xref: /linux/drivers/net/ethernet/intel/ice/ice_main.c (revision e85f99aa7760e74bb5a7e8515948f99c264a275f)
1 // SPDX-License-Identifier: GPL-2.0
2 /* Copyright (c) 2018, Intel Corporation. */
3 
4 /* Intel(R) Ethernet Connection E800 Series Linux Driver */
5 
6 #define pr_fmt(fmt) KBUILD_MODNAME ": " fmt
7 
8 #include <generated/utsrelease.h>
9 #include "ice.h"
10 #include "ice_base.h"
11 #include "ice_lib.h"
12 #include "ice_fltr.h"
13 #include "ice_dcb_lib.h"
14 #include "ice_dcb_nl.h"
15 #include "ice_devlink.h"
16 
17 #define DRV_SUMMARY	"Intel(R) Ethernet Connection E800 Series Linux Driver"
18 static const char ice_driver_string[] = DRV_SUMMARY;
19 static const char ice_copyright[] = "Copyright (c) 2018, Intel Corporation.";
20 
21 /* DDP Package file located in firmware search paths (e.g. /lib/firmware/) */
22 #define ICE_DDP_PKG_PATH	"intel/ice/ddp/"
23 #define ICE_DDP_PKG_FILE	ICE_DDP_PKG_PATH "ice.pkg"
24 
25 MODULE_AUTHOR("Intel Corporation, <linux.nics@intel.com>");
26 MODULE_DESCRIPTION(DRV_SUMMARY);
27 MODULE_LICENSE("GPL v2");
28 MODULE_FIRMWARE(ICE_DDP_PKG_FILE);
29 
30 static int debug = -1;
31 module_param(debug, int, 0644);
32 #ifndef CONFIG_DYNAMIC_DEBUG
33 MODULE_PARM_DESC(debug, "netif level (0=none,...,16=all), hw debug_mask (0x8XXXXXXX)");
34 #else
35 MODULE_PARM_DESC(debug, "netif level (0=none,...,16=all)");
36 #endif /* !CONFIG_DYNAMIC_DEBUG */
37 
38 static struct workqueue_struct *ice_wq;
39 static const struct net_device_ops ice_netdev_safe_mode_ops;
40 static const struct net_device_ops ice_netdev_ops;
41 static int ice_vsi_open(struct ice_vsi *vsi);
42 
43 static void ice_rebuild(struct ice_pf *pf, enum ice_reset_req reset_type);
44 
45 static void ice_vsi_release_all(struct ice_pf *pf);
46 
47 /**
48  * ice_get_tx_pending - returns number of Tx descriptors not processed
49  * @ring: the ring of descriptors
50  */
51 static u16 ice_get_tx_pending(struct ice_ring *ring)
52 {
53 	u16 head, tail;
54 
55 	head = ring->next_to_clean;
56 	tail = ring->next_to_use;
57 
58 	if (head != tail)
59 		return (head < tail) ?
60 			tail - head : (tail + ring->count - head);
61 	return 0;
62 }
63 
64 /**
65  * ice_check_for_hang_subtask - check for and recover hung queues
66  * @pf: pointer to PF struct
67  */
68 static void ice_check_for_hang_subtask(struct ice_pf *pf)
69 {
70 	struct ice_vsi *vsi = NULL;
71 	struct ice_hw *hw;
72 	unsigned int i;
73 	int packets;
74 	u32 v;
75 
76 	ice_for_each_vsi(pf, v)
77 		if (pf->vsi[v] && pf->vsi[v]->type == ICE_VSI_PF) {
78 			vsi = pf->vsi[v];
79 			break;
80 		}
81 
82 	if (!vsi || test_bit(__ICE_DOWN, vsi->state))
83 		return;
84 
85 	if (!(vsi->netdev && netif_carrier_ok(vsi->netdev)))
86 		return;
87 
88 	hw = &vsi->back->hw;
89 
90 	for (i = 0; i < vsi->num_txq; i++) {
91 		struct ice_ring *tx_ring = vsi->tx_rings[i];
92 
93 		if (tx_ring && tx_ring->desc) {
94 			/* If packet counter has not changed the queue is
95 			 * likely stalled, so force an interrupt for this
96 			 * queue.
97 			 *
98 			 * prev_pkt would be negative if there was no
99 			 * pending work.
100 			 */
101 			packets = tx_ring->stats.pkts & INT_MAX;
102 			if (tx_ring->tx_stats.prev_pkt == packets) {
103 				/* Trigger sw interrupt to revive the queue */
104 				ice_trigger_sw_intr(hw, tx_ring->q_vector);
105 				continue;
106 			}
107 
108 			/* Memory barrier between read of packet count and call
109 			 * to ice_get_tx_pending()
110 			 */
111 			smp_rmb();
112 			tx_ring->tx_stats.prev_pkt =
113 			    ice_get_tx_pending(tx_ring) ? packets : -1;
114 		}
115 	}
116 }
117 
118 /**
119  * ice_init_mac_fltr - Set initial MAC filters
120  * @pf: board private structure
121  *
122  * Set initial set of MAC filters for PF VSI; configure filters for permanent
123  * address and broadcast address. If an error is encountered, netdevice will be
124  * unregistered.
125  */
126 static int ice_init_mac_fltr(struct ice_pf *pf)
127 {
128 	enum ice_status status;
129 	struct ice_vsi *vsi;
130 	u8 *perm_addr;
131 
132 	vsi = ice_get_main_vsi(pf);
133 	if (!vsi)
134 		return -EINVAL;
135 
136 	perm_addr = vsi->port_info->mac.perm_addr;
137 	status = ice_fltr_add_mac_and_broadcast(vsi, perm_addr, ICE_FWD_TO_VSI);
138 	if (!status)
139 		return 0;
140 
141 	/* We aren't useful with no MAC filters, so unregister if we
142 	 * had an error
143 	 */
144 	if (vsi->netdev->reg_state == NETREG_REGISTERED) {
145 		dev_err(ice_pf_to_dev(pf), "Could not add MAC filters error %s. Unregistering device\n",
146 			ice_stat_str(status));
147 		unregister_netdev(vsi->netdev);
148 		free_netdev(vsi->netdev);
149 		vsi->netdev = NULL;
150 	}
151 
152 	return -EIO;
153 }
154 
155 /**
156  * ice_add_mac_to_sync_list - creates list of MAC addresses to be synced
157  * @netdev: the net device on which the sync is happening
158  * @addr: MAC address to sync
159  *
160  * This is a callback function which is called by the in kernel device sync
161  * functions (like __dev_uc_sync, __dev_mc_sync, etc). This function only
162  * populates the tmp_sync_list, which is later used by ice_add_mac to add the
163  * MAC filters from the hardware.
164  */
165 static int ice_add_mac_to_sync_list(struct net_device *netdev, const u8 *addr)
166 {
167 	struct ice_netdev_priv *np = netdev_priv(netdev);
168 	struct ice_vsi *vsi = np->vsi;
169 
170 	if (ice_fltr_add_mac_to_list(vsi, &vsi->tmp_sync_list, addr,
171 				     ICE_FWD_TO_VSI))
172 		return -EINVAL;
173 
174 	return 0;
175 }
176 
177 /**
178  * ice_add_mac_to_unsync_list - creates list of MAC addresses to be unsynced
179  * @netdev: the net device on which the unsync is happening
180  * @addr: MAC address to unsync
181  *
182  * This is a callback function which is called by the in kernel device unsync
183  * functions (like __dev_uc_unsync, __dev_mc_unsync, etc). This function only
184  * populates the tmp_unsync_list, which is later used by ice_remove_mac to
185  * delete the MAC filters from the hardware.
186  */
187 static int ice_add_mac_to_unsync_list(struct net_device *netdev, const u8 *addr)
188 {
189 	struct ice_netdev_priv *np = netdev_priv(netdev);
190 	struct ice_vsi *vsi = np->vsi;
191 
192 	if (ice_fltr_add_mac_to_list(vsi, &vsi->tmp_unsync_list, addr,
193 				     ICE_FWD_TO_VSI))
194 		return -EINVAL;
195 
196 	return 0;
197 }
198 
199 /**
200  * ice_vsi_fltr_changed - check if filter state changed
201  * @vsi: VSI to be checked
202  *
203  * returns true if filter state has changed, false otherwise.
204  */
205 static bool ice_vsi_fltr_changed(struct ice_vsi *vsi)
206 {
207 	return test_bit(ICE_VSI_FLAG_UMAC_FLTR_CHANGED, vsi->flags) ||
208 	       test_bit(ICE_VSI_FLAG_MMAC_FLTR_CHANGED, vsi->flags) ||
209 	       test_bit(ICE_VSI_FLAG_VLAN_FLTR_CHANGED, vsi->flags);
210 }
211 
212 /**
213  * ice_cfg_promisc - Enable or disable promiscuous mode for a given PF
214  * @vsi: the VSI being configured
215  * @promisc_m: mask of promiscuous config bits
216  * @set_promisc: enable or disable promisc flag request
217  *
218  */
219 static int ice_cfg_promisc(struct ice_vsi *vsi, u8 promisc_m, bool set_promisc)
220 {
221 	struct ice_hw *hw = &vsi->back->hw;
222 	enum ice_status status = 0;
223 
224 	if (vsi->type != ICE_VSI_PF)
225 		return 0;
226 
227 	if (vsi->vlan_ena) {
228 		status = ice_set_vlan_vsi_promisc(hw, vsi->idx, promisc_m,
229 						  set_promisc);
230 	} else {
231 		if (set_promisc)
232 			status = ice_set_vsi_promisc(hw, vsi->idx, promisc_m,
233 						     0);
234 		else
235 			status = ice_clear_vsi_promisc(hw, vsi->idx, promisc_m,
236 						       0);
237 	}
238 
239 	if (status)
240 		return -EIO;
241 
242 	return 0;
243 }
244 
245 /**
246  * ice_vsi_sync_fltr - Update the VSI filter list to the HW
247  * @vsi: ptr to the VSI
248  *
249  * Push any outstanding VSI filter changes through the AdminQ.
250  */
251 static int ice_vsi_sync_fltr(struct ice_vsi *vsi)
252 {
253 	struct device *dev = ice_pf_to_dev(vsi->back);
254 	struct net_device *netdev = vsi->netdev;
255 	bool promisc_forced_on = false;
256 	struct ice_pf *pf = vsi->back;
257 	struct ice_hw *hw = &pf->hw;
258 	enum ice_status status = 0;
259 	u32 changed_flags = 0;
260 	u8 promisc_m;
261 	int err = 0;
262 
263 	if (!vsi->netdev)
264 		return -EINVAL;
265 
266 	while (test_and_set_bit(__ICE_CFG_BUSY, vsi->state))
267 		usleep_range(1000, 2000);
268 
269 	changed_flags = vsi->current_netdev_flags ^ vsi->netdev->flags;
270 	vsi->current_netdev_flags = vsi->netdev->flags;
271 
272 	INIT_LIST_HEAD(&vsi->tmp_sync_list);
273 	INIT_LIST_HEAD(&vsi->tmp_unsync_list);
274 
275 	if (ice_vsi_fltr_changed(vsi)) {
276 		clear_bit(ICE_VSI_FLAG_UMAC_FLTR_CHANGED, vsi->flags);
277 		clear_bit(ICE_VSI_FLAG_MMAC_FLTR_CHANGED, vsi->flags);
278 		clear_bit(ICE_VSI_FLAG_VLAN_FLTR_CHANGED, vsi->flags);
279 
280 		/* grab the netdev's addr_list_lock */
281 		netif_addr_lock_bh(netdev);
282 		__dev_uc_sync(netdev, ice_add_mac_to_sync_list,
283 			      ice_add_mac_to_unsync_list);
284 		__dev_mc_sync(netdev, ice_add_mac_to_sync_list,
285 			      ice_add_mac_to_unsync_list);
286 		/* our temp lists are populated. release lock */
287 		netif_addr_unlock_bh(netdev);
288 	}
289 
290 	/* Remove MAC addresses in the unsync list */
291 	status = ice_fltr_remove_mac_list(vsi, &vsi->tmp_unsync_list);
292 	ice_fltr_free_list(dev, &vsi->tmp_unsync_list);
293 	if (status) {
294 		netdev_err(netdev, "Failed to delete MAC filters\n");
295 		/* if we failed because of alloc failures, just bail */
296 		if (status == ICE_ERR_NO_MEMORY) {
297 			err = -ENOMEM;
298 			goto out;
299 		}
300 	}
301 
302 	/* Add MAC addresses in the sync list */
303 	status = ice_fltr_add_mac_list(vsi, &vsi->tmp_sync_list);
304 	ice_fltr_free_list(dev, &vsi->tmp_sync_list);
305 	/* If filter is added successfully or already exists, do not go into
306 	 * 'if' condition and report it as error. Instead continue processing
307 	 * rest of the function.
308 	 */
309 	if (status && status != ICE_ERR_ALREADY_EXISTS) {
310 		netdev_err(netdev, "Failed to add MAC filters\n");
311 		/* If there is no more space for new umac filters, VSI
312 		 * should go into promiscuous mode. There should be some
313 		 * space reserved for promiscuous filters.
314 		 */
315 		if (hw->adminq.sq_last_status == ICE_AQ_RC_ENOSPC &&
316 		    !test_and_set_bit(__ICE_FLTR_OVERFLOW_PROMISC,
317 				      vsi->state)) {
318 			promisc_forced_on = true;
319 			netdev_warn(netdev, "Reached MAC filter limit, forcing promisc mode on VSI %d\n",
320 				    vsi->vsi_num);
321 		} else {
322 			err = -EIO;
323 			goto out;
324 		}
325 	}
326 	/* check for changes in promiscuous modes */
327 	if (changed_flags & IFF_ALLMULTI) {
328 		if (vsi->current_netdev_flags & IFF_ALLMULTI) {
329 			if (vsi->vlan_ena)
330 				promisc_m = ICE_MCAST_VLAN_PROMISC_BITS;
331 			else
332 				promisc_m = ICE_MCAST_PROMISC_BITS;
333 
334 			err = ice_cfg_promisc(vsi, promisc_m, true);
335 			if (err) {
336 				netdev_err(netdev, "Error setting Multicast promiscuous mode on VSI %i\n",
337 					   vsi->vsi_num);
338 				vsi->current_netdev_flags &= ~IFF_ALLMULTI;
339 				goto out_promisc;
340 			}
341 		} else {
342 			/* !(vsi->current_netdev_flags & IFF_ALLMULTI) */
343 			if (vsi->vlan_ena)
344 				promisc_m = ICE_MCAST_VLAN_PROMISC_BITS;
345 			else
346 				promisc_m = ICE_MCAST_PROMISC_BITS;
347 
348 			err = ice_cfg_promisc(vsi, promisc_m, false);
349 			if (err) {
350 				netdev_err(netdev, "Error clearing Multicast promiscuous mode on VSI %i\n",
351 					   vsi->vsi_num);
352 				vsi->current_netdev_flags |= IFF_ALLMULTI;
353 				goto out_promisc;
354 			}
355 		}
356 	}
357 
358 	if (((changed_flags & IFF_PROMISC) || promisc_forced_on) ||
359 	    test_bit(ICE_VSI_FLAG_PROMISC_CHANGED, vsi->flags)) {
360 		clear_bit(ICE_VSI_FLAG_PROMISC_CHANGED, vsi->flags);
361 		if (vsi->current_netdev_flags & IFF_PROMISC) {
362 			/* Apply Rx filter rule to get traffic from wire */
363 			if (!ice_is_dflt_vsi_in_use(pf->first_sw)) {
364 				err = ice_set_dflt_vsi(pf->first_sw, vsi);
365 				if (err && err != -EEXIST) {
366 					netdev_err(netdev, "Error %d setting default VSI %i Rx rule\n",
367 						   err, vsi->vsi_num);
368 					vsi->current_netdev_flags &=
369 						~IFF_PROMISC;
370 					goto out_promisc;
371 				}
372 			}
373 		} else {
374 			/* Clear Rx filter to remove traffic from wire */
375 			if (ice_is_vsi_dflt_vsi(pf->first_sw, vsi)) {
376 				err = ice_clear_dflt_vsi(pf->first_sw);
377 				if (err) {
378 					netdev_err(netdev, "Error %d clearing default VSI %i Rx rule\n",
379 						   err, vsi->vsi_num);
380 					vsi->current_netdev_flags |=
381 						IFF_PROMISC;
382 					goto out_promisc;
383 				}
384 			}
385 		}
386 	}
387 	goto exit;
388 
389 out_promisc:
390 	set_bit(ICE_VSI_FLAG_PROMISC_CHANGED, vsi->flags);
391 	goto exit;
392 out:
393 	/* if something went wrong then set the changed flag so we try again */
394 	set_bit(ICE_VSI_FLAG_UMAC_FLTR_CHANGED, vsi->flags);
395 	set_bit(ICE_VSI_FLAG_MMAC_FLTR_CHANGED, vsi->flags);
396 exit:
397 	clear_bit(__ICE_CFG_BUSY, vsi->state);
398 	return err;
399 }
400 
401 /**
402  * ice_sync_fltr_subtask - Sync the VSI filter list with HW
403  * @pf: board private structure
404  */
405 static void ice_sync_fltr_subtask(struct ice_pf *pf)
406 {
407 	int v;
408 
409 	if (!pf || !(test_bit(ICE_FLAG_FLTR_SYNC, pf->flags)))
410 		return;
411 
412 	clear_bit(ICE_FLAG_FLTR_SYNC, pf->flags);
413 
414 	ice_for_each_vsi(pf, v)
415 		if (pf->vsi[v] && ice_vsi_fltr_changed(pf->vsi[v]) &&
416 		    ice_vsi_sync_fltr(pf->vsi[v])) {
417 			/* come back and try again later */
418 			set_bit(ICE_FLAG_FLTR_SYNC, pf->flags);
419 			break;
420 		}
421 }
422 
423 /**
424  * ice_pf_dis_all_vsi - Pause all VSIs on a PF
425  * @pf: the PF
426  * @locked: is the rtnl_lock already held
427  */
428 static void ice_pf_dis_all_vsi(struct ice_pf *pf, bool locked)
429 {
430 	int v;
431 
432 	ice_for_each_vsi(pf, v)
433 		if (pf->vsi[v])
434 			ice_dis_vsi(pf->vsi[v], locked);
435 }
436 
437 /**
438  * ice_prepare_for_reset - prep for the core to reset
439  * @pf: board private structure
440  *
441  * Inform or close all dependent features in prep for reset.
442  */
443 static void
444 ice_prepare_for_reset(struct ice_pf *pf)
445 {
446 	struct ice_hw *hw = &pf->hw;
447 	unsigned int i;
448 
449 	/* already prepared for reset */
450 	if (test_bit(__ICE_PREPARED_FOR_RESET, pf->state))
451 		return;
452 
453 	/* Notify VFs of impending reset */
454 	if (ice_check_sq_alive(hw, &hw->mailboxq))
455 		ice_vc_notify_reset(pf);
456 
457 	/* Disable VFs until reset is completed */
458 	ice_for_each_vf(pf, i)
459 		ice_set_vf_state_qs_dis(&pf->vf[i]);
460 
461 	/* clear SW filtering DB */
462 	ice_clear_hw_tbls(hw);
463 	/* disable the VSIs and their queues that are not already DOWN */
464 	ice_pf_dis_all_vsi(pf, false);
465 
466 	if (hw->port_info)
467 		ice_sched_clear_port(hw->port_info);
468 
469 	ice_shutdown_all_ctrlq(hw);
470 
471 	set_bit(__ICE_PREPARED_FOR_RESET, pf->state);
472 }
473 
474 /**
475  * ice_do_reset - Initiate one of many types of resets
476  * @pf: board private structure
477  * @reset_type: reset type requested
478  * before this function was called.
479  */
480 static void ice_do_reset(struct ice_pf *pf, enum ice_reset_req reset_type)
481 {
482 	struct device *dev = ice_pf_to_dev(pf);
483 	struct ice_hw *hw = &pf->hw;
484 
485 	dev_dbg(dev, "reset_type 0x%x requested\n", reset_type);
486 	WARN_ON(in_interrupt());
487 
488 	ice_prepare_for_reset(pf);
489 
490 	/* trigger the reset */
491 	if (ice_reset(hw, reset_type)) {
492 		dev_err(dev, "reset %d failed\n", reset_type);
493 		set_bit(__ICE_RESET_FAILED, pf->state);
494 		clear_bit(__ICE_RESET_OICR_RECV, pf->state);
495 		clear_bit(__ICE_PREPARED_FOR_RESET, pf->state);
496 		clear_bit(__ICE_PFR_REQ, pf->state);
497 		clear_bit(__ICE_CORER_REQ, pf->state);
498 		clear_bit(__ICE_GLOBR_REQ, pf->state);
499 		return;
500 	}
501 
502 	/* PFR is a bit of a special case because it doesn't result in an OICR
503 	 * interrupt. So for PFR, rebuild after the reset and clear the reset-
504 	 * associated state bits.
505 	 */
506 	if (reset_type == ICE_RESET_PFR) {
507 		pf->pfr_count++;
508 		ice_rebuild(pf, reset_type);
509 		clear_bit(__ICE_PREPARED_FOR_RESET, pf->state);
510 		clear_bit(__ICE_PFR_REQ, pf->state);
511 		ice_reset_all_vfs(pf, true);
512 	}
513 }
514 
515 /**
516  * ice_reset_subtask - Set up for resetting the device and driver
517  * @pf: board private structure
518  */
519 static void ice_reset_subtask(struct ice_pf *pf)
520 {
521 	enum ice_reset_req reset_type = ICE_RESET_INVAL;
522 
523 	/* When a CORER/GLOBR/EMPR is about to happen, the hardware triggers an
524 	 * OICR interrupt. The OICR handler (ice_misc_intr) determines what type
525 	 * of reset is pending and sets bits in pf->state indicating the reset
526 	 * type and __ICE_RESET_OICR_RECV. So, if the latter bit is set
527 	 * prepare for pending reset if not already (for PF software-initiated
528 	 * global resets the software should already be prepared for it as
529 	 * indicated by __ICE_PREPARED_FOR_RESET; for global resets initiated
530 	 * by firmware or software on other PFs, that bit is not set so prepare
531 	 * for the reset now), poll for reset done, rebuild and return.
532 	 */
533 	if (test_bit(__ICE_RESET_OICR_RECV, pf->state)) {
534 		/* Perform the largest reset requested */
535 		if (test_and_clear_bit(__ICE_CORER_RECV, pf->state))
536 			reset_type = ICE_RESET_CORER;
537 		if (test_and_clear_bit(__ICE_GLOBR_RECV, pf->state))
538 			reset_type = ICE_RESET_GLOBR;
539 		if (test_and_clear_bit(__ICE_EMPR_RECV, pf->state))
540 			reset_type = ICE_RESET_EMPR;
541 		/* return if no valid reset type requested */
542 		if (reset_type == ICE_RESET_INVAL)
543 			return;
544 		ice_prepare_for_reset(pf);
545 
546 		/* make sure we are ready to rebuild */
547 		if (ice_check_reset(&pf->hw)) {
548 			set_bit(__ICE_RESET_FAILED, pf->state);
549 		} else {
550 			/* done with reset. start rebuild */
551 			pf->hw.reset_ongoing = false;
552 			ice_rebuild(pf, reset_type);
553 			/* clear bit to resume normal operations, but
554 			 * ICE_NEEDS_RESTART bit is set in case rebuild failed
555 			 */
556 			clear_bit(__ICE_RESET_OICR_RECV, pf->state);
557 			clear_bit(__ICE_PREPARED_FOR_RESET, pf->state);
558 			clear_bit(__ICE_PFR_REQ, pf->state);
559 			clear_bit(__ICE_CORER_REQ, pf->state);
560 			clear_bit(__ICE_GLOBR_REQ, pf->state);
561 			ice_reset_all_vfs(pf, true);
562 		}
563 
564 		return;
565 	}
566 
567 	/* No pending resets to finish processing. Check for new resets */
568 	if (test_bit(__ICE_PFR_REQ, pf->state))
569 		reset_type = ICE_RESET_PFR;
570 	if (test_bit(__ICE_CORER_REQ, pf->state))
571 		reset_type = ICE_RESET_CORER;
572 	if (test_bit(__ICE_GLOBR_REQ, pf->state))
573 		reset_type = ICE_RESET_GLOBR;
574 	/* If no valid reset type requested just return */
575 	if (reset_type == ICE_RESET_INVAL)
576 		return;
577 
578 	/* reset if not already down or busy */
579 	if (!test_bit(__ICE_DOWN, pf->state) &&
580 	    !test_bit(__ICE_CFG_BUSY, pf->state)) {
581 		ice_do_reset(pf, reset_type);
582 	}
583 }
584 
585 /**
586  * ice_print_topo_conflict - print topology conflict message
587  * @vsi: the VSI whose topology status is being checked
588  */
589 static void ice_print_topo_conflict(struct ice_vsi *vsi)
590 {
591 	switch (vsi->port_info->phy.link_info.topo_media_conflict) {
592 	case ICE_AQ_LINK_TOPO_CONFLICT:
593 	case ICE_AQ_LINK_MEDIA_CONFLICT:
594 	case ICE_AQ_LINK_TOPO_UNREACH_PRT:
595 	case ICE_AQ_LINK_TOPO_UNDRUTIL_PRT:
596 	case ICE_AQ_LINK_TOPO_UNDRUTIL_MEDIA:
597 		netdev_info(vsi->netdev, "Possible mis-configuration of the Ethernet port detected, please use the Intel(R) Ethernet Port Configuration Tool application to address the issue.\n");
598 		break;
599 	case ICE_AQ_LINK_TOPO_UNSUPP_MEDIA:
600 		netdev_info(vsi->netdev, "Rx/Tx is disabled on this device because an unsupported module type was detected. Refer to the Intel(R) Ethernet Adapters and Devices User Guide for a list of supported modules.\n");
601 		break;
602 	default:
603 		break;
604 	}
605 }
606 
607 /**
608  * ice_print_link_msg - print link up or down message
609  * @vsi: the VSI whose link status is being queried
610  * @isup: boolean for if the link is now up or down
611  */
612 void ice_print_link_msg(struct ice_vsi *vsi, bool isup)
613 {
614 	struct ice_aqc_get_phy_caps_data *caps;
615 	const char *an_advertised;
616 	enum ice_status status;
617 	const char *fec_req;
618 	const char *speed;
619 	const char *fec;
620 	const char *fc;
621 	const char *an;
622 
623 	if (!vsi)
624 		return;
625 
626 	if (vsi->current_isup == isup)
627 		return;
628 
629 	vsi->current_isup = isup;
630 
631 	if (!isup) {
632 		netdev_info(vsi->netdev, "NIC Link is Down\n");
633 		return;
634 	}
635 
636 	switch (vsi->port_info->phy.link_info.link_speed) {
637 	case ICE_AQ_LINK_SPEED_100GB:
638 		speed = "100 G";
639 		break;
640 	case ICE_AQ_LINK_SPEED_50GB:
641 		speed = "50 G";
642 		break;
643 	case ICE_AQ_LINK_SPEED_40GB:
644 		speed = "40 G";
645 		break;
646 	case ICE_AQ_LINK_SPEED_25GB:
647 		speed = "25 G";
648 		break;
649 	case ICE_AQ_LINK_SPEED_20GB:
650 		speed = "20 G";
651 		break;
652 	case ICE_AQ_LINK_SPEED_10GB:
653 		speed = "10 G";
654 		break;
655 	case ICE_AQ_LINK_SPEED_5GB:
656 		speed = "5 G";
657 		break;
658 	case ICE_AQ_LINK_SPEED_2500MB:
659 		speed = "2.5 G";
660 		break;
661 	case ICE_AQ_LINK_SPEED_1000MB:
662 		speed = "1 G";
663 		break;
664 	case ICE_AQ_LINK_SPEED_100MB:
665 		speed = "100 M";
666 		break;
667 	default:
668 		speed = "Unknown";
669 		break;
670 	}
671 
672 	switch (vsi->port_info->fc.current_mode) {
673 	case ICE_FC_FULL:
674 		fc = "Rx/Tx";
675 		break;
676 	case ICE_FC_TX_PAUSE:
677 		fc = "Tx";
678 		break;
679 	case ICE_FC_RX_PAUSE:
680 		fc = "Rx";
681 		break;
682 	case ICE_FC_NONE:
683 		fc = "None";
684 		break;
685 	default:
686 		fc = "Unknown";
687 		break;
688 	}
689 
690 	/* Get FEC mode based on negotiated link info */
691 	switch (vsi->port_info->phy.link_info.fec_info) {
692 	case ICE_AQ_LINK_25G_RS_528_FEC_EN:
693 	case ICE_AQ_LINK_25G_RS_544_FEC_EN:
694 		fec = "RS-FEC";
695 		break;
696 	case ICE_AQ_LINK_25G_KR_FEC_EN:
697 		fec = "FC-FEC/BASE-R";
698 		break;
699 	default:
700 		fec = "NONE";
701 		break;
702 	}
703 
704 	/* check if autoneg completed, might be false due to not supported */
705 	if (vsi->port_info->phy.link_info.an_info & ICE_AQ_AN_COMPLETED)
706 		an = "True";
707 	else
708 		an = "False";
709 
710 	/* Get FEC mode requested based on PHY caps last SW configuration */
711 	caps = kzalloc(sizeof(*caps), GFP_KERNEL);
712 	if (!caps) {
713 		fec_req = "Unknown";
714 		an_advertised = "Unknown";
715 		goto done;
716 	}
717 
718 	status = ice_aq_get_phy_caps(vsi->port_info, false,
719 				     ICE_AQC_REPORT_SW_CFG, caps, NULL);
720 	if (status)
721 		netdev_info(vsi->netdev, "Get phy capability failed.\n");
722 
723 	an_advertised = ice_is_phy_caps_an_enabled(caps) ? "On" : "Off";
724 
725 	if (caps->link_fec_options & ICE_AQC_PHY_FEC_25G_RS_528_REQ ||
726 	    caps->link_fec_options & ICE_AQC_PHY_FEC_25G_RS_544_REQ)
727 		fec_req = "RS-FEC";
728 	else if (caps->link_fec_options & ICE_AQC_PHY_FEC_10G_KR_40G_KR4_REQ ||
729 		 caps->link_fec_options & ICE_AQC_PHY_FEC_25G_KR_REQ)
730 		fec_req = "FC-FEC/BASE-R";
731 	else
732 		fec_req = "NONE";
733 
734 	kfree(caps);
735 
736 done:
737 	netdev_info(vsi->netdev, "NIC Link is up %sbps Full Duplex, Requested FEC: %s, Negotiated FEC: %s, Autoneg Advertised: %s, Autoneg Negotiated: %s, Flow Control: %s\n",
738 		    speed, fec_req, fec, an_advertised, an, fc);
739 	ice_print_topo_conflict(vsi);
740 }
741 
742 /**
743  * ice_vsi_link_event - update the VSI's netdev
744  * @vsi: the VSI on which the link event occurred
745  * @link_up: whether or not the VSI needs to be set up or down
746  */
747 static void ice_vsi_link_event(struct ice_vsi *vsi, bool link_up)
748 {
749 	if (!vsi)
750 		return;
751 
752 	if (test_bit(__ICE_DOWN, vsi->state) || !vsi->netdev)
753 		return;
754 
755 	if (vsi->type == ICE_VSI_PF) {
756 		if (link_up == netif_carrier_ok(vsi->netdev))
757 			return;
758 
759 		if (link_up) {
760 			netif_carrier_on(vsi->netdev);
761 			netif_tx_wake_all_queues(vsi->netdev);
762 		} else {
763 			netif_carrier_off(vsi->netdev);
764 			netif_tx_stop_all_queues(vsi->netdev);
765 		}
766 	}
767 }
768 
769 /**
770  * ice_link_event - process the link event
771  * @pf: PF that the link event is associated with
772  * @pi: port_info for the port that the link event is associated with
773  * @link_up: true if the physical link is up and false if it is down
774  * @link_speed: current link speed received from the link event
775  *
776  * Returns 0 on success and negative on failure
777  */
778 static int
779 ice_link_event(struct ice_pf *pf, struct ice_port_info *pi, bool link_up,
780 	       u16 link_speed)
781 {
782 	struct device *dev = ice_pf_to_dev(pf);
783 	struct ice_phy_info *phy_info;
784 	struct ice_vsi *vsi;
785 	u16 old_link_speed;
786 	bool old_link;
787 	int result;
788 
789 	phy_info = &pi->phy;
790 	phy_info->link_info_old = phy_info->link_info;
791 
792 	old_link = !!(phy_info->link_info_old.link_info & ICE_AQ_LINK_UP);
793 	old_link_speed = phy_info->link_info_old.link_speed;
794 
795 	/* update the link info structures and re-enable link events,
796 	 * don't bail on failure due to other book keeping needed
797 	 */
798 	result = ice_update_link_info(pi);
799 	if (result)
800 		dev_dbg(dev, "Failed to update link status and re-enable link events for port %d\n",
801 			pi->lport);
802 
803 	vsi = ice_get_main_vsi(pf);
804 	if (!vsi || !vsi->port_info)
805 		return -EINVAL;
806 
807 	/* turn off PHY if media was removed */
808 	if (!test_bit(ICE_FLAG_NO_MEDIA, pf->flags) &&
809 	    !(pi->phy.link_info.link_info & ICE_AQ_MEDIA_AVAILABLE)) {
810 		set_bit(ICE_FLAG_NO_MEDIA, pf->flags);
811 
812 		result = ice_aq_set_link_restart_an(pi, false, NULL);
813 		if (result) {
814 			dev_dbg(dev, "Failed to set link down, VSI %d error %d\n",
815 				vsi->vsi_num, result);
816 			return result;
817 		}
818 	}
819 
820 	/* if the old link up/down and speed is the same as the new */
821 	if (link_up == old_link && link_speed == old_link_speed)
822 		return result;
823 
824 	ice_dcb_rebuild(pf);
825 	ice_vsi_link_event(vsi, link_up);
826 	ice_print_link_msg(vsi, link_up);
827 
828 	ice_vc_notify_link_state(pf);
829 
830 	return result;
831 }
832 
833 /**
834  * ice_watchdog_subtask - periodic tasks not using event driven scheduling
835  * @pf: board private structure
836  */
837 static void ice_watchdog_subtask(struct ice_pf *pf)
838 {
839 	int i;
840 
841 	/* if interface is down do nothing */
842 	if (test_bit(__ICE_DOWN, pf->state) ||
843 	    test_bit(__ICE_CFG_BUSY, pf->state))
844 		return;
845 
846 	/* make sure we don't do these things too often */
847 	if (time_before(jiffies,
848 			pf->serv_tmr_prev + pf->serv_tmr_period))
849 		return;
850 
851 	pf->serv_tmr_prev = jiffies;
852 
853 	/* Update the stats for active netdevs so the network stack
854 	 * can look at updated numbers whenever it cares to
855 	 */
856 	ice_update_pf_stats(pf);
857 	ice_for_each_vsi(pf, i)
858 		if (pf->vsi[i] && pf->vsi[i]->netdev)
859 			ice_update_vsi_stats(pf->vsi[i]);
860 }
861 
862 /**
863  * ice_init_link_events - enable/initialize link events
864  * @pi: pointer to the port_info instance
865  *
866  * Returns -EIO on failure, 0 on success
867  */
868 static int ice_init_link_events(struct ice_port_info *pi)
869 {
870 	u16 mask;
871 
872 	mask = ~((u16)(ICE_AQ_LINK_EVENT_UPDOWN | ICE_AQ_LINK_EVENT_MEDIA_NA |
873 		       ICE_AQ_LINK_EVENT_MODULE_QUAL_FAIL));
874 
875 	if (ice_aq_set_event_mask(pi->hw, pi->lport, mask, NULL)) {
876 		dev_dbg(ice_hw_to_dev(pi->hw), "Failed to set link event mask for port %d\n",
877 			pi->lport);
878 		return -EIO;
879 	}
880 
881 	if (ice_aq_get_link_info(pi, true, NULL, NULL)) {
882 		dev_dbg(ice_hw_to_dev(pi->hw), "Failed to enable link events for port %d\n",
883 			pi->lport);
884 		return -EIO;
885 	}
886 
887 	return 0;
888 }
889 
890 /**
891  * ice_handle_link_event - handle link event via ARQ
892  * @pf: PF that the link event is associated with
893  * @event: event structure containing link status info
894  */
895 static int
896 ice_handle_link_event(struct ice_pf *pf, struct ice_rq_event_info *event)
897 {
898 	struct ice_aqc_get_link_status_data *link_data;
899 	struct ice_port_info *port_info;
900 	int status;
901 
902 	link_data = (struct ice_aqc_get_link_status_data *)event->msg_buf;
903 	port_info = pf->hw.port_info;
904 	if (!port_info)
905 		return -EINVAL;
906 
907 	status = ice_link_event(pf, port_info,
908 				!!(link_data->link_info & ICE_AQ_LINK_UP),
909 				le16_to_cpu(link_data->link_speed));
910 	if (status)
911 		dev_dbg(ice_pf_to_dev(pf), "Could not process link event, error %d\n",
912 			status);
913 
914 	return status;
915 }
916 
917 /**
918  * __ice_clean_ctrlq - helper function to clean controlq rings
919  * @pf: ptr to struct ice_pf
920  * @q_type: specific Control queue type
921  */
922 static int __ice_clean_ctrlq(struct ice_pf *pf, enum ice_ctl_q q_type)
923 {
924 	struct device *dev = ice_pf_to_dev(pf);
925 	struct ice_rq_event_info event;
926 	struct ice_hw *hw = &pf->hw;
927 	struct ice_ctl_q_info *cq;
928 	u16 pending, i = 0;
929 	const char *qtype;
930 	u32 oldval, val;
931 
932 	/* Do not clean control queue if/when PF reset fails */
933 	if (test_bit(__ICE_RESET_FAILED, pf->state))
934 		return 0;
935 
936 	switch (q_type) {
937 	case ICE_CTL_Q_ADMIN:
938 		cq = &hw->adminq;
939 		qtype = "Admin";
940 		break;
941 	case ICE_CTL_Q_MAILBOX:
942 		cq = &hw->mailboxq;
943 		qtype = "Mailbox";
944 		break;
945 	default:
946 		dev_warn(dev, "Unknown control queue type 0x%x\n", q_type);
947 		return 0;
948 	}
949 
950 	/* check for error indications - PF_xx_AxQLEN register layout for
951 	 * FW/MBX/SB are identical so just use defines for PF_FW_AxQLEN.
952 	 */
953 	val = rd32(hw, cq->rq.len);
954 	if (val & (PF_FW_ARQLEN_ARQVFE_M | PF_FW_ARQLEN_ARQOVFL_M |
955 		   PF_FW_ARQLEN_ARQCRIT_M)) {
956 		oldval = val;
957 		if (val & PF_FW_ARQLEN_ARQVFE_M)
958 			dev_dbg(dev, "%s Receive Queue VF Error detected\n",
959 				qtype);
960 		if (val & PF_FW_ARQLEN_ARQOVFL_M) {
961 			dev_dbg(dev, "%s Receive Queue Overflow Error detected\n",
962 				qtype);
963 		}
964 		if (val & PF_FW_ARQLEN_ARQCRIT_M)
965 			dev_dbg(dev, "%s Receive Queue Critical Error detected\n",
966 				qtype);
967 		val &= ~(PF_FW_ARQLEN_ARQVFE_M | PF_FW_ARQLEN_ARQOVFL_M |
968 			 PF_FW_ARQLEN_ARQCRIT_M);
969 		if (oldval != val)
970 			wr32(hw, cq->rq.len, val);
971 	}
972 
973 	val = rd32(hw, cq->sq.len);
974 	if (val & (PF_FW_ATQLEN_ATQVFE_M | PF_FW_ATQLEN_ATQOVFL_M |
975 		   PF_FW_ATQLEN_ATQCRIT_M)) {
976 		oldval = val;
977 		if (val & PF_FW_ATQLEN_ATQVFE_M)
978 			dev_dbg(dev, "%s Send Queue VF Error detected\n",
979 				qtype);
980 		if (val & PF_FW_ATQLEN_ATQOVFL_M) {
981 			dev_dbg(dev, "%s Send Queue Overflow Error detected\n",
982 				qtype);
983 		}
984 		if (val & PF_FW_ATQLEN_ATQCRIT_M)
985 			dev_dbg(dev, "%s Send Queue Critical Error detected\n",
986 				qtype);
987 		val &= ~(PF_FW_ATQLEN_ATQVFE_M | PF_FW_ATQLEN_ATQOVFL_M |
988 			 PF_FW_ATQLEN_ATQCRIT_M);
989 		if (oldval != val)
990 			wr32(hw, cq->sq.len, val);
991 	}
992 
993 	event.buf_len = cq->rq_buf_size;
994 	event.msg_buf = kzalloc(event.buf_len, GFP_KERNEL);
995 	if (!event.msg_buf)
996 		return 0;
997 
998 	do {
999 		enum ice_status ret;
1000 		u16 opcode;
1001 
1002 		ret = ice_clean_rq_elem(hw, cq, &event, &pending);
1003 		if (ret == ICE_ERR_AQ_NO_WORK)
1004 			break;
1005 		if (ret) {
1006 			dev_err(dev, "%s Receive Queue event error %s\n", qtype,
1007 				ice_stat_str(ret));
1008 			break;
1009 		}
1010 
1011 		opcode = le16_to_cpu(event.desc.opcode);
1012 
1013 		switch (opcode) {
1014 		case ice_aqc_opc_get_link_status:
1015 			if (ice_handle_link_event(pf, &event))
1016 				dev_err(dev, "Could not handle link event\n");
1017 			break;
1018 		case ice_aqc_opc_event_lan_overflow:
1019 			ice_vf_lan_overflow_event(pf, &event);
1020 			break;
1021 		case ice_mbx_opc_send_msg_to_pf:
1022 			ice_vc_process_vf_msg(pf, &event);
1023 			break;
1024 		case ice_aqc_opc_fw_logging:
1025 			ice_output_fw_log(hw, &event.desc, event.msg_buf);
1026 			break;
1027 		case ice_aqc_opc_lldp_set_mib_change:
1028 			ice_dcb_process_lldp_set_mib_change(pf, &event);
1029 			break;
1030 		default:
1031 			dev_dbg(dev, "%s Receive Queue unknown event 0x%04x ignored\n",
1032 				qtype, opcode);
1033 			break;
1034 		}
1035 	} while (pending && (i++ < ICE_DFLT_IRQ_WORK));
1036 
1037 	kfree(event.msg_buf);
1038 
1039 	return pending && (i == ICE_DFLT_IRQ_WORK);
1040 }
1041 
1042 /**
1043  * ice_ctrlq_pending - check if there is a difference between ntc and ntu
1044  * @hw: pointer to hardware info
1045  * @cq: control queue information
1046  *
1047  * returns true if there are pending messages in a queue, false if there aren't
1048  */
1049 static bool ice_ctrlq_pending(struct ice_hw *hw, struct ice_ctl_q_info *cq)
1050 {
1051 	u16 ntu;
1052 
1053 	ntu = (u16)(rd32(hw, cq->rq.head) & cq->rq.head_mask);
1054 	return cq->rq.next_to_clean != ntu;
1055 }
1056 
1057 /**
1058  * ice_clean_adminq_subtask - clean the AdminQ rings
1059  * @pf: board private structure
1060  */
1061 static void ice_clean_adminq_subtask(struct ice_pf *pf)
1062 {
1063 	struct ice_hw *hw = &pf->hw;
1064 
1065 	if (!test_bit(__ICE_ADMINQ_EVENT_PENDING, pf->state))
1066 		return;
1067 
1068 	if (__ice_clean_ctrlq(pf, ICE_CTL_Q_ADMIN))
1069 		return;
1070 
1071 	clear_bit(__ICE_ADMINQ_EVENT_PENDING, pf->state);
1072 
1073 	/* There might be a situation where new messages arrive to a control
1074 	 * queue between processing the last message and clearing the
1075 	 * EVENT_PENDING bit. So before exiting, check queue head again (using
1076 	 * ice_ctrlq_pending) and process new messages if any.
1077 	 */
1078 	if (ice_ctrlq_pending(hw, &hw->adminq))
1079 		__ice_clean_ctrlq(pf, ICE_CTL_Q_ADMIN);
1080 
1081 	ice_flush(hw);
1082 }
1083 
1084 /**
1085  * ice_clean_mailboxq_subtask - clean the MailboxQ rings
1086  * @pf: board private structure
1087  */
1088 static void ice_clean_mailboxq_subtask(struct ice_pf *pf)
1089 {
1090 	struct ice_hw *hw = &pf->hw;
1091 
1092 	if (!test_bit(__ICE_MAILBOXQ_EVENT_PENDING, pf->state))
1093 		return;
1094 
1095 	if (__ice_clean_ctrlq(pf, ICE_CTL_Q_MAILBOX))
1096 		return;
1097 
1098 	clear_bit(__ICE_MAILBOXQ_EVENT_PENDING, pf->state);
1099 
1100 	if (ice_ctrlq_pending(hw, &hw->mailboxq))
1101 		__ice_clean_ctrlq(pf, ICE_CTL_Q_MAILBOX);
1102 
1103 	ice_flush(hw);
1104 }
1105 
1106 /**
1107  * ice_service_task_schedule - schedule the service task to wake up
1108  * @pf: board private structure
1109  *
1110  * If not already scheduled, this puts the task into the work queue.
1111  */
1112 void ice_service_task_schedule(struct ice_pf *pf)
1113 {
1114 	if (!test_bit(__ICE_SERVICE_DIS, pf->state) &&
1115 	    !test_and_set_bit(__ICE_SERVICE_SCHED, pf->state) &&
1116 	    !test_bit(__ICE_NEEDS_RESTART, pf->state))
1117 		queue_work(ice_wq, &pf->serv_task);
1118 }
1119 
1120 /**
1121  * ice_service_task_complete - finish up the service task
1122  * @pf: board private structure
1123  */
1124 static void ice_service_task_complete(struct ice_pf *pf)
1125 {
1126 	WARN_ON(!test_bit(__ICE_SERVICE_SCHED, pf->state));
1127 
1128 	/* force memory (pf->state) to sync before next service task */
1129 	smp_mb__before_atomic();
1130 	clear_bit(__ICE_SERVICE_SCHED, pf->state);
1131 }
1132 
1133 /**
1134  * ice_service_task_stop - stop service task and cancel works
1135  * @pf: board private structure
1136  *
1137  * Return 0 if the __ICE_SERVICE_DIS bit was not already set,
1138  * 1 otherwise.
1139  */
1140 static int ice_service_task_stop(struct ice_pf *pf)
1141 {
1142 	int ret;
1143 
1144 	ret = test_and_set_bit(__ICE_SERVICE_DIS, pf->state);
1145 
1146 	if (pf->serv_tmr.function)
1147 		del_timer_sync(&pf->serv_tmr);
1148 	if (pf->serv_task.func)
1149 		cancel_work_sync(&pf->serv_task);
1150 
1151 	clear_bit(__ICE_SERVICE_SCHED, pf->state);
1152 	return ret;
1153 }
1154 
1155 /**
1156  * ice_service_task_restart - restart service task and schedule works
1157  * @pf: board private structure
1158  *
1159  * This function is needed for suspend and resume works (e.g WoL scenario)
1160  */
1161 static void ice_service_task_restart(struct ice_pf *pf)
1162 {
1163 	clear_bit(__ICE_SERVICE_DIS, pf->state);
1164 	ice_service_task_schedule(pf);
1165 }
1166 
1167 /**
1168  * ice_service_timer - timer callback to schedule service task
1169  * @t: pointer to timer_list
1170  */
1171 static void ice_service_timer(struct timer_list *t)
1172 {
1173 	struct ice_pf *pf = from_timer(pf, t, serv_tmr);
1174 
1175 	mod_timer(&pf->serv_tmr, round_jiffies(pf->serv_tmr_period + jiffies));
1176 	ice_service_task_schedule(pf);
1177 }
1178 
1179 /**
1180  * ice_handle_mdd_event - handle malicious driver detect event
1181  * @pf: pointer to the PF structure
1182  *
1183  * Called from service task. OICR interrupt handler indicates MDD event.
1184  * VF MDD logging is guarded by net_ratelimit. Additional PF and VF log
1185  * messages are wrapped by netif_msg_[rx|tx]_err. Since VF Rx MDD events
1186  * disable the queue, the PF can be configured to reset the VF using ethtool
1187  * private flag mdd-auto-reset-vf.
1188  */
1189 static void ice_handle_mdd_event(struct ice_pf *pf)
1190 {
1191 	struct device *dev = ice_pf_to_dev(pf);
1192 	struct ice_hw *hw = &pf->hw;
1193 	unsigned int i;
1194 	u32 reg;
1195 
1196 	if (!test_and_clear_bit(__ICE_MDD_EVENT_PENDING, pf->state)) {
1197 		/* Since the VF MDD event logging is rate limited, check if
1198 		 * there are pending MDD events.
1199 		 */
1200 		ice_print_vfs_mdd_events(pf);
1201 		return;
1202 	}
1203 
1204 	/* find what triggered an MDD event */
1205 	reg = rd32(hw, GL_MDET_TX_PQM);
1206 	if (reg & GL_MDET_TX_PQM_VALID_M) {
1207 		u8 pf_num = (reg & GL_MDET_TX_PQM_PF_NUM_M) >>
1208 				GL_MDET_TX_PQM_PF_NUM_S;
1209 		u16 vf_num = (reg & GL_MDET_TX_PQM_VF_NUM_M) >>
1210 				GL_MDET_TX_PQM_VF_NUM_S;
1211 		u8 event = (reg & GL_MDET_TX_PQM_MAL_TYPE_M) >>
1212 				GL_MDET_TX_PQM_MAL_TYPE_S;
1213 		u16 queue = ((reg & GL_MDET_TX_PQM_QNUM_M) >>
1214 				GL_MDET_TX_PQM_QNUM_S);
1215 
1216 		if (netif_msg_tx_err(pf))
1217 			dev_info(dev, "Malicious Driver Detection event %d on TX queue %d PF# %d VF# %d\n",
1218 				 event, queue, pf_num, vf_num);
1219 		wr32(hw, GL_MDET_TX_PQM, 0xffffffff);
1220 	}
1221 
1222 	reg = rd32(hw, GL_MDET_TX_TCLAN);
1223 	if (reg & GL_MDET_TX_TCLAN_VALID_M) {
1224 		u8 pf_num = (reg & GL_MDET_TX_TCLAN_PF_NUM_M) >>
1225 				GL_MDET_TX_TCLAN_PF_NUM_S;
1226 		u16 vf_num = (reg & GL_MDET_TX_TCLAN_VF_NUM_M) >>
1227 				GL_MDET_TX_TCLAN_VF_NUM_S;
1228 		u8 event = (reg & GL_MDET_TX_TCLAN_MAL_TYPE_M) >>
1229 				GL_MDET_TX_TCLAN_MAL_TYPE_S;
1230 		u16 queue = ((reg & GL_MDET_TX_TCLAN_QNUM_M) >>
1231 				GL_MDET_TX_TCLAN_QNUM_S);
1232 
1233 		if (netif_msg_tx_err(pf))
1234 			dev_info(dev, "Malicious Driver Detection event %d on TX queue %d PF# %d VF# %d\n",
1235 				 event, queue, pf_num, vf_num);
1236 		wr32(hw, GL_MDET_TX_TCLAN, 0xffffffff);
1237 	}
1238 
1239 	reg = rd32(hw, GL_MDET_RX);
1240 	if (reg & GL_MDET_RX_VALID_M) {
1241 		u8 pf_num = (reg & GL_MDET_RX_PF_NUM_M) >>
1242 				GL_MDET_RX_PF_NUM_S;
1243 		u16 vf_num = (reg & GL_MDET_RX_VF_NUM_M) >>
1244 				GL_MDET_RX_VF_NUM_S;
1245 		u8 event = (reg & GL_MDET_RX_MAL_TYPE_M) >>
1246 				GL_MDET_RX_MAL_TYPE_S;
1247 		u16 queue = ((reg & GL_MDET_RX_QNUM_M) >>
1248 				GL_MDET_RX_QNUM_S);
1249 
1250 		if (netif_msg_rx_err(pf))
1251 			dev_info(dev, "Malicious Driver Detection event %d on RX queue %d PF# %d VF# %d\n",
1252 				 event, queue, pf_num, vf_num);
1253 		wr32(hw, GL_MDET_RX, 0xffffffff);
1254 	}
1255 
1256 	/* check to see if this PF caused an MDD event */
1257 	reg = rd32(hw, PF_MDET_TX_PQM);
1258 	if (reg & PF_MDET_TX_PQM_VALID_M) {
1259 		wr32(hw, PF_MDET_TX_PQM, 0xFFFF);
1260 		if (netif_msg_tx_err(pf))
1261 			dev_info(dev, "Malicious Driver Detection event TX_PQM detected on PF\n");
1262 	}
1263 
1264 	reg = rd32(hw, PF_MDET_TX_TCLAN);
1265 	if (reg & PF_MDET_TX_TCLAN_VALID_M) {
1266 		wr32(hw, PF_MDET_TX_TCLAN, 0xFFFF);
1267 		if (netif_msg_tx_err(pf))
1268 			dev_info(dev, "Malicious Driver Detection event TX_TCLAN detected on PF\n");
1269 	}
1270 
1271 	reg = rd32(hw, PF_MDET_RX);
1272 	if (reg & PF_MDET_RX_VALID_M) {
1273 		wr32(hw, PF_MDET_RX, 0xFFFF);
1274 		if (netif_msg_rx_err(pf))
1275 			dev_info(dev, "Malicious Driver Detection event RX detected on PF\n");
1276 	}
1277 
1278 	/* Check to see if one of the VFs caused an MDD event, and then
1279 	 * increment counters and set print pending
1280 	 */
1281 	ice_for_each_vf(pf, i) {
1282 		struct ice_vf *vf = &pf->vf[i];
1283 
1284 		reg = rd32(hw, VP_MDET_TX_PQM(i));
1285 		if (reg & VP_MDET_TX_PQM_VALID_M) {
1286 			wr32(hw, VP_MDET_TX_PQM(i), 0xFFFF);
1287 			vf->mdd_tx_events.count++;
1288 			set_bit(__ICE_MDD_VF_PRINT_PENDING, pf->state);
1289 			if (netif_msg_tx_err(pf))
1290 				dev_info(dev, "Malicious Driver Detection event TX_PQM detected on VF %d\n",
1291 					 i);
1292 		}
1293 
1294 		reg = rd32(hw, VP_MDET_TX_TCLAN(i));
1295 		if (reg & VP_MDET_TX_TCLAN_VALID_M) {
1296 			wr32(hw, VP_MDET_TX_TCLAN(i), 0xFFFF);
1297 			vf->mdd_tx_events.count++;
1298 			set_bit(__ICE_MDD_VF_PRINT_PENDING, pf->state);
1299 			if (netif_msg_tx_err(pf))
1300 				dev_info(dev, "Malicious Driver Detection event TX_TCLAN detected on VF %d\n",
1301 					 i);
1302 		}
1303 
1304 		reg = rd32(hw, VP_MDET_TX_TDPU(i));
1305 		if (reg & VP_MDET_TX_TDPU_VALID_M) {
1306 			wr32(hw, VP_MDET_TX_TDPU(i), 0xFFFF);
1307 			vf->mdd_tx_events.count++;
1308 			set_bit(__ICE_MDD_VF_PRINT_PENDING, pf->state);
1309 			if (netif_msg_tx_err(pf))
1310 				dev_info(dev, "Malicious Driver Detection event TX_TDPU detected on VF %d\n",
1311 					 i);
1312 		}
1313 
1314 		reg = rd32(hw, VP_MDET_RX(i));
1315 		if (reg & VP_MDET_RX_VALID_M) {
1316 			wr32(hw, VP_MDET_RX(i), 0xFFFF);
1317 			vf->mdd_rx_events.count++;
1318 			set_bit(__ICE_MDD_VF_PRINT_PENDING, pf->state);
1319 			if (netif_msg_rx_err(pf))
1320 				dev_info(dev, "Malicious Driver Detection event RX detected on VF %d\n",
1321 					 i);
1322 
1323 			/* Since the queue is disabled on VF Rx MDD events, the
1324 			 * PF can be configured to reset the VF through ethtool
1325 			 * private flag mdd-auto-reset-vf.
1326 			 */
1327 			if (test_bit(ICE_FLAG_MDD_AUTO_RESET_VF, pf->flags)) {
1328 				/* VF MDD event counters will be cleared by
1329 				 * reset, so print the event prior to reset.
1330 				 */
1331 				ice_print_vf_rx_mdd_event(vf);
1332 				ice_reset_vf(&pf->vf[i], false);
1333 			}
1334 		}
1335 	}
1336 
1337 	ice_print_vfs_mdd_events(pf);
1338 }
1339 
1340 /**
1341  * ice_force_phys_link_state - Force the physical link state
1342  * @vsi: VSI to force the physical link state to up/down
1343  * @link_up: true/false indicates to set the physical link to up/down
1344  *
1345  * Force the physical link state by getting the current PHY capabilities from
1346  * hardware and setting the PHY config based on the determined capabilities. If
1347  * link changes a link event will be triggered because both the Enable Automatic
1348  * Link Update and LESM Enable bits are set when setting the PHY capabilities.
1349  *
1350  * Returns 0 on success, negative on failure
1351  */
1352 static int ice_force_phys_link_state(struct ice_vsi *vsi, bool link_up)
1353 {
1354 	struct ice_aqc_get_phy_caps_data *pcaps;
1355 	struct ice_aqc_set_phy_cfg_data *cfg;
1356 	struct ice_port_info *pi;
1357 	struct device *dev;
1358 	int retcode;
1359 
1360 	if (!vsi || !vsi->port_info || !vsi->back)
1361 		return -EINVAL;
1362 	if (vsi->type != ICE_VSI_PF)
1363 		return 0;
1364 
1365 	dev = ice_pf_to_dev(vsi->back);
1366 
1367 	pi = vsi->port_info;
1368 
1369 	pcaps = kzalloc(sizeof(*pcaps), GFP_KERNEL);
1370 	if (!pcaps)
1371 		return -ENOMEM;
1372 
1373 	retcode = ice_aq_get_phy_caps(pi, false, ICE_AQC_REPORT_SW_CFG, pcaps,
1374 				      NULL);
1375 	if (retcode) {
1376 		dev_err(dev, "Failed to get phy capabilities, VSI %d error %d\n",
1377 			vsi->vsi_num, retcode);
1378 		retcode = -EIO;
1379 		goto out;
1380 	}
1381 
1382 	/* No change in link */
1383 	if (link_up == !!(pcaps->caps & ICE_AQC_PHY_EN_LINK) &&
1384 	    link_up == !!(pi->phy.link_info.link_info & ICE_AQ_LINK_UP))
1385 		goto out;
1386 
1387 	/* Use the current user PHY configuration. The current user PHY
1388 	 * configuration is initialized during probe from PHY capabilities
1389 	 * software mode, and updated on set PHY configuration.
1390 	 */
1391 	cfg = kmemdup(&pi->phy.curr_user_phy_cfg, sizeof(*cfg), GFP_KERNEL);
1392 	if (!cfg) {
1393 		retcode = -ENOMEM;
1394 		goto out;
1395 	}
1396 
1397 	cfg->caps |= ICE_AQ_PHY_ENA_AUTO_LINK_UPDT;
1398 	if (link_up)
1399 		cfg->caps |= ICE_AQ_PHY_ENA_LINK;
1400 	else
1401 		cfg->caps &= ~ICE_AQ_PHY_ENA_LINK;
1402 
1403 	retcode = ice_aq_set_phy_cfg(&vsi->back->hw, pi, cfg, NULL);
1404 	if (retcode) {
1405 		dev_err(dev, "Failed to set phy config, VSI %d error %d\n",
1406 			vsi->vsi_num, retcode);
1407 		retcode = -EIO;
1408 	}
1409 
1410 	kfree(cfg);
1411 out:
1412 	kfree(pcaps);
1413 	return retcode;
1414 }
1415 
1416 /**
1417  * ice_init_nvm_phy_type - Initialize the NVM PHY type
1418  * @pi: port info structure
1419  *
1420  * Initialize nvm_phy_type_[low|high] for link lenient mode support
1421  */
1422 static int ice_init_nvm_phy_type(struct ice_port_info *pi)
1423 {
1424 	struct ice_aqc_get_phy_caps_data *pcaps;
1425 	struct ice_pf *pf = pi->hw->back;
1426 	enum ice_status status;
1427 	int err = 0;
1428 
1429 	pcaps = kzalloc(sizeof(*pcaps), GFP_KERNEL);
1430 	if (!pcaps)
1431 		return -ENOMEM;
1432 
1433 	status = ice_aq_get_phy_caps(pi, false, ICE_AQC_REPORT_NVM_CAP, pcaps,
1434 				     NULL);
1435 
1436 	if (status) {
1437 		dev_err(ice_pf_to_dev(pf), "Get PHY capability failed.\n");
1438 		err = -EIO;
1439 		goto out;
1440 	}
1441 
1442 	pf->nvm_phy_type_hi = pcaps->phy_type_high;
1443 	pf->nvm_phy_type_lo = pcaps->phy_type_low;
1444 
1445 out:
1446 	kfree(pcaps);
1447 	return err;
1448 }
1449 
1450 /**
1451  * ice_init_link_dflt_override - Initialize link default override
1452  * @pi: port info structure
1453  *
1454  * Initialize link default override and PHY total port shutdown during probe
1455  */
1456 static void ice_init_link_dflt_override(struct ice_port_info *pi)
1457 {
1458 	struct ice_link_default_override_tlv *ldo;
1459 	struct ice_pf *pf = pi->hw->back;
1460 
1461 	ldo = &pf->link_dflt_override;
1462 	if (ice_get_link_default_override(ldo, pi))
1463 		return;
1464 
1465 	if (!(ldo->options & ICE_LINK_OVERRIDE_PORT_DIS))
1466 		return;
1467 
1468 	/* Enable Total Port Shutdown (override/replace link-down-on-close
1469 	 * ethtool private flag) for ports with Port Disable bit set.
1470 	 */
1471 	set_bit(ICE_FLAG_TOTAL_PORT_SHUTDOWN_ENA, pf->flags);
1472 	set_bit(ICE_FLAG_LINK_DOWN_ON_CLOSE_ENA, pf->flags);
1473 }
1474 
1475 /**
1476  * ice_init_phy_cfg_dflt_override - Initialize PHY cfg default override settings
1477  * @pi: port info structure
1478  *
1479  * If default override is enabled, initialized the user PHY cfg speed and FEC
1480  * settings using the default override mask from the NVM.
1481  *
1482  * The PHY should only be configured with the default override settings the
1483  * first time media is available. The __ICE_LINK_DEFAULT_OVERRIDE_PENDING state
1484  * is used to indicate that the user PHY cfg default override is initialized
1485  * and the PHY has not been configured with the default override settings. The
1486  * state is set here, and cleared in ice_configure_phy the first time the PHY is
1487  * configured.
1488  */
1489 static void ice_init_phy_cfg_dflt_override(struct ice_port_info *pi)
1490 {
1491 	struct ice_link_default_override_tlv *ldo;
1492 	struct ice_aqc_set_phy_cfg_data *cfg;
1493 	struct ice_phy_info *phy = &pi->phy;
1494 	struct ice_pf *pf = pi->hw->back;
1495 
1496 	ldo = &pf->link_dflt_override;
1497 
1498 	/* If link default override is enabled, use to mask NVM PHY capabilities
1499 	 * for speed and FEC default configuration.
1500 	 */
1501 	cfg = &phy->curr_user_phy_cfg;
1502 
1503 	if (ldo->phy_type_low || ldo->phy_type_high) {
1504 		cfg->phy_type_low = pf->nvm_phy_type_lo &
1505 				    cpu_to_le64(ldo->phy_type_low);
1506 		cfg->phy_type_high = pf->nvm_phy_type_hi &
1507 				     cpu_to_le64(ldo->phy_type_high);
1508 	}
1509 	cfg->link_fec_opt = ldo->fec_options;
1510 	phy->curr_user_fec_req = ICE_FEC_AUTO;
1511 
1512 	set_bit(__ICE_LINK_DEFAULT_OVERRIDE_PENDING, pf->state);
1513 }
1514 
1515 /**
1516  * ice_init_phy_user_cfg - Initialize the PHY user configuration
1517  * @pi: port info structure
1518  *
1519  * Initialize the current user PHY configuration, speed, FEC, and FC requested
1520  * mode to default. The PHY defaults are from get PHY capabilities topology
1521  * with media so call when media is first available. An error is returned if
1522  * called when media is not available. The PHY initialization completed state is
1523  * set here.
1524  *
1525  * These configurations are used when setting PHY
1526  * configuration. The user PHY configuration is updated on set PHY
1527  * configuration. Returns 0 on success, negative on failure
1528  */
1529 static int ice_init_phy_user_cfg(struct ice_port_info *pi)
1530 {
1531 	struct ice_aqc_get_phy_caps_data *pcaps;
1532 	struct ice_phy_info *phy = &pi->phy;
1533 	struct ice_pf *pf = pi->hw->back;
1534 	enum ice_status status;
1535 	struct ice_vsi *vsi;
1536 	int err = 0;
1537 
1538 	if (!(phy->link_info.link_info & ICE_AQ_MEDIA_AVAILABLE))
1539 		return -EIO;
1540 
1541 	vsi = ice_get_main_vsi(pf);
1542 	if (!vsi)
1543 		return -EINVAL;
1544 
1545 	pcaps = kzalloc(sizeof(*pcaps), GFP_KERNEL);
1546 	if (!pcaps)
1547 		return -ENOMEM;
1548 
1549 	status = ice_aq_get_phy_caps(pi, false, ICE_AQC_REPORT_TOPO_CAP, pcaps,
1550 				     NULL);
1551 	if (status) {
1552 		dev_err(ice_pf_to_dev(pf), "Get PHY capability failed.\n");
1553 		err = -EIO;
1554 		goto err_out;
1555 	}
1556 
1557 	ice_copy_phy_caps_to_cfg(pi, pcaps, &pi->phy.curr_user_phy_cfg);
1558 
1559 	/* check if lenient mode is supported and enabled */
1560 	if (ice_fw_supports_link_override(&vsi->back->hw) &&
1561 	    !(pcaps->module_compliance_enforcement &
1562 	      ICE_AQC_MOD_ENFORCE_STRICT_MODE)) {
1563 		set_bit(ICE_FLAG_LINK_LENIENT_MODE_ENA, pf->flags);
1564 
1565 		/* if link default override is enabled, initialize user PHY
1566 		 * configuration with link default override values
1567 		 */
1568 		if (pf->link_dflt_override.options & ICE_LINK_OVERRIDE_EN) {
1569 			ice_init_phy_cfg_dflt_override(pi);
1570 			goto out;
1571 		}
1572 	}
1573 
1574 	/* if link default override is not enabled, initialize PHY using
1575 	 * topology with media
1576 	 */
1577 	phy->curr_user_fec_req = ice_caps_to_fec_mode(pcaps->caps,
1578 						      pcaps->link_fec_options);
1579 	phy->curr_user_fc_req = ice_caps_to_fc_mode(pcaps->caps);
1580 
1581 out:
1582 	phy->curr_user_speed_req = ICE_AQ_LINK_SPEED_M;
1583 	set_bit(__ICE_PHY_INIT_COMPLETE, pf->state);
1584 err_out:
1585 	kfree(pcaps);
1586 	return err;
1587 }
1588 
1589 /**
1590  * ice_configure_phy - configure PHY
1591  * @vsi: VSI of PHY
1592  *
1593  * Set the PHY configuration. If the current PHY configuration is the same as
1594  * the curr_user_phy_cfg, then do nothing to avoid link flap. Otherwise
1595  * configure the based get PHY capabilities for topology with media.
1596  */
1597 static int ice_configure_phy(struct ice_vsi *vsi)
1598 {
1599 	struct device *dev = ice_pf_to_dev(vsi->back);
1600 	struct ice_aqc_get_phy_caps_data *pcaps;
1601 	struct ice_aqc_set_phy_cfg_data *cfg;
1602 	struct ice_port_info *pi;
1603 	enum ice_status status;
1604 	int err = 0;
1605 
1606 	pi = vsi->port_info;
1607 	if (!pi)
1608 		return -EINVAL;
1609 
1610 	/* Ensure we have media as we cannot configure a medialess port */
1611 	if (!(pi->phy.link_info.link_info & ICE_AQ_MEDIA_AVAILABLE))
1612 		return -EPERM;
1613 
1614 	ice_print_topo_conflict(vsi);
1615 
1616 	if (vsi->port_info->phy.link_info.topo_media_conflict ==
1617 	    ICE_AQ_LINK_TOPO_UNSUPP_MEDIA)
1618 		return -EPERM;
1619 
1620 	if (test_bit(ICE_FLAG_LINK_DOWN_ON_CLOSE_ENA, vsi->back->flags))
1621 		return ice_force_phys_link_state(vsi, true);
1622 
1623 	pcaps = kzalloc(sizeof(*pcaps), GFP_KERNEL);
1624 	if (!pcaps)
1625 		return -ENOMEM;
1626 
1627 	/* Get current PHY config */
1628 	status = ice_aq_get_phy_caps(pi, false, ICE_AQC_REPORT_SW_CFG, pcaps,
1629 				     NULL);
1630 	if (status) {
1631 		dev_err(dev, "Failed to get PHY configuration, VSI %d error %s\n",
1632 			vsi->vsi_num, ice_stat_str(status));
1633 		err = -EIO;
1634 		goto done;
1635 	}
1636 
1637 	/* If PHY enable link is configured and configuration has not changed,
1638 	 * there's nothing to do
1639 	 */
1640 	if (pcaps->caps & ICE_AQC_PHY_EN_LINK &&
1641 	    ice_phy_caps_equals_cfg(pcaps, &pi->phy.curr_user_phy_cfg))
1642 		goto done;
1643 
1644 	/* Use PHY topology as baseline for configuration */
1645 	memset(pcaps, 0, sizeof(*pcaps));
1646 	status = ice_aq_get_phy_caps(pi, false, ICE_AQC_REPORT_TOPO_CAP, pcaps,
1647 				     NULL);
1648 	if (status) {
1649 		dev_err(dev, "Failed to get PHY topology, VSI %d error %s\n",
1650 			vsi->vsi_num, ice_stat_str(status));
1651 		err = -EIO;
1652 		goto done;
1653 	}
1654 
1655 	cfg = kzalloc(sizeof(*cfg), GFP_KERNEL);
1656 	if (!cfg) {
1657 		err = -ENOMEM;
1658 		goto done;
1659 	}
1660 
1661 	ice_copy_phy_caps_to_cfg(pi, pcaps, cfg);
1662 
1663 	/* Speed - If default override pending, use curr_user_phy_cfg set in
1664 	 * ice_init_phy_user_cfg_ldo.
1665 	 */
1666 	if (test_and_clear_bit(__ICE_LINK_DEFAULT_OVERRIDE_PENDING,
1667 			       vsi->back->state)) {
1668 		cfg->phy_type_low = pi->phy.curr_user_phy_cfg.phy_type_low;
1669 		cfg->phy_type_high = pi->phy.curr_user_phy_cfg.phy_type_high;
1670 	} else {
1671 		u64 phy_low = 0, phy_high = 0;
1672 
1673 		ice_update_phy_type(&phy_low, &phy_high,
1674 				    pi->phy.curr_user_speed_req);
1675 		cfg->phy_type_low = pcaps->phy_type_low & cpu_to_le64(phy_low);
1676 		cfg->phy_type_high = pcaps->phy_type_high &
1677 				     cpu_to_le64(phy_high);
1678 	}
1679 
1680 	/* Can't provide what was requested; use PHY capabilities */
1681 	if (!cfg->phy_type_low && !cfg->phy_type_high) {
1682 		cfg->phy_type_low = pcaps->phy_type_low;
1683 		cfg->phy_type_high = pcaps->phy_type_high;
1684 	}
1685 
1686 	/* FEC */
1687 	ice_cfg_phy_fec(pi, cfg, pi->phy.curr_user_fec_req);
1688 
1689 	/* Can't provide what was requested; use PHY capabilities */
1690 	if (cfg->link_fec_opt !=
1691 	    (cfg->link_fec_opt & pcaps->link_fec_options)) {
1692 		cfg->caps |= pcaps->caps & ICE_AQC_PHY_EN_AUTO_FEC;
1693 		cfg->link_fec_opt = pcaps->link_fec_options;
1694 	}
1695 
1696 	/* Flow Control - always supported; no need to check against
1697 	 * capabilities
1698 	 */
1699 	ice_cfg_phy_fc(pi, cfg, pi->phy.curr_user_fc_req);
1700 
1701 	/* Enable link and link update */
1702 	cfg->caps |= ICE_AQ_PHY_ENA_AUTO_LINK_UPDT | ICE_AQ_PHY_ENA_LINK;
1703 
1704 	status = ice_aq_set_phy_cfg(&vsi->back->hw, pi, cfg, NULL);
1705 	if (status) {
1706 		dev_err(dev, "Failed to set phy config, VSI %d error %s\n",
1707 			vsi->vsi_num, ice_stat_str(status));
1708 		err = -EIO;
1709 	}
1710 
1711 	kfree(cfg);
1712 done:
1713 	kfree(pcaps);
1714 	return err;
1715 }
1716 
1717 /**
1718  * ice_check_media_subtask - Check for media
1719  * @pf: pointer to PF struct
1720  *
1721  * If media is available, then initialize PHY user configuration if it is not
1722  * been, and configure the PHY if the interface is up.
1723  */
1724 static void ice_check_media_subtask(struct ice_pf *pf)
1725 {
1726 	struct ice_port_info *pi;
1727 	struct ice_vsi *vsi;
1728 	int err;
1729 
1730 	/* No need to check for media if it's already present */
1731 	if (!test_bit(ICE_FLAG_NO_MEDIA, pf->flags))
1732 		return;
1733 
1734 	vsi = ice_get_main_vsi(pf);
1735 	if (!vsi)
1736 		return;
1737 
1738 	/* Refresh link info and check if media is present */
1739 	pi = vsi->port_info;
1740 	err = ice_update_link_info(pi);
1741 	if (err)
1742 		return;
1743 
1744 	if (pi->phy.link_info.link_info & ICE_AQ_MEDIA_AVAILABLE) {
1745 		if (!test_bit(__ICE_PHY_INIT_COMPLETE, pf->state))
1746 			ice_init_phy_user_cfg(pi);
1747 
1748 		/* PHY settings are reset on media insertion, reconfigure
1749 		 * PHY to preserve settings.
1750 		 */
1751 		if (test_bit(__ICE_DOWN, vsi->state) &&
1752 		    test_bit(ICE_FLAG_LINK_DOWN_ON_CLOSE_ENA, vsi->back->flags))
1753 			return;
1754 
1755 		err = ice_configure_phy(vsi);
1756 		if (!err)
1757 			clear_bit(ICE_FLAG_NO_MEDIA, pf->flags);
1758 
1759 		/* A Link Status Event will be generated; the event handler
1760 		 * will complete bringing the interface up
1761 		 */
1762 	}
1763 }
1764 
1765 /**
1766  * ice_service_task - manage and run subtasks
1767  * @work: pointer to work_struct contained by the PF struct
1768  */
1769 static void ice_service_task(struct work_struct *work)
1770 {
1771 	struct ice_pf *pf = container_of(work, struct ice_pf, serv_task);
1772 	unsigned long start_time = jiffies;
1773 
1774 	/* subtasks */
1775 
1776 	/* process reset requests first */
1777 	ice_reset_subtask(pf);
1778 
1779 	/* bail if a reset/recovery cycle is pending or rebuild failed */
1780 	if (ice_is_reset_in_progress(pf->state) ||
1781 	    test_bit(__ICE_SUSPENDED, pf->state) ||
1782 	    test_bit(__ICE_NEEDS_RESTART, pf->state)) {
1783 		ice_service_task_complete(pf);
1784 		return;
1785 	}
1786 
1787 	ice_clean_adminq_subtask(pf);
1788 	ice_check_media_subtask(pf);
1789 	ice_check_for_hang_subtask(pf);
1790 	ice_sync_fltr_subtask(pf);
1791 	ice_handle_mdd_event(pf);
1792 	ice_watchdog_subtask(pf);
1793 
1794 	if (ice_is_safe_mode(pf)) {
1795 		ice_service_task_complete(pf);
1796 		return;
1797 	}
1798 
1799 	ice_process_vflr_event(pf);
1800 	ice_clean_mailboxq_subtask(pf);
1801 	ice_sync_arfs_fltrs(pf);
1802 	/* Clear __ICE_SERVICE_SCHED flag to allow scheduling next event */
1803 	ice_service_task_complete(pf);
1804 
1805 	/* If the tasks have taken longer than one service timer period
1806 	 * or there is more work to be done, reset the service timer to
1807 	 * schedule the service task now.
1808 	 */
1809 	if (time_after(jiffies, (start_time + pf->serv_tmr_period)) ||
1810 	    test_bit(__ICE_MDD_EVENT_PENDING, pf->state) ||
1811 	    test_bit(__ICE_VFLR_EVENT_PENDING, pf->state) ||
1812 	    test_bit(__ICE_MAILBOXQ_EVENT_PENDING, pf->state) ||
1813 	    test_bit(__ICE_ADMINQ_EVENT_PENDING, pf->state))
1814 		mod_timer(&pf->serv_tmr, jiffies);
1815 }
1816 
1817 /**
1818  * ice_set_ctrlq_len - helper function to set controlq length
1819  * @hw: pointer to the HW instance
1820  */
1821 static void ice_set_ctrlq_len(struct ice_hw *hw)
1822 {
1823 	hw->adminq.num_rq_entries = ICE_AQ_LEN;
1824 	hw->adminq.num_sq_entries = ICE_AQ_LEN;
1825 	hw->adminq.rq_buf_size = ICE_AQ_MAX_BUF_LEN;
1826 	hw->adminq.sq_buf_size = ICE_AQ_MAX_BUF_LEN;
1827 	hw->mailboxq.num_rq_entries = PF_MBX_ARQLEN_ARQLEN_M;
1828 	hw->mailboxq.num_sq_entries = ICE_MBXSQ_LEN;
1829 	hw->mailboxq.rq_buf_size = ICE_MBXQ_MAX_BUF_LEN;
1830 	hw->mailboxq.sq_buf_size = ICE_MBXQ_MAX_BUF_LEN;
1831 }
1832 
1833 /**
1834  * ice_schedule_reset - schedule a reset
1835  * @pf: board private structure
1836  * @reset: reset being requested
1837  */
1838 int ice_schedule_reset(struct ice_pf *pf, enum ice_reset_req reset)
1839 {
1840 	struct device *dev = ice_pf_to_dev(pf);
1841 
1842 	/* bail out if earlier reset has failed */
1843 	if (test_bit(__ICE_RESET_FAILED, pf->state)) {
1844 		dev_dbg(dev, "earlier reset has failed\n");
1845 		return -EIO;
1846 	}
1847 	/* bail if reset/recovery already in progress */
1848 	if (ice_is_reset_in_progress(pf->state)) {
1849 		dev_dbg(dev, "Reset already in progress\n");
1850 		return -EBUSY;
1851 	}
1852 
1853 	switch (reset) {
1854 	case ICE_RESET_PFR:
1855 		set_bit(__ICE_PFR_REQ, pf->state);
1856 		break;
1857 	case ICE_RESET_CORER:
1858 		set_bit(__ICE_CORER_REQ, pf->state);
1859 		break;
1860 	case ICE_RESET_GLOBR:
1861 		set_bit(__ICE_GLOBR_REQ, pf->state);
1862 		break;
1863 	default:
1864 		return -EINVAL;
1865 	}
1866 
1867 	ice_service_task_schedule(pf);
1868 	return 0;
1869 }
1870 
1871 /**
1872  * ice_irq_affinity_notify - Callback for affinity changes
1873  * @notify: context as to what irq was changed
1874  * @mask: the new affinity mask
1875  *
1876  * This is a callback function used by the irq_set_affinity_notifier function
1877  * so that we may register to receive changes to the irq affinity masks.
1878  */
1879 static void
1880 ice_irq_affinity_notify(struct irq_affinity_notify *notify,
1881 			const cpumask_t *mask)
1882 {
1883 	struct ice_q_vector *q_vector =
1884 		container_of(notify, struct ice_q_vector, affinity_notify);
1885 
1886 	cpumask_copy(&q_vector->affinity_mask, mask);
1887 }
1888 
1889 /**
1890  * ice_irq_affinity_release - Callback for affinity notifier release
1891  * @ref: internal core kernel usage
1892  *
1893  * This is a callback function used by the irq_set_affinity_notifier function
1894  * to inform the current notification subscriber that they will no longer
1895  * receive notifications.
1896  */
1897 static void ice_irq_affinity_release(struct kref __always_unused *ref) {}
1898 
1899 /**
1900  * ice_vsi_ena_irq - Enable IRQ for the given VSI
1901  * @vsi: the VSI being configured
1902  */
1903 static int ice_vsi_ena_irq(struct ice_vsi *vsi)
1904 {
1905 	struct ice_hw *hw = &vsi->back->hw;
1906 	int i;
1907 
1908 	ice_for_each_q_vector(vsi, i)
1909 		ice_irq_dynamic_ena(hw, vsi, vsi->q_vectors[i]);
1910 
1911 	ice_flush(hw);
1912 	return 0;
1913 }
1914 
1915 /**
1916  * ice_vsi_req_irq_msix - get MSI-X vectors from the OS for the VSI
1917  * @vsi: the VSI being configured
1918  * @basename: name for the vector
1919  */
1920 static int ice_vsi_req_irq_msix(struct ice_vsi *vsi, char *basename)
1921 {
1922 	int q_vectors = vsi->num_q_vectors;
1923 	struct ice_pf *pf = vsi->back;
1924 	int base = vsi->base_vector;
1925 	struct device *dev;
1926 	int rx_int_idx = 0;
1927 	int tx_int_idx = 0;
1928 	int vector, err;
1929 	int irq_num;
1930 
1931 	dev = ice_pf_to_dev(pf);
1932 	for (vector = 0; vector < q_vectors; vector++) {
1933 		struct ice_q_vector *q_vector = vsi->q_vectors[vector];
1934 
1935 		irq_num = pf->msix_entries[base + vector].vector;
1936 
1937 		if (q_vector->tx.ring && q_vector->rx.ring) {
1938 			snprintf(q_vector->name, sizeof(q_vector->name) - 1,
1939 				 "%s-%s-%d", basename, "TxRx", rx_int_idx++);
1940 			tx_int_idx++;
1941 		} else if (q_vector->rx.ring) {
1942 			snprintf(q_vector->name, sizeof(q_vector->name) - 1,
1943 				 "%s-%s-%d", basename, "rx", rx_int_idx++);
1944 		} else if (q_vector->tx.ring) {
1945 			snprintf(q_vector->name, sizeof(q_vector->name) - 1,
1946 				 "%s-%s-%d", basename, "tx", tx_int_idx++);
1947 		} else {
1948 			/* skip this unused q_vector */
1949 			continue;
1950 		}
1951 		err = devm_request_irq(dev, irq_num, vsi->irq_handler, 0,
1952 				       q_vector->name, q_vector);
1953 		if (err) {
1954 			netdev_err(vsi->netdev, "MSIX request_irq failed, error: %d\n",
1955 				   err);
1956 			goto free_q_irqs;
1957 		}
1958 
1959 		/* register for affinity change notifications */
1960 		if (!IS_ENABLED(CONFIG_RFS_ACCEL)) {
1961 			struct irq_affinity_notify *affinity_notify;
1962 
1963 			affinity_notify = &q_vector->affinity_notify;
1964 			affinity_notify->notify = ice_irq_affinity_notify;
1965 			affinity_notify->release = ice_irq_affinity_release;
1966 			irq_set_affinity_notifier(irq_num, affinity_notify);
1967 		}
1968 
1969 		/* assign the mask for this irq */
1970 		irq_set_affinity_hint(irq_num, &q_vector->affinity_mask);
1971 	}
1972 
1973 	vsi->irqs_ready = true;
1974 	return 0;
1975 
1976 free_q_irqs:
1977 	while (vector) {
1978 		vector--;
1979 		irq_num = pf->msix_entries[base + vector].vector;
1980 		if (!IS_ENABLED(CONFIG_RFS_ACCEL))
1981 			irq_set_affinity_notifier(irq_num, NULL);
1982 		irq_set_affinity_hint(irq_num, NULL);
1983 		devm_free_irq(dev, irq_num, &vsi->q_vectors[vector]);
1984 	}
1985 	return err;
1986 }
1987 
1988 /**
1989  * ice_xdp_alloc_setup_rings - Allocate and setup Tx rings for XDP
1990  * @vsi: VSI to setup Tx rings used by XDP
1991  *
1992  * Return 0 on success and negative value on error
1993  */
1994 static int ice_xdp_alloc_setup_rings(struct ice_vsi *vsi)
1995 {
1996 	struct device *dev = ice_pf_to_dev(vsi->back);
1997 	int i;
1998 
1999 	for (i = 0; i < vsi->num_xdp_txq; i++) {
2000 		u16 xdp_q_idx = vsi->alloc_txq + i;
2001 		struct ice_ring *xdp_ring;
2002 
2003 		xdp_ring = kzalloc(sizeof(*xdp_ring), GFP_KERNEL);
2004 
2005 		if (!xdp_ring)
2006 			goto free_xdp_rings;
2007 
2008 		xdp_ring->q_index = xdp_q_idx;
2009 		xdp_ring->reg_idx = vsi->txq_map[xdp_q_idx];
2010 		xdp_ring->ring_active = false;
2011 		xdp_ring->vsi = vsi;
2012 		xdp_ring->netdev = NULL;
2013 		xdp_ring->dev = dev;
2014 		xdp_ring->count = vsi->num_tx_desc;
2015 		WRITE_ONCE(vsi->xdp_rings[i], xdp_ring);
2016 		if (ice_setup_tx_ring(xdp_ring))
2017 			goto free_xdp_rings;
2018 		ice_set_ring_xdp(xdp_ring);
2019 		xdp_ring->xsk_umem = ice_xsk_umem(xdp_ring);
2020 	}
2021 
2022 	return 0;
2023 
2024 free_xdp_rings:
2025 	for (; i >= 0; i--)
2026 		if (vsi->xdp_rings[i] && vsi->xdp_rings[i]->desc)
2027 			ice_free_tx_ring(vsi->xdp_rings[i]);
2028 	return -ENOMEM;
2029 }
2030 
2031 /**
2032  * ice_vsi_assign_bpf_prog - set or clear bpf prog pointer on VSI
2033  * @vsi: VSI to set the bpf prog on
2034  * @prog: the bpf prog pointer
2035  */
2036 static void ice_vsi_assign_bpf_prog(struct ice_vsi *vsi, struct bpf_prog *prog)
2037 {
2038 	struct bpf_prog *old_prog;
2039 	int i;
2040 
2041 	old_prog = xchg(&vsi->xdp_prog, prog);
2042 	if (old_prog)
2043 		bpf_prog_put(old_prog);
2044 
2045 	ice_for_each_rxq(vsi, i)
2046 		WRITE_ONCE(vsi->rx_rings[i]->xdp_prog, vsi->xdp_prog);
2047 }
2048 
2049 /**
2050  * ice_prepare_xdp_rings - Allocate, configure and setup Tx rings for XDP
2051  * @vsi: VSI to bring up Tx rings used by XDP
2052  * @prog: bpf program that will be assigned to VSI
2053  *
2054  * Return 0 on success and negative value on error
2055  */
2056 int ice_prepare_xdp_rings(struct ice_vsi *vsi, struct bpf_prog *prog)
2057 {
2058 	u16 max_txqs[ICE_MAX_TRAFFIC_CLASS] = { 0 };
2059 	int xdp_rings_rem = vsi->num_xdp_txq;
2060 	struct ice_pf *pf = vsi->back;
2061 	struct ice_qs_cfg xdp_qs_cfg = {
2062 		.qs_mutex = &pf->avail_q_mutex,
2063 		.pf_map = pf->avail_txqs,
2064 		.pf_map_size = pf->max_pf_txqs,
2065 		.q_count = vsi->num_xdp_txq,
2066 		.scatter_count = ICE_MAX_SCATTER_TXQS,
2067 		.vsi_map = vsi->txq_map,
2068 		.vsi_map_offset = vsi->alloc_txq,
2069 		.mapping_mode = ICE_VSI_MAP_CONTIG
2070 	};
2071 	enum ice_status status;
2072 	struct device *dev;
2073 	int i, v_idx;
2074 
2075 	dev = ice_pf_to_dev(pf);
2076 	vsi->xdp_rings = devm_kcalloc(dev, vsi->num_xdp_txq,
2077 				      sizeof(*vsi->xdp_rings), GFP_KERNEL);
2078 	if (!vsi->xdp_rings)
2079 		return -ENOMEM;
2080 
2081 	vsi->xdp_mapping_mode = xdp_qs_cfg.mapping_mode;
2082 	if (__ice_vsi_get_qs(&xdp_qs_cfg))
2083 		goto err_map_xdp;
2084 
2085 	if (ice_xdp_alloc_setup_rings(vsi))
2086 		goto clear_xdp_rings;
2087 
2088 	/* follow the logic from ice_vsi_map_rings_to_vectors */
2089 	ice_for_each_q_vector(vsi, v_idx) {
2090 		struct ice_q_vector *q_vector = vsi->q_vectors[v_idx];
2091 		int xdp_rings_per_v, q_id, q_base;
2092 
2093 		xdp_rings_per_v = DIV_ROUND_UP(xdp_rings_rem,
2094 					       vsi->num_q_vectors - v_idx);
2095 		q_base = vsi->num_xdp_txq - xdp_rings_rem;
2096 
2097 		for (q_id = q_base; q_id < (q_base + xdp_rings_per_v); q_id++) {
2098 			struct ice_ring *xdp_ring = vsi->xdp_rings[q_id];
2099 
2100 			xdp_ring->q_vector = q_vector;
2101 			xdp_ring->next = q_vector->tx.ring;
2102 			q_vector->tx.ring = xdp_ring;
2103 		}
2104 		xdp_rings_rem -= xdp_rings_per_v;
2105 	}
2106 
2107 	/* omit the scheduler update if in reset path; XDP queues will be
2108 	 * taken into account at the end of ice_vsi_rebuild, where
2109 	 * ice_cfg_vsi_lan is being called
2110 	 */
2111 	if (ice_is_reset_in_progress(pf->state))
2112 		return 0;
2113 
2114 	/* tell the Tx scheduler that right now we have
2115 	 * additional queues
2116 	 */
2117 	for (i = 0; i < vsi->tc_cfg.numtc; i++)
2118 		max_txqs[i] = vsi->num_txq + vsi->num_xdp_txq;
2119 
2120 	status = ice_cfg_vsi_lan(vsi->port_info, vsi->idx, vsi->tc_cfg.ena_tc,
2121 				 max_txqs);
2122 	if (status) {
2123 		dev_err(dev, "Failed VSI LAN queue config for XDP, error: %s\n",
2124 			ice_stat_str(status));
2125 		goto clear_xdp_rings;
2126 	}
2127 	ice_vsi_assign_bpf_prog(vsi, prog);
2128 
2129 	return 0;
2130 clear_xdp_rings:
2131 	for (i = 0; i < vsi->num_xdp_txq; i++)
2132 		if (vsi->xdp_rings[i]) {
2133 			kfree_rcu(vsi->xdp_rings[i], rcu);
2134 			vsi->xdp_rings[i] = NULL;
2135 		}
2136 
2137 err_map_xdp:
2138 	mutex_lock(&pf->avail_q_mutex);
2139 	for (i = 0; i < vsi->num_xdp_txq; i++) {
2140 		clear_bit(vsi->txq_map[i + vsi->alloc_txq], pf->avail_txqs);
2141 		vsi->txq_map[i + vsi->alloc_txq] = ICE_INVAL_Q_INDEX;
2142 	}
2143 	mutex_unlock(&pf->avail_q_mutex);
2144 
2145 	devm_kfree(dev, vsi->xdp_rings);
2146 	return -ENOMEM;
2147 }
2148 
2149 /**
2150  * ice_destroy_xdp_rings - undo the configuration made by ice_prepare_xdp_rings
2151  * @vsi: VSI to remove XDP rings
2152  *
2153  * Detach XDP rings from irq vectors, clean up the PF bitmap and free
2154  * resources
2155  */
2156 int ice_destroy_xdp_rings(struct ice_vsi *vsi)
2157 {
2158 	u16 max_txqs[ICE_MAX_TRAFFIC_CLASS] = { 0 };
2159 	struct ice_pf *pf = vsi->back;
2160 	int i, v_idx;
2161 
2162 	/* q_vectors are freed in reset path so there's no point in detaching
2163 	 * rings; in case of rebuild being triggered not from reset reset bits
2164 	 * in pf->state won't be set, so additionally check first q_vector
2165 	 * against NULL
2166 	 */
2167 	if (ice_is_reset_in_progress(pf->state) || !vsi->q_vectors[0])
2168 		goto free_qmap;
2169 
2170 	ice_for_each_q_vector(vsi, v_idx) {
2171 		struct ice_q_vector *q_vector = vsi->q_vectors[v_idx];
2172 		struct ice_ring *ring;
2173 
2174 		ice_for_each_ring(ring, q_vector->tx)
2175 			if (!ring->tx_buf || !ice_ring_is_xdp(ring))
2176 				break;
2177 
2178 		/* restore the value of last node prior to XDP setup */
2179 		q_vector->tx.ring = ring;
2180 	}
2181 
2182 free_qmap:
2183 	mutex_lock(&pf->avail_q_mutex);
2184 	for (i = 0; i < vsi->num_xdp_txq; i++) {
2185 		clear_bit(vsi->txq_map[i + vsi->alloc_txq], pf->avail_txqs);
2186 		vsi->txq_map[i + vsi->alloc_txq] = ICE_INVAL_Q_INDEX;
2187 	}
2188 	mutex_unlock(&pf->avail_q_mutex);
2189 
2190 	for (i = 0; i < vsi->num_xdp_txq; i++)
2191 		if (vsi->xdp_rings[i]) {
2192 			if (vsi->xdp_rings[i]->desc)
2193 				ice_free_tx_ring(vsi->xdp_rings[i]);
2194 			kfree_rcu(vsi->xdp_rings[i], rcu);
2195 			vsi->xdp_rings[i] = NULL;
2196 		}
2197 
2198 	devm_kfree(ice_pf_to_dev(pf), vsi->xdp_rings);
2199 	vsi->xdp_rings = NULL;
2200 
2201 	if (ice_is_reset_in_progress(pf->state) || !vsi->q_vectors[0])
2202 		return 0;
2203 
2204 	ice_vsi_assign_bpf_prog(vsi, NULL);
2205 
2206 	/* notify Tx scheduler that we destroyed XDP queues and bring
2207 	 * back the old number of child nodes
2208 	 */
2209 	for (i = 0; i < vsi->tc_cfg.numtc; i++)
2210 		max_txqs[i] = vsi->num_txq;
2211 
2212 	/* change number of XDP Tx queues to 0 */
2213 	vsi->num_xdp_txq = 0;
2214 
2215 	return ice_cfg_vsi_lan(vsi->port_info, vsi->idx, vsi->tc_cfg.ena_tc,
2216 			       max_txqs);
2217 }
2218 
2219 /**
2220  * ice_xdp_setup_prog - Add or remove XDP eBPF program
2221  * @vsi: VSI to setup XDP for
2222  * @prog: XDP program
2223  * @extack: netlink extended ack
2224  */
2225 static int
2226 ice_xdp_setup_prog(struct ice_vsi *vsi, struct bpf_prog *prog,
2227 		   struct netlink_ext_ack *extack)
2228 {
2229 	int frame_size = vsi->netdev->mtu + ICE_ETH_PKT_HDR_PAD;
2230 	bool if_running = netif_running(vsi->netdev);
2231 	int ret = 0, xdp_ring_err = 0;
2232 
2233 	if (frame_size > vsi->rx_buf_len) {
2234 		NL_SET_ERR_MSG_MOD(extack, "MTU too large for loading XDP");
2235 		return -EOPNOTSUPP;
2236 	}
2237 
2238 	/* need to stop netdev while setting up the program for Rx rings */
2239 	if (if_running && !test_and_set_bit(__ICE_DOWN, vsi->state)) {
2240 		ret = ice_down(vsi);
2241 		if (ret) {
2242 			NL_SET_ERR_MSG_MOD(extack, "Preparing device for XDP attach failed");
2243 			return ret;
2244 		}
2245 	}
2246 
2247 	if (!ice_is_xdp_ena_vsi(vsi) && prog) {
2248 		vsi->num_xdp_txq = vsi->alloc_rxq;
2249 		xdp_ring_err = ice_prepare_xdp_rings(vsi, prog);
2250 		if (xdp_ring_err)
2251 			NL_SET_ERR_MSG_MOD(extack, "Setting up XDP Tx resources failed");
2252 	} else if (ice_is_xdp_ena_vsi(vsi) && !prog) {
2253 		xdp_ring_err = ice_destroy_xdp_rings(vsi);
2254 		if (xdp_ring_err)
2255 			NL_SET_ERR_MSG_MOD(extack, "Freeing XDP Tx resources failed");
2256 	} else {
2257 		ice_vsi_assign_bpf_prog(vsi, prog);
2258 	}
2259 
2260 	if (if_running)
2261 		ret = ice_up(vsi);
2262 
2263 	if (!ret && prog && vsi->xsk_umems) {
2264 		int i;
2265 
2266 		ice_for_each_rxq(vsi, i) {
2267 			struct ice_ring *rx_ring = vsi->rx_rings[i];
2268 
2269 			if (rx_ring->xsk_umem)
2270 				napi_schedule(&rx_ring->q_vector->napi);
2271 		}
2272 	}
2273 
2274 	return (ret || xdp_ring_err) ? -ENOMEM : 0;
2275 }
2276 
2277 /**
2278  * ice_xdp - implements XDP handler
2279  * @dev: netdevice
2280  * @xdp: XDP command
2281  */
2282 static int ice_xdp(struct net_device *dev, struct netdev_bpf *xdp)
2283 {
2284 	struct ice_netdev_priv *np = netdev_priv(dev);
2285 	struct ice_vsi *vsi = np->vsi;
2286 
2287 	if (vsi->type != ICE_VSI_PF) {
2288 		NL_SET_ERR_MSG_MOD(xdp->extack, "XDP can be loaded only on PF VSI");
2289 		return -EINVAL;
2290 	}
2291 
2292 	switch (xdp->command) {
2293 	case XDP_SETUP_PROG:
2294 		return ice_xdp_setup_prog(vsi, xdp->prog, xdp->extack);
2295 	case XDP_SETUP_XSK_UMEM:
2296 		return ice_xsk_umem_setup(vsi, xdp->xsk.umem,
2297 					  xdp->xsk.queue_id);
2298 	default:
2299 		return -EINVAL;
2300 	}
2301 }
2302 
2303 /**
2304  * ice_ena_misc_vector - enable the non-queue interrupts
2305  * @pf: board private structure
2306  */
2307 static void ice_ena_misc_vector(struct ice_pf *pf)
2308 {
2309 	struct ice_hw *hw = &pf->hw;
2310 	u32 val;
2311 
2312 	/* Disable anti-spoof detection interrupt to prevent spurious event
2313 	 * interrupts during a function reset. Anti-spoof functionally is
2314 	 * still supported.
2315 	 */
2316 	val = rd32(hw, GL_MDCK_TX_TDPU);
2317 	val |= GL_MDCK_TX_TDPU_RCU_ANTISPOOF_ITR_DIS_M;
2318 	wr32(hw, GL_MDCK_TX_TDPU, val);
2319 
2320 	/* clear things first */
2321 	wr32(hw, PFINT_OICR_ENA, 0);	/* disable all */
2322 	rd32(hw, PFINT_OICR);		/* read to clear */
2323 
2324 	val = (PFINT_OICR_ECC_ERR_M |
2325 	       PFINT_OICR_MAL_DETECT_M |
2326 	       PFINT_OICR_GRST_M |
2327 	       PFINT_OICR_PCI_EXCEPTION_M |
2328 	       PFINT_OICR_VFLR_M |
2329 	       PFINT_OICR_HMC_ERR_M |
2330 	       PFINT_OICR_PE_CRITERR_M);
2331 
2332 	wr32(hw, PFINT_OICR_ENA, val);
2333 
2334 	/* SW_ITR_IDX = 0, but don't change INTENA */
2335 	wr32(hw, GLINT_DYN_CTL(pf->oicr_idx),
2336 	     GLINT_DYN_CTL_SW_ITR_INDX_M | GLINT_DYN_CTL_INTENA_MSK_M);
2337 }
2338 
2339 /**
2340  * ice_misc_intr - misc interrupt handler
2341  * @irq: interrupt number
2342  * @data: pointer to a q_vector
2343  */
2344 static irqreturn_t ice_misc_intr(int __always_unused irq, void *data)
2345 {
2346 	struct ice_pf *pf = (struct ice_pf *)data;
2347 	struct ice_hw *hw = &pf->hw;
2348 	irqreturn_t ret = IRQ_NONE;
2349 	struct device *dev;
2350 	u32 oicr, ena_mask;
2351 
2352 	dev = ice_pf_to_dev(pf);
2353 	set_bit(__ICE_ADMINQ_EVENT_PENDING, pf->state);
2354 	set_bit(__ICE_MAILBOXQ_EVENT_PENDING, pf->state);
2355 
2356 	oicr = rd32(hw, PFINT_OICR);
2357 	ena_mask = rd32(hw, PFINT_OICR_ENA);
2358 
2359 	if (oicr & PFINT_OICR_SWINT_M) {
2360 		ena_mask &= ~PFINT_OICR_SWINT_M;
2361 		pf->sw_int_count++;
2362 	}
2363 
2364 	if (oicr & PFINT_OICR_MAL_DETECT_M) {
2365 		ena_mask &= ~PFINT_OICR_MAL_DETECT_M;
2366 		set_bit(__ICE_MDD_EVENT_PENDING, pf->state);
2367 	}
2368 	if (oicr & PFINT_OICR_VFLR_M) {
2369 		/* disable any further VFLR event notifications */
2370 		if (test_bit(__ICE_VF_RESETS_DISABLED, pf->state)) {
2371 			u32 reg = rd32(hw, PFINT_OICR_ENA);
2372 
2373 			reg &= ~PFINT_OICR_VFLR_M;
2374 			wr32(hw, PFINT_OICR_ENA, reg);
2375 		} else {
2376 			ena_mask &= ~PFINT_OICR_VFLR_M;
2377 			set_bit(__ICE_VFLR_EVENT_PENDING, pf->state);
2378 		}
2379 	}
2380 
2381 	if (oicr & PFINT_OICR_GRST_M) {
2382 		u32 reset;
2383 
2384 		/* we have a reset warning */
2385 		ena_mask &= ~PFINT_OICR_GRST_M;
2386 		reset = (rd32(hw, GLGEN_RSTAT) & GLGEN_RSTAT_RESET_TYPE_M) >>
2387 			GLGEN_RSTAT_RESET_TYPE_S;
2388 
2389 		if (reset == ICE_RESET_CORER)
2390 			pf->corer_count++;
2391 		else if (reset == ICE_RESET_GLOBR)
2392 			pf->globr_count++;
2393 		else if (reset == ICE_RESET_EMPR)
2394 			pf->empr_count++;
2395 		else
2396 			dev_dbg(dev, "Invalid reset type %d\n", reset);
2397 
2398 		/* If a reset cycle isn't already in progress, we set a bit in
2399 		 * pf->state so that the service task can start a reset/rebuild.
2400 		 * We also make note of which reset happened so that peer
2401 		 * devices/drivers can be informed.
2402 		 */
2403 		if (!test_and_set_bit(__ICE_RESET_OICR_RECV, pf->state)) {
2404 			if (reset == ICE_RESET_CORER)
2405 				set_bit(__ICE_CORER_RECV, pf->state);
2406 			else if (reset == ICE_RESET_GLOBR)
2407 				set_bit(__ICE_GLOBR_RECV, pf->state);
2408 			else
2409 				set_bit(__ICE_EMPR_RECV, pf->state);
2410 
2411 			/* There are couple of different bits at play here.
2412 			 * hw->reset_ongoing indicates whether the hardware is
2413 			 * in reset. This is set to true when a reset interrupt
2414 			 * is received and set back to false after the driver
2415 			 * has determined that the hardware is out of reset.
2416 			 *
2417 			 * __ICE_RESET_OICR_RECV in pf->state indicates
2418 			 * that a post reset rebuild is required before the
2419 			 * driver is operational again. This is set above.
2420 			 *
2421 			 * As this is the start of the reset/rebuild cycle, set
2422 			 * both to indicate that.
2423 			 */
2424 			hw->reset_ongoing = true;
2425 		}
2426 	}
2427 
2428 	if (oicr & PFINT_OICR_HMC_ERR_M) {
2429 		ena_mask &= ~PFINT_OICR_HMC_ERR_M;
2430 		dev_dbg(dev, "HMC Error interrupt - info 0x%x, data 0x%x\n",
2431 			rd32(hw, PFHMC_ERRORINFO),
2432 			rd32(hw, PFHMC_ERRORDATA));
2433 	}
2434 
2435 	/* Report any remaining unexpected interrupts */
2436 	oicr &= ena_mask;
2437 	if (oicr) {
2438 		dev_dbg(dev, "unhandled interrupt oicr=0x%08x\n", oicr);
2439 		/* If a critical error is pending there is no choice but to
2440 		 * reset the device.
2441 		 */
2442 		if (oicr & (PFINT_OICR_PE_CRITERR_M |
2443 			    PFINT_OICR_PCI_EXCEPTION_M |
2444 			    PFINT_OICR_ECC_ERR_M)) {
2445 			set_bit(__ICE_PFR_REQ, pf->state);
2446 			ice_service_task_schedule(pf);
2447 		}
2448 	}
2449 	ret = IRQ_HANDLED;
2450 
2451 	ice_service_task_schedule(pf);
2452 	ice_irq_dynamic_ena(hw, NULL, NULL);
2453 
2454 	return ret;
2455 }
2456 
2457 /**
2458  * ice_dis_ctrlq_interrupts - disable control queue interrupts
2459  * @hw: pointer to HW structure
2460  */
2461 static void ice_dis_ctrlq_interrupts(struct ice_hw *hw)
2462 {
2463 	/* disable Admin queue Interrupt causes */
2464 	wr32(hw, PFINT_FW_CTL,
2465 	     rd32(hw, PFINT_FW_CTL) & ~PFINT_FW_CTL_CAUSE_ENA_M);
2466 
2467 	/* disable Mailbox queue Interrupt causes */
2468 	wr32(hw, PFINT_MBX_CTL,
2469 	     rd32(hw, PFINT_MBX_CTL) & ~PFINT_MBX_CTL_CAUSE_ENA_M);
2470 
2471 	/* disable Control queue Interrupt causes */
2472 	wr32(hw, PFINT_OICR_CTL,
2473 	     rd32(hw, PFINT_OICR_CTL) & ~PFINT_OICR_CTL_CAUSE_ENA_M);
2474 
2475 	ice_flush(hw);
2476 }
2477 
2478 /**
2479  * ice_free_irq_msix_misc - Unroll misc vector setup
2480  * @pf: board private structure
2481  */
2482 static void ice_free_irq_msix_misc(struct ice_pf *pf)
2483 {
2484 	struct ice_hw *hw = &pf->hw;
2485 
2486 	ice_dis_ctrlq_interrupts(hw);
2487 
2488 	/* disable OICR interrupt */
2489 	wr32(hw, PFINT_OICR_ENA, 0);
2490 	ice_flush(hw);
2491 
2492 	if (pf->msix_entries) {
2493 		synchronize_irq(pf->msix_entries[pf->oicr_idx].vector);
2494 		devm_free_irq(ice_pf_to_dev(pf),
2495 			      pf->msix_entries[pf->oicr_idx].vector, pf);
2496 	}
2497 
2498 	pf->num_avail_sw_msix += 1;
2499 	ice_free_res(pf->irq_tracker, pf->oicr_idx, ICE_RES_MISC_VEC_ID);
2500 }
2501 
2502 /**
2503  * ice_ena_ctrlq_interrupts - enable control queue interrupts
2504  * @hw: pointer to HW structure
2505  * @reg_idx: HW vector index to associate the control queue interrupts with
2506  */
2507 static void ice_ena_ctrlq_interrupts(struct ice_hw *hw, u16 reg_idx)
2508 {
2509 	u32 val;
2510 
2511 	val = ((reg_idx & PFINT_OICR_CTL_MSIX_INDX_M) |
2512 	       PFINT_OICR_CTL_CAUSE_ENA_M);
2513 	wr32(hw, PFINT_OICR_CTL, val);
2514 
2515 	/* enable Admin queue Interrupt causes */
2516 	val = ((reg_idx & PFINT_FW_CTL_MSIX_INDX_M) |
2517 	       PFINT_FW_CTL_CAUSE_ENA_M);
2518 	wr32(hw, PFINT_FW_CTL, val);
2519 
2520 	/* enable Mailbox queue Interrupt causes */
2521 	val = ((reg_idx & PFINT_MBX_CTL_MSIX_INDX_M) |
2522 	       PFINT_MBX_CTL_CAUSE_ENA_M);
2523 	wr32(hw, PFINT_MBX_CTL, val);
2524 
2525 	ice_flush(hw);
2526 }
2527 
2528 /**
2529  * ice_req_irq_msix_misc - Setup the misc vector to handle non queue events
2530  * @pf: board private structure
2531  *
2532  * This sets up the handler for MSIX 0, which is used to manage the
2533  * non-queue interrupts, e.g. AdminQ and errors. This is not used
2534  * when in MSI or Legacy interrupt mode.
2535  */
2536 static int ice_req_irq_msix_misc(struct ice_pf *pf)
2537 {
2538 	struct device *dev = ice_pf_to_dev(pf);
2539 	struct ice_hw *hw = &pf->hw;
2540 	int oicr_idx, err = 0;
2541 
2542 	if (!pf->int_name[0])
2543 		snprintf(pf->int_name, sizeof(pf->int_name) - 1, "%s-%s:misc",
2544 			 dev_driver_string(dev), dev_name(dev));
2545 
2546 	/* Do not request IRQ but do enable OICR interrupt since settings are
2547 	 * lost during reset. Note that this function is called only during
2548 	 * rebuild path and not while reset is in progress.
2549 	 */
2550 	if (ice_is_reset_in_progress(pf->state))
2551 		goto skip_req_irq;
2552 
2553 	/* reserve one vector in irq_tracker for misc interrupts */
2554 	oicr_idx = ice_get_res(pf, pf->irq_tracker, 1, ICE_RES_MISC_VEC_ID);
2555 	if (oicr_idx < 0)
2556 		return oicr_idx;
2557 
2558 	pf->num_avail_sw_msix -= 1;
2559 	pf->oicr_idx = (u16)oicr_idx;
2560 
2561 	err = devm_request_irq(dev, pf->msix_entries[pf->oicr_idx].vector,
2562 			       ice_misc_intr, 0, pf->int_name, pf);
2563 	if (err) {
2564 		dev_err(dev, "devm_request_irq for %s failed: %d\n",
2565 			pf->int_name, err);
2566 		ice_free_res(pf->irq_tracker, 1, ICE_RES_MISC_VEC_ID);
2567 		pf->num_avail_sw_msix += 1;
2568 		return err;
2569 	}
2570 
2571 skip_req_irq:
2572 	ice_ena_misc_vector(pf);
2573 
2574 	ice_ena_ctrlq_interrupts(hw, pf->oicr_idx);
2575 	wr32(hw, GLINT_ITR(ICE_RX_ITR, pf->oicr_idx),
2576 	     ITR_REG_ALIGN(ICE_ITR_8K) >> ICE_ITR_GRAN_S);
2577 
2578 	ice_flush(hw);
2579 	ice_irq_dynamic_ena(hw, NULL, NULL);
2580 
2581 	return 0;
2582 }
2583 
2584 /**
2585  * ice_napi_add - register NAPI handler for the VSI
2586  * @vsi: VSI for which NAPI handler is to be registered
2587  *
2588  * This function is only called in the driver's load path. Registering the NAPI
2589  * handler is done in ice_vsi_alloc_q_vector() for all other cases (i.e. resume,
2590  * reset/rebuild, etc.)
2591  */
2592 static void ice_napi_add(struct ice_vsi *vsi)
2593 {
2594 	int v_idx;
2595 
2596 	if (!vsi->netdev)
2597 		return;
2598 
2599 	ice_for_each_q_vector(vsi, v_idx)
2600 		netif_napi_add(vsi->netdev, &vsi->q_vectors[v_idx]->napi,
2601 			       ice_napi_poll, NAPI_POLL_WEIGHT);
2602 }
2603 
2604 /**
2605  * ice_set_ops - set netdev and ethtools ops for the given netdev
2606  * @netdev: netdev instance
2607  */
2608 static void ice_set_ops(struct net_device *netdev)
2609 {
2610 	struct ice_pf *pf = ice_netdev_to_pf(netdev);
2611 
2612 	if (ice_is_safe_mode(pf)) {
2613 		netdev->netdev_ops = &ice_netdev_safe_mode_ops;
2614 		ice_set_ethtool_safe_mode_ops(netdev);
2615 		return;
2616 	}
2617 
2618 	netdev->netdev_ops = &ice_netdev_ops;
2619 	ice_set_ethtool_ops(netdev);
2620 }
2621 
2622 /**
2623  * ice_set_netdev_features - set features for the given netdev
2624  * @netdev: netdev instance
2625  */
2626 static void ice_set_netdev_features(struct net_device *netdev)
2627 {
2628 	struct ice_pf *pf = ice_netdev_to_pf(netdev);
2629 	netdev_features_t csumo_features;
2630 	netdev_features_t vlano_features;
2631 	netdev_features_t dflt_features;
2632 	netdev_features_t tso_features;
2633 
2634 	if (ice_is_safe_mode(pf)) {
2635 		/* safe mode */
2636 		netdev->features = NETIF_F_SG | NETIF_F_HIGHDMA;
2637 		netdev->hw_features = netdev->features;
2638 		return;
2639 	}
2640 
2641 	dflt_features = NETIF_F_SG	|
2642 			NETIF_F_HIGHDMA	|
2643 			NETIF_F_NTUPLE	|
2644 			NETIF_F_RXHASH;
2645 
2646 	csumo_features = NETIF_F_RXCSUM	  |
2647 			 NETIF_F_IP_CSUM  |
2648 			 NETIF_F_SCTP_CRC |
2649 			 NETIF_F_IPV6_CSUM;
2650 
2651 	vlano_features = NETIF_F_HW_VLAN_CTAG_FILTER |
2652 			 NETIF_F_HW_VLAN_CTAG_TX     |
2653 			 NETIF_F_HW_VLAN_CTAG_RX;
2654 
2655 	tso_features = NETIF_F_TSO			|
2656 		       NETIF_F_TSO_ECN			|
2657 		       NETIF_F_TSO6			|
2658 		       NETIF_F_GSO_GRE			|
2659 		       NETIF_F_GSO_UDP_TUNNEL		|
2660 		       NETIF_F_GSO_GRE_CSUM		|
2661 		       NETIF_F_GSO_UDP_TUNNEL_CSUM	|
2662 		       NETIF_F_GSO_PARTIAL		|
2663 		       NETIF_F_GSO_IPXIP4		|
2664 		       NETIF_F_GSO_IPXIP6		|
2665 		       NETIF_F_GSO_UDP_L4;
2666 
2667 	netdev->gso_partial_features |= NETIF_F_GSO_UDP_TUNNEL_CSUM |
2668 					NETIF_F_GSO_GRE_CSUM;
2669 	/* set features that user can change */
2670 	netdev->hw_features = dflt_features | csumo_features |
2671 			      vlano_features | tso_features;
2672 
2673 	/* add support for HW_CSUM on packets with MPLS header */
2674 	netdev->mpls_features =  NETIF_F_HW_CSUM;
2675 
2676 	/* enable features */
2677 	netdev->features |= netdev->hw_features;
2678 	/* encap and VLAN devices inherit default, csumo and tso features */
2679 	netdev->hw_enc_features |= dflt_features | csumo_features |
2680 				   tso_features;
2681 	netdev->vlan_features |= dflt_features | csumo_features |
2682 				 tso_features;
2683 }
2684 
2685 /**
2686  * ice_cfg_netdev - Allocate, configure and register a netdev
2687  * @vsi: the VSI associated with the new netdev
2688  *
2689  * Returns 0 on success, negative value on failure
2690  */
2691 static int ice_cfg_netdev(struct ice_vsi *vsi)
2692 {
2693 	struct ice_pf *pf = vsi->back;
2694 	struct ice_netdev_priv *np;
2695 	struct net_device *netdev;
2696 	u8 mac_addr[ETH_ALEN];
2697 	int err;
2698 
2699 	err = ice_devlink_create_port(pf);
2700 	if (err)
2701 		return err;
2702 
2703 	netdev = alloc_etherdev_mqs(sizeof(*np), vsi->alloc_txq,
2704 				    vsi->alloc_rxq);
2705 	if (!netdev) {
2706 		err = -ENOMEM;
2707 		goto err_destroy_devlink_port;
2708 	}
2709 
2710 	vsi->netdev = netdev;
2711 	np = netdev_priv(netdev);
2712 	np->vsi = vsi;
2713 
2714 	ice_set_netdev_features(netdev);
2715 
2716 	ice_set_ops(netdev);
2717 
2718 	if (vsi->type == ICE_VSI_PF) {
2719 		SET_NETDEV_DEV(netdev, ice_pf_to_dev(pf));
2720 		ether_addr_copy(mac_addr, vsi->port_info->mac.perm_addr);
2721 		ether_addr_copy(netdev->dev_addr, mac_addr);
2722 		ether_addr_copy(netdev->perm_addr, mac_addr);
2723 	}
2724 
2725 	netdev->priv_flags |= IFF_UNICAST_FLT;
2726 
2727 	/* Setup netdev TC information */
2728 	ice_vsi_cfg_netdev_tc(vsi, vsi->tc_cfg.ena_tc);
2729 
2730 	/* setup watchdog timeout value to be 5 second */
2731 	netdev->watchdog_timeo = 5 * HZ;
2732 
2733 	netdev->min_mtu = ETH_MIN_MTU;
2734 	netdev->max_mtu = ICE_MAX_MTU;
2735 
2736 	err = register_netdev(vsi->netdev);
2737 	if (err)
2738 		goto err_free_netdev;
2739 
2740 	devlink_port_type_eth_set(&pf->devlink_port, vsi->netdev);
2741 
2742 	netif_carrier_off(vsi->netdev);
2743 
2744 	/* make sure transmit queues start off as stopped */
2745 	netif_tx_stop_all_queues(vsi->netdev);
2746 
2747 	return 0;
2748 
2749 err_free_netdev:
2750 	free_netdev(vsi->netdev);
2751 	vsi->netdev = NULL;
2752 err_destroy_devlink_port:
2753 	ice_devlink_destroy_port(pf);
2754 	return err;
2755 }
2756 
2757 /**
2758  * ice_fill_rss_lut - Fill the RSS lookup table with default values
2759  * @lut: Lookup table
2760  * @rss_table_size: Lookup table size
2761  * @rss_size: Range of queue number for hashing
2762  */
2763 void ice_fill_rss_lut(u8 *lut, u16 rss_table_size, u16 rss_size)
2764 {
2765 	u16 i;
2766 
2767 	for (i = 0; i < rss_table_size; i++)
2768 		lut[i] = i % rss_size;
2769 }
2770 
2771 /**
2772  * ice_pf_vsi_setup - Set up a PF VSI
2773  * @pf: board private structure
2774  * @pi: pointer to the port_info instance
2775  *
2776  * Returns pointer to the successfully allocated VSI software struct
2777  * on success, otherwise returns NULL on failure.
2778  */
2779 static struct ice_vsi *
2780 ice_pf_vsi_setup(struct ice_pf *pf, struct ice_port_info *pi)
2781 {
2782 	return ice_vsi_setup(pf, pi, ICE_VSI_PF, ICE_INVAL_VFID);
2783 }
2784 
2785 /**
2786  * ice_ctrl_vsi_setup - Set up a control VSI
2787  * @pf: board private structure
2788  * @pi: pointer to the port_info instance
2789  *
2790  * Returns pointer to the successfully allocated VSI software struct
2791  * on success, otherwise returns NULL on failure.
2792  */
2793 static struct ice_vsi *
2794 ice_ctrl_vsi_setup(struct ice_pf *pf, struct ice_port_info *pi)
2795 {
2796 	return ice_vsi_setup(pf, pi, ICE_VSI_CTRL, ICE_INVAL_VFID);
2797 }
2798 
2799 /**
2800  * ice_lb_vsi_setup - Set up a loopback VSI
2801  * @pf: board private structure
2802  * @pi: pointer to the port_info instance
2803  *
2804  * Returns pointer to the successfully allocated VSI software struct
2805  * on success, otherwise returns NULL on failure.
2806  */
2807 struct ice_vsi *
2808 ice_lb_vsi_setup(struct ice_pf *pf, struct ice_port_info *pi)
2809 {
2810 	return ice_vsi_setup(pf, pi, ICE_VSI_LB, ICE_INVAL_VFID);
2811 }
2812 
2813 /**
2814  * ice_vlan_rx_add_vid - Add a VLAN ID filter to HW offload
2815  * @netdev: network interface to be adjusted
2816  * @proto: unused protocol
2817  * @vid: VLAN ID to be added
2818  *
2819  * net_device_ops implementation for adding VLAN IDs
2820  */
2821 static int
2822 ice_vlan_rx_add_vid(struct net_device *netdev, __always_unused __be16 proto,
2823 		    u16 vid)
2824 {
2825 	struct ice_netdev_priv *np = netdev_priv(netdev);
2826 	struct ice_vsi *vsi = np->vsi;
2827 	int ret;
2828 
2829 	if (vid >= VLAN_N_VID) {
2830 		netdev_err(netdev, "VLAN id requested %d is out of range %d\n",
2831 			   vid, VLAN_N_VID);
2832 		return -EINVAL;
2833 	}
2834 
2835 	if (vsi->info.pvid)
2836 		return -EINVAL;
2837 
2838 	/* VLAN 0 is added by default during load/reset */
2839 	if (!vid)
2840 		return 0;
2841 
2842 	/* Enable VLAN pruning when a VLAN other than 0 is added */
2843 	if (!ice_vsi_is_vlan_pruning_ena(vsi)) {
2844 		ret = ice_cfg_vlan_pruning(vsi, true, false);
2845 		if (ret)
2846 			return ret;
2847 	}
2848 
2849 	/* Add a switch rule for this VLAN ID so its corresponding VLAN tagged
2850 	 * packets aren't pruned by the device's internal switch on Rx
2851 	 */
2852 	ret = ice_vsi_add_vlan(vsi, vid, ICE_FWD_TO_VSI);
2853 	if (!ret) {
2854 		vsi->vlan_ena = true;
2855 		set_bit(ICE_VSI_FLAG_VLAN_FLTR_CHANGED, vsi->flags);
2856 	}
2857 
2858 	return ret;
2859 }
2860 
2861 /**
2862  * ice_vlan_rx_kill_vid - Remove a VLAN ID filter from HW offload
2863  * @netdev: network interface to be adjusted
2864  * @proto: unused protocol
2865  * @vid: VLAN ID to be removed
2866  *
2867  * net_device_ops implementation for removing VLAN IDs
2868  */
2869 static int
2870 ice_vlan_rx_kill_vid(struct net_device *netdev, __always_unused __be16 proto,
2871 		     u16 vid)
2872 {
2873 	struct ice_netdev_priv *np = netdev_priv(netdev);
2874 	struct ice_vsi *vsi = np->vsi;
2875 	int ret;
2876 
2877 	if (vsi->info.pvid)
2878 		return -EINVAL;
2879 
2880 	/* don't allow removal of VLAN 0 */
2881 	if (!vid)
2882 		return 0;
2883 
2884 	/* Make sure ice_vsi_kill_vlan is successful before updating VLAN
2885 	 * information
2886 	 */
2887 	ret = ice_vsi_kill_vlan(vsi, vid);
2888 	if (ret)
2889 		return ret;
2890 
2891 	/* Disable pruning when VLAN 0 is the only VLAN rule */
2892 	if (vsi->num_vlan == 1 && ice_vsi_is_vlan_pruning_ena(vsi))
2893 		ret = ice_cfg_vlan_pruning(vsi, false, false);
2894 
2895 	vsi->vlan_ena = false;
2896 	set_bit(ICE_VSI_FLAG_VLAN_FLTR_CHANGED, vsi->flags);
2897 	return ret;
2898 }
2899 
2900 /**
2901  * ice_setup_pf_sw - Setup the HW switch on startup or after reset
2902  * @pf: board private structure
2903  *
2904  * Returns 0 on success, negative value on failure
2905  */
2906 static int ice_setup_pf_sw(struct ice_pf *pf)
2907 {
2908 	struct ice_vsi *vsi;
2909 	int status = 0;
2910 
2911 	if (ice_is_reset_in_progress(pf->state))
2912 		return -EBUSY;
2913 
2914 	vsi = ice_pf_vsi_setup(pf, pf->hw.port_info);
2915 	if (!vsi) {
2916 		status = -ENOMEM;
2917 		goto unroll_vsi_setup;
2918 	}
2919 
2920 	status = ice_cfg_netdev(vsi);
2921 	if (status) {
2922 		status = -ENODEV;
2923 		goto unroll_vsi_setup;
2924 	}
2925 	/* netdev has to be configured before setting frame size */
2926 	ice_vsi_cfg_frame_size(vsi);
2927 
2928 	/* Setup DCB netlink interface */
2929 	ice_dcbnl_setup(vsi);
2930 
2931 	/* registering the NAPI handler requires both the queues and
2932 	 * netdev to be created, which are done in ice_pf_vsi_setup()
2933 	 * and ice_cfg_netdev() respectively
2934 	 */
2935 	ice_napi_add(vsi);
2936 
2937 	status = ice_set_cpu_rx_rmap(vsi);
2938 	if (status) {
2939 		dev_err(ice_pf_to_dev(pf), "Failed to set CPU Rx map VSI %d error %d\n",
2940 			vsi->vsi_num, status);
2941 		status = -EINVAL;
2942 		goto unroll_napi_add;
2943 	}
2944 	status = ice_init_mac_fltr(pf);
2945 	if (status)
2946 		goto free_cpu_rx_map;
2947 
2948 	return status;
2949 
2950 free_cpu_rx_map:
2951 	ice_free_cpu_rx_rmap(vsi);
2952 
2953 unroll_napi_add:
2954 	if (vsi) {
2955 		ice_napi_del(vsi);
2956 		if (vsi->netdev) {
2957 			if (vsi->netdev->reg_state == NETREG_REGISTERED)
2958 				unregister_netdev(vsi->netdev);
2959 			free_netdev(vsi->netdev);
2960 			vsi->netdev = NULL;
2961 		}
2962 	}
2963 
2964 unroll_vsi_setup:
2965 	if (vsi) {
2966 		ice_vsi_free_q_vectors(vsi);
2967 		ice_vsi_delete(vsi);
2968 		ice_vsi_put_qs(vsi);
2969 		ice_vsi_clear(vsi);
2970 	}
2971 	return status;
2972 }
2973 
2974 /**
2975  * ice_get_avail_q_count - Get count of queues in use
2976  * @pf_qmap: bitmap to get queue use count from
2977  * @lock: pointer to a mutex that protects access to pf_qmap
2978  * @size: size of the bitmap
2979  */
2980 static u16
2981 ice_get_avail_q_count(unsigned long *pf_qmap, struct mutex *lock, u16 size)
2982 {
2983 	unsigned long bit;
2984 	u16 count = 0;
2985 
2986 	mutex_lock(lock);
2987 	for_each_clear_bit(bit, pf_qmap, size)
2988 		count++;
2989 	mutex_unlock(lock);
2990 
2991 	return count;
2992 }
2993 
2994 /**
2995  * ice_get_avail_txq_count - Get count of Tx queues in use
2996  * @pf: pointer to an ice_pf instance
2997  */
2998 u16 ice_get_avail_txq_count(struct ice_pf *pf)
2999 {
3000 	return ice_get_avail_q_count(pf->avail_txqs, &pf->avail_q_mutex,
3001 				     pf->max_pf_txqs);
3002 }
3003 
3004 /**
3005  * ice_get_avail_rxq_count - Get count of Rx queues in use
3006  * @pf: pointer to an ice_pf instance
3007  */
3008 u16 ice_get_avail_rxq_count(struct ice_pf *pf)
3009 {
3010 	return ice_get_avail_q_count(pf->avail_rxqs, &pf->avail_q_mutex,
3011 				     pf->max_pf_rxqs);
3012 }
3013 
3014 /**
3015  * ice_deinit_pf - Unrolls initialziations done by ice_init_pf
3016  * @pf: board private structure to initialize
3017  */
3018 static void ice_deinit_pf(struct ice_pf *pf)
3019 {
3020 	ice_service_task_stop(pf);
3021 	mutex_destroy(&pf->sw_mutex);
3022 	mutex_destroy(&pf->tc_mutex);
3023 	mutex_destroy(&pf->avail_q_mutex);
3024 
3025 	if (pf->avail_txqs) {
3026 		bitmap_free(pf->avail_txqs);
3027 		pf->avail_txqs = NULL;
3028 	}
3029 
3030 	if (pf->avail_rxqs) {
3031 		bitmap_free(pf->avail_rxqs);
3032 		pf->avail_rxqs = NULL;
3033 	}
3034 }
3035 
3036 /**
3037  * ice_set_pf_caps - set PFs capability flags
3038  * @pf: pointer to the PF instance
3039  */
3040 static void ice_set_pf_caps(struct ice_pf *pf)
3041 {
3042 	struct ice_hw_func_caps *func_caps = &pf->hw.func_caps;
3043 
3044 	clear_bit(ICE_FLAG_DCB_CAPABLE, pf->flags);
3045 	if (func_caps->common_cap.dcb)
3046 		set_bit(ICE_FLAG_DCB_CAPABLE, pf->flags);
3047 	clear_bit(ICE_FLAG_SRIOV_CAPABLE, pf->flags);
3048 	if (func_caps->common_cap.sr_iov_1_1) {
3049 		set_bit(ICE_FLAG_SRIOV_CAPABLE, pf->flags);
3050 		pf->num_vfs_supported = min_t(int, func_caps->num_allocd_vfs,
3051 					      ICE_MAX_VF_COUNT);
3052 	}
3053 	clear_bit(ICE_FLAG_RSS_ENA, pf->flags);
3054 	if (func_caps->common_cap.rss_table_size)
3055 		set_bit(ICE_FLAG_RSS_ENA, pf->flags);
3056 
3057 	clear_bit(ICE_FLAG_FD_ENA, pf->flags);
3058 	if (func_caps->fd_fltr_guar > 0 || func_caps->fd_fltr_best_effort > 0) {
3059 		u16 unused;
3060 
3061 		/* ctrl_vsi_idx will be set to a valid value when flow director
3062 		 * is setup by ice_init_fdir
3063 		 */
3064 		pf->ctrl_vsi_idx = ICE_NO_VSI;
3065 		set_bit(ICE_FLAG_FD_ENA, pf->flags);
3066 		/* force guaranteed filter pool for PF */
3067 		ice_alloc_fd_guar_item(&pf->hw, &unused,
3068 				       func_caps->fd_fltr_guar);
3069 		/* force shared filter pool for PF */
3070 		ice_alloc_fd_shrd_item(&pf->hw, &unused,
3071 				       func_caps->fd_fltr_best_effort);
3072 	}
3073 
3074 	pf->max_pf_txqs = func_caps->common_cap.num_txq;
3075 	pf->max_pf_rxqs = func_caps->common_cap.num_rxq;
3076 }
3077 
3078 /**
3079  * ice_init_pf - Initialize general software structures (struct ice_pf)
3080  * @pf: board private structure to initialize
3081  */
3082 static int ice_init_pf(struct ice_pf *pf)
3083 {
3084 	ice_set_pf_caps(pf);
3085 
3086 	mutex_init(&pf->sw_mutex);
3087 	mutex_init(&pf->tc_mutex);
3088 
3089 	/* setup service timer and periodic service task */
3090 	timer_setup(&pf->serv_tmr, ice_service_timer, 0);
3091 	pf->serv_tmr_period = HZ;
3092 	INIT_WORK(&pf->serv_task, ice_service_task);
3093 	clear_bit(__ICE_SERVICE_SCHED, pf->state);
3094 
3095 	mutex_init(&pf->avail_q_mutex);
3096 	pf->avail_txqs = bitmap_zalloc(pf->max_pf_txqs, GFP_KERNEL);
3097 	if (!pf->avail_txqs)
3098 		return -ENOMEM;
3099 
3100 	pf->avail_rxqs = bitmap_zalloc(pf->max_pf_rxqs, GFP_KERNEL);
3101 	if (!pf->avail_rxqs) {
3102 		devm_kfree(ice_pf_to_dev(pf), pf->avail_txqs);
3103 		pf->avail_txqs = NULL;
3104 		return -ENOMEM;
3105 	}
3106 
3107 	return 0;
3108 }
3109 
3110 /**
3111  * ice_ena_msix_range - Request a range of MSIX vectors from the OS
3112  * @pf: board private structure
3113  *
3114  * compute the number of MSIX vectors required (v_budget) and request from
3115  * the OS. Return the number of vectors reserved or negative on failure
3116  */
3117 static int ice_ena_msix_range(struct ice_pf *pf)
3118 {
3119 	struct device *dev = ice_pf_to_dev(pf);
3120 	int v_left, v_actual, v_budget = 0;
3121 	int needed, err, i;
3122 
3123 	v_left = pf->hw.func_caps.common_cap.num_msix_vectors;
3124 
3125 	/* reserve one vector for miscellaneous handler */
3126 	needed = 1;
3127 	if (v_left < needed)
3128 		goto no_hw_vecs_left_err;
3129 	v_budget += needed;
3130 	v_left -= needed;
3131 
3132 	/* reserve vectors for LAN traffic */
3133 	needed = min_t(int, num_online_cpus(), v_left);
3134 	if (v_left < needed)
3135 		goto no_hw_vecs_left_err;
3136 	pf->num_lan_msix = needed;
3137 	v_budget += needed;
3138 	v_left -= needed;
3139 
3140 	/* reserve one vector for flow director */
3141 	if (test_bit(ICE_FLAG_FD_ENA, pf->flags)) {
3142 		needed = ICE_FDIR_MSIX;
3143 		if (v_left < needed)
3144 			goto no_hw_vecs_left_err;
3145 		v_budget += needed;
3146 		v_left -= needed;
3147 	}
3148 
3149 	pf->msix_entries = devm_kcalloc(dev, v_budget,
3150 					sizeof(*pf->msix_entries), GFP_KERNEL);
3151 
3152 	if (!pf->msix_entries) {
3153 		err = -ENOMEM;
3154 		goto exit_err;
3155 	}
3156 
3157 	for (i = 0; i < v_budget; i++)
3158 		pf->msix_entries[i].entry = i;
3159 
3160 	/* actually reserve the vectors */
3161 	v_actual = pci_enable_msix_range(pf->pdev, pf->msix_entries,
3162 					 ICE_MIN_MSIX, v_budget);
3163 
3164 	if (v_actual < 0) {
3165 		dev_err(dev, "unable to reserve MSI-X vectors\n");
3166 		err = v_actual;
3167 		goto msix_err;
3168 	}
3169 
3170 	if (v_actual < v_budget) {
3171 		dev_warn(dev, "not enough OS MSI-X vectors. requested = %d, obtained = %d\n",
3172 			 v_budget, v_actual);
3173 /* 2 vectors each for LAN and RDMA (traffic + OICR), one for flow director */
3174 #define ICE_MIN_LAN_VECS 2
3175 #define ICE_MIN_RDMA_VECS 2
3176 #define ICE_MIN_VECS (ICE_MIN_LAN_VECS + ICE_MIN_RDMA_VECS + 1)
3177 
3178 		if (v_actual < ICE_MIN_LAN_VECS) {
3179 			/* error if we can't get minimum vectors */
3180 			pci_disable_msix(pf->pdev);
3181 			err = -ERANGE;
3182 			goto msix_err;
3183 		} else {
3184 			pf->num_lan_msix = ICE_MIN_LAN_VECS;
3185 		}
3186 	}
3187 
3188 	return v_actual;
3189 
3190 msix_err:
3191 	devm_kfree(dev, pf->msix_entries);
3192 	goto exit_err;
3193 
3194 no_hw_vecs_left_err:
3195 	dev_err(dev, "not enough device MSI-X vectors. requested = %d, available = %d\n",
3196 		needed, v_left);
3197 	err = -ERANGE;
3198 exit_err:
3199 	pf->num_lan_msix = 0;
3200 	return err;
3201 }
3202 
3203 /**
3204  * ice_dis_msix - Disable MSI-X interrupt setup in OS
3205  * @pf: board private structure
3206  */
3207 static void ice_dis_msix(struct ice_pf *pf)
3208 {
3209 	pci_disable_msix(pf->pdev);
3210 	devm_kfree(ice_pf_to_dev(pf), pf->msix_entries);
3211 	pf->msix_entries = NULL;
3212 }
3213 
3214 /**
3215  * ice_clear_interrupt_scheme - Undo things done by ice_init_interrupt_scheme
3216  * @pf: board private structure
3217  */
3218 static void ice_clear_interrupt_scheme(struct ice_pf *pf)
3219 {
3220 	ice_dis_msix(pf);
3221 
3222 	if (pf->irq_tracker) {
3223 		devm_kfree(ice_pf_to_dev(pf), pf->irq_tracker);
3224 		pf->irq_tracker = NULL;
3225 	}
3226 }
3227 
3228 /**
3229  * ice_init_interrupt_scheme - Determine proper interrupt scheme
3230  * @pf: board private structure to initialize
3231  */
3232 static int ice_init_interrupt_scheme(struct ice_pf *pf)
3233 {
3234 	int vectors;
3235 
3236 	vectors = ice_ena_msix_range(pf);
3237 
3238 	if (vectors < 0)
3239 		return vectors;
3240 
3241 	/* set up vector assignment tracking */
3242 	pf->irq_tracker =
3243 		devm_kzalloc(ice_pf_to_dev(pf), sizeof(*pf->irq_tracker) +
3244 			     (sizeof(u16) * vectors), GFP_KERNEL);
3245 	if (!pf->irq_tracker) {
3246 		ice_dis_msix(pf);
3247 		return -ENOMEM;
3248 	}
3249 
3250 	/* populate SW interrupts pool with number of OS granted IRQs. */
3251 	pf->num_avail_sw_msix = (u16)vectors;
3252 	pf->irq_tracker->num_entries = (u16)vectors;
3253 	pf->irq_tracker->end = pf->irq_tracker->num_entries;
3254 
3255 	return 0;
3256 }
3257 
3258 /**
3259  * ice_is_wol_supported - get NVM state of WoL
3260  * @pf: board private structure
3261  *
3262  * Check if WoL is supported based on the HW configuration.
3263  * Returns true if NVM supports and enables WoL for this port, false otherwise
3264  */
3265 bool ice_is_wol_supported(struct ice_pf *pf)
3266 {
3267 	struct ice_hw *hw = &pf->hw;
3268 	u16 wol_ctrl;
3269 
3270 	/* A bit set to 1 in the NVM Software Reserved Word 2 (WoL control
3271 	 * word) indicates WoL is not supported on the corresponding PF ID.
3272 	 */
3273 	if (ice_read_sr_word(hw, ICE_SR_NVM_WOL_CFG, &wol_ctrl))
3274 		return false;
3275 
3276 	return !(BIT(hw->pf_id) & wol_ctrl);
3277 }
3278 
3279 /**
3280  * ice_vsi_recfg_qs - Change the number of queues on a VSI
3281  * @vsi: VSI being changed
3282  * @new_rx: new number of Rx queues
3283  * @new_tx: new number of Tx queues
3284  *
3285  * Only change the number of queues if new_tx, or new_rx is non-0.
3286  *
3287  * Returns 0 on success.
3288  */
3289 int ice_vsi_recfg_qs(struct ice_vsi *vsi, int new_rx, int new_tx)
3290 {
3291 	struct ice_pf *pf = vsi->back;
3292 	int err = 0, timeout = 50;
3293 
3294 	if (!new_rx && !new_tx)
3295 		return -EINVAL;
3296 
3297 	while (test_and_set_bit(__ICE_CFG_BUSY, pf->state)) {
3298 		timeout--;
3299 		if (!timeout)
3300 			return -EBUSY;
3301 		usleep_range(1000, 2000);
3302 	}
3303 
3304 	if (new_tx)
3305 		vsi->req_txq = (u16)new_tx;
3306 	if (new_rx)
3307 		vsi->req_rxq = (u16)new_rx;
3308 
3309 	/* set for the next time the netdev is started */
3310 	if (!netif_running(vsi->netdev)) {
3311 		ice_vsi_rebuild(vsi, false);
3312 		dev_dbg(ice_pf_to_dev(pf), "Link is down, queue count change happens when link is brought up\n");
3313 		goto done;
3314 	}
3315 
3316 	ice_vsi_close(vsi);
3317 	ice_vsi_rebuild(vsi, false);
3318 	ice_pf_dcb_recfg(pf);
3319 	ice_vsi_open(vsi);
3320 done:
3321 	clear_bit(__ICE_CFG_BUSY, pf->state);
3322 	return err;
3323 }
3324 
3325 /**
3326  * ice_log_pkg_init - log result of DDP package load
3327  * @hw: pointer to hardware info
3328  * @status: status of package load
3329  */
3330 static void
3331 ice_log_pkg_init(struct ice_hw *hw, enum ice_status *status)
3332 {
3333 	struct ice_pf *pf = (struct ice_pf *)hw->back;
3334 	struct device *dev = ice_pf_to_dev(pf);
3335 
3336 	switch (*status) {
3337 	case ICE_SUCCESS:
3338 		/* The package download AdminQ command returned success because
3339 		 * this download succeeded or ICE_ERR_AQ_NO_WORK since there is
3340 		 * already a package loaded on the device.
3341 		 */
3342 		if (hw->pkg_ver.major == hw->active_pkg_ver.major &&
3343 		    hw->pkg_ver.minor == hw->active_pkg_ver.minor &&
3344 		    hw->pkg_ver.update == hw->active_pkg_ver.update &&
3345 		    hw->pkg_ver.draft == hw->active_pkg_ver.draft &&
3346 		    !memcmp(hw->pkg_name, hw->active_pkg_name,
3347 			    sizeof(hw->pkg_name))) {
3348 			if (hw->pkg_dwnld_status == ICE_AQ_RC_EEXIST)
3349 				dev_info(dev, "DDP package already present on device: %s version %d.%d.%d.%d\n",
3350 					 hw->active_pkg_name,
3351 					 hw->active_pkg_ver.major,
3352 					 hw->active_pkg_ver.minor,
3353 					 hw->active_pkg_ver.update,
3354 					 hw->active_pkg_ver.draft);
3355 			else
3356 				dev_info(dev, "The DDP package was successfully loaded: %s version %d.%d.%d.%d\n",
3357 					 hw->active_pkg_name,
3358 					 hw->active_pkg_ver.major,
3359 					 hw->active_pkg_ver.minor,
3360 					 hw->active_pkg_ver.update,
3361 					 hw->active_pkg_ver.draft);
3362 		} else if (hw->active_pkg_ver.major != ICE_PKG_SUPP_VER_MAJ ||
3363 			   hw->active_pkg_ver.minor != ICE_PKG_SUPP_VER_MNR) {
3364 			dev_err(dev, "The device has a DDP package that is not supported by the driver.  The device has package '%s' version %d.%d.x.x.  The driver requires version %d.%d.x.x.  Entering Safe Mode.\n",
3365 				hw->active_pkg_name,
3366 				hw->active_pkg_ver.major,
3367 				hw->active_pkg_ver.minor,
3368 				ICE_PKG_SUPP_VER_MAJ, ICE_PKG_SUPP_VER_MNR);
3369 			*status = ICE_ERR_NOT_SUPPORTED;
3370 		} else if (hw->active_pkg_ver.major == ICE_PKG_SUPP_VER_MAJ &&
3371 			   hw->active_pkg_ver.minor == ICE_PKG_SUPP_VER_MNR) {
3372 			dev_info(dev, "The driver could not load the DDP package file because a compatible DDP package is already present on the device.  The device has package '%s' version %d.%d.%d.%d.  The package file found by the driver: '%s' version %d.%d.%d.%d.\n",
3373 				 hw->active_pkg_name,
3374 				 hw->active_pkg_ver.major,
3375 				 hw->active_pkg_ver.minor,
3376 				 hw->active_pkg_ver.update,
3377 				 hw->active_pkg_ver.draft,
3378 				 hw->pkg_name,
3379 				 hw->pkg_ver.major,
3380 				 hw->pkg_ver.minor,
3381 				 hw->pkg_ver.update,
3382 				 hw->pkg_ver.draft);
3383 		} else {
3384 			dev_err(dev, "An unknown error occurred when loading the DDP package, please reboot the system.  If the problem persists, update the NVM.  Entering Safe Mode.\n");
3385 			*status = ICE_ERR_NOT_SUPPORTED;
3386 		}
3387 		break;
3388 	case ICE_ERR_FW_DDP_MISMATCH:
3389 		dev_err(dev, "The firmware loaded on the device is not compatible with the DDP package.  Please update the device's NVM.  Entering safe mode.\n");
3390 		break;
3391 	case ICE_ERR_BUF_TOO_SHORT:
3392 	case ICE_ERR_CFG:
3393 		dev_err(dev, "The DDP package file is invalid. Entering Safe Mode.\n");
3394 		break;
3395 	case ICE_ERR_NOT_SUPPORTED:
3396 		/* Package File version not supported */
3397 		if (hw->pkg_ver.major > ICE_PKG_SUPP_VER_MAJ ||
3398 		    (hw->pkg_ver.major == ICE_PKG_SUPP_VER_MAJ &&
3399 		     hw->pkg_ver.minor > ICE_PKG_SUPP_VER_MNR))
3400 			dev_err(dev, "The DDP package file version is higher than the driver supports.  Please use an updated driver.  Entering Safe Mode.\n");
3401 		else if (hw->pkg_ver.major < ICE_PKG_SUPP_VER_MAJ ||
3402 			 (hw->pkg_ver.major == ICE_PKG_SUPP_VER_MAJ &&
3403 			  hw->pkg_ver.minor < ICE_PKG_SUPP_VER_MNR))
3404 			dev_err(dev, "The DDP package file version is lower than the driver supports.  The driver requires version %d.%d.x.x.  Please use an updated DDP Package file.  Entering Safe Mode.\n",
3405 				ICE_PKG_SUPP_VER_MAJ, ICE_PKG_SUPP_VER_MNR);
3406 		break;
3407 	case ICE_ERR_AQ_ERROR:
3408 		switch (hw->pkg_dwnld_status) {
3409 		case ICE_AQ_RC_ENOSEC:
3410 		case ICE_AQ_RC_EBADSIG:
3411 			dev_err(dev, "The DDP package could not be loaded because its signature is not valid.  Please use a valid DDP Package.  Entering Safe Mode.\n");
3412 			return;
3413 		case ICE_AQ_RC_ESVN:
3414 			dev_err(dev, "The DDP Package could not be loaded because its security revision is too low.  Please use an updated DDP Package.  Entering Safe Mode.\n");
3415 			return;
3416 		case ICE_AQ_RC_EBADMAN:
3417 		case ICE_AQ_RC_EBADBUF:
3418 			dev_err(dev, "An error occurred on the device while loading the DDP package.  The device will be reset.\n");
3419 			/* poll for reset to complete */
3420 			if (ice_check_reset(hw))
3421 				dev_err(dev, "Error resetting device. Please reload the driver\n");
3422 			return;
3423 		default:
3424 			break;
3425 		}
3426 		fallthrough;
3427 	default:
3428 		dev_err(dev, "An unknown error (%d) occurred when loading the DDP package.  Entering Safe Mode.\n",
3429 			*status);
3430 		break;
3431 	}
3432 }
3433 
3434 /**
3435  * ice_load_pkg - load/reload the DDP Package file
3436  * @firmware: firmware structure when firmware requested or NULL for reload
3437  * @pf: pointer to the PF instance
3438  *
3439  * Called on probe and post CORER/GLOBR rebuild to load DDP Package and
3440  * initialize HW tables.
3441  */
3442 static void
3443 ice_load_pkg(const struct firmware *firmware, struct ice_pf *pf)
3444 {
3445 	enum ice_status status = ICE_ERR_PARAM;
3446 	struct device *dev = ice_pf_to_dev(pf);
3447 	struct ice_hw *hw = &pf->hw;
3448 
3449 	/* Load DDP Package */
3450 	if (firmware && !hw->pkg_copy) {
3451 		status = ice_copy_and_init_pkg(hw, firmware->data,
3452 					       firmware->size);
3453 		ice_log_pkg_init(hw, &status);
3454 	} else if (!firmware && hw->pkg_copy) {
3455 		/* Reload package during rebuild after CORER/GLOBR reset */
3456 		status = ice_init_pkg(hw, hw->pkg_copy, hw->pkg_size);
3457 		ice_log_pkg_init(hw, &status);
3458 	} else {
3459 		dev_err(dev, "The DDP package file failed to load. Entering Safe Mode.\n");
3460 	}
3461 
3462 	if (status) {
3463 		/* Safe Mode */
3464 		clear_bit(ICE_FLAG_ADV_FEATURES, pf->flags);
3465 		return;
3466 	}
3467 
3468 	/* Successful download package is the precondition for advanced
3469 	 * features, hence setting the ICE_FLAG_ADV_FEATURES flag
3470 	 */
3471 	set_bit(ICE_FLAG_ADV_FEATURES, pf->flags);
3472 }
3473 
3474 /**
3475  * ice_verify_cacheline_size - verify driver's assumption of 64 Byte cache lines
3476  * @pf: pointer to the PF structure
3477  *
3478  * There is no error returned here because the driver should be able to handle
3479  * 128 Byte cache lines, so we only print a warning in case issues are seen,
3480  * specifically with Tx.
3481  */
3482 static void ice_verify_cacheline_size(struct ice_pf *pf)
3483 {
3484 	if (rd32(&pf->hw, GLPCI_CNF2) & GLPCI_CNF2_CACHELINE_SIZE_M)
3485 		dev_warn(ice_pf_to_dev(pf), "%d Byte cache line assumption is invalid, driver may have Tx timeouts!\n",
3486 			 ICE_CACHE_LINE_BYTES);
3487 }
3488 
3489 /**
3490  * ice_send_version - update firmware with driver version
3491  * @pf: PF struct
3492  *
3493  * Returns ICE_SUCCESS on success, else error code
3494  */
3495 static enum ice_status ice_send_version(struct ice_pf *pf)
3496 {
3497 	struct ice_driver_ver dv;
3498 
3499 	dv.major_ver = 0xff;
3500 	dv.minor_ver = 0xff;
3501 	dv.build_ver = 0xff;
3502 	dv.subbuild_ver = 0;
3503 	strscpy((char *)dv.driver_string, UTS_RELEASE,
3504 		sizeof(dv.driver_string));
3505 	return ice_aq_send_driver_ver(&pf->hw, &dv, NULL);
3506 }
3507 
3508 /**
3509  * ice_init_fdir - Initialize flow director VSI and configuration
3510  * @pf: pointer to the PF instance
3511  *
3512  * returns 0 on success, negative on error
3513  */
3514 static int ice_init_fdir(struct ice_pf *pf)
3515 {
3516 	struct device *dev = ice_pf_to_dev(pf);
3517 	struct ice_vsi *ctrl_vsi;
3518 	int err;
3519 
3520 	/* Side Band Flow Director needs to have a control VSI.
3521 	 * Allocate it and store it in the PF.
3522 	 */
3523 	ctrl_vsi = ice_ctrl_vsi_setup(pf, pf->hw.port_info);
3524 	if (!ctrl_vsi) {
3525 		dev_dbg(dev, "could not create control VSI\n");
3526 		return -ENOMEM;
3527 	}
3528 
3529 	err = ice_vsi_open_ctrl(ctrl_vsi);
3530 	if (err) {
3531 		dev_dbg(dev, "could not open control VSI\n");
3532 		goto err_vsi_open;
3533 	}
3534 
3535 	mutex_init(&pf->hw.fdir_fltr_lock);
3536 
3537 	err = ice_fdir_create_dflt_rules(pf);
3538 	if (err)
3539 		goto err_fdir_rule;
3540 
3541 	return 0;
3542 
3543 err_fdir_rule:
3544 	ice_fdir_release_flows(&pf->hw);
3545 	ice_vsi_close(ctrl_vsi);
3546 err_vsi_open:
3547 	ice_vsi_release(ctrl_vsi);
3548 	if (pf->ctrl_vsi_idx != ICE_NO_VSI) {
3549 		pf->vsi[pf->ctrl_vsi_idx] = NULL;
3550 		pf->ctrl_vsi_idx = ICE_NO_VSI;
3551 	}
3552 	return err;
3553 }
3554 
3555 /**
3556  * ice_get_opt_fw_name - return optional firmware file name or NULL
3557  * @pf: pointer to the PF instance
3558  */
3559 static char *ice_get_opt_fw_name(struct ice_pf *pf)
3560 {
3561 	/* Optional firmware name same as default with additional dash
3562 	 * followed by a EUI-64 identifier (PCIe Device Serial Number)
3563 	 */
3564 	struct pci_dev *pdev = pf->pdev;
3565 	char *opt_fw_filename;
3566 	u64 dsn;
3567 
3568 	/* Determine the name of the optional file using the DSN (two
3569 	 * dwords following the start of the DSN Capability).
3570 	 */
3571 	dsn = pci_get_dsn(pdev);
3572 	if (!dsn)
3573 		return NULL;
3574 
3575 	opt_fw_filename = kzalloc(NAME_MAX, GFP_KERNEL);
3576 	if (!opt_fw_filename)
3577 		return NULL;
3578 
3579 	snprintf(opt_fw_filename, NAME_MAX, "%sice-%016llx.pkg",
3580 		 ICE_DDP_PKG_PATH, dsn);
3581 
3582 	return opt_fw_filename;
3583 }
3584 
3585 /**
3586  * ice_request_fw - Device initialization routine
3587  * @pf: pointer to the PF instance
3588  */
3589 static void ice_request_fw(struct ice_pf *pf)
3590 {
3591 	char *opt_fw_filename = ice_get_opt_fw_name(pf);
3592 	const struct firmware *firmware = NULL;
3593 	struct device *dev = ice_pf_to_dev(pf);
3594 	int err = 0;
3595 
3596 	/* optional device-specific DDP (if present) overrides the default DDP
3597 	 * package file. kernel logs a debug message if the file doesn't exist,
3598 	 * and warning messages for other errors.
3599 	 */
3600 	if (opt_fw_filename) {
3601 		err = firmware_request_nowarn(&firmware, opt_fw_filename, dev);
3602 		if (err) {
3603 			kfree(opt_fw_filename);
3604 			goto dflt_pkg_load;
3605 		}
3606 
3607 		/* request for firmware was successful. Download to device */
3608 		ice_load_pkg(firmware, pf);
3609 		kfree(opt_fw_filename);
3610 		release_firmware(firmware);
3611 		return;
3612 	}
3613 
3614 dflt_pkg_load:
3615 	err = request_firmware(&firmware, ICE_DDP_PKG_FILE, dev);
3616 	if (err) {
3617 		dev_err(dev, "The DDP package file was not found or could not be read. Entering Safe Mode\n");
3618 		return;
3619 	}
3620 
3621 	/* request for firmware was successful. Download to device */
3622 	ice_load_pkg(firmware, pf);
3623 	release_firmware(firmware);
3624 }
3625 
3626 /**
3627  * ice_print_wake_reason - show the wake up cause in the log
3628  * @pf: pointer to the PF struct
3629  */
3630 static void ice_print_wake_reason(struct ice_pf *pf)
3631 {
3632 	u32 wus = pf->wakeup_reason;
3633 	const char *wake_str;
3634 
3635 	/* if no wake event, nothing to print */
3636 	if (!wus)
3637 		return;
3638 
3639 	if (wus & PFPM_WUS_LNKC_M)
3640 		wake_str = "Link\n";
3641 	else if (wus & PFPM_WUS_MAG_M)
3642 		wake_str = "Magic Packet\n";
3643 	else if (wus & PFPM_WUS_MNG_M)
3644 		wake_str = "Management\n";
3645 	else if (wus & PFPM_WUS_FW_RST_WK_M)
3646 		wake_str = "Firmware Reset\n";
3647 	else
3648 		wake_str = "Unknown\n";
3649 
3650 	dev_info(ice_pf_to_dev(pf), "Wake reason: %s", wake_str);
3651 }
3652 
3653 /**
3654  * ice_probe - Device initialization routine
3655  * @pdev: PCI device information struct
3656  * @ent: entry in ice_pci_tbl
3657  *
3658  * Returns 0 on success, negative on failure
3659  */
3660 static int
3661 ice_probe(struct pci_dev *pdev, const struct pci_device_id __always_unused *ent)
3662 {
3663 	struct device *dev = &pdev->dev;
3664 	struct ice_pf *pf;
3665 	struct ice_hw *hw;
3666 	int err;
3667 
3668 	/* this driver uses devres, see
3669 	 * Documentation/driver-api/driver-model/devres.rst
3670 	 */
3671 	err = pcim_enable_device(pdev);
3672 	if (err)
3673 		return err;
3674 
3675 	err = pcim_iomap_regions(pdev, BIT(ICE_BAR0), pci_name(pdev));
3676 	if (err) {
3677 		dev_err(dev, "BAR0 I/O map error %d\n", err);
3678 		return err;
3679 	}
3680 
3681 	pf = ice_allocate_pf(dev);
3682 	if (!pf)
3683 		return -ENOMEM;
3684 
3685 	/* set up for high or low DMA */
3686 	err = dma_set_mask_and_coherent(dev, DMA_BIT_MASK(64));
3687 	if (err)
3688 		err = dma_set_mask_and_coherent(dev, DMA_BIT_MASK(32));
3689 	if (err) {
3690 		dev_err(dev, "DMA configuration failed: 0x%x\n", err);
3691 		return err;
3692 	}
3693 
3694 	pci_enable_pcie_error_reporting(pdev);
3695 	pci_set_master(pdev);
3696 
3697 	pf->pdev = pdev;
3698 	pci_set_drvdata(pdev, pf);
3699 	set_bit(__ICE_DOWN, pf->state);
3700 	/* Disable service task until DOWN bit is cleared */
3701 	set_bit(__ICE_SERVICE_DIS, pf->state);
3702 
3703 	hw = &pf->hw;
3704 	hw->hw_addr = pcim_iomap_table(pdev)[ICE_BAR0];
3705 	pci_save_state(pdev);
3706 
3707 	hw->back = pf;
3708 	hw->vendor_id = pdev->vendor;
3709 	hw->device_id = pdev->device;
3710 	pci_read_config_byte(pdev, PCI_REVISION_ID, &hw->revision_id);
3711 	hw->subsystem_vendor_id = pdev->subsystem_vendor;
3712 	hw->subsystem_device_id = pdev->subsystem_device;
3713 	hw->bus.device = PCI_SLOT(pdev->devfn);
3714 	hw->bus.func = PCI_FUNC(pdev->devfn);
3715 	ice_set_ctrlq_len(hw);
3716 
3717 	pf->msg_enable = netif_msg_init(debug, ICE_DFLT_NETIF_M);
3718 
3719 	err = ice_devlink_register(pf);
3720 	if (err) {
3721 		dev_err(dev, "ice_devlink_register failed: %d\n", err);
3722 		goto err_exit_unroll;
3723 	}
3724 
3725 #ifndef CONFIG_DYNAMIC_DEBUG
3726 	if (debug < -1)
3727 		hw->debug_mask = debug;
3728 #endif
3729 
3730 	err = ice_init_hw(hw);
3731 	if (err) {
3732 		dev_err(dev, "ice_init_hw failed: %d\n", err);
3733 		err = -EIO;
3734 		goto err_exit_unroll;
3735 	}
3736 
3737 	ice_request_fw(pf);
3738 
3739 	/* if ice_request_fw fails, ICE_FLAG_ADV_FEATURES bit won't be
3740 	 * set in pf->state, which will cause ice_is_safe_mode to return
3741 	 * true
3742 	 */
3743 	if (ice_is_safe_mode(pf)) {
3744 		dev_err(dev, "Package download failed. Advanced features disabled - Device now in Safe Mode\n");
3745 		/* we already got function/device capabilities but these don't
3746 		 * reflect what the driver needs to do in safe mode. Instead of
3747 		 * adding conditional logic everywhere to ignore these
3748 		 * device/function capabilities, override them.
3749 		 */
3750 		ice_set_safe_mode_caps(hw);
3751 	}
3752 
3753 	err = ice_init_pf(pf);
3754 	if (err) {
3755 		dev_err(dev, "ice_init_pf failed: %d\n", err);
3756 		goto err_init_pf_unroll;
3757 	}
3758 
3759 	ice_devlink_init_regions(pf);
3760 
3761 	pf->num_alloc_vsi = hw->func_caps.guar_num_vsi;
3762 	if (!pf->num_alloc_vsi) {
3763 		err = -EIO;
3764 		goto err_init_pf_unroll;
3765 	}
3766 
3767 	pf->vsi = devm_kcalloc(dev, pf->num_alloc_vsi, sizeof(*pf->vsi),
3768 			       GFP_KERNEL);
3769 	if (!pf->vsi) {
3770 		err = -ENOMEM;
3771 		goto err_init_pf_unroll;
3772 	}
3773 
3774 	err = ice_init_interrupt_scheme(pf);
3775 	if (err) {
3776 		dev_err(dev, "ice_init_interrupt_scheme failed: %d\n", err);
3777 		err = -EIO;
3778 		goto err_init_vsi_unroll;
3779 	}
3780 
3781 	/* In case of MSIX we are going to setup the misc vector right here
3782 	 * to handle admin queue events etc. In case of legacy and MSI
3783 	 * the misc functionality and queue processing is combined in
3784 	 * the same vector and that gets setup at open.
3785 	 */
3786 	err = ice_req_irq_msix_misc(pf);
3787 	if (err) {
3788 		dev_err(dev, "setup of misc vector failed: %d\n", err);
3789 		goto err_init_interrupt_unroll;
3790 	}
3791 
3792 	/* create switch struct for the switch element created by FW on boot */
3793 	pf->first_sw = devm_kzalloc(dev, sizeof(*pf->first_sw), GFP_KERNEL);
3794 	if (!pf->first_sw) {
3795 		err = -ENOMEM;
3796 		goto err_msix_misc_unroll;
3797 	}
3798 
3799 	if (hw->evb_veb)
3800 		pf->first_sw->bridge_mode = BRIDGE_MODE_VEB;
3801 	else
3802 		pf->first_sw->bridge_mode = BRIDGE_MODE_VEPA;
3803 
3804 	pf->first_sw->pf = pf;
3805 
3806 	/* record the sw_id available for later use */
3807 	pf->first_sw->sw_id = hw->port_info->sw_id;
3808 
3809 	err = ice_setup_pf_sw(pf);
3810 	if (err) {
3811 		dev_err(dev, "probe failed due to setup PF switch: %d\n", err);
3812 		goto err_alloc_sw_unroll;
3813 	}
3814 
3815 	clear_bit(__ICE_SERVICE_DIS, pf->state);
3816 
3817 	/* tell the firmware we are up */
3818 	err = ice_send_version(pf);
3819 	if (err) {
3820 		dev_err(dev, "probe failed sending driver version %s. error: %d\n",
3821 			UTS_RELEASE, err);
3822 		goto err_alloc_sw_unroll;
3823 	}
3824 
3825 	/* since everything is good, start the service timer */
3826 	mod_timer(&pf->serv_tmr, round_jiffies(jiffies + pf->serv_tmr_period));
3827 
3828 	err = ice_init_link_events(pf->hw.port_info);
3829 	if (err) {
3830 		dev_err(dev, "ice_init_link_events failed: %d\n", err);
3831 		goto err_alloc_sw_unroll;
3832 	}
3833 
3834 	err = ice_init_nvm_phy_type(pf->hw.port_info);
3835 	if (err) {
3836 		dev_err(dev, "ice_init_nvm_phy_type failed: %d\n", err);
3837 		goto err_alloc_sw_unroll;
3838 	}
3839 
3840 	err = ice_update_link_info(pf->hw.port_info);
3841 	if (err) {
3842 		dev_err(dev, "ice_update_link_info failed: %d\n", err);
3843 		goto err_alloc_sw_unroll;
3844 	}
3845 
3846 	ice_init_link_dflt_override(pf->hw.port_info);
3847 
3848 	/* if media available, initialize PHY settings */
3849 	if (pf->hw.port_info->phy.link_info.link_info &
3850 	    ICE_AQ_MEDIA_AVAILABLE) {
3851 		err = ice_init_phy_user_cfg(pf->hw.port_info);
3852 		if (err) {
3853 			dev_err(dev, "ice_init_phy_user_cfg failed: %d\n", err);
3854 			goto err_alloc_sw_unroll;
3855 		}
3856 
3857 		if (!test_bit(ICE_FLAG_LINK_DOWN_ON_CLOSE_ENA, pf->flags)) {
3858 			struct ice_vsi *vsi = ice_get_main_vsi(pf);
3859 
3860 			if (vsi)
3861 				ice_configure_phy(vsi);
3862 		}
3863 	} else {
3864 		set_bit(ICE_FLAG_NO_MEDIA, pf->flags);
3865 	}
3866 
3867 	ice_verify_cacheline_size(pf);
3868 
3869 	/* Save wakeup reason register for later use */
3870 	pf->wakeup_reason = rd32(hw, PFPM_WUS);
3871 
3872 	/* check for a power management event */
3873 	ice_print_wake_reason(pf);
3874 
3875 	/* clear wake status, all bits */
3876 	wr32(hw, PFPM_WUS, U32_MAX);
3877 
3878 	/* Disable WoL at init, wait for user to enable */
3879 	device_set_wakeup_enable(dev, false);
3880 
3881 	/* If no DDP driven features have to be setup, we are done with probe */
3882 	if (ice_is_safe_mode(pf))
3883 		goto probe_done;
3884 
3885 	/* initialize DDP driven features */
3886 
3887 	/* Note: Flow director init failure is non-fatal to load */
3888 	if (ice_init_fdir(pf))
3889 		dev_err(dev, "could not initialize flow director\n");
3890 
3891 	/* Note: DCB init failure is non-fatal to load */
3892 	if (ice_init_pf_dcb(pf, false)) {
3893 		clear_bit(ICE_FLAG_DCB_CAPABLE, pf->flags);
3894 		clear_bit(ICE_FLAG_DCB_ENA, pf->flags);
3895 	} else {
3896 		ice_cfg_lldp_mib_change(&pf->hw, true);
3897 	}
3898 
3899 	/* print PCI link speed and width */
3900 	pcie_print_link_status(pf->pdev);
3901 
3902 probe_done:
3903 	/* ready to go, so clear down state bit */
3904 	clear_bit(__ICE_DOWN, pf->state);
3905 	return 0;
3906 
3907 err_alloc_sw_unroll:
3908 	ice_devlink_destroy_port(pf);
3909 	set_bit(__ICE_SERVICE_DIS, pf->state);
3910 	set_bit(__ICE_DOWN, pf->state);
3911 	devm_kfree(dev, pf->first_sw);
3912 err_msix_misc_unroll:
3913 	ice_free_irq_msix_misc(pf);
3914 err_init_interrupt_unroll:
3915 	ice_clear_interrupt_scheme(pf);
3916 err_init_vsi_unroll:
3917 	devm_kfree(dev, pf->vsi);
3918 err_init_pf_unroll:
3919 	ice_deinit_pf(pf);
3920 	ice_devlink_destroy_regions(pf);
3921 	ice_deinit_hw(hw);
3922 err_exit_unroll:
3923 	ice_devlink_unregister(pf);
3924 	pci_disable_pcie_error_reporting(pdev);
3925 	pci_disable_device(pdev);
3926 	return err;
3927 }
3928 
3929 /**
3930  * ice_set_wake - enable or disable Wake on LAN
3931  * @pf: pointer to the PF struct
3932  *
3933  * Simple helper for WoL control
3934  */
3935 static void ice_set_wake(struct ice_pf *pf)
3936 {
3937 	struct ice_hw *hw = &pf->hw;
3938 	bool wol = pf->wol_ena;
3939 
3940 	/* clear wake state, otherwise new wake events won't fire */
3941 	wr32(hw, PFPM_WUS, U32_MAX);
3942 
3943 	/* enable / disable APM wake up, no RMW needed */
3944 	wr32(hw, PFPM_APM, wol ? PFPM_APM_APME_M : 0);
3945 
3946 	/* set magic packet filter enabled */
3947 	wr32(hw, PFPM_WUFC, wol ? PFPM_WUFC_MAG_M : 0);
3948 }
3949 
3950 /**
3951  * ice_setup_magic_mc_wake - setup device to wake on multicast magic packet
3952  * @pf: pointer to the PF struct
3953  *
3954  * Issue firmware command to enable multicast magic wake, making
3955  * sure that any locally administered address (LAA) is used for
3956  * wake, and that PF reset doesn't undo the LAA.
3957  */
3958 static void ice_setup_mc_magic_wake(struct ice_pf *pf)
3959 {
3960 	struct device *dev = ice_pf_to_dev(pf);
3961 	struct ice_hw *hw = &pf->hw;
3962 	enum ice_status status;
3963 	u8 mac_addr[ETH_ALEN];
3964 	struct ice_vsi *vsi;
3965 	u8 flags;
3966 
3967 	if (!pf->wol_ena)
3968 		return;
3969 
3970 	vsi = ice_get_main_vsi(pf);
3971 	if (!vsi)
3972 		return;
3973 
3974 	/* Get current MAC address in case it's an LAA */
3975 	if (vsi->netdev)
3976 		ether_addr_copy(mac_addr, vsi->netdev->dev_addr);
3977 	else
3978 		ether_addr_copy(mac_addr, vsi->port_info->mac.perm_addr);
3979 
3980 	flags = ICE_AQC_MAN_MAC_WR_MC_MAG_EN |
3981 		ICE_AQC_MAN_MAC_UPDATE_LAA_WOL |
3982 		ICE_AQC_MAN_MAC_WR_WOL_LAA_PFR_KEEP;
3983 
3984 	status = ice_aq_manage_mac_write(hw, mac_addr, flags, NULL);
3985 	if (status)
3986 		dev_err(dev, "Failed to enable Multicast Magic Packet wake, err %s aq_err %s\n",
3987 			ice_stat_str(status),
3988 			ice_aq_str(hw->adminq.sq_last_status));
3989 }
3990 
3991 /**
3992  * ice_remove - Device removal routine
3993  * @pdev: PCI device information struct
3994  */
3995 static void ice_remove(struct pci_dev *pdev)
3996 {
3997 	struct ice_pf *pf = pci_get_drvdata(pdev);
3998 	int i;
3999 
4000 	if (!pf)
4001 		return;
4002 
4003 	for (i = 0; i < ICE_MAX_RESET_WAIT; i++) {
4004 		if (!ice_is_reset_in_progress(pf->state))
4005 			break;
4006 		msleep(100);
4007 	}
4008 
4009 	if (test_bit(ICE_FLAG_SRIOV_ENA, pf->flags)) {
4010 		set_bit(__ICE_VF_RESETS_DISABLED, pf->state);
4011 		ice_free_vfs(pf);
4012 	}
4013 
4014 	set_bit(__ICE_DOWN, pf->state);
4015 	ice_service_task_stop(pf);
4016 
4017 	mutex_destroy(&(&pf->hw)->fdir_fltr_lock);
4018 	if (!ice_is_safe_mode(pf))
4019 		ice_remove_arfs(pf);
4020 	ice_setup_mc_magic_wake(pf);
4021 	ice_devlink_destroy_port(pf);
4022 	ice_vsi_release_all(pf);
4023 	ice_set_wake(pf);
4024 	ice_free_irq_msix_misc(pf);
4025 	ice_for_each_vsi(pf, i) {
4026 		if (!pf->vsi[i])
4027 			continue;
4028 		ice_vsi_free_q_vectors(pf->vsi[i]);
4029 	}
4030 	ice_deinit_pf(pf);
4031 	ice_devlink_destroy_regions(pf);
4032 	ice_deinit_hw(&pf->hw);
4033 	ice_devlink_unregister(pf);
4034 
4035 	/* Issue a PFR as part of the prescribed driver unload flow.  Do not
4036 	 * do it via ice_schedule_reset() since there is no need to rebuild
4037 	 * and the service task is already stopped.
4038 	 */
4039 	ice_reset(&pf->hw, ICE_RESET_PFR);
4040 	pci_wait_for_pending_transaction(pdev);
4041 	ice_clear_interrupt_scheme(pf);
4042 	pci_disable_pcie_error_reporting(pdev);
4043 	pci_disable_device(pdev);
4044 }
4045 
4046 /**
4047  * ice_shutdown - PCI callback for shutting down device
4048  * @pdev: PCI device information struct
4049  */
4050 static void ice_shutdown(struct pci_dev *pdev)
4051 {
4052 	struct ice_pf *pf = pci_get_drvdata(pdev);
4053 
4054 	ice_remove(pdev);
4055 
4056 	if (system_state == SYSTEM_POWER_OFF) {
4057 		pci_wake_from_d3(pdev, pf->wol_ena);
4058 		pci_set_power_state(pdev, PCI_D3hot);
4059 	}
4060 }
4061 
4062 #ifdef CONFIG_PM
4063 /**
4064  * ice_prepare_for_shutdown - prep for PCI shutdown
4065  * @pf: board private structure
4066  *
4067  * Inform or close all dependent features in prep for PCI device shutdown
4068  */
4069 static void ice_prepare_for_shutdown(struct ice_pf *pf)
4070 {
4071 	struct ice_hw *hw = &pf->hw;
4072 	u32 v;
4073 
4074 	/* Notify VFs of impending reset */
4075 	if (ice_check_sq_alive(hw, &hw->mailboxq))
4076 		ice_vc_notify_reset(pf);
4077 
4078 	dev_dbg(ice_pf_to_dev(pf), "Tearing down internal switch for shutdown\n");
4079 
4080 	/* disable the VSIs and their queues that are not already DOWN */
4081 	ice_pf_dis_all_vsi(pf, false);
4082 
4083 	ice_for_each_vsi(pf, v)
4084 		if (pf->vsi[v])
4085 			pf->vsi[v]->vsi_num = 0;
4086 
4087 	ice_shutdown_all_ctrlq(hw);
4088 }
4089 
4090 /**
4091  * ice_reinit_interrupt_scheme - Reinitialize interrupt scheme
4092  * @pf: board private structure to reinitialize
4093  *
4094  * This routine reinitialize interrupt scheme that was cleared during
4095  * power management suspend callback.
4096  *
4097  * This should be called during resume routine to re-allocate the q_vectors
4098  * and reacquire interrupts.
4099  */
4100 static int ice_reinit_interrupt_scheme(struct ice_pf *pf)
4101 {
4102 	struct device *dev = ice_pf_to_dev(pf);
4103 	int ret, v;
4104 
4105 	/* Since we clear MSIX flag during suspend, we need to
4106 	 * set it back during resume...
4107 	 */
4108 
4109 	ret = ice_init_interrupt_scheme(pf);
4110 	if (ret) {
4111 		dev_err(dev, "Failed to re-initialize interrupt %d\n", ret);
4112 		return ret;
4113 	}
4114 
4115 	/* Remap vectors and rings, after successful re-init interrupts */
4116 	ice_for_each_vsi(pf, v) {
4117 		if (!pf->vsi[v])
4118 			continue;
4119 
4120 		ret = ice_vsi_alloc_q_vectors(pf->vsi[v]);
4121 		if (ret)
4122 			goto err_reinit;
4123 		ice_vsi_map_rings_to_vectors(pf->vsi[v]);
4124 	}
4125 
4126 	ret = ice_req_irq_msix_misc(pf);
4127 	if (ret) {
4128 		dev_err(dev, "Setting up misc vector failed after device suspend %d\n",
4129 			ret);
4130 		goto err_reinit;
4131 	}
4132 
4133 	return 0;
4134 
4135 err_reinit:
4136 	while (v--)
4137 		if (pf->vsi[v])
4138 			ice_vsi_free_q_vectors(pf->vsi[v]);
4139 
4140 	return ret;
4141 }
4142 
4143 /**
4144  * ice_suspend
4145  * @dev: generic device information structure
4146  *
4147  * Power Management callback to quiesce the device and prepare
4148  * for D3 transition.
4149  */
4150 static int ice_suspend(struct device *dev)
4151 {
4152 	struct pci_dev *pdev = to_pci_dev(dev);
4153 	struct ice_pf *pf;
4154 	int disabled, v;
4155 
4156 	pf = pci_get_drvdata(pdev);
4157 
4158 	if (!ice_pf_state_is_nominal(pf)) {
4159 		dev_err(dev, "Device is not ready, no need to suspend it\n");
4160 		return -EBUSY;
4161 	}
4162 
4163 	/* Stop watchdog tasks until resume completion.
4164 	 * Even though it is most likely that the service task is
4165 	 * disabled if the device is suspended or down, the service task's
4166 	 * state is controlled by a different state bit, and we should
4167 	 * store and honor whatever state that bit is in at this point.
4168 	 */
4169 	disabled = ice_service_task_stop(pf);
4170 
4171 	/* Already suspended?, then there is nothing to do */
4172 	if (test_and_set_bit(__ICE_SUSPENDED, pf->state)) {
4173 		if (!disabled)
4174 			ice_service_task_restart(pf);
4175 		return 0;
4176 	}
4177 
4178 	if (test_bit(__ICE_DOWN, pf->state) ||
4179 	    ice_is_reset_in_progress(pf->state)) {
4180 		dev_err(dev, "can't suspend device in reset or already down\n");
4181 		if (!disabled)
4182 			ice_service_task_restart(pf);
4183 		return 0;
4184 	}
4185 
4186 	ice_setup_mc_magic_wake(pf);
4187 
4188 	ice_prepare_for_shutdown(pf);
4189 
4190 	ice_set_wake(pf);
4191 
4192 	/* Free vectors, clear the interrupt scheme and release IRQs
4193 	 * for proper hibernation, especially with large number of CPUs.
4194 	 * Otherwise hibernation might fail when mapping all the vectors back
4195 	 * to CPU0.
4196 	 */
4197 	ice_free_irq_msix_misc(pf);
4198 	ice_for_each_vsi(pf, v) {
4199 		if (!pf->vsi[v])
4200 			continue;
4201 		ice_vsi_free_q_vectors(pf->vsi[v]);
4202 	}
4203 	ice_clear_interrupt_scheme(pf);
4204 
4205 	pci_wake_from_d3(pdev, pf->wol_ena);
4206 	pci_set_power_state(pdev, PCI_D3hot);
4207 	return 0;
4208 }
4209 
4210 /**
4211  * ice_resume - PM callback for waking up from D3
4212  * @dev: generic device information structure
4213  */
4214 static int ice_resume(struct device *dev)
4215 {
4216 	struct pci_dev *pdev = to_pci_dev(dev);
4217 	enum ice_reset_req reset_type;
4218 	struct ice_pf *pf;
4219 	struct ice_hw *hw;
4220 	int ret;
4221 
4222 	pci_set_power_state(pdev, PCI_D0);
4223 	pci_restore_state(pdev);
4224 	pci_save_state(pdev);
4225 
4226 	if (!pci_device_is_present(pdev))
4227 		return -ENODEV;
4228 
4229 	ret = pci_enable_device_mem(pdev);
4230 	if (ret) {
4231 		dev_err(dev, "Cannot enable device after suspend\n");
4232 		return ret;
4233 	}
4234 
4235 	pf = pci_get_drvdata(pdev);
4236 	hw = &pf->hw;
4237 
4238 	pf->wakeup_reason = rd32(hw, PFPM_WUS);
4239 	ice_print_wake_reason(pf);
4240 
4241 	/* We cleared the interrupt scheme when we suspended, so we need to
4242 	 * restore it now to resume device functionality.
4243 	 */
4244 	ret = ice_reinit_interrupt_scheme(pf);
4245 	if (ret)
4246 		dev_err(dev, "Cannot restore interrupt scheme: %d\n", ret);
4247 
4248 	clear_bit(__ICE_DOWN, pf->state);
4249 	/* Now perform PF reset and rebuild */
4250 	reset_type = ICE_RESET_PFR;
4251 	/* re-enable service task for reset, but allow reset to schedule it */
4252 	clear_bit(__ICE_SERVICE_DIS, pf->state);
4253 
4254 	if (ice_schedule_reset(pf, reset_type))
4255 		dev_err(dev, "Reset during resume failed.\n");
4256 
4257 	clear_bit(__ICE_SUSPENDED, pf->state);
4258 	ice_service_task_restart(pf);
4259 
4260 	/* Restart the service task */
4261 	mod_timer(&pf->serv_tmr, round_jiffies(jiffies + pf->serv_tmr_period));
4262 
4263 	return 0;
4264 }
4265 #endif /* CONFIG_PM */
4266 
4267 /**
4268  * ice_pci_err_detected - warning that PCI error has been detected
4269  * @pdev: PCI device information struct
4270  * @err: the type of PCI error
4271  *
4272  * Called to warn that something happened on the PCI bus and the error handling
4273  * is in progress.  Allows the driver to gracefully prepare/handle PCI errors.
4274  */
4275 static pci_ers_result_t
4276 ice_pci_err_detected(struct pci_dev *pdev, enum pci_channel_state err)
4277 {
4278 	struct ice_pf *pf = pci_get_drvdata(pdev);
4279 
4280 	if (!pf) {
4281 		dev_err(&pdev->dev, "%s: unrecoverable device error %d\n",
4282 			__func__, err);
4283 		return PCI_ERS_RESULT_DISCONNECT;
4284 	}
4285 
4286 	if (!test_bit(__ICE_SUSPENDED, pf->state)) {
4287 		ice_service_task_stop(pf);
4288 
4289 		if (!test_bit(__ICE_PREPARED_FOR_RESET, pf->state)) {
4290 			set_bit(__ICE_PFR_REQ, pf->state);
4291 			ice_prepare_for_reset(pf);
4292 		}
4293 	}
4294 
4295 	return PCI_ERS_RESULT_NEED_RESET;
4296 }
4297 
4298 /**
4299  * ice_pci_err_slot_reset - a PCI slot reset has just happened
4300  * @pdev: PCI device information struct
4301  *
4302  * Called to determine if the driver can recover from the PCI slot reset by
4303  * using a register read to determine if the device is recoverable.
4304  */
4305 static pci_ers_result_t ice_pci_err_slot_reset(struct pci_dev *pdev)
4306 {
4307 	struct ice_pf *pf = pci_get_drvdata(pdev);
4308 	pci_ers_result_t result;
4309 	int err;
4310 	u32 reg;
4311 
4312 	err = pci_enable_device_mem(pdev);
4313 	if (err) {
4314 		dev_err(&pdev->dev, "Cannot re-enable PCI device after reset, error %d\n",
4315 			err);
4316 		result = PCI_ERS_RESULT_DISCONNECT;
4317 	} else {
4318 		pci_set_master(pdev);
4319 		pci_restore_state(pdev);
4320 		pci_save_state(pdev);
4321 		pci_wake_from_d3(pdev, false);
4322 
4323 		/* Check for life */
4324 		reg = rd32(&pf->hw, GLGEN_RTRIG);
4325 		if (!reg)
4326 			result = PCI_ERS_RESULT_RECOVERED;
4327 		else
4328 			result = PCI_ERS_RESULT_DISCONNECT;
4329 	}
4330 
4331 	err = pci_aer_clear_nonfatal_status(pdev);
4332 	if (err)
4333 		dev_dbg(&pdev->dev, "pci_aer_clear_nonfatal_status() failed, error %d\n",
4334 			err);
4335 		/* non-fatal, continue */
4336 
4337 	return result;
4338 }
4339 
4340 /**
4341  * ice_pci_err_resume - restart operations after PCI error recovery
4342  * @pdev: PCI device information struct
4343  *
4344  * Called to allow the driver to bring things back up after PCI error and/or
4345  * reset recovery have finished
4346  */
4347 static void ice_pci_err_resume(struct pci_dev *pdev)
4348 {
4349 	struct ice_pf *pf = pci_get_drvdata(pdev);
4350 
4351 	if (!pf) {
4352 		dev_err(&pdev->dev, "%s failed, device is unrecoverable\n",
4353 			__func__);
4354 		return;
4355 	}
4356 
4357 	if (test_bit(__ICE_SUSPENDED, pf->state)) {
4358 		dev_dbg(&pdev->dev, "%s failed to resume normal operations!\n",
4359 			__func__);
4360 		return;
4361 	}
4362 
4363 	ice_do_reset(pf, ICE_RESET_PFR);
4364 	ice_service_task_restart(pf);
4365 	mod_timer(&pf->serv_tmr, round_jiffies(jiffies + pf->serv_tmr_period));
4366 }
4367 
4368 /**
4369  * ice_pci_err_reset_prepare - prepare device driver for PCI reset
4370  * @pdev: PCI device information struct
4371  */
4372 static void ice_pci_err_reset_prepare(struct pci_dev *pdev)
4373 {
4374 	struct ice_pf *pf = pci_get_drvdata(pdev);
4375 
4376 	if (!test_bit(__ICE_SUSPENDED, pf->state)) {
4377 		ice_service_task_stop(pf);
4378 
4379 		if (!test_bit(__ICE_PREPARED_FOR_RESET, pf->state)) {
4380 			set_bit(__ICE_PFR_REQ, pf->state);
4381 			ice_prepare_for_reset(pf);
4382 		}
4383 	}
4384 }
4385 
4386 /**
4387  * ice_pci_err_reset_done - PCI reset done, device driver reset can begin
4388  * @pdev: PCI device information struct
4389  */
4390 static void ice_pci_err_reset_done(struct pci_dev *pdev)
4391 {
4392 	ice_pci_err_resume(pdev);
4393 }
4394 
4395 /* ice_pci_tbl - PCI Device ID Table
4396  *
4397  * Wildcard entries (PCI_ANY_ID) should come last
4398  * Last entry must be all 0s
4399  *
4400  * { Vendor ID, Device ID, SubVendor ID, SubDevice ID,
4401  *   Class, Class Mask, private data (not used) }
4402  */
4403 static const struct pci_device_id ice_pci_tbl[] = {
4404 	{ PCI_VDEVICE(INTEL, ICE_DEV_ID_E810C_BACKPLANE), 0 },
4405 	{ PCI_VDEVICE(INTEL, ICE_DEV_ID_E810C_QSFP), 0 },
4406 	{ PCI_VDEVICE(INTEL, ICE_DEV_ID_E810C_SFP), 0 },
4407 	{ PCI_VDEVICE(INTEL, ICE_DEV_ID_E810_XXV_SFP), 0 },
4408 	{ PCI_VDEVICE(INTEL, ICE_DEV_ID_E823C_BACKPLANE), 0 },
4409 	{ PCI_VDEVICE(INTEL, ICE_DEV_ID_E823C_QSFP), 0 },
4410 	{ PCI_VDEVICE(INTEL, ICE_DEV_ID_E823C_SFP), 0 },
4411 	{ PCI_VDEVICE(INTEL, ICE_DEV_ID_E823C_10G_BASE_T), 0 },
4412 	{ PCI_VDEVICE(INTEL, ICE_DEV_ID_E823C_SGMII), 0 },
4413 	{ PCI_VDEVICE(INTEL, ICE_DEV_ID_E822C_BACKPLANE), 0 },
4414 	{ PCI_VDEVICE(INTEL, ICE_DEV_ID_E822C_QSFP), 0 },
4415 	{ PCI_VDEVICE(INTEL, ICE_DEV_ID_E822C_SFP), 0 },
4416 	{ PCI_VDEVICE(INTEL, ICE_DEV_ID_E822C_10G_BASE_T), 0 },
4417 	{ PCI_VDEVICE(INTEL, ICE_DEV_ID_E822C_SGMII), 0 },
4418 	{ PCI_VDEVICE(INTEL, ICE_DEV_ID_E822L_BACKPLANE), 0 },
4419 	{ PCI_VDEVICE(INTEL, ICE_DEV_ID_E822L_SFP), 0 },
4420 	{ PCI_VDEVICE(INTEL, ICE_DEV_ID_E822L_10G_BASE_T), 0 },
4421 	{ PCI_VDEVICE(INTEL, ICE_DEV_ID_E822L_SGMII), 0 },
4422 	{ PCI_VDEVICE(INTEL, ICE_DEV_ID_E823L_BACKPLANE), 0 },
4423 	{ PCI_VDEVICE(INTEL, ICE_DEV_ID_E823L_SFP), 0 },
4424 	{ PCI_VDEVICE(INTEL, ICE_DEV_ID_E823L_10G_BASE_T), 0 },
4425 	{ PCI_VDEVICE(INTEL, ICE_DEV_ID_E823L_1GBE), 0 },
4426 	{ PCI_VDEVICE(INTEL, ICE_DEV_ID_E823L_QSFP), 0 },
4427 	/* required last entry */
4428 	{ 0, }
4429 };
4430 MODULE_DEVICE_TABLE(pci, ice_pci_tbl);
4431 
4432 static __maybe_unused SIMPLE_DEV_PM_OPS(ice_pm_ops, ice_suspend, ice_resume);
4433 
4434 static const struct pci_error_handlers ice_pci_err_handler = {
4435 	.error_detected = ice_pci_err_detected,
4436 	.slot_reset = ice_pci_err_slot_reset,
4437 	.reset_prepare = ice_pci_err_reset_prepare,
4438 	.reset_done = ice_pci_err_reset_done,
4439 	.resume = ice_pci_err_resume
4440 };
4441 
4442 static struct pci_driver ice_driver = {
4443 	.name = KBUILD_MODNAME,
4444 	.id_table = ice_pci_tbl,
4445 	.probe = ice_probe,
4446 	.remove = ice_remove,
4447 #ifdef CONFIG_PM
4448 	.driver.pm = &ice_pm_ops,
4449 #endif /* CONFIG_PM */
4450 	.shutdown = ice_shutdown,
4451 	.sriov_configure = ice_sriov_configure,
4452 	.err_handler = &ice_pci_err_handler
4453 };
4454 
4455 /**
4456  * ice_module_init - Driver registration routine
4457  *
4458  * ice_module_init is the first routine called when the driver is
4459  * loaded. All it does is register with the PCI subsystem.
4460  */
4461 static int __init ice_module_init(void)
4462 {
4463 	int status;
4464 
4465 	pr_info("%s\n", ice_driver_string);
4466 	pr_info("%s\n", ice_copyright);
4467 
4468 	ice_wq = alloc_workqueue("%s", WQ_MEM_RECLAIM, 0, KBUILD_MODNAME);
4469 	if (!ice_wq) {
4470 		pr_err("Failed to create workqueue\n");
4471 		return -ENOMEM;
4472 	}
4473 
4474 	status = pci_register_driver(&ice_driver);
4475 	if (status) {
4476 		pr_err("failed to register PCI driver, err %d\n", status);
4477 		destroy_workqueue(ice_wq);
4478 	}
4479 
4480 	return status;
4481 }
4482 module_init(ice_module_init);
4483 
4484 /**
4485  * ice_module_exit - Driver exit cleanup routine
4486  *
4487  * ice_module_exit is called just before the driver is removed
4488  * from memory.
4489  */
4490 static void __exit ice_module_exit(void)
4491 {
4492 	pci_unregister_driver(&ice_driver);
4493 	destroy_workqueue(ice_wq);
4494 	pr_info("module unloaded\n");
4495 }
4496 module_exit(ice_module_exit);
4497 
4498 /**
4499  * ice_set_mac_address - NDO callback to set MAC address
4500  * @netdev: network interface device structure
4501  * @pi: pointer to an address structure
4502  *
4503  * Returns 0 on success, negative on failure
4504  */
4505 static int ice_set_mac_address(struct net_device *netdev, void *pi)
4506 {
4507 	struct ice_netdev_priv *np = netdev_priv(netdev);
4508 	struct ice_vsi *vsi = np->vsi;
4509 	struct ice_pf *pf = vsi->back;
4510 	struct ice_hw *hw = &pf->hw;
4511 	struct sockaddr *addr = pi;
4512 	enum ice_status status;
4513 	u8 flags = 0;
4514 	int err = 0;
4515 	u8 *mac;
4516 
4517 	mac = (u8 *)addr->sa_data;
4518 
4519 	if (!is_valid_ether_addr(mac))
4520 		return -EADDRNOTAVAIL;
4521 
4522 	if (ether_addr_equal(netdev->dev_addr, mac)) {
4523 		netdev_warn(netdev, "already using mac %pM\n", mac);
4524 		return 0;
4525 	}
4526 
4527 	if (test_bit(__ICE_DOWN, pf->state) ||
4528 	    ice_is_reset_in_progress(pf->state)) {
4529 		netdev_err(netdev, "can't set mac %pM. device not ready\n",
4530 			   mac);
4531 		return -EBUSY;
4532 	}
4533 
4534 	/* Clean up old MAC filter. Not an error if old filter doesn't exist */
4535 	status = ice_fltr_remove_mac(vsi, netdev->dev_addr, ICE_FWD_TO_VSI);
4536 	if (status && status != ICE_ERR_DOES_NOT_EXIST) {
4537 		err = -EADDRNOTAVAIL;
4538 		goto err_update_filters;
4539 	}
4540 
4541 	/* Add filter for new MAC. If filter exists, just return success */
4542 	status = ice_fltr_add_mac(vsi, mac, ICE_FWD_TO_VSI);
4543 	if (status == ICE_ERR_ALREADY_EXISTS) {
4544 		netdev_dbg(netdev, "filter for MAC %pM already exists\n", mac);
4545 		return 0;
4546 	}
4547 
4548 	/* error if the new filter addition failed */
4549 	if (status)
4550 		err = -EADDRNOTAVAIL;
4551 
4552 err_update_filters:
4553 	if (err) {
4554 		netdev_err(netdev, "can't set MAC %pM. filter update failed\n",
4555 			   mac);
4556 		return err;
4557 	}
4558 
4559 	/* change the netdev's MAC address */
4560 	memcpy(netdev->dev_addr, mac, netdev->addr_len);
4561 	netdev_dbg(vsi->netdev, "updated MAC address to %pM\n",
4562 		   netdev->dev_addr);
4563 
4564 	/* write new MAC address to the firmware */
4565 	flags = ICE_AQC_MAN_MAC_UPDATE_LAA_WOL;
4566 	status = ice_aq_manage_mac_write(hw, mac, flags, NULL);
4567 	if (status) {
4568 		netdev_err(netdev, "can't set MAC %pM. write to firmware failed error %s\n",
4569 			   mac, ice_stat_str(status));
4570 	}
4571 	return 0;
4572 }
4573 
4574 /**
4575  * ice_set_rx_mode - NDO callback to set the netdev filters
4576  * @netdev: network interface device structure
4577  */
4578 static void ice_set_rx_mode(struct net_device *netdev)
4579 {
4580 	struct ice_netdev_priv *np = netdev_priv(netdev);
4581 	struct ice_vsi *vsi = np->vsi;
4582 
4583 	if (!vsi)
4584 		return;
4585 
4586 	/* Set the flags to synchronize filters
4587 	 * ndo_set_rx_mode may be triggered even without a change in netdev
4588 	 * flags
4589 	 */
4590 	set_bit(ICE_VSI_FLAG_UMAC_FLTR_CHANGED, vsi->flags);
4591 	set_bit(ICE_VSI_FLAG_MMAC_FLTR_CHANGED, vsi->flags);
4592 	set_bit(ICE_FLAG_FLTR_SYNC, vsi->back->flags);
4593 
4594 	/* schedule our worker thread which will take care of
4595 	 * applying the new filter changes
4596 	 */
4597 	ice_service_task_schedule(vsi->back);
4598 }
4599 
4600 /**
4601  * ice_set_tx_maxrate - NDO callback to set the maximum per-queue bitrate
4602  * @netdev: network interface device structure
4603  * @queue_index: Queue ID
4604  * @maxrate: maximum bandwidth in Mbps
4605  */
4606 static int
4607 ice_set_tx_maxrate(struct net_device *netdev, int queue_index, u32 maxrate)
4608 {
4609 	struct ice_netdev_priv *np = netdev_priv(netdev);
4610 	struct ice_vsi *vsi = np->vsi;
4611 	enum ice_status status;
4612 	u16 q_handle;
4613 	u8 tc;
4614 
4615 	/* Validate maxrate requested is within permitted range */
4616 	if (maxrate && (maxrate > (ICE_SCHED_MAX_BW / 1000))) {
4617 		netdev_err(netdev, "Invalid max rate %d specified for the queue %d\n",
4618 			   maxrate, queue_index);
4619 		return -EINVAL;
4620 	}
4621 
4622 	q_handle = vsi->tx_rings[queue_index]->q_handle;
4623 	tc = ice_dcb_get_tc(vsi, queue_index);
4624 
4625 	/* Set BW back to default, when user set maxrate to 0 */
4626 	if (!maxrate)
4627 		status = ice_cfg_q_bw_dflt_lmt(vsi->port_info, vsi->idx, tc,
4628 					       q_handle, ICE_MAX_BW);
4629 	else
4630 		status = ice_cfg_q_bw_lmt(vsi->port_info, vsi->idx, tc,
4631 					  q_handle, ICE_MAX_BW, maxrate * 1000);
4632 	if (status) {
4633 		netdev_err(netdev, "Unable to set Tx max rate, error %s\n",
4634 			   ice_stat_str(status));
4635 		return -EIO;
4636 	}
4637 
4638 	return 0;
4639 }
4640 
4641 /**
4642  * ice_fdb_add - add an entry to the hardware database
4643  * @ndm: the input from the stack
4644  * @tb: pointer to array of nladdr (unused)
4645  * @dev: the net device pointer
4646  * @addr: the MAC address entry being added
4647  * @vid: VLAN ID
4648  * @flags: instructions from stack about fdb operation
4649  * @extack: netlink extended ack
4650  */
4651 static int
4652 ice_fdb_add(struct ndmsg *ndm, struct nlattr __always_unused *tb[],
4653 	    struct net_device *dev, const unsigned char *addr, u16 vid,
4654 	    u16 flags, struct netlink_ext_ack __always_unused *extack)
4655 {
4656 	int err;
4657 
4658 	if (vid) {
4659 		netdev_err(dev, "VLANs aren't supported yet for dev_uc|mc_add()\n");
4660 		return -EINVAL;
4661 	}
4662 	if (ndm->ndm_state && !(ndm->ndm_state & NUD_PERMANENT)) {
4663 		netdev_err(dev, "FDB only supports static addresses\n");
4664 		return -EINVAL;
4665 	}
4666 
4667 	if (is_unicast_ether_addr(addr) || is_link_local_ether_addr(addr))
4668 		err = dev_uc_add_excl(dev, addr);
4669 	else if (is_multicast_ether_addr(addr))
4670 		err = dev_mc_add_excl(dev, addr);
4671 	else
4672 		err = -EINVAL;
4673 
4674 	/* Only return duplicate errors if NLM_F_EXCL is set */
4675 	if (err == -EEXIST && !(flags & NLM_F_EXCL))
4676 		err = 0;
4677 
4678 	return err;
4679 }
4680 
4681 /**
4682  * ice_fdb_del - delete an entry from the hardware database
4683  * @ndm: the input from the stack
4684  * @tb: pointer to array of nladdr (unused)
4685  * @dev: the net device pointer
4686  * @addr: the MAC address entry being added
4687  * @vid: VLAN ID
4688  */
4689 static int
4690 ice_fdb_del(struct ndmsg *ndm, __always_unused struct nlattr *tb[],
4691 	    struct net_device *dev, const unsigned char *addr,
4692 	    __always_unused u16 vid)
4693 {
4694 	int err;
4695 
4696 	if (ndm->ndm_state & NUD_PERMANENT) {
4697 		netdev_err(dev, "FDB only supports static addresses\n");
4698 		return -EINVAL;
4699 	}
4700 
4701 	if (is_unicast_ether_addr(addr))
4702 		err = dev_uc_del(dev, addr);
4703 	else if (is_multicast_ether_addr(addr))
4704 		err = dev_mc_del(dev, addr);
4705 	else
4706 		err = -EINVAL;
4707 
4708 	return err;
4709 }
4710 
4711 /**
4712  * ice_set_features - set the netdev feature flags
4713  * @netdev: ptr to the netdev being adjusted
4714  * @features: the feature set that the stack is suggesting
4715  */
4716 static int
4717 ice_set_features(struct net_device *netdev, netdev_features_t features)
4718 {
4719 	struct ice_netdev_priv *np = netdev_priv(netdev);
4720 	struct ice_vsi *vsi = np->vsi;
4721 	struct ice_pf *pf = vsi->back;
4722 	int ret = 0;
4723 
4724 	/* Don't set any netdev advanced features with device in Safe Mode */
4725 	if (ice_is_safe_mode(vsi->back)) {
4726 		dev_err(ice_pf_to_dev(vsi->back), "Device is in Safe Mode - not enabling advanced netdev features\n");
4727 		return ret;
4728 	}
4729 
4730 	/* Do not change setting during reset */
4731 	if (ice_is_reset_in_progress(pf->state)) {
4732 		dev_err(ice_pf_to_dev(vsi->back), "Device is resetting, changing advanced netdev features temporarily unavailable.\n");
4733 		return -EBUSY;
4734 	}
4735 
4736 	/* Multiple features can be changed in one call so keep features in
4737 	 * separate if/else statements to guarantee each feature is checked
4738 	 */
4739 	if (features & NETIF_F_RXHASH && !(netdev->features & NETIF_F_RXHASH))
4740 		ret = ice_vsi_manage_rss_lut(vsi, true);
4741 	else if (!(features & NETIF_F_RXHASH) &&
4742 		 netdev->features & NETIF_F_RXHASH)
4743 		ret = ice_vsi_manage_rss_lut(vsi, false);
4744 
4745 	if ((features & NETIF_F_HW_VLAN_CTAG_RX) &&
4746 	    !(netdev->features & NETIF_F_HW_VLAN_CTAG_RX))
4747 		ret = ice_vsi_manage_vlan_stripping(vsi, true);
4748 	else if (!(features & NETIF_F_HW_VLAN_CTAG_RX) &&
4749 		 (netdev->features & NETIF_F_HW_VLAN_CTAG_RX))
4750 		ret = ice_vsi_manage_vlan_stripping(vsi, false);
4751 
4752 	if ((features & NETIF_F_HW_VLAN_CTAG_TX) &&
4753 	    !(netdev->features & NETIF_F_HW_VLAN_CTAG_TX))
4754 		ret = ice_vsi_manage_vlan_insertion(vsi);
4755 	else if (!(features & NETIF_F_HW_VLAN_CTAG_TX) &&
4756 		 (netdev->features & NETIF_F_HW_VLAN_CTAG_TX))
4757 		ret = ice_vsi_manage_vlan_insertion(vsi);
4758 
4759 	if ((features & NETIF_F_HW_VLAN_CTAG_FILTER) &&
4760 	    !(netdev->features & NETIF_F_HW_VLAN_CTAG_FILTER))
4761 		ret = ice_cfg_vlan_pruning(vsi, true, false);
4762 	else if (!(features & NETIF_F_HW_VLAN_CTAG_FILTER) &&
4763 		 (netdev->features & NETIF_F_HW_VLAN_CTAG_FILTER))
4764 		ret = ice_cfg_vlan_pruning(vsi, false, false);
4765 
4766 	if ((features & NETIF_F_NTUPLE) &&
4767 	    !(netdev->features & NETIF_F_NTUPLE)) {
4768 		ice_vsi_manage_fdir(vsi, true);
4769 		ice_init_arfs(vsi);
4770 	} else if (!(features & NETIF_F_NTUPLE) &&
4771 		 (netdev->features & NETIF_F_NTUPLE)) {
4772 		ice_vsi_manage_fdir(vsi, false);
4773 		ice_clear_arfs(vsi);
4774 	}
4775 
4776 	return ret;
4777 }
4778 
4779 /**
4780  * ice_vsi_vlan_setup - Setup VLAN offload properties on a VSI
4781  * @vsi: VSI to setup VLAN properties for
4782  */
4783 static int ice_vsi_vlan_setup(struct ice_vsi *vsi)
4784 {
4785 	int ret = 0;
4786 
4787 	if (vsi->netdev->features & NETIF_F_HW_VLAN_CTAG_RX)
4788 		ret = ice_vsi_manage_vlan_stripping(vsi, true);
4789 	if (vsi->netdev->features & NETIF_F_HW_VLAN_CTAG_TX)
4790 		ret = ice_vsi_manage_vlan_insertion(vsi);
4791 
4792 	return ret;
4793 }
4794 
4795 /**
4796  * ice_vsi_cfg - Setup the VSI
4797  * @vsi: the VSI being configured
4798  *
4799  * Return 0 on success and negative value on error
4800  */
4801 int ice_vsi_cfg(struct ice_vsi *vsi)
4802 {
4803 	int err;
4804 
4805 	if (vsi->netdev) {
4806 		ice_set_rx_mode(vsi->netdev);
4807 
4808 		err = ice_vsi_vlan_setup(vsi);
4809 
4810 		if (err)
4811 			return err;
4812 	}
4813 	ice_vsi_cfg_dcb_rings(vsi);
4814 
4815 	err = ice_vsi_cfg_lan_txqs(vsi);
4816 	if (!err && ice_is_xdp_ena_vsi(vsi))
4817 		err = ice_vsi_cfg_xdp_txqs(vsi);
4818 	if (!err)
4819 		err = ice_vsi_cfg_rxqs(vsi);
4820 
4821 	return err;
4822 }
4823 
4824 /**
4825  * ice_napi_enable_all - Enable NAPI for all q_vectors in the VSI
4826  * @vsi: the VSI being configured
4827  */
4828 static void ice_napi_enable_all(struct ice_vsi *vsi)
4829 {
4830 	int q_idx;
4831 
4832 	if (!vsi->netdev)
4833 		return;
4834 
4835 	ice_for_each_q_vector(vsi, q_idx) {
4836 		struct ice_q_vector *q_vector = vsi->q_vectors[q_idx];
4837 
4838 		if (q_vector->rx.ring || q_vector->tx.ring)
4839 			napi_enable(&q_vector->napi);
4840 	}
4841 }
4842 
4843 /**
4844  * ice_up_complete - Finish the last steps of bringing up a connection
4845  * @vsi: The VSI being configured
4846  *
4847  * Return 0 on success and negative value on error
4848  */
4849 static int ice_up_complete(struct ice_vsi *vsi)
4850 {
4851 	struct ice_pf *pf = vsi->back;
4852 	int err;
4853 
4854 	ice_vsi_cfg_msix(vsi);
4855 
4856 	/* Enable only Rx rings, Tx rings were enabled by the FW when the
4857 	 * Tx queue group list was configured and the context bits were
4858 	 * programmed using ice_vsi_cfg_txqs
4859 	 */
4860 	err = ice_vsi_start_all_rx_rings(vsi);
4861 	if (err)
4862 		return err;
4863 
4864 	clear_bit(__ICE_DOWN, vsi->state);
4865 	ice_napi_enable_all(vsi);
4866 	ice_vsi_ena_irq(vsi);
4867 
4868 	if (vsi->port_info &&
4869 	    (vsi->port_info->phy.link_info.link_info & ICE_AQ_LINK_UP) &&
4870 	    vsi->netdev) {
4871 		ice_print_link_msg(vsi, true);
4872 		netif_tx_start_all_queues(vsi->netdev);
4873 		netif_carrier_on(vsi->netdev);
4874 	}
4875 
4876 	ice_service_task_schedule(pf);
4877 
4878 	return 0;
4879 }
4880 
4881 /**
4882  * ice_up - Bring the connection back up after being down
4883  * @vsi: VSI being configured
4884  */
4885 int ice_up(struct ice_vsi *vsi)
4886 {
4887 	int err;
4888 
4889 	err = ice_vsi_cfg(vsi);
4890 	if (!err)
4891 		err = ice_up_complete(vsi);
4892 
4893 	return err;
4894 }
4895 
4896 /**
4897  * ice_fetch_u64_stats_per_ring - get packets and bytes stats per ring
4898  * @ring: Tx or Rx ring to read stats from
4899  * @pkts: packets stats counter
4900  * @bytes: bytes stats counter
4901  *
4902  * This function fetches stats from the ring considering the atomic operations
4903  * that needs to be performed to read u64 values in 32 bit machine.
4904  */
4905 static void
4906 ice_fetch_u64_stats_per_ring(struct ice_ring *ring, u64 *pkts, u64 *bytes)
4907 {
4908 	unsigned int start;
4909 	*pkts = 0;
4910 	*bytes = 0;
4911 
4912 	if (!ring)
4913 		return;
4914 	do {
4915 		start = u64_stats_fetch_begin_irq(&ring->syncp);
4916 		*pkts = ring->stats.pkts;
4917 		*bytes = ring->stats.bytes;
4918 	} while (u64_stats_fetch_retry_irq(&ring->syncp, start));
4919 }
4920 
4921 /**
4922  * ice_update_vsi_tx_ring_stats - Update VSI Tx ring stats counters
4923  * @vsi: the VSI to be updated
4924  * @rings: rings to work on
4925  * @count: number of rings
4926  */
4927 static void
4928 ice_update_vsi_tx_ring_stats(struct ice_vsi *vsi, struct ice_ring **rings,
4929 			     u16 count)
4930 {
4931 	struct rtnl_link_stats64 *vsi_stats = &vsi->net_stats;
4932 	u16 i;
4933 
4934 	for (i = 0; i < count; i++) {
4935 		struct ice_ring *ring;
4936 		u64 pkts, bytes;
4937 
4938 		ring = READ_ONCE(rings[i]);
4939 		ice_fetch_u64_stats_per_ring(ring, &pkts, &bytes);
4940 		vsi_stats->tx_packets += pkts;
4941 		vsi_stats->tx_bytes += bytes;
4942 		vsi->tx_restart += ring->tx_stats.restart_q;
4943 		vsi->tx_busy += ring->tx_stats.tx_busy;
4944 		vsi->tx_linearize += ring->tx_stats.tx_linearize;
4945 	}
4946 }
4947 
4948 /**
4949  * ice_update_vsi_ring_stats - Update VSI stats counters
4950  * @vsi: the VSI to be updated
4951  */
4952 static void ice_update_vsi_ring_stats(struct ice_vsi *vsi)
4953 {
4954 	struct rtnl_link_stats64 *vsi_stats = &vsi->net_stats;
4955 	struct ice_ring *ring;
4956 	u64 pkts, bytes;
4957 	int i;
4958 
4959 	/* reset netdev stats */
4960 	vsi_stats->tx_packets = 0;
4961 	vsi_stats->tx_bytes = 0;
4962 	vsi_stats->rx_packets = 0;
4963 	vsi_stats->rx_bytes = 0;
4964 
4965 	/* reset non-netdev (extended) stats */
4966 	vsi->tx_restart = 0;
4967 	vsi->tx_busy = 0;
4968 	vsi->tx_linearize = 0;
4969 	vsi->rx_buf_failed = 0;
4970 	vsi->rx_page_failed = 0;
4971 
4972 	rcu_read_lock();
4973 
4974 	/* update Tx rings counters */
4975 	ice_update_vsi_tx_ring_stats(vsi, vsi->tx_rings, vsi->num_txq);
4976 
4977 	/* update Rx rings counters */
4978 	ice_for_each_rxq(vsi, i) {
4979 		ring = READ_ONCE(vsi->rx_rings[i]);
4980 		ice_fetch_u64_stats_per_ring(ring, &pkts, &bytes);
4981 		vsi_stats->rx_packets += pkts;
4982 		vsi_stats->rx_bytes += bytes;
4983 		vsi->rx_buf_failed += ring->rx_stats.alloc_buf_failed;
4984 		vsi->rx_page_failed += ring->rx_stats.alloc_page_failed;
4985 	}
4986 
4987 	/* update XDP Tx rings counters */
4988 	if (ice_is_xdp_ena_vsi(vsi))
4989 		ice_update_vsi_tx_ring_stats(vsi, vsi->xdp_rings,
4990 					     vsi->num_xdp_txq);
4991 
4992 	rcu_read_unlock();
4993 }
4994 
4995 /**
4996  * ice_update_vsi_stats - Update VSI stats counters
4997  * @vsi: the VSI to be updated
4998  */
4999 void ice_update_vsi_stats(struct ice_vsi *vsi)
5000 {
5001 	struct rtnl_link_stats64 *cur_ns = &vsi->net_stats;
5002 	struct ice_eth_stats *cur_es = &vsi->eth_stats;
5003 	struct ice_pf *pf = vsi->back;
5004 
5005 	if (test_bit(__ICE_DOWN, vsi->state) ||
5006 	    test_bit(__ICE_CFG_BUSY, pf->state))
5007 		return;
5008 
5009 	/* get stats as recorded by Tx/Rx rings */
5010 	ice_update_vsi_ring_stats(vsi);
5011 
5012 	/* get VSI stats as recorded by the hardware */
5013 	ice_update_eth_stats(vsi);
5014 
5015 	cur_ns->tx_errors = cur_es->tx_errors;
5016 	cur_ns->rx_dropped = cur_es->rx_discards;
5017 	cur_ns->tx_dropped = cur_es->tx_discards;
5018 	cur_ns->multicast = cur_es->rx_multicast;
5019 
5020 	/* update some more netdev stats if this is main VSI */
5021 	if (vsi->type == ICE_VSI_PF) {
5022 		cur_ns->rx_crc_errors = pf->stats.crc_errors;
5023 		cur_ns->rx_errors = pf->stats.crc_errors +
5024 				    pf->stats.illegal_bytes +
5025 				    pf->stats.rx_len_errors +
5026 				    pf->stats.rx_undersize +
5027 				    pf->hw_csum_rx_error +
5028 				    pf->stats.rx_jabber +
5029 				    pf->stats.rx_fragments +
5030 				    pf->stats.rx_oversize;
5031 		cur_ns->rx_length_errors = pf->stats.rx_len_errors;
5032 		/* record drops from the port level */
5033 		cur_ns->rx_missed_errors = pf->stats.eth.rx_discards;
5034 	}
5035 }
5036 
5037 /**
5038  * ice_update_pf_stats - Update PF port stats counters
5039  * @pf: PF whose stats needs to be updated
5040  */
5041 void ice_update_pf_stats(struct ice_pf *pf)
5042 {
5043 	struct ice_hw_port_stats *prev_ps, *cur_ps;
5044 	struct ice_hw *hw = &pf->hw;
5045 	u16 fd_ctr_base;
5046 	u8 port;
5047 
5048 	port = hw->port_info->lport;
5049 	prev_ps = &pf->stats_prev;
5050 	cur_ps = &pf->stats;
5051 
5052 	ice_stat_update40(hw, GLPRT_GORCL(port), pf->stat_prev_loaded,
5053 			  &prev_ps->eth.rx_bytes,
5054 			  &cur_ps->eth.rx_bytes);
5055 
5056 	ice_stat_update40(hw, GLPRT_UPRCL(port), pf->stat_prev_loaded,
5057 			  &prev_ps->eth.rx_unicast,
5058 			  &cur_ps->eth.rx_unicast);
5059 
5060 	ice_stat_update40(hw, GLPRT_MPRCL(port), pf->stat_prev_loaded,
5061 			  &prev_ps->eth.rx_multicast,
5062 			  &cur_ps->eth.rx_multicast);
5063 
5064 	ice_stat_update40(hw, GLPRT_BPRCL(port), pf->stat_prev_loaded,
5065 			  &prev_ps->eth.rx_broadcast,
5066 			  &cur_ps->eth.rx_broadcast);
5067 
5068 	ice_stat_update32(hw, PRTRPB_RDPC, pf->stat_prev_loaded,
5069 			  &prev_ps->eth.rx_discards,
5070 			  &cur_ps->eth.rx_discards);
5071 
5072 	ice_stat_update40(hw, GLPRT_GOTCL(port), pf->stat_prev_loaded,
5073 			  &prev_ps->eth.tx_bytes,
5074 			  &cur_ps->eth.tx_bytes);
5075 
5076 	ice_stat_update40(hw, GLPRT_UPTCL(port), pf->stat_prev_loaded,
5077 			  &prev_ps->eth.tx_unicast,
5078 			  &cur_ps->eth.tx_unicast);
5079 
5080 	ice_stat_update40(hw, GLPRT_MPTCL(port), pf->stat_prev_loaded,
5081 			  &prev_ps->eth.tx_multicast,
5082 			  &cur_ps->eth.tx_multicast);
5083 
5084 	ice_stat_update40(hw, GLPRT_BPTCL(port), pf->stat_prev_loaded,
5085 			  &prev_ps->eth.tx_broadcast,
5086 			  &cur_ps->eth.tx_broadcast);
5087 
5088 	ice_stat_update32(hw, GLPRT_TDOLD(port), pf->stat_prev_loaded,
5089 			  &prev_ps->tx_dropped_link_down,
5090 			  &cur_ps->tx_dropped_link_down);
5091 
5092 	ice_stat_update40(hw, GLPRT_PRC64L(port), pf->stat_prev_loaded,
5093 			  &prev_ps->rx_size_64, &cur_ps->rx_size_64);
5094 
5095 	ice_stat_update40(hw, GLPRT_PRC127L(port), pf->stat_prev_loaded,
5096 			  &prev_ps->rx_size_127, &cur_ps->rx_size_127);
5097 
5098 	ice_stat_update40(hw, GLPRT_PRC255L(port), pf->stat_prev_loaded,
5099 			  &prev_ps->rx_size_255, &cur_ps->rx_size_255);
5100 
5101 	ice_stat_update40(hw, GLPRT_PRC511L(port), pf->stat_prev_loaded,
5102 			  &prev_ps->rx_size_511, &cur_ps->rx_size_511);
5103 
5104 	ice_stat_update40(hw, GLPRT_PRC1023L(port), pf->stat_prev_loaded,
5105 			  &prev_ps->rx_size_1023, &cur_ps->rx_size_1023);
5106 
5107 	ice_stat_update40(hw, GLPRT_PRC1522L(port), pf->stat_prev_loaded,
5108 			  &prev_ps->rx_size_1522, &cur_ps->rx_size_1522);
5109 
5110 	ice_stat_update40(hw, GLPRT_PRC9522L(port), pf->stat_prev_loaded,
5111 			  &prev_ps->rx_size_big, &cur_ps->rx_size_big);
5112 
5113 	ice_stat_update40(hw, GLPRT_PTC64L(port), pf->stat_prev_loaded,
5114 			  &prev_ps->tx_size_64, &cur_ps->tx_size_64);
5115 
5116 	ice_stat_update40(hw, GLPRT_PTC127L(port), pf->stat_prev_loaded,
5117 			  &prev_ps->tx_size_127, &cur_ps->tx_size_127);
5118 
5119 	ice_stat_update40(hw, GLPRT_PTC255L(port), pf->stat_prev_loaded,
5120 			  &prev_ps->tx_size_255, &cur_ps->tx_size_255);
5121 
5122 	ice_stat_update40(hw, GLPRT_PTC511L(port), pf->stat_prev_loaded,
5123 			  &prev_ps->tx_size_511, &cur_ps->tx_size_511);
5124 
5125 	ice_stat_update40(hw, GLPRT_PTC1023L(port), pf->stat_prev_loaded,
5126 			  &prev_ps->tx_size_1023, &cur_ps->tx_size_1023);
5127 
5128 	ice_stat_update40(hw, GLPRT_PTC1522L(port), pf->stat_prev_loaded,
5129 			  &prev_ps->tx_size_1522, &cur_ps->tx_size_1522);
5130 
5131 	ice_stat_update40(hw, GLPRT_PTC9522L(port), pf->stat_prev_loaded,
5132 			  &prev_ps->tx_size_big, &cur_ps->tx_size_big);
5133 
5134 	fd_ctr_base = hw->fd_ctr_base;
5135 
5136 	ice_stat_update40(hw,
5137 			  GLSTAT_FD_CNT0L(ICE_FD_SB_STAT_IDX(fd_ctr_base)),
5138 			  pf->stat_prev_loaded, &prev_ps->fd_sb_match,
5139 			  &cur_ps->fd_sb_match);
5140 	ice_stat_update32(hw, GLPRT_LXONRXC(port), pf->stat_prev_loaded,
5141 			  &prev_ps->link_xon_rx, &cur_ps->link_xon_rx);
5142 
5143 	ice_stat_update32(hw, GLPRT_LXOFFRXC(port), pf->stat_prev_loaded,
5144 			  &prev_ps->link_xoff_rx, &cur_ps->link_xoff_rx);
5145 
5146 	ice_stat_update32(hw, GLPRT_LXONTXC(port), pf->stat_prev_loaded,
5147 			  &prev_ps->link_xon_tx, &cur_ps->link_xon_tx);
5148 
5149 	ice_stat_update32(hw, GLPRT_LXOFFTXC(port), pf->stat_prev_loaded,
5150 			  &prev_ps->link_xoff_tx, &cur_ps->link_xoff_tx);
5151 
5152 	ice_update_dcb_stats(pf);
5153 
5154 	ice_stat_update32(hw, GLPRT_CRCERRS(port), pf->stat_prev_loaded,
5155 			  &prev_ps->crc_errors, &cur_ps->crc_errors);
5156 
5157 	ice_stat_update32(hw, GLPRT_ILLERRC(port), pf->stat_prev_loaded,
5158 			  &prev_ps->illegal_bytes, &cur_ps->illegal_bytes);
5159 
5160 	ice_stat_update32(hw, GLPRT_MLFC(port), pf->stat_prev_loaded,
5161 			  &prev_ps->mac_local_faults,
5162 			  &cur_ps->mac_local_faults);
5163 
5164 	ice_stat_update32(hw, GLPRT_MRFC(port), pf->stat_prev_loaded,
5165 			  &prev_ps->mac_remote_faults,
5166 			  &cur_ps->mac_remote_faults);
5167 
5168 	ice_stat_update32(hw, GLPRT_RLEC(port), pf->stat_prev_loaded,
5169 			  &prev_ps->rx_len_errors, &cur_ps->rx_len_errors);
5170 
5171 	ice_stat_update32(hw, GLPRT_RUC(port), pf->stat_prev_loaded,
5172 			  &prev_ps->rx_undersize, &cur_ps->rx_undersize);
5173 
5174 	ice_stat_update32(hw, GLPRT_RFC(port), pf->stat_prev_loaded,
5175 			  &prev_ps->rx_fragments, &cur_ps->rx_fragments);
5176 
5177 	ice_stat_update32(hw, GLPRT_ROC(port), pf->stat_prev_loaded,
5178 			  &prev_ps->rx_oversize, &cur_ps->rx_oversize);
5179 
5180 	ice_stat_update32(hw, GLPRT_RJC(port), pf->stat_prev_loaded,
5181 			  &prev_ps->rx_jabber, &cur_ps->rx_jabber);
5182 
5183 	cur_ps->fd_sb_status = test_bit(ICE_FLAG_FD_ENA, pf->flags) ? 1 : 0;
5184 
5185 	pf->stat_prev_loaded = true;
5186 }
5187 
5188 /**
5189  * ice_get_stats64 - get statistics for network device structure
5190  * @netdev: network interface device structure
5191  * @stats: main device statistics structure
5192  */
5193 static
5194 void ice_get_stats64(struct net_device *netdev, struct rtnl_link_stats64 *stats)
5195 {
5196 	struct ice_netdev_priv *np = netdev_priv(netdev);
5197 	struct rtnl_link_stats64 *vsi_stats;
5198 	struct ice_vsi *vsi = np->vsi;
5199 
5200 	vsi_stats = &vsi->net_stats;
5201 
5202 	if (!vsi->num_txq || !vsi->num_rxq)
5203 		return;
5204 
5205 	/* netdev packet/byte stats come from ring counter. These are obtained
5206 	 * by summing up ring counters (done by ice_update_vsi_ring_stats).
5207 	 * But, only call the update routine and read the registers if VSI is
5208 	 * not down.
5209 	 */
5210 	if (!test_bit(__ICE_DOWN, vsi->state))
5211 		ice_update_vsi_ring_stats(vsi);
5212 	stats->tx_packets = vsi_stats->tx_packets;
5213 	stats->tx_bytes = vsi_stats->tx_bytes;
5214 	stats->rx_packets = vsi_stats->rx_packets;
5215 	stats->rx_bytes = vsi_stats->rx_bytes;
5216 
5217 	/* The rest of the stats can be read from the hardware but instead we
5218 	 * just return values that the watchdog task has already obtained from
5219 	 * the hardware.
5220 	 */
5221 	stats->multicast = vsi_stats->multicast;
5222 	stats->tx_errors = vsi_stats->tx_errors;
5223 	stats->tx_dropped = vsi_stats->tx_dropped;
5224 	stats->rx_errors = vsi_stats->rx_errors;
5225 	stats->rx_dropped = vsi_stats->rx_dropped;
5226 	stats->rx_crc_errors = vsi_stats->rx_crc_errors;
5227 	stats->rx_length_errors = vsi_stats->rx_length_errors;
5228 }
5229 
5230 /**
5231  * ice_napi_disable_all - Disable NAPI for all q_vectors in the VSI
5232  * @vsi: VSI having NAPI disabled
5233  */
5234 static void ice_napi_disable_all(struct ice_vsi *vsi)
5235 {
5236 	int q_idx;
5237 
5238 	if (!vsi->netdev)
5239 		return;
5240 
5241 	ice_for_each_q_vector(vsi, q_idx) {
5242 		struct ice_q_vector *q_vector = vsi->q_vectors[q_idx];
5243 
5244 		if (q_vector->rx.ring || q_vector->tx.ring)
5245 			napi_disable(&q_vector->napi);
5246 	}
5247 }
5248 
5249 /**
5250  * ice_down - Shutdown the connection
5251  * @vsi: The VSI being stopped
5252  */
5253 int ice_down(struct ice_vsi *vsi)
5254 {
5255 	int i, tx_err, rx_err, link_err = 0;
5256 
5257 	/* Caller of this function is expected to set the
5258 	 * vsi->state __ICE_DOWN bit
5259 	 */
5260 	if (vsi->netdev) {
5261 		netif_carrier_off(vsi->netdev);
5262 		netif_tx_disable(vsi->netdev);
5263 	}
5264 
5265 	ice_vsi_dis_irq(vsi);
5266 
5267 	tx_err = ice_vsi_stop_lan_tx_rings(vsi, ICE_NO_RESET, 0);
5268 	if (tx_err)
5269 		netdev_err(vsi->netdev, "Failed stop Tx rings, VSI %d error %d\n",
5270 			   vsi->vsi_num, tx_err);
5271 	if (!tx_err && ice_is_xdp_ena_vsi(vsi)) {
5272 		tx_err = ice_vsi_stop_xdp_tx_rings(vsi);
5273 		if (tx_err)
5274 			netdev_err(vsi->netdev, "Failed stop XDP rings, VSI %d error %d\n",
5275 				   vsi->vsi_num, tx_err);
5276 	}
5277 
5278 	rx_err = ice_vsi_stop_all_rx_rings(vsi);
5279 	if (rx_err)
5280 		netdev_err(vsi->netdev, "Failed stop Rx rings, VSI %d error %d\n",
5281 			   vsi->vsi_num, rx_err);
5282 
5283 	ice_napi_disable_all(vsi);
5284 
5285 	if (test_bit(ICE_FLAG_LINK_DOWN_ON_CLOSE_ENA, vsi->back->flags)) {
5286 		link_err = ice_force_phys_link_state(vsi, false);
5287 		if (link_err)
5288 			netdev_err(vsi->netdev, "Failed to set physical link down, VSI %d error %d\n",
5289 				   vsi->vsi_num, link_err);
5290 	}
5291 
5292 	ice_for_each_txq(vsi, i)
5293 		ice_clean_tx_ring(vsi->tx_rings[i]);
5294 
5295 	ice_for_each_rxq(vsi, i)
5296 		ice_clean_rx_ring(vsi->rx_rings[i]);
5297 
5298 	if (tx_err || rx_err || link_err) {
5299 		netdev_err(vsi->netdev, "Failed to close VSI 0x%04X on switch 0x%04X\n",
5300 			   vsi->vsi_num, vsi->vsw->sw_id);
5301 		return -EIO;
5302 	}
5303 
5304 	return 0;
5305 }
5306 
5307 /**
5308  * ice_vsi_setup_tx_rings - Allocate VSI Tx queue resources
5309  * @vsi: VSI having resources allocated
5310  *
5311  * Return 0 on success, negative on failure
5312  */
5313 int ice_vsi_setup_tx_rings(struct ice_vsi *vsi)
5314 {
5315 	int i, err = 0;
5316 
5317 	if (!vsi->num_txq) {
5318 		dev_err(ice_pf_to_dev(vsi->back), "VSI %d has 0 Tx queues\n",
5319 			vsi->vsi_num);
5320 		return -EINVAL;
5321 	}
5322 
5323 	ice_for_each_txq(vsi, i) {
5324 		struct ice_ring *ring = vsi->tx_rings[i];
5325 
5326 		if (!ring)
5327 			return -EINVAL;
5328 
5329 		ring->netdev = vsi->netdev;
5330 		err = ice_setup_tx_ring(ring);
5331 		if (err)
5332 			break;
5333 	}
5334 
5335 	return err;
5336 }
5337 
5338 /**
5339  * ice_vsi_setup_rx_rings - Allocate VSI Rx queue resources
5340  * @vsi: VSI having resources allocated
5341  *
5342  * Return 0 on success, negative on failure
5343  */
5344 int ice_vsi_setup_rx_rings(struct ice_vsi *vsi)
5345 {
5346 	int i, err = 0;
5347 
5348 	if (!vsi->num_rxq) {
5349 		dev_err(ice_pf_to_dev(vsi->back), "VSI %d has 0 Rx queues\n",
5350 			vsi->vsi_num);
5351 		return -EINVAL;
5352 	}
5353 
5354 	ice_for_each_rxq(vsi, i) {
5355 		struct ice_ring *ring = vsi->rx_rings[i];
5356 
5357 		if (!ring)
5358 			return -EINVAL;
5359 
5360 		ring->netdev = vsi->netdev;
5361 		err = ice_setup_rx_ring(ring);
5362 		if (err)
5363 			break;
5364 	}
5365 
5366 	return err;
5367 }
5368 
5369 /**
5370  * ice_vsi_open_ctrl - open control VSI for use
5371  * @vsi: the VSI to open
5372  *
5373  * Initialization of the Control VSI
5374  *
5375  * Returns 0 on success, negative value on error
5376  */
5377 int ice_vsi_open_ctrl(struct ice_vsi *vsi)
5378 {
5379 	char int_name[ICE_INT_NAME_STR_LEN];
5380 	struct ice_pf *pf = vsi->back;
5381 	struct device *dev;
5382 	int err;
5383 
5384 	dev = ice_pf_to_dev(pf);
5385 	/* allocate descriptors */
5386 	err = ice_vsi_setup_tx_rings(vsi);
5387 	if (err)
5388 		goto err_setup_tx;
5389 
5390 	err = ice_vsi_setup_rx_rings(vsi);
5391 	if (err)
5392 		goto err_setup_rx;
5393 
5394 	err = ice_vsi_cfg(vsi);
5395 	if (err)
5396 		goto err_setup_rx;
5397 
5398 	snprintf(int_name, sizeof(int_name) - 1, "%s-%s:ctrl",
5399 		 dev_driver_string(dev), dev_name(dev));
5400 	err = ice_vsi_req_irq_msix(vsi, int_name);
5401 	if (err)
5402 		goto err_setup_rx;
5403 
5404 	ice_vsi_cfg_msix(vsi);
5405 
5406 	err = ice_vsi_start_all_rx_rings(vsi);
5407 	if (err)
5408 		goto err_up_complete;
5409 
5410 	clear_bit(__ICE_DOWN, vsi->state);
5411 	ice_vsi_ena_irq(vsi);
5412 
5413 	return 0;
5414 
5415 err_up_complete:
5416 	ice_down(vsi);
5417 err_setup_rx:
5418 	ice_vsi_free_rx_rings(vsi);
5419 err_setup_tx:
5420 	ice_vsi_free_tx_rings(vsi);
5421 
5422 	return err;
5423 }
5424 
5425 /**
5426  * ice_vsi_open - Called when a network interface is made active
5427  * @vsi: the VSI to open
5428  *
5429  * Initialization of the VSI
5430  *
5431  * Returns 0 on success, negative value on error
5432  */
5433 static int ice_vsi_open(struct ice_vsi *vsi)
5434 {
5435 	char int_name[ICE_INT_NAME_STR_LEN];
5436 	struct ice_pf *pf = vsi->back;
5437 	int err;
5438 
5439 	/* allocate descriptors */
5440 	err = ice_vsi_setup_tx_rings(vsi);
5441 	if (err)
5442 		goto err_setup_tx;
5443 
5444 	err = ice_vsi_setup_rx_rings(vsi);
5445 	if (err)
5446 		goto err_setup_rx;
5447 
5448 	err = ice_vsi_cfg(vsi);
5449 	if (err)
5450 		goto err_setup_rx;
5451 
5452 	snprintf(int_name, sizeof(int_name) - 1, "%s-%s",
5453 		 dev_driver_string(ice_pf_to_dev(pf)), vsi->netdev->name);
5454 	err = ice_vsi_req_irq_msix(vsi, int_name);
5455 	if (err)
5456 		goto err_setup_rx;
5457 
5458 	/* Notify the stack of the actual queue counts. */
5459 	err = netif_set_real_num_tx_queues(vsi->netdev, vsi->num_txq);
5460 	if (err)
5461 		goto err_set_qs;
5462 
5463 	err = netif_set_real_num_rx_queues(vsi->netdev, vsi->num_rxq);
5464 	if (err)
5465 		goto err_set_qs;
5466 
5467 	err = ice_up_complete(vsi);
5468 	if (err)
5469 		goto err_up_complete;
5470 
5471 	return 0;
5472 
5473 err_up_complete:
5474 	ice_down(vsi);
5475 err_set_qs:
5476 	ice_vsi_free_irq(vsi);
5477 err_setup_rx:
5478 	ice_vsi_free_rx_rings(vsi);
5479 err_setup_tx:
5480 	ice_vsi_free_tx_rings(vsi);
5481 
5482 	return err;
5483 }
5484 
5485 /**
5486  * ice_vsi_release_all - Delete all VSIs
5487  * @pf: PF from which all VSIs are being removed
5488  */
5489 static void ice_vsi_release_all(struct ice_pf *pf)
5490 {
5491 	int err, i;
5492 
5493 	if (!pf->vsi)
5494 		return;
5495 
5496 	ice_for_each_vsi(pf, i) {
5497 		if (!pf->vsi[i])
5498 			continue;
5499 
5500 		err = ice_vsi_release(pf->vsi[i]);
5501 		if (err)
5502 			dev_dbg(ice_pf_to_dev(pf), "Failed to release pf->vsi[%d], err %d, vsi_num = %d\n",
5503 				i, err, pf->vsi[i]->vsi_num);
5504 	}
5505 }
5506 
5507 /**
5508  * ice_vsi_rebuild_by_type - Rebuild VSI of a given type
5509  * @pf: pointer to the PF instance
5510  * @type: VSI type to rebuild
5511  *
5512  * Iterates through the pf->vsi array and rebuilds VSIs of the requested type
5513  */
5514 static int ice_vsi_rebuild_by_type(struct ice_pf *pf, enum ice_vsi_type type)
5515 {
5516 	struct device *dev = ice_pf_to_dev(pf);
5517 	enum ice_status status;
5518 	int i, err;
5519 
5520 	ice_for_each_vsi(pf, i) {
5521 		struct ice_vsi *vsi = pf->vsi[i];
5522 
5523 		if (!vsi || vsi->type != type)
5524 			continue;
5525 
5526 		/* rebuild the VSI */
5527 		err = ice_vsi_rebuild(vsi, true);
5528 		if (err) {
5529 			dev_err(dev, "rebuild VSI failed, err %d, VSI index %d, type %s\n",
5530 				err, vsi->idx, ice_vsi_type_str(type));
5531 			return err;
5532 		}
5533 
5534 		/* replay filters for the VSI */
5535 		status = ice_replay_vsi(&pf->hw, vsi->idx);
5536 		if (status) {
5537 			dev_err(dev, "replay VSI failed, status %s, VSI index %d, type %s\n",
5538 				ice_stat_str(status), vsi->idx,
5539 				ice_vsi_type_str(type));
5540 			return -EIO;
5541 		}
5542 
5543 		/* Re-map HW VSI number, using VSI handle that has been
5544 		 * previously validated in ice_replay_vsi() call above
5545 		 */
5546 		vsi->vsi_num = ice_get_hw_vsi_num(&pf->hw, vsi->idx);
5547 
5548 		/* enable the VSI */
5549 		err = ice_ena_vsi(vsi, false);
5550 		if (err) {
5551 			dev_err(dev, "enable VSI failed, err %d, VSI index %d, type %s\n",
5552 				err, vsi->idx, ice_vsi_type_str(type));
5553 			return err;
5554 		}
5555 
5556 		dev_info(dev, "VSI rebuilt. VSI index %d, type %s\n", vsi->idx,
5557 			 ice_vsi_type_str(type));
5558 	}
5559 
5560 	return 0;
5561 }
5562 
5563 /**
5564  * ice_update_pf_netdev_link - Update PF netdev link status
5565  * @pf: pointer to the PF instance
5566  */
5567 static void ice_update_pf_netdev_link(struct ice_pf *pf)
5568 {
5569 	bool link_up;
5570 	int i;
5571 
5572 	ice_for_each_vsi(pf, i) {
5573 		struct ice_vsi *vsi = pf->vsi[i];
5574 
5575 		if (!vsi || vsi->type != ICE_VSI_PF)
5576 			return;
5577 
5578 		ice_get_link_status(pf->vsi[i]->port_info, &link_up);
5579 		if (link_up) {
5580 			netif_carrier_on(pf->vsi[i]->netdev);
5581 			netif_tx_wake_all_queues(pf->vsi[i]->netdev);
5582 		} else {
5583 			netif_carrier_off(pf->vsi[i]->netdev);
5584 			netif_tx_stop_all_queues(pf->vsi[i]->netdev);
5585 		}
5586 	}
5587 }
5588 
5589 /**
5590  * ice_rebuild - rebuild after reset
5591  * @pf: PF to rebuild
5592  * @reset_type: type of reset
5593  *
5594  * Do not rebuild VF VSI in this flow because that is already handled via
5595  * ice_reset_all_vfs(). This is because requirements for resetting a VF after a
5596  * PFR/CORER/GLOBER/etc. are different than the normal flow. Also, we don't want
5597  * to reset/rebuild all the VF VSI twice.
5598  */
5599 static void ice_rebuild(struct ice_pf *pf, enum ice_reset_req reset_type)
5600 {
5601 	struct device *dev = ice_pf_to_dev(pf);
5602 	struct ice_hw *hw = &pf->hw;
5603 	enum ice_status ret;
5604 	int err;
5605 
5606 	if (test_bit(__ICE_DOWN, pf->state))
5607 		goto clear_recovery;
5608 
5609 	dev_dbg(dev, "rebuilding PF after reset_type=%d\n", reset_type);
5610 
5611 	ret = ice_init_all_ctrlq(hw);
5612 	if (ret) {
5613 		dev_err(dev, "control queues init failed %s\n",
5614 			ice_stat_str(ret));
5615 		goto err_init_ctrlq;
5616 	}
5617 
5618 	/* if DDP was previously loaded successfully */
5619 	if (!ice_is_safe_mode(pf)) {
5620 		/* reload the SW DB of filter tables */
5621 		if (reset_type == ICE_RESET_PFR)
5622 			ice_fill_blk_tbls(hw);
5623 		else
5624 			/* Reload DDP Package after CORER/GLOBR reset */
5625 			ice_load_pkg(NULL, pf);
5626 	}
5627 
5628 	ret = ice_clear_pf_cfg(hw);
5629 	if (ret) {
5630 		dev_err(dev, "clear PF configuration failed %s\n",
5631 			ice_stat_str(ret));
5632 		goto err_init_ctrlq;
5633 	}
5634 
5635 	if (pf->first_sw->dflt_vsi_ena)
5636 		dev_info(dev, "Clearing default VSI, re-enable after reset completes\n");
5637 	/* clear the default VSI configuration if it exists */
5638 	pf->first_sw->dflt_vsi = NULL;
5639 	pf->first_sw->dflt_vsi_ena = false;
5640 
5641 	ice_clear_pxe_mode(hw);
5642 
5643 	ret = ice_get_caps(hw);
5644 	if (ret) {
5645 		dev_err(dev, "ice_get_caps failed %s\n", ice_stat_str(ret));
5646 		goto err_init_ctrlq;
5647 	}
5648 
5649 	ret = ice_aq_set_mac_cfg(hw, ICE_AQ_SET_MAC_FRAME_SIZE_MAX, NULL);
5650 	if (ret) {
5651 		dev_err(dev, "set_mac_cfg failed %s\n", ice_stat_str(ret));
5652 		goto err_init_ctrlq;
5653 	}
5654 
5655 	err = ice_sched_init_port(hw->port_info);
5656 	if (err)
5657 		goto err_sched_init_port;
5658 
5659 	err = ice_update_link_info(hw->port_info);
5660 	if (err)
5661 		dev_err(dev, "Get link status error %d\n", err);
5662 
5663 	/* start misc vector */
5664 	err = ice_req_irq_msix_misc(pf);
5665 	if (err) {
5666 		dev_err(dev, "misc vector setup failed: %d\n", err);
5667 		goto err_sched_init_port;
5668 	}
5669 
5670 	if (test_bit(ICE_FLAG_FD_ENA, pf->flags)) {
5671 		wr32(hw, PFQF_FD_ENA, PFQF_FD_ENA_FD_ENA_M);
5672 		if (!rd32(hw, PFQF_FD_SIZE)) {
5673 			u16 unused, guar, b_effort;
5674 
5675 			guar = hw->func_caps.fd_fltr_guar;
5676 			b_effort = hw->func_caps.fd_fltr_best_effort;
5677 
5678 			/* force guaranteed filter pool for PF */
5679 			ice_alloc_fd_guar_item(hw, &unused, guar);
5680 			/* force shared filter pool for PF */
5681 			ice_alloc_fd_shrd_item(hw, &unused, b_effort);
5682 		}
5683 	}
5684 
5685 	if (test_bit(ICE_FLAG_DCB_ENA, pf->flags))
5686 		ice_dcb_rebuild(pf);
5687 
5688 	/* rebuild PF VSI */
5689 	err = ice_vsi_rebuild_by_type(pf, ICE_VSI_PF);
5690 	if (err) {
5691 		dev_err(dev, "PF VSI rebuild failed: %d\n", err);
5692 		goto err_vsi_rebuild;
5693 	}
5694 
5695 	/* If Flow Director is active */
5696 	if (test_bit(ICE_FLAG_FD_ENA, pf->flags)) {
5697 		err = ice_vsi_rebuild_by_type(pf, ICE_VSI_CTRL);
5698 		if (err) {
5699 			dev_err(dev, "control VSI rebuild failed: %d\n", err);
5700 			goto err_vsi_rebuild;
5701 		}
5702 
5703 		/* replay HW Flow Director recipes */
5704 		if (hw->fdir_prof)
5705 			ice_fdir_replay_flows(hw);
5706 
5707 		/* replay Flow Director filters */
5708 		ice_fdir_replay_fltrs(pf);
5709 
5710 		ice_rebuild_arfs(pf);
5711 	}
5712 
5713 	ice_update_pf_netdev_link(pf);
5714 
5715 	/* tell the firmware we are up */
5716 	ret = ice_send_version(pf);
5717 	if (ret) {
5718 		dev_err(dev, "Rebuild failed due to error sending driver version: %s\n",
5719 			ice_stat_str(ret));
5720 		goto err_vsi_rebuild;
5721 	}
5722 
5723 	ice_replay_post(hw);
5724 
5725 	/* if we get here, reset flow is successful */
5726 	clear_bit(__ICE_RESET_FAILED, pf->state);
5727 	return;
5728 
5729 err_vsi_rebuild:
5730 err_sched_init_port:
5731 	ice_sched_cleanup_all(hw);
5732 err_init_ctrlq:
5733 	ice_shutdown_all_ctrlq(hw);
5734 	set_bit(__ICE_RESET_FAILED, pf->state);
5735 clear_recovery:
5736 	/* set this bit in PF state to control service task scheduling */
5737 	set_bit(__ICE_NEEDS_RESTART, pf->state);
5738 	dev_err(dev, "Rebuild failed, unload and reload driver\n");
5739 }
5740 
5741 /**
5742  * ice_max_xdp_frame_size - returns the maximum allowed frame size for XDP
5743  * @vsi: Pointer to VSI structure
5744  */
5745 static int ice_max_xdp_frame_size(struct ice_vsi *vsi)
5746 {
5747 	if (PAGE_SIZE >= 8192 || test_bit(ICE_FLAG_LEGACY_RX, vsi->back->flags))
5748 		return ICE_RXBUF_2048 - XDP_PACKET_HEADROOM;
5749 	else
5750 		return ICE_RXBUF_3072;
5751 }
5752 
5753 /**
5754  * ice_change_mtu - NDO callback to change the MTU
5755  * @netdev: network interface device structure
5756  * @new_mtu: new value for maximum frame size
5757  *
5758  * Returns 0 on success, negative on failure
5759  */
5760 static int ice_change_mtu(struct net_device *netdev, int new_mtu)
5761 {
5762 	struct ice_netdev_priv *np = netdev_priv(netdev);
5763 	struct ice_vsi *vsi = np->vsi;
5764 	struct ice_pf *pf = vsi->back;
5765 	u8 count = 0;
5766 
5767 	if (new_mtu == (int)netdev->mtu) {
5768 		netdev_warn(netdev, "MTU is already %u\n", netdev->mtu);
5769 		return 0;
5770 	}
5771 
5772 	if (ice_is_xdp_ena_vsi(vsi)) {
5773 		int frame_size = ice_max_xdp_frame_size(vsi);
5774 
5775 		if (new_mtu + ICE_ETH_PKT_HDR_PAD > frame_size) {
5776 			netdev_err(netdev, "max MTU for XDP usage is %d\n",
5777 				   frame_size - ICE_ETH_PKT_HDR_PAD);
5778 			return -EINVAL;
5779 		}
5780 	}
5781 
5782 	if (new_mtu < (int)netdev->min_mtu) {
5783 		netdev_err(netdev, "new MTU invalid. min_mtu is %d\n",
5784 			   netdev->min_mtu);
5785 		return -EINVAL;
5786 	} else if (new_mtu > (int)netdev->max_mtu) {
5787 		netdev_err(netdev, "new MTU invalid. max_mtu is %d\n",
5788 			   netdev->min_mtu);
5789 		return -EINVAL;
5790 	}
5791 	/* if a reset is in progress, wait for some time for it to complete */
5792 	do {
5793 		if (ice_is_reset_in_progress(pf->state)) {
5794 			count++;
5795 			usleep_range(1000, 2000);
5796 		} else {
5797 			break;
5798 		}
5799 
5800 	} while (count < 100);
5801 
5802 	if (count == 100) {
5803 		netdev_err(netdev, "can't change MTU. Device is busy\n");
5804 		return -EBUSY;
5805 	}
5806 
5807 	netdev->mtu = (unsigned int)new_mtu;
5808 
5809 	/* if VSI is up, bring it down and then back up */
5810 	if (!test_and_set_bit(__ICE_DOWN, vsi->state)) {
5811 		int err;
5812 
5813 		err = ice_down(vsi);
5814 		if (err) {
5815 			netdev_err(netdev, "change MTU if_up err %d\n", err);
5816 			return err;
5817 		}
5818 
5819 		err = ice_up(vsi);
5820 		if (err) {
5821 			netdev_err(netdev, "change MTU if_up err %d\n", err);
5822 			return err;
5823 		}
5824 	}
5825 
5826 	netdev_dbg(netdev, "changed MTU to %d\n", new_mtu);
5827 	return 0;
5828 }
5829 
5830 /**
5831  * ice_aq_str - convert AQ err code to a string
5832  * @aq_err: the AQ error code to convert
5833  */
5834 const char *ice_aq_str(enum ice_aq_err aq_err)
5835 {
5836 	switch (aq_err) {
5837 	case ICE_AQ_RC_OK:
5838 		return "OK";
5839 	case ICE_AQ_RC_EPERM:
5840 		return "ICE_AQ_RC_EPERM";
5841 	case ICE_AQ_RC_ENOENT:
5842 		return "ICE_AQ_RC_ENOENT";
5843 	case ICE_AQ_RC_ENOMEM:
5844 		return "ICE_AQ_RC_ENOMEM";
5845 	case ICE_AQ_RC_EBUSY:
5846 		return "ICE_AQ_RC_EBUSY";
5847 	case ICE_AQ_RC_EEXIST:
5848 		return "ICE_AQ_RC_EEXIST";
5849 	case ICE_AQ_RC_EINVAL:
5850 		return "ICE_AQ_RC_EINVAL";
5851 	case ICE_AQ_RC_ENOSPC:
5852 		return "ICE_AQ_RC_ENOSPC";
5853 	case ICE_AQ_RC_ENOSYS:
5854 		return "ICE_AQ_RC_ENOSYS";
5855 	case ICE_AQ_RC_EMODE:
5856 		return "ICE_AQ_RC_EMODE";
5857 	case ICE_AQ_RC_ENOSEC:
5858 		return "ICE_AQ_RC_ENOSEC";
5859 	case ICE_AQ_RC_EBADSIG:
5860 		return "ICE_AQ_RC_EBADSIG";
5861 	case ICE_AQ_RC_ESVN:
5862 		return "ICE_AQ_RC_ESVN";
5863 	case ICE_AQ_RC_EBADMAN:
5864 		return "ICE_AQ_RC_EBADMAN";
5865 	case ICE_AQ_RC_EBADBUF:
5866 		return "ICE_AQ_RC_EBADBUF";
5867 	}
5868 
5869 	return "ICE_AQ_RC_UNKNOWN";
5870 }
5871 
5872 /**
5873  * ice_stat_str - convert status err code to a string
5874  * @stat_err: the status error code to convert
5875  */
5876 const char *ice_stat_str(enum ice_status stat_err)
5877 {
5878 	switch (stat_err) {
5879 	case ICE_SUCCESS:
5880 		return "OK";
5881 	case ICE_ERR_PARAM:
5882 		return "ICE_ERR_PARAM";
5883 	case ICE_ERR_NOT_IMPL:
5884 		return "ICE_ERR_NOT_IMPL";
5885 	case ICE_ERR_NOT_READY:
5886 		return "ICE_ERR_NOT_READY";
5887 	case ICE_ERR_NOT_SUPPORTED:
5888 		return "ICE_ERR_NOT_SUPPORTED";
5889 	case ICE_ERR_BAD_PTR:
5890 		return "ICE_ERR_BAD_PTR";
5891 	case ICE_ERR_INVAL_SIZE:
5892 		return "ICE_ERR_INVAL_SIZE";
5893 	case ICE_ERR_DEVICE_NOT_SUPPORTED:
5894 		return "ICE_ERR_DEVICE_NOT_SUPPORTED";
5895 	case ICE_ERR_RESET_FAILED:
5896 		return "ICE_ERR_RESET_FAILED";
5897 	case ICE_ERR_FW_API_VER:
5898 		return "ICE_ERR_FW_API_VER";
5899 	case ICE_ERR_NO_MEMORY:
5900 		return "ICE_ERR_NO_MEMORY";
5901 	case ICE_ERR_CFG:
5902 		return "ICE_ERR_CFG";
5903 	case ICE_ERR_OUT_OF_RANGE:
5904 		return "ICE_ERR_OUT_OF_RANGE";
5905 	case ICE_ERR_ALREADY_EXISTS:
5906 		return "ICE_ERR_ALREADY_EXISTS";
5907 	case ICE_ERR_NVM_CHECKSUM:
5908 		return "ICE_ERR_NVM_CHECKSUM";
5909 	case ICE_ERR_BUF_TOO_SHORT:
5910 		return "ICE_ERR_BUF_TOO_SHORT";
5911 	case ICE_ERR_NVM_BLANK_MODE:
5912 		return "ICE_ERR_NVM_BLANK_MODE";
5913 	case ICE_ERR_IN_USE:
5914 		return "ICE_ERR_IN_USE";
5915 	case ICE_ERR_MAX_LIMIT:
5916 		return "ICE_ERR_MAX_LIMIT";
5917 	case ICE_ERR_RESET_ONGOING:
5918 		return "ICE_ERR_RESET_ONGOING";
5919 	case ICE_ERR_HW_TABLE:
5920 		return "ICE_ERR_HW_TABLE";
5921 	case ICE_ERR_DOES_NOT_EXIST:
5922 		return "ICE_ERR_DOES_NOT_EXIST";
5923 	case ICE_ERR_FW_DDP_MISMATCH:
5924 		return "ICE_ERR_FW_DDP_MISMATCH";
5925 	case ICE_ERR_AQ_ERROR:
5926 		return "ICE_ERR_AQ_ERROR";
5927 	case ICE_ERR_AQ_TIMEOUT:
5928 		return "ICE_ERR_AQ_TIMEOUT";
5929 	case ICE_ERR_AQ_FULL:
5930 		return "ICE_ERR_AQ_FULL";
5931 	case ICE_ERR_AQ_NO_WORK:
5932 		return "ICE_ERR_AQ_NO_WORK";
5933 	case ICE_ERR_AQ_EMPTY:
5934 		return "ICE_ERR_AQ_EMPTY";
5935 	case ICE_ERR_AQ_FW_CRITICAL:
5936 		return "ICE_ERR_AQ_FW_CRITICAL";
5937 	}
5938 
5939 	return "ICE_ERR_UNKNOWN";
5940 }
5941 
5942 /**
5943  * ice_set_rss - Set RSS keys and lut
5944  * @vsi: Pointer to VSI structure
5945  * @seed: RSS hash seed
5946  * @lut: Lookup table
5947  * @lut_size: Lookup table size
5948  *
5949  * Returns 0 on success, negative on failure
5950  */
5951 int ice_set_rss(struct ice_vsi *vsi, u8 *seed, u8 *lut, u16 lut_size)
5952 {
5953 	struct ice_pf *pf = vsi->back;
5954 	struct ice_hw *hw = &pf->hw;
5955 	enum ice_status status;
5956 	struct device *dev;
5957 
5958 	dev = ice_pf_to_dev(pf);
5959 	if (seed) {
5960 		struct ice_aqc_get_set_rss_keys *buf =
5961 				  (struct ice_aqc_get_set_rss_keys *)seed;
5962 
5963 		status = ice_aq_set_rss_key(hw, vsi->idx, buf);
5964 
5965 		if (status) {
5966 			dev_err(dev, "Cannot set RSS key, err %s aq_err %s\n",
5967 				ice_stat_str(status),
5968 				ice_aq_str(hw->adminq.sq_last_status));
5969 			return -EIO;
5970 		}
5971 	}
5972 
5973 	if (lut) {
5974 		status = ice_aq_set_rss_lut(hw, vsi->idx, vsi->rss_lut_type,
5975 					    lut, lut_size);
5976 		if (status) {
5977 			dev_err(dev, "Cannot set RSS lut, err %s aq_err %s\n",
5978 				ice_stat_str(status),
5979 				ice_aq_str(hw->adminq.sq_last_status));
5980 			return -EIO;
5981 		}
5982 	}
5983 
5984 	return 0;
5985 }
5986 
5987 /**
5988  * ice_get_rss - Get RSS keys and lut
5989  * @vsi: Pointer to VSI structure
5990  * @seed: Buffer to store the keys
5991  * @lut: Buffer to store the lookup table entries
5992  * @lut_size: Size of buffer to store the lookup table entries
5993  *
5994  * Returns 0 on success, negative on failure
5995  */
5996 int ice_get_rss(struct ice_vsi *vsi, u8 *seed, u8 *lut, u16 lut_size)
5997 {
5998 	struct ice_pf *pf = vsi->back;
5999 	struct ice_hw *hw = &pf->hw;
6000 	enum ice_status status;
6001 	struct device *dev;
6002 
6003 	dev = ice_pf_to_dev(pf);
6004 	if (seed) {
6005 		struct ice_aqc_get_set_rss_keys *buf =
6006 				  (struct ice_aqc_get_set_rss_keys *)seed;
6007 
6008 		status = ice_aq_get_rss_key(hw, vsi->idx, buf);
6009 		if (status) {
6010 			dev_err(dev, "Cannot get RSS key, err %s aq_err %s\n",
6011 				ice_stat_str(status),
6012 				ice_aq_str(hw->adminq.sq_last_status));
6013 			return -EIO;
6014 		}
6015 	}
6016 
6017 	if (lut) {
6018 		status = ice_aq_get_rss_lut(hw, vsi->idx, vsi->rss_lut_type,
6019 					    lut, lut_size);
6020 		if (status) {
6021 			dev_err(dev, "Cannot get RSS lut, err %s aq_err %s\n",
6022 				ice_stat_str(status),
6023 				ice_aq_str(hw->adminq.sq_last_status));
6024 			return -EIO;
6025 		}
6026 	}
6027 
6028 	return 0;
6029 }
6030 
6031 /**
6032  * ice_bridge_getlink - Get the hardware bridge mode
6033  * @skb: skb buff
6034  * @pid: process ID
6035  * @seq: RTNL message seq
6036  * @dev: the netdev being configured
6037  * @filter_mask: filter mask passed in
6038  * @nlflags: netlink flags passed in
6039  *
6040  * Return the bridge mode (VEB/VEPA)
6041  */
6042 static int
6043 ice_bridge_getlink(struct sk_buff *skb, u32 pid, u32 seq,
6044 		   struct net_device *dev, u32 filter_mask, int nlflags)
6045 {
6046 	struct ice_netdev_priv *np = netdev_priv(dev);
6047 	struct ice_vsi *vsi = np->vsi;
6048 	struct ice_pf *pf = vsi->back;
6049 	u16 bmode;
6050 
6051 	bmode = pf->first_sw->bridge_mode;
6052 
6053 	return ndo_dflt_bridge_getlink(skb, pid, seq, dev, bmode, 0, 0, nlflags,
6054 				       filter_mask, NULL);
6055 }
6056 
6057 /**
6058  * ice_vsi_update_bridge_mode - Update VSI for switching bridge mode (VEB/VEPA)
6059  * @vsi: Pointer to VSI structure
6060  * @bmode: Hardware bridge mode (VEB/VEPA)
6061  *
6062  * Returns 0 on success, negative on failure
6063  */
6064 static int ice_vsi_update_bridge_mode(struct ice_vsi *vsi, u16 bmode)
6065 {
6066 	struct ice_aqc_vsi_props *vsi_props;
6067 	struct ice_hw *hw = &vsi->back->hw;
6068 	struct ice_vsi_ctx *ctxt;
6069 	enum ice_status status;
6070 	int ret = 0;
6071 
6072 	vsi_props = &vsi->info;
6073 
6074 	ctxt = kzalloc(sizeof(*ctxt), GFP_KERNEL);
6075 	if (!ctxt)
6076 		return -ENOMEM;
6077 
6078 	ctxt->info = vsi->info;
6079 
6080 	if (bmode == BRIDGE_MODE_VEB)
6081 		/* change from VEPA to VEB mode */
6082 		ctxt->info.sw_flags |= ICE_AQ_VSI_SW_FLAG_ALLOW_LB;
6083 	else
6084 		/* change from VEB to VEPA mode */
6085 		ctxt->info.sw_flags &= ~ICE_AQ_VSI_SW_FLAG_ALLOW_LB;
6086 	ctxt->info.valid_sections = cpu_to_le16(ICE_AQ_VSI_PROP_SW_VALID);
6087 
6088 	status = ice_update_vsi(hw, vsi->idx, ctxt, NULL);
6089 	if (status) {
6090 		dev_err(ice_pf_to_dev(vsi->back), "update VSI for bridge mode failed, bmode = %d err %s aq_err %s\n",
6091 			bmode, ice_stat_str(status),
6092 			ice_aq_str(hw->adminq.sq_last_status));
6093 		ret = -EIO;
6094 		goto out;
6095 	}
6096 	/* Update sw flags for book keeping */
6097 	vsi_props->sw_flags = ctxt->info.sw_flags;
6098 
6099 out:
6100 	kfree(ctxt);
6101 	return ret;
6102 }
6103 
6104 /**
6105  * ice_bridge_setlink - Set the hardware bridge mode
6106  * @dev: the netdev being configured
6107  * @nlh: RTNL message
6108  * @flags: bridge setlink flags
6109  * @extack: netlink extended ack
6110  *
6111  * Sets the bridge mode (VEB/VEPA) of the switch to which the netdev (VSI) is
6112  * hooked up to. Iterates through the PF VSI list and sets the loopback mode (if
6113  * not already set for all VSIs connected to this switch. And also update the
6114  * unicast switch filter rules for the corresponding switch of the netdev.
6115  */
6116 static int
6117 ice_bridge_setlink(struct net_device *dev, struct nlmsghdr *nlh,
6118 		   u16 __always_unused flags,
6119 		   struct netlink_ext_ack __always_unused *extack)
6120 {
6121 	struct ice_netdev_priv *np = netdev_priv(dev);
6122 	struct ice_pf *pf = np->vsi->back;
6123 	struct nlattr *attr, *br_spec;
6124 	struct ice_hw *hw = &pf->hw;
6125 	enum ice_status status;
6126 	struct ice_sw *pf_sw;
6127 	int rem, v, err = 0;
6128 
6129 	pf_sw = pf->first_sw;
6130 	/* find the attribute in the netlink message */
6131 	br_spec = nlmsg_find_attr(nlh, sizeof(struct ifinfomsg), IFLA_AF_SPEC);
6132 
6133 	nla_for_each_nested(attr, br_spec, rem) {
6134 		__u16 mode;
6135 
6136 		if (nla_type(attr) != IFLA_BRIDGE_MODE)
6137 			continue;
6138 		mode = nla_get_u16(attr);
6139 		if (mode != BRIDGE_MODE_VEPA && mode != BRIDGE_MODE_VEB)
6140 			return -EINVAL;
6141 		/* Continue  if bridge mode is not being flipped */
6142 		if (mode == pf_sw->bridge_mode)
6143 			continue;
6144 		/* Iterates through the PF VSI list and update the loopback
6145 		 * mode of the VSI
6146 		 */
6147 		ice_for_each_vsi(pf, v) {
6148 			if (!pf->vsi[v])
6149 				continue;
6150 			err = ice_vsi_update_bridge_mode(pf->vsi[v], mode);
6151 			if (err)
6152 				return err;
6153 		}
6154 
6155 		hw->evb_veb = (mode == BRIDGE_MODE_VEB);
6156 		/* Update the unicast switch filter rules for the corresponding
6157 		 * switch of the netdev
6158 		 */
6159 		status = ice_update_sw_rule_bridge_mode(hw);
6160 		if (status) {
6161 			netdev_err(dev, "switch rule update failed, mode = %d err %s aq_err %s\n",
6162 				   mode, ice_stat_str(status),
6163 				   ice_aq_str(hw->adminq.sq_last_status));
6164 			/* revert hw->evb_veb */
6165 			hw->evb_veb = (pf_sw->bridge_mode == BRIDGE_MODE_VEB);
6166 			return -EIO;
6167 		}
6168 
6169 		pf_sw->bridge_mode = mode;
6170 	}
6171 
6172 	return 0;
6173 }
6174 
6175 /**
6176  * ice_tx_timeout - Respond to a Tx Hang
6177  * @netdev: network interface device structure
6178  * @txqueue: Tx queue
6179  */
6180 static void ice_tx_timeout(struct net_device *netdev, unsigned int txqueue)
6181 {
6182 	struct ice_netdev_priv *np = netdev_priv(netdev);
6183 	struct ice_ring *tx_ring = NULL;
6184 	struct ice_vsi *vsi = np->vsi;
6185 	struct ice_pf *pf = vsi->back;
6186 	u32 i;
6187 
6188 	pf->tx_timeout_count++;
6189 
6190 	/* Check if PFC is enabled for the TC to which the queue belongs
6191 	 * to. If yes then Tx timeout is not caused by a hung queue, no
6192 	 * need to reset and rebuild
6193 	 */
6194 	if (ice_is_pfc_causing_hung_q(pf, txqueue)) {
6195 		dev_info(ice_pf_to_dev(pf), "Fake Tx hang detected on queue %u, timeout caused by PFC storm\n",
6196 			 txqueue);
6197 		return;
6198 	}
6199 
6200 	/* now that we have an index, find the tx_ring struct */
6201 	for (i = 0; i < vsi->num_txq; i++)
6202 		if (vsi->tx_rings[i] && vsi->tx_rings[i]->desc)
6203 			if (txqueue == vsi->tx_rings[i]->q_index) {
6204 				tx_ring = vsi->tx_rings[i];
6205 				break;
6206 			}
6207 
6208 	/* Reset recovery level if enough time has elapsed after last timeout.
6209 	 * Also ensure no new reset action happens before next timeout period.
6210 	 */
6211 	if (time_after(jiffies, (pf->tx_timeout_last_recovery + HZ * 20)))
6212 		pf->tx_timeout_recovery_level = 1;
6213 	else if (time_before(jiffies, (pf->tx_timeout_last_recovery +
6214 				       netdev->watchdog_timeo)))
6215 		return;
6216 
6217 	if (tx_ring) {
6218 		struct ice_hw *hw = &pf->hw;
6219 		u32 head, val = 0;
6220 
6221 		head = (rd32(hw, QTX_COMM_HEAD(vsi->txq_map[txqueue])) &
6222 			QTX_COMM_HEAD_HEAD_M) >> QTX_COMM_HEAD_HEAD_S;
6223 		/* Read interrupt register */
6224 		val = rd32(hw, GLINT_DYN_CTL(tx_ring->q_vector->reg_idx));
6225 
6226 		netdev_info(netdev, "tx_timeout: VSI_num: %d, Q %u, NTC: 0x%x, HW_HEAD: 0x%x, NTU: 0x%x, INT: 0x%x\n",
6227 			    vsi->vsi_num, txqueue, tx_ring->next_to_clean,
6228 			    head, tx_ring->next_to_use, val);
6229 	}
6230 
6231 	pf->tx_timeout_last_recovery = jiffies;
6232 	netdev_info(netdev, "tx_timeout recovery level %d, txqueue %u\n",
6233 		    pf->tx_timeout_recovery_level, txqueue);
6234 
6235 	switch (pf->tx_timeout_recovery_level) {
6236 	case 1:
6237 		set_bit(__ICE_PFR_REQ, pf->state);
6238 		break;
6239 	case 2:
6240 		set_bit(__ICE_CORER_REQ, pf->state);
6241 		break;
6242 	case 3:
6243 		set_bit(__ICE_GLOBR_REQ, pf->state);
6244 		break;
6245 	default:
6246 		netdev_err(netdev, "tx_timeout recovery unsuccessful, device is in unrecoverable state.\n");
6247 		set_bit(__ICE_DOWN, pf->state);
6248 		set_bit(__ICE_NEEDS_RESTART, vsi->state);
6249 		set_bit(__ICE_SERVICE_DIS, pf->state);
6250 		break;
6251 	}
6252 
6253 	ice_service_task_schedule(pf);
6254 	pf->tx_timeout_recovery_level++;
6255 }
6256 
6257 /**
6258  * ice_udp_tunnel_add - Get notifications about UDP tunnel ports that come up
6259  * @netdev: This physical port's netdev
6260  * @ti: Tunnel endpoint information
6261  */
6262 static void
6263 ice_udp_tunnel_add(struct net_device *netdev, struct udp_tunnel_info *ti)
6264 {
6265 	struct ice_netdev_priv *np = netdev_priv(netdev);
6266 	struct ice_vsi *vsi = np->vsi;
6267 	struct ice_pf *pf = vsi->back;
6268 	enum ice_tunnel_type tnl_type;
6269 	u16 port = ntohs(ti->port);
6270 	enum ice_status status;
6271 
6272 	switch (ti->type) {
6273 	case UDP_TUNNEL_TYPE_VXLAN:
6274 		tnl_type = TNL_VXLAN;
6275 		break;
6276 	case UDP_TUNNEL_TYPE_GENEVE:
6277 		tnl_type = TNL_GENEVE;
6278 		break;
6279 	default:
6280 		netdev_err(netdev, "Unknown tunnel type\n");
6281 		return;
6282 	}
6283 
6284 	status = ice_create_tunnel(&pf->hw, tnl_type, port);
6285 	if (status == ICE_ERR_OUT_OF_RANGE)
6286 		netdev_info(netdev, "Max tunneled UDP ports reached, port %d not added\n",
6287 			    port);
6288 	else if (status)
6289 		netdev_err(netdev, "Error adding UDP tunnel - %s\n",
6290 			   ice_stat_str(status));
6291 }
6292 
6293 /**
6294  * ice_udp_tunnel_del - Get notifications about UDP tunnel ports that go away
6295  * @netdev: This physical port's netdev
6296  * @ti: Tunnel endpoint information
6297  */
6298 static void
6299 ice_udp_tunnel_del(struct net_device *netdev, struct udp_tunnel_info *ti)
6300 {
6301 	struct ice_netdev_priv *np = netdev_priv(netdev);
6302 	struct ice_vsi *vsi = np->vsi;
6303 	struct ice_pf *pf = vsi->back;
6304 	u16 port = ntohs(ti->port);
6305 	enum ice_status status;
6306 	bool retval;
6307 
6308 	retval = ice_tunnel_port_in_use(&pf->hw, port, NULL);
6309 	if (!retval) {
6310 		netdev_info(netdev, "port %d not found in UDP tunnels list\n",
6311 			    port);
6312 		return;
6313 	}
6314 
6315 	status = ice_destroy_tunnel(&pf->hw, port, false);
6316 	if (status)
6317 		netdev_err(netdev, "error deleting port %d from UDP tunnels list\n",
6318 			   port);
6319 }
6320 
6321 /**
6322  * ice_open - Called when a network interface becomes active
6323  * @netdev: network interface device structure
6324  *
6325  * The open entry point is called when a network interface is made
6326  * active by the system (IFF_UP). At this point all resources needed
6327  * for transmit and receive operations are allocated, the interrupt
6328  * handler is registered with the OS, the netdev watchdog is enabled,
6329  * and the stack is notified that the interface is ready.
6330  *
6331  * Returns 0 on success, negative value on failure
6332  */
6333 int ice_open(struct net_device *netdev)
6334 {
6335 	struct ice_netdev_priv *np = netdev_priv(netdev);
6336 	struct ice_vsi *vsi = np->vsi;
6337 	struct ice_pf *pf = vsi->back;
6338 	struct ice_port_info *pi;
6339 	int err;
6340 
6341 	if (test_bit(__ICE_NEEDS_RESTART, pf->state)) {
6342 		netdev_err(netdev, "driver needs to be unloaded and reloaded\n");
6343 		return -EIO;
6344 	}
6345 
6346 	if (test_bit(__ICE_DOWN, pf->state)) {
6347 		netdev_err(netdev, "device is not ready yet\n");
6348 		return -EBUSY;
6349 	}
6350 
6351 	netif_carrier_off(netdev);
6352 
6353 	pi = vsi->port_info;
6354 	err = ice_update_link_info(pi);
6355 	if (err) {
6356 		netdev_err(netdev, "Failed to get link info, error %d\n",
6357 			   err);
6358 		return err;
6359 	}
6360 
6361 	/* Set PHY if there is media, otherwise, turn off PHY */
6362 	if (pi->phy.link_info.link_info & ICE_AQ_MEDIA_AVAILABLE) {
6363 		clear_bit(ICE_FLAG_NO_MEDIA, pf->flags);
6364 		if (!test_bit(__ICE_PHY_INIT_COMPLETE, pf->state)) {
6365 			err = ice_init_phy_user_cfg(pi);
6366 			if (err) {
6367 				netdev_err(netdev, "Failed to initialize PHY settings, error %d\n",
6368 					   err);
6369 				return err;
6370 			}
6371 		}
6372 
6373 		err = ice_configure_phy(vsi);
6374 		if (err) {
6375 			netdev_err(netdev, "Failed to set physical link up, error %d\n",
6376 				   err);
6377 			return err;
6378 		}
6379 	} else {
6380 		set_bit(ICE_FLAG_NO_MEDIA, pf->flags);
6381 		err = ice_aq_set_link_restart_an(pi, false, NULL);
6382 		if (err) {
6383 			netdev_err(netdev, "Failed to set PHY state, VSI %d error %d\n",
6384 				   vsi->vsi_num, err);
6385 			return err;
6386 		}
6387 	}
6388 
6389 	err = ice_vsi_open(vsi);
6390 	if (err)
6391 		netdev_err(netdev, "Failed to open VSI 0x%04X on switch 0x%04X\n",
6392 			   vsi->vsi_num, vsi->vsw->sw_id);
6393 
6394 	/* Update existing tunnels information */
6395 	udp_tunnel_get_rx_info(netdev);
6396 
6397 	return err;
6398 }
6399 
6400 /**
6401  * ice_stop - Disables a network interface
6402  * @netdev: network interface device structure
6403  *
6404  * The stop entry point is called when an interface is de-activated by the OS,
6405  * and the netdevice enters the DOWN state. The hardware is still under the
6406  * driver's control, but the netdev interface is disabled.
6407  *
6408  * Returns success only - not allowed to fail
6409  */
6410 int ice_stop(struct net_device *netdev)
6411 {
6412 	struct ice_netdev_priv *np = netdev_priv(netdev);
6413 	struct ice_vsi *vsi = np->vsi;
6414 
6415 	ice_vsi_close(vsi);
6416 
6417 	return 0;
6418 }
6419 
6420 /**
6421  * ice_features_check - Validate encapsulated packet conforms to limits
6422  * @skb: skb buffer
6423  * @netdev: This port's netdev
6424  * @features: Offload features that the stack believes apply
6425  */
6426 static netdev_features_t
6427 ice_features_check(struct sk_buff *skb,
6428 		   struct net_device __always_unused *netdev,
6429 		   netdev_features_t features)
6430 {
6431 	size_t len;
6432 
6433 	/* No point in doing any of this if neither checksum nor GSO are
6434 	 * being requested for this frame. We can rule out both by just
6435 	 * checking for CHECKSUM_PARTIAL
6436 	 */
6437 	if (skb->ip_summed != CHECKSUM_PARTIAL)
6438 		return features;
6439 
6440 	/* We cannot support GSO if the MSS is going to be less than
6441 	 * 64 bytes. If it is then we need to drop support for GSO.
6442 	 */
6443 	if (skb_is_gso(skb) && (skb_shinfo(skb)->gso_size < 64))
6444 		features &= ~NETIF_F_GSO_MASK;
6445 
6446 	len = skb_network_header(skb) - skb->data;
6447 	if (len > ICE_TXD_MACLEN_MAX || len & 0x1)
6448 		goto out_rm_features;
6449 
6450 	len = skb_transport_header(skb) - skb_network_header(skb);
6451 	if (len > ICE_TXD_IPLEN_MAX || len & 0x1)
6452 		goto out_rm_features;
6453 
6454 	if (skb->encapsulation) {
6455 		len = skb_inner_network_header(skb) - skb_transport_header(skb);
6456 		if (len > ICE_TXD_L4LEN_MAX || len & 0x1)
6457 			goto out_rm_features;
6458 
6459 		len = skb_inner_transport_header(skb) -
6460 		      skb_inner_network_header(skb);
6461 		if (len > ICE_TXD_IPLEN_MAX || len & 0x1)
6462 			goto out_rm_features;
6463 	}
6464 
6465 	return features;
6466 out_rm_features:
6467 	return features & ~(NETIF_F_CSUM_MASK | NETIF_F_GSO_MASK);
6468 }
6469 
6470 static const struct net_device_ops ice_netdev_safe_mode_ops = {
6471 	.ndo_open = ice_open,
6472 	.ndo_stop = ice_stop,
6473 	.ndo_start_xmit = ice_start_xmit,
6474 	.ndo_set_mac_address = ice_set_mac_address,
6475 	.ndo_validate_addr = eth_validate_addr,
6476 	.ndo_change_mtu = ice_change_mtu,
6477 	.ndo_get_stats64 = ice_get_stats64,
6478 	.ndo_tx_timeout = ice_tx_timeout,
6479 };
6480 
6481 static const struct net_device_ops ice_netdev_ops = {
6482 	.ndo_open = ice_open,
6483 	.ndo_stop = ice_stop,
6484 	.ndo_start_xmit = ice_start_xmit,
6485 	.ndo_features_check = ice_features_check,
6486 	.ndo_set_rx_mode = ice_set_rx_mode,
6487 	.ndo_set_mac_address = ice_set_mac_address,
6488 	.ndo_validate_addr = eth_validate_addr,
6489 	.ndo_change_mtu = ice_change_mtu,
6490 	.ndo_get_stats64 = ice_get_stats64,
6491 	.ndo_set_tx_maxrate = ice_set_tx_maxrate,
6492 	.ndo_set_vf_spoofchk = ice_set_vf_spoofchk,
6493 	.ndo_set_vf_mac = ice_set_vf_mac,
6494 	.ndo_get_vf_config = ice_get_vf_cfg,
6495 	.ndo_set_vf_trust = ice_set_vf_trust,
6496 	.ndo_set_vf_vlan = ice_set_vf_port_vlan,
6497 	.ndo_set_vf_link_state = ice_set_vf_link_state,
6498 	.ndo_get_vf_stats = ice_get_vf_stats,
6499 	.ndo_vlan_rx_add_vid = ice_vlan_rx_add_vid,
6500 	.ndo_vlan_rx_kill_vid = ice_vlan_rx_kill_vid,
6501 	.ndo_set_features = ice_set_features,
6502 	.ndo_bridge_getlink = ice_bridge_getlink,
6503 	.ndo_bridge_setlink = ice_bridge_setlink,
6504 	.ndo_fdb_add = ice_fdb_add,
6505 	.ndo_fdb_del = ice_fdb_del,
6506 #ifdef CONFIG_RFS_ACCEL
6507 	.ndo_rx_flow_steer = ice_rx_flow_steer,
6508 #endif
6509 	.ndo_tx_timeout = ice_tx_timeout,
6510 	.ndo_bpf = ice_xdp,
6511 	.ndo_xdp_xmit = ice_xdp_xmit,
6512 	.ndo_xsk_wakeup = ice_xsk_wakeup,
6513 	.ndo_udp_tunnel_add = ice_udp_tunnel_add,
6514 	.ndo_udp_tunnel_del = ice_udp_tunnel_del,
6515 };
6516