1 // SPDX-License-Identifier: GPL-2.0
2 /* Copyright (c) 2018, Intel Corporation. */
3 
4 #include "ice.h"
5 #include "ice_vf_lib_private.h"
6 #include "ice_base.h"
7 #include "ice_lib.h"
8 #include "ice_fltr.h"
9 #include "ice_dcb_lib.h"
10 #include "ice_flow.h"
11 #include "ice_eswitch.h"
12 #include "ice_virtchnl_allowlist.h"
13 #include "ice_flex_pipe.h"
14 #include "ice_vf_vsi_vlan_ops.h"
15 #include "ice_vlan.h"
16 
17 /**
18  * ice_free_vf_entries - Free all VF entries from the hash table
19  * @pf: pointer to the PF structure
20  *
21  * Iterate over the VF hash table, removing and releasing all VF entries.
22  * Called during VF teardown or as cleanup during failed VF initialization.
23  */
24 static void ice_free_vf_entries(struct ice_pf *pf)
25 {
26 	struct ice_vfs *vfs = &pf->vfs;
27 	struct hlist_node *tmp;
28 	struct ice_vf *vf;
29 	unsigned int bkt;
30 
31 	/* Remove all VFs from the hash table and release their main
32 	 * reference. Once all references to the VF are dropped, ice_put_vf()
33 	 * will call ice_release_vf which will remove the VF memory.
34 	 */
35 	lockdep_assert_held(&vfs->table_lock);
36 
37 	hash_for_each_safe(vfs->table, bkt, tmp, vf, entry) {
38 		hash_del_rcu(&vf->entry);
39 		ice_put_vf(vf);
40 	}
41 }
42 
43 /**
44  * ice_free_vf_res - Free a VF's resources
45  * @vf: pointer to the VF info
46  */
47 static void ice_free_vf_res(struct ice_vf *vf)
48 {
49 	struct ice_pf *pf = vf->pf;
50 	int i, last_vector_idx;
51 
52 	/* First, disable VF's configuration API to prevent OS from
53 	 * accessing the VF's VSI after it's freed or invalidated.
54 	 */
55 	clear_bit(ICE_VF_STATE_INIT, vf->vf_states);
56 	ice_vf_fdir_exit(vf);
57 	/* free VF control VSI */
58 	if (vf->ctrl_vsi_idx != ICE_NO_VSI)
59 		ice_vf_ctrl_vsi_release(vf);
60 
61 	/* free VSI and disconnect it from the parent uplink */
62 	if (vf->lan_vsi_idx != ICE_NO_VSI) {
63 		ice_vf_vsi_release(vf);
64 		vf->num_mac = 0;
65 	}
66 
67 	last_vector_idx = vf->first_vector_idx + pf->vfs.num_msix_per - 1;
68 
69 	/* clear VF MDD event information */
70 	memset(&vf->mdd_tx_events, 0, sizeof(vf->mdd_tx_events));
71 	memset(&vf->mdd_rx_events, 0, sizeof(vf->mdd_rx_events));
72 
73 	/* Disable interrupts so that VF starts in a known state */
74 	for (i = vf->first_vector_idx; i <= last_vector_idx; i++) {
75 		wr32(&pf->hw, GLINT_DYN_CTL(i), GLINT_DYN_CTL_CLEARPBA_M);
76 		ice_flush(&pf->hw);
77 	}
78 	/* reset some of the state variables keeping track of the resources */
79 	clear_bit(ICE_VF_STATE_MC_PROMISC, vf->vf_states);
80 	clear_bit(ICE_VF_STATE_UC_PROMISC, vf->vf_states);
81 }
82 
83 /**
84  * ice_dis_vf_mappings
85  * @vf: pointer to the VF structure
86  */
87 static void ice_dis_vf_mappings(struct ice_vf *vf)
88 {
89 	struct ice_pf *pf = vf->pf;
90 	struct ice_vsi *vsi;
91 	struct device *dev;
92 	int first, last, v;
93 	struct ice_hw *hw;
94 
95 	hw = &pf->hw;
96 	vsi = ice_get_vf_vsi(vf);
97 	if (WARN_ON(!vsi))
98 		return;
99 
100 	dev = ice_pf_to_dev(pf);
101 	wr32(hw, VPINT_ALLOC(vf->vf_id), 0);
102 	wr32(hw, VPINT_ALLOC_PCI(vf->vf_id), 0);
103 
104 	first = vf->first_vector_idx;
105 	last = first + pf->vfs.num_msix_per - 1;
106 	for (v = first; v <= last; v++) {
107 		u32 reg;
108 
109 		reg = (((1 << GLINT_VECT2FUNC_IS_PF_S) &
110 			GLINT_VECT2FUNC_IS_PF_M) |
111 		       ((hw->pf_id << GLINT_VECT2FUNC_PF_NUM_S) &
112 			GLINT_VECT2FUNC_PF_NUM_M));
113 		wr32(hw, GLINT_VECT2FUNC(v), reg);
114 	}
115 
116 	if (vsi->tx_mapping_mode == ICE_VSI_MAP_CONTIG)
117 		wr32(hw, VPLAN_TX_QBASE(vf->vf_id), 0);
118 	else
119 		dev_err(dev, "Scattered mode for VF Tx queues is not yet implemented\n");
120 
121 	if (vsi->rx_mapping_mode == ICE_VSI_MAP_CONTIG)
122 		wr32(hw, VPLAN_RX_QBASE(vf->vf_id), 0);
123 	else
124 		dev_err(dev, "Scattered mode for VF Rx queues is not yet implemented\n");
125 }
126 
127 /**
128  * ice_sriov_free_msix_res - Reset/free any used MSIX resources
129  * @pf: pointer to the PF structure
130  *
131  * Since no MSIX entries are taken from the pf->irq_tracker then just clear
132  * the pf->sriov_base_vector.
133  *
134  * Returns 0 on success, and -EINVAL on error.
135  */
136 static int ice_sriov_free_msix_res(struct ice_pf *pf)
137 {
138 	if (!pf)
139 		return -EINVAL;
140 
141 	pf->sriov_base_vector = 0;
142 
143 	return 0;
144 }
145 
146 /**
147  * ice_free_vfs - Free all VFs
148  * @pf: pointer to the PF structure
149  */
150 void ice_free_vfs(struct ice_pf *pf)
151 {
152 	struct device *dev = ice_pf_to_dev(pf);
153 	struct ice_vfs *vfs = &pf->vfs;
154 	struct ice_hw *hw = &pf->hw;
155 	struct ice_vf *vf;
156 	unsigned int bkt;
157 
158 	if (!ice_has_vfs(pf))
159 		return;
160 
161 	while (test_and_set_bit(ICE_VF_DIS, pf->state))
162 		usleep_range(1000, 2000);
163 
164 	/* Disable IOV before freeing resources. This lets any VF drivers
165 	 * running in the host get themselves cleaned up before we yank
166 	 * the carpet out from underneath their feet.
167 	 */
168 	if (!pci_vfs_assigned(pf->pdev))
169 		pci_disable_sriov(pf->pdev);
170 	else
171 		dev_warn(dev, "VFs are assigned - not disabling SR-IOV\n");
172 
173 	mutex_lock(&vfs->table_lock);
174 
175 	ice_eswitch_release(pf);
176 
177 	ice_for_each_vf(pf, bkt, vf) {
178 		mutex_lock(&vf->cfg_lock);
179 
180 		ice_dis_vf_qs(vf);
181 
182 		if (test_bit(ICE_VF_STATE_INIT, vf->vf_states)) {
183 			/* disable VF qp mappings and set VF disable state */
184 			ice_dis_vf_mappings(vf);
185 			set_bit(ICE_VF_STATE_DIS, vf->vf_states);
186 			ice_free_vf_res(vf);
187 		}
188 
189 		if (!pci_vfs_assigned(pf->pdev)) {
190 			u32 reg_idx, bit_idx;
191 
192 			reg_idx = (hw->func_caps.vf_base_id + vf->vf_id) / 32;
193 			bit_idx = (hw->func_caps.vf_base_id + vf->vf_id) % 32;
194 			wr32(hw, GLGEN_VFLRSTAT(reg_idx), BIT(bit_idx));
195 		}
196 
197 		/* clear malicious info since the VF is getting released */
198 		list_del(&vf->mbx_info.list_entry);
199 
200 		mutex_unlock(&vf->cfg_lock);
201 	}
202 
203 	if (ice_sriov_free_msix_res(pf))
204 		dev_err(dev, "Failed to free MSIX resources used by SR-IOV\n");
205 
206 	vfs->num_qps_per = 0;
207 	ice_free_vf_entries(pf);
208 
209 	mutex_unlock(&vfs->table_lock);
210 
211 	clear_bit(ICE_VF_DIS, pf->state);
212 	clear_bit(ICE_FLAG_SRIOV_ENA, pf->flags);
213 }
214 
215 /**
216  * ice_vf_vsi_setup - Set up a VF VSI
217  * @vf: VF to setup VSI for
218  *
219  * Returns pointer to the successfully allocated VSI struct on success,
220  * otherwise returns NULL on failure.
221  */
222 static struct ice_vsi *ice_vf_vsi_setup(struct ice_vf *vf)
223 {
224 	struct ice_vsi_cfg_params params = {};
225 	struct ice_pf *pf = vf->pf;
226 	struct ice_vsi *vsi;
227 
228 	params.type = ICE_VSI_VF;
229 	params.pi = ice_vf_get_port_info(vf);
230 	params.vf = vf;
231 	params.flags = ICE_VSI_FLAG_INIT;
232 
233 	vsi = ice_vsi_setup(pf, &params);
234 
235 	if (!vsi) {
236 		dev_err(ice_pf_to_dev(pf), "Failed to create VF VSI\n");
237 		ice_vf_invalidate_vsi(vf);
238 		return NULL;
239 	}
240 
241 	vf->lan_vsi_idx = vsi->idx;
242 	vf->lan_vsi_num = vsi->vsi_num;
243 
244 	return vsi;
245 }
246 
247 /**
248  * ice_calc_vf_first_vector_idx - Calculate MSIX vector index in the PF space
249  * @pf: pointer to PF structure
250  * @vf: pointer to VF that the first MSIX vector index is being calculated for
251  *
252  * This returns the first MSIX vector index in PF space that is used by this VF.
253  * This index is used when accessing PF relative registers such as
254  * GLINT_VECT2FUNC and GLINT_DYN_CTL.
255  * This will always be the OICR index in the AVF driver so any functionality
256  * using vf->first_vector_idx for queue configuration will have to increment by
257  * 1 to avoid meddling with the OICR index.
258  */
259 static int ice_calc_vf_first_vector_idx(struct ice_pf *pf, struct ice_vf *vf)
260 {
261 	return pf->sriov_base_vector + vf->vf_id * pf->vfs.num_msix_per;
262 }
263 
264 /**
265  * ice_ena_vf_msix_mappings - enable VF MSIX mappings in hardware
266  * @vf: VF to enable MSIX mappings for
267  *
268  * Some of the registers need to be indexed/configured using hardware global
269  * device values and other registers need 0-based values, which represent PF
270  * based values.
271  */
272 static void ice_ena_vf_msix_mappings(struct ice_vf *vf)
273 {
274 	int device_based_first_msix, device_based_last_msix;
275 	int pf_based_first_msix, pf_based_last_msix, v;
276 	struct ice_pf *pf = vf->pf;
277 	int device_based_vf_id;
278 	struct ice_hw *hw;
279 	u32 reg;
280 
281 	hw = &pf->hw;
282 	pf_based_first_msix = vf->first_vector_idx;
283 	pf_based_last_msix = (pf_based_first_msix + pf->vfs.num_msix_per) - 1;
284 
285 	device_based_first_msix = pf_based_first_msix +
286 		pf->hw.func_caps.common_cap.msix_vector_first_id;
287 	device_based_last_msix =
288 		(device_based_first_msix + pf->vfs.num_msix_per) - 1;
289 	device_based_vf_id = vf->vf_id + hw->func_caps.vf_base_id;
290 
291 	reg = (((device_based_first_msix << VPINT_ALLOC_FIRST_S) &
292 		VPINT_ALLOC_FIRST_M) |
293 	       ((device_based_last_msix << VPINT_ALLOC_LAST_S) &
294 		VPINT_ALLOC_LAST_M) | VPINT_ALLOC_VALID_M);
295 	wr32(hw, VPINT_ALLOC(vf->vf_id), reg);
296 
297 	reg = (((device_based_first_msix << VPINT_ALLOC_PCI_FIRST_S)
298 		 & VPINT_ALLOC_PCI_FIRST_M) |
299 	       ((device_based_last_msix << VPINT_ALLOC_PCI_LAST_S) &
300 		VPINT_ALLOC_PCI_LAST_M) | VPINT_ALLOC_PCI_VALID_M);
301 	wr32(hw, VPINT_ALLOC_PCI(vf->vf_id), reg);
302 
303 	/* map the interrupts to its functions */
304 	for (v = pf_based_first_msix; v <= pf_based_last_msix; v++) {
305 		reg = (((device_based_vf_id << GLINT_VECT2FUNC_VF_NUM_S) &
306 			GLINT_VECT2FUNC_VF_NUM_M) |
307 		       ((hw->pf_id << GLINT_VECT2FUNC_PF_NUM_S) &
308 			GLINT_VECT2FUNC_PF_NUM_M));
309 		wr32(hw, GLINT_VECT2FUNC(v), reg);
310 	}
311 
312 	/* Map mailbox interrupt to VF MSI-X vector 0 */
313 	wr32(hw, VPINT_MBX_CTL(device_based_vf_id), VPINT_MBX_CTL_CAUSE_ENA_M);
314 }
315 
316 /**
317  * ice_ena_vf_q_mappings - enable Rx/Tx queue mappings for a VF
318  * @vf: VF to enable the mappings for
319  * @max_txq: max Tx queues allowed on the VF's VSI
320  * @max_rxq: max Rx queues allowed on the VF's VSI
321  */
322 static void ice_ena_vf_q_mappings(struct ice_vf *vf, u16 max_txq, u16 max_rxq)
323 {
324 	struct device *dev = ice_pf_to_dev(vf->pf);
325 	struct ice_vsi *vsi = ice_get_vf_vsi(vf);
326 	struct ice_hw *hw = &vf->pf->hw;
327 	u32 reg;
328 
329 	if (WARN_ON(!vsi))
330 		return;
331 
332 	/* set regardless of mapping mode */
333 	wr32(hw, VPLAN_TXQ_MAPENA(vf->vf_id), VPLAN_TXQ_MAPENA_TX_ENA_M);
334 
335 	/* VF Tx queues allocation */
336 	if (vsi->tx_mapping_mode == ICE_VSI_MAP_CONTIG) {
337 		/* set the VF PF Tx queue range
338 		 * VFNUMQ value should be set to (number of queues - 1). A value
339 		 * of 0 means 1 queue and a value of 255 means 256 queues
340 		 */
341 		reg = (((vsi->txq_map[0] << VPLAN_TX_QBASE_VFFIRSTQ_S) &
342 			VPLAN_TX_QBASE_VFFIRSTQ_M) |
343 		       (((max_txq - 1) << VPLAN_TX_QBASE_VFNUMQ_S) &
344 			VPLAN_TX_QBASE_VFNUMQ_M));
345 		wr32(hw, VPLAN_TX_QBASE(vf->vf_id), reg);
346 	} else {
347 		dev_err(dev, "Scattered mode for VF Tx queues is not yet implemented\n");
348 	}
349 
350 	/* set regardless of mapping mode */
351 	wr32(hw, VPLAN_RXQ_MAPENA(vf->vf_id), VPLAN_RXQ_MAPENA_RX_ENA_M);
352 
353 	/* VF Rx queues allocation */
354 	if (vsi->rx_mapping_mode == ICE_VSI_MAP_CONTIG) {
355 		/* set the VF PF Rx queue range
356 		 * VFNUMQ value should be set to (number of queues - 1). A value
357 		 * of 0 means 1 queue and a value of 255 means 256 queues
358 		 */
359 		reg = (((vsi->rxq_map[0] << VPLAN_RX_QBASE_VFFIRSTQ_S) &
360 			VPLAN_RX_QBASE_VFFIRSTQ_M) |
361 		       (((max_rxq - 1) << VPLAN_RX_QBASE_VFNUMQ_S) &
362 			VPLAN_RX_QBASE_VFNUMQ_M));
363 		wr32(hw, VPLAN_RX_QBASE(vf->vf_id), reg);
364 	} else {
365 		dev_err(dev, "Scattered mode for VF Rx queues is not yet implemented\n");
366 	}
367 }
368 
369 /**
370  * ice_ena_vf_mappings - enable VF MSIX and queue mapping
371  * @vf: pointer to the VF structure
372  */
373 static void ice_ena_vf_mappings(struct ice_vf *vf)
374 {
375 	struct ice_vsi *vsi = ice_get_vf_vsi(vf);
376 
377 	if (WARN_ON(!vsi))
378 		return;
379 
380 	ice_ena_vf_msix_mappings(vf);
381 	ice_ena_vf_q_mappings(vf, vsi->alloc_txq, vsi->alloc_rxq);
382 }
383 
384 /**
385  * ice_calc_vf_reg_idx - Calculate the VF's register index in the PF space
386  * @vf: VF to calculate the register index for
387  * @q_vector: a q_vector associated to the VF
388  */
389 int ice_calc_vf_reg_idx(struct ice_vf *vf, struct ice_q_vector *q_vector)
390 {
391 	struct ice_pf *pf;
392 
393 	if (!vf || !q_vector)
394 		return -EINVAL;
395 
396 	pf = vf->pf;
397 
398 	/* always add one to account for the OICR being the first MSIX */
399 	return pf->sriov_base_vector + pf->vfs.num_msix_per * vf->vf_id +
400 		q_vector->v_idx + 1;
401 }
402 
403 /**
404  * ice_sriov_set_msix_res - Set any used MSIX resources
405  * @pf: pointer to PF structure
406  * @num_msix_needed: number of MSIX vectors needed for all SR-IOV VFs
407  *
408  * This function allows SR-IOV resources to be taken from the end of the PF's
409  * allowed HW MSIX vectors so that the irq_tracker will not be affected. We
410  * just set the pf->sriov_base_vector and return success.
411  *
412  * If there are not enough resources available, return an error. This should
413  * always be caught by ice_set_per_vf_res().
414  *
415  * Return 0 on success, and -EINVAL when there are not enough MSIX vectors
416  * in the PF's space available for SR-IOV.
417  */
418 static int ice_sriov_set_msix_res(struct ice_pf *pf, u16 num_msix_needed)
419 {
420 	u16 total_vectors = pf->hw.func_caps.common_cap.num_msix_vectors;
421 	int vectors_used = ice_get_max_used_msix_vector(pf);
422 	int sriov_base_vector;
423 
424 	sriov_base_vector = total_vectors - num_msix_needed;
425 
426 	/* make sure we only grab irq_tracker entries from the list end and
427 	 * that we have enough available MSIX vectors
428 	 */
429 	if (sriov_base_vector < vectors_used)
430 		return -EINVAL;
431 
432 	pf->sriov_base_vector = sriov_base_vector;
433 
434 	return 0;
435 }
436 
437 /**
438  * ice_set_per_vf_res - check if vectors and queues are available
439  * @pf: pointer to the PF structure
440  * @num_vfs: the number of SR-IOV VFs being configured
441  *
442  * First, determine HW interrupts from common pool. If we allocate fewer VFs, we
443  * get more vectors and can enable more queues per VF. Note that this does not
444  * grab any vectors from the SW pool already allocated. Also note, that all
445  * vector counts include one for each VF's miscellaneous interrupt vector
446  * (i.e. OICR).
447  *
448  * Minimum VFs - 2 vectors, 1 queue pair
449  * Small VFs - 5 vectors, 4 queue pairs
450  * Medium VFs - 17 vectors, 16 queue pairs
451  *
452  * Second, determine number of queue pairs per VF by starting with a pre-defined
453  * maximum each VF supports. If this is not possible, then we adjust based on
454  * queue pairs available on the device.
455  *
456  * Lastly, set queue and MSI-X VF variables tracked by the PF so it can be used
457  * by each VF during VF initialization and reset.
458  */
459 static int ice_set_per_vf_res(struct ice_pf *pf, u16 num_vfs)
460 {
461 	int vectors_used = ice_get_max_used_msix_vector(pf);
462 	u16 num_msix_per_vf, num_txq, num_rxq, avail_qs;
463 	int msix_avail_per_vf, msix_avail_for_sriov;
464 	struct device *dev = ice_pf_to_dev(pf);
465 	int err;
466 
467 	lockdep_assert_held(&pf->vfs.table_lock);
468 
469 	if (!num_vfs)
470 		return -EINVAL;
471 
472 	/* determine MSI-X resources per VF */
473 	msix_avail_for_sriov = pf->hw.func_caps.common_cap.num_msix_vectors -
474 		vectors_used;
475 	msix_avail_per_vf = msix_avail_for_sriov / num_vfs;
476 	if (msix_avail_per_vf >= ICE_NUM_VF_MSIX_MED) {
477 		num_msix_per_vf = ICE_NUM_VF_MSIX_MED;
478 	} else if (msix_avail_per_vf >= ICE_NUM_VF_MSIX_SMALL) {
479 		num_msix_per_vf = ICE_NUM_VF_MSIX_SMALL;
480 	} else if (msix_avail_per_vf >= ICE_NUM_VF_MSIX_MULTIQ_MIN) {
481 		num_msix_per_vf = ICE_NUM_VF_MSIX_MULTIQ_MIN;
482 	} else if (msix_avail_per_vf >= ICE_MIN_INTR_PER_VF) {
483 		num_msix_per_vf = ICE_MIN_INTR_PER_VF;
484 	} else {
485 		dev_err(dev, "Only %d MSI-X interrupts available for SR-IOV. Not enough to support minimum of %d MSI-X interrupts per VF for %d VFs\n",
486 			msix_avail_for_sriov, ICE_MIN_INTR_PER_VF,
487 			num_vfs);
488 		return -ENOSPC;
489 	}
490 
491 	num_txq = min_t(u16, num_msix_per_vf - ICE_NONQ_VECS_VF,
492 			ICE_MAX_RSS_QS_PER_VF);
493 	avail_qs = ice_get_avail_txq_count(pf) / num_vfs;
494 	if (!avail_qs)
495 		num_txq = 0;
496 	else if (num_txq > avail_qs)
497 		num_txq = rounddown_pow_of_two(avail_qs);
498 
499 	num_rxq = min_t(u16, num_msix_per_vf - ICE_NONQ_VECS_VF,
500 			ICE_MAX_RSS_QS_PER_VF);
501 	avail_qs = ice_get_avail_rxq_count(pf) / num_vfs;
502 	if (!avail_qs)
503 		num_rxq = 0;
504 	else if (num_rxq > avail_qs)
505 		num_rxq = rounddown_pow_of_two(avail_qs);
506 
507 	if (num_txq < ICE_MIN_QS_PER_VF || num_rxq < ICE_MIN_QS_PER_VF) {
508 		dev_err(dev, "Not enough queues to support minimum of %d queue pairs per VF for %d VFs\n",
509 			ICE_MIN_QS_PER_VF, num_vfs);
510 		return -ENOSPC;
511 	}
512 
513 	err = ice_sriov_set_msix_res(pf, num_msix_per_vf * num_vfs);
514 	if (err) {
515 		dev_err(dev, "Unable to set MSI-X resources for %d VFs, err %d\n",
516 			num_vfs, err);
517 		return err;
518 	}
519 
520 	/* only allow equal Tx/Rx queue count (i.e. queue pairs) */
521 	pf->vfs.num_qps_per = min_t(int, num_txq, num_rxq);
522 	pf->vfs.num_msix_per = num_msix_per_vf;
523 	dev_info(dev, "Enabling %d VFs with %d vectors and %d queues per VF\n",
524 		 num_vfs, pf->vfs.num_msix_per, pf->vfs.num_qps_per);
525 
526 	return 0;
527 }
528 
529 /**
530  * ice_init_vf_vsi_res - initialize/setup VF VSI resources
531  * @vf: VF to initialize/setup the VSI for
532  *
533  * This function creates a VSI for the VF, adds a VLAN 0 filter, and sets up the
534  * VF VSI's broadcast filter and is only used during initial VF creation.
535  */
536 static int ice_init_vf_vsi_res(struct ice_vf *vf)
537 {
538 	struct ice_pf *pf = vf->pf;
539 	struct ice_vsi *vsi;
540 	int err;
541 
542 	vf->first_vector_idx = ice_calc_vf_first_vector_idx(pf, vf);
543 
544 	vsi = ice_vf_vsi_setup(vf);
545 	if (!vsi)
546 		return -ENOMEM;
547 
548 	err = ice_vf_init_host_cfg(vf, vsi);
549 	if (err)
550 		goto release_vsi;
551 
552 	return 0;
553 
554 release_vsi:
555 	ice_vf_vsi_release(vf);
556 	return err;
557 }
558 
559 /**
560  * ice_start_vfs - start VFs so they are ready to be used by SR-IOV
561  * @pf: PF the VFs are associated with
562  */
563 static int ice_start_vfs(struct ice_pf *pf)
564 {
565 	struct ice_hw *hw = &pf->hw;
566 	unsigned int bkt, it_cnt;
567 	struct ice_vf *vf;
568 	int retval;
569 
570 	lockdep_assert_held(&pf->vfs.table_lock);
571 
572 	it_cnt = 0;
573 	ice_for_each_vf(pf, bkt, vf) {
574 		vf->vf_ops->clear_reset_trigger(vf);
575 
576 		retval = ice_init_vf_vsi_res(vf);
577 		if (retval) {
578 			dev_err(ice_pf_to_dev(pf), "Failed to initialize VSI resources for VF %d, error %d\n",
579 				vf->vf_id, retval);
580 			goto teardown;
581 		}
582 
583 		set_bit(ICE_VF_STATE_INIT, vf->vf_states);
584 		ice_ena_vf_mappings(vf);
585 		wr32(hw, VFGEN_RSTAT(vf->vf_id), VIRTCHNL_VFR_VFACTIVE);
586 		it_cnt++;
587 	}
588 
589 	ice_flush(hw);
590 	return 0;
591 
592 teardown:
593 	ice_for_each_vf(pf, bkt, vf) {
594 		if (it_cnt == 0)
595 			break;
596 
597 		ice_dis_vf_mappings(vf);
598 		ice_vf_vsi_release(vf);
599 		it_cnt--;
600 	}
601 
602 	return retval;
603 }
604 
605 /**
606  * ice_sriov_free_vf - Free VF memory after all references are dropped
607  * @vf: pointer to VF to free
608  *
609  * Called by ice_put_vf through ice_release_vf once the last reference to a VF
610  * structure has been dropped.
611  */
612 static void ice_sriov_free_vf(struct ice_vf *vf)
613 {
614 	mutex_destroy(&vf->cfg_lock);
615 
616 	kfree_rcu(vf, rcu);
617 }
618 
619 /**
620  * ice_sriov_clear_reset_state - clears VF Reset status register
621  * @vf: the vf to configure
622  */
623 static void ice_sriov_clear_reset_state(struct ice_vf *vf)
624 {
625 	struct ice_hw *hw = &vf->pf->hw;
626 
627 	/* Clear the reset status register so that VF immediately sees that
628 	 * the device is resetting, even if hardware hasn't yet gotten around
629 	 * to clearing VFGEN_RSTAT for us.
630 	 */
631 	wr32(hw, VFGEN_RSTAT(vf->vf_id), VIRTCHNL_VFR_INPROGRESS);
632 }
633 
634 /**
635  * ice_sriov_clear_mbx_register - clears SRIOV VF's mailbox registers
636  * @vf: the vf to configure
637  */
638 static void ice_sriov_clear_mbx_register(struct ice_vf *vf)
639 {
640 	struct ice_pf *pf = vf->pf;
641 
642 	wr32(&pf->hw, VF_MBX_ARQLEN(vf->vf_id), 0);
643 	wr32(&pf->hw, VF_MBX_ATQLEN(vf->vf_id), 0);
644 }
645 
646 /**
647  * ice_sriov_trigger_reset_register - trigger VF reset for SRIOV VF
648  * @vf: pointer to VF structure
649  * @is_vflr: true if reset occurred due to VFLR
650  *
651  * Trigger and cleanup after a VF reset for a SR-IOV VF.
652  */
653 static void ice_sriov_trigger_reset_register(struct ice_vf *vf, bool is_vflr)
654 {
655 	struct ice_pf *pf = vf->pf;
656 	u32 reg, reg_idx, bit_idx;
657 	unsigned int vf_abs_id, i;
658 	struct device *dev;
659 	struct ice_hw *hw;
660 
661 	dev = ice_pf_to_dev(pf);
662 	hw = &pf->hw;
663 	vf_abs_id = vf->vf_id + hw->func_caps.vf_base_id;
664 
665 	/* In the case of a VFLR, HW has already reset the VF and we just need
666 	 * to clean up. Otherwise we must first trigger the reset using the
667 	 * VFRTRIG register.
668 	 */
669 	if (!is_vflr) {
670 		reg = rd32(hw, VPGEN_VFRTRIG(vf->vf_id));
671 		reg |= VPGEN_VFRTRIG_VFSWR_M;
672 		wr32(hw, VPGEN_VFRTRIG(vf->vf_id), reg);
673 	}
674 
675 	/* clear the VFLR bit in GLGEN_VFLRSTAT */
676 	reg_idx = (vf_abs_id) / 32;
677 	bit_idx = (vf_abs_id) % 32;
678 	wr32(hw, GLGEN_VFLRSTAT(reg_idx), BIT(bit_idx));
679 	ice_flush(hw);
680 
681 	wr32(hw, PF_PCI_CIAA,
682 	     VF_DEVICE_STATUS | (vf_abs_id << PF_PCI_CIAA_VF_NUM_S));
683 	for (i = 0; i < ICE_PCI_CIAD_WAIT_COUNT; i++) {
684 		reg = rd32(hw, PF_PCI_CIAD);
685 		/* no transactions pending so stop polling */
686 		if ((reg & VF_TRANS_PENDING_M) == 0)
687 			break;
688 
689 		dev_err(dev, "VF %u PCI transactions stuck\n", vf->vf_id);
690 		udelay(ICE_PCI_CIAD_WAIT_DELAY_US);
691 	}
692 }
693 
694 /**
695  * ice_sriov_poll_reset_status - poll SRIOV VF reset status
696  * @vf: pointer to VF structure
697  *
698  * Returns true when reset is successful, else returns false
699  */
700 static bool ice_sriov_poll_reset_status(struct ice_vf *vf)
701 {
702 	struct ice_pf *pf = vf->pf;
703 	unsigned int i;
704 	u32 reg;
705 
706 	for (i = 0; i < 10; i++) {
707 		/* VF reset requires driver to first reset the VF and then
708 		 * poll the status register to make sure that the reset
709 		 * completed successfully.
710 		 */
711 		reg = rd32(&pf->hw, VPGEN_VFRSTAT(vf->vf_id));
712 		if (reg & VPGEN_VFRSTAT_VFRD_M)
713 			return true;
714 
715 		/* only sleep if the reset is not done */
716 		usleep_range(10, 20);
717 	}
718 	return false;
719 }
720 
721 /**
722  * ice_sriov_clear_reset_trigger - enable VF to access hardware
723  * @vf: VF to enabled hardware access for
724  */
725 static void ice_sriov_clear_reset_trigger(struct ice_vf *vf)
726 {
727 	struct ice_hw *hw = &vf->pf->hw;
728 	u32 reg;
729 
730 	reg = rd32(hw, VPGEN_VFRTRIG(vf->vf_id));
731 	reg &= ~VPGEN_VFRTRIG_VFSWR_M;
732 	wr32(hw, VPGEN_VFRTRIG(vf->vf_id), reg);
733 	ice_flush(hw);
734 }
735 
736 /**
737  * ice_sriov_create_vsi - Create a new VSI for a VF
738  * @vf: VF to create the VSI for
739  *
740  * This is called by ice_vf_recreate_vsi to create the new VSI after the old
741  * VSI has been released.
742  */
743 static int ice_sriov_create_vsi(struct ice_vf *vf)
744 {
745 	struct ice_vsi *vsi;
746 
747 	vsi = ice_vf_vsi_setup(vf);
748 	if (!vsi)
749 		return -ENOMEM;
750 
751 	return 0;
752 }
753 
754 /**
755  * ice_sriov_post_vsi_rebuild - tasks to do after the VF's VSI have been rebuilt
756  * @vf: VF to perform tasks on
757  */
758 static void ice_sriov_post_vsi_rebuild(struct ice_vf *vf)
759 {
760 	ice_ena_vf_mappings(vf);
761 	wr32(&vf->pf->hw, VFGEN_RSTAT(vf->vf_id), VIRTCHNL_VFR_VFACTIVE);
762 }
763 
764 static const struct ice_vf_ops ice_sriov_vf_ops = {
765 	.reset_type = ICE_VF_RESET,
766 	.free = ice_sriov_free_vf,
767 	.clear_reset_state = ice_sriov_clear_reset_state,
768 	.clear_mbx_register = ice_sriov_clear_mbx_register,
769 	.trigger_reset_register = ice_sriov_trigger_reset_register,
770 	.poll_reset_status = ice_sriov_poll_reset_status,
771 	.clear_reset_trigger = ice_sriov_clear_reset_trigger,
772 	.irq_close = NULL,
773 	.create_vsi = ice_sriov_create_vsi,
774 	.post_vsi_rebuild = ice_sriov_post_vsi_rebuild,
775 };
776 
777 /**
778  * ice_create_vf_entries - Allocate and insert VF entries
779  * @pf: pointer to the PF structure
780  * @num_vfs: the number of VFs to allocate
781  *
782  * Allocate new VF entries and insert them into the hash table. Set some
783  * basic default fields for initializing the new VFs.
784  *
785  * After this function exits, the hash table will have num_vfs entries
786  * inserted.
787  *
788  * Returns 0 on success or an integer error code on failure.
789  */
790 static int ice_create_vf_entries(struct ice_pf *pf, u16 num_vfs)
791 {
792 	struct ice_vfs *vfs = &pf->vfs;
793 	struct ice_vf *vf;
794 	u16 vf_id;
795 	int err;
796 
797 	lockdep_assert_held(&vfs->table_lock);
798 
799 	for (vf_id = 0; vf_id < num_vfs; vf_id++) {
800 		vf = kzalloc(sizeof(*vf), GFP_KERNEL);
801 		if (!vf) {
802 			err = -ENOMEM;
803 			goto err_free_entries;
804 		}
805 		kref_init(&vf->refcnt);
806 
807 		vf->pf = pf;
808 		vf->vf_id = vf_id;
809 
810 		/* set sriov vf ops for VFs created during SRIOV flow */
811 		vf->vf_ops = &ice_sriov_vf_ops;
812 
813 		ice_initialize_vf_entry(vf);
814 
815 		vf->vf_sw_id = pf->first_sw;
816 
817 		hash_add_rcu(vfs->table, &vf->entry, vf_id);
818 	}
819 
820 	return 0;
821 
822 err_free_entries:
823 	ice_free_vf_entries(pf);
824 	return err;
825 }
826 
827 /**
828  * ice_ena_vfs - enable VFs so they are ready to be used
829  * @pf: pointer to the PF structure
830  * @num_vfs: number of VFs to enable
831  */
832 static int ice_ena_vfs(struct ice_pf *pf, u16 num_vfs)
833 {
834 	struct device *dev = ice_pf_to_dev(pf);
835 	struct ice_hw *hw = &pf->hw;
836 	int ret;
837 
838 	/* Disable global interrupt 0 so we don't try to handle the VFLR. */
839 	wr32(hw, GLINT_DYN_CTL(pf->oicr_irq.index),
840 	     ICE_ITR_NONE << GLINT_DYN_CTL_ITR_INDX_S);
841 	set_bit(ICE_OICR_INTR_DIS, pf->state);
842 	ice_flush(hw);
843 
844 	ret = pci_enable_sriov(pf->pdev, num_vfs);
845 	if (ret)
846 		goto err_unroll_intr;
847 
848 	mutex_lock(&pf->vfs.table_lock);
849 
850 	ret = ice_set_per_vf_res(pf, num_vfs);
851 	if (ret) {
852 		dev_err(dev, "Not enough resources for %d VFs, err %d. Try with fewer number of VFs\n",
853 			num_vfs, ret);
854 		goto err_unroll_sriov;
855 	}
856 
857 	ret = ice_create_vf_entries(pf, num_vfs);
858 	if (ret) {
859 		dev_err(dev, "Failed to allocate VF entries for %d VFs\n",
860 			num_vfs);
861 		goto err_unroll_sriov;
862 	}
863 
864 	ret = ice_start_vfs(pf);
865 	if (ret) {
866 		dev_err(dev, "Failed to start %d VFs, err %d\n", num_vfs, ret);
867 		ret = -EAGAIN;
868 		goto err_unroll_vf_entries;
869 	}
870 
871 	clear_bit(ICE_VF_DIS, pf->state);
872 
873 	ret = ice_eswitch_configure(pf);
874 	if (ret) {
875 		dev_err(dev, "Failed to configure eswitch, err %d\n", ret);
876 		goto err_unroll_sriov;
877 	}
878 
879 	/* rearm global interrupts */
880 	if (test_and_clear_bit(ICE_OICR_INTR_DIS, pf->state))
881 		ice_irq_dynamic_ena(hw, NULL, NULL);
882 
883 	mutex_unlock(&pf->vfs.table_lock);
884 
885 	return 0;
886 
887 err_unroll_vf_entries:
888 	ice_free_vf_entries(pf);
889 err_unroll_sriov:
890 	mutex_unlock(&pf->vfs.table_lock);
891 	pci_disable_sriov(pf->pdev);
892 err_unroll_intr:
893 	/* rearm interrupts here */
894 	ice_irq_dynamic_ena(hw, NULL, NULL);
895 	clear_bit(ICE_OICR_INTR_DIS, pf->state);
896 	return ret;
897 }
898 
899 /**
900  * ice_pci_sriov_ena - Enable or change number of VFs
901  * @pf: pointer to the PF structure
902  * @num_vfs: number of VFs to allocate
903  *
904  * Returns 0 on success and negative on failure
905  */
906 static int ice_pci_sriov_ena(struct ice_pf *pf, int num_vfs)
907 {
908 	struct device *dev = ice_pf_to_dev(pf);
909 	int err;
910 
911 	if (!num_vfs) {
912 		ice_free_vfs(pf);
913 		return 0;
914 	}
915 
916 	if (num_vfs > pf->vfs.num_supported) {
917 		dev_err(dev, "Can't enable %d VFs, max VFs supported is %d\n",
918 			num_vfs, pf->vfs.num_supported);
919 		return -EOPNOTSUPP;
920 	}
921 
922 	dev_info(dev, "Enabling %d VFs\n", num_vfs);
923 	err = ice_ena_vfs(pf, num_vfs);
924 	if (err) {
925 		dev_err(dev, "Failed to enable SR-IOV: %d\n", err);
926 		return err;
927 	}
928 
929 	set_bit(ICE_FLAG_SRIOV_ENA, pf->flags);
930 	return 0;
931 }
932 
933 /**
934  * ice_check_sriov_allowed - check if SR-IOV is allowed based on various checks
935  * @pf: PF to enabled SR-IOV on
936  */
937 static int ice_check_sriov_allowed(struct ice_pf *pf)
938 {
939 	struct device *dev = ice_pf_to_dev(pf);
940 
941 	if (!test_bit(ICE_FLAG_SRIOV_CAPABLE, pf->flags)) {
942 		dev_err(dev, "This device is not capable of SR-IOV\n");
943 		return -EOPNOTSUPP;
944 	}
945 
946 	if (ice_is_safe_mode(pf)) {
947 		dev_err(dev, "SR-IOV cannot be configured - Device is in Safe Mode\n");
948 		return -EOPNOTSUPP;
949 	}
950 
951 	if (!ice_pf_state_is_nominal(pf)) {
952 		dev_err(dev, "Cannot enable SR-IOV, device not ready\n");
953 		return -EBUSY;
954 	}
955 
956 	return 0;
957 }
958 
959 /**
960  * ice_sriov_configure - Enable or change number of VFs via sysfs
961  * @pdev: pointer to a pci_dev structure
962  * @num_vfs: number of VFs to allocate or 0 to free VFs
963  *
964  * This function is called when the user updates the number of VFs in sysfs. On
965  * success return whatever num_vfs was set to by the caller. Return negative on
966  * failure.
967  */
968 int ice_sriov_configure(struct pci_dev *pdev, int num_vfs)
969 {
970 	struct ice_pf *pf = pci_get_drvdata(pdev);
971 	struct device *dev = ice_pf_to_dev(pf);
972 	int err;
973 
974 	err = ice_check_sriov_allowed(pf);
975 	if (err)
976 		return err;
977 
978 	if (!num_vfs) {
979 		if (!pci_vfs_assigned(pdev)) {
980 			ice_free_vfs(pf);
981 			return 0;
982 		}
983 
984 		dev_err(dev, "can't free VFs because some are assigned to VMs.\n");
985 		return -EBUSY;
986 	}
987 
988 	err = ice_pci_sriov_ena(pf, num_vfs);
989 	if (err)
990 		return err;
991 
992 	return num_vfs;
993 }
994 
995 /**
996  * ice_process_vflr_event - Free VF resources via IRQ calls
997  * @pf: pointer to the PF structure
998  *
999  * called from the VFLR IRQ handler to
1000  * free up VF resources and state variables
1001  */
1002 void ice_process_vflr_event(struct ice_pf *pf)
1003 {
1004 	struct ice_hw *hw = &pf->hw;
1005 	struct ice_vf *vf;
1006 	unsigned int bkt;
1007 	u32 reg;
1008 
1009 	if (!test_and_clear_bit(ICE_VFLR_EVENT_PENDING, pf->state) ||
1010 	    !ice_has_vfs(pf))
1011 		return;
1012 
1013 	mutex_lock(&pf->vfs.table_lock);
1014 	ice_for_each_vf(pf, bkt, vf) {
1015 		u32 reg_idx, bit_idx;
1016 
1017 		reg_idx = (hw->func_caps.vf_base_id + vf->vf_id) / 32;
1018 		bit_idx = (hw->func_caps.vf_base_id + vf->vf_id) % 32;
1019 		/* read GLGEN_VFLRSTAT register to find out the flr VFs */
1020 		reg = rd32(hw, GLGEN_VFLRSTAT(reg_idx));
1021 		if (reg & BIT(bit_idx))
1022 			/* GLGEN_VFLRSTAT bit will be cleared in ice_reset_vf */
1023 			ice_reset_vf(vf, ICE_VF_RESET_VFLR | ICE_VF_RESET_LOCK);
1024 	}
1025 	mutex_unlock(&pf->vfs.table_lock);
1026 }
1027 
1028 /**
1029  * ice_get_vf_from_pfq - get the VF who owns the PF space queue passed in
1030  * @pf: PF used to index all VFs
1031  * @pfq: queue index relative to the PF's function space
1032  *
1033  * If no VF is found who owns the pfq then return NULL, otherwise return a
1034  * pointer to the VF who owns the pfq
1035  *
1036  * If this function returns non-NULL, it acquires a reference count of the VF
1037  * structure. The caller is responsible for calling ice_put_vf() to drop this
1038  * reference.
1039  */
1040 static struct ice_vf *ice_get_vf_from_pfq(struct ice_pf *pf, u16 pfq)
1041 {
1042 	struct ice_vf *vf;
1043 	unsigned int bkt;
1044 
1045 	rcu_read_lock();
1046 	ice_for_each_vf_rcu(pf, bkt, vf) {
1047 		struct ice_vsi *vsi;
1048 		u16 rxq_idx;
1049 
1050 		vsi = ice_get_vf_vsi(vf);
1051 		if (!vsi)
1052 			continue;
1053 
1054 		ice_for_each_rxq(vsi, rxq_idx)
1055 			if (vsi->rxq_map[rxq_idx] == pfq) {
1056 				struct ice_vf *found;
1057 
1058 				if (kref_get_unless_zero(&vf->refcnt))
1059 					found = vf;
1060 				else
1061 					found = NULL;
1062 				rcu_read_unlock();
1063 				return found;
1064 			}
1065 	}
1066 	rcu_read_unlock();
1067 
1068 	return NULL;
1069 }
1070 
1071 /**
1072  * ice_globalq_to_pfq - convert from global queue index to PF space queue index
1073  * @pf: PF used for conversion
1074  * @globalq: global queue index used to convert to PF space queue index
1075  */
1076 static u32 ice_globalq_to_pfq(struct ice_pf *pf, u32 globalq)
1077 {
1078 	return globalq - pf->hw.func_caps.common_cap.rxq_first_id;
1079 }
1080 
1081 /**
1082  * ice_vf_lan_overflow_event - handle LAN overflow event for a VF
1083  * @pf: PF that the LAN overflow event happened on
1084  * @event: structure holding the event information for the LAN overflow event
1085  *
1086  * Determine if the LAN overflow event was caused by a VF queue. If it was not
1087  * caused by a VF, do nothing. If a VF caused this LAN overflow event trigger a
1088  * reset on the offending VF.
1089  */
1090 void
1091 ice_vf_lan_overflow_event(struct ice_pf *pf, struct ice_rq_event_info *event)
1092 {
1093 	u32 gldcb_rtctq, queue;
1094 	struct ice_vf *vf;
1095 
1096 	gldcb_rtctq = le32_to_cpu(event->desc.params.lan_overflow.prtdcb_ruptq);
1097 	dev_dbg(ice_pf_to_dev(pf), "GLDCB_RTCTQ: 0x%08x\n", gldcb_rtctq);
1098 
1099 	/* event returns device global Rx queue number */
1100 	queue = (gldcb_rtctq & GLDCB_RTCTQ_RXQNUM_M) >>
1101 		GLDCB_RTCTQ_RXQNUM_S;
1102 
1103 	vf = ice_get_vf_from_pfq(pf, ice_globalq_to_pfq(pf, queue));
1104 	if (!vf)
1105 		return;
1106 
1107 	ice_reset_vf(vf, ICE_VF_RESET_NOTIFY | ICE_VF_RESET_LOCK);
1108 	ice_put_vf(vf);
1109 }
1110 
1111 /**
1112  * ice_set_vf_spoofchk
1113  * @netdev: network interface device structure
1114  * @vf_id: VF identifier
1115  * @ena: flag to enable or disable feature
1116  *
1117  * Enable or disable VF spoof checking
1118  */
1119 int ice_set_vf_spoofchk(struct net_device *netdev, int vf_id, bool ena)
1120 {
1121 	struct ice_netdev_priv *np = netdev_priv(netdev);
1122 	struct ice_pf *pf = np->vsi->back;
1123 	struct ice_vsi *vf_vsi;
1124 	struct device *dev;
1125 	struct ice_vf *vf;
1126 	int ret;
1127 
1128 	dev = ice_pf_to_dev(pf);
1129 
1130 	vf = ice_get_vf_by_id(pf, vf_id);
1131 	if (!vf)
1132 		return -EINVAL;
1133 
1134 	ret = ice_check_vf_ready_for_reset(vf);
1135 	if (ret)
1136 		goto out_put_vf;
1137 
1138 	vf_vsi = ice_get_vf_vsi(vf);
1139 	if (!vf_vsi) {
1140 		netdev_err(netdev, "VSI %d for VF %d is null\n",
1141 			   vf->lan_vsi_idx, vf->vf_id);
1142 		ret = -EINVAL;
1143 		goto out_put_vf;
1144 	}
1145 
1146 	if (vf_vsi->type != ICE_VSI_VF) {
1147 		netdev_err(netdev, "Type %d of VSI %d for VF %d is no ICE_VSI_VF\n",
1148 			   vf_vsi->type, vf_vsi->vsi_num, vf->vf_id);
1149 		ret = -ENODEV;
1150 		goto out_put_vf;
1151 	}
1152 
1153 	if (ena == vf->spoofchk) {
1154 		dev_dbg(dev, "VF spoofchk already %s\n", ena ? "ON" : "OFF");
1155 		ret = 0;
1156 		goto out_put_vf;
1157 	}
1158 
1159 	ret = ice_vsi_apply_spoofchk(vf_vsi, ena);
1160 	if (ret)
1161 		dev_err(dev, "Failed to set spoofchk %s for VF %d VSI %d\n error %d\n",
1162 			ena ? "ON" : "OFF", vf->vf_id, vf_vsi->vsi_num, ret);
1163 	else
1164 		vf->spoofchk = ena;
1165 
1166 out_put_vf:
1167 	ice_put_vf(vf);
1168 	return ret;
1169 }
1170 
1171 /**
1172  * ice_get_vf_cfg
1173  * @netdev: network interface device structure
1174  * @vf_id: VF identifier
1175  * @ivi: VF configuration structure
1176  *
1177  * return VF configuration
1178  */
1179 int
1180 ice_get_vf_cfg(struct net_device *netdev, int vf_id, struct ifla_vf_info *ivi)
1181 {
1182 	struct ice_pf *pf = ice_netdev_to_pf(netdev);
1183 	struct ice_vf *vf;
1184 	int ret;
1185 
1186 	vf = ice_get_vf_by_id(pf, vf_id);
1187 	if (!vf)
1188 		return -EINVAL;
1189 
1190 	ret = ice_check_vf_ready_for_cfg(vf);
1191 	if (ret)
1192 		goto out_put_vf;
1193 
1194 	ivi->vf = vf_id;
1195 	ether_addr_copy(ivi->mac, vf->hw_lan_addr);
1196 
1197 	/* VF configuration for VLAN and applicable QoS */
1198 	ivi->vlan = ice_vf_get_port_vlan_id(vf);
1199 	ivi->qos = ice_vf_get_port_vlan_prio(vf);
1200 	if (ice_vf_is_port_vlan_ena(vf))
1201 		ivi->vlan_proto = cpu_to_be16(ice_vf_get_port_vlan_tpid(vf));
1202 
1203 	ivi->trusted = vf->trusted;
1204 	ivi->spoofchk = vf->spoofchk;
1205 	if (!vf->link_forced)
1206 		ivi->linkstate = IFLA_VF_LINK_STATE_AUTO;
1207 	else if (vf->link_up)
1208 		ivi->linkstate = IFLA_VF_LINK_STATE_ENABLE;
1209 	else
1210 		ivi->linkstate = IFLA_VF_LINK_STATE_DISABLE;
1211 	ivi->max_tx_rate = vf->max_tx_rate;
1212 	ivi->min_tx_rate = vf->min_tx_rate;
1213 
1214 out_put_vf:
1215 	ice_put_vf(vf);
1216 	return ret;
1217 }
1218 
1219 /**
1220  * ice_set_vf_mac
1221  * @netdev: network interface device structure
1222  * @vf_id: VF identifier
1223  * @mac: MAC address
1224  *
1225  * program VF MAC address
1226  */
1227 int ice_set_vf_mac(struct net_device *netdev, int vf_id, u8 *mac)
1228 {
1229 	struct ice_pf *pf = ice_netdev_to_pf(netdev);
1230 	struct ice_vf *vf;
1231 	int ret;
1232 
1233 	if (is_multicast_ether_addr(mac)) {
1234 		netdev_err(netdev, "%pM not a valid unicast address\n", mac);
1235 		return -EINVAL;
1236 	}
1237 
1238 	vf = ice_get_vf_by_id(pf, vf_id);
1239 	if (!vf)
1240 		return -EINVAL;
1241 
1242 	/* nothing left to do, unicast MAC already set */
1243 	if (ether_addr_equal(vf->dev_lan_addr, mac) &&
1244 	    ether_addr_equal(vf->hw_lan_addr, mac)) {
1245 		ret = 0;
1246 		goto out_put_vf;
1247 	}
1248 
1249 	ret = ice_check_vf_ready_for_reset(vf);
1250 	if (ret)
1251 		goto out_put_vf;
1252 
1253 	mutex_lock(&vf->cfg_lock);
1254 
1255 	/* VF is notified of its new MAC via the PF's response to the
1256 	 * VIRTCHNL_OP_GET_VF_RESOURCES message after the VF has been reset
1257 	 */
1258 	ether_addr_copy(vf->dev_lan_addr, mac);
1259 	ether_addr_copy(vf->hw_lan_addr, mac);
1260 	if (is_zero_ether_addr(mac)) {
1261 		/* VF will send VIRTCHNL_OP_ADD_ETH_ADDR message with its MAC */
1262 		vf->pf_set_mac = false;
1263 		netdev_info(netdev, "Removing MAC on VF %d. VF driver will be reinitialized\n",
1264 			    vf->vf_id);
1265 	} else {
1266 		/* PF will add MAC rule for the VF */
1267 		vf->pf_set_mac = true;
1268 		netdev_info(netdev, "Setting MAC %pM on VF %d. VF driver will be reinitialized\n",
1269 			    mac, vf_id);
1270 	}
1271 
1272 	ice_reset_vf(vf, ICE_VF_RESET_NOTIFY);
1273 	mutex_unlock(&vf->cfg_lock);
1274 
1275 out_put_vf:
1276 	ice_put_vf(vf);
1277 	return ret;
1278 }
1279 
1280 /**
1281  * ice_set_vf_trust
1282  * @netdev: network interface device structure
1283  * @vf_id: VF identifier
1284  * @trusted: Boolean value to enable/disable trusted VF
1285  *
1286  * Enable or disable a given VF as trusted
1287  */
1288 int ice_set_vf_trust(struct net_device *netdev, int vf_id, bool trusted)
1289 {
1290 	struct ice_pf *pf = ice_netdev_to_pf(netdev);
1291 	struct ice_vf *vf;
1292 	int ret;
1293 
1294 	vf = ice_get_vf_by_id(pf, vf_id);
1295 	if (!vf)
1296 		return -EINVAL;
1297 
1298 	if (ice_is_eswitch_mode_switchdev(pf)) {
1299 		dev_info(ice_pf_to_dev(pf), "Trusted VF is forbidden in switchdev mode\n");
1300 		return -EOPNOTSUPP;
1301 	}
1302 
1303 	ret = ice_check_vf_ready_for_reset(vf);
1304 	if (ret)
1305 		goto out_put_vf;
1306 
1307 	/* Check if already trusted */
1308 	if (trusted == vf->trusted) {
1309 		ret = 0;
1310 		goto out_put_vf;
1311 	}
1312 
1313 	mutex_lock(&vf->cfg_lock);
1314 
1315 	vf->trusted = trusted;
1316 	ice_reset_vf(vf, ICE_VF_RESET_NOTIFY);
1317 	dev_info(ice_pf_to_dev(pf), "VF %u is now %strusted\n",
1318 		 vf_id, trusted ? "" : "un");
1319 
1320 	mutex_unlock(&vf->cfg_lock);
1321 
1322 out_put_vf:
1323 	ice_put_vf(vf);
1324 	return ret;
1325 }
1326 
1327 /**
1328  * ice_set_vf_link_state
1329  * @netdev: network interface device structure
1330  * @vf_id: VF identifier
1331  * @link_state: required link state
1332  *
1333  * Set VF's link state, irrespective of physical link state status
1334  */
1335 int ice_set_vf_link_state(struct net_device *netdev, int vf_id, int link_state)
1336 {
1337 	struct ice_pf *pf = ice_netdev_to_pf(netdev);
1338 	struct ice_vf *vf;
1339 	int ret;
1340 
1341 	vf = ice_get_vf_by_id(pf, vf_id);
1342 	if (!vf)
1343 		return -EINVAL;
1344 
1345 	ret = ice_check_vf_ready_for_cfg(vf);
1346 	if (ret)
1347 		goto out_put_vf;
1348 
1349 	switch (link_state) {
1350 	case IFLA_VF_LINK_STATE_AUTO:
1351 		vf->link_forced = false;
1352 		break;
1353 	case IFLA_VF_LINK_STATE_ENABLE:
1354 		vf->link_forced = true;
1355 		vf->link_up = true;
1356 		break;
1357 	case IFLA_VF_LINK_STATE_DISABLE:
1358 		vf->link_forced = true;
1359 		vf->link_up = false;
1360 		break;
1361 	default:
1362 		ret = -EINVAL;
1363 		goto out_put_vf;
1364 	}
1365 
1366 	ice_vc_notify_vf_link_state(vf);
1367 
1368 out_put_vf:
1369 	ice_put_vf(vf);
1370 	return ret;
1371 }
1372 
1373 /**
1374  * ice_calc_all_vfs_min_tx_rate - calculate cumulative min Tx rate on all VFs
1375  * @pf: PF associated with VFs
1376  */
1377 static int ice_calc_all_vfs_min_tx_rate(struct ice_pf *pf)
1378 {
1379 	struct ice_vf *vf;
1380 	unsigned int bkt;
1381 	int rate = 0;
1382 
1383 	rcu_read_lock();
1384 	ice_for_each_vf_rcu(pf, bkt, vf)
1385 		rate += vf->min_tx_rate;
1386 	rcu_read_unlock();
1387 
1388 	return rate;
1389 }
1390 
1391 /**
1392  * ice_min_tx_rate_oversubscribed - check if min Tx rate causes oversubscription
1393  * @vf: VF trying to configure min_tx_rate
1394  * @min_tx_rate: min Tx rate in Mbps
1395  *
1396  * Check if the min_tx_rate being passed in will cause oversubscription of total
1397  * min_tx_rate based on the current link speed and all other VFs configured
1398  * min_tx_rate
1399  *
1400  * Return true if the passed min_tx_rate would cause oversubscription, else
1401  * return false
1402  */
1403 static bool
1404 ice_min_tx_rate_oversubscribed(struct ice_vf *vf, int min_tx_rate)
1405 {
1406 	struct ice_vsi *vsi = ice_get_vf_vsi(vf);
1407 	int all_vfs_min_tx_rate;
1408 	int link_speed_mbps;
1409 
1410 	if (WARN_ON(!vsi))
1411 		return false;
1412 
1413 	link_speed_mbps = ice_get_link_speed_mbps(vsi);
1414 	all_vfs_min_tx_rate = ice_calc_all_vfs_min_tx_rate(vf->pf);
1415 
1416 	/* this VF's previous rate is being overwritten */
1417 	all_vfs_min_tx_rate -= vf->min_tx_rate;
1418 
1419 	if (all_vfs_min_tx_rate + min_tx_rate > link_speed_mbps) {
1420 		dev_err(ice_pf_to_dev(vf->pf), "min_tx_rate of %d Mbps on VF %u would cause oversubscription of %d Mbps based on the current link speed %d Mbps\n",
1421 			min_tx_rate, vf->vf_id,
1422 			all_vfs_min_tx_rate + min_tx_rate - link_speed_mbps,
1423 			link_speed_mbps);
1424 		return true;
1425 	}
1426 
1427 	return false;
1428 }
1429 
1430 /**
1431  * ice_set_vf_bw - set min/max VF bandwidth
1432  * @netdev: network interface device structure
1433  * @vf_id: VF identifier
1434  * @min_tx_rate: Minimum Tx rate in Mbps
1435  * @max_tx_rate: Maximum Tx rate in Mbps
1436  */
1437 int
1438 ice_set_vf_bw(struct net_device *netdev, int vf_id, int min_tx_rate,
1439 	      int max_tx_rate)
1440 {
1441 	struct ice_pf *pf = ice_netdev_to_pf(netdev);
1442 	struct ice_vsi *vsi;
1443 	struct device *dev;
1444 	struct ice_vf *vf;
1445 	int ret;
1446 
1447 	dev = ice_pf_to_dev(pf);
1448 
1449 	vf = ice_get_vf_by_id(pf, vf_id);
1450 	if (!vf)
1451 		return -EINVAL;
1452 
1453 	ret = ice_check_vf_ready_for_cfg(vf);
1454 	if (ret)
1455 		goto out_put_vf;
1456 
1457 	vsi = ice_get_vf_vsi(vf);
1458 	if (!vsi) {
1459 		ret = -EINVAL;
1460 		goto out_put_vf;
1461 	}
1462 
1463 	if (min_tx_rate && ice_is_dcb_active(pf)) {
1464 		dev_err(dev, "DCB on PF is currently enabled. VF min Tx rate limiting not allowed on this PF.\n");
1465 		ret = -EOPNOTSUPP;
1466 		goto out_put_vf;
1467 	}
1468 
1469 	if (ice_min_tx_rate_oversubscribed(vf, min_tx_rate)) {
1470 		ret = -EINVAL;
1471 		goto out_put_vf;
1472 	}
1473 
1474 	if (vf->min_tx_rate != (unsigned int)min_tx_rate) {
1475 		ret = ice_set_min_bw_limit(vsi, (u64)min_tx_rate * 1000);
1476 		if (ret) {
1477 			dev_err(dev, "Unable to set min-tx-rate for VF %d\n",
1478 				vf->vf_id);
1479 			goto out_put_vf;
1480 		}
1481 
1482 		vf->min_tx_rate = min_tx_rate;
1483 	}
1484 
1485 	if (vf->max_tx_rate != (unsigned int)max_tx_rate) {
1486 		ret = ice_set_max_bw_limit(vsi, (u64)max_tx_rate * 1000);
1487 		if (ret) {
1488 			dev_err(dev, "Unable to set max-tx-rate for VF %d\n",
1489 				vf->vf_id);
1490 			goto out_put_vf;
1491 		}
1492 
1493 		vf->max_tx_rate = max_tx_rate;
1494 	}
1495 
1496 out_put_vf:
1497 	ice_put_vf(vf);
1498 	return ret;
1499 }
1500 
1501 /**
1502  * ice_get_vf_stats - populate some stats for the VF
1503  * @netdev: the netdev of the PF
1504  * @vf_id: the host OS identifier (0-255)
1505  * @vf_stats: pointer to the OS memory to be initialized
1506  */
1507 int ice_get_vf_stats(struct net_device *netdev, int vf_id,
1508 		     struct ifla_vf_stats *vf_stats)
1509 {
1510 	struct ice_pf *pf = ice_netdev_to_pf(netdev);
1511 	struct ice_eth_stats *stats;
1512 	struct ice_vsi *vsi;
1513 	struct ice_vf *vf;
1514 	int ret;
1515 
1516 	vf = ice_get_vf_by_id(pf, vf_id);
1517 	if (!vf)
1518 		return -EINVAL;
1519 
1520 	ret = ice_check_vf_ready_for_cfg(vf);
1521 	if (ret)
1522 		goto out_put_vf;
1523 
1524 	vsi = ice_get_vf_vsi(vf);
1525 	if (!vsi) {
1526 		ret = -EINVAL;
1527 		goto out_put_vf;
1528 	}
1529 
1530 	ice_update_eth_stats(vsi);
1531 	stats = &vsi->eth_stats;
1532 
1533 	memset(vf_stats, 0, sizeof(*vf_stats));
1534 
1535 	vf_stats->rx_packets = stats->rx_unicast + stats->rx_broadcast +
1536 		stats->rx_multicast;
1537 	vf_stats->tx_packets = stats->tx_unicast + stats->tx_broadcast +
1538 		stats->tx_multicast;
1539 	vf_stats->rx_bytes   = stats->rx_bytes;
1540 	vf_stats->tx_bytes   = stats->tx_bytes;
1541 	vf_stats->broadcast  = stats->rx_broadcast;
1542 	vf_stats->multicast  = stats->rx_multicast;
1543 	vf_stats->rx_dropped = stats->rx_discards;
1544 	vf_stats->tx_dropped = stats->tx_discards;
1545 
1546 out_put_vf:
1547 	ice_put_vf(vf);
1548 	return ret;
1549 }
1550 
1551 /**
1552  * ice_is_supported_port_vlan_proto - make sure the vlan_proto is supported
1553  * @hw: hardware structure used to check the VLAN mode
1554  * @vlan_proto: VLAN TPID being checked
1555  *
1556  * If the device is configured in Double VLAN Mode (DVM), then both ETH_P_8021Q
1557  * and ETH_P_8021AD are supported. If the device is configured in Single VLAN
1558  * Mode (SVM), then only ETH_P_8021Q is supported.
1559  */
1560 static bool
1561 ice_is_supported_port_vlan_proto(struct ice_hw *hw, u16 vlan_proto)
1562 {
1563 	bool is_supported = false;
1564 
1565 	switch (vlan_proto) {
1566 	case ETH_P_8021Q:
1567 		is_supported = true;
1568 		break;
1569 	case ETH_P_8021AD:
1570 		if (ice_is_dvm_ena(hw))
1571 			is_supported = true;
1572 		break;
1573 	}
1574 
1575 	return is_supported;
1576 }
1577 
1578 /**
1579  * ice_set_vf_port_vlan
1580  * @netdev: network interface device structure
1581  * @vf_id: VF identifier
1582  * @vlan_id: VLAN ID being set
1583  * @qos: priority setting
1584  * @vlan_proto: VLAN protocol
1585  *
1586  * program VF Port VLAN ID and/or QoS
1587  */
1588 int
1589 ice_set_vf_port_vlan(struct net_device *netdev, int vf_id, u16 vlan_id, u8 qos,
1590 		     __be16 vlan_proto)
1591 {
1592 	struct ice_pf *pf = ice_netdev_to_pf(netdev);
1593 	u16 local_vlan_proto = ntohs(vlan_proto);
1594 	struct device *dev;
1595 	struct ice_vf *vf;
1596 	int ret;
1597 
1598 	dev = ice_pf_to_dev(pf);
1599 
1600 	if (vlan_id >= VLAN_N_VID || qos > 7) {
1601 		dev_err(dev, "Invalid Port VLAN parameters for VF %d, ID %d, QoS %d\n",
1602 			vf_id, vlan_id, qos);
1603 		return -EINVAL;
1604 	}
1605 
1606 	if (!ice_is_supported_port_vlan_proto(&pf->hw, local_vlan_proto)) {
1607 		dev_err(dev, "VF VLAN protocol 0x%04x is not supported\n",
1608 			local_vlan_proto);
1609 		return -EPROTONOSUPPORT;
1610 	}
1611 
1612 	vf = ice_get_vf_by_id(pf, vf_id);
1613 	if (!vf)
1614 		return -EINVAL;
1615 
1616 	ret = ice_check_vf_ready_for_reset(vf);
1617 	if (ret)
1618 		goto out_put_vf;
1619 
1620 	if (ice_vf_get_port_vlan_prio(vf) == qos &&
1621 	    ice_vf_get_port_vlan_tpid(vf) == local_vlan_proto &&
1622 	    ice_vf_get_port_vlan_id(vf) == vlan_id) {
1623 		/* duplicate request, so just return success */
1624 		dev_dbg(dev, "Duplicate port VLAN %u, QoS %u, TPID 0x%04x request\n",
1625 			vlan_id, qos, local_vlan_proto);
1626 		ret = 0;
1627 		goto out_put_vf;
1628 	}
1629 
1630 	mutex_lock(&vf->cfg_lock);
1631 
1632 	vf->port_vlan_info = ICE_VLAN(local_vlan_proto, vlan_id, qos);
1633 	if (ice_vf_is_port_vlan_ena(vf))
1634 		dev_info(dev, "Setting VLAN %u, QoS %u, TPID 0x%04x on VF %d\n",
1635 			 vlan_id, qos, local_vlan_proto, vf_id);
1636 	else
1637 		dev_info(dev, "Clearing port VLAN on VF %d\n", vf_id);
1638 
1639 	ice_reset_vf(vf, ICE_VF_RESET_NOTIFY);
1640 	mutex_unlock(&vf->cfg_lock);
1641 
1642 out_put_vf:
1643 	ice_put_vf(vf);
1644 	return ret;
1645 }
1646 
1647 /**
1648  * ice_print_vf_rx_mdd_event - print VF Rx malicious driver detect event
1649  * @vf: pointer to the VF structure
1650  */
1651 void ice_print_vf_rx_mdd_event(struct ice_vf *vf)
1652 {
1653 	struct ice_pf *pf = vf->pf;
1654 	struct device *dev;
1655 
1656 	dev = ice_pf_to_dev(pf);
1657 
1658 	dev_info(dev, "%d Rx Malicious Driver Detection events detected on PF %d VF %d MAC %pM. mdd-auto-reset-vfs=%s\n",
1659 		 vf->mdd_rx_events.count, pf->hw.pf_id, vf->vf_id,
1660 		 vf->dev_lan_addr,
1661 		 test_bit(ICE_FLAG_MDD_AUTO_RESET_VF, pf->flags)
1662 			  ? "on" : "off");
1663 }
1664 
1665 /**
1666  * ice_print_vfs_mdd_events - print VFs malicious driver detect event
1667  * @pf: pointer to the PF structure
1668  *
1669  * Called from ice_handle_mdd_event to rate limit and print VFs MDD events.
1670  */
1671 void ice_print_vfs_mdd_events(struct ice_pf *pf)
1672 {
1673 	struct device *dev = ice_pf_to_dev(pf);
1674 	struct ice_hw *hw = &pf->hw;
1675 	struct ice_vf *vf;
1676 	unsigned int bkt;
1677 
1678 	/* check that there are pending MDD events to print */
1679 	if (!test_and_clear_bit(ICE_MDD_VF_PRINT_PENDING, pf->state))
1680 		return;
1681 
1682 	/* VF MDD event logs are rate limited to one second intervals */
1683 	if (time_is_after_jiffies(pf->vfs.last_printed_mdd_jiffies + HZ * 1))
1684 		return;
1685 
1686 	pf->vfs.last_printed_mdd_jiffies = jiffies;
1687 
1688 	mutex_lock(&pf->vfs.table_lock);
1689 	ice_for_each_vf(pf, bkt, vf) {
1690 		/* only print Rx MDD event message if there are new events */
1691 		if (vf->mdd_rx_events.count != vf->mdd_rx_events.last_printed) {
1692 			vf->mdd_rx_events.last_printed =
1693 							vf->mdd_rx_events.count;
1694 			ice_print_vf_rx_mdd_event(vf);
1695 		}
1696 
1697 		/* only print Tx MDD event message if there are new events */
1698 		if (vf->mdd_tx_events.count != vf->mdd_tx_events.last_printed) {
1699 			vf->mdd_tx_events.last_printed =
1700 							vf->mdd_tx_events.count;
1701 
1702 			dev_info(dev, "%d Tx Malicious Driver Detection events detected on PF %d VF %d MAC %pM.\n",
1703 				 vf->mdd_tx_events.count, hw->pf_id, vf->vf_id,
1704 				 vf->dev_lan_addr);
1705 		}
1706 	}
1707 	mutex_unlock(&pf->vfs.table_lock);
1708 }
1709 
1710 /**
1711  * ice_restore_all_vfs_msi_state - restore VF MSI state after PF FLR
1712  * @pdev: pointer to a pci_dev structure
1713  *
1714  * Called when recovering from a PF FLR to restore interrupt capability to
1715  * the VFs.
1716  */
1717 void ice_restore_all_vfs_msi_state(struct pci_dev *pdev)
1718 {
1719 	u16 vf_id;
1720 	int pos;
1721 
1722 	if (!pci_num_vf(pdev))
1723 		return;
1724 
1725 	pos = pci_find_ext_capability(pdev, PCI_EXT_CAP_ID_SRIOV);
1726 	if (pos) {
1727 		struct pci_dev *vfdev;
1728 
1729 		pci_read_config_word(pdev, pos + PCI_SRIOV_VF_DID,
1730 				     &vf_id);
1731 		vfdev = pci_get_device(pdev->vendor, vf_id, NULL);
1732 		while (vfdev) {
1733 			if (vfdev->is_virtfn && vfdev->physfn == pdev)
1734 				pci_restore_msi_state(vfdev);
1735 			vfdev = pci_get_device(pdev->vendor, vf_id,
1736 					       vfdev);
1737 		}
1738 	}
1739 }
1740