters are only allocated/assigned to the PF VSI which
519 * passes the traffic. The CTRL VSI is only used to add/delete filters
520 * so we don't allocate resources to it
521 */
522
523 /* FD filters from guaranteed pool per VSI */
524 g_val = pf->hw.func_caps.fd_fltr_guar;
525 if (!g_val)
526 return -EPERM;
527
528 /* FD filters from best effort pool */
529 b_val = pf->hw.func_caps.fd_fltr_best_effort;
530 if (!b_val)
531 return -EPERM;
532
533 if (!(vsi->type == ICE_VSI_PF || vsi->type == ICE_VSI_VF))
534 return -EPERM;
535
536 if (!test_bit(ICE_FLAG_FD_ENA, pf->flags))
537 return -EPERM;
538
539 vsi->num_gfltr = g_val / pf->num_alloc_vsi;
540
541 /* each VSI gets same "best_effort" quota */
542 vsi->num_bfltr = b_val;
543
544 if (vsi->type == ICE_VSI_VF) {
545 vsi->num_gfltr = 0;
546
547 /* each VSI gets same "best_effort" quota */
548 vsi->num_bfltr = b_val;
549 }
550
551 return 0;
552}
553
554/**
555 * ice_vsi_get_qs - Assign queues from PF to VSI
556 * @vsi: the VSI to assign queues to
557 *
558 * Returns 0 on success and a negative value on error
559 */
560static int ice_vsi_get_qs(struct ice_vsi *vsi)
561{
562 struct ice_pf *pf = vsi->back;
563 struct ice_qs_cfg tx_qs_cfg = {
564 .qs_mutex = &pf->avail_q_mutex,
565 .pf_map = pf->avail_txqs,
566 .pf_map_size = pf->max_pf_txqs,
567 .q_count = vsi->alloc_txq,
568 .scatter_count = ICE_MAX_SCATTER_TXQS,
569 .vsi_map = vsi->txq_map,
570 .vsi_map_offset = 0,
571 .mapping_mode = ICE_VSI_MAP_CONTIG
572 };
573 struct ice_qs_cfg rx_qs_cfg = {
574 .qs_mutex = &pf->avail_q_mutex,
575 .pf_map = pf->avail_rxqs,
576 .pf_map_size = pf->max_pf_rxqs,
577 .q_count = vsi->alloc_rxq,
578 .scatter_count = ICE_MAX_SCATTER_RXQS,
579 .vsi_map = vsi->rxq_map,
580 .vsi_map_offset = 0,
581 .mapping_mode = ICE_VSI_MAP_CONTIG
582 };
583 int ret;
584
585 ret = __ice_vsi_get_qs(&tx_qs_cfg);
586 if (ret)
587 return ret;
588 vsi->tx_mapping_mode = tx_qs_cfg.mapping_mode;
589
590 ret = __ice_vsi_get_qs(&rx_qs_cfg);
591 if (ret)
592 return ret;
593 vsi->rx_mapping_mode = rx_qs_cfg.mapping_mode;
594
595 return 0;
596}
597
598/**
599 * ice_vsi_put_qs - Release queues from VSI to PF
600 * @vsi: the VSI that is going to release queues
601 */
602static void ice_vsi_put_qs(struct ice_vsi *vsi)
603{
604 struct ice_pf *pf = vsi->back;
605 int i;
606
607 mutex_lock(&pf->avail_q_mutex);
608
609 for (i = 0; i < vsi->alloc_txq; i++) {
610 clear_bit(vsi->txq_map[i], pf->avail_txqs);
611 vsi->txq_map[i] = ICE_INVAL_Q_INDEX;
612 }
613
614 for (i = 0; i < vsi->alloc_rxq; i++) {
615 clear_bit(vsi->rxq_map[i], pf->avail_rxqs);
616 vsi->rxq_map[i] = ICE_INVAL_Q_INDEX;
617 }
618
619 mutex_unlock(&pf->avail_q_mutex);
620}
621
622/**
623 * ice_is_safe_mode
624 * @pf: pointer to the PF struct
625 *
626 * returns true if driver is in safe mode, false otherwise
627 */
628bool ice_is_safe_mode(struct ice_pf *pf)
629{
630 return !test_bit(ICE_FLAG_ADV_FEATURES, pf->flags);
631}
632
633/**
634 * ice_is_aux_ena
635 * @pf: pointer to the PF struct
636 *
637 * returns true if AUX devices/drivers are supported, false otherwise
638 */
639bool ice_is_aux_ena(struct ice_pf *pf)
640{
641 return test_bit(ICE_FLAG_AUX_ENA, pf->flags);
642}
643
644/**
645 * ice_vsi_clean_rss_flow_fld - Delete RSS configuration
646 * @vsi: the VSI being cleaned up
647 *
648 * This function deletes RSS input set for all flows that were configured
649 * for this VSI
650 */
651static void ice_vsi_clean_rss_flow_fld(struct ice_vsi *vsi)
652{
653 struct ice_pf *pf = vsi->back;
654 enum ice_status status;
655
656 if (ice_is_safe_mode(pf))
657 return;
658
659 status = ice_rem_vsi_rss_cfg(&pf->hw, vsi->idx);
660 if (status)
661 dev_dbg(ice_pf_to_dev(pf), "ice_rem_vsi_rss_cfg failed for vsi = %d, error = %s\n",
662 vsi->vsi_num, ice_stat_str(status));
663}
664
665/**
666 * ice_rss_clean - Delete RSS related VSI structures and configuration
667 * @vsi: the VSI being removed
668 */
669static void ice_rss_clean(struct ice_vsi *vsi)
670{
671 struct ice_pf *pf = vsi->back;
672 struct device *dev;
673
674 dev = ice_pf_to_dev(pf);
675
676 if (vsi->rss_hkey_user)
677 devm_kfree(dev, vsi->rss_hkey_user);
678 if (vsi->rss_lut_user)
679 devm_kfree(dev, vsi->rss_lut_user);
680
681 ice_vsi_clean_rss_flow_fld(vsi);
682 /* remove RSS replay list */
683 if (!ice_is_safe_mode(pf))
684 ice_rem_vsi_rss_list(&pf->hw, vsi->idx);
685}
686
687/**
688 * ice_vsi_set_rss_params - Setup RSS capabilities per VSI type
689 * @vsi: the VSI being configured
690 */
691static void ice_vsi_set_rss_params(struct ice_vsi *vsi)
692{
693 struct ice_hw_common_caps *cap;
694 struct ice_pf *pf = vsi->back;
695
696 if (!test_bit(ICE_FLAG_RSS_ENA, pf->flags)) {
697 vsi->rss_size = 1;
698 return;
699 }
700
701 cap = &pf->hw.func_caps.common_cap;
702 switch (vsi->type) {
703 case ICE_VSI_PF:
704 /* PF VSI will inherit RSS instance of PF */
705 vsi->rss_table_size = (u16)cap->rss_table_size;
706 vsi->rss_size = min_t(u16, num_online_cpus(),
707 BIT(cap->rss_table_entry_width));
708 vsi->rss_lut_type = ICE_AQC_GSET_RSS_LUT_TABLE_TYPE_PF;
709 break;
710 case ICE_VSI_VF:
711 /* VF VSI will get a small RSS table.
712 * For VSI_LUT, LUT size should be set to 64 bytes.
713 */
714 vsi->rss_table_size = ICE_VSIQF_HLUT_ARRAY_SIZE;
715 vsi->rss_size = ICE_MAX_RSS_QS_PER_VF;
716 vsi->rss_lut_type = ICE_AQC_GSET_RSS_LUT_TABLE_TYPE_VSI;
717 break;
718 case ICE_VSI_LB:
719 break;
720 default:
721 dev_dbg(ice_pf_to_dev(pf), "Unsupported VSI type %s\n",
722 ice_vsi_type_str(vsi->type));
723 break;
724 }
725}
726
727/**
728 * ice_set_dflt_vsi_ctx - Set default VSI context before adding a VSI
729 * @ctxt: the VSI context being set
730 *
731 * This initializes a default VSI context for all sections except the Queues.
732 */
733static void ice_set_dflt_vsi_ctx(struct ice_vsi_ctx *ctxt)
734{
735 u32 table = 0;
736
737 memset(&ctxt->info, 0, sizeof(ctxt->info));
738 /* VSI's should be allocated from shared pool */
739 ctxt->alloc_from_pool = true;
740 /* Src pruning enabled by default */
741 ctxt->info.sw_flags = ICE_AQ_VSI_SW_FLAG_SRC_PRUNE;
742 /* Traffic from VSI can be sent to LAN */
743 ctxt->info.sw_flags2 = ICE_AQ_VSI_SW_FLAG_LAN_ENA;
744 /* By default bits 3 and 4 in vlan_flags are 0's which results in legacy
745 * behavior (show VLAN, DEI, and UP) in descriptor. Also, allow all
746 * packets untagged/tagged.
747 */
748 ctxt->info.vlan_flags = ((ICE_AQ_VSI_VLAN_MODE_ALL &
749 ICE_AQ_VSI_VLAN_MODE_M) >>
750 ICE_AQ_VSI_VLAN_MODE_S);
751 /* Have 1:1 UP mapping for both ingress/egress tables */
752 table |= ICE_UP_TABLE_TRANSLATE(0, 0);
753 table |= ICE_UP_TABLE_TRANSLATE(1, 1);
754 table |= ICE_UP_TABLE_TRANSLATE(2, 2);
755 table |= ICE_UP_TABLE_TRANSLATE(3, 3);
756 table |= ICE_UP_TABLE_TRANSLATE(4, 4);
757 table |= ICE_UP_TABLE_TRANSLATE(5, 5);
758 table |= ICE_UP_TABLE_TRANSLATE(6, 6);
759 table |= ICE_UP_TABLE_TRANSLATE(7, 7);
760 ctxt->info.ingress_table = cpu_to_le32(table);
761 ctxt->info.egress_table = cpu_to_le32(table);
762 /* Have 1:1 UP mapping for outer to inner UP table */
763 ctxt->info.outer_up_table = cpu_to_le32(table);
764 /* No Outer tag support outer_tag_flags remains to zero */
765}
766
767/**
768 * ice_vsi_setup_q_map - Setup a VSI queue map
769 * @vsi: the VSI being configured
770 * @ctxt: VSI context structure
771 */
772static void ice_vsi_setup_q_map(struct ice_vsi *vsi, struct ice_vsi_ctx *ctxt)
773{
774 u16 offset = 0, qmap = 0, tx_count = 0, pow = 0;
775 u16 num_txq_per_tc, num_rxq_per_tc;
776 u16 qcount_tx = vsi->alloc_txq;
777 u16 qcount_rx = vsi->alloc_rxq;
778 bool ena_tc0 = false;
779 u8 netdev_tc = 0;
780 int i;
781
782 /* at least TC0 should be enabled by default */
783 if (vsi->tc_cfg.numtc) {
784 if (!(vsi->tc_cfg.ena_tc & BIT(0)))
785 ena_tc0 = true;
786 } else {
787 ena_tc0 = true;
788 }
789
790 if (ena_tc0) {
791 vsi->tc_cfg.numtc++;
792 vsi->tc_cfg.ena_tc |= 1;
793 }
794
795 num_rxq_per_tc = min_t(u16, qcount_rx / vsi->tc_cfg.numtc, ICE_MAX_RXQS_PER_TC);
796 if (!num_rxq_per_tc)
797 num_rxq_per_tc = 1;
798 num_txq_per_tc = qcount_tx / vsi->tc_cfg.numtc;
799 if (!num_txq_per_tc)
800 num_txq_per_tc = 1;
801
802 /* find the (rounded up) power-of-2 of qcount */
803 pow = (u16)order_base_2(num_rxq_per_tc);
804
805 /* TC mapping is a function of the number of Rx queues assigned to the
806 * VSI for each traffic class and the offset of these queues.
807 * The first 10 bits are for queue offset for TC0, next 4 bits for no:of
808 * queues allocated to TC0. No:of queues is a power-of-2.
809 *
810 * If TC is not enabled, the queue offset is set to 0, and allocate one
811 * queue, this way, traffic for the given TC will be sent to the default
812 * queue.
813 *
814 * Setup number and offset of Rx queues for all TCs for the VSI
815 */
816 ice_for_each_traffic_class(i) {
817 if (!(vsi->tc_cfg.ena_tc & BIT(i))) {
818 /* TC is not enabled */
819 vsi->tc_cfg.tc_info[i].qoffset = 0;
820 vsi->tc_cfg.tc_info[i].qcount_rx = 1;
821 vsi->tc_cfg.tc_info[i].qcount_tx = 1;
822 vsi->tc_cfg.tc_info[i].netdev_tc = 0;
823 ctxt->info.tc_mapping[i] = 0;
824 continue;
825 }
826
827 /* TC is enabled */
828 vsi->tc_cfg.tc_info[i].qoffset = offset;
829 vsi->tc_cfg.tc_info[i].qcount_rx = num_rxq_per_tc;
830 vsi->tc_cfg.tc_info[i].qcount_tx = num_txq_per_tc;
831 vsi->tc_cfg.tc_info[i].netdev_tc = netdev_tc++;
832
833 qmap = ((offset << ICE_AQ_VSI_TC_Q_OFFSET_S) &
834 ICE_AQ_VSI_TC_Q_OFFSET_M) |
835 ((pow << ICE_AQ_VSI_TC_Q_NUM_S) &
836 ICE_AQ_VSI_TC_Q_NUM_M);
837 offset += num_rxq_per_tc;
838 tx_count += num_txq_per_tc;
839 ctxt->info.tc_mapping[i] = cpu_to_le16(qmap);
840 }
841
842 /* if offset is non-zero, means it is calculated correctly based on
843 * enabled TCs for a given VSI otherwise qcount_rx will always
844 * be correct and non-zero because it is based off - VSI's
845 * allocated Rx queues which is at least 1 (hence qcount_tx will be
846 * at least 1)
847 */
848 if (offset)
849 vsi->num_rxq = offset;
850 else
851 vsi->num_rxq = num_rxq_per_tc;
852
853 vsi->num_txq = tx_count;
854
855 if (vsi->type == ICE_VSI_VF && vsi->num_txq != vsi->num_rxq) {
856 dev_dbg(ice_pf_to_dev(vsi->back), "VF VSI should have same number of Tx and Rx queues. Hence making them equal\n");
857 /* since there is a chance that num_rxq could have been changed
858 * in the above for loop, make num_txq equal to num_rxq.
859 */
860 vsi->num_txq = vsi->num_rxq;
861 }
862
863 /* Rx queue mapping */
864 ctxt->info.mapping_flags |= cpu_to_le16(ICE_AQ_VSI_Q_MAP_CONTIG);
865 /* q_mapping buffer holds the info for the first queue allocated for
866 * this VSI in the PF space and also the number of queues associated
867 * with this VSI.
868 */
869 ctxt->info.q_mapping[0] = cpu_to_le16(vsi->rxq_map[0]);
870 ctxt->info.q_mapping[1] = cpu_to_le16(vsi->num_rxq);
871}
872
873/**
874 * ice_set_fd_vsi_ctx - Set FD VSI context before adding a VSI
875 * @ctxt: the VSI context being set
876 * @vsi: the VSI being configured
877 */
878static void ice_set_fd_vsi_ctx(struct ice_vsi_ctx *ctxt, struct ice_vsi *vsi)
879{
880 u8 dflt_q_group, dflt_q_prio;
881 u16 dflt_q, report_q, val;
882
883 if (vsi->type != ICE_VSI_PF && vsi->type != ICE_VSI_CTRL &&
884 vsi->type != ICE_VSI_VF)
885 return;
886
887 val = ICE_AQ_VSI_PROP_FLOW_DIR_VALID;
888 ctxt->info.valid_sections |= cpu_to_le16(val);
889 dflt_q = 0;
890 dflt_q_group = 0;
891 report_q = 0;
892 dflt_q_prio = 0;
893
894 /* enable flow director filtering/programming */
895 val = ICE_AQ_VSI_FD_ENABLE | ICE_AQ_VSI_FD_PROG_ENABLE;
896 ctxt->info.fd_options = cpu_to_le16(val);
897 /* max of allocated flow director filters */
898 ctxt->info.max_fd_fltr_dedicated =
899 cpu_to_le16(vsi->num_gfltr);
900 /* max of shared flow director filters any VSI may program */
901 ctxt->info.max_fd_fltr_shared =
902 cpu_to_le16(vsi->num_bfltr);
903 /* default queue index within the VSI of the default FD */
904 val = ((dflt_q << ICE_AQ_VSI_FD_DEF_Q_S) &
905 ICE_AQ_VSI_FD_DEF_Q_M);
906 /* target queue or queue group to the FD filter */
907 val |= ((dflt_q_group << ICE_AQ_VSI_FD_DEF_GRP_S) &
908 ICE_AQ_VSI_FD_DEF_GRP_M);
909 ctxt->info.fd_def_q = cpu_to_le16(val);
910 /* queue index on which FD filter completion is reported */
911 val = ((report_q << ICE_AQ_VSI_FD_REPORT_Q_S) &
912 ICE_AQ_VSI_FD_REPORT_Q_M);
913 /* priority of the default qindex action */
914 val |= ((dflt_q_prio << ICE_AQ_VSI_FD_DEF_PRIORITY_S) &
915 ICE_AQ_VSI_FD_DEF_PRIORITY_M);
916 ctxt->info.fd_report_opt = cpu_to_le16(val);
917}
918
919/**
920 * ice_set_rss_vsi_ctx - Set RSS VSI context before adding a VSI
921 * @ctxt: the VSI context being set
922 * @vsi: the VSI being configured
923 */
924static void ice_set_rss_vsi_ctx(struct ice_vsi_ctx *ctxt, struct ice_vsi *vsi)
925{
926 u8 lut_type, hash_type;
927 struct device *dev;
928 struct ice_pf *pf;
929
930 pf = vsi->back;
931 dev = ice_pf_to_dev(pf);
932
933 switch (vsi->type) {
934 case ICE_VSI_PF:
935 /* PF VSI will inherit RSS instance of PF */
936 lut_type = ICE_AQ_VSI_Q_OPT_RSS_LUT_PF;
937 hash_type = ICE_AQ_VSI_Q_OPT_RSS_TPLZ;
938 break;
939 case ICE_VSI_VF:
940 /* VF VSI will gets a small RSS table which is a VSI LUT type */
941 lut_type = ICE_AQ_VSI_Q_OPT_RSS_LUT_VSI;
942 hash_type = ICE_AQ_VSI_Q_OPT_RSS_TPLZ;
943 break;
944 default:
945 dev_dbg(dev, "Unsupported VSI type %s\n",
946 ice_vsi_type_str(vsi->type));
947 return;
948 }
949
950 ctxt->info.q_opt_rss = ((lut_type << ICE_AQ_VSI_Q_OPT_RSS_LUT_S) &
951 ICE_AQ_VSI_Q_OPT_RSS_LUT_M) |
952 ((hash_type << ICE_AQ_VSI_Q_OPT_RSS_HASH_S) &
953 ICE_AQ_VSI_Q_OPT_RSS_HASH_M);
954}
955
956/**
957 * ice_vsi_init - Create and initialize a VSI
958 * @vsi: the VSI being configured
959 * @init_vsi: is this call creating a VSI
960 *
961 * This initializes a VSI context depending on the VSI type to be added and
962 * passes it down to the add_vsi aq command to create a new VSI.
963 */
964static int ice_vsi_init(struct ice_vsi *vsi, bool init_vsi)
965{
966 struct ice_pf *pf = vsi->back;
967 struct ice_hw *hw = &pf->hw;
968 struct ice_vsi_ctx *ctxt;
969 struct device *dev;
970 int ret = 0;
971
972 dev = ice_pf_to_dev(pf);
973 ctxt = kzalloc(sizeof(*ctxt), GFP_KERNEL);
974 if (!ctxt)
975 return -ENOMEM;
976
977 switch (vsi->type) {
978 case ICE_VSI_CTRL:
979 case ICE_VSI_LB:
980 case ICE_VSI_PF:
981 ctxt->flags = ICE_AQ_VSI_TYPE_PF;
982 break;
983 case ICE_VSI_VF:
984 ctxt->flags = ICE_AQ_VSI_TYPE_VF;
985 /* VF number here is the absolute VF number (0-255) */
986 ctxt->vf_num = vsi->vf_id + hw->func_caps.vf_base_id;
987 break;
988 default:
989 ret = -ENODEV;
990 goto out;
991 }
992
993 ice_set_dflt_vsi_ctx(ctxt);
994 if (test_bit(ICE_FLAG_FD_ENA, pf->flags))
995 ice_set_fd_vsi_ctx(ctxt, vsi);
996 /* if the switch is in VEB mode, allow VSI loopback */
997 if (vsi->vsw->bridge_mode == BRIDGE_MODE_VEB)
998 ctxt->info.sw_flags |= ICE_AQ_VSI_SW_FLAG_ALLOW_LB;
999
1000 /* Set LUT type and HASH type if RSS is enabled */
1001 if (test_bit(ICE_FLAG_RSS_ENA, pf->flags) &&
1002 vsi->type != ICE_VSI_CTRL) {
1003 ice_set_rss_vsi_ctx(ctxt, vsi);
1004 /* if updating VSI context, make sure to set valid_section:
1005 * to indicate which section of VSI context being updated
1006 */
1007 if (!init_vsi)
1008 ctxt->info.valid_sections |=
1009 cpu_to_le16(ICE_AQ_VSI_PROP_Q_OPT_VALID);
1010 }
1011
1012 ctxt->info.sw_id = vsi->port_info->sw_id;
1013 ice_vsi_setup_q_map(vsi, ctxt);
1014 if (!init_vsi) /* means VSI being updated */
1015 /* must to indicate which section of VSI context are
1016 * being modified
1017 */
1018 ctxt->info.valid_sections |=
1019 cpu_to_le16(ICE_AQ_VSI_PROP_RXQ_MAP_VALID);
1020
1021 /* enable/disable MAC and VLAN anti-spoof when spoofchk is on/off
1022 * respectively
1023 */
1024 if (vsi->type == ICE_VSI_VF) {
1025 ctxt->info.valid_sections |=
1026 cpu_to_le16(ICE_AQ_VSI_PROP_SECURITY_VALID);
1027 if (pf->vf[vsi->vf_id].spoofchk) {
1028 ctxt->info.sec_flags |=
1029 ICE_AQ_VSI_SEC_FLAG_ENA_MAC_ANTI_SPOOF |
1030 (ICE_AQ_VSI_SEC_TX_VLAN_PRUNE_ENA <<
1031 ICE_AQ_VSI_SEC_TX_PRUNE_ENA_S);
1032 } else {
1033 ctxt->info.sec_flags &=
1034 ~(ICE_AQ_VSI_SEC_FLAG_ENA_MAC_ANTI_SPOOF |
1035 (ICE_AQ_VSI_SEC_TX_VLAN_PRUNE_ENA <<
1036 ICE_AQ_VSI_SEC_TX_PRUNE_ENA_S));
1037 }
1038 }
1039
1040 /* Allow control frames out of main VSI */
1041 if (vsi->type == ICE_VSI_PF) {
1042 ctxt->info.sec_flags |= ICE_AQ_VSI_SEC_FLAG_ALLOW_DEST_OVRD;
1043 ctxt->info.valid_sections |=
1044 cpu_to_le16(ICE_AQ_VSI_PROP_SECURITY_VALID);
1045 }
1046
1047 if (init_vsi) {
1048 ret = ice_add_vsi(hw, vsi->idx, ctxt, NULL);
1049 if (ret) {
1050 dev_err(dev, "Add VSI failed, err %d\n", ret);
1051 ret = -EIO;
1052 goto out;
1053 }
1054 } else {
1055 ret = ice_update_vsi(hw, vsi->idx, ctxt, NULL);
1056 if (ret) {
1057 dev_err(dev, "Update VSI failed, err %d\n", ret);
1058 ret = -EIO;
1059 goto out;
1060 }
1061 }
1062
1063 /* keep context for update VSI operations */
1064 vsi->info = ctxt->info;
1065
1066 /* record VSI number returned */
1067 vsi->vsi_num = ctxt->vsi_num;
1068
1069out:
1070 kfree(ctxt);
1071 return ret;
1072}
1073
1074/**
1075 * ice_free_res - free a block of resources
1076 * @res: pointer to the resource
1077 * @index: starting index previously returned by ice_get_res
1078 * @id: identifier to track owner
1079 *
1080 * Returns number of resources freed
1081 */
1082int ice_free_res(struct ice_res_tracker *res, u16 index, u16 id)
1083{
1084 int count = 0;
1085 int i;
1086
1087 if (!res || index >= res->end)
1088 return -EINVAL;
1089
1090 id |= ICE_RES_VALID_BIT;
1091 for (i = index; i < res->end && res->list[i] == id; i++) {
1092 res->list[i] = 0;
1093 count++;
1094 }
1095
1096 return count;
1097}
1098
1099/**
1100 * ice_search_res - Search the tracker for a block of resources
1101 * @res: pointer to the resource
1102 * @needed: size of the block needed
1103 * @id: identifier to track owner
1104 *
1105 * Returns the base item index of the block, or -ENOMEM for error
1106 */
1107static int ice_search_res(struct ice_res_tracker *res, u16 needed, u16 id)
1108{
1109 u16 start = 0, end = 0;
1110
1111 if (needed > res->end)
1112 return -ENOMEM;
1113
1114 id |= ICE_RES_VALID_BIT;
1115
1116 do {
1117 /* skip already allocated entries */
1118 if (res->list[end++] & ICE_RES_VALID_BIT) {
1119 start = end;
1120 if ((start + needed) > res->end)
1121 break;
1122 }
1123
1124 if (end == (start + needed)) {
1125 int i = start;
1126
1127 /* there was enough, so assign it to the requestor */
1128 while (i != end)
1129 res->list[i++] = id;
1130
1131 return start;
1132 }
1133 } while (end < res->end);
1134
1135 return -ENOMEM;
1136}
1137
1138/**
1139 * ice_get_free_res_count - Get free count from a resource tracker
1140 * @res: Resource tracker instance
1141 */
1142static u16 ice_get_free_res_count(struct ice_res_tracker *res)
1143{
1144 u16 i, count = 0;
1145
1146 for (i = 0; i < res->end; i++)
1147 if (!(res->list[i] & ICE_RES_VALID_BIT))
1148 count++;
1149
1150 return count;
1151}
1152
1153/**
1154 * ice_get_res - get a block of resources
1155 * @pf: board private structure
1156 * @res: pointer to the resource
1157 * @needed: size of the block needed
1158 * @id: identifier to track owner
1159 *
1160 * Returns the base item index of the block, or negative for error
1161 */
1162int
1163ice_get_res(struct ice_pf *pf, struct ice_res_tracker *res, u16 needed, u16 id)
1164{
1165 if (!res || !pf)
1166 return -EINVAL;
1167
1168 if (!needed || needed > res->num_entries || id >= ICE_RES_VALID_BIT) {
1169 dev_err(ice_pf_to_dev(pf), "param err: needed=%d, num_entries = %d id=0x%04x\n",
1170 needed, res->num_entries, id);
1171 return -EINVAL;
1172 }
1173
1174 return ice_search_res(res, needed, id);
1175}
1176
1177/**
1178 * ice_vsi_setup_vector_base - Set up the base vector for the given VSI
1179 * @vsi: ptr to the VSI
1180 *
1181 * This should only be called after ice_vsi_alloc() which allocates the
1182 * corresponding SW VSI structure and initializes num_queue_pairs for the
1183 * newly allocated VSI.
1184 *
1185 * Returns 0 on success or negative on failure
1186 */
1187static int ice_vsi_setup_vector_base(struct ice_vsi *vsi)
1188{
1189 struct ice_pf *pf = vsi->back;
1190 struct device *dev;
1191 u16 num_q_vectors;
1192 int base;
1193
1194 dev = ice_pf_to_dev(pf);
1195 /* SRIOV doesn't grab irq_tracker entries for each VSI */
1196 if (vsi->type == ICE_VSI_VF)
1197 return 0;
1198
1199 if (vsi->base_vector) {
1200 dev_dbg(dev, "VSI %d has non-zero base vector %d\n",
1201 vsi->vsi_num, vsi->base_vector);
1202 return -EEXIST;
1203 }
1204
1205 num_q_vectors = vsi->num_q_vectors;
1206 /* reserve slots from OS requested IRQs */
1207 if (vsi->type == ICE_VSI_CTRL && vsi->vf_id != ICE_INVAL_VFID) {
1208 int i;
1209
1210 ice_for_each_vf(pf, i) {
1211 struct ice_vf *vf = &pf->vf[i];
1212
1213 if (i != vsi->vf_id && vf->ctrl_vsi_idx != ICE_NO_VSI) {
1214 base = pf->vsi[vf->ctrl_vsi_idx]->base_vector;
1215 break;
1216 }
1217 }
1218 if (i == pf->num_alloc_vfs)
1219 base = ice_get_res(pf, pf->irq_tracker, num_q_vectors,
1220 ICE_RES_VF_CTRL_VEC_ID);
1221 } else {
1222 base = ice_get_res(pf, pf->irq_tracker, num_q_vectors,
1223 vsi->idx);
1224 }
1225
1226 if (base < 0) {
1227 dev_err(dev, "%d MSI-X interrupts available. %s %d failed to get %d MSI-X vectors\n",
1228 ice_get_free_res_count(pf->irq_tracker),
1229 ice_vsi_type_str(vsi->type), vsi->idx, num_q_vectors);
1230 return -ENOENT;
1231 }
1232 vsi->base_vector = (u16)base;
1233 pf->num_avail_sw_msix -= num_q_vectors;
1234
1235 return 0;
1236}
1237
1238/**
1239 * ice_vsi_clear_rings - Deallocates the Tx and Rx rings for VSI
1240 * @vsi: the VSI having rings deallocated
1241 */
1242static void ice_vsi_clear_rings(struct ice_vsi *vsi)
1243{
1244 int i;
1245
1246 /* Avoid stale references by clearing map from vector to ring */
1247 if (vsi->q_vectors) {
1248 ice_for_each_q_vector(vsi, i) {
1249 struct ice_q_vector *q_vector = vsi->q_vectors[i];
1250
1251 if (q_vector) {
1252 q_vector->tx.ring = NULL;
1253 q_vector->rx.ring = NULL;
1254 }
1255 }
1256 }
1257
1258 if (vsi->tx_rings) {
1259 for (i = 0; i < vsi->alloc_txq; i++) {
1260 if (vsi->tx_rings[i]) {
1261 kfree_rcu(vsi->tx_rings[i], rcu);
1262 WRITE_ONCE(vsi->tx_rings[i], NULL);
1263 }
1264 }
1265 }
1266 if (vsi->rx_rings) {
1267 for (i = 0; i < vsi->alloc_rxq; i++) {
1268 if (vsi->rx_rings[i]) {
1269 kfree_rcu(vsi->rx_rings[i], rcu);
1270 WRITE_ONCE(vsi->rx_rings[i], NULL);
1271 }
1272 }
1273 }
1274}
1275
1276/**
1277 * ice_vsi_alloc_rings - Allocates Tx and Rx rings for the VSI
1278 * @vsi: VSI which is having rings allocated
1279 */
1280static int ice_vsi_alloc_rings(struct ice_vsi *vsi)
1281{
1282 struct ice_pf *pf = vsi->back;
1283 struct device *dev;
1284 u16 i;
1285
1286 dev = ice_pf_to_dev(pf);
1287 /* Allocate Tx rings */
1288 for (i = 0; i < vsi->alloc_txq; i++) {
1289 struct ice_ring *ring;
1290
1291 /* allocate with kzalloc(), free with kfree_rcu() */
1292 ring = kzalloc(sizeof(*ring), GFP_KERNEL);
1293
1294 if (!ring)
1295 goto err_out;
1296
1297 ring->q_index = i;
1298 ring->reg_idx = vsi->txq_map[i];
1299 ring->ring_active = false;
1300 ring->vsi = vsi;
1301 ring->tx_tstamps = &pf->ptp.port.tx;
1302 ring->dev = dev;
1303 ring->count = vsi->num_tx_desc;
1304 WRITE_ONCE(vsi->tx_rings[i], ring);
1305 }
1306
1307 /* Allocate Rx rings */
1308 for (i = 0; i < vsi->alloc_rxq; i++) {
1309 struct ice_ring *ring;
1310
1311 /* allocate with kzalloc(), free with kfree_rcu() */
1312 ring = kzalloc(sizeof(*ring), GFP_KERNEL);
1313 if (!ring)
1314 goto err_out;
1315
1316 ring->q_index = i;
1317 ring->reg_idx = vsi->rxq_map[i];
1318 ring->ring_active = false;
1319 ring->vsi = vsi;
1320 ring->netdev = vsi->netdev;
1321 ring->dev = dev;
1322 ring->count = vsi->num_rx_desc;
1323 WRITE_ONCE(vsi->rx_rings[i], ring);
1324 }
1325
1326 return 0;
1327
1328err_out:
1329 ice_vsi_clear_rings(vsi);
1330 return -ENOMEM;
1331}
1332
1333/**
1334 * ice_vsi_manage_rss_lut - disable/enable RSS
1335 * @vsi: the VSI being changed
1336 * @ena: boolean value indicating if this is an enable or disable request
1337 *
1338 * In the event of disable request for RSS, this function will zero out RSS
1339 * LUT, while in the event of enable request for RSS, it will reconfigure RSS
1340 * LUT.
1341 */
1342void ice_vsi_manage_rss_lut(struct ice_vsi *vsi, bool ena)
1343{
1344 u8 *lut;
1345
1346 lut = kzalloc(vsi->rss_table_size, GFP_KERNEL);
1347 if (!lut)
1348 return;
1349
1350 if (ena) {
1351 if (vsi->rss_lut_user)
1352 memcpy(lut, vsi->rss_lut_user, vsi->rss_table_size);
1353 else
1354 ice_fill_rss_lut(lut, vsi->rss_table_size,
1355 vsi->rss_size);
1356 }
1357
1358 ice_set_rss_lut(vsi, lut, vsi->rss_table_size);
1359 kfree(lut);
1360}
1361
1362/**
1363 * ice_vsi_cfg_rss_lut_key - Configure RSS params for a VSI
1364 * @vsi: VSI to be configured
1365 */
1366static int ice_vsi_cfg_rss_lut_key(struct ice_vsi *vsi)
1367{
1368 struct ice_pf *pf = vsi->back;
1369 struct device *dev;
1370 u8 *lut, *key;
1371 int err;
1372
1373 dev = ice_pf_to_dev(pf);
1374 vsi->rss_size = min_t(u16, vsi->rss_size, vsi->num_rxq);
1375
1376 lut = kzalloc(vsi->rss_table_size, GFP_KERNEL);
1377 if (!lut)
1378 return -ENOMEM;
1379
1380 if (vsi->rss_lut_user)
1381 memcpy(lut, vsi->rss_lut_user, vsi->rss_table_size);
1382 else
1383 ice_fill_rss_lut(lut, vsi->rss_table_size, vsi->rss_size);
1384
1385 err = ice_set_rss_lut(vsi, lut, vsi->rss_table_size);
1386 if (err) {
1387 dev_err(dev, "set_rss_lut failed, error %d\n", err);
1388 goto ice_vsi_cfg_rss_exit;
1389 }
1390
1391 key = kzalloc(ICE_GET_SET_RSS_KEY_EXTEND_KEY_SIZE, GFP_KERNEL);
1392 if (!key) {
1393 err = -ENOMEM;
1394 goto ice_vsi_cfg_rss_exit;
1395 }
1396
1397 if (vsi->rss_hkey_user)
1398 memcpy(key, vsi->rss_hkey_user, ICE_GET_SET_RSS_KEY_EXTEND_KEY_SIZE);
1399 else
1400 netdev_rss_key_fill((void *)key, ICE_GET_SET_RSS_KEY_EXTEND_KEY_SIZE);
1401
1402 err = ice_set_rss_key(vsi, key);
1403 if (err)
1404 dev_err(dev, "set_rss_key failed, error %d\n", err);
1405
1406 kfree(key);
1407ice_vsi_cfg_rss_exit:
1408 kfree(lut);
1409 return err;
1410}
1411
1412/**
1413 * ice_vsi_set_vf_rss_flow_fld - Sets VF VSI RSS input set for different flows
1414 * @vsi: VSI to be configured
1415 *
1416 * This function will only be called during the VF VSI setup. Upon successful
1417 * completion of package download, this function will configure default RSS
1418 * input sets for VF VSI.
1419 */
1420static void ice_vsi_set_vf_rss_flow_fld(struct ice_vsi *vsi)
1421{
1422 struct ice_pf *pf = vsi->back;
1423 enum ice_status status;
1424 struct device *dev;
1425
1426 dev = ice_pf_to_dev(pf);
1427 if (ice_is_safe_mode(pf)) {
1428 dev_dbg(dev, "Advanced RSS disabled. Package download failed, vsi num = %d\n",
1429 vsi->vsi_num);
1430 return;
1431 }
1432
1433 status = ice_add_avf_rss_cfg(&pf->hw, vsi->idx, ICE_DEFAULT_RSS_HENA);
1434 if (status)
1435 dev_dbg(dev, "ice_add_avf_rss_cfg failed for vsi = %d, error = %s\n",
1436 vsi->vsi_num, ice_stat_str(status));
1437}
1438
1439/**
1440 * ice_vsi_set_rss_flow_fld - Sets RSS input set for different flows
1441 * @vsi: VSI to be configured
1442 *
1443 * This function will only be called after successful download package call
1444 * during initialization of PF. Since the downloaded package will erase the
1445 * RSS section, this function will configure RSS input sets for different
1446 * flow types. The last profile added has the highest priority, therefore 2
1447 * tuple profiles (i.e. IPv4 src/dst) are added before 4 tuple profiles
1448 * (i.e. IPv4 src/dst TCP src/dst port).
1449 */
1450static void ice_vsi_set_rss_flow_fld(struct ice_vsi *vsi)
1451{
1452 u16 vsi_handle = vsi->idx, vsi_num = vsi->vsi_num;
1453 struct ice_pf *pf = vsi->back;
1454 struct ice_hw *hw = &pf->hw;
1455 enum ice_status status;
1456 struct device *dev;
1457
1458 dev = ice_pf_to_dev(pf);
1459 if (ice_is_safe_mode(pf)) {
1460 dev_dbg(dev, "Advanced RSS disabled. Package download failed, vsi num = %d\n",
1461 vsi_num);
1462 return;
1463 }
1464 /* configure RSS for IPv4 with input set IP src/dst */
1465 status = ice_add_rss_cfg(hw, vsi_handle, ICE_FLOW_HASH_IPV4,
1466 ICE_FLOW_SEG_HDR_IPV4);
1467 if (status)
1468 dev_dbg(dev, "ice_add_rss_cfg failed for ipv4 flow, vsi = %d, error = %s\n",
1469 vsi_num, ice_stat_str(status));
1470
1471 /* configure RSS for IPv6 with input set IPv6 src/dst */
1472 status = ice_add_rss_cfg(hw, vsi_handle, ICE_FLOW_HASH_IPV6,
1473 ICE_FLOW_SEG_HDR_IPV6);
1474 if (status)
1475 dev_dbg(dev, "ice_add_rss_cfg failed for ipv6 flow, vsi = %d, error = %s\n",
1476 vsi_num, ice_stat_str(status));
1477
1478 /* configure RSS for tcp4 with input set IP src/dst, TCP src/dst */
1479 status = ice_add_rss_cfg(hw, vsi_handle, ICE_HASH_TCP_IPV4,
1480 ICE_FLOW_SEG_HDR_TCP | ICE_FLOW_SEG_HDR_IPV4);
1481 if (status)
1482 dev_dbg(dev, "ice_add_rss_cfg failed for tcp4 flow, vsi = %d, error = %s\n",
1483 vsi_num, ice_stat_str(status));
1484
1485 /* configure RSS for udp4 with input set IP src/dst, UDP src/dst */
1486 status = ice_add_rss_cfg(hw, vsi_handle, ICE_HASH_UDP_IPV4,
1487 ICE_FLOW_SEG_HDR_UDP | ICE_FLOW_SEG_HDR_IPV4);
1488 if (status)
1489 dev_dbg(dev, "ice_add_rss_cfg failed for udp4 flow, vsi = %d, error = %s\n",
1490 vsi_num, ice_stat_str(status));
1491
1492 /* configure RSS for sctp4 with input set IP src/dst */
1493 status = ice_add_rss_cfg(hw, vsi_handle, ICE_FLOW_HASH_IPV4,
1494 ICE_FLOW_SEG_HDR_SCTP | ICE_FLOW_SEG_HDR_IPV4);
1495 if (status)
1496 dev_dbg(dev, "ice_add_rss_cfg failed for sctp4 flow, vsi = %d, error = %s\n",
1497 vsi_num, ice_stat_str(status));
1498
1499 /* configure RSS for tcp6 with input set IPv6 src/dst, TCP src/dst */
1500 status = ice_add_rss_cfg(hw, vsi_handle, ICE_HASH_TCP_IPV6,
1501 ICE_FLOW_SEG_HDR_TCP | ICE_FLOW_SEG_HDR_IPV6);
1502 if (status)
1503 dev_dbg(dev, "ice_add_rss_cfg failed for tcp6 flow, vsi = %d, error = %s\n",
1504 vsi_num, ice_stat_str(status));
1505
1506 /* configure RSS for udp6 with input set IPv6 src/dst, UDP src/dst */
1507 status = ice_add_rss_cfg(hw, vsi_handle, ICE_HASH_UDP_IPV6,
1508 ICE_FLOW_SEG_HDR_UDP | ICE_FLOW_SEG_HDR_IPV6);
1509 if (status)
1510 dev_dbg(dev, "ice_add_rss_cfg failed for udp6 flow, vsi = %d, error = %s\n",
1511 vsi_num, ice_stat_str(status));
1512
1513 /* configure RSS for sctp6 with input set IPv6 src/dst */
1514 status = ice_add_rss_cfg(hw, vsi_handle, ICE_FLOW_HASH_IPV6,
1515 ICE_FLOW_SEG_HDR_SCTP | ICE_FLOW_SEG_HDR_IPV6);
1516 if (status)
1517 dev_dbg(dev, "ice_add_rss_cfg failed for sctp6 flow, vsi = %d, error = %s\n",
1518 vsi_num, ice_stat_str(status));
1519}
1520
1521/**
1522 * ice_pf_state_is_nominal - checks the PF for nominal state
1523 * @pf: pointer to PF to check
1524 *
1525 * Check the PF's state for a collection of bits that would indicate
1526 * the PF is in a state that would inhibit normal operation for
1527 * driver functionality.
1528 *
1529 * Returns true if PF is in a nominal state, false otherwise
1530 */
1531bool ice_pf_state_is_nominal(struct ice_pf *pf)
1532{
1533 DECLARE_BITMAP(check_bits, ICE_STATE_NBITS) = { 0 };
1534
1535 if (!pf)
1536 return false;
1537
1538 bitmap_set(check_bits, 0, ICE_STATE_NOMINAL_CHECK_BITS);
1539 if (bitmap_intersects(pf->state, check_bits, ICE_STATE_NBITS))
1540 return false;
1541
1542 return true;
1543}
1544
1545/**
1546 * ice_update_eth_stats - Update VSI-specific ethernet statistics counters
1547 * @vsi: the VSI to be updated
1548 */
1549void ice_update_eth_stats(struct ice_vsi *vsi)
1550{
1551 struct ice_eth_stats *prev_es, *cur_es;
1552 struct ice_hw *hw = &vsi->back->hw;
1553 u16 vsi_num = vsi->vsi_num; /* HW absolute index of a VSI */
1554
1555 prev_es = &vsi->eth_stats_prev;
1556 cur_es = &vsi->eth_stats;
1557
1558 ice_stat_update40(hw, GLV_GORCL(vsi_num), vsi->stat_offsets_loaded,
1559 &prev_es->rx_bytes, &cur_es->rx_bytes);
1560
1561 ice_stat_update40(hw, GLV_UPRCL(vsi_num), vsi->stat_offsets_loaded,
1562 &prev_es->rx_unicast, &cur_es->rx_unicast);
1563
1564 ice_stat_update40(hw, GLV_MPRCL(vsi_num), vsi->stat_offsets_loaded,
1565 &prev_es->rx_multicast, &cur_es->rx_multicast);
1566
1567 ice_stat_update40(hw, GLV_BPRCL(vsi_num), vsi->stat_offsets_loaded,
1568 &prev_es->rx_broadcast, &cur_es->rx_broadcast);
1569
1570 ice_stat_update32(hw, GLV_RDPC(vsi_num), vsi->stat_offsets_loaded,
1571 &prev_es->rx_discards, &cur_es->rx_discards);
1572
1573 ice_stat_update40(hw, GLV_GOTCL(vsi_num), vsi->stat_offsets_loaded,
1574 &prev_es->tx_bytes, &cur_es->tx_bytes);
1575
1576 ice_stat_update40(hw, GLV_UPTCL(vsi_num), vsi->stat_offsets_loaded,
1577 &prev_es->tx_unicast, &cur_es->tx_unicast);
1578
1579 ice_stat_update40(hw, GLV_MPTCL(vsi_num), vsi->stat_offsets_loaded,
1580 &prev_es->tx_multicast, &cur_es->tx_multicast);
1581
1582 ice_stat_update40(hw, GLV_BPTCL(vsi_num), vsi->stat_offsets_loaded,
1583 &prev_es->tx_broadcast, &cur_es->tx_broadcast);
1584
1585 ice_stat_update32(hw, GLV_TEPC(vsi_num), vsi->stat_offsets_loaded,
1586 &prev_es->tx_errors, &cur_es->tx_errors);
1587
1588 vsi->stat_offsets_loaded = true;
1589}
1590
1591/**
1592 * ice_vsi_add_vlan - Add VSI membership for given VLAN
1593 * @vsi: the VSI being configured
1594 * @vid: VLAN ID to be added
1595 * @action: filter action to be performed on match
1596 */
1597int
1598ice_vsi_add_vlan(struct ice_vsi *vsi, u16 vid, enum ice_sw_fwd_act_type action)
1599{
1600 struct ice_pf *pf = vsi->back;
1601 struct device *dev;
1602 int err = 0;
1603
1604 dev = ice_pf_to_dev(pf);
1605
1606 if (!ice_fltr_add_vlan(vsi, vid, action)) {
1607 vsi->num_vlan++;
1608 } else {
1609 err = -ENODEV;
1610 dev_err(dev, "Failure Adding VLAN %d on VSI %i\n", vid,
1611 vsi->vsi_num);
1612 }
1613
1614 return err;
1615}
1616
1617/**
1618 * ice_vsi_kill_vlan - Remove VSI membership for a given VLAN
1619 * @vsi: the VSI being configured
1620 * @vid: VLAN ID to be removed
1621 *
1622 * Returns 0 on success and negative on failure
1623 */
1624int ice_vsi_kill_vlan(struct ice_vsi *vsi, u16 vid)
1625{
1626 struct ice_pf *pf = vsi->back;
1627 enum ice_status status;
1628 struct device *dev;
1629 int err = 0;
1630
1631 dev = ice_pf_to_dev(pf);
1632
1633 status = ice_fltr_remove_vlan(vsi, vid, ICE_FWD_TO_VSI);
1634 if (!status) {
1635 vsi->num_vlan--;
1636 } else if (status == ICE_ERR_DOES_NOT_EXIST) {
1637 dev_dbg(dev, "Failed to remove VLAN %d on VSI %i, it does not exist, status: %s\n",
1638 vid, vsi->vsi_num, ice_stat_str(status));
1639 } else {
1640 dev_err(dev, "Error removing VLAN %d on vsi %i error: %s\n",
1641 vid, vsi->vsi_num, ice_stat_str(status));
1642 err = -EIO;
1643 }
1644
1645 return err;
1646}
1647
1648/**
1649 * ice_vsi_cfg_frame_size - setup max frame size and Rx buffer length
1650 * @vsi: VSI
1651 */
1652void ice_vsi_cfg_frame_size(struct ice_vsi *vsi)
1653{
1654 if (!vsi->netdev || test_bit(ICE_FLAG_LEGACY_RX, vsi->back->flags)) {
1655 vsi->max_frame = ICE_AQ_SET_MAC_FRAME_SIZE_MAX;
1656 vsi->rx_buf_len = ICE_RXBUF_2048;
1657#if (PAGE_SIZE < 8192)
1658 } else if (!ICE_2K_TOO_SMALL_WITH_PADDING &&
1659 (vsi->netdev->mtu <= ETH_DATA_LEN)) {
1660 vsi->max_frame = ICE_RXBUF_1536 - NET_IP_ALIGN;
1661 vsi->rx_buf_len = ICE_RXBUF_1536 - NET_IP_ALIGN;
1662#endif
1663 } else {
1664 vsi->max_frame = ICE_AQ_SET_MAC_FRAME_SIZE_MAX;
1665#if (PAGE_SIZE < 8192)
1666 vsi->rx_buf_len = ICE_RXBUF_3072;
1667#else
1668 vsi->rx_buf_len = ICE_RXBUF_2048;
1669#endif
1670 }
1671}
1672
1673/**
1674 * ice_write_qrxflxp_cntxt - write/configure QRXFLXP_CNTXT register
1675 * @hw: HW pointer
1676 * @pf_q: index of the Rx queue in the PF's queue space
1677 * @rxdid: flexible descriptor RXDID
1678 * @prio: priority for the RXDID for this queue
1679 * @ena_ts: true to enable timestamp and false to disable timestamp
1680 */
1681void
1682ice_write_qrxflxp_cntxt(struct ice_hw *hw, u16 pf_q, u32 rxdid, u32 prio,
1683 bool ena_ts)
1684{
1685 int regval = rd32(hw, QRXFLXP_CNTXT(pf_q));
1686
1687 /* clear any previous values */
1688 regval &= ~(QRXFLXP_CNTXT_RXDID_IDX_M |
1689 QRXFLXP_CNTXT_RXDID_PRIO_M |
1690 QRXFLXP_CNTXT_TS_M);
1691
1692 regval |= (rxdid << QRXFLXP_CNTXT_RXDID_IDX_S) &
1693 QRXFLXP_CNTXT_RXDID_IDX_M;
1694
1695 regval |= (prio << QRXFLXP_CNTXT_RXDID_PRIO_S) &
1696 QRXFLXP_CNTXT_RXDID_PRIO_M;
1697
1698 if (ena_ts)
1699 /* Enable TimeSync on this queue */
1700 regval |= QRXFLXP_CNTXT_TS_M;
1701
1702 wr32(hw, QRXFLXP_CNTXT(pf_q), regval);
1703}
1704
1705int ice_vsi_cfg_single_rxq(struct ice_vsi *vsi, u16 q_idx)
1706{
1707 if (q_idx >= vsi->num_rxq)
1708 return -EINVAL;
1709
1710 return ice_vsi_cfg_rxq(vsi->rx_rings[q_idx]);
1711}
1712
1713int ice_vsi_cfg_single_txq(struct ice_vsi *vsi, struct ice_ring **tx_rings, u16 q_idx)
1714{
1715 struct ice_aqc_add_tx_qgrp *qg_buf;
1716 int err;
1717
1718 if (q_idx >= vsi->alloc_txq || !tx_rings || !tx_rings[q_idx])
1719 return -EINVAL;
1720
1721 qg_buf = kzalloc(struct_size(qg_buf, txqs, 1), GFP_KERNEL);
1722 if (!qg_buf)
1723 return -ENOMEM;
1724
1725 qg_buf->num_txqs = 1;
1726
1727 err = ice_vsi_cfg_txq(vsi, tx_rings[q_idx], qg_buf);
1728 kfree(qg_buf);
1729 return err;
1730}
1731
1732/**
1733 * ice_vsi_cfg_rxqs - Configure the VSI for Rx
1734 * @vsi: the VSI being configured
1735 *
1736 * Return 0 on success and a negative value on error
1737 * Configure the Rx VSI for operation.
1738 */
1739int ice_vsi_cfg_rxqs(struct ice_vsi *vsi)
1740{
1741 u16 i;
1742
1743 if (vsi->type == ICE_VSI_VF)
1744 goto setup_rings;
1745
1746 ice_vsi_cfg_frame_size(vsi);
1747setup_rings:
1748 /* set up individual rings */
1749 ice_for_each_rxq(vsi, i) {
1750 int err = ice_vsi_cfg_rxq(vsi->rx_rings[i]);
1751
1752 if (err)
1753 return err;
1754 }
1755
1756 return 0;
1757}
1758
1759/**
1760 * ice_vsi_cfg_txqs - Configure the VSI for Tx
1761 * @vsi: the VSI being configured
1762 * @rings: Tx ring array to be configured
1763 * @count: number of Tx ring array elements
1764 *
1765 * Return 0 on success and a negative value on error
1766 * Configure the Tx VSI for operation.
1767 */
1768static int
1769ice_vsi_cfg_txqs(struct ice_vsi *vsi, struct ice_ring **rings, u16 count)
1770{
1771 struct ice_aqc_add_tx_qgrp *qg_buf;
1772 u16 q_idx = 0;
1773 int err = 0;
1774
1775 qg_buf = kzalloc(struct_size(qg_buf, txqs, 1), GFP_KERNEL);
1776 if (!qg_buf)
1777 return -ENOMEM;
1778
1779 qg_buf->num_txqs = 1;
1780
1781 for (q_idx = 0; q_idx < count; q_idx++) {
1782 err = ice_vsi_cfg_txq(vsi, rings[q_idx], qg_buf);
1783 if (err)
1784 goto err_cfg_txqs;
1785 }
1786
1787err_cfg_txqs:
1788 kfree(qg_buf);
1789 return err;
1790}
1791
1792/**
1793 * ice_vsi_cfg_lan_txqs - Configure the VSI for Tx
1794 * @vsi: the VSI being configured
1795 *
1796 * Return 0 on success and a negative value on error
1797 * Configure the Tx VSI for operation.
1798 */
1799int ice_vsi_cfg_lan_txqs(struct ice_vsi *vsi)
1800{
1801 return ice_vsi_cfg_txqs(vsi, vsi->tx_rings, vsi->num_txq);
1802}
1803
1804/**
1805 * ice_vsi_cfg_xdp_txqs - Configure Tx queues dedicated for XDP in given VSI
1806 * @vsi: the VSI being configured
1807 *
1808 * Return 0 on success and a negative value on error
1809 * Configure the Tx queues dedicated for XDP in given VSI for operation.
1810 */
1811int ice_vsi_cfg_xdp_txqs(struct ice_vsi *vsi)
1812{
1813 int ret;
1814 int i;
1815
1816 ret = ice_vsi_cfg_txqs(vsi, vsi->xdp_rings, vsi->num_xdp_txq);
1817 if (ret)
1818 return ret;
1819
1820 for (i = 0; i < vsi->num_xdp_txq; i++)
1821 vsi->xdp_rings[i]->xsk_pool = ice_xsk_pool(vsi->xdp_rings[i]);
1822
1823 return ret;
1824}
1825
1826/**
1827 * ice_intrl_usec_to_reg - convert interrupt rate limit to register value
1828 * @intrl: interrupt rate limit in usecs
1829 * @gran: interrupt rate limit granularity in usecs
1830 *
1831 * This function converts a decimal interrupt rate limit in usecs to the format
1832 * expected by firmware.
1833 */
1834static u32 ice_intrl_usec_to_reg(u8 intrl, u8 gran)
1835{
1836 u32 val = intrl / gran;
1837
1838 if (val)
1839 return val | GLINT_RATE_INTRL_ENA_M;
1840 return 0;
1841}
1842
1843/**
1844 * ice_write_intrl - write throttle rate limit to interrupt specific register
1845 * @q_vector: pointer to interrupt specific structure
1846 * @intrl: throttle rate limit in microseconds to write
1847 */
1848void ice_write_intrl(struct ice_q_vector *q_vector, u8 intrl)
1849{
1850 struct ice_hw *hw = &q_vector->vsi->back->hw;
1851
1852 wr32(hw, GLINT_RATE(q_vector->reg_idx),
1853 ice_intrl_usec_to_reg(intrl, ICE_INTRL_GRAN_ABOVE_25));
1854}
1855
1856/**
1857 * __ice_write_itr - write throttle rate to register
1858 * @q_vector: pointer to interrupt data structure
1859 * @rc: pointer to ring container
1860 * @itr: throttle rate in microseconds to write
1861 */
1862static void __ice_write_itr(struct ice_q_vector *q_vector,
1863 struct ice_ring_container *rc, u16 itr)
1864{
1865 struct ice_hw *hw = &q_vector->vsi->back->hw;
1866
1867 wr32(hw, GLINT_ITR(rc->itr_idx, q_vector->reg_idx),
1868 ITR_REG_ALIGN(itr) >> ICE_ITR_GRAN_S);
1869}
1870
1871/**
1872 * ice_write_itr - write throttle rate to queue specific register
1873 * @rc: pointer to ring container
1874 * @itr: throttle rate in microseconds to write
1875 */
1876void ice_write_itr(struct ice_ring_container *rc, u16 itr)
1877{
1878 struct ice_q_vector *q_vector;
1879
1880 if (!rc->ring)
1881 return;
1882
1883 q_vector = rc->ring->q_vector;
1884
1885 __ice_write_itr(q_vector, rc, itr);
1886}
1887
1888/**
1889 * ice_vsi_cfg_msix - MSIX mode Interrupt Config in the HW
1890 * @vsi: the VSI being configured
1891 *
1892 * This configures MSIX mode interrupts for the PF VSI, and should not be used
1893 * for the VF VSI.
1894 */
1895void ice_vsi_cfg_msix(struct ice_vsi *vsi)
1896{
1897 struct ice_pf *pf = vsi->back;
1898 struct ice_hw *hw = &pf->hw;
1899 u16 txq = 0, rxq = 0;
1900 int i, q;
1901
1902 for (i = 0; i < vsi->num_q_vectors; i++) {
1903 struct ice_q_vector *q_vector = vsi->q_vectors[i];
1904 u16 reg_idx = q_vector->reg_idx;
1905
1906 ice_cfg_itr(hw, q_vector);
1907
1908 /* Both Transmit Queue Interrupt Cause Control register
1909 * and Receive Queue Interrupt Cause control register
1910 * expects MSIX_INDX field to be the vector index
1911 * within the function space and not the absolute
1912 * vector index across PF or across device.
1913 * For SR-IOV VF VSIs queue vector index always starts
1914 * with 1 since first vector index(0) is used for OICR
1915 * in VF space. Since VMDq and other PF VSIs are within
1916 * the PF function space, use the vector index that is
1917 * tracked for this PF.
1918 */
1919 for (q = 0; q < q_vector->num_ring_tx; q++) {
1920 ice_cfg_txq_interrupt(vsi, txq, reg_idx,
1921 q_vector->tx.itr_idx);
1922 txq++;
1923 }
1924
1925 for (q = 0; q < q_vector->num_ring_rx; q++) {
1926 ice_cfg_rxq_interrupt(vsi, rxq, reg_idx,
1927 q_vector->rx.itr_idx);
1928 rxq++;
1929 }
1930 }
1931}
1932
1933/**
1934 * ice_vsi_manage_vlan_insertion - Manage VLAN insertion for the VSI for Tx
1935 * @vsi: the VSI being changed
1936 */
1937int ice_vsi_manage_vlan_insertion(struct ice_vsi *vsi)
1938{
1939 struct ice_hw *hw = &vsi->back->hw;
1940 struct ice_vsi_ctx *ctxt;
1941 enum ice_status status;
1942 int ret = 0;
1943
1944 ctxt = kzalloc(sizeof(*ctxt), GFP_KERNEL);
1945 if (!ctxt)
1946 return -ENOMEM;
1947
1948 /* Here we are configuring the VSI to let the driver add VLAN tags by
1949 * setting vlan_flags to ICE_AQ_VSI_VLAN_MODE_ALL. The actual VLAN tag
1950 * insertion happens in the Tx hot path, in ice_tx_map.
1951 */
1952 ctxt->info.vlan_flags = ICE_AQ_VSI_VLAN_MODE_ALL;
1953
1954 /* Preserve existing VLAN strip setting */
1955 ctxt->info.vlan_flags |= (vsi->info.vlan_flags &
1956 ICE_AQ_VSI_VLAN_EMOD_M);
1957
1958 ctxt->info.valid_sections = cpu_to_le16(ICE_AQ_VSI_PROP_VLAN_VALID);
1959
1960 status = ice_update_vsi(hw, vsi->idx, ctxt, NULL);
1961 if (status) {
1962 dev_err(ice_pf_to_dev(vsi->back), "update VSI for VLAN insert failed, err %s aq_err %s\n",
1963 ice_stat_str(status),
1964 ice_aq_str(hw->adminq.sq_last_status));
1965 ret = -EIO;
1966 goto out;
1967 }
1968
1969 vsi->info.vlan_flags = ctxt->info.vlan_flags;
1970out:
1971 kfree(ctxt);
1972 return ret;
1973}
1974
1975/**
1976 * ice_vsi_manage_vlan_stripping - Manage VLAN stripping for the VSI for Rx
1977 * @vsi: the VSI being changed
1978 * @ena: boolean value indicating if this is a enable or disable request
1979 */
1980int ice_vsi_manage_vlan_stripping(struct ice_vsi *vsi, bool ena)
1981{
1982 struct ice_hw *hw = &vsi->back->hw;
1983 struct ice_vsi_ctx *ctxt;
1984 enum ice_status status;
1985 int ret = 0;
1986
1987 /* do not allow modifying VLAN stripping when a port VLAN is configured
1988 * on this VSI
1989 */
1990 if (vsi->info.pvid)
1991 return 0;
1992
1993 ctxt = kzalloc(sizeof(*ctxt), GFP_KERNEL);
1994 if (!ctxt)
1995 return -ENOMEM;
1996
1997 /* Here we are configuring what the VSI should do with the VLAN tag in
1998 * the Rx packet. We can either leave the tag in the packet or put it in
1999 * the Rx descriptor.
2000 */
2001 if (ena)
2002 /* Strip VLAN tag from Rx packet and put it in the desc */
2003 ctxt->info.vlan_flags = ICE_AQ_VSI_VLAN_EMOD_STR_BOTH;
2004 else
2005 /* Disable stripping. Leave tag in packet */
2006 ctxt->info.vlan_flags = ICE_AQ_VSI_VLAN_EMOD_NOTHING;
2007
2008 /* Allow all packets untagged/tagged */
2009 ctxt->info.vlan_flags |= ICE_AQ_VSI_VLAN_MODE_ALL;
2010
2011 ctxt->info.valid_sections = cpu_to_le16(ICE_AQ_VSI_PROP_VLAN_VALID);
2012
2013 status = ice_update_vsi(hw, vsi->idx, ctxt, NULL);
2014 if (status) {
2015 dev_err(ice_pf_to_dev(vsi->back), "update VSI for VLAN strip failed, ena = %d err %s aq_err %s\n",
2016 ena, ice_stat_str(status),
2017 ice_aq_str(hw->adminq.sq_last_status));
2018 ret = -EIO;
2019 goto out;
2020 }
2021
2022 vsi->info.vlan_flags = ctxt->info.vlan_flags;
2023out:
2024 kfree(ctxt);
2025 return ret;
2026}
2027
2028/**
2029 * ice_vsi_start_all_rx_rings - start/enable all of a VSI's Rx rings
2030 * @vsi: the VSI whose rings are to be enabled
2031 *
2032 * Returns 0 on success and a negative value on error
2033 */
2034int ice_vsi_start_all_rx_rings(struct ice_vsi *vsi)
2035{
2036 return ice_vsi_ctrl_all_rx_rings(vsi, true);
2037}
2038
2039/**
2040 * ice_vsi_stop_all_rx_rings - stop/disable all of a VSI's Rx rings
2041 * @vsi: the VSI whose rings are to be disabled
2042 *
2043 * Returns 0 on success and a negative value on error
2044 */
2045int ice_vsi_stop_all_rx_rings(struct ice_vsi *vsi)
2046{
2047 return ice_vsi_ctrl_all_rx_rings(vsi, false);
2048}
2049
2050/**
2051 * ice_vsi_stop_tx_rings - Disable Tx rings
2052 * @vsi: the VSI being configured
2053 * @rst_src: reset source
2054 * @rel_vmvf_num: Relative ID of VF/VM
2055 * @rings: Tx ring array to be stopped
2056 * @count: number of Tx ring array elements
2057 */
2058static int
2059ice_vsi_stop_tx_rings(struct ice_vsi *vsi, enum ice_disq_rst_src rst_src,
2060 u16 rel_vmvf_num, struct ice_ring **rings, u16 count)
2061{
2062 u16 q_idx;
2063
2064 if (vsi->num_txq > ICE_LAN_TXQ_MAX_QDIS)
2065 return -EINVAL;
2066
2067 for (q_idx = 0; q_idx < count; q_idx++) {
2068 struct ice_txq_meta txq_meta = { };
2069 int status;
2070
2071 if (!rings || !rings[q_idx])
2072 return -EINVAL;
2073
2074 ice_fill_txq_meta(vsi, rings[q_idx], &txq_meta);
2075 status = ice_vsi_stop_tx_ring(vsi, rst_src, rel_vmvf_num,
2076 rings[q_idx], &txq_meta);
2077
2078 if (status)
2079 return status;
2080 }
2081
2082 return 0;
2083}
2084
2085/**
2086 * ice_vsi_stop_lan_tx_rings - Disable LAN Tx rings
2087 * @vsi: the VSI being configured
2088 * @rst_src: reset source
2089 * @rel_vmvf_num: Relative ID of VF/VM
2090 */
2091int
2092ice_vsi_stop_lan_tx_rings(struct ice_vsi *vsi, enum ice_disq_rst_src rst_src,
2093 u16 rel_vmvf_num)
2094{
2095 return ice_vsi_stop_tx_rings(vsi, rst_src, rel_vmvf_num, vsi->tx_rings, vsi->num_txq);
2096}
2097
2098/**
2099 * ice_vsi_stop_xdp_tx_rings - Disable XDP Tx rings
2100 * @vsi: the VSI being configured
2101 */
2102int ice_vsi_stop_xdp_tx_rings(struct ice_vsi *vsi)
2103{
2104 return ice_vsi_stop_tx_rings(vsi, ICE_NO_RESET, 0, vsi->xdp_rings, vsi->num_xdp_txq);
2105}
2106
2107/**
2108 * ice_vsi_is_vlan_pruning_ena - check if VLAN pruning is enabled or not
2109 * @vsi: VSI to check whether or not VLAN pruning is enabled.
2110 *
2111 * returns true if Rx VLAN pruning is enabled and false otherwise.
2112 */
2113bool ice_vsi_is_vlan_pruning_ena(struct ice_vsi *vsi)
2114{
2115 if (!vsi)
2116 return false;
2117
2118 return (vsi->info.sw_flags2 & ICE_AQ_VSI_SW_FLAG_RX_VLAN_PRUNE_ENA);
2119}
2120
2121/**
2122 * ice_cfg_vlan_pruning - enable or disable VLAN pruning on the VSI
2123 * @vsi: VSI to enable or disable VLAN pruning on
2124 * @ena: set to true to enable VLAN pruning and false to disable it
2125 * @vlan_promisc: enable valid security flags if not in VLAN promiscuous mode
2126 *
2127 * returns 0 if VSI is updated, negative otherwise
2128 */
2129int ice_cfg_vlan_pruning(struct ice_vsi *vsi, bool ena, bool vlan_promisc)
2130{
2131 struct ice_vsi_ctx *ctxt;
2132 struct ice_pf *pf;
2133 int status;
2134
2135 if (!vsi)
2136 return -EINVAL;
2137
2138 /* Don't enable VLAN pruning if the netdev is currently in promiscuous
2139 * mode. VLAN pruning will be enabled when the interface exits
2140 * promiscuous mode if any VLAN filters are active.
2141 */
2142 if (vsi->netdev && vsi->netdev->flags & IFF_PROMISC && ena)
2143 return 0;
2144
2145 pf = vsi->back;
2146 ctxt = kzalloc(sizeof(*ctxt), GFP_KERNEL);
2147 if (!ctxt)
2148 return -ENOMEM;
2149
2150 ctxt->info = vsi->info;
2151
2152 if (ena)
2153 ctxt->info.sw_flags2 |= ICE_AQ_VSI_SW_FLAG_RX_VLAN_PRUNE_ENA;
2154 else
2155 ctxt->info.sw_flags2 &= ~ICE_AQ_VSI_SW_FLAG_RX_VLAN_PRUNE_ENA;
2156
2157 if (!vlan_promisc)
2158 ctxt->info.valid_sections =
2159 cpu_to_le16(ICE_AQ_VSI_PROP_SW_VALID);
2160
2161 status = ice_update_vsi(&pf->hw, vsi->idx, ctxt, NULL);
2162 if (status) {
2163 netdev_err(vsi->netdev, "%sabling VLAN pruning on VSI handle: %d, VSI HW ID: %d failed, err = %s, aq_err = %s\n",
2164 ena ? "En" : "Dis", vsi->idx, vsi->vsi_num,
2165 ice_stat_str(status),
2166 ice_aq_str(pf->hw.adminq.sq_last_status));
2167 goto err_out;
2168 }
2169
2170 vsi->info.sw_flags2 = ctxt->info.sw_flags2;
2171
2172 kfree(ctxt);
2173 return 0;
2174
2175err_out:
2176 kfree(ctxt);
2177 return -EIO;
2178}
2179
2180static void ice_vsi_set_tc_cfg(struct ice_vsi *vsi)
2181{
2182 struct ice_dcbx_cfg *cfg = &vsi->port_info->qos_cfg.local_dcbx_cfg;
2183
2184 vsi->tc_cfg.ena_tc = ice_dcb_get_ena_tc(cfg);
2185 vsi->tc_cfg.numtc = ice_dcb_get_num_tc(cfg);
2186}
2187
2188/**
2189 * ice_vsi_set_q_vectors_reg_idx - set the HW register index for all q_vectors
2190 * @vsi: VSI to set the q_vectors register index on
2191 */
2192static int
2193ice_vsi_set_q_vectors_reg_idx(struct ice_vsi *vsi)
2194{
2195 u16 i;
2196
2197 if (!vsi || !vsi->q_vectors)
2198 return -EINVAL;
2199
2200 ice_for_each_q_vector(vsi, i) {
2201 struct ice_q_vector *q_vector = vsi->q_vectors[i];
2202
2203 if (!q_vector) {
2204 dev_err(ice_pf_to_dev(vsi->back), "Failed to set reg_idx on q_vector %d VSI %d\n",
2205 i, vsi->vsi_num);
2206 goto clear_reg_idx;
2207 }
2208
2209 if (vsi->type == ICE_VSI_VF) {
2210 struct ice_vf *vf = &vsi->back->vf[vsi->vf_id];
2211
2212 q_vector->reg_idx = ice_calc_vf_reg_idx(vf, q_vector);
2213 } else {
2214 q_vector->reg_idx =
2215 q_vector->v_idx + vsi->base_vector;
2216 }
2217 }
2218
2219 return 0;
2220
2221clear_reg_idx:
2222 ice_for_each_q_vector(vsi, i) {
2223 struct ice_q_vector *q_vector = vsi->q_vectors[i];
2224
2225 if (q_vector)
2226 q_vector->reg_idx = 0;
2227 }
2228
2229 return -EINVAL;
2230}
2231
2232/**
2233 * ice_cfg_sw_lldp - Config switch rules for LLDP packet handling
2234 * @vsi: the VSI being configured
2235 * @tx: bool to determine Tx or Rx rule
2236 * @create: bool to determine create or remove Rule
2237 */
2238void ice_cfg_sw_lldp(struct ice_vsi *vsi, bool tx, bool create)
2239{
2240 enum ice_status (*eth_fltr)(struct ice_vsi *v, u16 type, u16 flag,
2241 enum ice_sw_fwd_act_type act);
2242 struct ice_pf *pf = vsi->back;
2243 enum ice_status status;
2244 struct device *dev;
2245
2246 dev = ice_pf_to_dev(pf);
2247 eth_fltr = create ? ice_fltr_add_eth : ice_fltr_remove_eth;
2248
2249 if (tx) {
2250 status = eth_fltr(vsi, ETH_P_LLDP, ICE_FLTR_TX,
2251 ICE_DROP_PACKET);
2252 } else {
2253 if (ice_fw_supports_lldp_fltr_ctrl(&pf->hw)) {
2254 status = ice_lldp_fltr_add_remove(&pf->hw, vsi->vsi_num,
2255 create);
2256 } else {
2257 status = eth_fltr(vsi, ETH_P_LLDP, ICE_FLTR_RX,
2258 ICE_FWD_TO_VSI);
2259 }
2260 }
2261
2262 if (status)
2263 dev_dbg(dev, "Fail %s %s LLDP rule on VSI %i error: %s\n",
2264 create ? "adding" : "removing", tx ? "TX" : "RX",
2265 vsi->vsi_num, ice_stat_str(status));
2266}
2267
2268/**
2269 * ice_set_agg_vsi - sets up scheduler aggregator node and move VSI into it
2270 * @vsi: pointer to the VSI
2271 *
2272 * This function will allocate new scheduler aggregator now if needed and will
2273 * move specified VSI into it.
2274 */
2275static void ice_set_agg_vsi(struct ice_vsi *vsi)
2276{
2277 struct device *dev = ice_pf_to_dev(vsi->back);
2278 struct ice_agg_node *agg_node_iter = NULL;
2279 u32 agg_id = ICE_INVALID_AGG_NODE_ID;
2280 struct ice_agg_node *agg_node = NULL;
2281 int node_offset, max_agg_nodes = 0;
2282 struct ice_port_info *port_info;
2283 struct ice_pf *pf = vsi->back;
2284 u32 agg_node_id_start = 0;
2285 enum ice_status status;
2286
2287 /* create (as needed) scheduler aggregator node and move VSI into
2288 * corresponding aggregator node
2289 * - PF aggregator node to contains VSIs of type _PF and _CTRL
2290 * - VF aggregator nodes will contain VF VSI
2291 */
2292 port_info = pf->hw.port_info;
2293 if (!port_info)
2294 return;
2295
2296 switch (vsi->type) {
2297 case ICE_VSI_CTRL:
2298 case ICE_VSI_LB:
2299 case ICE_VSI_PF:
2300 max_agg_nodes = ICE_MAX_PF_AGG_NODES;
2301 agg_node_id_start = ICE_PF_AGG_NODE_ID_START;
2302 agg_node_iter = &pf->pf_agg_node[0];
2303 break;
2304 case ICE_VSI_VF:
2305 /* user can create 'n' VFs on a given PF, but since max children
2306 * per aggregator node can be only 64. Following code handles
2307 * aggregator(s) for VF VSIs, either selects a agg_node which
2308 * was already created provided num_vsis < 64, otherwise
2309 * select next available node, which will be created
2310 */
2311 max_agg_nodes = ICE_MAX_VF_AGG_NODES;
2312 agg_node_id_start = ICE_VF_AGG_NODE_ID_START;
2313 agg_node_iter = &pf->vf_agg_node[0];
2314 break;
2315 default:
2316 /* other VSI type, handle later if needed */
2317 dev_dbg(dev, "unexpected VSI type %s\n",
2318 ice_vsi_type_str(vsi->type));
2319 return;
2320 }
2321
2322 /* find the appropriate aggregator node */
2323 for (node_offset = 0; node_offset < max_agg_nodes; node_offset++) {
2324 /* see if we can find space in previously created
2325 * node if num_vsis < 64, otherwise skip
2326 */
2327 if (agg_node_iter->num_vsis &&
2328 agg_node_iter->num_vsis == ICE_MAX_VSIS_IN_AGG_NODE) {
2329 agg_node_iter++;
2330 continue;
2331 }
2332
2333 if (agg_node_iter->valid &&
2334 agg_node_iter->agg_id != ICE_INVALID_AGG_NODE_ID) {
2335 agg_id = agg_node_iter->agg_id;
2336 agg_node = agg_node_iter;
2337 break;
2338 }
2339
2340 /* find unclaimed agg_id */
2341 if (agg_node_iter->agg_id == ICE_INVALID_AGG_NODE_ID) {
2342 agg_id = node_offset + agg_node_id_start;
2343 agg_node = agg_node_iter;
2344 break;
2345 }
2346 /* move to next agg_node */
2347 agg_node_iter++;
2348 }
2349
2350 if (!agg_node)
2351 return;
2352
2353 /* if selected aggregator node was not created, create it */
2354 if (!agg_node->valid) {
2355 status = ice_cfg_agg(port_info, agg_id, ICE_AGG_TYPE_AGG,
2356 (u8)vsi->tc_cfg.ena_tc);
2357 if (status) {
2358 dev_err(dev, "unable to create aggregator node with agg_id %u\n",
2359 agg_id);
2360 return;
2361 }
2362 /* aggregator node is created, store the neeeded info */
2363 agg_node->valid = true;
2364 agg_node->agg_id = agg_id;
2365 }
2366
2367 /* move VSI to corresponding aggregator node */
2368 status = ice_move_vsi_to_agg(port_info, agg_id, vsi->idx,
2369 (u8)vsi->tc_cfg.ena_tc);
2370 if (status) {
2371 dev_err(dev, "unable to move VSI idx %u into aggregator %u node",
2372 vsi->idx, agg_id);
2373 return;
2374 }
2375
2376 /* keep active children count for aggregator node */
2377 agg_node->num_vsis++;
2378
2379 /* cache the 'agg_id' in VSI, so that after reset - VSI will be moved
2380 * to aggregator node
2381 */
2382 vsi->agg_node = agg_node;
2383 dev_dbg(dev, "successfully moved VSI idx %u tc_bitmap 0x%x) into aggregator node %d which has num_vsis %u\n",
2384 vsi->idx, vsi->tc_cfg.ena_tc, vsi->agg_node->agg_id,
2385 vsi->agg_node->num_vsis);
2386}
2387
2388/**
2389 * ice_vsi_setup - Set up a VSI by a given type
2390 * @pf: board private structure
2391 * @pi: pointer to the port_info instance
2392 * @vsi_type: VSI type
2393 * @vf_id: defines VF ID to which this VSI connects. This field is meant to be
2394 * used only for ICE_VSI_VF VSI type. For other VSI types, should
2395 * fill-in ICE_INVAL_VFID as input.
2396 *
2397 * This allocates the sw VSI structure and its queue resources.
2398 *
2399 * Returns pointer to the successfully allocated and configured VSI sw struct on
2400 * success, NULL on failure.
2401 */
2402struct ice_vsi *
2403ice_vsi_setup(struct ice_pf *pf, struct ice_port_info *pi,
2404 enum ice_vsi_type vsi_type, u16 vf_id)
2405{
2406 u16 max_txqs[ICE_MAX_TRAFFIC_CLASS] = { 0 };
2407 struct device *dev = ice_pf_to_dev(pf);
2408 enum ice_status status;
2409 struct ice_vsi *vsi;
2410 int ret, i;
2411
2412 if (vsi_type == ICE_VSI_VF || vsi_type == ICE_VSI_CTRL)
2413 vsi = ice_vsi_alloc(pf, vsi_type, vf_id);
2414 else
2415 vsi = ice_vsi_alloc(pf, vsi_type, ICE_INVAL_VFID);
2416
2417 if (!vsi) {
2418 dev_err(dev, "could not allocate VSI\n");
2419 return NULL;
2420 }
2421
2422 vsi->port_info = pi;
2423 vsi->vsw = pf->first_sw;
2424 if (vsi->type == ICE_VSI_PF)
2425 vsi->ethtype = ETH_P_PAUSE;
2426
2427 if (vsi->type == ICE_VSI_VF || vsi->type == ICE_VSI_CTRL)
2428 vsi->vf_id = vf_id;
2429
2430 ice_alloc_fd_res(vsi);
2431
2432 if (ice_vsi_get_qs(vsi)) {
2433 dev_err(dev, "Failed to allocate queues. vsi->idx = %d\n",
2434 vsi->idx);
2435 goto unroll_vsi_alloc;
2436 }
2437
2438 /* set RSS capabilities */
2439 ice_vsi_set_rss_params(vsi);
2440
2441 /* set TC configuration */
2442 ice_vsi_set_tc_cfg(vsi);
2443
2444 /* create the VSI */
2445 ret = ice_vsi_init(vsi, true);
2446 if (ret)
2447 goto unroll_get_qs;
2448
2449 switch (vsi->type) {
2450 case ICE_VSI_CTRL:
2451 case ICE_VSI_PF:
2452 ret = ice_vsi_alloc_q_vectors(vsi);
2453 if (ret)
2454 goto unroll_vsi_init;
2455
2456 ret = ice_vsi_setup_vector_base(vsi);
2457 if (ret)
2458 goto unroll_alloc_q_vector;
2459
2460 ret = ice_vsi_set_q_vectors_reg_idx(vsi);
2461 if (ret)
2462 goto unroll_vector_base;
2463
2464 ret = ice_vsi_alloc_rings(vsi);
2465 if (ret)
2466 goto unroll_vector_base;
2467
2468 /* Always add VLAN ID 0 switch rule by default. This is needed
2469 * in order to allow all untagged and 0 tagged priority traffic
2470 * if Rx VLAN pruning is enabled. Also there are cases where we
2471 * don't get the call to add VLAN 0 via ice_vlan_rx_add_vid()
2472 * so this handles those cases (i.e. adding the PF to a bridge
2473 * without the 8021q module loaded).
2474 */
2475 ret = ice_vsi_add_vlan(vsi, 0, ICE_FWD_TO_VSI);
2476 if (ret)
2477 goto unroll_clear_rings;
2478
2479 ice_vsi_map_rings_to_vectors(vsi);
2480
2481 /* ICE_VSI_CTRL does not need RSS so skip RSS processing */
2482 if (vsi->type != ICE_VSI_CTRL)
2483 /* Do not exit if configuring RSS had an issue, at
2484 * least receive traffic on first queue. Hence no
2485 * need to capture return value
2486 */
2487 if (test_bit(ICE_FLAG_RSS_ENA, pf->flags)) {
2488 ice_vsi_cfg_rss_lut_key(vsi);
2489 ice_vsi_set_rss_flow_fld(vsi);
2490 }
2491 ice_init_arfs(vsi);
2492 break;
2493 case ICE_VSI_VF:
2494 /* VF driver will take care of creating netdev for this type and
2495 * map queues to vectors through Virtchnl, PF driver only
2496 * creates a VSI and corresponding structures for bookkeeping
2497 * purpose
2498 */
2499 ret = ice_vsi_alloc_q_vectors(vsi);
2500 if (ret)
2501 goto unroll_vsi_init;
2502
2503 ret = ice_vsi_alloc_rings(vsi);
2504 if (ret)
2505 goto unroll_alloc_q_vector;
2506
2507 ret = ice_vsi_set_q_vectors_reg_idx(vsi);
2508 if (ret)
2509 goto unroll_vector_base;
2510
2511 /* Do not exit if configuring RSS had an issue, at least
2512 * receive traffic on first queue. Hence no need to capture
2513 * return value
2514 */
2515 if (test_bit(ICE_FLAG_RSS_ENA, pf->flags)) {
2516 ice_vsi_cfg_rss_lut_key(vsi);
2517 ice_vsi_set_vf_rss_flow_fld(vsi);
2518 }
2519 break;
2520 case ICE_VSI_LB:
2521 ret = ice_vsi_alloc_rings(vsi);
2522 if (ret)
2523 goto unroll_vsi_init;
2524 break;
2525 default:
2526 /* clean up the resources and exit */
2527 goto unroll_vsi_init;
2528 }
2529
2530 /* configure VSI nodes based on number of queues and TC's */
2531 for (i = 0; i < vsi->tc_cfg.numtc; i++)
2532 max_txqs[i] = vsi->alloc_txq;
2533
2534 status = ice_cfg_vsi_lan(vsi->port_info, vsi->idx, vsi->tc_cfg.ena_tc,
2535 max_txqs);
2536 if (status) {
2537 dev_err(dev, "VSI %d failed lan queue config, error %s\n",
2538 vsi->vsi_num, ice_stat_str(status));
2539 goto unroll_clear_rings;
2540 }
2541
2542 /* Add switch rule to drop all Tx Flow Control Frames, of look up
2543 * type ETHERTYPE from VSIs, and restrict malicious VF from sending
2544 * out PAUSE or PFC frames. If enabled, FW can still send FC frames.
2545 * The rule is added once for PF VSI in order to create appropriate
2546 * recipe, since VSI/VSI list is ignored with drop action...
2547 * Also add rules to handle LLDP Tx packets. Tx LLDP packets need to
2548 * be dropped so that VFs cannot send LLDP packets to reconfig DCB
2549 * settings in the HW.
2550 */
2551 if (!ice_is_safe_mode(pf))
2552 if (vsi->type == ICE_VSI_PF) {
2553 ice_fltr_add_eth(vsi, ETH_P_PAUSE, ICE_FLTR_TX,
2554 ICE_DROP_PACKET);
2555 ice_cfg_sw_lldp(vsi, true, true);
2556 }
2557
2558 if (!vsi->agg_node)
2559 ice_set_agg_vsi(vsi);
2560 return vsi;
2561
2562unroll_clear_rings:
2563 ice_vsi_clear_rings(vsi);
2564unroll_vector_base:
2565 /* reclaim SW interrupts back to the common pool */
2566 ice_free_res(pf->irq_tracker, vsi->base_vector, vsi->idx);
2567 pf->num_avail_sw_msix += vsi->num_q_vectors;
2568unroll_alloc_q_vector:
2569 ice_vsi_free_q_vectors(vsi);
2570unroll_vsi_init:
2571 ice_vsi_delete(vsi);
2572unroll_get_qs:
2573 ice_vsi_put_qs(vsi);
2574unroll_vsi_alloc:
2575 if (vsi_type == ICE_VSI_VF)
2576 ice_enable_lag(pf->lag);
2577 ice_vsi_clear(vsi);
2578
2579 return NULL;
2580}
2581
2582/**
2583 * ice_vsi_release_msix - Clear the queue to Interrupt mapping in HW
2584 * @vsi: the VSI being cleaned up
2585 */
2586static void ice_vsi_release_msix(struct ice_vsi *vsi)
2587{
2588 struct ice_pf *pf = vsi->back;
2589 struct ice_hw *hw = &pf->hw;
2590 u32 txq = 0;
2591 u32 rxq = 0;
2592 int i, q;
2593
2594 for (i = 0; i < vsi->num_q_vectors; i++) {
2595 struct ice_q_vector *q_vector = vsi->q_vectors[i];
2596
2597 ice_write_intrl(q_vector, 0);
2598 for (q = 0; q < q_vector->num_ring_tx; q++) {
2599 ice_write_itr(&q_vector->tx, 0);
2600 wr32(hw, QINT_TQCTL(vsi->txq_map[txq]), 0);
2601 if (ice_is_xdp_ena_vsi(vsi)) {
2602 u32 xdp_txq = txq + vsi->num_xdp_txq;
2603
2604 wr32(hw, QINT_TQCTL(vsi->txq_map[xdp_txq]), 0);
2605 }
2606 txq++;
2607 }
2608
2609 for (q = 0; q < q_vector->num_ring_rx; q++) {
2610 ice_write_itr(&q_vector->rx, 0);
2611 wr32(hw, QINT_RQCTL(vsi->rxq_map[rxq]), 0);
2612 rxq++;
2613 }
2614 }
2615
2616 ice_flush(hw);
2617}
2618
2619/**
2620 * ice_vsi_free_irq - Free the IRQ association with the OS
2621 * @vsi: the VSI being configured
2622 */
2623void ice_vsi_free_irq(struct ice_vsi *vsi)
2624{
2625 struct ice_pf *pf = vsi->back;
2626 int base = vsi->base_vector;
2627 int i;
2628
2629 if (!vsi->q_vectors || !vsi->irqs_ready)
2630 return;
2631
2632 ice_vsi_release_msix(vsi);
2633 if (vsi->type == ICE_VSI_VF)
2634 return;
2635
2636 vsi->irqs_ready = false;
2637 ice_for_each_q_vector(vsi, i) {
2638 u16 vector = i + base;
2639 int irq_num;
2640
2641 irq_num = pf->msix_entries[vector].vector;
2642
2643 /* free only the irqs that were actually requested */
2644 if (!vsi->q_vectors[i] ||
2645 !(vsi->q_vectors[i]->num_ring_tx ||
2646 vsi->q_vectors[i]->num_ring_rx))
2647 continue;
2648
2649 /* clear the affinity notifier in the IRQ descriptor */
2650 irq_set_affinity_notifier(irq_num, NULL);
2651
2652 /* clear the affinity_mask in the IRQ descriptor */
2653 irq_set_affinity_hint(irq_num, NULL);
2654 synchronize_irq(irq_num);
2655 devm_free_irq(ice_pf_to_dev(pf), irq_num, vsi->q_vectors[i]);
2656 }
2657}
2658
2659/**
2660 * ice_vsi_free_tx_rings - Free Tx resources for VSI queues
2661 * @vsi: the VSI having resources freed
2662 */
2663void ice_vsi_free_tx_rings(struct ice_vsi *vsi)
2664{
2665 int i;
2666
2667 if (!vsi->tx_rings)
2668 return;
2669
2670 ice_for_each_txq(vsi, i)
2671 if (vsi->tx_rings[i] && vsi->tx_rings[i]->desc)
2672 ice_free_tx_ring(vsi->tx_rings[i]);
2673}
2674
2675/**
2676 * ice_vsi_free_rx_rings - Free Rx resources for VSI queues
2677 * @vsi: the VSI having resources freed
2678 */
2679void ice_vsi_free_rx_rings(struct ice_vsi *vsi)
2680{
2681 int i;
2682
2683 if (!vsi->rx_rings)
2684 return;
2685
2686 ice_for_each_rxq(vsi, i)
2687 if (vsi->rx_rings[i] && vsi->rx_rings[i]->desc)
2688 ice_free_rx_ring(vsi->rx_rings[i]);
2689}
2690
2691/**
2692 * ice_vsi_close - Shut down a VSI
2693 * @vsi: the VSI being shut down
2694 */
2695void ice_vsi_close(struct ice_vsi *vsi)
2696{
2697 if (!test_and_set_bit(ICE_VSI_DOWN, vsi->state))
2698 ice_down(vsi);
2699
2700 ice_vsi_free_irq(vsi);
2701 ice_vsi_free_tx_rings(vsi);
2702 ice_vsi_free_rx_rings(vsi);
2703}
2704
2705/**
2706 * ice_ena_vsi - resume a VSI
2707 * @vsi: the VSI being resume
2708 * @locked: is the rtnl_lock already held
2709 */
2710int ice_ena_vsi(struct ice_vsi *vsi, bool locked)
2711{
2712 int err = 0;
2713
2714 if (!test_bit(ICE_VSI_NEEDS_RESTART, vsi->state))
2715 return 0;
2716
2717 clear_bit(ICE_VSI_NEEDS_RESTART, vsi->state);
2718
2719 if (vsi->netdev && vsi->type == ICE_VSI_PF) {
2720 if (netif_running(vsi->netdev)) {
2721 if (!locked)
2722 rtnl_lock();
2723
2724 err = ice_open_internal(vsi->netdev);
2725
2726 if (!locked)
2727 rtnl_unlock();
2728 }
2729 } else if (vsi->type == ICE_VSI_CTRL) {
2730 err = ice_vsi_open_ctrl(vsi);
2731 }
2732
2733 return err;
2734}
2735
2736/**
2737 * ice_dis_vsi - pause a VSI
2738 * @vsi: the VSI being paused
2739 * @locked: is the rtnl_lock already held
2740 */
2741void ice_dis_vsi(struct ice_vsi *vsi, bool locked)
2742{
2743 if (test_bit(ICE_VSI_DOWN, vsi->state))
2744 return;
2745
2746 set_bit(ICE_VSI_NEEDS_RESTART, vsi->state);
2747
2748 if (vsi->type == ICE_VSI_PF && vsi->netdev) {
2749 if (netif_running(vsi->netdev)) {
2750 if (!locked)
2751 rtnl_lock();
2752
2753 ice_vsi_close(vsi);
2754
2755 if (!locked)
2756 rtnl_unlock();
2757 } else {
2758 ice_vsi_close(vsi);
2759 }
2760 } else if (vsi->type == ICE_VSI_CTRL) {
2761 ice_vsi_close(vsi);
2762 }
2763}
2764
2765/**
2766 * ice_vsi_dis_irq - Mask off queue interrupt generation on the VSI
2767 * @vsi: the VSI being un-configured
2768 */
2769void ice_vsi_dis_irq(struct ice_vsi *vsi)
2770{
2771 int base = vsi->base_vector;
2772 struct ice_pf *pf = vsi->back;
2773 struct ice_hw *hw = &pf->hw;
2774 u32 val;
2775 int i;
2776
2777 /* disable interrupt causation from each queue */
2778 if (vsi->tx_rings) {
2779 ice_for_each_txq(vsi, i) {
2780 if (vsi->tx_rings[i]) {
2781 u16 reg;
2782
2783 reg = vsi->tx_rings[i]->reg_idx;
2784 val = rd32(hw, QINT_TQCTL(reg));
2785 val &= ~QINT_TQCTL_CAUSE_ENA_M;
2786 wr32(hw, QINT_TQCTL(reg), val);
2787 }
2788 }
2789 }
2790
2791 if (vsi->rx_rings) {
2792 ice_for_each_rxq(vsi, i) {
2793 if (vsi->rx_rings[i]) {
2794 u16 reg;
2795
2796 reg = vsi->rx_rings[i]->reg_idx;
2797 val = rd32(hw, QINT_RQCTL(reg));
2798 val &= ~QINT_RQCTL_CAUSE_ENA_M;
2799 wr32(hw, QINT_RQCTL(reg), val);
2800 }
2801 }
2802 }
2803
2804 /* disable each interrupt */
2805 ice_for_each_q_vector(vsi, i) {
2806 if (!vsi->q_vectors[i])
2807 continue;
2808 wr32(hw, GLINT_DYN_CTL(vsi->q_vectors[i]->reg_idx), 0);
2809 }
2810
2811 ice_flush(hw);
2812
2813 /* don't call synchronize_irq() for VF's from the host */
2814 if (vsi->type == ICE_VSI_VF)
2815 return;
2816
2817 ice_for_each_q_vector(vsi, i)
2818 synchronize_irq(pf->msix_entries[i + base].vector);
2819}
2820
2821/**
2822 * ice_napi_del - Remove NAPI handler for the VSI
2823 * @vsi: VSI for which NAPI handler is to be removed
2824 */
2825void ice_napi_del(struct ice_vsi *vsi)
2826{
2827 int v_idx;
2828
2829 if (!vsi->netdev)
2830 return;
2831
2832 ice_for_each_q_vector(vsi, v_idx)
2833 netif_napi_del(&vsi->q_vectors[v_idx]->napi);
2834}
2835
2836/**
2837 * ice_vsi_release - Delete a VSI and free its resources
2838 * @vsi: the VSI being removed
2839 *
2840 * Returns 0 on success or < 0 on error
2841 */
2842int ice_vsi_release(struct ice_vsi *vsi)
2843{
2844 enum ice_status err;
2845 struct ice_pf *pf;
2846
2847 if (!vsi->back)
2848 return -ENODEV;
2849 pf = vsi->back;
2850
2851 /* do not unregister while driver is in the reset recovery pending
2852 * state. Since reset/rebuild happens through PF service task workqueue,
2853 * it's not a good idea to unregister netdev that is associated to the
2854 * PF that is running the work queue items currently. This is done to
2855 * avoid check_flush_dependency() warning on this wq
2856 */
2857 if (vsi->netdev && !ice_is_reset_in_progress(pf->state) &&
2858 (test_bit(ICE_VSI_NETDEV_REGISTERED, vsi->state))) {
2859 unregister_netdev(vsi->netdev);
2860 clear_bit(ICE_VSI_NETDEV_REGISTERED, vsi->state);
2861 }
2862
2863 ice_devlink_destroy_port(vsi);
2864
2865 if (test_bit(ICE_FLAG_RSS_ENA, pf->flags))
2866 ice_rss_clean(vsi);
2867
2868 /* Disable VSI and free resources */
2869 if (vsi->type != ICE_VSI_LB)
2870 ice_vsi_dis_irq(vsi);
2871 ice_vsi_close(vsi);
2872
2873 /* SR-IOV determines needed MSIX resources all at once instead of per
2874 * VSI since when VFs are spawned we know how many VFs there are and how
2875 * many interrupts each VF needs. SR-IOV MSIX resources are also
2876 * cleared in the same manner.
2877 */
2878 if (vsi->type == ICE_VSI_CTRL && vsi->vf_id != ICE_INVAL_VFID) {
2879 int i;
2880
2881 ice_for_each_vf(pf, i) {
2882 struct ice_vf *vf = &pf->vf[i];
2883
2884 if (i != vsi->vf_id && vf->ctrl_vsi_idx != ICE_NO_VSI)
2885 break;
2886 }
2887 if (i == pf->num_alloc_vfs) {
2888 /* No other VFs left that have control VSI, reclaim SW
2889 * interrupts back to the common pool
2890 */
2891 ice_free_res(pf->irq_tracker, vsi->base_vector,
2892 ICE_RES_VF_CTRL_VEC_ID);
2893 pf->num_avail_sw_msix += vsi->num_q_vectors;
2894 }
2895 } else if (vsi->type != ICE_VSI_VF) {
2896 /* reclaim SW interrupts back to the common pool */
2897 ice_free_res(pf->irq_tracker, vsi->base_vector, vsi->idx);
2898 pf->num_avail_sw_msix += vsi->num_q_vectors;
2899 }
2900
2901 if (!ice_is_safe_mode(pf)) {
2902 if (vsi->type == ICE_VSI_PF) {
2903 ice_fltr_remove_eth(vsi, ETH_P_PAUSE, ICE_FLTR_TX,
2904 ICE_DROP_PACKET);
2905 ice_cfg_sw_lldp(vsi, true, false);
2906 /* The Rx rule will only exist to remove if the LLDP FW
2907 * engine is currently stopped
2908 */
2909 if (!test_bit(ICE_FLAG_FW_LLDP_AGENT, pf->flags))
2910 ice_cfg_sw_lldp(vsi, false, false);
2911 }
2912 }
2913
2914 ice_fltr_remove_all(vsi);
2915 ice_rm_vsi_lan_cfg(vsi->port_info, vsi->idx);
2916 err = ice_rm_vsi_rdma_cfg(vsi->port_info, vsi->idx);
2917 if (err)
2918 dev_err(ice_pf_to_dev(vsi->back), "Failed to remove RDMA scheduler config for VSI %u, err %d\n",
2919 vsi->vsi_num, err);
2920 ice_vsi_delete(vsi);
2921 ice_vsi_free_q_vectors(vsi);
2922
2923 if (vsi->netdev) {
2924 if (test_bit(ICE_VSI_NETDEV_REGISTERED, vsi->state)) {
2925 unregister_netdev(vsi->netdev);
2926 clear_bit(ICE_VSI_NETDEV_REGISTERED, vsi->state);
2927 }
2928 if (test_bit(ICE_VSI_NETDEV_ALLOCD, vsi->state)) {
2929 free_netdev(vsi->netdev);
2930 vsi->netdev = NULL;
2931 clear_bit(ICE_VSI_NETDEV_ALLOCD, vsi->state);
2932 }
2933 }
2934
2935 if (vsi->type == ICE_VSI_VF &&
2936 vsi->agg_node && vsi->agg_node->valid)
2937 vsi->agg_node->num_vsis--;
2938 ice_vsi_clear_rings(vsi);
2939
2940 ice_vsi_put_qs(vsi);
2941
2942 /* retain SW VSI data structure since it is needed to unregister and
2943 * free VSI netdev when PF is not in reset recovery pending state,\
2944 * for ex: during rmmod.
2945 */
2946 if (!ice_is_reset_in_progress(pf->state))
2947 ice_vsi_clear(vsi);
2948
2949 return 0;
2950}
2951
2952/**
2953 * ice_vsi_rebuild_get_coalesce - get coalesce from all q_vectors
2954 * @vsi: VSI connected with q_vectors
2955 * @coalesce: array of struct with stored coalesce
2956 *
2957 * Returns array size.
2958 */
2959static int
2960ice_vsi_rebuild_get_coalesce(struct ice_vsi *vsi,
2961 struct ice_coalesce_stored *coalesce)
2962{
2963 int i;
2964
2965 ice_for_each_q_vector(vsi, i) {
2966 struct ice_q_vector *q_vector = vsi->q_vectors[i];
2967
2968 coalesce[i].itr_tx = q_vector->tx.itr_setting;
2969 coalesce[i].itr_rx = q_vector->rx.itr_setting;
2970 coalesce[i].intrl = q_vector->intrl;
2971
2972 if (i < vsi->num_txq)
2973 coalesce[i].tx_valid = true;
2974 if (i < vsi->num_rxq)
2975 coalesce[i].rx_valid = true;
2976 }
2977
2978 return vsi->num_q_vectors;
2979}
2980
2981/**
2982 * ice_vsi_rebuild_set_coalesce - set coalesce from earlier saved arrays
2983 * @vsi: VSI connected with q_vectors
2984 * @coalesce: pointer to array of struct with stored coalesce
2985 * @size: size of coalesce array
2986 *
2987 * Before this function, ice_vsi_rebuild_get_coalesce should be called to save
2988 * ITR params in arrays. If size is 0 or coalesce wasn't stored set coalesce
2989 * to default value.
2990 */
2991static void
2992ice_vsi_rebuild_set_coalesce(struct ice_vsi *vsi,
2993 struct ice_coalesce_stored *coalesce, int size)
2994{
2995 struct ice_ring_container *rc;
2996 int i;
2997
2998 if ((size && !coalesce) || !vsi)
2999 return;
3000
3001 /* There are a couple of cases that have to be handled here:
3002 * 1. The case where the number of queue vectors stays the same, but
3003 * the number of Tx or Rx rings changes (the first for loop)
3004 * 2. The case where the number of queue vectors increased (the
3005 * second for loop)
3006 */
3007 for (i = 0; i < size && i < vsi->num_q_vectors; i++) {
3008 /* There are 2 cases to handle here and they are the same for
3009 * both Tx and Rx:
3010 * if the entry was valid previously (coalesce[i].[tr]x_valid
3011 * and the loop variable is less than the number of rings
3012 * allocated, then write the previous values
3013 *
3014 * if the entry was not valid previously, but the number of
3015 * rings is less than are allocated (this means the number of
3016 * rings increased from previously), then write out the
3017 * values in the first element
3018 *
3019 * Also, always write the ITR, even if in ITR_IS_DYNAMIC
3020 * as there is no harm because the dynamic algorithm
3021 * will just overwrite.
3022 */
3023 if (i < vsi->alloc_rxq && coalesce[i].rx_valid) {
3024 rc = &vsi->q_vectors[i]->rx;
3025 rc->itr_setting = coalesce[i].itr_rx;
3026 ice_write_itr(rc, rc->itr_setting);
3027 } else if (i < vsi->alloc_rxq) {
3028 rc = &vsi->q_vectors[i]->rx;
3029 rc->itr_setting = coalesce[0].itr_rx;
3030 ice_write_itr(rc, rc->itr_setting);
3031 }
3032
3033 if (i < vsi->alloc_txq && coalesce[i].tx_valid) {
3034 rc = &vsi->q_vectors[i]->tx;
3035 rc->itr_setting = coalesce[i].itr_tx;
3036 ice_write_itr(rc, rc->itr_setting);
3037 } else if (i < vsi->alloc_txq) {
3038 rc = &vsi->q_vectors[i]->tx;
3039 rc->itr_setting = coalesce[0].itr_tx;
3040 ice_write_itr(rc, rc->itr_setting);
3041 }
3042
3043 vsi->q_vectors[i]->intrl = coalesce[i].intrl;
3044 ice_write_intrl(vsi->q_vectors[i], coalesce[i].intrl);
3045 }
3046
3047 /* the number of queue vectors increased so write whatever is in
3048 * the first element
3049 */
3050 for (; i < vsi->num_q_vectors; i++) {
3051 /* transmit */
3052 rc = &vsi->q_vectors[i]->tx;
3053 rc->itr_setting = coalesce[0].itr_tx;
3054 ice_write_itr(rc, rc->itr_setting);
3055
3056 /* receive */
3057 rc = &vsi->q_vectors[i]->rx;
3058 rc->itr_setting = coalesce[0].itr_rx;
3059 ice_write_itr(rc, rc->itr_setting);
3060
3061 vsi->q_vectors[i]->intrl = coalesce[0].intrl;
3062 ice_write_intrl(vsi->q_vectors[i], coalesce[0].intrl);
3063 }
3064}
3065
3066/**
3067 * ice_vsi_rebuild - Rebuild VSI after reset
3068 * @vsi: VSI to be rebuild
3069 * @init_vsi: is this an initialization or a reconfigure of the VSI
3070 *
3071 * Returns 0 on success and negative value on failure
3072 */
3073int ice_vsi_rebuild(struct ice_vsi *vsi, bool init_vsi)
3074{
3075 u16 max_txqs[ICE_MAX_TRAFFIC_CLASS] = { 0 };
3076 struct ice_coalesce_stored *coalesce;
3077 int prev_num_q_vectors = 0;
3078 struct ice_vf *vf = NULL;
3079 enum ice_vsi_type vtype;
3080 enum ice_status status;
3081 struct ice_pf *pf;
3082 int ret, i;
3083
3084 if (!vsi)
3085 return -EINVAL;
3086
3087 pf = vsi->back;
3088 vtype = vsi->type;
3089 if (vtype == ICE_VSI_VF)
3090 vf = &pf->vf[vsi->vf_id];
3091
3092 coalesce = kcalloc(vsi->num_q_vectors,
3093 sizeof(struct ice_coalesce_stored), GFP_KERNEL);
3094 if (!coalesce)
3095 return -ENOMEM;
3096
3097 prev_num_q_vectors = ice_vsi_rebuild_get_coalesce(vsi, coalesce);
3098
3099 ice_rm_vsi_lan_cfg(vsi->port_info, vsi->idx);
3100 ret = ice_rm_vsi_rdma_cfg(vsi->port_info, vsi->idx);
3101 if (ret)
3102 dev_err(ice_pf_to_dev(vsi->back), "Failed to remove RDMA scheduler config for VSI %u, err %d\n",
3103 vsi->vsi_num, ret);
3104 ice_vsi_free_q_vectors(vsi);
3105
3106 /* SR-IOV determines needed MSIX resources all at once instead of per
3107 * VSI since when VFs are spawned we know how many VFs there are and how
3108 * many interrupts each VF needs. SR-IOV MSIX resources are also
3109 * cleared in the same manner.
3110 */
3111 if (vtype != ICE_VSI_VF) {
3112 /* reclaim SW interrupts back to the common pool */
3113 ice_free_res(pf->irq_tracker, vsi->base_vector, vsi->idx);
3114 pf->num_avail_sw_msix += vsi->num_q_vectors;
3115 vsi->base_vector = 0;
3116 }
3117
3118 if (ice_is_xdp_ena_vsi(vsi))
3119 /* return value check can be skipped here, it always returns
3120 * 0 if reset is in progress
3121 */
3122 ice_destroy_xdp_rings(vsi);
3123 ice_vsi_put_qs(vsi);
3124 ice_vsi_clear_rings(vsi);
3125 ice_vsi_free_arrays(vsi);
3126 if (vtype == ICE_VSI_VF)
3127 ice_vsi_set_num_qs(vsi, vf->vf_id);
3128 else
3129 ice_vsi_set_num_qs(vsi, ICE_INVAL_VFID);
3130
3131 ret = ice_vsi_alloc_arrays(vsi);
3132 if (ret < 0)
3133 goto err_vsi;
3134
3135 ice_vsi_get_qs(vsi);
3136
3137 ice_alloc_fd_res(vsi);
3138 ice_vsi_set_tc_cfg(vsi);
3139
3140 /* Initialize VSI struct elements and create VSI in FW */
3141 ret = ice_vsi_init(vsi, init_vsi);
3142 if (ret < 0)
3143 goto err_vsi;
3144
3145 switch (vtype) {
3146 case ICE_VSI_CTRL:
3147 case ICE_VSI_PF:
3148 ret = ice_vsi_alloc_q_vectors(vsi);
3149 if (ret)
3150 goto err_rings;
3151
3152 ret = ice_vsi_setup_vector_base(vsi);
3153 if (ret)
3154 goto err_vectors;
3155
3156 ret = ice_vsi_set_q_vectors_reg_idx(vsi);
3157 if (ret)
3158 goto err_vectors;
3159
3160 ret = ice_vsi_alloc_rings(vsi);
3161 if (ret)
3162 goto err_vectors;
3163
3164 ice_vsi_map_rings_to_vectors(vsi);
3165 if (ice_is_xdp_ena_vsi(vsi)) {
3166 vsi->num_xdp_txq = vsi->alloc_rxq;
3167 ret = ice_prepare_xdp_rings(vsi, vsi->xdp_prog);
3168 if (ret)
3169 goto err_vectors;
3170 }
3171 /* ICE_VSI_CTRL does not need RSS so skip RSS processing */
3172 if (vtype != ICE_VSI_CTRL)
3173 /* Do not exit if configuring RSS had an issue, at
3174 * least receive traffic on first queue. Hence no
3175 * need to capture return value
3176 */
3177 if (test_bit(ICE_FLAG_RSS_ENA, pf->flags))
3178 ice_vsi_cfg_rss_lut_key(vsi);
3179 break;
3180 case ICE_VSI_VF:
3181 ret = ice_vsi_alloc_q_vectors(vsi);
3182 if (ret)
3183 goto err_rings;
3184
3185 ret = ice_vsi_set_q_vectors_reg_idx(vsi);
3186 if (ret)
3187 goto err_vectors;
3188
3189 ret = ice_vsi_alloc_rings(vsi);
3190 if (ret)
3191 goto err_vectors;
3192
3193 break;
3194 default:
3195 break;
3196 }
3197
3198 /* configure VSI nodes based on number of queues and TC's */
3199 for (i = 0; i < vsi->tc_cfg.numtc; i++) {
3200 max_txqs[i] = vsi->alloc_txq;
3201
3202 if (ice_is_xdp_ena_vsi(vsi))
3203 max_txqs[i] += vsi->num_xdp_txq;
3204 }
3205
3206 status = ice_cfg_vsi_lan(vsi->port_info, vsi->idx, vsi->tc_cfg.ena_tc,
3207 max_txqs);
3208 if (status) {
3209 dev_err(ice_pf_to_dev(pf), "VSI %d failed lan queue config, error %s\n",
3210 vsi->vsi_num, ice_stat_str(status));
3211 if (init_vsi) {
3212 ret = -EIO;
3213 goto err_vectors;
3214 } else {
3215 return ice_schedule_reset(pf, ICE_RESET_PFR);
3216 }
3217 }
3218 ice_vsi_rebuild_set_coalesce(vsi, coalesce, prev_num_q_vectors);
3219 kfree(coalesce);
3220
3221 return 0;
3222
3223err_vectors:
3224 ice_vsi_free_q_vectors(vsi);
3225err_rings:
3226 if (vsi->netdev) {
3227 vsi->current_netdev_flags = 0;
3228 unregister_netdev(vsi->netdev);
3229 free_netdev(vsi->netdev);
3230 vsi->netdev = NULL;
3231 }
3232err_vsi:
3233 ice_vsi_clear(vsi);
3234 set_bit(ICE_RESET_FAILED, pf->state);
3235 kfree(coalesce);
3236 return ret;
3237}
3238
3239/**
3240 * ice_is_reset_in_progress - check for a reset in progress
3241 * @state: PF state field
3242 */
3243bool ice_is_reset_in_progress(unsigned long *state)
3244{
3245 return test_bit(ICE_RESET_OICR_RECV, state) ||
3246 test_bit(ICE_PFR_REQ, state) ||
3247 test_bit(ICE_CORER_REQ, state) ||
3248 test_bit(ICE_GLOBR_REQ, state);
3249}
3250
3251/**
3252 * ice_wait_for_reset - Wait for driver to finish reset and rebuild
3253 * @pf: pointer to the PF structure
3254 * @timeout: length of time to wait, in jiffies
3255 *
3256 * Wait (sleep) for a short time until the driver finishes cleaning up from
3257 * a device reset. The caller must be able to sleep. Use this to delay
3258 * operations that could fail while the driver is cleaning up after a device
3259 * reset.
3260 *
3261 * Returns 0 on success, -EBUSY if the reset is not finished within the
3262 * timeout, and -ERESTARTSYS if the thread was interrupted.
3263 */
3264int ice_wait_for_reset(struct ice_pf *pf, unsigned long timeout)
3265{
3266 long ret;
3267
3268 ret = wait_event_interruptible_timeout(pf->reset_wait_queue,
3269 !ice_is_reset_in_progress(pf->state),
3270 timeout);
3271 if (ret < 0)
3272 return ret;
3273 else if (!ret)
3274 return -EBUSY;
3275 else
3276 return 0;
3277}
3278
3279#ifdef CONFIG_DCB
3280/**
3281 * ice_vsi_update_q_map - update our copy of the VSI info with new queue map
3282 * @vsi: VSI being configured
3283 * @ctx: the context buffer returned from AQ VSI update command
3284 */
3285static void ice_vsi_update_q_map(struct ice_vsi *vsi, struct ice_vsi_ctx *ctx)
3286{
3287 vsi->info.mapping_flags = ctx->info.mapping_flags;
3288 memcpy(&vsi->info.q_mapping, &ctx->info.q_mapping,
3289 sizeof(vsi->info.q_mapping));
3290 memcpy(&vsi->info.tc_mapping, ctx->info.tc_mapping,
3291 sizeof(vsi->info.tc_mapping));
3292}
3293
3294/**
3295 * ice_vsi_cfg_tc - Configure VSI Tx Sched for given TC map
3296 * @vsi: VSI to be configured
3297 * @ena_tc: TC bitmap
3298 *
3299 * VSI queues expected to be quiesced before calling this function
3300 */
3301int ice_vsi_cfg_tc(struct ice_vsi *vsi, u8 ena_tc)
3302{
3303 u16 max_txqs[ICE_MAX_TRAFFIC_CLASS] = { 0 };
3304 struct ice_pf *pf = vsi->back;
3305 struct ice_vsi_ctx *ctx;
3306 enum ice_status status;
3307 struct device *dev;
3308 int i, ret = 0;
3309 u8 num_tc = 0;
3310
3311 dev = ice_pf_to_dev(pf);
3312
3313 ice_for_each_traffic_class(i) {
3314 /* build bitmap of enabled TCs */
3315 if (ena_tc & BIT(i))
3316 num_tc++;
3317 /* populate max_txqs per TC */
3318 max_txqs[i] = vsi->alloc_txq;
3319 }
3320
3321 vsi->tc_cfg.ena_tc = ena_tc;
3322 vsi->tc_cfg.numtc = num_tc;
3323
3324 ctx = kzalloc(sizeof(*ctx), GFP_KERNEL);
3325 if (!ctx)
3326 return -ENOMEM;
3327
3328 ctx->vf_num = 0;
3329 ctx->info = vsi->info;
3330
3331 ice_vsi_setup_q_map(vsi, ctx);
3332
3333 /* must to indicate which section of VSI context are being modified */
3334 ctx->info.valid_sections = cpu_to_le16(ICE_AQ_VSI_PROP_RXQ_MAP_VALID);
3335 status = ice_update_vsi(&pf->hw, vsi->idx, ctx, NULL);
3336 if (status) {
3337 dev_info(dev, "Failed VSI Update\n");
3338 ret = -EIO;
3339 goto out;
3340 }
3341
3342 status = ice_cfg_vsi_lan(vsi->port_info, vsi->idx, vsi->tc_cfg.ena_tc,
3343 max_txqs);
3344
3345 if (status) {
3346 dev_err(dev, "VSI %d failed TC config, error %s\n",
3347 vsi->vsi_num, ice_stat_str(status));
3348 ret = -EIO;
3349 goto out;
3350 }
3351 ice_vsi_update_q_map(vsi, ctx);
3352 vsi->info.valid_sections = 0;
3353
3354 ice_vsi_cfg_netdev_tc(vsi, ena_tc);
3355out:
3356 kfree(ctx);
3357 return ret;
3358}
3359#endif /* CONFIG_DCB */
3360
3361/**
3362 * ice_update_ring_stats - Update ring statistics
3363 * @ring: ring to update
3364 * @pkts: number of processed packets
3365 * @bytes: number of processed bytes
3366 *
3367 * This function assumes that caller has acquired a u64_stats_sync lock.
3368 */
3369static void ice_update_ring_stats(struct ice_ring *ring, u64 pkts, u64 bytes)
3370{
3371 ring->stats.bytes += bytes;
3372 ring->stats.pkts += pkts;
3373}
3374
3375/**
3376 * ice_update_tx_ring_stats - Update Tx ring specific counters
3377 * @tx_ring: ring to update
3378 * @pkts: number of processed packets
3379 * @bytes: number of processed bytes
3380 */
3381void ice_update_tx_ring_stats(struct ice_ring *tx_ring, u64 pkts, u64 bytes)
3382{
3383 u64_stats_update_begin(&tx_ring->syncp);
3384 ice_update_ring_stats(tx_ring, pkts, bytes);
3385 u64_stats_update_end(&tx_ring->syncp);
3386}
3387
3388/**
3389 * ice_update_rx_ring_stats - Update Rx ring specific counters
3390 * @rx_ring: ring to update
3391 * @pkts: number of processed packets
3392 * @bytes: number of processed bytes
3393 */
3394void ice_update_rx_ring_stats(struct ice_ring *rx_ring, u64 pkts, u64 bytes)
3395{
3396 u64_stats_update_begin(&rx_ring->syncp);
3397 ice_update_ring_stats(rx_ring, pkts, bytes);
3398 u64_stats_update_end(&rx_ring->syncp);
3399}
3400
3401/**
3402 * ice_status_to_errno - convert from enum ice_status to Linux errno
3403 * @err: ice_status value to convert
3404 */
3405int ice_status_to_errno(enum ice_status err)
3406{
3407 switch (err) {
3408 case ICE_SUCCESS:
3409 return 0;
3410 case ICE_ERR_DOES_NOT_EXIST:
3411 return -ENOENT;
3412 case ICE_ERR_OUT_OF_RANGE:
3413 case ICE_ERR_AQ_ERROR:
3414 case ICE_ERR_AQ_TIMEOUT:
3415 case ICE_ERR_AQ_EMPTY:
3416 case ICE_ERR_AQ_FW_CRITICAL:
3417 return -EIO;
3418 case ICE_ERR_PARAM:
3419 case ICE_ERR_INVAL_SIZE:
3420 return -EINVAL;
3421 case ICE_ERR_NO_MEMORY:
3422 return -ENOMEM;
3423 case ICE_ERR_MAX_LIMIT:
3424 return -EAGAIN;
3425 case ICE_ERR_RESET_ONGOING:
3426 return -EBUSY;
3427 case ICE_ERR_AQ_FULL:
3428 return -ENOSPC;
3429 default:
3430 return -EINVAL;
3431 }
3432}
3433
3434/**
3435 * ice_is_dflt_vsi_in_use - check if the default forwarding VSI is being used
3436 * @sw: switch to check if its default forwarding VSI is free
3437 *
3438 * Return true if the default forwarding VSI is already being used, else returns
3439 * false signalling that it's available to use.
3440 */
3441bool ice_is_dflt_vsi_in_use(struct ice_sw *sw)
3442{
3443 return (sw->dflt_vsi && sw->dflt_vsi_ena);
3444}
3445
3446/**
3447 * ice_is_vsi_dflt_vsi - check if the VSI passed in is the default VSI
3448 * @sw: switch for the default forwarding VSI to compare against
3449 * @vsi: VSI to compare against default forwarding VSI
3450 *
3451 * If this VSI passed in is the default forwarding VSI then return true, else
3452 * return false
3453 */
3454bool ice_is_vsi_dflt_vsi(struct ice_sw *sw, struct ice_vsi *vsi)
3455{
3456 return (sw->dflt_vsi == vsi && sw->dflt_vsi_ena);
3457}
3458
3459/**
3460 * ice_set_dflt_vsi - set the default forwarding VSI
3461 * @sw: switch used to assign the default forwarding VSI
3462 * @vsi: VSI getting set as the default forwarding VSI on the switch
3463 *
3464 * If the VSI passed in is already the default VSI and it's enabled just return
3465 * success.
3466 *
3467 * If there is already a default VSI on the switch and it's enabled then return
3468 * -EEXIST since there can only be one default VSI per switch.
3469 *
3470 * Otherwise try to set the VSI passed in as the switch's default VSI and
3471 * return the result.
3472 */
3473int ice_set_dflt_vsi(struct ice_sw *sw, struct ice_vsi *vsi)
3474{
3475 enum ice_status status;
3476 struct device *dev;
3477
3478 if (!sw || !vsi)
3479 return -EINVAL;
3480
3481 dev = ice_pf_to_dev(vsi->back);
3482
3483 /* the VSI passed in is already the default VSI */
3484 if (ice_is_vsi_dflt_vsi(sw, vsi)) {
3485 dev_dbg(dev, "VSI %d passed in is already the default forwarding VSI, nothing to do\n",
3486 vsi->vsi_num);
3487 return 0;
3488 }
3489
3490 /* another VSI is already the default VSI for this switch */
3491 if (ice_is_dflt_vsi_in_use(sw)) {
3492 dev_err(dev, "Default forwarding VSI %d already in use, disable it and try again\n",
3493 sw->dflt_vsi->vsi_num);
3494 return -EEXIST;
3495 }
3496
3497 status = ice_cfg_dflt_vsi(&vsi->back->hw, vsi->idx, true, ICE_FLTR_RX);
3498 if (status) {
3499 dev_err(dev, "Failed to set VSI %d as the default forwarding VSI, error %s\n",
3500 vsi->vsi_num, ice_stat_str(status));
3501 return -EIO;
3502 }
3503
3504 sw->dflt_vsi = vsi;
3505 sw->dflt_vsi_ena = true;
3506
3507 return 0;
3508}
3509
3510/**
3511 * ice_clear_dflt_vsi - clear the default forwarding VSI
3512 * @sw: switch used to clear the default VSI
3513 *
3514 * If the switch has no default VSI or it's not enabled then return error.
3515 *
3516 * Otherwise try to clear the default VSI and return the result.
3517 */
3518int ice_clear_dflt_vsi(struct ice_sw *sw)
3519{
3520 struct ice_vsi *dflt_vsi;
3521 enum ice_status status;
3522 struct device *dev;
3523
3524 if (!sw)
3525 return -EINVAL;
3526
3527 dev = ice_pf_to_dev(sw->pf);
3528
3529 dflt_vsi = sw->dflt_vsi;
3530
3531 /* there is no default VSI configured */
3532 if (!ice_is_dflt_vsi_in_use(sw))
3533 return -ENODEV;
3534
3535 status = ice_cfg_dflt_vsi(&dflt_vsi->back->hw, dflt_vsi->idx, false,
3536 ICE_FLTR_RX);
3537 if (status) {
3538 dev_err(dev, "Failed to clear the default forwarding VSI %d, error %s\n",
3539 dflt_vsi->vsi_num, ice_stat_str(status));
3540 return -EIO;
3541 }
3542
3543 sw->dflt_vsi = NULL;
3544 sw->dflt_vsi_ena = false;
3545
3546 return 0;
3547}
3548
3549/**
3550 * ice_set_link - turn on/off physical link
3551 * @vsi: VSI to modify physical link on
3552 * @ena: turn on/off physical link
3553 */
3554int ice_set_link(struct ice_vsi *vsi, bool ena)
3555{
3556 struct device *dev = ice_pf_to_dev(vsi->back);
3557 struct ice_port_info *pi = vsi->port_info;
3558 struct ice_hw *hw = pi->hw;
3559 enum ice_status status;
3560
3561 if (vsi->type != ICE_VSI_PF)
3562 return -EINVAL;
3563
3564 status = ice_aq_set_link_restart_an(pi, ena, NULL);
3565
3566 /* if link is owned by manageability, FW will return ICE_AQ_RC_EMODE.
3567 * this is not a fatal error, so print a warning message and return
3568 * a success code. Return an error if FW returns an error code other
3569 * than ICE_AQ_RC_EMODE
3570 */
3571 if (status == ICE_ERR_AQ_ERROR) {
3572 if (hw->adminq.sq_last_status == ICE_AQ_RC_EMODE)
3573 dev_warn(dev, "can't set link to %s, err %s aq_err %s. not fatal, continuing\n",
3574 (ena ? "ON" : "OFF"), ice_stat_str(status),
3575 ice_aq_str(hw->adminq.sq_last_status));
3576 } else if (status) {
3577 dev_err(dev, "can't set link to %s, err %s aq_err %s\n",
3578 (ena ? "ON" : "OFF"), ice_stat_str(status),
3579 ice_aq_str(hw->adminq.sq_last_status));
3580 return -EIO;
3581 }
3582
3583 return 0;
3584}