xref: /freebsd/sys/dev/ice/ice_lib.c (revision 2e4311906d8c8dc7a7c726345268253bca6d4acc)
1 /* SPDX-License-Identifier: BSD-3-Clause */
2 /*  Copyright (c) 2021, Intel Corporation
3  *  All rights reserved.
4  *
5  *  Redistribution and use in source and binary forms, with or without
6  *  modification, are permitted provided that the following conditions are met:
7  *
8  *   1. Redistributions of source code must retain the above copyright notice,
9  *      this list of conditions and the following disclaimer.
10  *
11  *   2. Redistributions in binary form must reproduce the above copyright
12  *      notice, this list of conditions and the following disclaimer in the
13  *      documentation and/or other materials provided with the distribution.
14  *
15  *   3. Neither the name of the Intel Corporation nor the names of its
16  *      contributors may be used to endorse or promote products derived from
17  *      this software without specific prior written permission.
18  *
19  *  THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
20  *  AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
21  *  IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
22  *  ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE
23  *  LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
24  *  CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
25  *  SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
26  *  INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
27  *  CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
28  *  ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
29  *  POSSIBILITY OF SUCH DAMAGE.
30  */
31 /*$FreeBSD$*/
32 
33 /**
34  * @file ice_lib.c
35  * @brief Generic device setup and sysctl functions
36  *
37  * Library of generic device functions not specific to the networking stack.
38  *
39  * This includes hardware initialization functions, as well as handlers for
40  * many of the device sysctls used to probe driver status or tune specific
41  * behaviors.
42  */
43 
44 #include "ice_lib.h"
45 #include "ice_iflib.h"
46 #include <dev/pci/pcivar.h>
47 #include <dev/pci/pcireg.h>
48 #include <machine/resource.h>
49 #include <net/if_dl.h>
50 #include <sys/firmware.h>
51 #include <sys/priv.h>
52 #include <sys/limits.h>
53 
54 /**
55  * @var M_ICE
56  * @brief main ice driver allocation type
57  *
58  * malloc(9) allocation type used by the majority of memory allocations in the
59  * ice driver.
60  */
61 MALLOC_DEFINE(M_ICE, "ice", "Intel(R) 100Gb Network Driver lib allocations");
62 
63 /*
64  * Helper function prototypes
65  */
66 static int ice_get_next_vsi(struct ice_vsi **all_vsi, int size);
67 static void ice_set_default_vsi_ctx(struct ice_vsi_ctx *ctx);
68 static void ice_set_rss_vsi_ctx(struct ice_vsi_ctx *ctx, enum ice_vsi_type type);
69 static int ice_setup_vsi_qmap(struct ice_vsi *vsi, struct ice_vsi_ctx *ctx);
70 static int ice_setup_tx_ctx(struct ice_tx_queue *txq,
71 			    struct ice_tlan_ctx *tlan_ctx, u16 pf_q);
72 static int ice_setup_rx_ctx(struct ice_rx_queue *rxq);
73 static int ice_is_rxq_ready(struct ice_hw *hw, int pf_q, u32 *reg);
74 static void ice_free_fltr_list(struct ice_list_head *list);
75 static int ice_add_mac_to_list(struct ice_vsi *vsi, struct ice_list_head *list,
76 			       const u8 *addr, enum ice_sw_fwd_act_type action);
77 static void ice_check_ctrlq_errors(struct ice_softc *sc, const char *qname,
78 				   struct ice_ctl_q_info *cq);
79 static void ice_process_link_event(struct ice_softc *sc, struct ice_rq_event_info *e);
80 static void ice_process_ctrlq_event(struct ice_softc *sc, const char *qname,
81 				    struct ice_rq_event_info *event);
82 static void ice_nvm_version_str(struct ice_hw *hw, struct sbuf *buf);
83 static void ice_active_pkg_version_str(struct ice_hw *hw, struct sbuf *buf);
84 static void ice_os_pkg_version_str(struct ice_hw *hw, struct sbuf *buf);
85 static bool ice_filter_is_mcast(struct ice_vsi *vsi, struct ice_fltr_info *info);
86 static u_int ice_sync_one_mcast_filter(void *p, struct sockaddr_dl *sdl, u_int errors);
87 static void ice_add_debug_tunables(struct ice_softc *sc);
88 static void ice_add_debug_sysctls(struct ice_softc *sc);
89 static void ice_vsi_set_rss_params(struct ice_vsi *vsi);
90 static void ice_get_default_rss_key(u8 *seed);
91 static int  ice_set_rss_key(struct ice_vsi *vsi);
92 static int  ice_set_rss_lut(struct ice_vsi *vsi);
93 static void ice_set_rss_flow_flds(struct ice_vsi *vsi);
94 static void ice_clean_vsi_rss_cfg(struct ice_vsi *vsi);
95 static const char *ice_aq_speed_to_str(struct ice_port_info *pi);
96 static const char *ice_requested_fec_mode(struct ice_port_info *pi);
97 static const char *ice_negotiated_fec_mode(struct ice_port_info *pi);
98 static const char *ice_autoneg_mode(struct ice_port_info *pi);
99 static const char *ice_flowcontrol_mode(struct ice_port_info *pi);
100 static void ice_print_bus_link_data(device_t dev, struct ice_hw *hw);
101 static void ice_set_pci_link_status_data(struct ice_hw *hw, u16 link_status);
102 static uint8_t ice_pcie_bandwidth_check(struct ice_softc *sc);
103 static uint64_t ice_pcie_bus_speed_to_rate(enum ice_pcie_bus_speed speed);
104 static int ice_pcie_lnk_width_to_int(enum ice_pcie_link_width width);
105 static uint64_t ice_phy_types_to_max_rate(struct ice_port_info *pi);
106 static void ice_add_sysctls_sw_stats(struct ice_vsi *vsi,
107 				     struct sysctl_ctx_list *ctx,
108 				     struct sysctl_oid *parent);
109 static void ice_setup_vsi_common(struct ice_softc *sc, struct ice_vsi *vsi,
110 				 enum ice_vsi_type type, int idx,
111 				 bool dynamic);
112 static void ice_handle_mib_change_event(struct ice_softc *sc,
113 				 struct ice_rq_event_info *event);
114 static void
115 ice_handle_lan_overflow_event(struct ice_softc *sc,
116 			      struct ice_rq_event_info *event);
117 static int ice_add_ethertype_to_list(struct ice_vsi *vsi,
118 				     struct ice_list_head *list,
119 				     u16 ethertype, u16 direction,
120 				     enum ice_sw_fwd_act_type action);
121 static void ice_add_rx_lldp_filter(struct ice_softc *sc);
122 static void ice_del_rx_lldp_filter(struct ice_softc *sc);
123 static u16 ice_aq_phy_types_to_link_speeds(u64 phy_type_low,
124 					   u64 phy_type_high);
125 struct ice_phy_data;
126 static int
127 ice_intersect_phy_types_and_speeds(struct ice_softc *sc,
128 				   struct ice_phy_data *phy_data);
129 static int
130 ice_apply_saved_phy_req_to_cfg(struct ice_softc *sc,
131 			       struct ice_aqc_set_phy_cfg_data *cfg);
132 static int
133 ice_apply_saved_fec_req_to_cfg(struct ice_softc *sc,
134 			       struct ice_aqc_set_phy_cfg_data *cfg);
135 static void
136 ice_apply_saved_fc_req_to_cfg(struct ice_port_info *pi,
137 			      struct ice_aqc_set_phy_cfg_data *cfg);
138 static void
139 ice_print_ldo_tlv(struct ice_softc *sc,
140 		  struct ice_link_default_override_tlv *tlv);
141 static void
142 ice_sysctl_speeds_to_aq_phy_types(u16 sysctl_speeds, u64 *phy_type_low,
143 				  u64 *phy_type_high);
144 static u16 ice_apply_supported_speed_filter(u16 report_speeds);
145 static void
146 ice_handle_health_status_event(struct ice_softc *sc,
147 			       struct ice_rq_event_info *event);
148 static void
149 ice_print_health_status_string(device_t dev,
150 			       struct ice_aqc_health_status_elem *elem);
151 
152 static int ice_module_init(void);
153 static int ice_module_exit(void);
154 
155 /*
156  * package version comparison functions
157  */
158 static bool pkg_ver_empty(struct ice_pkg_ver *pkg_ver, u8 *pkg_name);
159 static int pkg_ver_compatible(struct ice_pkg_ver *pkg_ver);
160 
161 /*
162  * dynamic sysctl handlers
163  */
164 static int ice_sysctl_show_fw(SYSCTL_HANDLER_ARGS);
165 static int ice_sysctl_pkg_version(SYSCTL_HANDLER_ARGS);
166 static int ice_sysctl_os_pkg_version(SYSCTL_HANDLER_ARGS);
167 static int ice_sysctl_dump_mac_filters(SYSCTL_HANDLER_ARGS);
168 static int ice_sysctl_dump_vlan_filters(SYSCTL_HANDLER_ARGS);
169 static int ice_sysctl_dump_ethertype_filters(SYSCTL_HANDLER_ARGS);
170 static int ice_sysctl_dump_ethertype_mac_filters(SYSCTL_HANDLER_ARGS);
171 static int ice_sysctl_current_speed(SYSCTL_HANDLER_ARGS);
172 static int ice_sysctl_request_reset(SYSCTL_HANDLER_ARGS);
173 static int ice_sysctl_dump_state_flags(SYSCTL_HANDLER_ARGS);
174 static int ice_sysctl_fec_config(SYSCTL_HANDLER_ARGS);
175 static int ice_sysctl_fc_config(SYSCTL_HANDLER_ARGS);
176 static int ice_sysctl_negotiated_fc(SYSCTL_HANDLER_ARGS);
177 static int ice_sysctl_negotiated_fec(SYSCTL_HANDLER_ARGS);
178 static int ice_sysctl_phy_type_low(SYSCTL_HANDLER_ARGS);
179 static int ice_sysctl_phy_type_high(SYSCTL_HANDLER_ARGS);
180 static int __ice_sysctl_phy_type_handler(SYSCTL_HANDLER_ARGS,
181 					 bool is_phy_type_high);
182 static int ice_sysctl_advertise_speed(SYSCTL_HANDLER_ARGS);
183 static int ice_sysctl_rx_itr(SYSCTL_HANDLER_ARGS);
184 static int ice_sysctl_tx_itr(SYSCTL_HANDLER_ARGS);
185 static int ice_sysctl_fw_lldp_agent(SYSCTL_HANDLER_ARGS);
186 static int ice_sysctl_fw_cur_lldp_persist_status(SYSCTL_HANDLER_ARGS);
187 static int ice_sysctl_fw_dflt_lldp_persist_status(SYSCTL_HANDLER_ARGS);
188 static int ice_sysctl_phy_caps(SYSCTL_HANDLER_ARGS, u8 report_mode);
189 static int ice_sysctl_phy_sw_caps(SYSCTL_HANDLER_ARGS);
190 static int ice_sysctl_phy_nvm_caps(SYSCTL_HANDLER_ARGS);
191 static int ice_sysctl_phy_topo_caps(SYSCTL_HANDLER_ARGS);
192 static int ice_sysctl_phy_link_status(SYSCTL_HANDLER_ARGS);
193 static int ice_sysctl_read_i2c_diag_data(SYSCTL_HANDLER_ARGS);
194 static int ice_sysctl_tx_cso_stat(SYSCTL_HANDLER_ARGS);
195 static int ice_sysctl_rx_cso_stat(SYSCTL_HANDLER_ARGS);
196 static int ice_sysctl_pba_number(SYSCTL_HANDLER_ARGS);
197 static int ice_sysctl_rx_errors_stat(SYSCTL_HANDLER_ARGS);
198 
199 /**
200  * ice_map_bar - Map PCIe BAR memory
201  * @dev: the PCIe device
202  * @bar: the BAR info structure
203  * @bar_num: PCIe BAR number
204  *
205  * Maps the specified PCIe BAR. Stores the mapping data in struct
206  * ice_bar_info.
207  */
208 int
209 ice_map_bar(device_t dev, struct ice_bar_info *bar, int bar_num)
210 {
211 	if (bar->res != NULL) {
212 		device_printf(dev, "PCI BAR%d already mapped\n", bar_num);
213 		return (EDOOFUS);
214 	}
215 
216 	bar->rid = PCIR_BAR(bar_num);
217 	bar->res = bus_alloc_resource_any(dev, SYS_RES_MEMORY, &bar->rid,
218 					  RF_ACTIVE);
219 	if (!bar->res) {
220 		device_printf(dev, "PCI BAR%d mapping failed\n", bar_num);
221 		return (ENXIO);
222 	}
223 
224 	bar->tag = rman_get_bustag(bar->res);
225 	bar->handle = rman_get_bushandle(bar->res);
226 	bar->size = rman_get_size(bar->res);
227 
228 	return (0);
229 }
230 
231 /**
232  * ice_free_bar - Free PCIe BAR memory
233  * @dev: the PCIe device
234  * @bar: the BAR info structure
235  *
236  * Frees the specified PCIe BAR, releasing its resources.
237  */
238 void
239 ice_free_bar(device_t dev, struct ice_bar_info *bar)
240 {
241 	if (bar->res != NULL)
242 		bus_release_resource(dev, SYS_RES_MEMORY, bar->rid, bar->res);
243 	bar->res = NULL;
244 }
245 
246 /**
247  * ice_set_ctrlq_len - Configure ctrlq lengths for a device
248  * @hw: the device hardware structure
249  *
250  * Configures the control queues for the given device, setting up the
251  * specified lengths, prior to initializing hardware.
252  */
253 void
254 ice_set_ctrlq_len(struct ice_hw *hw)
255 {
256 	hw->adminq.num_rq_entries = ICE_AQ_LEN;
257 	hw->adminq.num_sq_entries = ICE_AQ_LEN;
258 	hw->adminq.rq_buf_size = ICE_AQ_MAX_BUF_LEN;
259 	hw->adminq.sq_buf_size = ICE_AQ_MAX_BUF_LEN;
260 
261 	hw->mailboxq.num_rq_entries = ICE_MBXQ_LEN;
262 	hw->mailboxq.num_sq_entries = ICE_MBXQ_LEN;
263 	hw->mailboxq.rq_buf_size = ICE_MBXQ_MAX_BUF_LEN;
264 	hw->mailboxq.sq_buf_size = ICE_MBXQ_MAX_BUF_LEN;
265 
266 }
267 
268 /**
269  * ice_get_next_vsi - Get the next available VSI slot
270  * @all_vsi: the VSI list
271  * @size: the size of the VSI list
272  *
273  * Returns the index to the first available VSI slot. Will return size (one
274  * past the last index) if there are no slots available.
275  */
276 static int
277 ice_get_next_vsi(struct ice_vsi **all_vsi, int size)
278 {
279 	int i;
280 
281 	for (i = 0; i < size; i++) {
282 		if (all_vsi[i] == NULL)
283 			return i;
284 	}
285 
286 	return size;
287 }
288 
289 /**
290  * ice_setup_vsi_common - Common VSI setup for both dynamic and static VSIs
291  * @sc: the device private softc structure
292  * @vsi: the VSI to setup
293  * @type: the VSI type of the new VSI
294  * @idx: the index in the all_vsi array to use
295  * @dynamic: whether this VSI memory was dynamically allocated
296  *
297  * Perform setup for a VSI that is common to both dynamically allocated VSIs
298  * and the static PF VSI which is embedded in the softc structure.
299  */
300 static void
301 ice_setup_vsi_common(struct ice_softc *sc, struct ice_vsi *vsi,
302 		     enum ice_vsi_type type, int idx, bool dynamic)
303 {
304 	/* Store important values in VSI struct */
305 	vsi->type = type;
306 	vsi->sc = sc;
307 	vsi->idx = idx;
308 	sc->all_vsi[idx] = vsi;
309 	vsi->dynamic = dynamic;
310 
311 	/* Setup the VSI tunables now */
312 	ice_add_vsi_tunables(vsi, sc->vsi_sysctls);
313 }
314 
315 /**
316  * ice_alloc_vsi - Allocate a dynamic VSI
317  * @sc: device softc structure
318  * @type: VSI type
319  *
320  * Allocates a new dynamic VSI structure and inserts it into the VSI list.
321  */
322 struct ice_vsi *
323 ice_alloc_vsi(struct ice_softc *sc, enum ice_vsi_type type)
324 {
325 	struct ice_vsi *vsi;
326 	int idx;
327 
328 	/* Find an open index for a new VSI to be allocated. If the returned
329 	 * index is >= the num_available_vsi then it means no slot is
330 	 * available.
331 	 */
332 	idx = ice_get_next_vsi(sc->all_vsi, sc->num_available_vsi);
333 	if (idx >= sc->num_available_vsi) {
334 		device_printf(sc->dev, "No available VSI slots\n");
335 		return NULL;
336 	}
337 
338 	vsi = (struct ice_vsi *)malloc(sizeof(*vsi), M_ICE, M_WAITOK|M_ZERO);
339 	if (!vsi) {
340 		device_printf(sc->dev, "Unable to allocate VSI memory\n");
341 		return NULL;
342 	}
343 
344 	ice_setup_vsi_common(sc, vsi, type, idx, true);
345 
346 	return vsi;
347 }
348 
349 /**
350  * ice_setup_pf_vsi - Setup the PF VSI
351  * @sc: the device private softc
352  *
353  * Setup the PF VSI structure which is embedded as sc->pf_vsi in the device
354  * private softc. Unlike other VSIs, the PF VSI memory is allocated as part of
355  * the softc memory, instead of being dynamically allocated at creation.
356  */
357 void
358 ice_setup_pf_vsi(struct ice_softc *sc)
359 {
360 	ice_setup_vsi_common(sc, &sc->pf_vsi, ICE_VSI_PF, 0, false);
361 }
362 
363 /**
364  * ice_alloc_vsi_qmap
365  * @vsi: VSI structure
366  * @max_tx_queues: Number of transmit queues to identify
367  * @max_rx_queues: Number of receive queues to identify
368  *
369  * Allocates a max_[t|r]x_queues array of words for the VSI where each
370  * word contains the index of the queue it represents.  In here, all
371  * words are initialized to an index of ICE_INVALID_RES_IDX, indicating
372  * all queues for this VSI are not yet assigned an index and thus,
373  * not ready for use.
374  *
375  * Returns an error code on failure.
376  */
377 int
378 ice_alloc_vsi_qmap(struct ice_vsi *vsi, const int max_tx_queues,
379 		   const int max_rx_queues)
380 {
381 	struct ice_softc *sc = vsi->sc;
382 	int i;
383 
384 	MPASS(max_tx_queues > 0);
385 	MPASS(max_rx_queues > 0);
386 
387 	/* Allocate Tx queue mapping memory */
388 	if (!(vsi->tx_qmap =
389 	      (u16 *) malloc(sizeof(u16) * max_tx_queues, M_ICE, M_WAITOK))) {
390 		device_printf(sc->dev, "Unable to allocate Tx qmap memory\n");
391 		return (ENOMEM);
392 	}
393 
394 	/* Allocate Rx queue mapping memory */
395 	if (!(vsi->rx_qmap =
396 	      (u16 *) malloc(sizeof(u16) * max_rx_queues, M_ICE, M_WAITOK))) {
397 		device_printf(sc->dev, "Unable to allocate Rx qmap memory\n");
398 		goto free_tx_qmap;
399 	}
400 
401 	/* Mark every queue map as invalid to start with */
402 	for (i = 0; i < max_tx_queues; i++) {
403 		vsi->tx_qmap[i] = ICE_INVALID_RES_IDX;
404 	}
405 	for (i = 0; i < max_rx_queues; i++) {
406 		vsi->rx_qmap[i] = ICE_INVALID_RES_IDX;
407 	}
408 
409 	return 0;
410 
411 free_tx_qmap:
412 	free(vsi->tx_qmap, M_ICE);
413 	vsi->tx_qmap = NULL;
414 
415 	return (ENOMEM);
416 }
417 
418 /**
419  * ice_free_vsi_qmaps - Free the PF qmaps associated with a VSI
420  * @vsi: the VSI private structure
421  *
422  * Frees the PF qmaps associated with the given VSI. Generally this will be
423  * called by ice_release_vsi, but may need to be called during attach cleanup,
424  * depending on when the qmaps were allocated.
425  */
426 void
427 ice_free_vsi_qmaps(struct ice_vsi *vsi)
428 {
429 	struct ice_softc *sc = vsi->sc;
430 
431 	if (vsi->tx_qmap) {
432 		ice_resmgr_release_map(&sc->tx_qmgr, vsi->tx_qmap,
433 					   vsi->num_tx_queues);
434 		free(vsi->tx_qmap, M_ICE);
435 		vsi->tx_qmap = NULL;
436 	}
437 
438 	if (vsi->rx_qmap) {
439 		ice_resmgr_release_map(&sc->rx_qmgr, vsi->rx_qmap,
440 					   vsi->num_rx_queues);
441 		free(vsi->rx_qmap, M_ICE);
442 		vsi->rx_qmap = NULL;
443 	}
444 }
445 
446 /**
447  * ice_set_default_vsi_ctx - Setup default VSI context parameters
448  * @ctx: the VSI context to initialize
449  *
450  * Initialize and prepare a default VSI context for configuring a new VSI.
451  */
452 static void
453 ice_set_default_vsi_ctx(struct ice_vsi_ctx *ctx)
454 {
455 	u32 table = 0;
456 
457 	memset(&ctx->info, 0, sizeof(ctx->info));
458 	/* VSI will be allocated from shared pool */
459 	ctx->alloc_from_pool = true;
460 	/* Enable source pruning by default */
461 	ctx->info.sw_flags = ICE_AQ_VSI_SW_FLAG_SRC_PRUNE;
462 	/* Traffic from VSI can be sent to LAN */
463 	ctx->info.sw_flags2 = ICE_AQ_VSI_SW_FLAG_LAN_ENA;
464 	/* Allow all packets untagged/tagged */
465 	ctx->info.inner_vlan_flags = ((ICE_AQ_VSI_INNER_VLAN_TX_MODE_ALL &
466 				       ICE_AQ_VSI_INNER_VLAN_TX_MODE_M) >>
467 				       ICE_AQ_VSI_INNER_VLAN_TX_MODE_S);
468 	/* Show VLAN/UP from packets in Rx descriptors */
469 	ctx->info.inner_vlan_flags |= ((ICE_AQ_VSI_INNER_VLAN_EMODE_STR_BOTH &
470 					ICE_AQ_VSI_INNER_VLAN_EMODE_M) >>
471 					ICE_AQ_VSI_INNER_VLAN_EMODE_S);
472 	/* Have 1:1 UP mapping for both ingress/egress tables */
473 	table |= ICE_UP_TABLE_TRANSLATE(0, 0);
474 	table |= ICE_UP_TABLE_TRANSLATE(1, 1);
475 	table |= ICE_UP_TABLE_TRANSLATE(2, 2);
476 	table |= ICE_UP_TABLE_TRANSLATE(3, 3);
477 	table |= ICE_UP_TABLE_TRANSLATE(4, 4);
478 	table |= ICE_UP_TABLE_TRANSLATE(5, 5);
479 	table |= ICE_UP_TABLE_TRANSLATE(6, 6);
480 	table |= ICE_UP_TABLE_TRANSLATE(7, 7);
481 	ctx->info.ingress_table = CPU_TO_LE32(table);
482 	ctx->info.egress_table = CPU_TO_LE32(table);
483 	/* Have 1:1 UP mapping for outer to inner UP table */
484 	ctx->info.outer_up_table = CPU_TO_LE32(table);
485 	/* No Outer tag support, so outer_vlan_flags remains zero */
486 }
487 
488 /**
489  * ice_set_rss_vsi_ctx - Setup VSI context parameters for RSS
490  * @ctx: the VSI context to configure
491  * @type: the VSI type
492  *
493  * Configures the VSI context for RSS, based on the VSI type.
494  */
495 static void
496 ice_set_rss_vsi_ctx(struct ice_vsi_ctx *ctx, enum ice_vsi_type type)
497 {
498 	u8 lut_type, hash_type;
499 
500 	switch (type) {
501 	case ICE_VSI_PF:
502 		lut_type = ICE_AQ_VSI_Q_OPT_RSS_LUT_PF;
503 		hash_type = ICE_AQ_VSI_Q_OPT_RSS_TPLZ;
504 		break;
505 	case ICE_VSI_VF:
506 		lut_type = ICE_AQ_VSI_Q_OPT_RSS_LUT_VSI;
507 		hash_type = ICE_AQ_VSI_Q_OPT_RSS_TPLZ;
508 		break;
509 	default:
510 		/* Other VSI types do not support RSS */
511 		return;
512 	}
513 
514 	ctx->info.q_opt_rss = (((lut_type << ICE_AQ_VSI_Q_OPT_RSS_LUT_S) &
515 				 ICE_AQ_VSI_Q_OPT_RSS_LUT_M) |
516 				((hash_type << ICE_AQ_VSI_Q_OPT_RSS_HASH_S) &
517 				 ICE_AQ_VSI_Q_OPT_RSS_HASH_M));
518 }
519 
520 /**
521  * ice_setup_vsi_qmap - Setup the queue mapping for a VSI
522  * @vsi: the VSI to configure
523  * @ctx: the VSI context to configure
524  *
525  * Configures the context for the given VSI, setting up how the firmware
526  * should map the queues for this VSI.
527  */
528 static int
529 ice_setup_vsi_qmap(struct ice_vsi *vsi, struct ice_vsi_ctx *ctx)
530 {
531 	int pow = 0;
532 	u16 qmap;
533 
534 	MPASS(vsi->rx_qmap != NULL);
535 
536 	/* TODO:
537 	 * Handle multiple Traffic Classes
538 	 * Handle scattered queues (for VFs)
539 	 */
540 	if (vsi->qmap_type != ICE_RESMGR_ALLOC_CONTIGUOUS)
541 		return (EOPNOTSUPP);
542 
543 	ctx->info.mapping_flags |= CPU_TO_LE16(ICE_AQ_VSI_Q_MAP_CONTIG);
544 
545 	ctx->info.q_mapping[0] = CPU_TO_LE16(vsi->rx_qmap[0]);
546 	ctx->info.q_mapping[1] = CPU_TO_LE16(vsi->num_rx_queues);
547 
548 
549 	/* Calculate the next power-of-2 of number of queues */
550 	if (vsi->num_rx_queues)
551 		pow = flsl(vsi->num_rx_queues - 1);
552 
553 	/* Assign all the queues to traffic class zero */
554 	qmap = (pow << ICE_AQ_VSI_TC_Q_NUM_S) & ICE_AQ_VSI_TC_Q_NUM_M;
555 	ctx->info.tc_mapping[0] = CPU_TO_LE16(qmap);
556 
557 	return 0;
558 }
559 
560 /**
561  * ice_initialize_vsi - Initialize a VSI for use
562  * @vsi: the vsi to initialize
563  *
564  * Initialize a VSI over the adminq and prepare it for operation.
565  */
566 int
567 ice_initialize_vsi(struct ice_vsi *vsi)
568 {
569 	struct ice_vsi_ctx ctx = { 0 };
570 	struct ice_hw *hw = &vsi->sc->hw;
571 	u16 max_txqs[ICE_MAX_TRAFFIC_CLASS] = { 0 };
572 	enum ice_status status;
573 	int err;
574 
575 	/* For now, we only have code supporting PF VSIs */
576 	switch (vsi->type) {
577 	case ICE_VSI_PF:
578 		ctx.flags = ICE_AQ_VSI_TYPE_PF;
579 		break;
580 	default:
581 		return (ENODEV);
582 	}
583 
584 	ice_set_default_vsi_ctx(&ctx);
585 	ice_set_rss_vsi_ctx(&ctx, vsi->type);
586 
587 	/* XXX: VSIs of other types may need different port info? */
588 	ctx.info.sw_id = hw->port_info->sw_id;
589 
590 	/* Set some RSS parameters based on the VSI type */
591 	ice_vsi_set_rss_params(vsi);
592 
593 	/* Initialize the Rx queue mapping for this VSI */
594 	err = ice_setup_vsi_qmap(vsi, &ctx);
595 	if (err) {
596 		return err;
597 	}
598 
599 	/* (Re-)add VSI to HW VSI handle list */
600 	status = ice_add_vsi(hw, vsi->idx, &ctx, NULL);
601 	if (status != 0) {
602 		device_printf(vsi->sc->dev,
603 		    "Add VSI AQ call failed, err %s aq_err %s\n",
604 		    ice_status_str(status),
605 		    ice_aq_str(hw->adminq.sq_last_status));
606 		return (EIO);
607 	}
608 	vsi->info = ctx.info;
609 
610 	/* TODO: DCB traffic class support? */
611 	max_txqs[0] = vsi->num_tx_queues;
612 
613 	status = ice_cfg_vsi_lan(hw->port_info, vsi->idx,
614 			      ICE_DFLT_TRAFFIC_CLASS, max_txqs);
615 	if (status) {
616 		device_printf(vsi->sc->dev,
617 		    "Failed VSI lan queue config, err %s aq_err %s\n",
618 		    ice_status_str(status),
619 		    ice_aq_str(hw->adminq.sq_last_status));
620 		ice_deinit_vsi(vsi);
621 		return (ENODEV);
622 	}
623 
624 	/* Reset VSI stats */
625 	ice_reset_vsi_stats(vsi);
626 
627 	return 0;
628 }
629 
630 /**
631  * ice_deinit_vsi - Tell firmware to release resources for a VSI
632  * @vsi: the VSI to release
633  *
634  * Helper function which requests the firmware to release the hardware
635  * resources associated with a given VSI.
636  */
637 void
638 ice_deinit_vsi(struct ice_vsi *vsi)
639 {
640 	struct ice_vsi_ctx ctx = { 0 };
641 	struct ice_softc *sc = vsi->sc;
642 	struct ice_hw *hw = &sc->hw;
643 	enum ice_status status;
644 
645 	/* Assert that the VSI pointer matches in the list */
646 	MPASS(vsi == sc->all_vsi[vsi->idx]);
647 
648 	ctx.info = vsi->info;
649 
650 	status = ice_rm_vsi_lan_cfg(hw->port_info, vsi->idx);
651 	if (status) {
652 		/*
653 		 * This should only fail if the VSI handle is invalid, or if
654 		 * any of the nodes have leaf nodes which are still in use.
655 		 */
656 		device_printf(sc->dev,
657 			      "Unable to remove scheduler nodes for VSI %d, err %s\n",
658 			      vsi->idx, ice_status_str(status));
659 	}
660 
661 	/* Tell firmware to release the VSI resources */
662 	status = ice_free_vsi(hw, vsi->idx, &ctx, false, NULL);
663 	if (status != 0) {
664 		device_printf(sc->dev,
665 		    "Free VSI %u AQ call failed, err %s aq_err %s\n",
666 		    vsi->idx, ice_status_str(status),
667 		    ice_aq_str(hw->adminq.sq_last_status));
668 	}
669 }
670 
671 /**
672  * ice_release_vsi - Release resources associated with a VSI
673  * @vsi: the VSI to release
674  *
675  * Release software and firmware resources associated with a VSI. Release the
676  * queue managers associated with this VSI. Also free the VSI structure memory
677  * if the VSI was allocated dynamically using ice_alloc_vsi().
678  */
679 void
680 ice_release_vsi(struct ice_vsi *vsi)
681 {
682 	struct ice_softc *sc = vsi->sc;
683 	int idx = vsi->idx;
684 
685 	/* Assert that the VSI pointer matches in the list */
686 	MPASS(vsi == sc->all_vsi[idx]);
687 
688 	/* Cleanup RSS configuration */
689 	if (ice_is_bit_set(sc->feat_en, ICE_FEATURE_RSS))
690 		ice_clean_vsi_rss_cfg(vsi);
691 
692 	ice_del_vsi_sysctl_ctx(vsi);
693 
694 	ice_deinit_vsi(vsi);
695 
696 	ice_free_vsi_qmaps(vsi);
697 
698 	if (vsi->dynamic) {
699 		free(sc->all_vsi[idx], M_ICE);
700 	}
701 
702 	sc->all_vsi[idx] = NULL;
703 }
704 
705 /**
706  * ice_aq_speed_to_rate - Convert AdminQ speed enum to baudrate
707  * @pi: port info data
708  *
709  * Returns the baudrate value for the current link speed of a given port.
710  */
711 uint64_t
712 ice_aq_speed_to_rate(struct ice_port_info *pi)
713 {
714 	switch (pi->phy.link_info.link_speed) {
715 	case ICE_AQ_LINK_SPEED_100GB:
716 		return IF_Gbps(100);
717 	case ICE_AQ_LINK_SPEED_50GB:
718 		return IF_Gbps(50);
719 	case ICE_AQ_LINK_SPEED_40GB:
720 		return IF_Gbps(40);
721 	case ICE_AQ_LINK_SPEED_25GB:
722 		return IF_Gbps(25);
723 	case ICE_AQ_LINK_SPEED_10GB:
724 		return IF_Gbps(10);
725 	case ICE_AQ_LINK_SPEED_5GB:
726 		return IF_Gbps(5);
727 	case ICE_AQ_LINK_SPEED_2500MB:
728 		return IF_Mbps(2500);
729 	case ICE_AQ_LINK_SPEED_1000MB:
730 		return IF_Mbps(1000);
731 	case ICE_AQ_LINK_SPEED_100MB:
732 		return IF_Mbps(100);
733 	case ICE_AQ_LINK_SPEED_10MB:
734 		return IF_Mbps(10);
735 	case ICE_AQ_LINK_SPEED_UNKNOWN:
736 	default:
737 		/* return 0 if we don't know the link speed */
738 		return 0;
739 	}
740 }
741 
742 /**
743  * ice_aq_speed_to_str - Convert AdminQ speed enum to string representation
744  * @pi: port info data
745  *
746  * Returns the string representation of the current link speed for a given
747  * port.
748  */
749 static const char *
750 ice_aq_speed_to_str(struct ice_port_info *pi)
751 {
752 	switch (pi->phy.link_info.link_speed) {
753 	case ICE_AQ_LINK_SPEED_100GB:
754 		return "100 Gbps";
755 	case ICE_AQ_LINK_SPEED_50GB:
756 		return "50 Gbps";
757 	case ICE_AQ_LINK_SPEED_40GB:
758 		return "40 Gbps";
759 	case ICE_AQ_LINK_SPEED_25GB:
760 		return "25 Gbps";
761 	case ICE_AQ_LINK_SPEED_20GB:
762 		return "20 Gbps";
763 	case ICE_AQ_LINK_SPEED_10GB:
764 		return "10 Gbps";
765 	case ICE_AQ_LINK_SPEED_5GB:
766 		return "5 Gbps";
767 	case ICE_AQ_LINK_SPEED_2500MB:
768 		return "2.5 Gbps";
769 	case ICE_AQ_LINK_SPEED_1000MB:
770 		return "1 Gbps";
771 	case ICE_AQ_LINK_SPEED_100MB:
772 		return "100 Mbps";
773 	case ICE_AQ_LINK_SPEED_10MB:
774 		return "10 Mbps";
775 	case ICE_AQ_LINK_SPEED_UNKNOWN:
776 	default:
777 		return "Unknown speed";
778 	}
779 }
780 
781 /**
782  * ice_get_phy_type_low - Get media associated with phy_type_low
783  * @phy_type_low: the low 64bits of phy_type from the AdminQ
784  *
785  * Given the lower 64bits of the phy_type from the hardware, return the
786  * ifm_active bit associated. Return IFM_UNKNOWN when phy_type_low is unknown.
787  * Note that only one of ice_get_phy_type_low or ice_get_phy_type_high should
788  * be called. If phy_type_low is zero, call ice_phy_type_high.
789  */
790 int
791 ice_get_phy_type_low(uint64_t phy_type_low)
792 {
793 	switch (phy_type_low) {
794 	case ICE_PHY_TYPE_LOW_100BASE_TX:
795 		return IFM_100_TX;
796 	case ICE_PHY_TYPE_LOW_100M_SGMII:
797 		return IFM_100_SGMII;
798 	case ICE_PHY_TYPE_LOW_1000BASE_T:
799 		return IFM_1000_T;
800 	case ICE_PHY_TYPE_LOW_1000BASE_SX:
801 		return IFM_1000_SX;
802 	case ICE_PHY_TYPE_LOW_1000BASE_LX:
803 		return IFM_1000_LX;
804 	case ICE_PHY_TYPE_LOW_1000BASE_KX:
805 		return IFM_1000_KX;
806 	case ICE_PHY_TYPE_LOW_1G_SGMII:
807 		return IFM_1000_SGMII;
808 	case ICE_PHY_TYPE_LOW_2500BASE_T:
809 		return IFM_2500_T;
810 	case ICE_PHY_TYPE_LOW_2500BASE_X:
811 		return IFM_2500_X;
812 	case ICE_PHY_TYPE_LOW_2500BASE_KX:
813 		return IFM_2500_KX;
814 	case ICE_PHY_TYPE_LOW_5GBASE_T:
815 		return IFM_5000_T;
816 	case ICE_PHY_TYPE_LOW_5GBASE_KR:
817 		return IFM_5000_KR;
818 	case ICE_PHY_TYPE_LOW_10GBASE_T:
819 		return IFM_10G_T;
820 	case ICE_PHY_TYPE_LOW_10G_SFI_DA:
821 		return IFM_10G_TWINAX;
822 	case ICE_PHY_TYPE_LOW_10GBASE_SR:
823 		return IFM_10G_SR;
824 	case ICE_PHY_TYPE_LOW_10GBASE_LR:
825 		return IFM_10G_LR;
826 	case ICE_PHY_TYPE_LOW_10GBASE_KR_CR1:
827 		return IFM_10G_KR;
828 	case ICE_PHY_TYPE_LOW_10G_SFI_AOC_ACC:
829 		return IFM_10G_AOC;
830 	case ICE_PHY_TYPE_LOW_10G_SFI_C2C:
831 		return IFM_10G_SFI;
832 	case ICE_PHY_TYPE_LOW_25GBASE_T:
833 		return IFM_25G_T;
834 	case ICE_PHY_TYPE_LOW_25GBASE_CR:
835 		return IFM_25G_CR;
836 	case ICE_PHY_TYPE_LOW_25GBASE_CR_S:
837 		return IFM_25G_CR_S;
838 	case ICE_PHY_TYPE_LOW_25GBASE_CR1:
839 		return IFM_25G_CR1;
840 	case ICE_PHY_TYPE_LOW_25GBASE_SR:
841 		return IFM_25G_SR;
842 	case ICE_PHY_TYPE_LOW_25GBASE_LR:
843 		return IFM_25G_LR;
844 	case ICE_PHY_TYPE_LOW_25GBASE_KR:
845 		return IFM_25G_KR;
846 	case ICE_PHY_TYPE_LOW_25GBASE_KR_S:
847 		return IFM_25G_KR_S;
848 	case ICE_PHY_TYPE_LOW_25GBASE_KR1:
849 		return IFM_25G_KR1;
850 	case ICE_PHY_TYPE_LOW_25G_AUI_AOC_ACC:
851 		return IFM_25G_AOC;
852 	case ICE_PHY_TYPE_LOW_25G_AUI_C2C:
853 		return IFM_25G_AUI;
854 	case ICE_PHY_TYPE_LOW_40GBASE_CR4:
855 		return IFM_40G_CR4;
856 	case ICE_PHY_TYPE_LOW_40GBASE_SR4:
857 		return IFM_40G_SR4;
858 	case ICE_PHY_TYPE_LOW_40GBASE_LR4:
859 		return IFM_40G_LR4;
860 	case ICE_PHY_TYPE_LOW_40GBASE_KR4:
861 		return IFM_40G_KR4;
862 	case ICE_PHY_TYPE_LOW_40G_XLAUI_AOC_ACC:
863 		return IFM_40G_XLAUI_AC;
864 	case ICE_PHY_TYPE_LOW_40G_XLAUI:
865 		return IFM_40G_XLAUI;
866 	case ICE_PHY_TYPE_LOW_50GBASE_CR2:
867 		return IFM_50G_CR2;
868 	case ICE_PHY_TYPE_LOW_50GBASE_SR2:
869 		return IFM_50G_SR2;
870 	case ICE_PHY_TYPE_LOW_50GBASE_LR2:
871 		return IFM_50G_LR2;
872 	case ICE_PHY_TYPE_LOW_50GBASE_KR2:
873 		return IFM_50G_KR2;
874 	case ICE_PHY_TYPE_LOW_50G_LAUI2_AOC_ACC:
875 		return IFM_50G_LAUI2_AC;
876 	case ICE_PHY_TYPE_LOW_50G_LAUI2:
877 		return IFM_50G_LAUI2;
878 	case ICE_PHY_TYPE_LOW_50G_AUI2_AOC_ACC:
879 		return IFM_50G_AUI2_AC;
880 	case ICE_PHY_TYPE_LOW_50G_AUI2:
881 		return IFM_50G_AUI2;
882 	case ICE_PHY_TYPE_LOW_50GBASE_CP:
883 		return IFM_50G_CP;
884 	case ICE_PHY_TYPE_LOW_50GBASE_SR:
885 		return IFM_50G_SR;
886 	case ICE_PHY_TYPE_LOW_50GBASE_FR:
887 		return IFM_50G_FR;
888 	case ICE_PHY_TYPE_LOW_50GBASE_LR:
889 		return IFM_50G_LR;
890 	case ICE_PHY_TYPE_LOW_50GBASE_KR_PAM4:
891 		return IFM_50G_KR_PAM4;
892 	case ICE_PHY_TYPE_LOW_50G_AUI1_AOC_ACC:
893 		return IFM_50G_AUI1_AC;
894 	case ICE_PHY_TYPE_LOW_50G_AUI1:
895 		return IFM_50G_AUI1;
896 	case ICE_PHY_TYPE_LOW_100GBASE_CR4:
897 		return IFM_100G_CR4;
898 	case ICE_PHY_TYPE_LOW_100GBASE_SR4:
899 		return IFM_100G_SR4;
900 	case ICE_PHY_TYPE_LOW_100GBASE_LR4:
901 		return IFM_100G_LR4;
902 	case ICE_PHY_TYPE_LOW_100GBASE_KR4:
903 		return IFM_100G_KR4;
904 	case ICE_PHY_TYPE_LOW_100G_CAUI4_AOC_ACC:
905 		return IFM_100G_CAUI4_AC;
906 	case ICE_PHY_TYPE_LOW_100G_CAUI4:
907 		return IFM_100G_CAUI4;
908 	case ICE_PHY_TYPE_LOW_100G_AUI4_AOC_ACC:
909 		return IFM_100G_AUI4_AC;
910 	case ICE_PHY_TYPE_LOW_100G_AUI4:
911 		return IFM_100G_AUI4;
912 	case ICE_PHY_TYPE_LOW_100GBASE_CR_PAM4:
913 		return IFM_100G_CR_PAM4;
914 	case ICE_PHY_TYPE_LOW_100GBASE_KR_PAM4:
915 		return IFM_100G_KR_PAM4;
916 	case ICE_PHY_TYPE_LOW_100GBASE_CP2:
917 		return IFM_100G_CP2;
918 	case ICE_PHY_TYPE_LOW_100GBASE_SR2:
919 		return IFM_100G_SR2;
920 	case ICE_PHY_TYPE_LOW_100GBASE_DR:
921 		return IFM_100G_DR;
922 	default:
923 		return IFM_UNKNOWN;
924 	}
925 }
926 
927 /**
928  * ice_get_phy_type_high - Get media associated with phy_type_high
929  * @phy_type_high: the upper 64bits of phy_type from the AdminQ
930  *
931  * Given the upper 64bits of the phy_type from the hardware, return the
932  * ifm_active bit associated. Return IFM_UNKNOWN on an unknown value. Note
933  * that only one of ice_get_phy_type_low or ice_get_phy_type_high should be
934  * called. If phy_type_high is zero, call ice_get_phy_type_low.
935  */
936 int
937 ice_get_phy_type_high(uint64_t phy_type_high)
938 {
939 	switch (phy_type_high) {
940 	case ICE_PHY_TYPE_HIGH_100GBASE_KR2_PAM4:
941 		return IFM_100G_KR2_PAM4;
942 	case ICE_PHY_TYPE_HIGH_100G_CAUI2_AOC_ACC:
943 		return IFM_100G_CAUI2_AC;
944 	case ICE_PHY_TYPE_HIGH_100G_CAUI2:
945 		return IFM_100G_CAUI2;
946 	case ICE_PHY_TYPE_HIGH_100G_AUI2_AOC_ACC:
947 		return IFM_100G_AUI2_AC;
948 	case ICE_PHY_TYPE_HIGH_100G_AUI2:
949 		return IFM_100G_AUI2;
950 	default:
951 		return IFM_UNKNOWN;
952 	}
953 }
954 
955 /**
956  * ice_phy_types_to_max_rate - Returns port's max supported baudrate
957  * @pi: port info struct
958  *
959  * ice_aq_get_phy_caps() w/ ICE_AQC_REPORT_TOPO_CAP_MEDIA parameter needs
960  * to have been called before this function for it to work.
961  */
962 static uint64_t
963 ice_phy_types_to_max_rate(struct ice_port_info *pi)
964 {
965 	uint64_t phy_low = pi->phy.phy_type_low;
966 	uint64_t phy_high = pi->phy.phy_type_high;
967 	uint64_t max_rate = 0;
968 	int bit;
969 
970 	/*
971 	 * These are based on the indices used in the BIT() macros for
972 	 * ICE_PHY_TYPE_LOW_*
973 	 */
974 	static const uint64_t phy_rates[] = {
975 	    IF_Mbps(100),
976 	    IF_Mbps(100),
977 	    IF_Gbps(1ULL),
978 	    IF_Gbps(1ULL),
979 	    IF_Gbps(1ULL),
980 	    IF_Gbps(1ULL),
981 	    IF_Gbps(1ULL),
982 	    IF_Mbps(2500ULL),
983 	    IF_Mbps(2500ULL),
984 	    IF_Mbps(2500ULL),
985 	    IF_Gbps(5ULL),
986 	    IF_Gbps(5ULL),
987 	    IF_Gbps(10ULL),
988 	    IF_Gbps(10ULL),
989 	    IF_Gbps(10ULL),
990 	    IF_Gbps(10ULL),
991 	    IF_Gbps(10ULL),
992 	    IF_Gbps(10ULL),
993 	    IF_Gbps(10ULL),
994 	    IF_Gbps(25ULL),
995 	    IF_Gbps(25ULL),
996 	    IF_Gbps(25ULL),
997 	    IF_Gbps(25ULL),
998 	    IF_Gbps(25ULL),
999 	    IF_Gbps(25ULL),
1000 	    IF_Gbps(25ULL),
1001 	    IF_Gbps(25ULL),
1002 	    IF_Gbps(25ULL),
1003 	    IF_Gbps(25ULL),
1004 	    IF_Gbps(25ULL),
1005 	    IF_Gbps(40ULL),
1006 	    IF_Gbps(40ULL),
1007 	    IF_Gbps(40ULL),
1008 	    IF_Gbps(40ULL),
1009 	    IF_Gbps(40ULL),
1010 	    IF_Gbps(40ULL),
1011 	    IF_Gbps(50ULL),
1012 	    IF_Gbps(50ULL),
1013 	    IF_Gbps(50ULL),
1014 	    IF_Gbps(50ULL),
1015 	    IF_Gbps(50ULL),
1016 	    IF_Gbps(50ULL),
1017 	    IF_Gbps(50ULL),
1018 	    IF_Gbps(50ULL),
1019 	    IF_Gbps(50ULL),
1020 	    IF_Gbps(50ULL),
1021 	    IF_Gbps(50ULL),
1022 	    IF_Gbps(50ULL),
1023 	    IF_Gbps(50ULL),
1024 	    IF_Gbps(50ULL),
1025 	    IF_Gbps(50ULL),
1026 	    IF_Gbps(100ULL),
1027 	    IF_Gbps(100ULL),
1028 	    IF_Gbps(100ULL),
1029 	    IF_Gbps(100ULL),
1030 	    IF_Gbps(100ULL),
1031 	    IF_Gbps(100ULL),
1032 	    IF_Gbps(100ULL),
1033 	    IF_Gbps(100ULL),
1034 	    IF_Gbps(100ULL),
1035 	    IF_Gbps(100ULL),
1036 	    IF_Gbps(100ULL),
1037 	    IF_Gbps(100ULL),
1038 	    IF_Gbps(100ULL),
1039 	    /* These rates are for ICE_PHY_TYPE_HIGH_* */
1040 	    IF_Gbps(100ULL),
1041 	    IF_Gbps(100ULL),
1042 	    IF_Gbps(100ULL),
1043 	    IF_Gbps(100ULL),
1044 	    IF_Gbps(100ULL)
1045 	};
1046 
1047 	/* coverity[address_of] */
1048 	for_each_set_bit(bit, &phy_high, 64)
1049 		if ((bit + 64) < (int)ARRAY_SIZE(phy_rates))
1050 			max_rate = uqmax(max_rate, phy_rates[(bit + 64)]);
1051 
1052 	/* coverity[address_of] */
1053 	for_each_set_bit(bit, &phy_low, 64)
1054 		max_rate = uqmax(max_rate, phy_rates[bit]);
1055 
1056 	return (max_rate);
1057 }
1058 
1059 /* The if_media type is split over the original 5 bit media variant field,
1060  * along with extended types using up extra bits in the options section.
1061  * We want to convert this split number into a bitmap index, so we reverse the
1062  * calculation of IFM_X here.
1063  */
1064 #define IFM_IDX(x) (((x) & IFM_TMASK) | \
1065 		    (((x) & IFM_ETH_XTYPE) >> IFM_ETH_XSHIFT))
1066 
1067 /**
1068  * ice_add_media_types - Add supported media types to the media structure
1069  * @sc: ice private softc structure
1070  * @media: ifmedia structure to setup
1071  *
1072  * Looks up the supported phy types, and initializes the various media types
1073  * available.
1074  *
1075  * @pre this function must be protected from being called while another thread
1076  * is accessing the ifmedia types.
1077  */
1078 enum ice_status
1079 ice_add_media_types(struct ice_softc *sc, struct ifmedia *media)
1080 {
1081 	struct ice_aqc_get_phy_caps_data pcaps = { 0 };
1082 	struct ice_port_info *pi = sc->hw.port_info;
1083 	enum ice_status status;
1084 	uint64_t phy_low, phy_high;
1085 	int bit;
1086 
1087 	ASSERT_CFG_LOCKED(sc);
1088 
1089 	/* the maximum possible media type index is 511. We probably don't
1090 	 * need most of this space, but this ensures future compatibility when
1091 	 * additional media types are used.
1092 	 */
1093 	ice_declare_bitmap(already_added, 511);
1094 
1095 	/* Remove all previous media types */
1096 	ifmedia_removeall(media);
1097 
1098 	status = ice_aq_get_phy_caps(pi, false, ICE_AQC_REPORT_ACTIVE_CFG,
1099 				     &pcaps, NULL);
1100 	if (status != ICE_SUCCESS) {
1101 		device_printf(sc->dev,
1102 		    "%s: ice_aq_get_phy_caps (ACTIVE) failed; status %s, aq_err %s\n",
1103 		    __func__, ice_status_str(status),
1104 		    ice_aq_str(sc->hw.adminq.sq_last_status));
1105 		return (status);
1106 	}
1107 	phy_low = le64toh(pcaps.phy_type_low);
1108 	phy_high = le64toh(pcaps.phy_type_high);
1109 
1110 	/* make sure the added bitmap is zero'd */
1111 	memset(already_added, 0, sizeof(already_added));
1112 
1113 	/* coverity[address_of] */
1114 	for_each_set_bit(bit, &phy_low, 64) {
1115 		uint64_t type = BIT_ULL(bit);
1116 		int ostype;
1117 
1118 		/* get the OS media type */
1119 		ostype = ice_get_phy_type_low(type);
1120 
1121 		/* don't bother adding the unknown type */
1122 		if (ostype == IFM_UNKNOWN)
1123 			continue;
1124 
1125 		/* only add each media type to the list once */
1126 		if (ice_is_bit_set(already_added, IFM_IDX(ostype)))
1127 			continue;
1128 
1129 		ifmedia_add(media, IFM_ETHER | ostype, 0, NULL);
1130 		ice_set_bit(IFM_IDX(ostype), already_added);
1131 	}
1132 
1133 	/* coverity[address_of] */
1134 	for_each_set_bit(bit, &phy_high, 64) {
1135 		uint64_t type = BIT_ULL(bit);
1136 		int ostype;
1137 
1138 		/* get the OS media type */
1139 		ostype = ice_get_phy_type_high(type);
1140 
1141 		/* don't bother adding the unknown type */
1142 		if (ostype == IFM_UNKNOWN)
1143 			continue;
1144 
1145 		/* only add each media type to the list once */
1146 		if (ice_is_bit_set(already_added, IFM_IDX(ostype)))
1147 			continue;
1148 
1149 		ifmedia_add(media, IFM_ETHER | ostype, 0, NULL);
1150 		ice_set_bit(IFM_IDX(ostype), already_added);
1151 	}
1152 
1153 	/* Use autoselect media by default */
1154 	ifmedia_add(media, IFM_ETHER | IFM_AUTO, 0, NULL);
1155 	ifmedia_set(media, IFM_ETHER | IFM_AUTO);
1156 
1157 	return (ICE_SUCCESS);
1158 }
1159 
1160 /**
1161  * ice_configure_rxq_interrupts - Configure HW Rx queues for MSI-X interrupts
1162  * @vsi: the VSI to configure
1163  *
1164  * Called when setting up MSI-X interrupts to configure the Rx hardware queues.
1165  */
1166 void
1167 ice_configure_rxq_interrupts(struct ice_vsi *vsi)
1168 {
1169 	struct ice_hw *hw = &vsi->sc->hw;
1170 	int i;
1171 
1172 	for (i = 0; i < vsi->num_rx_queues; i++) {
1173 		struct ice_rx_queue *rxq = &vsi->rx_queues[i];
1174 		u32 val;
1175 
1176 		val = (QINT_RQCTL_CAUSE_ENA_M |
1177 		       (ICE_RX_ITR << QINT_RQCTL_ITR_INDX_S) |
1178 		       (rxq->irqv->me << QINT_RQCTL_MSIX_INDX_S));
1179 		wr32(hw, QINT_RQCTL(vsi->rx_qmap[rxq->me]), val);
1180 	}
1181 
1182 	ice_flush(hw);
1183 }
1184 
1185 /**
1186  * ice_configure_txq_interrupts - Configure HW Tx queues for MSI-X interrupts
1187  * @vsi: the VSI to configure
1188  *
1189  * Called when setting up MSI-X interrupts to configure the Tx hardware queues.
1190  */
1191 void
1192 ice_configure_txq_interrupts(struct ice_vsi *vsi)
1193 {
1194 	struct ice_hw *hw = &vsi->sc->hw;
1195 	int i;
1196 
1197 	for (i = 0; i < vsi->num_tx_queues; i++) {
1198 		struct ice_tx_queue *txq = &vsi->tx_queues[i];
1199 		u32 val;
1200 
1201 		val = (QINT_TQCTL_CAUSE_ENA_M |
1202 		       (ICE_TX_ITR << QINT_TQCTL_ITR_INDX_S) |
1203 		       (txq->irqv->me << QINT_TQCTL_MSIX_INDX_S));
1204 		wr32(hw, QINT_TQCTL(vsi->tx_qmap[txq->me]), val);
1205 	}
1206 
1207 	ice_flush(hw);
1208 }
1209 
1210 /**
1211  * ice_flush_rxq_interrupts - Unconfigure Hw Rx queues MSI-X interrupt cause
1212  * @vsi: the VSI to configure
1213  *
1214  * Unset the CAUSE_ENA flag of the TQCTL register for each queue, then trigger
1215  * a software interrupt on that cause. This is required as part of the Rx
1216  * queue disable logic to dissociate the Rx queue from the interrupt.
1217  *
1218  * Note: this function must be called prior to disabling Rx queues with
1219  * ice_control_rx_queues, otherwise the Rx queue may not be disabled properly.
1220  */
1221 void
1222 ice_flush_rxq_interrupts(struct ice_vsi *vsi)
1223 {
1224 	struct ice_hw *hw = &vsi->sc->hw;
1225 	int i;
1226 
1227 	for (i = 0; i < vsi->num_rx_queues; i++) {
1228 		struct ice_rx_queue *rxq = &vsi->rx_queues[i];
1229 		u32 reg, val;
1230 
1231 		/* Clear the CAUSE_ENA flag */
1232 		reg = vsi->rx_qmap[rxq->me];
1233 		val = rd32(hw, QINT_RQCTL(reg));
1234 		val &= ~QINT_RQCTL_CAUSE_ENA_M;
1235 		wr32(hw, QINT_RQCTL(reg), val);
1236 
1237 		ice_flush(hw);
1238 
1239 		/* Trigger a software interrupt to complete interrupt
1240 		 * dissociation.
1241 		 */
1242 		wr32(hw, GLINT_DYN_CTL(rxq->irqv->me),
1243 		     GLINT_DYN_CTL_SWINT_TRIG_M | GLINT_DYN_CTL_INTENA_MSK_M);
1244 	}
1245 }
1246 
1247 /**
1248  * ice_flush_txq_interrupts - Unconfigure Hw Tx queues MSI-X interrupt cause
1249  * @vsi: the VSI to configure
1250  *
1251  * Unset the CAUSE_ENA flag of the TQCTL register for each queue, then trigger
1252  * a software interrupt on that cause. This is required as part of the Tx
1253  * queue disable logic to dissociate the Tx queue from the interrupt.
1254  *
1255  * Note: this function must be called prior to ice_vsi_disable_tx, otherwise
1256  * the Tx queue disable may not complete properly.
1257  */
1258 void
1259 ice_flush_txq_interrupts(struct ice_vsi *vsi)
1260 {
1261 	struct ice_hw *hw = &vsi->sc->hw;
1262 	int i;
1263 
1264 	for (i = 0; i < vsi->num_tx_queues; i++) {
1265 		struct ice_tx_queue *txq = &vsi->tx_queues[i];
1266 		u32 reg, val;
1267 
1268 		/* Clear the CAUSE_ENA flag */
1269 		reg = vsi->tx_qmap[txq->me];
1270 		val = rd32(hw, QINT_TQCTL(reg));
1271 		val &= ~QINT_TQCTL_CAUSE_ENA_M;
1272 		wr32(hw, QINT_TQCTL(reg), val);
1273 
1274 		ice_flush(hw);
1275 
1276 		/* Trigger a software interrupt to complete interrupt
1277 		 * dissociation.
1278 		 */
1279 		wr32(hw, GLINT_DYN_CTL(txq->irqv->me),
1280 		     GLINT_DYN_CTL_SWINT_TRIG_M | GLINT_DYN_CTL_INTENA_MSK_M);
1281 	}
1282 }
1283 
1284 /**
1285  * ice_configure_rx_itr - Configure the Rx ITR settings for this VSI
1286  * @vsi: the VSI to configure
1287  *
1288  * Program the hardware ITR registers with the settings for this VSI.
1289  */
1290 void
1291 ice_configure_rx_itr(struct ice_vsi *vsi)
1292 {
1293 	struct ice_hw *hw = &vsi->sc->hw;
1294 	int i;
1295 
1296 	/* TODO: Handle per-queue/per-vector ITR? */
1297 
1298 	for (i = 0; i < vsi->num_rx_queues; i++) {
1299 		struct ice_rx_queue *rxq = &vsi->rx_queues[i];
1300 
1301 		wr32(hw, GLINT_ITR(ICE_RX_ITR, rxq->irqv->me),
1302 		     ice_itr_to_reg(hw, vsi->rx_itr));
1303 	}
1304 
1305 	ice_flush(hw);
1306 }
1307 
1308 /**
1309  * ice_configure_tx_itr - Configure the Tx ITR settings for this VSI
1310  * @vsi: the VSI to configure
1311  *
1312  * Program the hardware ITR registers with the settings for this VSI.
1313  */
1314 void
1315 ice_configure_tx_itr(struct ice_vsi *vsi)
1316 {
1317 	struct ice_hw *hw = &vsi->sc->hw;
1318 	int i;
1319 
1320 	/* TODO: Handle per-queue/per-vector ITR? */
1321 
1322 	for (i = 0; i < vsi->num_tx_queues; i++) {
1323 		struct ice_tx_queue *txq = &vsi->tx_queues[i];
1324 
1325 		wr32(hw, GLINT_ITR(ICE_TX_ITR, txq->irqv->me),
1326 		     ice_itr_to_reg(hw, vsi->tx_itr));
1327 	}
1328 
1329 	ice_flush(hw);
1330 }
1331 
1332 /**
1333  * ice_setup_tx_ctx - Setup an ice_tlan_ctx structure for a queue
1334  * @txq: the Tx queue to configure
1335  * @tlan_ctx: the Tx LAN queue context structure to initialize
1336  * @pf_q: real queue number
1337  */
1338 static int
1339 ice_setup_tx_ctx(struct ice_tx_queue *txq, struct ice_tlan_ctx *tlan_ctx, u16 pf_q)
1340 {
1341 	struct ice_vsi *vsi = txq->vsi;
1342 	struct ice_softc *sc = vsi->sc;
1343 	struct ice_hw *hw = &sc->hw;
1344 
1345 	tlan_ctx->port_num = hw->port_info->lport;
1346 
1347 	/* number of descriptors in the queue */
1348 	tlan_ctx->qlen = txq->desc_count;
1349 
1350 	/* set the transmit queue base address, defined in 128 byte units */
1351 	tlan_ctx->base = txq->tx_paddr >> 7;
1352 
1353 	tlan_ctx->pf_num = hw->pf_id;
1354 
1355 	/* For now, we only have code supporting PF VSIs */
1356 	switch (vsi->type) {
1357 	case ICE_VSI_PF:
1358 		tlan_ctx->vmvf_type = ICE_TLAN_CTX_VMVF_TYPE_PF;
1359 		break;
1360 	default:
1361 		return (ENODEV);
1362 	}
1363 
1364 	tlan_ctx->src_vsi = ice_get_hw_vsi_num(hw, vsi->idx);
1365 
1366 	/* Enable TSO */
1367 	tlan_ctx->tso_ena = 1;
1368 	tlan_ctx->internal_usage_flag = 1;
1369 
1370 	tlan_ctx->tso_qnum = pf_q;
1371 
1372 	/*
1373 	 * Stick with the older legacy Tx queue interface, instead of the new
1374 	 * advanced queue interface.
1375 	 */
1376 	tlan_ctx->legacy_int = 1;
1377 
1378 	/* Descriptor WB mode */
1379 	tlan_ctx->wb_mode = 0;
1380 
1381 	return (0);
1382 }
1383 
1384 /**
1385  * ice_cfg_vsi_for_tx - Configure the hardware for Tx
1386  * @vsi: the VSI to configure
1387  *
1388  * Configure the device Tx queues through firmware AdminQ commands. After
1389  * this, Tx queues will be ready for transmit.
1390  */
1391 int
1392 ice_cfg_vsi_for_tx(struct ice_vsi *vsi)
1393 {
1394 	struct ice_aqc_add_tx_qgrp *qg;
1395 	struct ice_hw *hw = &vsi->sc->hw;
1396 	device_t dev = vsi->sc->dev;
1397 	enum ice_status status;
1398 	int i;
1399 	int err = 0;
1400 	u16 qg_size, pf_q;
1401 
1402 	qg_size = ice_struct_size(qg, txqs, 1);
1403 	qg = (struct ice_aqc_add_tx_qgrp *)malloc(qg_size, M_ICE, M_NOWAIT|M_ZERO);
1404 	if (!qg)
1405 		return (ENOMEM);
1406 
1407 	qg->num_txqs = 1;
1408 
1409 	for (i = 0; i < vsi->num_tx_queues; i++) {
1410 		struct ice_tlan_ctx tlan_ctx = { 0 };
1411 		struct ice_tx_queue *txq = &vsi->tx_queues[i];
1412 
1413 		pf_q = vsi->tx_qmap[txq->me];
1414 		qg->txqs[0].txq_id = htole16(pf_q);
1415 
1416 		err = ice_setup_tx_ctx(txq, &tlan_ctx, pf_q);
1417 		if (err)
1418 			goto free_txqg;
1419 
1420 		ice_set_ctx(hw, (u8 *)&tlan_ctx, qg->txqs[0].txq_ctx,
1421 			    ice_tlan_ctx_info);
1422 
1423 		status = ice_ena_vsi_txq(hw->port_info, vsi->idx, 0,
1424 					 i, 1, qg, qg_size, NULL);
1425 		if (status) {
1426 			device_printf(dev,
1427 				      "Failed to set LAN Tx queue context, err %s aq_err %s\n",
1428 				      ice_status_str(status), ice_aq_str(hw->adminq.sq_last_status));
1429 			err = ENODEV;
1430 			goto free_txqg;
1431 		}
1432 
1433 		/* Keep track of the Tx queue TEID */
1434 		if (pf_q == le16toh(qg->txqs[0].txq_id))
1435 			txq->q_teid = le32toh(qg->txqs[0].q_teid);
1436 	}
1437 
1438 free_txqg:
1439 	free(qg, M_ICE);
1440 
1441 	return (err);
1442 }
1443 
1444 /**
1445  * ice_setup_rx_ctx - Setup an Rx context structure for a receive queue
1446  * @rxq: the receive queue to program
1447  *
1448  * Setup an Rx queue context structure and program it into the hardware
1449  * registers. This is a necessary step for enabling the Rx queue.
1450  *
1451  * @pre the VSI associated with this queue must have initialized mbuf_sz
1452  */
1453 static int
1454 ice_setup_rx_ctx(struct ice_rx_queue *rxq)
1455 {
1456 	struct ice_rlan_ctx rlan_ctx = {0};
1457 	struct ice_vsi *vsi = rxq->vsi;
1458 	struct ice_softc *sc = vsi->sc;
1459 	struct ice_hw *hw = &sc->hw;
1460 	enum ice_status status;
1461 	u32 rxdid = ICE_RXDID_FLEX_NIC;
1462 	u32 regval;
1463 	u16 pf_q;
1464 
1465 	pf_q = vsi->rx_qmap[rxq->me];
1466 
1467 	/* set the receive queue base address, defined in 128 byte units */
1468 	rlan_ctx.base = rxq->rx_paddr >> 7;
1469 
1470 	rlan_ctx.qlen = rxq->desc_count;
1471 
1472 	rlan_ctx.dbuf = vsi->mbuf_sz >> ICE_RLAN_CTX_DBUF_S;
1473 
1474 	/* use 32 byte descriptors */
1475 	rlan_ctx.dsize = 1;
1476 
1477 	/* Strip the Ethernet CRC bytes before the packet is posted to the
1478 	 * host memory.
1479 	 */
1480 	rlan_ctx.crcstrip = 1;
1481 
1482 	rlan_ctx.l2tsel = 1;
1483 
1484 	/* don't do header splitting */
1485 	rlan_ctx.dtype = ICE_RX_DTYPE_NO_SPLIT;
1486 	rlan_ctx.hsplit_0 = ICE_RLAN_RX_HSPLIT_0_NO_SPLIT;
1487 	rlan_ctx.hsplit_1 = ICE_RLAN_RX_HSPLIT_1_NO_SPLIT;
1488 
1489 	/* strip VLAN from inner headers */
1490 	rlan_ctx.showiv = 1;
1491 
1492 	rlan_ctx.rxmax = min(vsi->max_frame_size,
1493 			     ICE_MAX_RX_SEGS * vsi->mbuf_sz);
1494 
1495 	rlan_ctx.lrxqthresh = 1;
1496 
1497 	if (vsi->type != ICE_VSI_VF) {
1498 		regval = rd32(hw, QRXFLXP_CNTXT(pf_q));
1499 		regval &= ~QRXFLXP_CNTXT_RXDID_IDX_M;
1500 		regval |= (rxdid << QRXFLXP_CNTXT_RXDID_IDX_S) &
1501 			QRXFLXP_CNTXT_RXDID_IDX_M;
1502 
1503 		regval &= ~QRXFLXP_CNTXT_RXDID_PRIO_M;
1504 		regval |= (0x03 << QRXFLXP_CNTXT_RXDID_PRIO_S) &
1505 			QRXFLXP_CNTXT_RXDID_PRIO_M;
1506 
1507 		wr32(hw, QRXFLXP_CNTXT(pf_q), regval);
1508 	}
1509 
1510 	status = ice_write_rxq_ctx(hw, &rlan_ctx, pf_q);
1511 	if (status) {
1512 		device_printf(sc->dev,
1513 			      "Failed to set LAN Rx queue context, err %s aq_err %s\n",
1514 			      ice_status_str(status), ice_aq_str(hw->adminq.sq_last_status));
1515 		return (EIO);
1516 	}
1517 
1518 	wr32(hw, rxq->tail, 0);
1519 
1520 	return 0;
1521 }
1522 
1523 /**
1524  * ice_cfg_vsi_for_rx - Configure the hardware for Rx
1525  * @vsi: the VSI to configure
1526  *
1527  * Prepare an Rx context descriptor and configure the device to receive
1528  * traffic.
1529  *
1530  * @pre the VSI must have initialized mbuf_sz
1531  */
1532 int
1533 ice_cfg_vsi_for_rx(struct ice_vsi *vsi)
1534 {
1535 	int i, err;
1536 
1537 	for (i = 0; i < vsi->num_rx_queues; i++) {
1538 		MPASS(vsi->mbuf_sz > 0);
1539 		err = ice_setup_rx_ctx(&vsi->rx_queues[i]);
1540 		if (err)
1541 			return err;
1542 	}
1543 
1544 	return (0);
1545 }
1546 
1547 /**
1548  * ice_is_rxq_ready - Check if an Rx queue is ready
1549  * @hw: ice hw structure
1550  * @pf_q: absolute PF queue index to check
1551  * @reg: on successful return, contains qrx_ctrl contents
1552  *
1553  * Reads the QRX_CTRL register and verifies if the queue is in a consistent
1554  * state. That is, QENA_REQ matches QENA_STAT. Used to check before making
1555  * a request to change the queue, as well as to verify the request has
1556  * finished. The queue should change status within a few microseconds, so we
1557  * use a small delay while polling the register.
1558  *
1559  * Returns an error code if the queue does not update after a few retries.
1560  */
1561 static int
1562 ice_is_rxq_ready(struct ice_hw *hw, int pf_q, u32 *reg)
1563 {
1564 	u32 qrx_ctrl, qena_req, qena_stat;
1565 	int i;
1566 
1567 	for (i = 0; i < ICE_Q_WAIT_RETRY_LIMIT; i++) {
1568 		qrx_ctrl = rd32(hw, QRX_CTRL(pf_q));
1569 		qena_req = (qrx_ctrl >> QRX_CTRL_QENA_REQ_S) & 1;
1570 		qena_stat = (qrx_ctrl >> QRX_CTRL_QENA_STAT_S) & 1;
1571 
1572 		/* if the request and status bits equal, then the queue is
1573 		 * fully disabled or enabled.
1574 		 */
1575 		if (qena_req == qena_stat) {
1576 			*reg = qrx_ctrl;
1577 			return (0);
1578 		}
1579 
1580 		/* wait a few microseconds before we check again */
1581 		DELAY(10);
1582 	}
1583 
1584 	return (ETIMEDOUT);
1585 }
1586 
1587 /**
1588  * ice_control_rx_queues - Configure hardware to start or stop the Rx queues
1589  * @vsi: VSI to enable/disable queues
1590  * @enable: true to enable queues, false to disable
1591  *
1592  * Control the Rx queues through the QRX_CTRL register, enabling or disabling
1593  * them. Wait for the appropriate time to ensure that the queues have actually
1594  * reached the expected state.
1595  */
1596 int
1597 ice_control_rx_queues(struct ice_vsi *vsi, bool enable)
1598 {
1599 	struct ice_hw *hw = &vsi->sc->hw;
1600 	device_t dev = vsi->sc->dev;
1601 	u32 qrx_ctrl = 0;
1602 	int i, err;
1603 
1604 	/* TODO: amortize waits by changing all queues up front and then
1605 	 * checking their status afterwards. This will become more necessary
1606 	 * when we have a large number of queues.
1607 	 */
1608 	for (i = 0; i < vsi->num_rx_queues; i++) {
1609 		struct ice_rx_queue *rxq = &vsi->rx_queues[i];
1610 		int pf_q = vsi->rx_qmap[rxq->me];
1611 
1612 		err = ice_is_rxq_ready(hw, pf_q, &qrx_ctrl);
1613 		if (err) {
1614 			device_printf(dev,
1615 				      "Rx queue %d is not ready\n",
1616 				      pf_q);
1617 			return err;
1618 		}
1619 
1620 		/* Skip if the queue is already in correct state */
1621 		if (enable == !!(qrx_ctrl & QRX_CTRL_QENA_STAT_M))
1622 			continue;
1623 
1624 		if (enable)
1625 			qrx_ctrl |= QRX_CTRL_QENA_REQ_M;
1626 		else
1627 			qrx_ctrl &= ~QRX_CTRL_QENA_REQ_M;
1628 		wr32(hw, QRX_CTRL(pf_q), qrx_ctrl);
1629 
1630 		/* wait for the queue to finalize the request */
1631 		err = ice_is_rxq_ready(hw, pf_q, &qrx_ctrl);
1632 		if (err) {
1633 			device_printf(dev,
1634 				      "Rx queue %d %sable timeout\n",
1635 				      pf_q, (enable ? "en" : "dis"));
1636 			return err;
1637 		}
1638 
1639 		/* this should never happen */
1640 		if (enable != !!(qrx_ctrl & QRX_CTRL_QENA_STAT_M)) {
1641 			device_printf(dev,
1642 				      "Rx queue %d invalid state\n",
1643 				      pf_q);
1644 			return (EDOOFUS);
1645 		}
1646 	}
1647 
1648 	return (0);
1649 }
1650 
1651 /**
1652  * ice_add_mac_to_list - Add MAC filter to a MAC filter list
1653  * @vsi: the VSI to forward to
1654  * @list: list which contains MAC filter entries
1655  * @addr: the MAC address to be added
1656  * @action: filter action to perform on match
1657  *
1658  * Adds a MAC address filter to the list which will be forwarded to firmware
1659  * to add a series of MAC address filters.
1660  *
1661  * Returns 0 on success, and an error code on failure.
1662  *
1663  */
1664 static int
1665 ice_add_mac_to_list(struct ice_vsi *vsi, struct ice_list_head *list,
1666 		    const u8 *addr, enum ice_sw_fwd_act_type action)
1667 {
1668 	struct ice_fltr_list_entry *entry;
1669 
1670 	entry = (__typeof(entry))malloc(sizeof(*entry), M_ICE, M_NOWAIT|M_ZERO);
1671 	if (!entry)
1672 		return (ENOMEM);
1673 
1674 	entry->fltr_info.flag = ICE_FLTR_TX;
1675 	entry->fltr_info.src_id = ICE_SRC_ID_VSI;
1676 	entry->fltr_info.lkup_type = ICE_SW_LKUP_MAC;
1677 	entry->fltr_info.fltr_act = action;
1678 	entry->fltr_info.vsi_handle = vsi->idx;
1679 	bcopy(addr, entry->fltr_info.l_data.mac.mac_addr, ETHER_ADDR_LEN);
1680 
1681 	LIST_ADD(&entry->list_entry, list);
1682 
1683 	return 0;
1684 }
1685 
1686 /**
1687  * ice_free_fltr_list - Free memory associated with a MAC address list
1688  * @list: the list to free
1689  *
1690  * Free the memory of each entry associated with the list.
1691  */
1692 static void
1693 ice_free_fltr_list(struct ice_list_head *list)
1694 {
1695 	struct ice_fltr_list_entry *e, *tmp;
1696 
1697 	LIST_FOR_EACH_ENTRY_SAFE(e, tmp, list, ice_fltr_list_entry, list_entry) {
1698 		LIST_DEL(&e->list_entry);
1699 		free(e, M_ICE);
1700 	}
1701 }
1702 
1703 /**
1704  * ice_add_vsi_mac_filter - Add a MAC address filter for a VSI
1705  * @vsi: the VSI to add the filter for
1706  * @addr: MAC address to add a filter for
1707  *
1708  * Add a MAC address filter for a given VSI. This is a wrapper around
1709  * ice_add_mac to simplify the interface. First, it only accepts a single
1710  * address, so we don't have to mess around with the list setup in other
1711  * functions. Second, it ignores the ICE_ERR_ALREADY_EXIST error, so that
1712  * callers don't need to worry about attempting to add the same filter twice.
1713  */
1714 int
1715 ice_add_vsi_mac_filter(struct ice_vsi *vsi, const u8 *addr)
1716 {
1717 	struct ice_list_head mac_addr_list;
1718 	struct ice_hw *hw = &vsi->sc->hw;
1719 	device_t dev = vsi->sc->dev;
1720 	enum ice_status status;
1721 	int err = 0;
1722 
1723 	INIT_LIST_HEAD(&mac_addr_list);
1724 
1725 	err = ice_add_mac_to_list(vsi, &mac_addr_list, addr, ICE_FWD_TO_VSI);
1726 	if (err)
1727 		goto free_mac_list;
1728 
1729 	status = ice_add_mac(hw, &mac_addr_list);
1730 	if (status == ICE_ERR_ALREADY_EXISTS) {
1731 		; /* Don't complain if we try to add a filter that already exists */
1732 	} else if (status) {
1733 		device_printf(dev,
1734 			      "Failed to add a filter for MAC %6D, err %s aq_err %s\n",
1735 			      addr, ":",
1736 			      ice_status_str(status),
1737 			      ice_aq_str(hw->adminq.sq_last_status));
1738 		err = (EIO);
1739 	}
1740 
1741 free_mac_list:
1742 	ice_free_fltr_list(&mac_addr_list);
1743 	return err;
1744 }
1745 
1746 /**
1747  * ice_cfg_pf_default_mac_filters - Setup default unicast and broadcast addrs
1748  * @sc: device softc structure
1749  *
1750  * Program the default unicast and broadcast filters for the PF VSI.
1751  */
1752 int
1753 ice_cfg_pf_default_mac_filters(struct ice_softc *sc)
1754 {
1755 	struct ice_vsi *vsi = &sc->pf_vsi;
1756 	struct ice_hw *hw = &sc->hw;
1757 	int err;
1758 
1759 	/* Add the LAN MAC address */
1760 	err = ice_add_vsi_mac_filter(vsi, hw->port_info->mac.lan_addr);
1761 	if (err)
1762 		return err;
1763 
1764 	/* Add the broadcast address */
1765 	err = ice_add_vsi_mac_filter(vsi, broadcastaddr);
1766 	if (err)
1767 		return err;
1768 
1769 	return (0);
1770 }
1771 
1772 /**
1773  * ice_remove_vsi_mac_filter - Remove a MAC address filter for a VSI
1774  * @vsi: the VSI to add the filter for
1775  * @addr: MAC address to remove a filter for
1776  *
1777  * Remove a MAC address filter from a given VSI. This is a wrapper around
1778  * ice_remove_mac to simplify the interface. First, it only accepts a single
1779  * address, so we don't have to mess around with the list setup in other
1780  * functions. Second, it ignores the ICE_ERR_DOES_NOT_EXIST error, so that
1781  * callers don't need to worry about attempting to remove filters which
1782  * haven't yet been added.
1783  */
1784 int
1785 ice_remove_vsi_mac_filter(struct ice_vsi *vsi, const u8 *addr)
1786 {
1787 	struct ice_list_head mac_addr_list;
1788 	struct ice_hw *hw = &vsi->sc->hw;
1789 	device_t dev = vsi->sc->dev;
1790 	enum ice_status status;
1791 	int err = 0;
1792 
1793 	INIT_LIST_HEAD(&mac_addr_list);
1794 
1795 	err = ice_add_mac_to_list(vsi, &mac_addr_list, addr, ICE_FWD_TO_VSI);
1796 	if (err)
1797 		goto free_mac_list;
1798 
1799 	status = ice_remove_mac(hw, &mac_addr_list);
1800 	if (status == ICE_ERR_DOES_NOT_EXIST) {
1801 		; /* Don't complain if we try to remove a filter that doesn't exist */
1802 	} else if (status) {
1803 		device_printf(dev,
1804 			      "Failed to remove a filter for MAC %6D, err %s aq_err %s\n",
1805 			      addr, ":",
1806 			      ice_status_str(status),
1807 			      ice_aq_str(hw->adminq.sq_last_status));
1808 		err = (EIO);
1809 	}
1810 
1811 free_mac_list:
1812 	ice_free_fltr_list(&mac_addr_list);
1813 	return err;
1814 }
1815 
1816 /**
1817  * ice_rm_pf_default_mac_filters - Remove default unicast and broadcast addrs
1818  * @sc: device softc structure
1819  *
1820  * Remove the default unicast and broadcast filters from the PF VSI.
1821  */
1822 int
1823 ice_rm_pf_default_mac_filters(struct ice_softc *sc)
1824 {
1825 	struct ice_vsi *vsi = &sc->pf_vsi;
1826 	struct ice_hw *hw = &sc->hw;
1827 	int err;
1828 
1829 	/* Remove the LAN MAC address */
1830 	err = ice_remove_vsi_mac_filter(vsi, hw->port_info->mac.lan_addr);
1831 	if (err)
1832 		return err;
1833 
1834 	/* Remove the broadcast address */
1835 	err = ice_remove_vsi_mac_filter(vsi, broadcastaddr);
1836 	if (err)
1837 		return (EIO);
1838 
1839 	return (0);
1840 }
1841 
1842 /**
1843  * ice_check_ctrlq_errors - Check for and report controlq errors
1844  * @sc: device private structure
1845  * @qname: name of the controlq
1846  * @cq: the controlq to check
1847  *
1848  * Check and report controlq errors. Currently all we do is report them to the
1849  * kernel message log, but we might want to improve this in the future, such
1850  * as to keep track of statistics.
1851  */
1852 static void
1853 ice_check_ctrlq_errors(struct ice_softc *sc, const char *qname,
1854 		       struct ice_ctl_q_info *cq)
1855 {
1856 	struct ice_hw *hw = &sc->hw;
1857 	u32 val;
1858 
1859 	/* Check for error indications. Note that all the controlqs use the
1860 	 * same register layout, so we use the PF_FW_AxQLEN defines only.
1861 	 */
1862 	val = rd32(hw, cq->rq.len);
1863 	if (val & (PF_FW_ARQLEN_ARQVFE_M | PF_FW_ARQLEN_ARQOVFL_M |
1864 		   PF_FW_ARQLEN_ARQCRIT_M)) {
1865 		if (val & PF_FW_ARQLEN_ARQVFE_M)
1866 			device_printf(sc->dev,
1867 				"%s Receive Queue VF Error detected\n", qname);
1868 		if (val & PF_FW_ARQLEN_ARQOVFL_M)
1869 			device_printf(sc->dev,
1870 				"%s Receive Queue Overflow Error detected\n",
1871 				qname);
1872 		if (val & PF_FW_ARQLEN_ARQCRIT_M)
1873 			device_printf(sc->dev,
1874 				"%s Receive Queue Critical Error detected\n",
1875 				qname);
1876 		val &= ~(PF_FW_ARQLEN_ARQVFE_M | PF_FW_ARQLEN_ARQOVFL_M |
1877 			 PF_FW_ARQLEN_ARQCRIT_M);
1878 		wr32(hw, cq->rq.len, val);
1879 	}
1880 
1881 	val = rd32(hw, cq->sq.len);
1882 	if (val & (PF_FW_ATQLEN_ATQVFE_M | PF_FW_ATQLEN_ATQOVFL_M |
1883 		   PF_FW_ATQLEN_ATQCRIT_M)) {
1884 		if (val & PF_FW_ATQLEN_ATQVFE_M)
1885 			device_printf(sc->dev,
1886 				"%s Send Queue VF Error detected\n", qname);
1887 		if (val & PF_FW_ATQLEN_ATQOVFL_M)
1888 			device_printf(sc->dev,
1889 				"%s Send Queue Overflow Error detected\n",
1890 				qname);
1891 		if (val & PF_FW_ATQLEN_ATQCRIT_M)
1892 			device_printf(sc->dev,
1893 				"%s Send Queue Critical Error detected\n",
1894 				qname);
1895 		val &= ~(PF_FW_ATQLEN_ATQVFE_M | PF_FW_ATQLEN_ATQOVFL_M |
1896 			 PF_FW_ATQLEN_ATQCRIT_M);
1897 		wr32(hw, cq->sq.len, val);
1898 	}
1899 }
1900 
1901 /**
1902  * ice_process_link_event - Process a link event indication from firmware
1903  * @sc: device softc structure
1904  * @e: the received event data
1905  *
1906  * Gets the current link status from hardware, and may print a message if an
1907  * unqualified is detected.
1908  */
1909 static void
1910 ice_process_link_event(struct ice_softc *sc,
1911 		       struct ice_rq_event_info __invariant_only *e)
1912 {
1913 	struct ice_port_info *pi = sc->hw.port_info;
1914 	struct ice_hw *hw = &sc->hw;
1915 	device_t dev = sc->dev;
1916 	enum ice_status status;
1917 
1918 	/* Sanity check that the data length matches */
1919 	MPASS(le16toh(e->desc.datalen) == sizeof(struct ice_aqc_get_link_status_data));
1920 
1921 	/*
1922 	 * Even though the adapter gets link status information inside the
1923 	 * event, it needs to send a Get Link Status AQ command in order
1924 	 * to re-enable link events.
1925 	 */
1926 	pi->phy.get_link_info = true;
1927 	ice_get_link_status(pi, &sc->link_up);
1928 
1929 	if (pi->phy.link_info.topo_media_conflict &
1930 	   (ICE_AQ_LINK_TOPO_CONFLICT | ICE_AQ_LINK_MEDIA_CONFLICT |
1931 	    ICE_AQ_LINK_TOPO_CORRUPT))
1932 		device_printf(dev,
1933 		    "Possible mis-configuration of the Ethernet port detected; please use the Intel (R) Ethernet Port Configuration Tool utility to address the issue.\n");
1934 
1935 	if ((pi->phy.link_info.link_info & ICE_AQ_MEDIA_AVAILABLE) &&
1936 	    !(pi->phy.link_info.link_info & ICE_AQ_LINK_UP)) {
1937 		if (!(pi->phy.link_info.an_info & ICE_AQ_QUALIFIED_MODULE))
1938 			device_printf(dev,
1939 			    "Link 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");
1940 		if (pi->phy.link_info.link_cfg_err & ICE_AQ_LINK_MODULE_POWER_UNSUPPORTED)
1941 			device_printf(dev,
1942 			    "The module's power requirements exceed the device's power supply. Cannot start link.\n");
1943 		if (pi->phy.link_info.link_cfg_err & ICE_AQ_LINK_INVAL_MAX_POWER_LIMIT)
1944 			device_printf(dev,
1945 			    "The installed module is incompatible with the device's NVM image. Cannot start link.\n");
1946 	}
1947 
1948 	if (!(pi->phy.link_info.link_info & ICE_AQ_MEDIA_AVAILABLE)) {
1949 		if (!ice_testandset_state(&sc->state, ICE_STATE_NO_MEDIA)) {
1950 			status = ice_aq_set_link_restart_an(pi, false, NULL);
1951 			if (status != ICE_SUCCESS)
1952 				device_printf(dev,
1953 				    "%s: ice_aq_set_link_restart_an: status %s, aq_err %s\n",
1954 				    __func__, ice_status_str(status),
1955 				    ice_aq_str(hw->adminq.sq_last_status));
1956 		}
1957 	}
1958 	/* ICE_STATE_NO_MEDIA is cleared when polling task detects media */
1959 
1960 	/* Indicate that link status must be reported again */
1961 	ice_clear_state(&sc->state, ICE_STATE_LINK_STATUS_REPORTED);
1962 
1963 	/* OS link info is updated elsewhere */
1964 }
1965 
1966 /**
1967  * ice_process_ctrlq_event - Respond to a controlq event
1968  * @sc: device private structure
1969  * @qname: the name for this controlq
1970  * @event: the event to process
1971  *
1972  * Perform actions in response to various controlq event notifications.
1973  */
1974 static void
1975 ice_process_ctrlq_event(struct ice_softc *sc, const char *qname,
1976 			struct ice_rq_event_info *event)
1977 {
1978 	u16 opcode;
1979 
1980 	opcode = le16toh(event->desc.opcode);
1981 
1982 	switch (opcode) {
1983 	case ice_aqc_opc_get_link_status:
1984 		ice_process_link_event(sc, event);
1985 		break;
1986 	case ice_mbx_opc_send_msg_to_pf:
1987 		/* TODO: handle IOV event */
1988 		break;
1989 	case ice_aqc_opc_lldp_set_mib_change:
1990 		ice_handle_mib_change_event(sc, event);
1991 		break;
1992 	case ice_aqc_opc_event_lan_overflow:
1993 		ice_handle_lan_overflow_event(sc, event);
1994 		break;
1995 	case ice_aqc_opc_get_health_status:
1996 		ice_handle_health_status_event(sc, event);
1997 		break;
1998 	default:
1999 		device_printf(sc->dev,
2000 			      "%s Receive Queue unhandled event 0x%04x ignored\n",
2001 			      qname, opcode);
2002 	}
2003 }
2004 
2005 /**
2006  * ice_process_ctrlq - helper function to process controlq rings
2007  * @sc: device private structure
2008  * @q_type: specific control queue type
2009  * @pending: return parameter to track remaining events
2010  *
2011  * Process controlq events for a given control queue type. Returns zero on
2012  * success, and an error code on failure. If successful, pending is the number
2013  * of remaining events left in the queue.
2014  */
2015 int
2016 ice_process_ctrlq(struct ice_softc *sc, enum ice_ctl_q q_type, u16 *pending)
2017 {
2018 	struct ice_rq_event_info event = { { 0 } };
2019 	struct ice_hw *hw = &sc->hw;
2020 	struct ice_ctl_q_info *cq;
2021 	enum ice_status status;
2022 	const char *qname;
2023 	int loop = 0;
2024 
2025 	switch (q_type) {
2026 	case ICE_CTL_Q_ADMIN:
2027 		cq = &hw->adminq;
2028 		qname = "Admin";
2029 		break;
2030 	case ICE_CTL_Q_MAILBOX:
2031 		cq = &hw->mailboxq;
2032 		qname = "Mailbox";
2033 		break;
2034 	default:
2035 		device_printf(sc->dev,
2036 			      "Unknown control queue type 0x%x\n",
2037 			      q_type);
2038 		return 0;
2039 	}
2040 
2041 	ice_check_ctrlq_errors(sc, qname, cq);
2042 
2043 	/*
2044 	 * Control queue processing happens during the admin task which may be
2045 	 * holding a non-sleepable lock, so we *must* use M_NOWAIT here.
2046 	 */
2047 	event.buf_len = cq->rq_buf_size;
2048 	event.msg_buf = (u8 *)malloc(event.buf_len, M_ICE, M_ZERO | M_NOWAIT);
2049 	if (!event.msg_buf) {
2050 		device_printf(sc->dev,
2051 			      "Unable to allocate memory for %s Receive Queue event\n",
2052 			      qname);
2053 		return (ENOMEM);
2054 	}
2055 
2056 	do {
2057 		status = ice_clean_rq_elem(hw, cq, &event, pending);
2058 		if (status == ICE_ERR_AQ_NO_WORK)
2059 			break;
2060 		if (status) {
2061 			if (q_type == ICE_CTL_Q_ADMIN)
2062 				device_printf(sc->dev,
2063 					      "%s Receive Queue event error %s\n",
2064 					      qname, ice_status_str(status));
2065 			else
2066 				device_printf(sc->dev,
2067 					      "%s Receive Queue event error %s\n",
2068 					      qname, ice_status_str(status));
2069 			free(event.msg_buf, M_ICE);
2070 			return (EIO);
2071 		}
2072 		/* XXX should we separate this handler by controlq type? */
2073 		ice_process_ctrlq_event(sc, qname, &event);
2074 	} while (*pending && (++loop < ICE_CTRLQ_WORK_LIMIT));
2075 
2076 	free(event.msg_buf, M_ICE);
2077 
2078 	return 0;
2079 }
2080 
2081 /**
2082  * pkg_ver_empty - Check if a package version is empty
2083  * @pkg_ver: the package version to check
2084  * @pkg_name: the package name to check
2085  *
2086  * Checks if the package version structure is empty. We consider a package
2087  * version as empty if none of the versions are non-zero and the name string
2088  * is null as well.
2089  *
2090  * This is used to check if the package version was initialized by the driver,
2091  * as we do not expect an actual DDP package file to have a zero'd version and
2092  * name.
2093  *
2094  * @returns true if the package version is valid, or false otherwise.
2095  */
2096 static bool
2097 pkg_ver_empty(struct ice_pkg_ver *pkg_ver, u8 *pkg_name)
2098 {
2099 	return (pkg_name[0] == '\0' &&
2100 		pkg_ver->major == 0 &&
2101 		pkg_ver->minor == 0 &&
2102 		pkg_ver->update == 0 &&
2103 		pkg_ver->draft == 0);
2104 }
2105 
2106 /**
2107  * pkg_ver_compatible - Check if the package version is compatible
2108  * @pkg_ver: the package version to check
2109  *
2110  * Compares the package version number to the driver's expected major/minor
2111  * version. Returns an integer indicating whether the version is older, newer,
2112  * or compatible with the driver.
2113  *
2114  * @returns 0 if the package version is compatible, -1 if the package version
2115  * is older, and 1 if the package version is newer than the driver version.
2116  */
2117 static int
2118 pkg_ver_compatible(struct ice_pkg_ver *pkg_ver)
2119 {
2120 	if (pkg_ver->major > ICE_PKG_SUPP_VER_MAJ)
2121 		return (1); /* newer */
2122 	else if ((pkg_ver->major == ICE_PKG_SUPP_VER_MAJ) &&
2123 		 (pkg_ver->minor > ICE_PKG_SUPP_VER_MNR))
2124 		return (1); /* newer */
2125 	else if ((pkg_ver->major == ICE_PKG_SUPP_VER_MAJ) &&
2126 		 (pkg_ver->minor == ICE_PKG_SUPP_VER_MNR))
2127 		return (0); /* compatible */
2128 	else
2129 		return (-1); /* older */
2130 }
2131 
2132 /**
2133  * ice_os_pkg_version_str - Format OS package version info into a sbuf
2134  * @hw: device hw structure
2135  * @buf: string buffer to store name/version string
2136  *
2137  * Formats the name and version of the OS DDP package as found in the ice_ddp
2138  * module into a string.
2139  *
2140  * @remark This will almost always be the same as the active package, but
2141  * could be different in some cases. Use ice_active_pkg_version_str to get the
2142  * version of the active DDP package.
2143  */
2144 static void
2145 ice_os_pkg_version_str(struct ice_hw *hw, struct sbuf *buf)
2146 {
2147 	char name_buf[ICE_PKG_NAME_SIZE];
2148 
2149 	/* If the OS DDP package info is empty, use "None" */
2150 	if (pkg_ver_empty(&hw->pkg_ver, hw->pkg_name)) {
2151 		sbuf_printf(buf, "None");
2152 		return;
2153 	}
2154 
2155 	/*
2156 	 * This should already be null-terminated, but since this is a raw
2157 	 * value from an external source, strlcpy() into a new buffer to
2158 	 * make sure.
2159 	 */
2160 	bzero(name_buf, sizeof(name_buf));
2161 	strlcpy(name_buf, (char *)hw->pkg_name, ICE_PKG_NAME_SIZE);
2162 
2163 	sbuf_printf(buf, "%s version %u.%u.%u.%u",
2164 	    name_buf,
2165 	    hw->pkg_ver.major,
2166 	    hw->pkg_ver.minor,
2167 	    hw->pkg_ver.update,
2168 	    hw->pkg_ver.draft);
2169 }
2170 
2171 /**
2172  * ice_active_pkg_version_str - Format active package version info into a sbuf
2173  * @hw: device hw structure
2174  * @buf: string buffer to store name/version string
2175  *
2176  * Formats the name and version of the active DDP package info into a string
2177  * buffer for use.
2178  */
2179 static void
2180 ice_active_pkg_version_str(struct ice_hw *hw, struct sbuf *buf)
2181 {
2182 	char name_buf[ICE_PKG_NAME_SIZE];
2183 
2184 	/* If the active DDP package info is empty, use "None" */
2185 	if (pkg_ver_empty(&hw->active_pkg_ver, hw->active_pkg_name)) {
2186 		sbuf_printf(buf, "None");
2187 		return;
2188 	}
2189 
2190 	/*
2191 	 * This should already be null-terminated, but since this is a raw
2192 	 * value from an external source, strlcpy() into a new buffer to
2193 	 * make sure.
2194 	 */
2195 	bzero(name_buf, sizeof(name_buf));
2196 	strlcpy(name_buf, (char *)hw->active_pkg_name, ICE_PKG_NAME_SIZE);
2197 
2198 	sbuf_printf(buf, "%s version %u.%u.%u.%u",
2199 	    name_buf,
2200 	    hw->active_pkg_ver.major,
2201 	    hw->active_pkg_ver.minor,
2202 	    hw->active_pkg_ver.update,
2203 	    hw->active_pkg_ver.draft);
2204 
2205 	if (hw->active_track_id != 0)
2206 		sbuf_printf(buf, ", track id 0x%08x", hw->active_track_id);
2207 }
2208 
2209 /**
2210  * ice_nvm_version_str - Format the NVM version information into a sbuf
2211  * @hw: device hw structure
2212  * @buf: string buffer to store version string
2213  *
2214  * Formats the NVM information including firmware version, API version, NVM
2215  * version, the EETRACK id, and OEM specific version information into a string
2216  * buffer.
2217  */
2218 static void
2219 ice_nvm_version_str(struct ice_hw *hw, struct sbuf *buf)
2220 {
2221 	struct ice_nvm_info *nvm = &hw->flash.nvm;
2222 	struct ice_orom_info *orom = &hw->flash.orom;
2223 	struct ice_netlist_info *netlist = &hw->flash.netlist;
2224 
2225 	/* Note that the netlist versions are stored in packed Binary Coded
2226 	 * Decimal format. The use of '%x' will correctly display these as
2227 	 * decimal numbers. This works because every 4 bits will be displayed
2228 	 * as a hexadecimal digit, and the BCD format will only use the values
2229 	 * 0-9.
2230 	 */
2231 	sbuf_printf(buf,
2232 		    "fw %u.%u.%u api %u.%u nvm %x.%02x etid %08x netlist %x.%x.%x-%x.%x.%x.%04x oem %u.%u.%u",
2233 		    hw->fw_maj_ver, hw->fw_min_ver, hw->fw_patch,
2234 		    hw->api_maj_ver, hw->api_min_ver,
2235 		    nvm->major, nvm->minor, nvm->eetrack,
2236 		    netlist->major, netlist->minor,
2237 		    netlist->type >> 16, netlist->type & 0xFFFF,
2238 		    netlist->rev, netlist->cust_ver, netlist->hash,
2239 		    orom->major, orom->build, orom->patch);
2240 }
2241 
2242 /**
2243  * ice_print_nvm_version - Print the NVM info to the kernel message log
2244  * @sc: the device softc structure
2245  *
2246  * Format and print an NVM version string using ice_nvm_version_str().
2247  */
2248 void
2249 ice_print_nvm_version(struct ice_softc *sc)
2250 {
2251 	struct ice_hw *hw = &sc->hw;
2252 	device_t dev = sc->dev;
2253 	struct sbuf *sbuf;
2254 
2255 	sbuf = sbuf_new_auto();
2256 	ice_nvm_version_str(hw, sbuf);
2257 	sbuf_finish(sbuf);
2258 	device_printf(dev, "%s\n", sbuf_data(sbuf));
2259 	sbuf_delete(sbuf);
2260 }
2261 
2262 /**
2263  * ice_update_vsi_hw_stats - Update VSI-specific ethernet statistics counters
2264  * @vsi: the VSI to be updated
2265  *
2266  * Reads hardware stats and updates the ice_vsi_hw_stats tracking structure with
2267  * the updated values.
2268  */
2269 void
2270 ice_update_vsi_hw_stats(struct ice_vsi *vsi)
2271 {
2272 	struct ice_eth_stats *prev_es, *cur_es;
2273 	struct ice_hw *hw = &vsi->sc->hw;
2274 	u16 vsi_num;
2275 
2276 	if (!ice_is_vsi_valid(hw, vsi->idx))
2277 		return;
2278 
2279 	vsi_num = ice_get_hw_vsi_num(hw, vsi->idx); /* HW absolute index of a VSI */
2280 	prev_es = &vsi->hw_stats.prev;
2281 	cur_es = &vsi->hw_stats.cur;
2282 
2283 #define ICE_VSI_STAT40(name, location) \
2284 	ice_stat_update40(hw, name ## L(vsi_num), \
2285 			  vsi->hw_stats.offsets_loaded, \
2286 			  &prev_es->location, &cur_es->location)
2287 
2288 #define ICE_VSI_STAT32(name, location) \
2289 	ice_stat_update32(hw, name(vsi_num), \
2290 			  vsi->hw_stats.offsets_loaded, \
2291 			  &prev_es->location, &cur_es->location)
2292 
2293 	ICE_VSI_STAT40(GLV_GORC, rx_bytes);
2294 	ICE_VSI_STAT40(GLV_UPRC, rx_unicast);
2295 	ICE_VSI_STAT40(GLV_MPRC, rx_multicast);
2296 	ICE_VSI_STAT40(GLV_BPRC, rx_broadcast);
2297 	ICE_VSI_STAT32(GLV_RDPC, rx_discards);
2298 	ICE_VSI_STAT40(GLV_GOTC, tx_bytes);
2299 	ICE_VSI_STAT40(GLV_UPTC, tx_unicast);
2300 	ICE_VSI_STAT40(GLV_MPTC, tx_multicast);
2301 	ICE_VSI_STAT40(GLV_BPTC, tx_broadcast);
2302 	ICE_VSI_STAT32(GLV_TEPC, tx_errors);
2303 
2304 	ice_stat_update_repc(hw, vsi->idx, vsi->hw_stats.offsets_loaded,
2305 			     cur_es);
2306 
2307 #undef ICE_VSI_STAT40
2308 #undef ICE_VSI_STAT32
2309 
2310 	vsi->hw_stats.offsets_loaded = true;
2311 }
2312 
2313 /**
2314  * ice_reset_vsi_stats - Reset VSI statistics counters
2315  * @vsi: VSI structure
2316  *
2317  * Resets the software tracking counters for the VSI statistics, and indicate
2318  * that the offsets haven't been loaded. This is intended to be called
2319  * post-reset so that VSI statistics count from zero again.
2320  */
2321 void
2322 ice_reset_vsi_stats(struct ice_vsi *vsi)
2323 {
2324 	/* Reset HW stats */
2325 	memset(&vsi->hw_stats.prev, 0, sizeof(vsi->hw_stats.prev));
2326 	memset(&vsi->hw_stats.cur, 0, sizeof(vsi->hw_stats.cur));
2327 	vsi->hw_stats.offsets_loaded = false;
2328 }
2329 
2330 /**
2331  * ice_update_pf_stats - Update port stats counters
2332  * @sc: device private softc structure
2333  *
2334  * Reads hardware statistics registers and updates the software tracking
2335  * structure with new values.
2336  */
2337 void
2338 ice_update_pf_stats(struct ice_softc *sc)
2339 {
2340 	struct ice_hw_port_stats *prev_ps, *cur_ps;
2341 	struct ice_hw *hw = &sc->hw;
2342 	u8 lport;
2343 
2344 	MPASS(hw->port_info);
2345 
2346 	prev_ps = &sc->stats.prev;
2347 	cur_ps = &sc->stats.cur;
2348 	lport = hw->port_info->lport;
2349 
2350 #define ICE_PF_STAT40(name, location) \
2351 	ice_stat_update40(hw, name ## L(lport), \
2352 			  sc->stats.offsets_loaded, \
2353 			  &prev_ps->location, &cur_ps->location)
2354 
2355 #define ICE_PF_STAT32(name, location) \
2356 	ice_stat_update32(hw, name(lport), \
2357 			  sc->stats.offsets_loaded, \
2358 			  &prev_ps->location, &cur_ps->location)
2359 
2360 	ICE_PF_STAT40(GLPRT_GORC, eth.rx_bytes);
2361 	ICE_PF_STAT40(GLPRT_UPRC, eth.rx_unicast);
2362 	ICE_PF_STAT40(GLPRT_MPRC, eth.rx_multicast);
2363 	ICE_PF_STAT40(GLPRT_BPRC, eth.rx_broadcast);
2364 	ICE_PF_STAT40(GLPRT_GOTC, eth.tx_bytes);
2365 	ICE_PF_STAT40(GLPRT_UPTC, eth.tx_unicast);
2366 	ICE_PF_STAT40(GLPRT_MPTC, eth.tx_multicast);
2367 	ICE_PF_STAT40(GLPRT_BPTC, eth.tx_broadcast);
2368 	/* This stat register doesn't have an lport */
2369 	ice_stat_update32(hw, PRTRPB_RDPC,
2370 			  sc->stats.offsets_loaded,
2371 			  &prev_ps->eth.rx_discards, &cur_ps->eth.rx_discards);
2372 
2373 	ICE_PF_STAT32(GLPRT_TDOLD, tx_dropped_link_down);
2374 	ICE_PF_STAT40(GLPRT_PRC64, rx_size_64);
2375 	ICE_PF_STAT40(GLPRT_PRC127, rx_size_127);
2376 	ICE_PF_STAT40(GLPRT_PRC255, rx_size_255);
2377 	ICE_PF_STAT40(GLPRT_PRC511, rx_size_511);
2378 	ICE_PF_STAT40(GLPRT_PRC1023, rx_size_1023);
2379 	ICE_PF_STAT40(GLPRT_PRC1522, rx_size_1522);
2380 	ICE_PF_STAT40(GLPRT_PRC9522, rx_size_big);
2381 	ICE_PF_STAT40(GLPRT_PTC64, tx_size_64);
2382 	ICE_PF_STAT40(GLPRT_PTC127, tx_size_127);
2383 	ICE_PF_STAT40(GLPRT_PTC255, tx_size_255);
2384 	ICE_PF_STAT40(GLPRT_PTC511, tx_size_511);
2385 	ICE_PF_STAT40(GLPRT_PTC1023, tx_size_1023);
2386 	ICE_PF_STAT40(GLPRT_PTC1522, tx_size_1522);
2387 	ICE_PF_STAT40(GLPRT_PTC9522, tx_size_big);
2388 
2389 	ICE_PF_STAT32(GLPRT_LXONRXC, link_xon_rx);
2390 	ICE_PF_STAT32(GLPRT_LXOFFRXC, link_xoff_rx);
2391 	ICE_PF_STAT32(GLPRT_LXONTXC, link_xon_tx);
2392 	ICE_PF_STAT32(GLPRT_LXOFFTXC, link_xoff_tx);
2393 	ICE_PF_STAT32(GLPRT_CRCERRS, crc_errors);
2394 	ICE_PF_STAT32(GLPRT_ILLERRC, illegal_bytes);
2395 	ICE_PF_STAT32(GLPRT_MLFC, mac_local_faults);
2396 	ICE_PF_STAT32(GLPRT_MRFC, mac_remote_faults);
2397 	ICE_PF_STAT32(GLPRT_RLEC, rx_len_errors);
2398 	ICE_PF_STAT32(GLPRT_RUC, rx_undersize);
2399 	ICE_PF_STAT32(GLPRT_RFC, rx_fragments);
2400 	ICE_PF_STAT32(GLPRT_ROC, rx_oversize);
2401 	ICE_PF_STAT32(GLPRT_RJC, rx_jabber);
2402 
2403 #undef ICE_PF_STAT40
2404 #undef ICE_PF_STAT32
2405 
2406 	sc->stats.offsets_loaded = true;
2407 }
2408 
2409 /**
2410  * ice_reset_pf_stats - Reset port stats counters
2411  * @sc: Device private softc structure
2412  *
2413  * Reset software tracking values for statistics to zero, and indicate that
2414  * offsets haven't been loaded. Intended to be called after a device reset so
2415  * that statistics count from zero again.
2416  */
2417 void
2418 ice_reset_pf_stats(struct ice_softc *sc)
2419 {
2420 	memset(&sc->stats.prev, 0, sizeof(sc->stats.prev));
2421 	memset(&sc->stats.cur, 0, sizeof(sc->stats.cur));
2422 	sc->stats.offsets_loaded = false;
2423 }
2424 
2425 /**
2426  * ice_sysctl_show_fw - sysctl callback to show firmware information
2427  * @oidp: sysctl oid structure
2428  * @arg1: pointer to private data structure
2429  * @arg2: unused
2430  * @req: sysctl request pointer
2431  *
2432  * Callback for the fw_version sysctl, to display the current firmware
2433  * information found at hardware init time.
2434  */
2435 static int
2436 ice_sysctl_show_fw(SYSCTL_HANDLER_ARGS)
2437 {
2438 	struct ice_softc *sc = (struct ice_softc *)arg1;
2439 	struct ice_hw *hw = &sc->hw;
2440 	struct sbuf *sbuf;
2441 
2442 	UNREFERENCED_PARAMETER(oidp);
2443 	UNREFERENCED_PARAMETER(arg2);
2444 
2445 	if (ice_driver_is_detaching(sc))
2446 		return (ESHUTDOWN);
2447 
2448 	sbuf = sbuf_new_for_sysctl(NULL, NULL, 128, req);
2449 	ice_nvm_version_str(hw, sbuf);
2450 	sbuf_finish(sbuf);
2451 	sbuf_delete(sbuf);
2452 
2453 	return (0);
2454 }
2455 
2456 /**
2457  * ice_sysctl_pba_number - sysctl callback to show PBA number
2458  * @oidp: sysctl oid structure
2459  * @arg1: pointer to private data structure
2460  * @arg2: unused
2461  * @req: sysctl request pointer
2462  *
2463  * Callback for the pba_number sysctl, used to read the Product Board Assembly
2464  * number for this device.
2465  */
2466 static int
2467 ice_sysctl_pba_number(SYSCTL_HANDLER_ARGS)
2468 {
2469 	struct ice_softc *sc = (struct ice_softc *)arg1;
2470 	struct ice_hw *hw = &sc->hw;
2471 	device_t dev = sc->dev;
2472 	u8 pba_string[32] = "";
2473 	enum ice_status status;
2474 
2475 	UNREFERENCED_PARAMETER(arg2);
2476 
2477 	if (ice_driver_is_detaching(sc))
2478 		return (ESHUTDOWN);
2479 
2480 	status = ice_read_pba_string(hw, pba_string, sizeof(pba_string));
2481 	if (status) {
2482 		device_printf(dev,
2483 		    "%s: failed to read PBA string from NVM; status %s, aq_err %s\n",
2484 		    __func__, ice_status_str(status),
2485 		    ice_aq_str(hw->adminq.sq_last_status));
2486 		return (EIO);
2487 	}
2488 
2489 	return sysctl_handle_string(oidp, pba_string, sizeof(pba_string), req);
2490 }
2491 
2492 /**
2493  * ice_sysctl_pkg_version - sysctl to show the active package version info
2494  * @oidp: sysctl oid structure
2495  * @arg1: pointer to private data structure
2496  * @arg2: unused
2497  * @req: sysctl request pointer
2498  *
2499  * Callback for the pkg_version sysctl, to display the active DDP package name
2500  * and version information.
2501  */
2502 static int
2503 ice_sysctl_pkg_version(SYSCTL_HANDLER_ARGS)
2504 {
2505 	struct ice_softc *sc = (struct ice_softc *)arg1;
2506 	struct ice_hw *hw = &sc->hw;
2507 	struct sbuf *sbuf;
2508 
2509 	UNREFERENCED_PARAMETER(oidp);
2510 	UNREFERENCED_PARAMETER(arg2);
2511 
2512 	if (ice_driver_is_detaching(sc))
2513 		return (ESHUTDOWN);
2514 
2515 	sbuf = sbuf_new_for_sysctl(NULL, NULL, 128, req);
2516 	ice_active_pkg_version_str(hw, sbuf);
2517 	sbuf_finish(sbuf);
2518 	sbuf_delete(sbuf);
2519 
2520 	return (0);
2521 }
2522 
2523 /**
2524  * ice_sysctl_os_pkg_version - sysctl to show the OS package version info
2525  * @oidp: sysctl oid structure
2526  * @arg1: pointer to private data structure
2527  * @arg2: unused
2528  * @req: sysctl request pointer
2529  *
2530  * Callback for the pkg_version sysctl, to display the OS DDP package name and
2531  * version info found in the ice_ddp module.
2532  */
2533 static int
2534 ice_sysctl_os_pkg_version(SYSCTL_HANDLER_ARGS)
2535 {
2536 	struct ice_softc *sc = (struct ice_softc *)arg1;
2537 	struct ice_hw *hw = &sc->hw;
2538 	struct sbuf *sbuf;
2539 
2540 	UNREFERENCED_PARAMETER(oidp);
2541 	UNREFERENCED_PARAMETER(arg2);
2542 
2543 	if (ice_driver_is_detaching(sc))
2544 		return (ESHUTDOWN);
2545 
2546 	sbuf = sbuf_new_for_sysctl(NULL, NULL, 128, req);
2547 	ice_os_pkg_version_str(hw, sbuf);
2548 	sbuf_finish(sbuf);
2549 	sbuf_delete(sbuf);
2550 
2551 	return (0);
2552 }
2553 
2554 /**
2555  * ice_sysctl_current_speed - sysctl callback to show current link speed
2556  * @oidp: sysctl oid structure
2557  * @arg1: pointer to private data structure
2558  * @arg2: unused
2559  * @req: sysctl request pointer
2560  *
2561  * Callback for the current_speed sysctl, to display the string representing
2562  * the current link speed.
2563  */
2564 static int
2565 ice_sysctl_current_speed(SYSCTL_HANDLER_ARGS)
2566 {
2567 	struct ice_softc *sc = (struct ice_softc *)arg1;
2568 	struct ice_hw *hw = &sc->hw;
2569 	struct sbuf *sbuf;
2570 
2571 	UNREFERENCED_PARAMETER(oidp);
2572 	UNREFERENCED_PARAMETER(arg2);
2573 
2574 	if (ice_driver_is_detaching(sc))
2575 		return (ESHUTDOWN);
2576 
2577 	sbuf = sbuf_new_for_sysctl(NULL, NULL, 10, req);
2578 	sbuf_printf(sbuf, "%s", ice_aq_speed_to_str(hw->port_info));
2579 	sbuf_finish(sbuf);
2580 	sbuf_delete(sbuf);
2581 
2582 	return (0);
2583 }
2584 
2585 /**
2586  * @var phy_link_speeds
2587  * @brief PHY link speed conversion array
2588  *
2589  * Array of link speeds to convert ICE_PHY_TYPE_LOW and ICE_PHY_TYPE_HIGH into
2590  * link speeds used by the link speed sysctls.
2591  *
2592  * @remark these are based on the indices used in the BIT() macros for the
2593  * ICE_PHY_TYPE_LOW_* and ICE_PHY_TYPE_HIGH_* definitions.
2594  */
2595 static const uint16_t phy_link_speeds[] = {
2596     ICE_AQ_LINK_SPEED_100MB,
2597     ICE_AQ_LINK_SPEED_100MB,
2598     ICE_AQ_LINK_SPEED_1000MB,
2599     ICE_AQ_LINK_SPEED_1000MB,
2600     ICE_AQ_LINK_SPEED_1000MB,
2601     ICE_AQ_LINK_SPEED_1000MB,
2602     ICE_AQ_LINK_SPEED_1000MB,
2603     ICE_AQ_LINK_SPEED_2500MB,
2604     ICE_AQ_LINK_SPEED_2500MB,
2605     ICE_AQ_LINK_SPEED_2500MB,
2606     ICE_AQ_LINK_SPEED_5GB,
2607     ICE_AQ_LINK_SPEED_5GB,
2608     ICE_AQ_LINK_SPEED_10GB,
2609     ICE_AQ_LINK_SPEED_10GB,
2610     ICE_AQ_LINK_SPEED_10GB,
2611     ICE_AQ_LINK_SPEED_10GB,
2612     ICE_AQ_LINK_SPEED_10GB,
2613     ICE_AQ_LINK_SPEED_10GB,
2614     ICE_AQ_LINK_SPEED_10GB,
2615     ICE_AQ_LINK_SPEED_25GB,
2616     ICE_AQ_LINK_SPEED_25GB,
2617     ICE_AQ_LINK_SPEED_25GB,
2618     ICE_AQ_LINK_SPEED_25GB,
2619     ICE_AQ_LINK_SPEED_25GB,
2620     ICE_AQ_LINK_SPEED_25GB,
2621     ICE_AQ_LINK_SPEED_25GB,
2622     ICE_AQ_LINK_SPEED_25GB,
2623     ICE_AQ_LINK_SPEED_25GB,
2624     ICE_AQ_LINK_SPEED_25GB,
2625     ICE_AQ_LINK_SPEED_25GB,
2626     ICE_AQ_LINK_SPEED_40GB,
2627     ICE_AQ_LINK_SPEED_40GB,
2628     ICE_AQ_LINK_SPEED_40GB,
2629     ICE_AQ_LINK_SPEED_40GB,
2630     ICE_AQ_LINK_SPEED_40GB,
2631     ICE_AQ_LINK_SPEED_40GB,
2632     ICE_AQ_LINK_SPEED_50GB,
2633     ICE_AQ_LINK_SPEED_50GB,
2634     ICE_AQ_LINK_SPEED_50GB,
2635     ICE_AQ_LINK_SPEED_50GB,
2636     ICE_AQ_LINK_SPEED_50GB,
2637     ICE_AQ_LINK_SPEED_50GB,
2638     ICE_AQ_LINK_SPEED_50GB,
2639     ICE_AQ_LINK_SPEED_50GB,
2640     ICE_AQ_LINK_SPEED_50GB,
2641     ICE_AQ_LINK_SPEED_50GB,
2642     ICE_AQ_LINK_SPEED_50GB,
2643     ICE_AQ_LINK_SPEED_50GB,
2644     ICE_AQ_LINK_SPEED_50GB,
2645     ICE_AQ_LINK_SPEED_50GB,
2646     ICE_AQ_LINK_SPEED_50GB,
2647     ICE_AQ_LINK_SPEED_100GB,
2648     ICE_AQ_LINK_SPEED_100GB,
2649     ICE_AQ_LINK_SPEED_100GB,
2650     ICE_AQ_LINK_SPEED_100GB,
2651     ICE_AQ_LINK_SPEED_100GB,
2652     ICE_AQ_LINK_SPEED_100GB,
2653     ICE_AQ_LINK_SPEED_100GB,
2654     ICE_AQ_LINK_SPEED_100GB,
2655     ICE_AQ_LINK_SPEED_100GB,
2656     ICE_AQ_LINK_SPEED_100GB,
2657     ICE_AQ_LINK_SPEED_100GB,
2658     ICE_AQ_LINK_SPEED_100GB,
2659     ICE_AQ_LINK_SPEED_100GB,
2660     /* These rates are for ICE_PHY_TYPE_HIGH_* */
2661     ICE_AQ_LINK_SPEED_100GB,
2662     ICE_AQ_LINK_SPEED_100GB,
2663     ICE_AQ_LINK_SPEED_100GB,
2664     ICE_AQ_LINK_SPEED_100GB,
2665     ICE_AQ_LINK_SPEED_100GB
2666 };
2667 
2668 #define ICE_SYSCTL_HELP_ADVERTISE_SPEED		\
2669 "\nControl advertised link speed."		\
2670 "\nFlags:"					\
2671 "\n\t   0x0 - Auto"				\
2672 "\n\t   0x1 - 10 Mb"				\
2673 "\n\t   0x2 - 100 Mb"				\
2674 "\n\t   0x4 - 1G"				\
2675 "\n\t   0x8 - 2.5G"				\
2676 "\n\t  0x10 - 5G"				\
2677 "\n\t  0x20 - 10G"				\
2678 "\n\t  0x40 - 20G"				\
2679 "\n\t  0x80 - 25G"				\
2680 "\n\t 0x100 - 40G"				\
2681 "\n\t 0x200 - 50G"				\
2682 "\n\t 0x400 - 100G"				\
2683 "\n\t0x8000 - Unknown"				\
2684 "\n\t"						\
2685 "\nUse \"sysctl -x\" to view flags properly."
2686 
2687 #define ICE_PHYS_100MB			\
2688     (ICE_PHY_TYPE_LOW_100BASE_TX |	\
2689      ICE_PHY_TYPE_LOW_100M_SGMII)
2690 #define ICE_PHYS_1000MB			\
2691     (ICE_PHY_TYPE_LOW_1000BASE_T |	\
2692      ICE_PHY_TYPE_LOW_1000BASE_SX |	\
2693      ICE_PHY_TYPE_LOW_1000BASE_LX |	\
2694      ICE_PHY_TYPE_LOW_1000BASE_KX |	\
2695      ICE_PHY_TYPE_LOW_1G_SGMII)
2696 #define ICE_PHYS_2500MB			\
2697     (ICE_PHY_TYPE_LOW_2500BASE_T |	\
2698      ICE_PHY_TYPE_LOW_2500BASE_X |	\
2699      ICE_PHY_TYPE_LOW_2500BASE_KX)
2700 #define ICE_PHYS_5GB			\
2701     (ICE_PHY_TYPE_LOW_5GBASE_T |	\
2702      ICE_PHY_TYPE_LOW_5GBASE_KR)
2703 #define ICE_PHYS_10GB			\
2704     (ICE_PHY_TYPE_LOW_10GBASE_T |	\
2705      ICE_PHY_TYPE_LOW_10G_SFI_DA |	\
2706      ICE_PHY_TYPE_LOW_10GBASE_SR |	\
2707      ICE_PHY_TYPE_LOW_10GBASE_LR |	\
2708      ICE_PHY_TYPE_LOW_10GBASE_KR_CR1 |	\
2709      ICE_PHY_TYPE_LOW_10G_SFI_AOC_ACC |	\
2710      ICE_PHY_TYPE_LOW_10G_SFI_C2C)
2711 #define ICE_PHYS_25GB			\
2712     (ICE_PHY_TYPE_LOW_25GBASE_T |	\
2713      ICE_PHY_TYPE_LOW_25GBASE_CR |	\
2714      ICE_PHY_TYPE_LOW_25GBASE_CR_S |	\
2715      ICE_PHY_TYPE_LOW_25GBASE_CR1 |	\
2716      ICE_PHY_TYPE_LOW_25GBASE_SR |	\
2717      ICE_PHY_TYPE_LOW_25GBASE_LR |	\
2718      ICE_PHY_TYPE_LOW_25GBASE_KR |	\
2719      ICE_PHY_TYPE_LOW_25GBASE_KR_S |	\
2720      ICE_PHY_TYPE_LOW_25GBASE_KR1 |	\
2721      ICE_PHY_TYPE_LOW_25G_AUI_AOC_ACC |	\
2722      ICE_PHY_TYPE_LOW_25G_AUI_C2C)
2723 #define ICE_PHYS_40GB			\
2724     (ICE_PHY_TYPE_LOW_40GBASE_CR4 |	\
2725      ICE_PHY_TYPE_LOW_40GBASE_SR4 |	\
2726      ICE_PHY_TYPE_LOW_40GBASE_LR4 |	\
2727      ICE_PHY_TYPE_LOW_40GBASE_KR4 |	\
2728      ICE_PHY_TYPE_LOW_40G_XLAUI_AOC_ACC | \
2729      ICE_PHY_TYPE_LOW_40G_XLAUI)
2730 #define ICE_PHYS_50GB			\
2731     (ICE_PHY_TYPE_LOW_50GBASE_CR2 |	\
2732      ICE_PHY_TYPE_LOW_50GBASE_SR2 |	\
2733      ICE_PHY_TYPE_LOW_50GBASE_LR2 |	\
2734      ICE_PHY_TYPE_LOW_50GBASE_KR2 |	\
2735      ICE_PHY_TYPE_LOW_50G_LAUI2_AOC_ACC | \
2736      ICE_PHY_TYPE_LOW_50G_LAUI2 |	\
2737      ICE_PHY_TYPE_LOW_50G_AUI2_AOC_ACC | \
2738      ICE_PHY_TYPE_LOW_50G_AUI2 |	\
2739      ICE_PHY_TYPE_LOW_50GBASE_CP |	\
2740      ICE_PHY_TYPE_LOW_50GBASE_SR |	\
2741      ICE_PHY_TYPE_LOW_50GBASE_FR |	\
2742      ICE_PHY_TYPE_LOW_50GBASE_LR |	\
2743      ICE_PHY_TYPE_LOW_50GBASE_KR_PAM4 |	\
2744      ICE_PHY_TYPE_LOW_50G_AUI1_AOC_ACC | \
2745      ICE_PHY_TYPE_LOW_50G_AUI1)
2746 #define ICE_PHYS_100GB_LOW		\
2747     (ICE_PHY_TYPE_LOW_100GBASE_CR4 |	\
2748      ICE_PHY_TYPE_LOW_100GBASE_SR4 |	\
2749      ICE_PHY_TYPE_LOW_100GBASE_LR4 |	\
2750      ICE_PHY_TYPE_LOW_100GBASE_KR4 |	\
2751      ICE_PHY_TYPE_LOW_100G_CAUI4_AOC_ACC | \
2752      ICE_PHY_TYPE_LOW_100G_CAUI4 |	\
2753      ICE_PHY_TYPE_LOW_100G_AUI4_AOC_ACC | \
2754      ICE_PHY_TYPE_LOW_100G_AUI4 |	\
2755      ICE_PHY_TYPE_LOW_100GBASE_CR_PAM4 | \
2756      ICE_PHY_TYPE_LOW_100GBASE_KR_PAM4 | \
2757      ICE_PHY_TYPE_LOW_100GBASE_CP2 |	\
2758      ICE_PHY_TYPE_LOW_100GBASE_SR2 |	\
2759      ICE_PHY_TYPE_LOW_100GBASE_DR)
2760 #define ICE_PHYS_100GB_HIGH		\
2761     (ICE_PHY_TYPE_HIGH_100GBASE_KR2_PAM4 | \
2762      ICE_PHY_TYPE_HIGH_100G_CAUI2_AOC_ACC | \
2763      ICE_PHY_TYPE_HIGH_100G_CAUI2 |	\
2764      ICE_PHY_TYPE_HIGH_100G_AUI2_AOC_ACC | \
2765      ICE_PHY_TYPE_HIGH_100G_AUI2)
2766 
2767 /**
2768  * ice_aq_phy_types_to_link_speeds - Convert the PHY Types to speeds
2769  * @phy_type_low: lower 64-bit PHY Type bitmask
2770  * @phy_type_high: upper 64-bit PHY Type bitmask
2771  *
2772  * Convert the PHY Type fields from Get PHY Abilities and Set PHY Config into
2773  * link speed flags. If phy_type_high has an unknown PHY type, then the return
2774  * value will include the "ICE_AQ_LINK_SPEED_UNKNOWN" flag as well.
2775  */
2776 static u16
2777 ice_aq_phy_types_to_link_speeds(u64 phy_type_low, u64 phy_type_high)
2778 {
2779 	u16 sysctl_speeds = 0;
2780 	int bit;
2781 
2782 	/* coverity[address_of] */
2783 	for_each_set_bit(bit, &phy_type_low, 64)
2784 		sysctl_speeds |= phy_link_speeds[bit];
2785 
2786 	/* coverity[address_of] */
2787 	for_each_set_bit(bit, &phy_type_high, 64) {
2788 		if ((bit + 64) < (int)ARRAY_SIZE(phy_link_speeds))
2789 			sysctl_speeds |= phy_link_speeds[bit + 64];
2790 		else
2791 			sysctl_speeds |= ICE_AQ_LINK_SPEED_UNKNOWN;
2792 	}
2793 
2794 	return (sysctl_speeds);
2795 }
2796 
2797 /**
2798  * ice_sysctl_speeds_to_aq_phy_types - Convert sysctl speed flags to AQ PHY flags
2799  * @sysctl_speeds: 16-bit sysctl speeds or AQ_LINK_SPEED flags
2800  * @phy_type_low: output parameter for lower AQ PHY flags
2801  * @phy_type_high: output parameter for higher AQ PHY flags
2802  *
2803  * Converts the given link speed flags into AQ PHY type flag sets appropriate
2804  * for use in a Set PHY Config command.
2805  */
2806 static void
2807 ice_sysctl_speeds_to_aq_phy_types(u16 sysctl_speeds, u64 *phy_type_low,
2808 				  u64 *phy_type_high)
2809 {
2810 	*phy_type_low = 0, *phy_type_high = 0;
2811 
2812 	if (sysctl_speeds & ICE_AQ_LINK_SPEED_100MB)
2813 		*phy_type_low |= ICE_PHYS_100MB;
2814 	if (sysctl_speeds & ICE_AQ_LINK_SPEED_1000MB)
2815 		*phy_type_low |= ICE_PHYS_1000MB;
2816 	if (sysctl_speeds & ICE_AQ_LINK_SPEED_2500MB)
2817 		*phy_type_low |= ICE_PHYS_2500MB;
2818 	if (sysctl_speeds & ICE_AQ_LINK_SPEED_5GB)
2819 		*phy_type_low |= ICE_PHYS_5GB;
2820 	if (sysctl_speeds & ICE_AQ_LINK_SPEED_10GB)
2821 		*phy_type_low |= ICE_PHYS_10GB;
2822 	if (sysctl_speeds & ICE_AQ_LINK_SPEED_25GB)
2823 		*phy_type_low |= ICE_PHYS_25GB;
2824 	if (sysctl_speeds & ICE_AQ_LINK_SPEED_40GB)
2825 		*phy_type_low |= ICE_PHYS_40GB;
2826 	if (sysctl_speeds & ICE_AQ_LINK_SPEED_50GB)
2827 		*phy_type_low |= ICE_PHYS_50GB;
2828 	if (sysctl_speeds & ICE_AQ_LINK_SPEED_100GB) {
2829 		*phy_type_low |= ICE_PHYS_100GB_LOW;
2830 		*phy_type_high |= ICE_PHYS_100GB_HIGH;
2831 	}
2832 }
2833 
2834 /**
2835  * @struct ice_phy_data
2836  * @brief PHY caps and link speeds
2837  *
2838  * Buffer providing report mode and user speeds;
2839  * returning intersection of PHY types and speeds.
2840  */
2841 struct ice_phy_data {
2842 	u64 phy_low_orig;     /* PHY low quad from report */
2843 	u64 phy_high_orig;    /* PHY high quad from report */
2844 	u64 phy_low_intr;     /* PHY low quad intersection with user speeds */
2845 	u64 phy_high_intr;    /* PHY high quad intersection with user speeds */
2846 	u16 user_speeds_orig; /* Input from caller - See ICE_AQ_LINK_SPEED_* */
2847 	u16 user_speeds_intr; /* Intersect with report speeds */
2848 	u8 report_mode;       /* See ICE_AQC_REPORT_* */
2849 };
2850 
2851 /**
2852  * ice_intersect_phy_types_and_speeds - Return intersection of link speeds
2853  * @sc: device private structure
2854  * @phy_data: device PHY data
2855  *
2856  * On read: Displays the currently supported speeds
2857  * On write: Sets the device's supported speeds
2858  * Valid input flags: see ICE_SYSCTL_HELP_ADVERTISE_SPEED
2859  */
2860 static int
2861 ice_intersect_phy_types_and_speeds(struct ice_softc *sc,
2862 				   struct ice_phy_data *phy_data)
2863 {
2864 	struct ice_aqc_get_phy_caps_data pcaps = { 0 };
2865 	const char *report_types[5] = { "w/o MEDIA",
2866 					"w/MEDIA",
2867 					"ACTIVE",
2868 					"EDOOFUS", /* Not used */
2869 					"DFLT" };
2870 	struct ice_hw *hw = &sc->hw;
2871 	struct ice_port_info *pi = hw->port_info;
2872 	enum ice_status status;
2873 	u16 report_speeds, temp_speeds;
2874 	u8 report_type;
2875 	bool apply_speed_filter = false;
2876 
2877 	switch (phy_data->report_mode) {
2878 	case ICE_AQC_REPORT_TOPO_CAP_NO_MEDIA:
2879 	case ICE_AQC_REPORT_TOPO_CAP_MEDIA:
2880 	case ICE_AQC_REPORT_ACTIVE_CFG:
2881 	case ICE_AQC_REPORT_DFLT_CFG:
2882 		report_type = phy_data->report_mode >> 1;
2883 		break;
2884 	default:
2885 		device_printf(sc->dev,
2886 		    "%s: phy_data.report_mode \"%u\" doesn't exist\n",
2887 		    __func__, phy_data->report_mode);
2888 		return (EINVAL);
2889 	}
2890 
2891 	/* 0 is treated as "Auto"; the driver will handle selecting the
2892 	 * correct speeds. Including, in some cases, applying an override
2893 	 * if provided.
2894 	 */
2895 	if (phy_data->user_speeds_orig == 0)
2896 		phy_data->user_speeds_orig = USHRT_MAX;
2897 	else if (ice_is_bit_set(sc->feat_en, ICE_FEATURE_LENIENT_LINK_MODE))
2898 		apply_speed_filter = true;
2899 
2900 	status = ice_aq_get_phy_caps(pi, false, phy_data->report_mode, &pcaps, NULL);
2901 	if (status != ICE_SUCCESS) {
2902 		device_printf(sc->dev,
2903 		    "%s: ice_aq_get_phy_caps (%s) failed; status %s, aq_err %s\n",
2904 		    __func__, report_types[report_type],
2905 		    ice_status_str(status),
2906 		    ice_aq_str(sc->hw.adminq.sq_last_status));
2907 		return (EIO);
2908 	}
2909 
2910 	phy_data->phy_low_orig = le64toh(pcaps.phy_type_low);
2911 	phy_data->phy_high_orig = le64toh(pcaps.phy_type_high);
2912 	report_speeds = ice_aq_phy_types_to_link_speeds(phy_data->phy_low_orig,
2913 	    phy_data->phy_high_orig);
2914 	if (apply_speed_filter) {
2915 		temp_speeds = ice_apply_supported_speed_filter(report_speeds);
2916 		if ((phy_data->user_speeds_orig & temp_speeds) == 0) {
2917 			device_printf(sc->dev,
2918 			    "User-specified speeds (\"0x%04X\") not supported\n",
2919 			    phy_data->user_speeds_orig);
2920 			return (EINVAL);
2921 		}
2922 		report_speeds = temp_speeds;
2923 	}
2924 	ice_sysctl_speeds_to_aq_phy_types(phy_data->user_speeds_orig,
2925 	    &phy_data->phy_low_intr, &phy_data->phy_high_intr);
2926 	phy_data->user_speeds_intr = phy_data->user_speeds_orig & report_speeds;
2927 	phy_data->phy_low_intr &= phy_data->phy_low_orig;
2928 	phy_data->phy_high_intr &= phy_data->phy_high_orig;
2929 
2930 	return (0);
2931  }
2932 
2933 /**
2934  * ice_sysctl_advertise_speed - Display/change link speeds supported by port
2935  * @oidp: sysctl oid structure
2936  * @arg1: pointer to private data structure
2937  * @arg2: unused
2938  * @req: sysctl request pointer
2939  *
2940  * On read: Displays the currently supported speeds
2941  * On write: Sets the device's supported speeds
2942  * Valid input flags: see ICE_SYSCTL_HELP_ADVERTISE_SPEED
2943  */
2944 static int
2945 ice_sysctl_advertise_speed(SYSCTL_HANDLER_ARGS)
2946 {
2947 	struct ice_softc *sc = (struct ice_softc *)arg1;
2948 	struct ice_port_info *pi = sc->hw.port_info;
2949 	struct ice_phy_data phy_data = { 0 };
2950 	device_t dev = sc->dev;
2951 	u16 sysctl_speeds;
2952 	int ret;
2953 
2954 	UNREFERENCED_PARAMETER(arg2);
2955 
2956 	if (ice_driver_is_detaching(sc))
2957 		return (ESHUTDOWN);
2958 
2959 	/* Get the current speeds from the adapter's "active" configuration. */
2960 	phy_data.report_mode = ICE_AQC_REPORT_ACTIVE_CFG;
2961 	ret = ice_intersect_phy_types_and_speeds(sc, &phy_data);
2962 	if (ret) {
2963 		/* Error message already printed within function */
2964 		return (ret);
2965 	}
2966 
2967 	sysctl_speeds = phy_data.user_speeds_intr;
2968 
2969 	ret = sysctl_handle_16(oidp, &sysctl_speeds, 0, req);
2970 	if ((ret) || (req->newptr == NULL))
2971 		return (ret);
2972 
2973 	if (sysctl_speeds > 0x7FF) {
2974 		device_printf(dev,
2975 			      "%s: \"%u\" is outside of the range of acceptable values.\n",
2976 			      __func__, sysctl_speeds);
2977 		return (EINVAL);
2978 	}
2979 
2980 	pi->phy.curr_user_speed_req = sysctl_speeds;
2981 
2982 	/* Apply settings requested by user */
2983 	return ice_apply_saved_phy_cfg(sc, ICE_APPLY_LS);
2984 }
2985 
2986 #define ICE_SYSCTL_HELP_FEC_CONFIG			\
2987 "\nDisplay or set the port's requested FEC mode."	\
2988 "\n\tauto - " ICE_FEC_STRING_AUTO			\
2989 "\n\tfc - " ICE_FEC_STRING_BASER			\
2990 "\n\trs - " ICE_FEC_STRING_RS				\
2991 "\n\tnone - " ICE_FEC_STRING_NONE			\
2992 "\nEither of the left or right strings above can be used to set the requested mode."
2993 
2994 /**
2995  * ice_sysctl_fec_config - Display/change the configured FEC mode
2996  * @oidp: sysctl oid structure
2997  * @arg1: pointer to private data structure
2998  * @arg2: unused
2999  * @req: sysctl request pointer
3000  *
3001  * On read: Displays the configured FEC mode
3002  * On write: Sets the device's FEC mode to the input string, if it's valid.
3003  * Valid input strings: see ICE_SYSCTL_HELP_FEC_CONFIG
3004  */
3005 static int
3006 ice_sysctl_fec_config(SYSCTL_HANDLER_ARGS)
3007 {
3008 	struct ice_softc *sc = (struct ice_softc *)arg1;
3009 	struct ice_port_info *pi = sc->hw.port_info;
3010 	enum ice_fec_mode new_mode;
3011 	device_t dev = sc->dev;
3012 	char req_fec[32];
3013 	int ret;
3014 
3015 	UNREFERENCED_PARAMETER(arg2);
3016 
3017 	if (ice_driver_is_detaching(sc))
3018 		return (ESHUTDOWN);
3019 
3020 	bzero(req_fec, sizeof(req_fec));
3021 	strlcpy(req_fec, ice_requested_fec_mode(pi), sizeof(req_fec));
3022 
3023 	ret = sysctl_handle_string(oidp, req_fec, sizeof(req_fec), req);
3024 	if ((ret) || (req->newptr == NULL))
3025 		return (ret);
3026 
3027 	if (strcmp(req_fec, "auto") == 0 ||
3028 	    strcmp(req_fec, ice_fec_str(ICE_FEC_AUTO)) == 0) {
3029 		new_mode = ICE_FEC_AUTO;
3030 	} else if (strcmp(req_fec, "fc") == 0 ||
3031 	    strcmp(req_fec, ice_fec_str(ICE_FEC_BASER)) == 0) {
3032 		new_mode = ICE_FEC_BASER;
3033 	} else if (strcmp(req_fec, "rs") == 0 ||
3034 	    strcmp(req_fec, ice_fec_str(ICE_FEC_RS)) == 0) {
3035 		new_mode = ICE_FEC_RS;
3036 	} else if (strcmp(req_fec, "none") == 0 ||
3037 	    strcmp(req_fec, ice_fec_str(ICE_FEC_NONE)) == 0) {
3038 		new_mode = ICE_FEC_NONE;
3039 	} else {
3040 		device_printf(dev,
3041 		    "%s: \"%s\" is not a valid FEC mode\n",
3042 		    __func__, req_fec);
3043 		return (EINVAL);
3044 	}
3045 
3046 	/* Cache user FEC mode for later link ups */
3047 	pi->phy.curr_user_fec_req = new_mode;
3048 
3049 	/* Apply settings requested by user */
3050 	return ice_apply_saved_phy_cfg(sc, ICE_APPLY_FEC);
3051 }
3052 
3053 /**
3054  * ice_sysctl_negotiated_fec - Display the negotiated FEC mode on the link
3055  * @oidp: sysctl oid structure
3056  * @arg1: pointer to private data structure
3057  * @arg2: unused
3058  * @req: sysctl request pointer
3059  *
3060  * On read: Displays the negotiated FEC mode, in a string
3061  */
3062 static int
3063 ice_sysctl_negotiated_fec(SYSCTL_HANDLER_ARGS)
3064 {
3065 	struct ice_softc *sc = (struct ice_softc *)arg1;
3066 	struct ice_hw *hw = &sc->hw;
3067 	char neg_fec[32];
3068 	int ret;
3069 
3070 	UNREFERENCED_PARAMETER(arg2);
3071 
3072 	if (ice_driver_is_detaching(sc))
3073 		return (ESHUTDOWN);
3074 
3075 	/* Copy const string into a buffer to drop const qualifier */
3076 	bzero(neg_fec, sizeof(neg_fec));
3077 	strlcpy(neg_fec, ice_negotiated_fec_mode(hw->port_info), sizeof(neg_fec));
3078 
3079 	ret = sysctl_handle_string(oidp, neg_fec, 0, req);
3080 	if (req->newptr != NULL)
3081 		return (EPERM);
3082 
3083 	return (ret);
3084 }
3085 
3086 #define ICE_SYSCTL_HELP_FC_CONFIG				\
3087 "\nDisplay or set the port's advertised flow control mode.\n"	\
3088 "\t0 - " ICE_FC_STRING_NONE					\
3089 "\n\t1 - " ICE_FC_STRING_RX					\
3090 "\n\t2 - " ICE_FC_STRING_TX					\
3091 "\n\t3 - " ICE_FC_STRING_FULL					\
3092 "\nEither the numbers or the strings above can be used to set the advertised mode."
3093 
3094 /**
3095  * ice_sysctl_fc_config - Display/change the advertised flow control mode
3096  * @oidp: sysctl oid structure
3097  * @arg1: pointer to private data structure
3098  * @arg2: unused
3099  * @req: sysctl request pointer
3100  *
3101  * On read: Displays the configured flow control mode
3102  * On write: Sets the device's flow control mode to the input, if it's valid.
3103  * Valid input strings: see ICE_SYSCTL_HELP_FC_CONFIG
3104  */
3105 static int
3106 ice_sysctl_fc_config(SYSCTL_HANDLER_ARGS)
3107 {
3108 	struct ice_softc *sc = (struct ice_softc *)arg1;
3109 	struct ice_port_info *pi = sc->hw.port_info;
3110 	struct ice_aqc_get_phy_caps_data pcaps = { 0 };
3111 	enum ice_fc_mode old_mode, new_mode;
3112 	struct ice_hw *hw = &sc->hw;
3113 	device_t dev = sc->dev;
3114 	enum ice_status status;
3115 	int ret, fc_num;
3116 	bool mode_set = false;
3117 	struct sbuf buf;
3118 	char *fc_str_end;
3119 	char fc_str[32];
3120 
3121 	UNREFERENCED_PARAMETER(arg2);
3122 
3123 	if (ice_driver_is_detaching(sc))
3124 		return (ESHUTDOWN);
3125 
3126 	status = ice_aq_get_phy_caps(pi, false, ICE_AQC_REPORT_ACTIVE_CFG,
3127 				     &pcaps, NULL);
3128 	if (status != ICE_SUCCESS) {
3129 		device_printf(dev,
3130 		    "%s: ice_aq_get_phy_caps failed; status %s, aq_err %s\n",
3131 		    __func__, ice_status_str(status),
3132 		    ice_aq_str(hw->adminq.sq_last_status));
3133 		return (EIO);
3134 	}
3135 
3136 	/* Convert HW response format to SW enum value */
3137 	if ((pcaps.caps & ICE_AQC_PHY_EN_TX_LINK_PAUSE) &&
3138 	    (pcaps.caps & ICE_AQC_PHY_EN_RX_LINK_PAUSE))
3139 		old_mode = ICE_FC_FULL;
3140 	else if (pcaps.caps & ICE_AQC_PHY_EN_TX_LINK_PAUSE)
3141 		old_mode = ICE_FC_TX_PAUSE;
3142 	else if (pcaps.caps & ICE_AQC_PHY_EN_RX_LINK_PAUSE)
3143 		old_mode = ICE_FC_RX_PAUSE;
3144 	else
3145 		old_mode = ICE_FC_NONE;
3146 
3147 	/* Create "old" string for output */
3148 	bzero(fc_str, sizeof(fc_str));
3149 	sbuf_new_for_sysctl(&buf, fc_str, sizeof(fc_str), req);
3150 	sbuf_printf(&buf, "%d<%s>", old_mode, ice_fc_str(old_mode));
3151 	sbuf_finish(&buf);
3152 	sbuf_delete(&buf);
3153 
3154 	ret = sysctl_handle_string(oidp, fc_str, sizeof(fc_str), req);
3155 	if ((ret) || (req->newptr == NULL))
3156 		return (ret);
3157 
3158 	/* Try to parse input as a string, first */
3159 	if (strcasecmp(ice_fc_str(ICE_FC_FULL), fc_str) == 0) {
3160 		new_mode = ICE_FC_FULL;
3161 		mode_set = true;
3162 	}
3163 	else if (strcasecmp(ice_fc_str(ICE_FC_TX_PAUSE), fc_str) == 0) {
3164 		new_mode = ICE_FC_TX_PAUSE;
3165 		mode_set = true;
3166 	}
3167 	else if (strcasecmp(ice_fc_str(ICE_FC_RX_PAUSE), fc_str) == 0) {
3168 		new_mode = ICE_FC_RX_PAUSE;
3169 		mode_set = true;
3170 	}
3171 	else if (strcasecmp(ice_fc_str(ICE_FC_NONE), fc_str) == 0) {
3172 		new_mode = ICE_FC_NONE;
3173 		mode_set = true;
3174 	}
3175 
3176 	/*
3177 	 * Then check if it's an integer, for compatibility with the method
3178 	 * used in older drivers.
3179 	 */
3180 	if (!mode_set) {
3181 		fc_num = strtol(fc_str, &fc_str_end, 0);
3182 		if (fc_str_end == fc_str)
3183 			fc_num = -1;
3184 		switch (fc_num) {
3185 		case 3:
3186 			new_mode = ICE_FC_FULL;
3187 			break;
3188 		case 2:
3189 			new_mode = ICE_FC_TX_PAUSE;
3190 			break;
3191 		case 1:
3192 			new_mode = ICE_FC_RX_PAUSE;
3193 			break;
3194 		case 0:
3195 			new_mode = ICE_FC_NONE;
3196 			break;
3197 		default:
3198 			device_printf(dev,
3199 			    "%s: \"%s\" is not a valid flow control mode\n",
3200 			    __func__, fc_str);
3201 			return (EINVAL);
3202 		}
3203 	}
3204 
3205 	/* Set the flow control mode in FW */
3206 	pi->phy.curr_user_fc_req = new_mode;
3207 
3208 	/* Apply settings requested by user */
3209 	return ice_apply_saved_phy_cfg(sc, ICE_APPLY_FC);
3210 }
3211 
3212 /**
3213  * ice_sysctl_negotiated_fc - Display currently negotiated FC mode
3214  * @oidp: sysctl oid structure
3215  * @arg1: pointer to private data structure
3216  * @arg2: unused
3217  * @req: sysctl request pointer
3218  *
3219  * On read: Displays the currently negotiated flow control settings.
3220  *
3221  * If link is not established, this will report ICE_FC_NONE, as no flow
3222  * control is negotiated while link is down.
3223  */
3224 static int
3225 ice_sysctl_negotiated_fc(SYSCTL_HANDLER_ARGS)
3226 {
3227 	struct ice_softc *sc = (struct ice_softc *)arg1;
3228 	struct ice_port_info *pi = sc->hw.port_info;
3229 	const char *negotiated_fc;
3230 
3231 	UNREFERENCED_PARAMETER(arg2);
3232 
3233 	if (ice_driver_is_detaching(sc))
3234 		return (ESHUTDOWN);
3235 
3236 	negotiated_fc = ice_flowcontrol_mode(pi);
3237 
3238 	return sysctl_handle_string(oidp, __DECONST(char *, negotiated_fc), 0, req);
3239 }
3240 
3241 /**
3242  * __ice_sysctl_phy_type_handler - Display/change supported PHY types/speeds
3243  * @oidp: sysctl oid structure
3244  * @arg1: pointer to private data structure
3245  * @arg2: unused
3246  * @req: sysctl request pointer
3247  * @is_phy_type_high: if true, handle the high PHY type instead of the low PHY type
3248  *
3249  * Private handler for phy_type_high and phy_type_low sysctls.
3250  */
3251 static int
3252 __ice_sysctl_phy_type_handler(SYSCTL_HANDLER_ARGS, bool is_phy_type_high)
3253 {
3254 	struct ice_softc *sc = (struct ice_softc *)arg1;
3255 	struct ice_aqc_get_phy_caps_data pcaps = { 0 };
3256 	struct ice_aqc_set_phy_cfg_data cfg = { 0 };
3257 	struct ice_hw *hw = &sc->hw;
3258 	device_t dev = sc->dev;
3259 	enum ice_status status;
3260 	uint64_t types;
3261 	int ret;
3262 
3263 	UNREFERENCED_PARAMETER(arg2);
3264 
3265 	if (ice_driver_is_detaching(sc))
3266 		return (ESHUTDOWN);
3267 
3268 	status = ice_aq_get_phy_caps(hw->port_info, false, ICE_AQC_REPORT_ACTIVE_CFG,
3269 				     &pcaps, NULL);
3270 	if (status != ICE_SUCCESS) {
3271 		device_printf(dev,
3272 		    "%s: ice_aq_get_phy_caps failed; status %s, aq_err %s\n",
3273 		    __func__, ice_status_str(status),
3274 		    ice_aq_str(hw->adminq.sq_last_status));
3275 		return (EIO);
3276 	}
3277 
3278 	if (is_phy_type_high)
3279 		types = pcaps.phy_type_high;
3280 	else
3281 		types = pcaps.phy_type_low;
3282 
3283 	ret = sysctl_handle_64(oidp, &types, sizeof(types), req);
3284 	if ((ret) || (req->newptr == NULL))
3285 		return (ret);
3286 
3287 	ice_copy_phy_caps_to_cfg(hw->port_info, &pcaps, &cfg);
3288 
3289 	if (is_phy_type_high)
3290 		cfg.phy_type_high = types & hw->port_info->phy.phy_type_high;
3291 	else
3292 		cfg.phy_type_low = types & hw->port_info->phy.phy_type_low;
3293 	cfg.caps |= ICE_AQ_PHY_ENA_AUTO_LINK_UPDT;
3294 
3295 	status = ice_aq_set_phy_cfg(hw, hw->port_info, &cfg, NULL);
3296 	if (status != ICE_SUCCESS) {
3297 		device_printf(dev,
3298 		    "%s: ice_aq_set_phy_cfg failed; status %s, aq_err %s\n",
3299 		    __func__, ice_status_str(status),
3300 		    ice_aq_str(hw->adminq.sq_last_status));
3301 		return (EIO);
3302 	}
3303 
3304 	return (0);
3305 
3306 }
3307 
3308 /**
3309  * ice_sysctl_phy_type_low - Display/change supported lower PHY types/speeds
3310  * @oidp: sysctl oid structure
3311  * @arg1: pointer to private data structure
3312  * @arg2: unused
3313  * @req: sysctl request pointer
3314  *
3315  * On read: Displays the currently supported lower PHY types
3316  * On write: Sets the device's supported low PHY types
3317  */
3318 static int
3319 ice_sysctl_phy_type_low(SYSCTL_HANDLER_ARGS)
3320 {
3321 	return __ice_sysctl_phy_type_handler(oidp, arg1, arg2, req, false);
3322 }
3323 
3324 /**
3325  * ice_sysctl_phy_type_high - Display/change supported higher PHY types/speeds
3326  * @oidp: sysctl oid structure
3327  * @arg1: pointer to private data structure
3328  * @arg2: unused
3329  * @req: sysctl request pointer
3330  *
3331  * On read: Displays the currently supported higher PHY types
3332  * On write: Sets the device's supported high PHY types
3333  */
3334 static int
3335 ice_sysctl_phy_type_high(SYSCTL_HANDLER_ARGS)
3336 {
3337 	return __ice_sysctl_phy_type_handler(oidp, arg1, arg2, req, true);
3338 }
3339 
3340 /**
3341  * ice_sysctl_phy_caps - Display response from Get PHY abililties
3342  * @oidp: sysctl oid structure
3343  * @arg1: pointer to private data structure
3344  * @arg2: unused
3345  * @req: sysctl request pointer
3346  * @report_mode: the mode to report
3347  *
3348  * On read: Display the response from Get PHY abillities with the given report
3349  * mode.
3350  */
3351 static int
3352 ice_sysctl_phy_caps(SYSCTL_HANDLER_ARGS, u8 report_mode)
3353 {
3354 	struct ice_softc *sc = (struct ice_softc *)arg1;
3355 	struct ice_aqc_get_phy_caps_data pcaps = { 0 };
3356 	struct ice_hw *hw = &sc->hw;
3357 	struct ice_port_info *pi = hw->port_info;
3358 	device_t dev = sc->dev;
3359 	enum ice_status status;
3360 	int ret;
3361 
3362 	UNREFERENCED_PARAMETER(arg2);
3363 
3364 	ret = priv_check(curthread, PRIV_DRIVER);
3365 	if (ret)
3366 		return (ret);
3367 
3368 	if (ice_driver_is_detaching(sc))
3369 		return (ESHUTDOWN);
3370 
3371 	status = ice_aq_get_phy_caps(pi, true, report_mode, &pcaps, NULL);
3372 	if (status != ICE_SUCCESS) {
3373 		device_printf(dev,
3374 		    "%s: ice_aq_get_phy_caps failed; status %s, aq_err %s\n",
3375 		    __func__, ice_status_str(status),
3376 		    ice_aq_str(hw->adminq.sq_last_status));
3377 		return (EIO);
3378 	}
3379 
3380 	ret = sysctl_handle_opaque(oidp, &pcaps, sizeof(pcaps), req);
3381 	if (req->newptr != NULL)
3382 		return (EPERM);
3383 
3384 	return (ret);
3385 }
3386 
3387 /**
3388  * ice_sysctl_phy_sw_caps - Display response from Get PHY abililties
3389  * @oidp: sysctl oid structure
3390  * @arg1: pointer to private data structure
3391  * @arg2: unused
3392  * @req: sysctl request pointer
3393  *
3394  * On read: Display the response from Get PHY abillities reporting the last
3395  * software configuration.
3396  */
3397 static int
3398 ice_sysctl_phy_sw_caps(SYSCTL_HANDLER_ARGS)
3399 {
3400 	return ice_sysctl_phy_caps(oidp, arg1, arg2, req,
3401 				   ICE_AQC_REPORT_ACTIVE_CFG);
3402 }
3403 
3404 /**
3405  * ice_sysctl_phy_nvm_caps - Display response from Get PHY abililties
3406  * @oidp: sysctl oid structure
3407  * @arg1: pointer to private data structure
3408  * @arg2: unused
3409  * @req: sysctl request pointer
3410  *
3411  * On read: Display the response from Get PHY abillities reporting the NVM
3412  * configuration.
3413  */
3414 static int
3415 ice_sysctl_phy_nvm_caps(SYSCTL_HANDLER_ARGS)
3416 {
3417 	return ice_sysctl_phy_caps(oidp, arg1, arg2, req,
3418 				   ICE_AQC_REPORT_TOPO_CAP_NO_MEDIA);
3419 }
3420 
3421 /**
3422  * ice_sysctl_phy_topo_caps - Display response from Get PHY abililties
3423  * @oidp: sysctl oid structure
3424  * @arg1: pointer to private data structure
3425  * @arg2: unused
3426  * @req: sysctl request pointer
3427  *
3428  * On read: Display the response from Get PHY abillities reporting the
3429  * topology configuration.
3430  */
3431 static int
3432 ice_sysctl_phy_topo_caps(SYSCTL_HANDLER_ARGS)
3433 {
3434 	return ice_sysctl_phy_caps(oidp, arg1, arg2, req,
3435 				   ICE_AQC_REPORT_TOPO_CAP_MEDIA);
3436 }
3437 
3438 /**
3439  * ice_sysctl_phy_link_status - Display response from Get Link Status
3440  * @oidp: sysctl oid structure
3441  * @arg1: pointer to private data structure
3442  * @arg2: unused
3443  * @req: sysctl request pointer
3444  *
3445  * On read: Display the response from firmware for the Get Link Status
3446  * request.
3447  */
3448 static int
3449 ice_sysctl_phy_link_status(SYSCTL_HANDLER_ARGS)
3450 {
3451 	struct ice_aqc_get_link_status_data link_data = { 0 };
3452 	struct ice_softc *sc = (struct ice_softc *)arg1;
3453 	struct ice_hw *hw = &sc->hw;
3454 	struct ice_port_info *pi = hw->port_info;
3455 	struct ice_aqc_get_link_status *resp;
3456 	struct ice_aq_desc desc;
3457 	device_t dev = sc->dev;
3458 	enum ice_status status;
3459 	int ret;
3460 
3461 	UNREFERENCED_PARAMETER(arg2);
3462 
3463 	/*
3464 	 * Ensure that only contexts with driver privilege are allowed to
3465 	 * access this information
3466 	 */
3467 	ret = priv_check(curthread, PRIV_DRIVER);
3468 	if (ret)
3469 		return (ret);
3470 
3471 	if (ice_driver_is_detaching(sc))
3472 		return (ESHUTDOWN);
3473 
3474 	ice_fill_dflt_direct_cmd_desc(&desc, ice_aqc_opc_get_link_status);
3475 	resp = &desc.params.get_link_status;
3476 	resp->lport_num = pi->lport;
3477 
3478 	status = ice_aq_send_cmd(hw, &desc, &link_data, sizeof(link_data), NULL);
3479 	if (status != ICE_SUCCESS) {
3480 		device_printf(dev,
3481 		    "%s: ice_aq_send_cmd failed; status %s, aq_err %s\n",
3482 		    __func__, ice_status_str(status),
3483 		    ice_aq_str(hw->adminq.sq_last_status));
3484 		return (EIO);
3485 	}
3486 
3487 	ret = sysctl_handle_opaque(oidp, &link_data, sizeof(link_data), req);
3488 	if (req->newptr != NULL)
3489 		return (EPERM);
3490 
3491 	return (ret);
3492 }
3493 
3494 /**
3495  * ice_sysctl_fw_cur_lldp_persist_status - Display current FW LLDP status
3496  * @oidp: sysctl oid structure
3497  * @arg1: pointer to private softc structure
3498  * @arg2: unused
3499  * @req: sysctl request pointer
3500  *
3501  * On read: Displays current persistent LLDP status.
3502  */
3503 static int
3504 ice_sysctl_fw_cur_lldp_persist_status(SYSCTL_HANDLER_ARGS)
3505 {
3506 	struct ice_softc *sc = (struct ice_softc *)arg1;
3507 	struct ice_hw *hw = &sc->hw;
3508 	device_t dev = sc->dev;
3509 	enum ice_status status;
3510 	struct sbuf *sbuf;
3511 	u32 lldp_state;
3512 
3513 	UNREFERENCED_PARAMETER(arg2);
3514 	UNREFERENCED_PARAMETER(oidp);
3515 
3516 	if (ice_driver_is_detaching(sc))
3517 		return (ESHUTDOWN);
3518 
3519 	status = ice_get_cur_lldp_persist_status(hw, &lldp_state);
3520 	if (status) {
3521 		device_printf(dev,
3522 		    "Could not acquire current LLDP persistence status, err %s aq_err %s\n",
3523 		    ice_status_str(status), ice_aq_str(hw->adminq.sq_last_status));
3524 		return (EIO);
3525 	}
3526 
3527 	sbuf = sbuf_new_for_sysctl(NULL, NULL, 128, req);
3528 	sbuf_printf(sbuf, "%s", ice_fw_lldp_status(lldp_state));
3529 	sbuf_finish(sbuf);
3530 	sbuf_delete(sbuf);
3531 
3532 	return (0);
3533 }
3534 
3535 /**
3536  * ice_sysctl_fw_dflt_lldp_persist_status - Display default FW LLDP status
3537  * @oidp: sysctl oid structure
3538  * @arg1: pointer to private softc structure
3539  * @arg2: unused
3540  * @req: sysctl request pointer
3541  *
3542  * On read: Displays default persistent LLDP status.
3543  */
3544 static int
3545 ice_sysctl_fw_dflt_lldp_persist_status(SYSCTL_HANDLER_ARGS)
3546 {
3547 	struct ice_softc *sc = (struct ice_softc *)arg1;
3548 	struct ice_hw *hw = &sc->hw;
3549 	device_t dev = sc->dev;
3550 	enum ice_status status;
3551 	struct sbuf *sbuf;
3552 	u32 lldp_state;
3553 
3554 	UNREFERENCED_PARAMETER(arg2);
3555 	UNREFERENCED_PARAMETER(oidp);
3556 
3557 	if (ice_driver_is_detaching(sc))
3558 		return (ESHUTDOWN);
3559 
3560 	status = ice_get_dflt_lldp_persist_status(hw, &lldp_state);
3561 	if (status) {
3562 		device_printf(dev,
3563 		    "Could not acquire default LLDP persistence status, err %s aq_err %s\n",
3564 		    ice_status_str(status), ice_aq_str(hw->adminq.sq_last_status));
3565 		return (EIO);
3566 	}
3567 
3568 	sbuf = sbuf_new_for_sysctl(NULL, NULL, 128, req);
3569 	sbuf_printf(sbuf, "%s", ice_fw_lldp_status(lldp_state));
3570 	sbuf_finish(sbuf);
3571 	sbuf_delete(sbuf);
3572 
3573 	return (0);
3574 }
3575 
3576 #define ICE_SYSCTL_HELP_FW_LLDP_AGENT	\
3577 "\nDisplay or change FW LLDP agent state:" \
3578 "\n\t0 - disabled"			\
3579 "\n\t1 - enabled"
3580 
3581 /**
3582  * ice_sysctl_fw_lldp_agent - Display or change the FW LLDP agent status
3583  * @oidp: sysctl oid structure
3584  * @arg1: pointer to private softc structure
3585  * @arg2: unused
3586  * @req: sysctl request pointer
3587  *
3588  * On read: Displays whether the FW LLDP agent is running
3589  * On write: Persistently enables or disables the FW LLDP agent
3590  */
3591 static int
3592 ice_sysctl_fw_lldp_agent(SYSCTL_HANDLER_ARGS)
3593 {
3594 	struct ice_softc *sc = (struct ice_softc *)arg1;
3595 	struct ice_hw *hw = &sc->hw;
3596 	device_t dev = sc->dev;
3597 	enum ice_status status;
3598 	int ret;
3599 	u32 old_state;
3600 	u8 fw_lldp_enabled;
3601 	bool retried_start_lldp = false;
3602 
3603 	UNREFERENCED_PARAMETER(arg2);
3604 
3605 	if (ice_driver_is_detaching(sc))
3606 		return (ESHUTDOWN);
3607 
3608 	status = ice_get_cur_lldp_persist_status(hw, &old_state);
3609 	if (status) {
3610 		device_printf(dev,
3611 		    "Could not acquire current LLDP persistence status, err %s aq_err %s\n",
3612 		    ice_status_str(status), ice_aq_str(hw->adminq.sq_last_status));
3613 		return (EIO);
3614 	}
3615 
3616 	if (old_state > ICE_LLDP_ADMINSTATUS_ENA_RXTX) {
3617 		status = ice_get_dflt_lldp_persist_status(hw, &old_state);
3618 		if (status) {
3619 			device_printf(dev,
3620 			    "Could not acquire default LLDP persistence status, err %s aq_err %s\n",
3621 			    ice_status_str(status),
3622 			    ice_aq_str(hw->adminq.sq_last_status));
3623 			return (EIO);
3624 		}
3625 	}
3626 	if (old_state == 0)
3627 		fw_lldp_enabled = false;
3628 	else
3629 		fw_lldp_enabled = true;
3630 
3631 	ret = sysctl_handle_bool(oidp, &fw_lldp_enabled, 0, req);
3632 	if ((ret) || (req->newptr == NULL))
3633 		return (ret);
3634 
3635 	if (old_state == 0 && fw_lldp_enabled == false)
3636 		return (0);
3637 
3638 	if (old_state != 0 && fw_lldp_enabled == true)
3639 		return (0);
3640 
3641 	if (fw_lldp_enabled == false) {
3642 		status = ice_aq_stop_lldp(hw, true, true, NULL);
3643 		/* EPERM is returned if the LLDP agent is already shutdown */
3644 		if (status && hw->adminq.sq_last_status != ICE_AQ_RC_EPERM) {
3645 			device_printf(dev,
3646 			    "%s: ice_aq_stop_lldp failed; status %s, aq_err %s\n",
3647 			    __func__, ice_status_str(status),
3648 			    ice_aq_str(hw->adminq.sq_last_status));
3649 			return (EIO);
3650 		}
3651 		ice_aq_set_dcb_parameters(hw, true, NULL);
3652 		hw->port_info->qos_cfg.is_sw_lldp = true;
3653 		ice_add_rx_lldp_filter(sc);
3654 	} else {
3655 		ice_del_rx_lldp_filter(sc);
3656 retry_start_lldp:
3657 		status = ice_aq_start_lldp(hw, true, NULL);
3658 		if (status) {
3659 			switch (hw->adminq.sq_last_status) {
3660 			/* EEXIST is returned if the LLDP agent is already started */
3661 			case ICE_AQ_RC_EEXIST:
3662 				break;
3663 			case ICE_AQ_RC_EAGAIN:
3664 				/* Retry command after a 2 second wait */
3665 				if (retried_start_lldp == false) {
3666 					retried_start_lldp = true;
3667 					pause("slldp", ICE_START_LLDP_RETRY_WAIT);
3668 					goto retry_start_lldp;
3669 				}
3670 				/* Fallthrough */
3671 			default:
3672 				device_printf(dev,
3673 				    "%s: ice_aq_start_lldp failed; status %s, aq_err %s\n",
3674 				    __func__, ice_status_str(status),
3675 				    ice_aq_str(hw->adminq.sq_last_status));
3676 				return (EIO);
3677 			}
3678 		}
3679 		hw->port_info->qos_cfg.is_sw_lldp = false;
3680 	}
3681 
3682 	return (ret);
3683 }
3684 
3685 /**
3686  * ice_add_device_sysctls - add device specific dynamic sysctls
3687  * @sc: device private structure
3688  *
3689  * Add per-device dynamic sysctls which show device configuration or enable
3690  * configuring device functionality. For tunable values which can be set prior
3691  * to load, see ice_add_device_tunables.
3692  *
3693  * This function depends on the sysctl layout setup by ice_add_device_tunables,
3694  * and likely should be called near the end of the attach process.
3695  */
3696 void
3697 ice_add_device_sysctls(struct ice_softc *sc)
3698 {
3699 	struct sysctl_oid *hw_node;
3700 	device_t dev = sc->dev;
3701 
3702 	struct sysctl_ctx_list *ctx = device_get_sysctl_ctx(dev);
3703 	struct sysctl_oid_list *ctx_list =
3704 	    SYSCTL_CHILDREN(device_get_sysctl_tree(dev));
3705 
3706 	SYSCTL_ADD_PROC(ctx, ctx_list,
3707 	    OID_AUTO, "fw_version", CTLTYPE_STRING | CTLFLAG_RD,
3708 	    sc, 0, ice_sysctl_show_fw, "A", "Firmware version");
3709 
3710 	SYSCTL_ADD_PROC(ctx, ctx_list,
3711 	    OID_AUTO, "pba_number", CTLTYPE_STRING | CTLFLAG_RD,
3712 	    sc, 0, ice_sysctl_pba_number, "A", "Product Board Assembly Number");
3713 
3714 	SYSCTL_ADD_PROC(ctx, ctx_list,
3715 	    OID_AUTO, "ddp_version", CTLTYPE_STRING | CTLFLAG_RD,
3716 	    sc, 0, ice_sysctl_pkg_version, "A", "Active DDP package name and version");
3717 
3718 	SYSCTL_ADD_PROC(ctx, ctx_list,
3719 	    OID_AUTO, "current_speed", CTLTYPE_STRING | CTLFLAG_RD,
3720 	    sc, 0, ice_sysctl_current_speed, "A", "Current Port Link Speed");
3721 
3722 	SYSCTL_ADD_PROC(ctx, ctx_list,
3723 	    OID_AUTO, "requested_fec", CTLTYPE_STRING | CTLFLAG_RW,
3724 	    sc, 0, ice_sysctl_fec_config, "A", ICE_SYSCTL_HELP_FEC_CONFIG);
3725 
3726 	SYSCTL_ADD_PROC(ctx, ctx_list,
3727 	    OID_AUTO, "negotiated_fec", CTLTYPE_STRING | CTLFLAG_RD,
3728 	    sc, 0, ice_sysctl_negotiated_fec, "A", "Current Negotiated FEC mode");
3729 
3730 	SYSCTL_ADD_PROC(ctx, ctx_list,
3731 	    OID_AUTO, "fc", CTLTYPE_STRING | CTLFLAG_RW,
3732 	    sc, 0, ice_sysctl_fc_config, "A", ICE_SYSCTL_HELP_FC_CONFIG);
3733 
3734 	SYSCTL_ADD_PROC(ctx, ctx_list,
3735 	    OID_AUTO, "advertise_speed", CTLTYPE_U16 | CTLFLAG_RW,
3736 	    sc, 0, ice_sysctl_advertise_speed, "SU", ICE_SYSCTL_HELP_ADVERTISE_SPEED);
3737 
3738 	SYSCTL_ADD_PROC(ctx, ctx_list,
3739 	    OID_AUTO, "fw_lldp_agent", CTLTYPE_U8 | CTLFLAG_RWTUN,
3740 	    sc, 0, ice_sysctl_fw_lldp_agent, "CU", ICE_SYSCTL_HELP_FW_LLDP_AGENT);
3741 
3742 	/* Differentiate software and hardware statistics, by keeping hw stats
3743 	 * in their own node. This isn't in ice_add_device_tunables, because
3744 	 * we won't have any CTLFLAG_TUN sysctls under this node.
3745 	 */
3746 	hw_node = SYSCTL_ADD_NODE(ctx, ctx_list, OID_AUTO, "hw", CTLFLAG_RD,
3747 				  NULL, "Port Hardware Statistics");
3748 
3749 	ice_add_sysctls_mac_stats(ctx, hw_node, &sc->stats.cur);
3750 
3751 	/* Add the main PF VSI stats now. Other VSIs will add their own stats
3752 	 * during creation
3753 	 */
3754 	ice_add_vsi_sysctls(&sc->pf_vsi);
3755 
3756 	/* Add sysctls related to debugging the device driver. This includes
3757 	 * sysctls which display additional internal driver state for use in
3758 	 * understanding what is happening within the driver.
3759 	 */
3760 	ice_add_debug_sysctls(sc);
3761 }
3762 
3763 /**
3764  * @enum hmc_error_type
3765  * @brief enumeration of HMC errors
3766  *
3767  * Enumeration defining the possible HMC errors that might occur.
3768  */
3769 enum hmc_error_type {
3770 	HMC_ERR_PMF_INVALID = 0,
3771 	HMC_ERR_VF_IDX_INVALID = 1,
3772 	HMC_ERR_VF_PARENT_PF_INVALID = 2,
3773 	/* 3 is reserved */
3774 	HMC_ERR_INDEX_TOO_BIG = 4,
3775 	HMC_ERR_ADDRESS_TOO_LARGE = 5,
3776 	HMC_ERR_SEGMENT_DESC_INVALID = 6,
3777 	HMC_ERR_SEGMENT_DESC_TOO_SMALL = 7,
3778 	HMC_ERR_PAGE_DESC_INVALID = 8,
3779 	HMC_ERR_UNSUPPORTED_REQUEST_COMPLETION = 9,
3780 	/* 10 is reserved */
3781 	HMC_ERR_INVALID_OBJECT_TYPE = 11,
3782 	/* 12 is reserved */
3783 };
3784 
3785 /**
3786  * ice_log_hmc_error - Log an HMC error message
3787  * @hw: device hw structure
3788  * @dev: the device to pass to device_printf()
3789  *
3790  * Log a message when an HMC error interrupt is triggered.
3791  */
3792 void
3793 ice_log_hmc_error(struct ice_hw *hw, device_t dev)
3794 {
3795 	u32 info, data;
3796 	u8 index, errtype, objtype;
3797 	bool isvf;
3798 
3799 	info = rd32(hw, PFHMC_ERRORINFO);
3800 	data = rd32(hw, PFHMC_ERRORDATA);
3801 
3802 	index = (u8)(info & PFHMC_ERRORINFO_PMF_INDEX_M);
3803 	errtype = (u8)((info & PFHMC_ERRORINFO_HMC_ERROR_TYPE_M) >>
3804 		       PFHMC_ERRORINFO_HMC_ERROR_TYPE_S);
3805 	objtype = (u8)((info & PFHMC_ERRORINFO_HMC_OBJECT_TYPE_M) >>
3806 		       PFHMC_ERRORINFO_HMC_OBJECT_TYPE_S);
3807 
3808 	isvf = info & PFHMC_ERRORINFO_PMF_ISVF_M;
3809 
3810 	device_printf(dev, "%s HMC Error detected on PMF index %d:\n",
3811 		      isvf ? "VF" : "PF", index);
3812 
3813 	device_printf(dev, "error type %d, object type %d, data 0x%08x\n",
3814 		      errtype, objtype, data);
3815 
3816 	switch (errtype) {
3817 	case HMC_ERR_PMF_INVALID:
3818 		device_printf(dev, "Private Memory Function is not valid\n");
3819 		break;
3820 	case HMC_ERR_VF_IDX_INVALID:
3821 		device_printf(dev, "Invalid Private Memory Function index for PE enabled VF\n");
3822 		break;
3823 	case HMC_ERR_VF_PARENT_PF_INVALID:
3824 		device_printf(dev, "Invalid parent PF for PE enabled VF\n");
3825 		break;
3826 	case HMC_ERR_INDEX_TOO_BIG:
3827 		device_printf(dev, "Object index too big\n");
3828 		break;
3829 	case HMC_ERR_ADDRESS_TOO_LARGE:
3830 		device_printf(dev, "Address extends beyond segment descriptor limit\n");
3831 		break;
3832 	case HMC_ERR_SEGMENT_DESC_INVALID:
3833 		device_printf(dev, "Segment descriptor is invalid\n");
3834 		break;
3835 	case HMC_ERR_SEGMENT_DESC_TOO_SMALL:
3836 		device_printf(dev, "Segment descriptor is too small\n");
3837 		break;
3838 	case HMC_ERR_PAGE_DESC_INVALID:
3839 		device_printf(dev, "Page descriptor is invalid\n");
3840 		break;
3841 	case HMC_ERR_UNSUPPORTED_REQUEST_COMPLETION:
3842 		device_printf(dev, "Unsupported Request completion received from PCIe\n");
3843 		break;
3844 	case HMC_ERR_INVALID_OBJECT_TYPE:
3845 		device_printf(dev, "Invalid object type\n");
3846 		break;
3847 	default:
3848 		device_printf(dev, "Unknown HMC error\n");
3849 	}
3850 
3851 	/* Clear the error indication */
3852 	wr32(hw, PFHMC_ERRORINFO, 0);
3853 }
3854 
3855 /**
3856  * @struct ice_sysctl_info
3857  * @brief sysctl information
3858  *
3859  * Structure used to simplify the process of defining the many similar
3860  * statistics sysctls.
3861  */
3862 struct ice_sysctl_info {
3863 	u64		*stat;
3864 	const char	*name;
3865 	const char	*description;
3866 };
3867 
3868 /**
3869  * ice_add_sysctls_eth_stats - Add sysctls for ethernet statistics
3870  * @ctx: sysctl ctx to use
3871  * @parent: the parent node to add sysctls under
3872  * @stats: the ethernet stats structure to source values from
3873  *
3874  * Adds statistics sysctls for the ethernet statistics of the MAC or a VSI.
3875  * Will add them under the parent node specified.
3876  *
3877  * Note that tx_errors is only meaningful for VSIs and not the global MAC/PF
3878  * statistics, so it is not included here. Similarly, rx_discards has different
3879  * descriptions for VSIs and MAC/PF stats, so it is also not included here.
3880  */
3881 void
3882 ice_add_sysctls_eth_stats(struct sysctl_ctx_list *ctx,
3883 			  struct sysctl_oid *parent,
3884 			  struct ice_eth_stats *stats)
3885 {
3886 	const struct ice_sysctl_info ctls[] = {
3887 		/* Rx Stats */
3888 		{ &stats->rx_bytes, "good_octets_rcvd", "Good Octets Received" },
3889 		{ &stats->rx_unicast, "ucast_pkts_rcvd", "Unicast Packets Received" },
3890 		{ &stats->rx_multicast, "mcast_pkts_rcvd", "Multicast Packets Received" },
3891 		{ &stats->rx_broadcast, "bcast_pkts_rcvd", "Broadcast Packets Received" },
3892 		/* Tx Stats */
3893 		{ &stats->tx_bytes, "good_octets_txd", "Good Octets Transmitted" },
3894 		{ &stats->tx_unicast, "ucast_pkts_txd", "Unicast Packets Transmitted" },
3895 		{ &stats->tx_multicast, "mcast_pkts_txd", "Multicast Packets Transmitted" },
3896 		{ &stats->tx_broadcast, "bcast_pkts_txd", "Broadcast Packets Transmitted" },
3897 		/* End */
3898 		{ 0, 0, 0 }
3899 	};
3900 
3901 	struct sysctl_oid_list *parent_list = SYSCTL_CHILDREN(parent);
3902 
3903 	const struct ice_sysctl_info *entry = ctls;
3904 	while (entry->stat != 0) {
3905 		SYSCTL_ADD_U64(ctx, parent_list, OID_AUTO, entry->name,
3906 			       CTLFLAG_RD | CTLFLAG_STATS, entry->stat, 0,
3907 			       entry->description);
3908 		entry++;
3909 	}
3910 }
3911 
3912 /**
3913  * ice_sysctl_tx_cso_stat - Display Tx checksum offload statistic
3914  * @oidp: sysctl oid structure
3915  * @arg1: pointer to private data structure
3916  * @arg2: Tx CSO stat to read
3917  * @req: sysctl request pointer
3918  *
3919  * On read: Sums the per-queue Tx CSO stat and displays it.
3920  */
3921 static int
3922 ice_sysctl_tx_cso_stat(SYSCTL_HANDLER_ARGS)
3923 {
3924 	struct ice_vsi *vsi = (struct ice_vsi *)arg1;
3925 	enum ice_tx_cso_stat type = (enum ice_tx_cso_stat)arg2;
3926 	u64 stat = 0;
3927 	int i;
3928 
3929 	if (ice_driver_is_detaching(vsi->sc))
3930 		return (ESHUTDOWN);
3931 
3932 	/* Check that the type is valid */
3933 	if (type >= ICE_CSO_STAT_TX_COUNT)
3934 		return (EDOOFUS);
3935 
3936 	/* Sum the stat for each of the Tx queues */
3937 	for (i = 0; i < vsi->num_tx_queues; i++)
3938 		stat += vsi->tx_queues[i].stats.cso[type];
3939 
3940 	return sysctl_handle_64(oidp, NULL, stat, req);
3941 }
3942 
3943 /**
3944  * ice_sysctl_rx_cso_stat - Display Rx checksum offload statistic
3945  * @oidp: sysctl oid structure
3946  * @arg1: pointer to private data structure
3947  * @arg2: Rx CSO stat to read
3948  * @req: sysctl request pointer
3949  *
3950  * On read: Sums the per-queue Rx CSO stat and displays it.
3951  */
3952 static int
3953 ice_sysctl_rx_cso_stat(SYSCTL_HANDLER_ARGS)
3954 {
3955 	struct ice_vsi *vsi = (struct ice_vsi *)arg1;
3956 	enum ice_rx_cso_stat type = (enum ice_rx_cso_stat)arg2;
3957 	u64 stat = 0;
3958 	int i;
3959 
3960 	if (ice_driver_is_detaching(vsi->sc))
3961 		return (ESHUTDOWN);
3962 
3963 	/* Check that the type is valid */
3964 	if (type >= ICE_CSO_STAT_RX_COUNT)
3965 		return (EDOOFUS);
3966 
3967 	/* Sum the stat for each of the Rx queues */
3968 	for (i = 0; i < vsi->num_rx_queues; i++)
3969 		stat += vsi->rx_queues[i].stats.cso[type];
3970 
3971 	return sysctl_handle_64(oidp, NULL, stat, req);
3972 }
3973 
3974 /**
3975  * ice_sysctl_rx_errors_stat - Display aggregate of Rx errors
3976  * @oidp: sysctl oid structure
3977  * @arg1: pointer to private data structure
3978  * @arg2: unused
3979  * @req: sysctl request pointer
3980  *
3981  * On read: Sums current values of Rx error statistics and
3982  * displays it.
3983  */
3984 static int
3985 ice_sysctl_rx_errors_stat(SYSCTL_HANDLER_ARGS)
3986 {
3987 	struct ice_vsi *vsi = (struct ice_vsi *)arg1;
3988 	struct ice_hw_port_stats *hs = &vsi->sc->stats.cur;
3989 	u64 stat = 0;
3990 	int i, type;
3991 
3992 	UNREFERENCED_PARAMETER(arg2);
3993 
3994 	if (ice_driver_is_detaching(vsi->sc))
3995 		return (ESHUTDOWN);
3996 
3997 	stat += hs->rx_undersize;
3998 	stat += hs->rx_fragments;
3999 	stat += hs->rx_oversize;
4000 	stat += hs->rx_jabber;
4001 	stat += hs->rx_len_errors;
4002 	stat += hs->crc_errors;
4003 	stat += hs->illegal_bytes;
4004 
4005 	/* Checksum error stats */
4006 	for (i = 0; i < vsi->num_rx_queues; i++)
4007 		for (type = ICE_CSO_STAT_RX_IP4_ERR;
4008 		     type < ICE_CSO_STAT_RX_COUNT;
4009 		     type++)
4010 			stat += vsi->rx_queues[i].stats.cso[type];
4011 
4012 	return sysctl_handle_64(oidp, NULL, stat, req);
4013 }
4014 
4015 /**
4016  * @struct ice_rx_cso_stat_info
4017  * @brief sysctl information for an Rx checksum offload statistic
4018  *
4019  * Structure used to simplify the process of defining the checksum offload
4020  * statistics.
4021  */
4022 struct ice_rx_cso_stat_info {
4023 	enum ice_rx_cso_stat	type;
4024 	const char		*name;
4025 	const char		*description;
4026 };
4027 
4028 /**
4029  * @struct ice_tx_cso_stat_info
4030  * @brief sysctl information for a Tx checksum offload statistic
4031  *
4032  * Structure used to simplify the process of defining the checksum offload
4033  * statistics.
4034  */
4035 struct ice_tx_cso_stat_info {
4036 	enum ice_tx_cso_stat	type;
4037 	const char		*name;
4038 	const char		*description;
4039 };
4040 
4041 /**
4042  * ice_add_sysctls_sw_stats - Add sysctls for software statistics
4043  * @vsi: pointer to the VSI to add sysctls for
4044  * @ctx: sysctl ctx to use
4045  * @parent: the parent node to add sysctls under
4046  *
4047  * Add statistics sysctls for software tracked statistics of a VSI.
4048  *
4049  * Currently this only adds checksum offload statistics, but more counters may
4050  * be added in the future.
4051  */
4052 static void
4053 ice_add_sysctls_sw_stats(struct ice_vsi *vsi,
4054 			 struct sysctl_ctx_list *ctx,
4055 			 struct sysctl_oid *parent)
4056 {
4057 	struct sysctl_oid *cso_node;
4058 	struct sysctl_oid_list *cso_list;
4059 
4060 	/* Tx CSO Stats */
4061 	const struct ice_tx_cso_stat_info tx_ctls[] = {
4062 		{ ICE_CSO_STAT_TX_TCP, "tx_tcp", "Transmit TCP Packets marked for HW checksum" },
4063 		{ ICE_CSO_STAT_TX_UDP, "tx_udp", "Transmit UDP Packets marked for HW checksum" },
4064 		{ ICE_CSO_STAT_TX_SCTP, "tx_sctp", "Transmit SCTP Packets marked for HW checksum" },
4065 		{ ICE_CSO_STAT_TX_IP4, "tx_ip4", "Transmit IPv4 Packets marked for HW checksum" },
4066 		{ ICE_CSO_STAT_TX_IP6, "tx_ip6", "Transmit IPv6 Packets marked for HW checksum" },
4067 		{ ICE_CSO_STAT_TX_L3_ERR, "tx_l3_err", "Transmit packets that driver failed to set L3 HW CSO bits for" },
4068 		{ ICE_CSO_STAT_TX_L4_ERR, "tx_l4_err", "Transmit packets that driver failed to set L4 HW CSO bits for" },
4069 		/* End */
4070 		{ ICE_CSO_STAT_TX_COUNT, 0, 0 }
4071 	};
4072 
4073 	/* Rx CSO Stats */
4074 	const struct ice_rx_cso_stat_info rx_ctls[] = {
4075 		{ ICE_CSO_STAT_RX_IP4_ERR, "rx_ip4_err", "Received packets with invalid IPv4 checksum indicated by HW" },
4076 		{ ICE_CSO_STAT_RX_IP6_ERR, "rx_ip6_err", "Received IPv6 packets with extension headers" },
4077 		{ ICE_CSO_STAT_RX_L3_ERR, "rx_l3_err", "Received packets with an unexpected invalid L3 checksum indicated by HW" },
4078 		{ ICE_CSO_STAT_RX_TCP_ERR, "rx_tcp_err", "Received packets with invalid TCP checksum indicated by HW" },
4079 		{ ICE_CSO_STAT_RX_UDP_ERR, "rx_udp_err", "Received packets with invalid UDP checksum indicated by HW" },
4080 		{ ICE_CSO_STAT_RX_SCTP_ERR, "rx_sctp_err", "Received packets with invalid SCTP checksum indicated by HW" },
4081 		{ ICE_CSO_STAT_RX_L4_ERR, "rx_l4_err", "Received packets with an unexpected invalid L4 checksum indicated by HW" },
4082 		/* End */
4083 		{ ICE_CSO_STAT_RX_COUNT, 0, 0 }
4084 	};
4085 
4086 	struct sysctl_oid_list *parent_list = SYSCTL_CHILDREN(parent);
4087 
4088 	/* Add a node for statistics tracked by software. */
4089 	cso_node = SYSCTL_ADD_NODE(ctx, parent_list, OID_AUTO, "cso", CTLFLAG_RD,
4090 				  NULL, "Checksum offload Statistics");
4091 	cso_list = SYSCTL_CHILDREN(cso_node);
4092 
4093 	const struct ice_tx_cso_stat_info *tx_entry = tx_ctls;
4094 	while (tx_entry->name && tx_entry->description) {
4095 		SYSCTL_ADD_PROC(ctx, cso_list, OID_AUTO, tx_entry->name,
4096 				CTLTYPE_U64 | CTLFLAG_RD | CTLFLAG_STATS,
4097 				vsi, tx_entry->type, ice_sysctl_tx_cso_stat, "QU",
4098 				tx_entry->description);
4099 		tx_entry++;
4100 	}
4101 
4102 	const struct ice_rx_cso_stat_info *rx_entry = rx_ctls;
4103 	while (rx_entry->name && rx_entry->description) {
4104 		SYSCTL_ADD_PROC(ctx, cso_list, OID_AUTO, rx_entry->name,
4105 				CTLTYPE_U64 | CTLFLAG_RD | CTLFLAG_STATS,
4106 				vsi, rx_entry->type, ice_sysctl_rx_cso_stat, "QU",
4107 				rx_entry->description);
4108 		rx_entry++;
4109 	}
4110 }
4111 
4112 /**
4113  * ice_add_vsi_sysctls - Add sysctls for a VSI
4114  * @vsi: pointer to VSI structure
4115  *
4116  * Add various sysctls for a given VSI.
4117  */
4118 void
4119 ice_add_vsi_sysctls(struct ice_vsi *vsi)
4120 {
4121 	struct sysctl_ctx_list *ctx = &vsi->ctx;
4122 	struct sysctl_oid *hw_node, *sw_node;
4123 	struct sysctl_oid_list *vsi_list, *hw_list, *sw_list;
4124 
4125 	vsi_list = SYSCTL_CHILDREN(vsi->vsi_node);
4126 
4127 	/* Keep hw stats in their own node. */
4128 	hw_node = SYSCTL_ADD_NODE(ctx, vsi_list, OID_AUTO, "hw", CTLFLAG_RD,
4129 				  NULL, "VSI Hardware Statistics");
4130 	hw_list = SYSCTL_CHILDREN(hw_node);
4131 
4132 	/* Add the ethernet statistics for this VSI */
4133 	ice_add_sysctls_eth_stats(ctx, hw_node, &vsi->hw_stats.cur);
4134 
4135 	SYSCTL_ADD_U64(ctx, hw_list, OID_AUTO, "rx_discards",
4136 			CTLFLAG_RD | CTLFLAG_STATS, &vsi->hw_stats.cur.rx_discards,
4137 			0, "Discarded Rx Packets (see rx_errors or rx_no_desc)");
4138 
4139 	SYSCTL_ADD_PROC(ctx, hw_list, OID_AUTO, "rx_errors",
4140 			CTLTYPE_U64 | CTLFLAG_RD | CTLFLAG_STATS,
4141 			vsi, 0, ice_sysctl_rx_errors_stat, "QU",
4142 			"Aggregate of all Rx errors");
4143 
4144 	SYSCTL_ADD_U64(ctx, hw_list, OID_AUTO, "rx_no_desc",
4145 		       CTLFLAG_RD | CTLFLAG_STATS, &vsi->hw_stats.cur.rx_no_desc,
4146 		       0, "Rx Packets Discarded Due To Lack Of Descriptors");
4147 
4148 	SYSCTL_ADD_U64(ctx, hw_list, OID_AUTO, "tx_errors",
4149 			CTLFLAG_RD | CTLFLAG_STATS, &vsi->hw_stats.cur.tx_errors,
4150 			0, "Tx Packets Discarded Due To Error");
4151 
4152 	/* Add a node for statistics tracked by software. */
4153 	sw_node = SYSCTL_ADD_NODE(ctx, vsi_list, OID_AUTO, "sw", CTLFLAG_RD,
4154 				  NULL, "VSI Software Statistics");
4155 	sw_list = SYSCTL_CHILDREN(sw_node);
4156 
4157 	ice_add_sysctls_sw_stats(vsi, ctx, sw_node);
4158 }
4159 
4160 /**
4161  * ice_add_sysctls_mac_stats - Add sysctls for global MAC statistics
4162  * @ctx: the sysctl ctx to use
4163  * @parent: parent node to add the sysctls under
4164  * @stats: the hw ports stat structure to pull values from
4165  *
4166  * Add global MAC statistics sysctls.
4167  */
4168 void
4169 ice_add_sysctls_mac_stats(struct sysctl_ctx_list *ctx,
4170 			  struct sysctl_oid *parent,
4171 			  struct ice_hw_port_stats *stats)
4172 {
4173 	struct sysctl_oid *mac_node;
4174 	struct sysctl_oid_list *parent_list, *mac_list;
4175 
4176 	parent_list = SYSCTL_CHILDREN(parent);
4177 
4178 	mac_node = SYSCTL_ADD_NODE(ctx, parent_list, OID_AUTO, "mac", CTLFLAG_RD,
4179 				   NULL, "Mac Hardware Statistics");
4180 	mac_list = SYSCTL_CHILDREN(mac_node);
4181 
4182 	/* add the common ethernet statistics */
4183 	ice_add_sysctls_eth_stats(ctx, mac_node, &stats->eth);
4184 
4185 	const struct ice_sysctl_info ctls[] = {
4186 		/* Packet Reception Stats */
4187 		{&stats->rx_size_64, "rx_frames_64", "64 byte frames received"},
4188 		{&stats->rx_size_127, "rx_frames_65_127", "65-127 byte frames received"},
4189 		{&stats->rx_size_255, "rx_frames_128_255", "128-255 byte frames received"},
4190 		{&stats->rx_size_511, "rx_frames_256_511", "256-511 byte frames received"},
4191 		{&stats->rx_size_1023, "rx_frames_512_1023", "512-1023 byte frames received"},
4192 		{&stats->rx_size_1522, "rx_frames_1024_1522", "1024-1522 byte frames received"},
4193 		{&stats->rx_size_big, "rx_frames_big", "1523-9522 byte frames received"},
4194 		{&stats->rx_undersize, "rx_undersize", "Undersized packets received"},
4195 		{&stats->rx_fragments, "rx_fragmented", "Fragmented packets received"},
4196 		{&stats->rx_oversize, "rx_oversized", "Oversized packets received"},
4197 		{&stats->rx_jabber, "rx_jabber", "Received Jabber"},
4198 		{&stats->rx_len_errors, "rx_length_errors", "Receive Length Errors"},
4199 		{&stats->eth.rx_discards, "rx_discards",
4200 		    "Discarded Rx Packets by Port (shortage of storage space)"},
4201 		/* Packet Transmission Stats */
4202 		{&stats->tx_size_64, "tx_frames_64", "64 byte frames transmitted"},
4203 		{&stats->tx_size_127, "tx_frames_65_127", "65-127 byte frames transmitted"},
4204 		{&stats->tx_size_255, "tx_frames_128_255", "128-255 byte frames transmitted"},
4205 		{&stats->tx_size_511, "tx_frames_256_511", "256-511 byte frames transmitted"},
4206 		{&stats->tx_size_1023, "tx_frames_512_1023", "512-1023 byte frames transmitted"},
4207 		{&stats->tx_size_1522, "tx_frames_1024_1522", "1024-1522 byte frames transmitted"},
4208 		{&stats->tx_size_big, "tx_frames_big", "1523-9522 byte frames transmitted"},
4209 		{&stats->tx_dropped_link_down, "tx_dropped", "Tx Dropped Due To Link Down"},
4210 		/* Flow control */
4211 		{&stats->link_xon_tx, "xon_txd", "Link XON transmitted"},
4212 		{&stats->link_xon_rx, "xon_recvd", "Link XON received"},
4213 		{&stats->link_xoff_tx, "xoff_txd", "Link XOFF transmitted"},
4214 		{&stats->link_xoff_rx, "xoff_recvd", "Link XOFF received"},
4215 		/* Other */
4216 		{&stats->crc_errors, "crc_errors", "CRC Errors"},
4217 		{&stats->illegal_bytes, "illegal_bytes", "Illegal Byte Errors"},
4218 		{&stats->mac_local_faults, "local_faults", "MAC Local Faults"},
4219 		{&stats->mac_remote_faults, "remote_faults", "MAC Remote Faults"},
4220 		/* End */
4221 		{ 0, 0, 0 }
4222 	};
4223 
4224 	const struct ice_sysctl_info *entry = ctls;
4225 	while (entry->stat != 0) {
4226 		SYSCTL_ADD_U64(ctx, mac_list, OID_AUTO, entry->name,
4227 			CTLFLAG_RD | CTLFLAG_STATS, entry->stat, 0,
4228 			entry->description);
4229 		entry++;
4230 	}
4231 }
4232 
4233 /**
4234  * ice_configure_misc_interrupts - enable 'other' interrupt causes
4235  * @sc: pointer to device private softc
4236  *
4237  * Enable various "other" interrupt causes, and associate them to interrupt 0,
4238  * which is our administrative interrupt.
4239  */
4240 void
4241 ice_configure_misc_interrupts(struct ice_softc *sc)
4242 {
4243 	struct ice_hw *hw = &sc->hw;
4244 	u32 val;
4245 
4246 	/* Read the OICR register to clear it */
4247 	rd32(hw, PFINT_OICR);
4248 
4249 	/* Enable useful "other" interrupt causes */
4250 	val = (PFINT_OICR_ECC_ERR_M |
4251 	       PFINT_OICR_MAL_DETECT_M |
4252 	       PFINT_OICR_GRST_M |
4253 	       PFINT_OICR_PCI_EXCEPTION_M |
4254 	       PFINT_OICR_VFLR_M |
4255 	       PFINT_OICR_HMC_ERR_M |
4256 	       PFINT_OICR_PE_CRITERR_M);
4257 
4258 	wr32(hw, PFINT_OICR_ENA, val);
4259 
4260 	/* Note that since we're using MSI-X index 0, and ITR index 0, we do
4261 	 * not explicitly program them when writing to the PFINT_*_CTL
4262 	 * registers. Nevertheless, these writes are associating the
4263 	 * interrupts with the ITR 0 vector
4264 	 */
4265 
4266 	/* Associate the OICR interrupt with ITR 0, and enable it */
4267 	wr32(hw, PFINT_OICR_CTL, PFINT_OICR_CTL_CAUSE_ENA_M);
4268 
4269 	/* Associate the Mailbox interrupt with ITR 0, and enable it */
4270 	wr32(hw, PFINT_MBX_CTL, PFINT_MBX_CTL_CAUSE_ENA_M);
4271 
4272 	/* Associate the AdminQ interrupt with ITR 0, and enable it */
4273 	wr32(hw, PFINT_FW_CTL, PFINT_FW_CTL_CAUSE_ENA_M);
4274 }
4275 
4276 /**
4277  * ice_filter_is_mcast - Check if info is a multicast filter
4278  * @vsi: vsi structure addresses are targeted towards
4279  * @info: filter info
4280  *
4281  * @returns true if the provided info is a multicast filter, and false
4282  * otherwise.
4283  */
4284 static bool
4285 ice_filter_is_mcast(struct ice_vsi *vsi, struct ice_fltr_info *info)
4286 {
4287 	const u8 *addr = info->l_data.mac.mac_addr;
4288 
4289 	/*
4290 	 * Check if this info matches a multicast filter added by
4291 	 * ice_add_mac_to_list
4292 	 */
4293 	if ((info->flag == ICE_FLTR_TX) &&
4294 	    (info->src_id == ICE_SRC_ID_VSI) &&
4295 	    (info->lkup_type == ICE_SW_LKUP_MAC) &&
4296 	    (info->vsi_handle == vsi->idx) &&
4297 	    ETHER_IS_MULTICAST(addr) && !ETHER_IS_BROADCAST(addr))
4298 		return true;
4299 
4300 	return false;
4301 }
4302 
4303 /**
4304  * @struct ice_mcast_sync_data
4305  * @brief data used by ice_sync_one_mcast_filter function
4306  *
4307  * Structure used to store data needed for processing by the
4308  * ice_sync_one_mcast_filter. This structure contains a linked list of filters
4309  * to be added, an error indication, and a pointer to the device softc.
4310  */
4311 struct ice_mcast_sync_data {
4312 	struct ice_list_head add_list;
4313 	struct ice_softc *sc;
4314 	int err;
4315 };
4316 
4317 /**
4318  * ice_sync_one_mcast_filter - Check if we need to program the filter
4319  * @p: void pointer to algorithm data
4320  * @sdl: link level socket address
4321  * @count: unused count value
4322  *
4323  * Called by if_foreach_llmaddr to operate on each filter in the ifp filter
4324  * list. For the given address, search our internal list to see if we have
4325  * found the filter. If not, add it to our list of filters that need to be
4326  * programmed.
4327  *
4328  * @returns (1) if we've actually setup the filter to be added
4329  */
4330 static u_int
4331 ice_sync_one_mcast_filter(void *p, struct sockaddr_dl *sdl,
4332 			  u_int __unused count)
4333 {
4334 	struct ice_mcast_sync_data *data = (struct ice_mcast_sync_data *)p;
4335 	struct ice_softc *sc = data->sc;
4336 	struct ice_hw *hw = &sc->hw;
4337 	struct ice_switch_info *sw = hw->switch_info;
4338 	const u8 *sdl_addr = (const u8 *)LLADDR(sdl);
4339 	struct ice_fltr_mgmt_list_entry *itr;
4340 	struct ice_list_head *rules;
4341 	int err;
4342 
4343 	rules = &sw->recp_list[ICE_SW_LKUP_MAC].filt_rules;
4344 
4345 	/*
4346 	 * If a previous filter already indicated an error, there is no need
4347 	 * for us to finish processing the rest of the filters.
4348 	 */
4349 	if (data->err)
4350 		return (0);
4351 
4352 	/* See if this filter has already been programmed */
4353 	LIST_FOR_EACH_ENTRY(itr, rules, ice_fltr_mgmt_list_entry, list_entry) {
4354 		struct ice_fltr_info *info = &itr->fltr_info;
4355 		const u8 *addr = info->l_data.mac.mac_addr;
4356 
4357 		/* Only check multicast filters */
4358 		if (!ice_filter_is_mcast(&sc->pf_vsi, info))
4359 			continue;
4360 
4361 		/*
4362 		 * If this filter matches, mark the internal filter as
4363 		 * "found", and exit.
4364 		 */
4365 		if (bcmp(addr, sdl_addr, ETHER_ADDR_LEN) == 0) {
4366 			itr->marker = ICE_FLTR_FOUND;
4367 			return (1);
4368 		}
4369 	}
4370 
4371 	/*
4372 	 * If we failed to locate the filter in our internal list, we need to
4373 	 * place it into our add list.
4374 	 */
4375 	err = ice_add_mac_to_list(&sc->pf_vsi, &data->add_list, sdl_addr,
4376 				  ICE_FWD_TO_VSI);
4377 	if (err) {
4378 		device_printf(sc->dev,
4379 			      "Failed to place MAC %6D onto add list, err %s\n",
4380 			      sdl_addr, ":", ice_err_str(err));
4381 		data->err = err;
4382 
4383 		return (0);
4384 	}
4385 
4386 	return (1);
4387 }
4388 
4389 /**
4390  * ice_sync_multicast_filters - Synchronize OS and internal filter list
4391  * @sc: device private structure
4392  *
4393  * Called in response to SIOCDELMULTI to synchronize the operating system
4394  * multicast address list with the internal list of filters programmed to
4395  * firmware.
4396  *
4397  * Works in one phase to find added and deleted filters using a marker bit on
4398  * the internal list.
4399  *
4400  * First, a loop over the internal list clears the marker bit. Second, for
4401  * each filter in the ifp list is checked. If we find it in the internal list,
4402  * the marker bit is set. Otherwise, the filter is added to the add list.
4403  * Third, a loop over the internal list determines if any filters have not
4404  * been found. Each of these is added to the delete list. Finally, the add and
4405  * delete lists are programmed to firmware to update the filters.
4406  *
4407  * @returns zero on success or an integer error code on failure.
4408  */
4409 int
4410 ice_sync_multicast_filters(struct ice_softc *sc)
4411 {
4412 	struct ice_hw *hw = &sc->hw;
4413 	struct ice_switch_info *sw = hw->switch_info;
4414 	struct ice_fltr_mgmt_list_entry *itr;
4415 	struct ice_mcast_sync_data data = {};
4416 	struct ice_list_head *rules, remove_list;
4417 	enum ice_status status;
4418 	int err = 0;
4419 
4420 	INIT_LIST_HEAD(&data.add_list);
4421 	INIT_LIST_HEAD(&remove_list);
4422 	data.sc = sc;
4423 	data.err = 0;
4424 
4425 	rules = &sw->recp_list[ICE_SW_LKUP_MAC].filt_rules;
4426 
4427 	/* Acquire the lock for the entire duration */
4428 	ice_acquire_lock(&sw->recp_list[ICE_SW_LKUP_MAC].filt_rule_lock);
4429 
4430 	/* (1) Reset the marker state for all filters */
4431 	LIST_FOR_EACH_ENTRY(itr, rules, ice_fltr_mgmt_list_entry, list_entry)
4432 		itr->marker = ICE_FLTR_NOT_FOUND;
4433 
4434 	/* (2) determine which filters need to be added and removed */
4435 	if_foreach_llmaddr(sc->ifp, ice_sync_one_mcast_filter, (void *)&data);
4436 	if (data.err) {
4437 		/* ice_sync_one_mcast_filter already prints an error */
4438 		err = data.err;
4439 		ice_release_lock(&sw->recp_list[ICE_SW_LKUP_MAC].filt_rule_lock);
4440 		goto free_filter_lists;
4441 	}
4442 
4443 	LIST_FOR_EACH_ENTRY(itr, rules, ice_fltr_mgmt_list_entry, list_entry) {
4444 		struct ice_fltr_info *info = &itr->fltr_info;
4445 		const u8 *addr = info->l_data.mac.mac_addr;
4446 
4447 		/* Only check multicast filters */
4448 		if (!ice_filter_is_mcast(&sc->pf_vsi, info))
4449 			continue;
4450 
4451 		/*
4452 		 * If the filter is not marked as found, then it must no
4453 		 * longer be in the ifp address list, so we need to remove it.
4454 		 */
4455 		if (itr->marker == ICE_FLTR_NOT_FOUND) {
4456 			err = ice_add_mac_to_list(&sc->pf_vsi, &remove_list,
4457 						  addr, ICE_FWD_TO_VSI);
4458 			if (err) {
4459 				device_printf(sc->dev,
4460 					      "Failed to place MAC %6D onto remove list, err %s\n",
4461 					      addr, ":", ice_err_str(err));
4462 				ice_release_lock(&sw->recp_list[ICE_SW_LKUP_MAC].filt_rule_lock);
4463 				goto free_filter_lists;
4464 			}
4465 		}
4466 	}
4467 
4468 	ice_release_lock(&sw->recp_list[ICE_SW_LKUP_MAC].filt_rule_lock);
4469 
4470 	status = ice_add_mac(hw, &data.add_list);
4471 	if (status) {
4472 		device_printf(sc->dev,
4473 			      "Could not add new MAC filters, err %s aq_err %s\n",
4474 			      ice_status_str(status), ice_aq_str(hw->adminq.sq_last_status));
4475 		err = (EIO);
4476 		goto free_filter_lists;
4477 	}
4478 
4479 	status = ice_remove_mac(hw, &remove_list);
4480 	if (status) {
4481 		device_printf(sc->dev,
4482 			      "Could not remove old MAC filters, err %s aq_err %s\n",
4483 			      ice_status_str(status), ice_aq_str(hw->adminq.sq_last_status));
4484 		err = (EIO);
4485 		goto free_filter_lists;
4486 	}
4487 
4488 free_filter_lists:
4489 	ice_free_fltr_list(&data.add_list);
4490 	ice_free_fltr_list(&remove_list);
4491 
4492 	return (err);
4493 }
4494 
4495 /**
4496  * ice_add_vlan_hw_filter - Add a VLAN filter for a given VSI
4497  * @vsi: The VSI to add the filter for
4498  * @vid: VLAN to add
4499  *
4500  * Programs a HW filter so that the given VSI will receive the specified VLAN.
4501  */
4502 enum ice_status
4503 ice_add_vlan_hw_filter(struct ice_vsi *vsi, u16 vid)
4504 {
4505 	struct ice_hw *hw = &vsi->sc->hw;
4506 	struct ice_list_head vlan_list;
4507 	struct ice_fltr_list_entry vlan_entry;
4508 
4509 	INIT_LIST_HEAD(&vlan_list);
4510 	memset(&vlan_entry, 0, sizeof(vlan_entry));
4511 
4512 	vlan_entry.fltr_info.lkup_type = ICE_SW_LKUP_VLAN;
4513 	vlan_entry.fltr_info.fltr_act = ICE_FWD_TO_VSI;
4514 	vlan_entry.fltr_info.flag = ICE_FLTR_TX;
4515 	vlan_entry.fltr_info.src_id = ICE_SRC_ID_VSI;
4516 	vlan_entry.fltr_info.vsi_handle = vsi->idx;
4517 	vlan_entry.fltr_info.l_data.vlan.vlan_id = vid;
4518 
4519 	LIST_ADD(&vlan_entry.list_entry, &vlan_list);
4520 
4521 	return ice_add_vlan(hw, &vlan_list);
4522 }
4523 
4524 /**
4525  * ice_remove_vlan_hw_filter - Remove a VLAN filter for a given VSI
4526  * @vsi: The VSI to add the filter for
4527  * @vid: VLAN to remove
4528  *
4529  * Removes a previously programmed HW filter for the specified VSI.
4530  */
4531 enum ice_status
4532 ice_remove_vlan_hw_filter(struct ice_vsi *vsi, u16 vid)
4533 {
4534 	struct ice_hw *hw = &vsi->sc->hw;
4535 	struct ice_list_head vlan_list;
4536 	struct ice_fltr_list_entry vlan_entry;
4537 
4538 	INIT_LIST_HEAD(&vlan_list);
4539 	memset(&vlan_entry, 0, sizeof(vlan_entry));
4540 
4541 	vlan_entry.fltr_info.lkup_type = ICE_SW_LKUP_VLAN;
4542 	vlan_entry.fltr_info.fltr_act = ICE_FWD_TO_VSI;
4543 	vlan_entry.fltr_info.flag = ICE_FLTR_TX;
4544 	vlan_entry.fltr_info.src_id = ICE_SRC_ID_VSI;
4545 	vlan_entry.fltr_info.vsi_handle = vsi->idx;
4546 	vlan_entry.fltr_info.l_data.vlan.vlan_id = vid;
4547 
4548 	LIST_ADD(&vlan_entry.list_entry, &vlan_list);
4549 
4550 	return ice_remove_vlan(hw, &vlan_list);
4551 }
4552 
4553 #define ICE_SYSCTL_HELP_RX_ITR			\
4554 "\nControl Rx interrupt throttle rate."		\
4555 "\n\t0-8160 - sets interrupt rate in usecs"	\
4556 "\n\t    -1 - reset the Rx itr to default"
4557 
4558 /**
4559  * ice_sysctl_rx_itr - Display or change the Rx ITR for a VSI
4560  * @oidp: sysctl oid structure
4561  * @arg1: pointer to private data structure
4562  * @arg2: unused
4563  * @req: sysctl request pointer
4564  *
4565  * On read: Displays the current Rx ITR value
4566  * on write: Sets the Rx ITR value, reconfiguring device if it is up
4567  */
4568 static int
4569 ice_sysctl_rx_itr(SYSCTL_HANDLER_ARGS)
4570 {
4571 	struct ice_vsi *vsi = (struct ice_vsi *)arg1;
4572 	struct ice_softc *sc = vsi->sc;
4573 	int increment, ret;
4574 
4575 	UNREFERENCED_PARAMETER(arg2);
4576 
4577 	if (ice_driver_is_detaching(sc))
4578 		return (ESHUTDOWN);
4579 
4580 	ret = sysctl_handle_16(oidp, &vsi->rx_itr, 0, req);
4581 	if ((ret) || (req->newptr == NULL))
4582 		return (ret);
4583 
4584 	if (vsi->rx_itr < 0)
4585 		vsi->rx_itr = ICE_DFLT_RX_ITR;
4586 	if (vsi->rx_itr > ICE_ITR_MAX)
4587 		vsi->rx_itr = ICE_ITR_MAX;
4588 
4589 	/* Assume 2usec increment if it hasn't been loaded yet */
4590 	increment = sc->hw.itr_gran ? : 2;
4591 
4592 	/* We need to round the value to the hardware's ITR granularity */
4593 	vsi->rx_itr = (vsi->rx_itr / increment ) * increment;
4594 
4595 	/* If the driver has finished initializing, then we need to reprogram
4596 	 * the ITR registers now. Otherwise, they will be programmed during
4597 	 * driver initialization.
4598 	 */
4599 	if (ice_test_state(&sc->state, ICE_STATE_DRIVER_INITIALIZED))
4600 		ice_configure_rx_itr(vsi);
4601 
4602 	return (0);
4603 }
4604 
4605 #define ICE_SYSCTL_HELP_TX_ITR			\
4606 "\nControl Tx interrupt throttle rate."		\
4607 "\n\t0-8160 - sets interrupt rate in usecs"	\
4608 "\n\t    -1 - reset the Tx itr to default"
4609 
4610 /**
4611  * ice_sysctl_tx_itr - Display or change the Tx ITR for a VSI
4612  * @oidp: sysctl oid structure
4613  * @arg1: pointer to private data structure
4614  * @arg2: unused
4615  * @req: sysctl request pointer
4616  *
4617  * On read: Displays the current Tx ITR value
4618  * on write: Sets the Tx ITR value, reconfiguring device if it is up
4619  */
4620 static int
4621 ice_sysctl_tx_itr(SYSCTL_HANDLER_ARGS)
4622 {
4623 	struct ice_vsi *vsi = (struct ice_vsi *)arg1;
4624 	struct ice_softc *sc = vsi->sc;
4625 	int increment, ret;
4626 
4627 	UNREFERENCED_PARAMETER(arg2);
4628 
4629 	if (ice_driver_is_detaching(sc))
4630 		return (ESHUTDOWN);
4631 
4632 	ret = sysctl_handle_16(oidp, &vsi->tx_itr, 0, req);
4633 	if ((ret) || (req->newptr == NULL))
4634 		return (ret);
4635 
4636 	/* Allow configuring a negative value to reset to the default */
4637 	if (vsi->tx_itr < 0)
4638 		vsi->tx_itr = ICE_DFLT_TX_ITR;
4639 	if (vsi->tx_itr > ICE_ITR_MAX)
4640 		vsi->tx_itr = ICE_ITR_MAX;
4641 
4642 	/* Assume 2usec increment if it hasn't been loaded yet */
4643 	increment = sc->hw.itr_gran ? : 2;
4644 
4645 	/* We need to round the value to the hardware's ITR granularity */
4646 	vsi->tx_itr = (vsi->tx_itr / increment ) * increment;
4647 
4648 	/* If the driver has finished initializing, then we need to reprogram
4649 	 * the ITR registers now. Otherwise, they will be programmed during
4650 	 * driver initialization.
4651 	 */
4652 	if (ice_test_state(&sc->state, ICE_STATE_DRIVER_INITIALIZED))
4653 		ice_configure_tx_itr(vsi);
4654 
4655 	return (0);
4656 }
4657 
4658 /**
4659  * ice_add_vsi_tunables - Add tunables and nodes for a VSI
4660  * @vsi: pointer to VSI structure
4661  * @parent: parent node to add the tunables under
4662  *
4663  * Create a sysctl context for the VSI, so that sysctls for the VSI can be
4664  * dynamically removed upon VSI removal.
4665  *
4666  * Add various tunables and set up the basic node structure for the VSI. Must
4667  * be called *prior* to ice_add_vsi_sysctls. It should be called as soon as
4668  * possible after the VSI memory is initialized.
4669  *
4670  * VSI specific sysctls with CTLFLAG_TUN should be initialized here so that
4671  * their values can be read from loader.conf prior to their first use in the
4672  * driver.
4673  */
4674 void
4675 ice_add_vsi_tunables(struct ice_vsi *vsi, struct sysctl_oid *parent)
4676 {
4677 	struct sysctl_oid_list *vsi_list;
4678 	char vsi_name[32], vsi_desc[32];
4679 
4680 	struct sysctl_oid_list *parent_list = SYSCTL_CHILDREN(parent);
4681 
4682 	/* Initialize the sysctl context for this VSI */
4683 	sysctl_ctx_init(&vsi->ctx);
4684 
4685 	/* Add a node to collect this VSI's statistics together */
4686 	snprintf(vsi_name, sizeof(vsi_name), "%u", vsi->idx);
4687 	snprintf(vsi_desc, sizeof(vsi_desc), "VSI %u", vsi->idx);
4688 	vsi->vsi_node = SYSCTL_ADD_NODE(&vsi->ctx, parent_list, OID_AUTO, vsi_name,
4689 					CTLFLAG_RD, NULL, vsi_desc);
4690 	vsi_list = SYSCTL_CHILDREN(vsi->vsi_node);
4691 
4692 	vsi->rx_itr = ICE_DFLT_TX_ITR;
4693 	SYSCTL_ADD_PROC(&vsi->ctx, vsi_list, OID_AUTO, "rx_itr",
4694 			CTLTYPE_S16 | CTLFLAG_RWTUN,
4695 			vsi, 0, ice_sysctl_rx_itr, "S",
4696 			ICE_SYSCTL_HELP_RX_ITR);
4697 
4698 	vsi->tx_itr = ICE_DFLT_TX_ITR;
4699 	SYSCTL_ADD_PROC(&vsi->ctx, vsi_list, OID_AUTO, "tx_itr",
4700 			CTLTYPE_S16 | CTLFLAG_RWTUN,
4701 			vsi, 0, ice_sysctl_tx_itr, "S",
4702 			ICE_SYSCTL_HELP_TX_ITR);
4703 }
4704 
4705 /**
4706  * ice_del_vsi_sysctl_ctx - Delete the sysctl context(s) of a VSI
4707  * @vsi: the VSI to remove contexts for
4708  *
4709  * Free the context for the VSI sysctls. This includes the main context, as
4710  * well as the per-queue sysctls.
4711  */
4712 void
4713 ice_del_vsi_sysctl_ctx(struct ice_vsi *vsi)
4714 {
4715 	device_t dev = vsi->sc->dev;
4716 	int err;
4717 
4718 	if (vsi->vsi_node) {
4719 		err = sysctl_ctx_free(&vsi->ctx);
4720 		if (err)
4721 			device_printf(dev, "failed to free VSI %d sysctl context, err %s\n",
4722 				      vsi->idx, ice_err_str(err));
4723 		vsi->vsi_node = NULL;
4724 	}
4725 }
4726 
4727 /**
4728  * ice_add_device_tunables - Add early tunable sysctls and sysctl nodes
4729  * @sc: device private structure
4730  *
4731  * Add per-device dynamic tunable sysctls, and setup the general sysctl trees
4732  * for re-use by ice_add_device_sysctls.
4733  *
4734  * In order for the sysctl fields to be initialized before use, this function
4735  * should be called as early as possible during attach activities.
4736  *
4737  * Any non-global sysctl marked as CTLFLAG_TUN should likely be initialized
4738  * here in this function, rather than later in ice_add_device_sysctls.
4739  *
4740  * To make things easier, this function is also expected to setup the various
4741  * sysctl nodes in addition to tunables so that other sysctls which can't be
4742  * initialized early can hook into the same nodes.
4743  */
4744 void
4745 ice_add_device_tunables(struct ice_softc *sc)
4746 {
4747 	device_t dev = sc->dev;
4748 
4749 	struct sysctl_ctx_list *ctx = device_get_sysctl_ctx(dev);
4750 	struct sysctl_oid_list *ctx_list =
4751 		SYSCTL_CHILDREN(device_get_sysctl_tree(dev));
4752 
4753 	sc->enable_health_events = ice_enable_health_events;
4754 
4755 	SYSCTL_ADD_BOOL(ctx, ctx_list, OID_AUTO, "enable_health_events",
4756 			CTLFLAG_RDTUN, &sc->enable_health_events, 0,
4757 			"Enable FW health event reporting for this PF");
4758 
4759 	/* Add a node to track VSI sysctls. Keep track of the node in the
4760 	 * softc so that we can hook other sysctls into it later. This
4761 	 * includes both the VSI statistics, as well as potentially dynamic
4762 	 * VSIs in the future.
4763 	 */
4764 
4765 	sc->vsi_sysctls = SYSCTL_ADD_NODE(ctx, ctx_list, OID_AUTO, "vsi",
4766 					  CTLFLAG_RD, NULL, "VSI Configuration and Statistics");
4767 
4768 	/* Add debug tunables */
4769 	ice_add_debug_tunables(sc);
4770 }
4771 
4772 /**
4773  * ice_sysctl_dump_mac_filters - Dump a list of all HW MAC Filters
4774  * @oidp: sysctl oid structure
4775  * @arg1: pointer to private data structure
4776  * @arg2: unused
4777  * @req: sysctl request pointer
4778  *
4779  * Callback for "mac_filters" sysctl to dump the programmed MAC filters.
4780  */
4781 static int
4782 ice_sysctl_dump_mac_filters(SYSCTL_HANDLER_ARGS)
4783 {
4784 	struct ice_softc *sc = (struct ice_softc *)arg1;
4785 	struct ice_hw *hw = &sc->hw;
4786 	struct ice_switch_info *sw = hw->switch_info;
4787 	struct ice_fltr_mgmt_list_entry *fm_entry;
4788 	struct ice_list_head *rule_head;
4789 	struct ice_lock *rule_lock;
4790 	struct ice_fltr_info *fi;
4791 	struct sbuf *sbuf;
4792 	int ret;
4793 
4794 	UNREFERENCED_PARAMETER(oidp);
4795 	UNREFERENCED_PARAMETER(arg2);
4796 
4797 	if (ice_driver_is_detaching(sc))
4798 		return (ESHUTDOWN);
4799 
4800 	/* Wire the old buffer so we can take a non-sleepable lock */
4801 	ret = sysctl_wire_old_buffer(req, 0);
4802 	if (ret)
4803 		return (ret);
4804 
4805 	sbuf = sbuf_new_for_sysctl(NULL, NULL, 128, req);
4806 
4807 	rule_lock = &sw->recp_list[ICE_SW_LKUP_MAC].filt_rule_lock;
4808 	rule_head = &sw->recp_list[ICE_SW_LKUP_MAC].filt_rules;
4809 
4810 	sbuf_printf(sbuf, "MAC Filter List");
4811 
4812 	ice_acquire_lock(rule_lock);
4813 
4814 	LIST_FOR_EACH_ENTRY(fm_entry, rule_head, ice_fltr_mgmt_list_entry, list_entry) {
4815 		fi = &fm_entry->fltr_info;
4816 
4817 		sbuf_printf(sbuf,
4818 			    "\nmac = %6D, vsi_handle = %3d, fw_act_flag = %5s, lb_en = %1d, lan_en = %1d, fltr_act = %15s, fltr_rule_id = %d",
4819 			    fi->l_data.mac.mac_addr, ":", fi->vsi_handle,
4820 			    ice_fltr_flag_str(fi->flag), fi->lb_en, fi->lan_en,
4821 			    ice_fwd_act_str(fi->fltr_act), fi->fltr_rule_id);
4822 
4823 		/* if we have a vsi_list_info, print some information about that */
4824 		if (fm_entry->vsi_list_info) {
4825 			sbuf_printf(sbuf,
4826 				    ", vsi_count = %3d, vsi_list_id = %3d, ref_cnt = %3d",
4827 				    fm_entry->vsi_count,
4828 				    fm_entry->vsi_list_info->vsi_list_id,
4829 				    fm_entry->vsi_list_info->ref_cnt);
4830 		}
4831 	}
4832 
4833 	ice_release_lock(rule_lock);
4834 
4835 	sbuf_finish(sbuf);
4836 	sbuf_delete(sbuf);
4837 
4838 	return (0);
4839 }
4840 
4841 /**
4842  * ice_sysctl_dump_vlan_filters - Dump a list of all HW VLAN Filters
4843  * @oidp: sysctl oid structure
4844  * @arg1: pointer to private data structure
4845  * @arg2: unused
4846  * @req: sysctl request pointer
4847  *
4848  * Callback for "vlan_filters" sysctl to dump the programmed VLAN filters.
4849  */
4850 static int
4851 ice_sysctl_dump_vlan_filters(SYSCTL_HANDLER_ARGS)
4852 {
4853 	struct ice_softc *sc = (struct ice_softc *)arg1;
4854 	struct ice_hw *hw = &sc->hw;
4855 	struct ice_switch_info *sw = hw->switch_info;
4856 	struct ice_fltr_mgmt_list_entry *fm_entry;
4857 	struct ice_list_head *rule_head;
4858 	struct ice_lock *rule_lock;
4859 	struct ice_fltr_info *fi;
4860 	struct sbuf *sbuf;
4861 	int ret;
4862 
4863 	UNREFERENCED_PARAMETER(oidp);
4864 	UNREFERENCED_PARAMETER(arg2);
4865 
4866 	if (ice_driver_is_detaching(sc))
4867 		return (ESHUTDOWN);
4868 
4869 	/* Wire the old buffer so we can take a non-sleepable lock */
4870 	ret = sysctl_wire_old_buffer(req, 0);
4871 	if (ret)
4872 		return (ret);
4873 
4874 	sbuf = sbuf_new_for_sysctl(NULL, NULL, 128, req);
4875 
4876 	rule_lock = &sw->recp_list[ICE_SW_LKUP_VLAN].filt_rule_lock;
4877 	rule_head = &sw->recp_list[ICE_SW_LKUP_VLAN].filt_rules;
4878 
4879 	sbuf_printf(sbuf, "VLAN Filter List");
4880 
4881 	ice_acquire_lock(rule_lock);
4882 
4883 	LIST_FOR_EACH_ENTRY(fm_entry, rule_head, ice_fltr_mgmt_list_entry, list_entry) {
4884 		fi = &fm_entry->fltr_info;
4885 
4886 		sbuf_printf(sbuf,
4887 			    "\nvlan_id = %4d, vsi_handle = %3d, fw_act_flag = %5s, lb_en = %1d, lan_en = %1d, fltr_act = %15s, fltr_rule_id = %4d",
4888 			    fi->l_data.vlan.vlan_id, fi->vsi_handle,
4889 			    ice_fltr_flag_str(fi->flag), fi->lb_en, fi->lan_en,
4890 			    ice_fwd_act_str(fi->fltr_act), fi->fltr_rule_id);
4891 
4892 		/* if we have a vsi_list_info, print some information about that */
4893 		if (fm_entry->vsi_list_info) {
4894 			sbuf_printf(sbuf,
4895 				    ", vsi_count = %3d, vsi_list_id = %3d, ref_cnt = %3d",
4896 				    fm_entry->vsi_count,
4897 				    fm_entry->vsi_list_info->vsi_list_id,
4898 				    fm_entry->vsi_list_info->ref_cnt);
4899 		}
4900 	}
4901 
4902 	ice_release_lock(rule_lock);
4903 
4904 	sbuf_finish(sbuf);
4905 	sbuf_delete(sbuf);
4906 
4907 	return (0);
4908 }
4909 
4910 /**
4911  * ice_sysctl_dump_ethertype_filters - Dump a list of all HW Ethertype filters
4912  * @oidp: sysctl oid structure
4913  * @arg1: pointer to private data structure
4914  * @arg2: unused
4915  * @req: sysctl request pointer
4916  *
4917  * Callback for "ethertype_filters" sysctl to dump the programmed Ethertype
4918  * filters.
4919  */
4920 static int
4921 ice_sysctl_dump_ethertype_filters(SYSCTL_HANDLER_ARGS)
4922 {
4923 	struct ice_softc *sc = (struct ice_softc *)arg1;
4924 	struct ice_hw *hw = &sc->hw;
4925 	struct ice_switch_info *sw = hw->switch_info;
4926 	struct ice_fltr_mgmt_list_entry *fm_entry;
4927 	struct ice_list_head *rule_head;
4928 	struct ice_lock *rule_lock;
4929 	struct ice_fltr_info *fi;
4930 	struct sbuf *sbuf;
4931 	int ret;
4932 
4933 	UNREFERENCED_PARAMETER(oidp);
4934 	UNREFERENCED_PARAMETER(arg2);
4935 
4936 	if (ice_driver_is_detaching(sc))
4937 		return (ESHUTDOWN);
4938 
4939 	/* Wire the old buffer so we can take a non-sleepable lock */
4940 	ret = sysctl_wire_old_buffer(req, 0);
4941 	if (ret)
4942 		return (ret);
4943 
4944 	sbuf = sbuf_new_for_sysctl(NULL, NULL, 128, req);
4945 
4946 	rule_lock = &sw->recp_list[ICE_SW_LKUP_ETHERTYPE].filt_rule_lock;
4947 	rule_head = &sw->recp_list[ICE_SW_LKUP_ETHERTYPE].filt_rules;
4948 
4949 	sbuf_printf(sbuf, "Ethertype Filter List");
4950 
4951 	ice_acquire_lock(rule_lock);
4952 
4953 	LIST_FOR_EACH_ENTRY(fm_entry, rule_head, ice_fltr_mgmt_list_entry, list_entry) {
4954 		fi = &fm_entry->fltr_info;
4955 
4956 		sbuf_printf(sbuf,
4957 			    "\nethertype = 0x%04x, vsi_handle = %3d, fw_act_flag = %5s, lb_en = %1d, lan_en = %1d, fltr_act = %15s, fltr_rule_id = %4d",
4958 			fi->l_data.ethertype_mac.ethertype,
4959 			fi->vsi_handle, ice_fltr_flag_str(fi->flag),
4960 			fi->lb_en, fi->lan_en, ice_fwd_act_str(fi->fltr_act),
4961 			fi->fltr_rule_id);
4962 
4963 		/* if we have a vsi_list_info, print some information about that */
4964 		if (fm_entry->vsi_list_info) {
4965 			sbuf_printf(sbuf,
4966 				    ", vsi_count = %3d, vsi_list_id = %3d, ref_cnt = %3d",
4967 				    fm_entry->vsi_count,
4968 				    fm_entry->vsi_list_info->vsi_list_id,
4969 				    fm_entry->vsi_list_info->ref_cnt);
4970 		}
4971 	}
4972 
4973 	ice_release_lock(rule_lock);
4974 
4975 	sbuf_finish(sbuf);
4976 	sbuf_delete(sbuf);
4977 
4978 	return (0);
4979 }
4980 
4981 /**
4982  * ice_sysctl_dump_ethertype_mac_filters - Dump a list of all HW Ethertype/MAC filters
4983  * @oidp: sysctl oid structure
4984  * @arg1: pointer to private data structure
4985  * @arg2: unused
4986  * @req: sysctl request pointer
4987  *
4988  * Callback for "ethertype_mac_filters" sysctl to dump the programmed
4989  * Ethertype/MAC filters.
4990  */
4991 static int
4992 ice_sysctl_dump_ethertype_mac_filters(SYSCTL_HANDLER_ARGS)
4993 {
4994 	struct ice_softc *sc = (struct ice_softc *)arg1;
4995 	struct ice_hw *hw = &sc->hw;
4996 	struct ice_switch_info *sw = hw->switch_info;
4997 	struct ice_fltr_mgmt_list_entry *fm_entry;
4998 	struct ice_list_head *rule_head;
4999 	struct ice_lock *rule_lock;
5000 	struct ice_fltr_info *fi;
5001 	struct sbuf *sbuf;
5002 	int ret;
5003 
5004 	UNREFERENCED_PARAMETER(oidp);
5005 	UNREFERENCED_PARAMETER(arg2);
5006 
5007 	if (ice_driver_is_detaching(sc))
5008 		return (ESHUTDOWN);
5009 
5010 	/* Wire the old buffer so we can take a non-sleepable lock */
5011 	ret = sysctl_wire_old_buffer(req, 0);
5012 	if (ret)
5013 		return (ret);
5014 
5015 	sbuf = sbuf_new_for_sysctl(NULL, NULL, 128, req);
5016 
5017 	rule_lock = &sw->recp_list[ICE_SW_LKUP_ETHERTYPE_MAC].filt_rule_lock;
5018 	rule_head = &sw->recp_list[ICE_SW_LKUP_ETHERTYPE_MAC].filt_rules;
5019 
5020 	sbuf_printf(sbuf, "Ethertype/MAC Filter List");
5021 
5022 	ice_acquire_lock(rule_lock);
5023 
5024 	LIST_FOR_EACH_ENTRY(fm_entry, rule_head, ice_fltr_mgmt_list_entry, list_entry) {
5025 		fi = &fm_entry->fltr_info;
5026 
5027 		sbuf_printf(sbuf,
5028 			    "\nethertype = 0x%04x, mac = %6D, vsi_handle = %3d, fw_act_flag = %5s, lb_en = %1d, lan_en = %1d, fltr_act = %15s, fltr_rule_id = %4d",
5029 			    fi->l_data.ethertype_mac.ethertype,
5030 			    fi->l_data.ethertype_mac.mac_addr, ":",
5031 			    fi->vsi_handle, ice_fltr_flag_str(fi->flag),
5032 			    fi->lb_en, fi->lan_en, ice_fwd_act_str(fi->fltr_act),
5033 			    fi->fltr_rule_id);
5034 
5035 		/* if we have a vsi_list_info, print some information about that */
5036 		if (fm_entry->vsi_list_info) {
5037 			sbuf_printf(sbuf,
5038 				    ", vsi_count = %3d, vsi_list_id = %3d, ref_cnt = %3d",
5039 				    fm_entry->vsi_count,
5040 				    fm_entry->vsi_list_info->vsi_list_id,
5041 				    fm_entry->vsi_list_info->ref_cnt);
5042 		}
5043 	}
5044 
5045 	ice_release_lock(rule_lock);
5046 
5047 	sbuf_finish(sbuf);
5048 	sbuf_delete(sbuf);
5049 
5050 	return (0);
5051 }
5052 
5053 /**
5054  * ice_sysctl_dump_state_flags - Dump device driver state flags
5055  * @oidp: sysctl oid structure
5056  * @arg1: pointer to private data structure
5057  * @arg2: unused
5058  * @req: sysctl request pointer
5059  *
5060  * Callback for "state" sysctl to display currently set driver state flags.
5061  */
5062 static int
5063 ice_sysctl_dump_state_flags(SYSCTL_HANDLER_ARGS)
5064 {
5065 	struct ice_softc *sc = (struct ice_softc *)arg1;
5066 	struct sbuf *sbuf;
5067 	u32 copied_state;
5068 	unsigned int i;
5069 	bool at_least_one = false;
5070 
5071 	UNREFERENCED_PARAMETER(oidp);
5072 	UNREFERENCED_PARAMETER(arg2);
5073 
5074 	if (ice_driver_is_detaching(sc))
5075 		return (ESHUTDOWN);
5076 
5077 	/* Make a copy of the state to ensure we display coherent values */
5078 	copied_state = atomic_load_acq_32(&sc->state);
5079 
5080 	sbuf = sbuf_new_for_sysctl(NULL, NULL, 128, req);
5081 
5082 	/* Add the string for each set state to the sbuf */
5083 	for (i = 0; i < 32; i++) {
5084 		if (copied_state & BIT(i)) {
5085 			const char *str = ice_state_to_str((enum ice_state)i);
5086 
5087 			at_least_one = true;
5088 
5089 			if (str)
5090 				sbuf_printf(sbuf, "\n%s", str);
5091 			else
5092 				sbuf_printf(sbuf, "\nBIT(%u)", i);
5093 		}
5094 	}
5095 
5096 	if (!at_least_one)
5097 		sbuf_printf(sbuf, "Nothing set");
5098 
5099 	sbuf_finish(sbuf);
5100 	sbuf_delete(sbuf);
5101 
5102 	return (0);
5103 }
5104 
5105 /**
5106  * ice_add_debug_tunables - Add tunables helpful for debugging the device driver
5107  * @sc: device private structure
5108  *
5109  * Add sysctl tunable values related to debugging the device driver. For now,
5110  * this means a tunable to set the debug mask early during driver load.
5111  *
5112  * The debug node will be marked CTLFLAG_SKIP unless INVARIANTS is defined, so
5113  * that in normal kernel builds, these will all be hidden, but on a debug
5114  * kernel they will be more easily visible.
5115  */
5116 static void
5117 ice_add_debug_tunables(struct ice_softc *sc)
5118 {
5119 	struct sysctl_oid_list *debug_list;
5120 	device_t dev = sc->dev;
5121 
5122 	struct sysctl_ctx_list *ctx = device_get_sysctl_ctx(dev);
5123 	struct sysctl_oid_list *ctx_list =
5124 	    SYSCTL_CHILDREN(device_get_sysctl_tree(dev));
5125 
5126 	sc->debug_sysctls = SYSCTL_ADD_NODE(ctx, ctx_list, OID_AUTO, "debug",
5127 					    ICE_CTLFLAG_DEBUG | CTLFLAG_RD,
5128 					    NULL, "Debug Sysctls");
5129 	debug_list = SYSCTL_CHILDREN(sc->debug_sysctls);
5130 
5131 	SYSCTL_ADD_U64(ctx, debug_list, OID_AUTO, "debug_mask",
5132 		       CTLFLAG_RW | CTLFLAG_TUN, &sc->hw.debug_mask, 0,
5133 		       "Debug message enable/disable mask");
5134 
5135 	/* Load the default value from the global sysctl first */
5136 	sc->enable_tx_fc_filter = ice_enable_tx_fc_filter;
5137 
5138 	SYSCTL_ADD_BOOL(ctx, debug_list, OID_AUTO, "enable_tx_fc_filter",
5139 			CTLFLAG_RDTUN, &sc->enable_tx_fc_filter, 0,
5140 			"Drop Ethertype 0x8808 control frames originating from software on this PF");
5141 
5142 	/* Load the default value from the global sysctl first */
5143 	sc->enable_tx_lldp_filter = ice_enable_tx_lldp_filter;
5144 
5145 	SYSCTL_ADD_BOOL(ctx, debug_list, OID_AUTO, "enable_tx_lldp_filter",
5146 			CTLFLAG_RDTUN, &sc->enable_tx_lldp_filter, 0,
5147 			"Drop Ethertype 0x88cc LLDP frames originating from software on this PF");
5148 
5149 }
5150 
5151 #define ICE_SYSCTL_HELP_REQUEST_RESET		\
5152 "\nRequest the driver to initiate a reset."	\
5153 "\n\tpfr - Initiate a PF reset"			\
5154 "\n\tcorer - Initiate a CORE reset"		\
5155 "\n\tglobr - Initiate a GLOBAL reset"
5156 
5157 /**
5158  * @var rl_sysctl_ticks
5159  * @brief timestamp for latest reset request sysctl call
5160  *
5161  * Helps rate-limit the call to the sysctl which resets the device
5162  */
5163 int rl_sysctl_ticks = 0;
5164 
5165 /**
5166  * ice_sysctl_request_reset - Request that the driver initiate a reset
5167  * @oidp: sysctl oid structure
5168  * @arg1: pointer to private data structure
5169  * @arg2: unused
5170  * @req: sysctl request pointer
5171  *
5172  * Callback for "request_reset" sysctl to request that the driver initiate
5173  * a reset. Expects to be passed one of the following strings
5174  *
5175  * "pfr" - Initiate a PF reset
5176  * "corer" - Initiate a CORE reset
5177  * "globr" - Initiate a Global reset
5178  */
5179 static int
5180 ice_sysctl_request_reset(SYSCTL_HANDLER_ARGS)
5181 {
5182 	struct ice_softc *sc = (struct ice_softc *)arg1;
5183 	struct ice_hw *hw = &sc->hw;
5184 	enum ice_status status;
5185 	enum ice_reset_req reset_type = ICE_RESET_INVAL;
5186 	const char *reset_message;
5187 	int ret;
5188 
5189 	/* Buffer to store the requested reset string. Must contain enough
5190 	 * space to store the largest expected reset string, which currently
5191 	 * means 6 bytes of space.
5192 	 */
5193 	char reset[6] = "";
5194 
5195 	UNREFERENCED_PARAMETER(arg2);
5196 
5197 	ret = priv_check(curthread, PRIV_DRIVER);
5198 	if (ret)
5199 		return (ret);
5200 
5201 	if (ice_driver_is_detaching(sc))
5202 		return (ESHUTDOWN);
5203 
5204 	/* Read in the requested reset type. */
5205 	ret = sysctl_handle_string(oidp, reset, sizeof(reset), req);
5206 	if ((ret) || (req->newptr == NULL))
5207 		return (ret);
5208 
5209 	if (strcmp(reset, "pfr") == 0) {
5210 		reset_message = "Requesting a PF reset";
5211 		reset_type = ICE_RESET_PFR;
5212 	} else if (strcmp(reset, "corer") == 0) {
5213 		reset_message = "Initiating a CORE reset";
5214 		reset_type = ICE_RESET_CORER;
5215 	} else if (strcmp(reset, "globr") == 0) {
5216 		reset_message = "Initiating a GLOBAL reset";
5217 		reset_type = ICE_RESET_GLOBR;
5218 	} else if (strcmp(reset, "empr") == 0) {
5219 		device_printf(sc->dev, "Triggering an EMP reset via software is not currently supported\n");
5220 		return (EOPNOTSUPP);
5221 	}
5222 
5223 	if (reset_type == ICE_RESET_INVAL) {
5224 		device_printf(sc->dev, "%s is not a valid reset request\n", reset);
5225 		return (EINVAL);
5226 	}
5227 
5228 	/*
5229 	 * Rate-limit the frequency at which this function is called.
5230 	 * Assuming this is called successfully once, typically,
5231 	 * everything should be handled within the allotted time frame.
5232 	 * However, in the odd setup situations, we've also put in
5233 	 * guards for when the reset has finished, but we're in the
5234 	 * process of rebuilding. And instead of queueing an intent,
5235 	 * simply error out and let the caller retry, if so desired.
5236 	 */
5237 	if (TICKS_2_MSEC(ticks - rl_sysctl_ticks) < 500) {
5238 		device_printf(sc->dev,
5239 		    "Call frequency too high. Operation aborted.\n");
5240 		return (EBUSY);
5241 	}
5242 	rl_sysctl_ticks = ticks;
5243 
5244 	if (TICKS_2_MSEC(ticks - sc->rebuild_ticks) < 100) {
5245 		device_printf(sc->dev, "Device rebuilding. Operation aborted.\n");
5246 		return (EBUSY);
5247 	}
5248 
5249 	if (rd32(hw, GLGEN_RSTAT) & GLGEN_RSTAT_DEVSTATE_M) {
5250 		device_printf(sc->dev, "Device in reset. Operation aborted.\n");
5251 		return (EBUSY);
5252 	}
5253 
5254 	device_printf(sc->dev, "%s\n", reset_message);
5255 
5256 	/* Initiate the PF reset during the admin status task */
5257 	if (reset_type == ICE_RESET_PFR) {
5258 		ice_set_state(&sc->state, ICE_STATE_RESET_PFR_REQ);
5259 		return (0);
5260 	}
5261 
5262 	/*
5263 	 * Other types of resets including CORE and GLOBAL resets trigger an
5264 	 * interrupt on all PFs. Initiate the reset now. Preparation and
5265 	 * rebuild logic will be handled by the admin status task.
5266 	 */
5267 	status = ice_reset(hw, reset_type);
5268 
5269 	/*
5270 	 * Resets can take a long time and we still don't want another call
5271 	 * to this function before we settle down.
5272 	 */
5273 	rl_sysctl_ticks = ticks;
5274 
5275 	if (status) {
5276 		device_printf(sc->dev, "failed to initiate device reset, err %s\n",
5277 			      ice_status_str(status));
5278 		ice_set_state(&sc->state, ICE_STATE_RESET_FAILED);
5279 		return (EFAULT);
5280 	}
5281 
5282 	return (0);
5283 }
5284 
5285 /**
5286  * ice_add_debug_sysctls - Add sysctls helpful for debugging the device driver
5287  * @sc: device private structure
5288  *
5289  * Add sysctls related to debugging the device driver. Generally these should
5290  * simply be sysctls which dump internal driver state, to aid in understanding
5291  * what the driver is doing.
5292  */
5293 static void
5294 ice_add_debug_sysctls(struct ice_softc *sc)
5295 {
5296 	struct sysctl_oid *sw_node;
5297 	struct sysctl_oid_list *debug_list, *sw_list;
5298 	device_t dev = sc->dev;
5299 
5300 	struct sysctl_ctx_list *ctx = device_get_sysctl_ctx(dev);
5301 
5302 	debug_list = SYSCTL_CHILDREN(sc->debug_sysctls);
5303 
5304 	SYSCTL_ADD_PROC(ctx, debug_list, OID_AUTO, "request_reset",
5305 			CTLTYPE_STRING | CTLFLAG_WR, sc, 0,
5306 			ice_sysctl_request_reset, "A",
5307 			ICE_SYSCTL_HELP_REQUEST_RESET);
5308 
5309 	SYSCTL_ADD_U32(ctx, debug_list, OID_AUTO, "pfr_count", CTLFLAG_RD,
5310 		       &sc->soft_stats.pfr_count, 0, "# of PF resets handled");
5311 
5312 	SYSCTL_ADD_U32(ctx, debug_list, OID_AUTO, "corer_count", CTLFLAG_RD,
5313 		       &sc->soft_stats.corer_count, 0, "# of CORE resets handled");
5314 
5315 	SYSCTL_ADD_U32(ctx, debug_list, OID_AUTO, "globr_count", CTLFLAG_RD,
5316 		       &sc->soft_stats.globr_count, 0, "# of Global resets handled");
5317 
5318 	SYSCTL_ADD_U32(ctx, debug_list, OID_AUTO, "empr_count", CTLFLAG_RD,
5319 		       &sc->soft_stats.empr_count, 0, "# of EMP resets handled");
5320 
5321 	SYSCTL_ADD_U32(ctx, debug_list, OID_AUTO, "tx_mdd_count", CTLFLAG_RD,
5322 		       &sc->soft_stats.tx_mdd_count, 0, "# of Tx MDD events detected");
5323 
5324 	SYSCTL_ADD_U32(ctx, debug_list, OID_AUTO, "rx_mdd_count", CTLFLAG_RD,
5325 		       &sc->soft_stats.rx_mdd_count, 0, "# of Rx MDD events detected");
5326 
5327 	SYSCTL_ADD_PROC(ctx, debug_list,
5328 	    OID_AUTO, "state", CTLTYPE_STRING | CTLFLAG_RD,
5329 	    sc, 0, ice_sysctl_dump_state_flags, "A", "Driver State Flags");
5330 
5331 	SYSCTL_ADD_PROC(ctx, debug_list,
5332 			OID_AUTO, "phy_type_low", CTLTYPE_U64 | CTLFLAG_RW,
5333 			sc, 0, ice_sysctl_phy_type_low, "QU",
5334 			"PHY type Low from Get PHY Caps/Set PHY Cfg");
5335 
5336 	SYSCTL_ADD_PROC(ctx, debug_list,
5337 			OID_AUTO, "phy_type_high", CTLTYPE_U64 | CTLFLAG_RW,
5338 			sc, 0, ice_sysctl_phy_type_high, "QU",
5339 			"PHY type High from Get PHY Caps/Set PHY Cfg");
5340 
5341 	SYSCTL_ADD_PROC(ctx, debug_list,
5342 			OID_AUTO, "phy_sw_caps", CTLTYPE_STRUCT | CTLFLAG_RD,
5343 			sc, 0, ice_sysctl_phy_sw_caps, "",
5344 			"Get PHY Capabilities (Software configuration)");
5345 
5346 	SYSCTL_ADD_PROC(ctx, debug_list,
5347 			OID_AUTO, "phy_nvm_caps", CTLTYPE_STRUCT | CTLFLAG_RD,
5348 			sc, 0, ice_sysctl_phy_nvm_caps, "",
5349 			"Get PHY Capabilities (NVM configuration)");
5350 
5351 	SYSCTL_ADD_PROC(ctx, debug_list,
5352 			OID_AUTO, "phy_topo_caps", CTLTYPE_STRUCT | CTLFLAG_RD,
5353 			sc, 0, ice_sysctl_phy_topo_caps, "",
5354 			"Get PHY Capabilities (Topology configuration)");
5355 
5356 	SYSCTL_ADD_PROC(ctx, debug_list,
5357 			OID_AUTO, "phy_link_status", CTLTYPE_STRUCT | CTLFLAG_RD,
5358 			sc, 0, ice_sysctl_phy_link_status, "",
5359 			"Get PHY Link Status");
5360 
5361 	SYSCTL_ADD_PROC(ctx, debug_list,
5362 			OID_AUTO, "read_i2c_diag_data", CTLTYPE_STRING | CTLFLAG_RD,
5363 			sc, 0, ice_sysctl_read_i2c_diag_data, "A",
5364 			"Dump selected diagnostic data from FW");
5365 
5366 	SYSCTL_ADD_U32(ctx, debug_list, OID_AUTO, "fw_build", CTLFLAG_RD,
5367 			&sc->hw.fw_build, 0, "FW Build ID");
5368 
5369 	SYSCTL_ADD_PROC(ctx, debug_list, OID_AUTO, "os_ddp_version", CTLTYPE_STRING | CTLFLAG_RD,
5370 			sc, 0, ice_sysctl_os_pkg_version, "A",
5371 			"DDP package name and version found in ice_ddp");
5372 
5373 	SYSCTL_ADD_PROC(ctx, debug_list,
5374 	    OID_AUTO, "cur_lldp_persist_status", CTLTYPE_STRING | CTLFLAG_RD,
5375 	    sc, 0, ice_sysctl_fw_cur_lldp_persist_status, "A", "Current LLDP persistent status");
5376 
5377 	SYSCTL_ADD_PROC(ctx, debug_list,
5378 	    OID_AUTO, "dflt_lldp_persist_status", CTLTYPE_STRING | CTLFLAG_RD,
5379 	    sc, 0, ice_sysctl_fw_dflt_lldp_persist_status, "A", "Default LLDP persistent status");
5380 
5381 	SYSCTL_ADD_PROC(ctx, debug_list,
5382 	    OID_AUTO, "negotiated_fc", CTLTYPE_STRING | CTLFLAG_RD,
5383 	    sc, 0, ice_sysctl_negotiated_fc, "A", "Current Negotiated Flow Control mode");
5384 
5385 	sw_node = SYSCTL_ADD_NODE(ctx, debug_list, OID_AUTO, "switch",
5386 				  CTLFLAG_RD, NULL, "Switch Configuration");
5387 	sw_list = SYSCTL_CHILDREN(sw_node);
5388 
5389 	SYSCTL_ADD_PROC(ctx, sw_list,
5390 	    OID_AUTO, "mac_filters", CTLTYPE_STRING | CTLFLAG_RD,
5391 	    sc, 0, ice_sysctl_dump_mac_filters, "A", "MAC Filters");
5392 
5393 	SYSCTL_ADD_PROC(ctx, sw_list,
5394 	    OID_AUTO, "vlan_filters", CTLTYPE_STRING | CTLFLAG_RD,
5395 	    sc, 0, ice_sysctl_dump_vlan_filters, "A", "VLAN Filters");
5396 
5397 	SYSCTL_ADD_PROC(ctx, sw_list,
5398 	    OID_AUTO, "ethertype_filters", CTLTYPE_STRING | CTLFLAG_RD,
5399 	    sc, 0, ice_sysctl_dump_ethertype_filters, "A", "Ethertype Filters");
5400 
5401 	SYSCTL_ADD_PROC(ctx, sw_list,
5402 	    OID_AUTO, "ethertype_mac_filters", CTLTYPE_STRING | CTLFLAG_RD,
5403 	    sc, 0, ice_sysctl_dump_ethertype_mac_filters, "A", "Ethertype/MAC Filters");
5404 
5405 }
5406 
5407 /**
5408  * ice_vsi_disable_tx - Disable (unconfigure) Tx queues for a VSI
5409  * @vsi: the VSI to disable
5410  *
5411  * Disables the Tx queues associated with this VSI. Essentially the opposite
5412  * of ice_cfg_vsi_for_tx.
5413  */
5414 int
5415 ice_vsi_disable_tx(struct ice_vsi *vsi)
5416 {
5417 	struct ice_softc *sc = vsi->sc;
5418 	struct ice_hw *hw = &sc->hw;
5419 	enum ice_status status;
5420 	u32 *q_teids;
5421 	u16 *q_ids, *q_handles;
5422 	int i, err = 0;
5423 
5424 	if (vsi->num_tx_queues > 255)
5425 		return (ENOSYS);
5426 
5427 	q_teids = (u32 *)malloc(sizeof(*q_teids) * vsi->num_tx_queues,
5428 				M_ICE, M_NOWAIT|M_ZERO);
5429 	if (!q_teids)
5430 		return (ENOMEM);
5431 
5432 	q_ids = (u16 *)malloc(sizeof(*q_ids) * vsi->num_tx_queues,
5433 				M_ICE, M_NOWAIT|M_ZERO);
5434 	if (!q_ids) {
5435 		err = (ENOMEM);
5436 		goto free_q_teids;
5437 	}
5438 
5439 	q_handles = (u16 *)malloc(sizeof(*q_handles) * vsi->num_tx_queues,
5440 				M_ICE, M_NOWAIT|M_ZERO);
5441 	if (!q_handles) {
5442 		err = (ENOMEM);
5443 		goto free_q_ids;
5444 	}
5445 
5446 
5447 	for (i = 0; i < vsi->num_tx_queues; i++) {
5448 		struct ice_tx_queue *txq = &vsi->tx_queues[i];
5449 
5450 		q_ids[i] = vsi->tx_qmap[i];
5451 		q_handles[i] = i;
5452 		q_teids[i] = txq->q_teid;
5453 	}
5454 
5455 	status = ice_dis_vsi_txq(hw->port_info, vsi->idx, 0, vsi->num_tx_queues,
5456 				 q_handles, q_ids, q_teids, ICE_NO_RESET, 0, NULL);
5457 	if (status == ICE_ERR_DOES_NOT_EXIST) {
5458 		; /* Queues have already been disabled, no need to report this as an error */
5459 	} else if (status == ICE_ERR_RESET_ONGOING) {
5460 		device_printf(sc->dev,
5461 			      "Reset in progress. LAN Tx queues already disabled\n");
5462 	} else if (status) {
5463 		device_printf(sc->dev,
5464 			      "Failed to disable LAN Tx queues: err %s aq_err %s\n",
5465 			      ice_status_str(status), ice_aq_str(hw->adminq.sq_last_status));
5466 		err = (ENODEV);
5467 	}
5468 
5469 /* free_q_handles: */
5470 	free(q_handles, M_ICE);
5471 free_q_ids:
5472 	free(q_ids, M_ICE);
5473 free_q_teids:
5474 	free(q_teids, M_ICE);
5475 
5476 	return err;
5477 }
5478 
5479 /**
5480  * ice_vsi_set_rss_params - Set the RSS parameters for the VSI
5481  * @vsi: the VSI to configure
5482  *
5483  * Sets the RSS table size and lookup table type for the VSI based on its
5484  * VSI type.
5485  */
5486 static void
5487 ice_vsi_set_rss_params(struct ice_vsi *vsi)
5488 {
5489 	struct ice_softc *sc = vsi->sc;
5490 	struct ice_hw_common_caps *cap;
5491 
5492 	cap = &sc->hw.func_caps.common_cap;
5493 
5494 	switch (vsi->type) {
5495 	case ICE_VSI_PF:
5496 		/* The PF VSI inherits RSS instance of the PF */
5497 		vsi->rss_table_size = cap->rss_table_size;
5498 		vsi->rss_lut_type = ICE_AQC_GSET_RSS_LUT_TABLE_TYPE_PF;
5499 		break;
5500 	case ICE_VSI_VF:
5501 		vsi->rss_table_size = ICE_VSIQF_HLUT_ARRAY_SIZE;
5502 		vsi->rss_lut_type = ICE_AQC_GSET_RSS_LUT_TABLE_TYPE_VSI;
5503 		break;
5504 	default:
5505 		device_printf(sc->dev,
5506 			      "VSI %d: RSS not supported for VSI type %d\n",
5507 			      vsi->idx, vsi->type);
5508 		break;
5509 	}
5510 }
5511 
5512 /**
5513  * ice_vsi_add_txqs_ctx - Create a sysctl context and node to store txq sysctls
5514  * @vsi: The VSI to add the context for
5515  *
5516  * Creates a sysctl context for storing txq sysctls. Additionally creates
5517  * a node rooted at the given VSI's main sysctl node. This context will be
5518  * used to store per-txq sysctls which may need to be released during the
5519  * driver's lifetime.
5520  */
5521 void
5522 ice_vsi_add_txqs_ctx(struct ice_vsi *vsi)
5523 {
5524 	struct sysctl_oid_list *vsi_list;
5525 
5526 	sysctl_ctx_init(&vsi->txqs_ctx);
5527 
5528 	vsi_list = SYSCTL_CHILDREN(vsi->vsi_node);
5529 
5530 	vsi->txqs_node = SYSCTL_ADD_NODE(&vsi->txqs_ctx, vsi_list, OID_AUTO, "txqs",
5531 					 CTLFLAG_RD, NULL, "Tx Queues");
5532 }
5533 
5534 /**
5535  * ice_vsi_add_rxqs_ctx - Create a sysctl context and node to store rxq sysctls
5536  * @vsi: The VSI to add the context for
5537  *
5538  * Creates a sysctl context for storing rxq sysctls. Additionally creates
5539  * a node rooted at the given VSI's main sysctl node. This context will be
5540  * used to store per-rxq sysctls which may need to be released during the
5541  * driver's lifetime.
5542  */
5543 void
5544 ice_vsi_add_rxqs_ctx(struct ice_vsi *vsi)
5545 {
5546 	struct sysctl_oid_list *vsi_list;
5547 
5548 	sysctl_ctx_init(&vsi->rxqs_ctx);
5549 
5550 	vsi_list = SYSCTL_CHILDREN(vsi->vsi_node);
5551 
5552 	vsi->rxqs_node = SYSCTL_ADD_NODE(&vsi->rxqs_ctx, vsi_list, OID_AUTO, "rxqs",
5553 					 CTLFLAG_RD, NULL, "Rx Queues");
5554 }
5555 
5556 /**
5557  * ice_vsi_del_txqs_ctx - Delete the Tx queue sysctl context for this VSI
5558  * @vsi: The VSI to delete from
5559  *
5560  * Frees the txq sysctl context created for storing the per-queue Tx sysctls.
5561  * Must be called prior to freeing the Tx queue memory, in order to avoid
5562  * having sysctls point at stale memory.
5563  */
5564 void
5565 ice_vsi_del_txqs_ctx(struct ice_vsi *vsi)
5566 {
5567 	device_t dev = vsi->sc->dev;
5568 	int err;
5569 
5570 	if (vsi->txqs_node) {
5571 		err = sysctl_ctx_free(&vsi->txqs_ctx);
5572 		if (err)
5573 			device_printf(dev, "failed to free VSI %d txqs_ctx, err %s\n",
5574 				      vsi->idx, ice_err_str(err));
5575 		vsi->txqs_node = NULL;
5576 	}
5577 }
5578 
5579 /**
5580  * ice_vsi_del_rxqs_ctx - Delete the Rx queue sysctl context for this VSI
5581  * @vsi: The VSI to delete from
5582  *
5583  * Frees the rxq sysctl context created for storing the per-queue Rx sysctls.
5584  * Must be called prior to freeing the Rx queue memory, in order to avoid
5585  * having sysctls point at stale memory.
5586  */
5587 void
5588 ice_vsi_del_rxqs_ctx(struct ice_vsi *vsi)
5589 {
5590 	device_t dev = vsi->sc->dev;
5591 	int err;
5592 
5593 	if (vsi->rxqs_node) {
5594 		err = sysctl_ctx_free(&vsi->rxqs_ctx);
5595 		if (err)
5596 			device_printf(dev, "failed to free VSI %d rxqs_ctx, err %s\n",
5597 				      vsi->idx, ice_err_str(err));
5598 		vsi->rxqs_node = NULL;
5599 	}
5600 }
5601 
5602 /**
5603  * ice_add_txq_sysctls - Add per-queue sysctls for a Tx queue
5604  * @txq: pointer to the Tx queue
5605  *
5606 * Add per-queue sysctls for a given Tx queue. Can't be called during
5607 * ice_add_vsi_sysctls, since the queue memory has not yet been setup.
5608  */
5609 void
5610 ice_add_txq_sysctls(struct ice_tx_queue *txq)
5611 {
5612 	struct ice_vsi *vsi = txq->vsi;
5613 	struct sysctl_ctx_list *ctx = &vsi->txqs_ctx;
5614 	struct sysctl_oid_list *txqs_list, *this_txq_list;
5615 	struct sysctl_oid *txq_node;
5616 	char txq_name[32], txq_desc[32];
5617 
5618 	const struct ice_sysctl_info ctls[] = {
5619 		{ &txq->stats.tx_packets, "tx_packets", "Queue Packets Transmitted" },
5620 		{ &txq->stats.tx_bytes, "tx_bytes", "Queue Bytes Transmitted" },
5621 		{ &txq->stats.mss_too_small, "mss_too_small", "TSO sends with an MSS less than 64" },
5622 		{ 0, 0, 0 }
5623 	};
5624 
5625 	const struct ice_sysctl_info *entry = ctls;
5626 
5627 	txqs_list = SYSCTL_CHILDREN(vsi->txqs_node);
5628 
5629 	snprintf(txq_name, sizeof(txq_name), "%u", txq->me);
5630 	snprintf(txq_desc, sizeof(txq_desc), "Tx Queue %u", txq->me);
5631 	txq_node = SYSCTL_ADD_NODE(ctx, txqs_list, OID_AUTO, txq_name,
5632 				   CTLFLAG_RD, NULL, txq_desc);
5633 	this_txq_list = SYSCTL_CHILDREN(txq_node);
5634 
5635 	/* Add the Tx queue statistics */
5636 	while (entry->stat != 0) {
5637 		SYSCTL_ADD_U64(ctx, this_txq_list, OID_AUTO, entry->name,
5638 			       CTLFLAG_RD | CTLFLAG_STATS, entry->stat, 0,
5639 			       entry->description);
5640 		entry++;
5641 	}
5642 }
5643 
5644 /**
5645  * ice_add_rxq_sysctls - Add per-queue sysctls for an Rx queue
5646  * @rxq: pointer to the Rx queue
5647  *
5648  * Add per-queue sysctls for a given Rx queue. Can't be called during
5649  * ice_add_vsi_sysctls, since the queue memory has not yet been setup.
5650  */
5651 void
5652 ice_add_rxq_sysctls(struct ice_rx_queue *rxq)
5653 {
5654 	struct ice_vsi *vsi = rxq->vsi;
5655 	struct sysctl_ctx_list *ctx = &vsi->rxqs_ctx;
5656 	struct sysctl_oid_list *rxqs_list, *this_rxq_list;
5657 	struct sysctl_oid *rxq_node;
5658 	char rxq_name[32], rxq_desc[32];
5659 
5660 	const struct ice_sysctl_info ctls[] = {
5661 		{ &rxq->stats.rx_packets, "rx_packets", "Queue Packets Received" },
5662 		{ &rxq->stats.rx_bytes, "rx_bytes", "Queue Bytes Received" },
5663 		{ &rxq->stats.desc_errs, "rx_desc_errs", "Queue Rx Descriptor Errors" },
5664 		{ 0, 0, 0 }
5665 	};
5666 
5667 	const struct ice_sysctl_info *entry = ctls;
5668 
5669 	rxqs_list = SYSCTL_CHILDREN(vsi->rxqs_node);
5670 
5671 	snprintf(rxq_name, sizeof(rxq_name), "%u", rxq->me);
5672 	snprintf(rxq_desc, sizeof(rxq_desc), "Rx Queue %u", rxq->me);
5673 	rxq_node = SYSCTL_ADD_NODE(ctx, rxqs_list, OID_AUTO, rxq_name,
5674 				   CTLFLAG_RD, NULL, rxq_desc);
5675 	this_rxq_list = SYSCTL_CHILDREN(rxq_node);
5676 
5677 	/* Add the Rx queue statistics */
5678 	while (entry->stat != 0) {
5679 		SYSCTL_ADD_U64(ctx, this_rxq_list, OID_AUTO, entry->name,
5680 			       CTLFLAG_RD | CTLFLAG_STATS, entry->stat, 0,
5681 			       entry->description);
5682 		entry++;
5683 	}
5684 }
5685 
5686 /**
5687  * ice_get_default_rss_key - Obtain a default RSS key
5688  * @seed: storage for the RSS key data
5689  *
5690  * Copies a pre-generated RSS key into the seed memory. The seed pointer must
5691  * point to a block of memory that is at least 40 bytes in size.
5692  *
5693  * The key isn't randomly generated each time this function is called because
5694  * that makes the RSS key change every time we reconfigure RSS. This does mean
5695  * that we're hard coding a possibly 'well known' key. We might want to
5696  * investigate randomly generating this key once during the first call.
5697  */
5698 static void
5699 ice_get_default_rss_key(u8 *seed)
5700 {
5701 	const u8 default_seed[ICE_AQC_GET_SET_RSS_KEY_DATA_RSS_KEY_SIZE] = {
5702 		0x39, 0xed, 0xff, 0x4d, 0x43, 0x58, 0x42, 0xc3, 0x5f, 0xb8,
5703 		0xa5, 0x32, 0x95, 0x65, 0x81, 0xcd, 0x36, 0x79, 0x71, 0x97,
5704 		0xde, 0xa4, 0x41, 0x40, 0x6f, 0x27, 0xe9, 0x81, 0x13, 0xa0,
5705 		0x95, 0x93, 0x5b, 0x1e, 0x9d, 0x27, 0x9d, 0x24, 0x84, 0xb5,
5706 	};
5707 
5708 	bcopy(default_seed, seed, ICE_AQC_GET_SET_RSS_KEY_DATA_RSS_KEY_SIZE);
5709 }
5710 
5711 /**
5712  * ice_set_rss_key - Configure a given VSI with the default RSS key
5713  * @vsi: the VSI to configure
5714  *
5715  * Program the hardware RSS key. We use rss_getkey to grab the kernel RSS key.
5716  * If the kernel RSS interface is not available, this will fall back to our
5717  * pre-generated hash seed from ice_get_default_rss_key().
5718  */
5719 static int
5720 ice_set_rss_key(struct ice_vsi *vsi)
5721 {
5722 	struct ice_aqc_get_set_rss_keys keydata = { .standard_rss_key = {0} };
5723 	struct ice_softc *sc = vsi->sc;
5724 	struct ice_hw *hw = &sc->hw;
5725 	enum ice_status status;
5726 
5727 	/*
5728 	 * If the RSS kernel interface is disabled, this will return the
5729 	 * default RSS key above.
5730 	 */
5731 	rss_getkey(keydata.standard_rss_key);
5732 
5733 	status = ice_aq_set_rss_key(hw, vsi->idx, &keydata);
5734 	if (status) {
5735 		device_printf(sc->dev,
5736 			      "ice_aq_set_rss_key status %s, error %s\n",
5737 			      ice_status_str(status), ice_aq_str(hw->adminq.sq_last_status));
5738 		return (EIO);
5739 	}
5740 
5741 	return (0);
5742 }
5743 
5744 /**
5745  * ice_set_rss_flow_flds - Program the RSS hash flows after package init
5746  * @vsi: the VSI to configure
5747  *
5748  * If the package file is initialized, the default RSS flows are reset. We
5749  * need to reprogram the expected hash configuration. We'll use
5750  * rss_gethashconfig() to determine which flows to enable. If RSS kernel
5751  * support is not enabled, this macro will fall back to suitable defaults.
5752  */
5753 static void
5754 ice_set_rss_flow_flds(struct ice_vsi *vsi)
5755 {
5756 	struct ice_softc *sc = vsi->sc;
5757 	struct ice_hw *hw = &sc->hw;
5758 	struct ice_rss_hash_cfg rss_cfg = { 0, 0, ICE_RSS_ANY_HEADERS, false };
5759 	device_t dev = sc->dev;
5760 	enum ice_status status;
5761 	u_int rss_hash_config;
5762 
5763 	rss_hash_config = rss_gethashconfig();
5764 
5765 	if (rss_hash_config & RSS_HASHTYPE_RSS_IPV4) {
5766 		rss_cfg.addl_hdrs = ICE_FLOW_SEG_HDR_IPV4;
5767 		rss_cfg.hash_flds = ICE_FLOW_HASH_IPV4;
5768 		status = ice_add_rss_cfg(hw, vsi->idx, &rss_cfg);
5769 		if (status)
5770 			device_printf(dev,
5771 				      "ice_add_rss_cfg on VSI %d failed for ipv4 flow, err %s aq_err %s\n",
5772 				      vsi->idx, ice_status_str(status), ice_aq_str(hw->adminq.sq_last_status));
5773 	}
5774 	if (rss_hash_config & RSS_HASHTYPE_RSS_TCP_IPV4) {
5775 		rss_cfg.addl_hdrs = ICE_FLOW_SEG_HDR_IPV4 | ICE_FLOW_SEG_HDR_TCP;
5776 		rss_cfg.hash_flds = ICE_HASH_TCP_IPV4;
5777 		status = ice_add_rss_cfg(hw, vsi->idx, &rss_cfg);
5778 		if (status)
5779 			device_printf(dev,
5780 				      "ice_add_rss_cfg on VSI %d failed for tcp4 flow, err %s aq_err %s\n",
5781 				      vsi->idx, ice_status_str(status), ice_aq_str(hw->adminq.sq_last_status));
5782 	}
5783 	if (rss_hash_config & RSS_HASHTYPE_RSS_UDP_IPV4) {
5784 		rss_cfg.addl_hdrs = ICE_FLOW_SEG_HDR_IPV4 | ICE_FLOW_SEG_HDR_UDP;
5785 		rss_cfg.hash_flds = ICE_HASH_UDP_IPV4;
5786 		status = ice_add_rss_cfg(hw, vsi->idx, &rss_cfg);
5787 		if (status)
5788 			device_printf(dev,
5789 				      "ice_add_rss_cfg on VSI %d failed for udp4 flow, err %s aq_err %s\n",
5790 				      vsi->idx, ice_status_str(status), ice_aq_str(hw->adminq.sq_last_status));
5791 	}
5792 	if (rss_hash_config & (RSS_HASHTYPE_RSS_IPV6 | RSS_HASHTYPE_RSS_IPV6_EX)) {
5793 		rss_cfg.addl_hdrs = ICE_FLOW_SEG_HDR_IPV6;
5794 		rss_cfg.hash_flds = ICE_FLOW_HASH_IPV6;
5795 		status = ice_add_rss_cfg(hw, vsi->idx, &rss_cfg);
5796 		if (status)
5797 			device_printf(dev,
5798 				      "ice_add_rss_cfg on VSI %d failed for ipv6 flow, err %s aq_err %s\n",
5799 				      vsi->idx, ice_status_str(status), ice_aq_str(hw->adminq.sq_last_status));
5800 	}
5801 	if (rss_hash_config & RSS_HASHTYPE_RSS_TCP_IPV6) {
5802 		rss_cfg.addl_hdrs = ICE_FLOW_SEG_HDR_IPV6 | ICE_FLOW_SEG_HDR_TCP;
5803 		rss_cfg.hash_flds = ICE_HASH_TCP_IPV6;
5804 		status = ice_add_rss_cfg(hw, vsi->idx, &rss_cfg);
5805 		if (status)
5806 			device_printf(dev,
5807 				      "ice_add_rss_cfg on VSI %d failed for tcp6 flow, err %s aq_err %s\n",
5808 				      vsi->idx, ice_status_str(status), ice_aq_str(hw->adminq.sq_last_status));
5809 	}
5810 	if (rss_hash_config & RSS_HASHTYPE_RSS_UDP_IPV6) {
5811 		rss_cfg.addl_hdrs = ICE_FLOW_SEG_HDR_IPV6 | ICE_FLOW_SEG_HDR_UDP;
5812 		rss_cfg.hash_flds = ICE_HASH_UDP_IPV6;
5813 		status = ice_add_rss_cfg(hw, vsi->idx, &rss_cfg);
5814 		if (status)
5815 			device_printf(dev,
5816 				      "ice_add_rss_cfg on VSI %d failed for udp6 flow, err %s aq_err %s\n",
5817 				      vsi->idx, ice_status_str(status), ice_aq_str(hw->adminq.sq_last_status));
5818 	}
5819 
5820 	/* Warn about RSS hash types which are not supported */
5821 	/* coverity[dead_error_condition] */
5822 	if (rss_hash_config & ~ICE_DEFAULT_RSS_HASH_CONFIG) {
5823 		device_printf(dev,
5824 			      "ice_add_rss_cfg on VSI %d could not configure every requested hash type\n",
5825 			      vsi->idx);
5826 	}
5827 }
5828 
5829 /**
5830  * ice_set_rss_lut - Program the RSS lookup table for a VSI
5831  * @vsi: the VSI to configure
5832  *
5833  * Programs the RSS lookup table for a given VSI. We use
5834  * rss_get_indirection_to_bucket which will use the indirection table provided
5835  * by the kernel RSS interface when available. If the kernel RSS interface is
5836  * not available, we will fall back to a simple round-robin fashion queue
5837  * assignment.
5838  */
5839 static int
5840 ice_set_rss_lut(struct ice_vsi *vsi)
5841 {
5842 	struct ice_softc *sc = vsi->sc;
5843 	struct ice_hw *hw = &sc->hw;
5844 	device_t dev = sc->dev;
5845 	struct ice_aq_get_set_rss_lut_params lut_params;
5846 	enum ice_status status;
5847 	int i, err = 0;
5848 	u8 *lut;
5849 
5850 	lut = (u8 *)malloc(vsi->rss_table_size, M_ICE, M_NOWAIT|M_ZERO);
5851 	if (!lut) {
5852 		device_printf(dev, "Failed to allocate RSS lut memory\n");
5853 		return (ENOMEM);
5854 	}
5855 
5856 	/* Populate the LUT with max no. of queues. If the RSS kernel
5857 	 * interface is disabled, this will assign the lookup table in
5858 	 * a simple round robin fashion
5859 	 */
5860 	for (i = 0; i < vsi->rss_table_size; i++) {
5861 		/* XXX: this needs to be changed if num_rx_queues ever counts
5862 		 * more than just the RSS queues */
5863 		lut[i] = rss_get_indirection_to_bucket(i) % vsi->num_rx_queues;
5864 	}
5865 
5866 	lut_params.vsi_handle = vsi->idx;
5867 	lut_params.lut_size = vsi->rss_table_size;
5868 	lut_params.lut_type = vsi->rss_lut_type;
5869 	lut_params.lut = lut;
5870 	lut_params.global_lut_id = 0;
5871 	status = ice_aq_set_rss_lut(hw, &lut_params);
5872 	if (status) {
5873 		device_printf(dev,
5874 			      "Cannot set RSS lut, err %s aq_err %s\n",
5875 			      ice_status_str(status), ice_aq_str(hw->adminq.sq_last_status));
5876 		err = (EIO);
5877 	}
5878 
5879 	free(lut, M_ICE);
5880 	return err;
5881 }
5882 
5883 /**
5884  * ice_config_rss - Configure RSS for a VSI
5885  * @vsi: the VSI to configure
5886  *
5887  * If FEATURE_RSS is enabled, configures the RSS lookup table and hash key for
5888  * a given VSI.
5889  */
5890 int
5891 ice_config_rss(struct ice_vsi *vsi)
5892 {
5893 	int err;
5894 
5895 	/* Nothing to do, if RSS is not enabled */
5896 	if (!ice_is_bit_set(vsi->sc->feat_en, ICE_FEATURE_RSS))
5897 		return 0;
5898 
5899 	err = ice_set_rss_key(vsi);
5900 	if (err)
5901 		return err;
5902 
5903 	ice_set_rss_flow_flds(vsi);
5904 
5905 	return ice_set_rss_lut(vsi);
5906 }
5907 
5908 /**
5909  * ice_log_pkg_init - Log a message about status of DDP initialization
5910  * @sc: the device softc pointer
5911  * @pkg_status: the status result of ice_copy_and_init_pkg
5912  *
5913  * Called by ice_load_pkg after an attempt to download the DDP package
5914  * contents to the device. Determines whether the download was successful or
5915  * not and logs an appropriate message for the system administrator.
5916  *
5917  * @post if a DDP package was previously downloaded on another port and it
5918  * is not compatible with this driver, pkg_status will be updated to reflect
5919  * this, and the driver will transition to safe mode.
5920  */
5921 void
5922 ice_log_pkg_init(struct ice_softc *sc, enum ice_status *pkg_status)
5923 {
5924 	struct ice_hw *hw = &sc->hw;
5925 	device_t dev = sc->dev;
5926 	struct sbuf *active_pkg, *os_pkg;
5927 
5928 	active_pkg = sbuf_new_auto();
5929 	ice_active_pkg_version_str(hw, active_pkg);
5930 	sbuf_finish(active_pkg);
5931 
5932 	os_pkg = sbuf_new_auto();
5933 	ice_os_pkg_version_str(hw, os_pkg);
5934 	sbuf_finish(os_pkg);
5935 
5936 	switch (*pkg_status) {
5937 	case ICE_SUCCESS:
5938 		/* The package download AdminQ command returned success because
5939 		 * this download succeeded or ICE_ERR_AQ_NO_WORK since there is
5940 		 * already a package loaded on the device.
5941 		 */
5942 		if (hw->pkg_ver.major == hw->active_pkg_ver.major &&
5943 		    hw->pkg_ver.minor == hw->active_pkg_ver.minor &&
5944 		    hw->pkg_ver.update == hw->active_pkg_ver.update &&
5945 		    hw->pkg_ver.draft == hw->active_pkg_ver.draft &&
5946 		    !memcmp(hw->pkg_name, hw->active_pkg_name,
5947 			    sizeof(hw->pkg_name))) {
5948 			switch (hw->pkg_dwnld_status) {
5949 			case ICE_AQ_RC_OK:
5950 				device_printf(dev,
5951 					      "The DDP package was successfully loaded: %s.\n",
5952 					      sbuf_data(active_pkg));
5953 				break;
5954 			case ICE_AQ_RC_EEXIST:
5955 				device_printf(dev,
5956 					      "DDP package already present on device: %s.\n",
5957 					      sbuf_data(active_pkg));
5958 				break;
5959 			default:
5960 				/* We do not expect this to occur, but the
5961 				 * extra messaging is here in case something
5962 				 * changes in the ice_init_pkg flow.
5963 				 */
5964 				device_printf(dev,
5965 					      "DDP package already present on device: %s.  An unexpected error occurred, pkg_dwnld_status %s.\n",
5966 					      sbuf_data(active_pkg),
5967 					      ice_aq_str(hw->pkg_dwnld_status));
5968 				break;
5969 			}
5970 		} else if (pkg_ver_compatible(&hw->active_pkg_ver) == 0) {
5971 			device_printf(dev,
5972 				      "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.  The ice_ddp module has package: %s.\n",
5973 				      sbuf_data(active_pkg),
5974 				      sbuf_data(os_pkg));
5975 		} else if (pkg_ver_compatible(&hw->active_pkg_ver) > 0) {
5976 			device_printf(dev,
5977 				      "The device has a DDP package that is higher than the driver supports.  The device has package %s.  The driver requires version %d.%d.x.x.  Entering Safe Mode.\n",
5978 				      sbuf_data(active_pkg),
5979 				      ICE_PKG_SUPP_VER_MAJ, ICE_PKG_SUPP_VER_MNR);
5980 			*pkg_status = ICE_ERR_NOT_SUPPORTED;
5981 		} else {
5982 			device_printf(dev,
5983 				      "The device has a DDP package that is lower than the driver supports.  The device has package %s.  The driver requires version %d.%d.x.x.  Entering Safe Mode.\n",
5984 				      sbuf_data(active_pkg),
5985 				      ICE_PKG_SUPP_VER_MAJ, ICE_PKG_SUPP_VER_MNR);
5986 			*pkg_status = ICE_ERR_NOT_SUPPORTED;
5987 		}
5988 		break;
5989 	case ICE_ERR_NOT_SUPPORTED:
5990 		/*
5991 		 * This assumes that the active_pkg_ver will not be
5992 		 * initialized if the ice_ddp package version is not
5993 		 * supported.
5994 		 */
5995 		if (pkg_ver_empty(&hw->active_pkg_ver, hw->active_pkg_name)) {
5996 			/* The ice_ddp version is not supported */
5997 			if (pkg_ver_compatible(&hw->pkg_ver) > 0) {
5998 				device_printf(dev,
5999 					      "The DDP package in the ice_ddp module is higher than the driver supports.  The ice_ddp module has package %s.  The driver requires version %d.%d.x.x.  Please use an updated driver.  Entering Safe Mode.\n",
6000 					      sbuf_data(os_pkg),
6001 					      ICE_PKG_SUPP_VER_MAJ, ICE_PKG_SUPP_VER_MNR);
6002 			} else if (pkg_ver_compatible(&hw->pkg_ver) < 0) {
6003 				device_printf(dev,
6004 					      "The DDP package in the ice_ddp module is lower than the driver supports.  The ice_ddp module has package %s.  The driver requires version %d.%d.x.x.  Please use an updated ice_ddp module.  Entering Safe Mode.\n",
6005 					      sbuf_data(os_pkg),
6006 					      ICE_PKG_SUPP_VER_MAJ, ICE_PKG_SUPP_VER_MNR);
6007 			} else {
6008 				device_printf(dev,
6009 					      "An unknown error (%s aq_err %s) occurred when loading the DDP package.  The ice_ddp module has package %s.  The device has package %s.  The driver requires version %d.%d.x.x.  Entering Safe Mode.\n",
6010 					      ice_status_str(*pkg_status),
6011 					      ice_aq_str(hw->pkg_dwnld_status),
6012 					      sbuf_data(os_pkg),
6013 					      sbuf_data(active_pkg),
6014 					      ICE_PKG_SUPP_VER_MAJ, ICE_PKG_SUPP_VER_MNR);
6015 			}
6016 		} else {
6017 			if (pkg_ver_compatible(&hw->active_pkg_ver) > 0) {
6018 				device_printf(dev,
6019 					      "The device has a DDP package that is higher than the driver supports.  The device has package %s.  The driver requires version %d.%d.x.x.  Entering Safe Mode.\n",
6020 					      sbuf_data(active_pkg),
6021 					      ICE_PKG_SUPP_VER_MAJ, ICE_PKG_SUPP_VER_MNR);
6022 			} else if (pkg_ver_compatible(&hw->active_pkg_ver) < 0) {
6023 				device_printf(dev,
6024 					      "The device has a DDP package that is lower than the driver supports.  The device has package %s.  The driver requires version %d.%d.x.x.  Entering Safe Mode.\n",
6025 					      sbuf_data(active_pkg),
6026 					      ICE_PKG_SUPP_VER_MAJ, ICE_PKG_SUPP_VER_MNR);
6027 			} else {
6028 				device_printf(dev,
6029 					      "An unknown error (%s aq_err %s) occurred when loading the DDP package.  The ice_ddp module has package %s.  The device has package %s.  The driver requires version %d.%d.x.x.  Entering Safe Mode.\n",
6030 					      ice_status_str(*pkg_status),
6031 					      ice_aq_str(hw->pkg_dwnld_status),
6032 					      sbuf_data(os_pkg),
6033 					      sbuf_data(active_pkg),
6034 					      ICE_PKG_SUPP_VER_MAJ, ICE_PKG_SUPP_VER_MNR);
6035 			}
6036 		}
6037 		break;
6038 	case ICE_ERR_CFG:
6039 	case ICE_ERR_BUF_TOO_SHORT:
6040 	case ICE_ERR_PARAM:
6041 		device_printf(dev,
6042 			      "The DDP package in the ice_ddp module is invalid.  Entering Safe Mode\n");
6043 		break;
6044 	case ICE_ERR_FW_DDP_MISMATCH:
6045 		device_printf(dev,
6046 			      "The firmware loaded on the device is not compatible with the DDP package.  Please update the device's NVM.  Entering safe mode.\n");
6047 		break;
6048 	case ICE_ERR_AQ_ERROR:
6049 		switch (hw->pkg_dwnld_status) {
6050 		case ICE_AQ_RC_ENOSEC:
6051 		case ICE_AQ_RC_EBADSIG:
6052 			device_printf(dev,
6053 				 "The DDP package in the ice_ddp module cannot be loaded because its signature is not valid.  Please use a valid ice_ddp module.  Entering Safe Mode.\n");
6054 			goto free_sbufs;
6055 		case ICE_AQ_RC_ESVN:
6056 			device_printf(dev,
6057 				 "The DDP package in the ice_ddp module could not be loaded because its security revision is too low.  Please use an updated ice_ddp module.  Entering Safe Mode.\n");
6058 			goto free_sbufs;
6059 		case ICE_AQ_RC_EBADMAN:
6060 		case ICE_AQ_RC_EBADBUF:
6061 			device_printf(dev,
6062 				 "An error occurred on the device while loading the DDP package.  Entering Safe Mode.\n");
6063 			goto free_sbufs;
6064 		default:
6065 			break;
6066 		}
6067 		/* fall-through */
6068 	default:
6069 		device_printf(dev,
6070 			 "An unknown error (%s aq_err %s) occurred when loading the DDP package.  Entering Safe Mode.\n",
6071 			 ice_status_str(*pkg_status),
6072 			 ice_aq_str(hw->pkg_dwnld_status));
6073 		break;
6074 	}
6075 
6076 free_sbufs:
6077 	sbuf_delete(active_pkg);
6078 	sbuf_delete(os_pkg);
6079 }
6080 
6081 /**
6082  * ice_load_pkg_file - Load the DDP package file using firmware_get
6083  * @sc: device private softc
6084  *
6085  * Use firmware_get to load the DDP package memory and then request that
6086  * firmware download the package contents and program the relevant hardware
6087  * bits.
6088  *
6089  * This function makes a copy of the DDP package memory which is tracked in
6090  * the ice_hw structure. The copy will be managed and released by
6091  * ice_deinit_hw(). This allows the firmware reference to be immediately
6092  * released using firmware_put.
6093  */
6094 void
6095 ice_load_pkg_file(struct ice_softc *sc)
6096 {
6097 	struct ice_hw *hw = &sc->hw;
6098 	device_t dev = sc->dev;
6099 	enum ice_status status;
6100 	const struct firmware *pkg;
6101 
6102 	pkg = firmware_get("ice_ddp");
6103 	if (!pkg) {
6104 		device_printf(dev, "The DDP package module (ice_ddp) failed to load or could not be found. Entering Safe Mode.\n");
6105 		if (cold)
6106 			device_printf(dev,
6107 				      "The DDP package module cannot be automatically loaded while booting. You may want to specify ice_ddp_load=\"YES\" in your loader.conf\n");
6108 		ice_set_bit(ICE_FEATURE_SAFE_MODE, sc->feat_cap);
6109 		ice_set_bit(ICE_FEATURE_SAFE_MODE, sc->feat_en);
6110 		return;
6111 	}
6112 
6113 	/* Copy and download the pkg contents */
6114 	status = ice_copy_and_init_pkg(hw, (const u8 *)pkg->data, pkg->datasize);
6115 
6116 	/* Release the firmware reference */
6117 	firmware_put(pkg, FIRMWARE_UNLOAD);
6118 
6119 	/* Check the active DDP package version and log a message */
6120 	ice_log_pkg_init(sc, &status);
6121 
6122 	/* Place the driver into safe mode */
6123 	if (status != ICE_SUCCESS) {
6124 		ice_set_bit(ICE_FEATURE_SAFE_MODE, sc->feat_cap);
6125 		ice_set_bit(ICE_FEATURE_SAFE_MODE, sc->feat_en);
6126 	}
6127 }
6128 
6129 /**
6130  * ice_get_ifnet_counter - Retrieve counter value for a given ifnet counter
6131  * @vsi: the vsi to retrieve the value for
6132  * @counter: the counter type to retrieve
6133  *
6134  * Returns the value for a given ifnet counter. To do so, we calculate the
6135  * value based on the matching hardware statistics.
6136  */
6137 uint64_t
6138 ice_get_ifnet_counter(struct ice_vsi *vsi, ift_counter counter)
6139 {
6140 	struct ice_hw_port_stats *hs = &vsi->sc->stats.cur;
6141 	struct ice_eth_stats *es = &vsi->hw_stats.cur;
6142 
6143 	/* For some statistics, especially those related to error flows, we do
6144 	 * not have per-VSI counters. In this case, we just report the global
6145 	 * counters.
6146 	 */
6147 
6148 	switch (counter) {
6149 	case IFCOUNTER_IPACKETS:
6150 		return (es->rx_unicast + es->rx_multicast + es->rx_broadcast);
6151 	case IFCOUNTER_IERRORS:
6152 		return (hs->crc_errors + hs->illegal_bytes +
6153 			hs->mac_local_faults + hs->mac_remote_faults +
6154 			hs->rx_len_errors + hs->rx_undersize +
6155 			hs->rx_oversize + hs->rx_fragments + hs->rx_jabber);
6156 	case IFCOUNTER_OPACKETS:
6157 		return (es->tx_unicast + es->tx_multicast + es->tx_broadcast);
6158 	case IFCOUNTER_OERRORS:
6159 		return (es->tx_errors);
6160 	case IFCOUNTER_COLLISIONS:
6161 		return (0);
6162 	case IFCOUNTER_IBYTES:
6163 		return (es->rx_bytes);
6164 	case IFCOUNTER_OBYTES:
6165 		return (es->tx_bytes);
6166 	case IFCOUNTER_IMCASTS:
6167 		return (es->rx_multicast);
6168 	case IFCOUNTER_OMCASTS:
6169 		return (es->tx_multicast);
6170 	case IFCOUNTER_IQDROPS:
6171 		return (es->rx_discards);
6172 	case IFCOUNTER_OQDROPS:
6173 		return (hs->tx_dropped_link_down);
6174 	case IFCOUNTER_NOPROTO:
6175 		return (es->rx_unknown_protocol);
6176 	default:
6177 		return if_get_counter_default(vsi->sc->ifp, counter);
6178 	}
6179 }
6180 
6181 /**
6182  * ice_save_pci_info - Save PCI configuration fields in HW struct
6183  * @hw: the ice_hw struct to save the PCI information in
6184  * @dev: the device to get the PCI information from
6185  *
6186  * This should only be called once, early in the device attach
6187  * process.
6188  */
6189 void
6190 ice_save_pci_info(struct ice_hw *hw, device_t dev)
6191 {
6192 	hw->vendor_id = pci_get_vendor(dev);
6193 	hw->device_id = pci_get_device(dev);
6194 	hw->subsystem_vendor_id = pci_get_subvendor(dev);
6195 	hw->subsystem_device_id = pci_get_subdevice(dev);
6196 	hw->revision_id = pci_get_revid(dev);
6197 	hw->bus.device = pci_get_slot(dev);
6198 	hw->bus.func = pci_get_function(dev);
6199 }
6200 
6201 /**
6202  * ice_replay_all_vsi_cfg - Replace configuration for all VSIs after reset
6203  * @sc: the device softc
6204  *
6205  * Replace the configuration for each VSI, and then cleanup replay
6206  * information. Called after a hardware reset in order to reconfigure the
6207  * active VSIs.
6208  */
6209 int
6210 ice_replay_all_vsi_cfg(struct ice_softc *sc)
6211 {
6212 	struct ice_hw *hw = &sc->hw;
6213 	enum ice_status status;
6214 	int i;
6215 
6216 	for (i = 0 ; i < sc->num_available_vsi; i++) {
6217 		struct ice_vsi *vsi = sc->all_vsi[i];
6218 
6219 		if (!vsi)
6220 			continue;
6221 
6222 		status = ice_replay_vsi(hw, vsi->idx);
6223 		if (status) {
6224 			device_printf(sc->dev, "Failed to replay VSI %d, err %s aq_err %s\n",
6225 				      vsi->idx, ice_status_str(status),
6226 				      ice_aq_str(hw->adminq.sq_last_status));
6227 			return (EIO);
6228 		}
6229 	}
6230 
6231 	/* Cleanup replay filters after successful reconfiguration */
6232 	ice_replay_post(hw);
6233 	return (0);
6234 }
6235 
6236 /**
6237  * ice_clean_vsi_rss_cfg - Cleanup RSS configuration for a given VSI
6238  * @vsi: pointer to the VSI structure
6239  *
6240  * Cleanup the advanced RSS configuration for a given VSI. This is necessary
6241  * during driver removal to ensure that all RSS resources are properly
6242  * released.
6243  *
6244  * @remark this function doesn't report an error as it is expected to be
6245  * called during driver reset and unload, and there isn't much the driver can
6246  * do if freeing RSS resources fails.
6247  */
6248 static void
6249 ice_clean_vsi_rss_cfg(struct ice_vsi *vsi)
6250 {
6251 	struct ice_softc *sc = vsi->sc;
6252 	struct ice_hw *hw = &sc->hw;
6253 	device_t dev = sc->dev;
6254 	enum ice_status status;
6255 
6256 	status = ice_rem_vsi_rss_cfg(hw, vsi->idx);
6257 	if (status)
6258 		device_printf(dev,
6259 			      "Failed to remove RSS configuration for VSI %d, err %s\n",
6260 			      vsi->idx, ice_status_str(status));
6261 
6262 	/* Remove this VSI from the RSS list */
6263 	ice_rem_vsi_rss_list(hw, vsi->idx);
6264 }
6265 
6266 /**
6267  * ice_clean_all_vsi_rss_cfg - Cleanup RSS configuration for all VSIs
6268  * @sc: the device softc pointer
6269  *
6270  * Cleanup the advanced RSS configuration for all VSIs on a given PF
6271  * interface.
6272  *
6273  * @remark This should be called while preparing for a reset, to cleanup stale
6274  * RSS configuration for all VSIs.
6275  */
6276 void
6277 ice_clean_all_vsi_rss_cfg(struct ice_softc *sc)
6278 {
6279 	int i;
6280 
6281 	/* No need to cleanup if RSS is not enabled */
6282 	if (!ice_is_bit_set(sc->feat_en, ICE_FEATURE_RSS))
6283 		return;
6284 
6285 	for (i = 0; i < sc->num_available_vsi; i++) {
6286 		struct ice_vsi *vsi = sc->all_vsi[i];
6287 
6288 		if (vsi)
6289 			ice_clean_vsi_rss_cfg(vsi);
6290 	}
6291 }
6292 
6293 /**
6294  * ice_requested_fec_mode - Return the requested FEC mode as a string
6295  * @pi: The port info structure
6296  *
6297  * Return a string representing the requested FEC mode.
6298  */
6299 static const char *
6300 ice_requested_fec_mode(struct ice_port_info *pi)
6301 {
6302 	struct ice_aqc_get_phy_caps_data pcaps = { 0 };
6303 	enum ice_status status;
6304 
6305 	status = ice_aq_get_phy_caps(pi, false, ICE_AQC_REPORT_ACTIVE_CFG,
6306 				     &pcaps, NULL);
6307 	if (status)
6308 		/* Just report unknown if we can't get capabilities */
6309 		return "Unknown";
6310 
6311 	/* Check if RS-FEC has been requested first */
6312 	if (pcaps.link_fec_options & (ICE_AQC_PHY_FEC_25G_RS_528_REQ |
6313 				      ICE_AQC_PHY_FEC_25G_RS_544_REQ))
6314 		return ice_fec_str(ICE_FEC_RS);
6315 
6316 	/* If RS FEC has not been requested, then check BASE-R */
6317 	if (pcaps.link_fec_options & (ICE_AQC_PHY_FEC_10G_KR_40G_KR4_REQ |
6318 				      ICE_AQC_PHY_FEC_25G_KR_REQ))
6319 		return ice_fec_str(ICE_FEC_BASER);
6320 
6321 	return ice_fec_str(ICE_FEC_NONE);
6322 }
6323 
6324 /**
6325  * ice_negotiated_fec_mode - Return the negotiated FEC mode as a string
6326  * @pi: The port info structure
6327  *
6328  * Return a string representing the current FEC mode.
6329  */
6330 static const char *
6331 ice_negotiated_fec_mode(struct ice_port_info *pi)
6332 {
6333 	/* First, check if RS has been requested first */
6334 	if (pi->phy.link_info.fec_info & (ICE_AQ_LINK_25G_RS_528_FEC_EN |
6335 					  ICE_AQ_LINK_25G_RS_544_FEC_EN))
6336 		return ice_fec_str(ICE_FEC_RS);
6337 
6338 	/* If RS FEC has not been requested, then check BASE-R */
6339 	if (pi->phy.link_info.fec_info & ICE_AQ_LINK_25G_KR_FEC_EN)
6340 		return ice_fec_str(ICE_FEC_BASER);
6341 
6342 	return ice_fec_str(ICE_FEC_NONE);
6343 }
6344 
6345 /**
6346  * ice_autoneg_mode - Return string indicating of autoneg completed
6347  * @pi: The port info structure
6348  *
6349  * Return "True" if autonegotiation is completed, "False" otherwise.
6350  */
6351 static const char *
6352 ice_autoneg_mode(struct ice_port_info *pi)
6353 {
6354 	if (pi->phy.link_info.an_info & ICE_AQ_AN_COMPLETED)
6355 		return "True";
6356 	else
6357 		return "False";
6358 }
6359 
6360 /**
6361  * ice_flowcontrol_mode - Return string indicating the Flow Control mode
6362  * @pi: The port info structure
6363  *
6364  * Returns the current Flow Control mode as a string.
6365  */
6366 static const char *
6367 ice_flowcontrol_mode(struct ice_port_info *pi)
6368 {
6369 	return ice_fc_str(pi->fc.current_mode);
6370 }
6371 
6372 /**
6373  * ice_link_up_msg - Log a link up message with associated info
6374  * @sc: the device private softc
6375  *
6376  * Log a link up message with LOG_NOTICE message level. Include information
6377  * about the duplex, FEC mode, autonegotiation and flow control.
6378  */
6379 void
6380 ice_link_up_msg(struct ice_softc *sc)
6381 {
6382 	struct ice_hw *hw = &sc->hw;
6383 	struct ifnet *ifp = sc->ifp;
6384 	const char *speed, *req_fec, *neg_fec, *autoneg, *flowcontrol;
6385 
6386 	speed = ice_aq_speed_to_str(hw->port_info);
6387 	req_fec = ice_requested_fec_mode(hw->port_info);
6388 	neg_fec = ice_negotiated_fec_mode(hw->port_info);
6389 	autoneg = ice_autoneg_mode(hw->port_info);
6390 	flowcontrol = ice_flowcontrol_mode(hw->port_info);
6391 
6392 	log(LOG_NOTICE, "%s: Link is up, %s Full Duplex, Requested FEC: %s, Negotiated FEC: %s, Autoneg: %s, Flow Control: %s\n",
6393 	    ifp->if_xname, speed, req_fec, neg_fec, autoneg, flowcontrol);
6394 }
6395 
6396 /**
6397  * ice_update_laa_mac - Update MAC address if Locally Administered
6398  * @sc: the device softc
6399  *
6400  * Update the device MAC address when a Locally Administered Address is
6401  * assigned.
6402  *
6403  * This function does *not* update the MAC filter list itself. Instead, it
6404  * should be called after ice_rm_pf_default_mac_filters, so that the previous
6405  * address filter will be removed, and before ice_cfg_pf_default_mac_filters,
6406  * so that the new address filter will be assigned.
6407  */
6408 int
6409 ice_update_laa_mac(struct ice_softc *sc)
6410 {
6411 	const u8 *lladdr = (const u8 *)IF_LLADDR(sc->ifp);
6412 	struct ice_hw *hw = &sc->hw;
6413 	enum ice_status status;
6414 
6415 	/* If the address is the same, then there is nothing to update */
6416 	if (!memcmp(lladdr, hw->port_info->mac.lan_addr, ETHER_ADDR_LEN))
6417 		return (0);
6418 
6419 	/* Reject Multicast addresses */
6420 	if (ETHER_IS_MULTICAST(lladdr))
6421 		return (EINVAL);
6422 
6423 	status = ice_aq_manage_mac_write(hw, lladdr, ICE_AQC_MAN_MAC_UPDATE_LAA_WOL, NULL);
6424 	if (status) {
6425 		device_printf(sc->dev, "Failed to write mac %6D to firmware, err %s aq_err %s\n",
6426 			      lladdr, ":", ice_status_str(status),
6427 			      ice_aq_str(hw->adminq.sq_last_status));
6428 		return (EFAULT);
6429 	}
6430 
6431 	/* Copy the address into place of the LAN address. */
6432 	bcopy(lladdr, hw->port_info->mac.lan_addr, ETHER_ADDR_LEN);
6433 
6434 	return (0);
6435 }
6436 
6437 /**
6438  * ice_get_and_print_bus_info - Save (PCI) bus info and print messages
6439  * @sc: device softc
6440  *
6441  * This will potentially print out a warning message if bus bandwidth
6442  * is insufficient for full-speed operation.
6443  *
6444  * This should only be called once, during the attach process, after
6445  * hw->port_info has been filled out with port link topology information
6446  * (from the Get PHY Capabilities Admin Queue command).
6447  */
6448 void
6449 ice_get_and_print_bus_info(struct ice_softc *sc)
6450 {
6451 	struct ice_hw *hw = &sc->hw;
6452 	device_t dev = sc->dev;
6453 	u16 pci_link_status;
6454 	int offset;
6455 
6456 	pci_find_cap(dev, PCIY_EXPRESS, &offset);
6457 	pci_link_status = pci_read_config(dev, offset + PCIER_LINK_STA, 2);
6458 
6459 	/* Fill out hw struct with PCIE link status info */
6460 	ice_set_pci_link_status_data(hw, pci_link_status);
6461 
6462 	/* Use info to print out bandwidth messages */
6463 	ice_print_bus_link_data(dev, hw);
6464 
6465 	if (ice_pcie_bandwidth_check(sc)) {
6466 		device_printf(dev,
6467 		    "PCI-Express bandwidth available for this device may be insufficient for optimal performance.\n");
6468 		device_printf(dev,
6469 		    "Please move the device to a different PCI-e link with more lanes and/or higher transfer rate.\n");
6470 	}
6471 }
6472 
6473 /**
6474  * ice_pcie_bus_speed_to_rate - Convert driver bus speed enum value to
6475  * a 64-bit baudrate.
6476  * @speed: enum value to convert
6477  *
6478  * This only goes up to PCIE Gen 4.
6479  */
6480 static uint64_t
6481 ice_pcie_bus_speed_to_rate(enum ice_pcie_bus_speed speed)
6482 {
6483 	/* If the PCI-E speed is Gen1 or Gen2, then report
6484 	 * only 80% of bus speed to account for encoding overhead.
6485 	 */
6486 	switch (speed) {
6487 	case ice_pcie_speed_2_5GT:
6488 		return IF_Gbps(2);
6489 	case ice_pcie_speed_5_0GT:
6490 		return IF_Gbps(4);
6491 	case ice_pcie_speed_8_0GT:
6492 		return IF_Gbps(8);
6493 	case ice_pcie_speed_16_0GT:
6494 		return IF_Gbps(16);
6495 	case ice_pcie_speed_unknown:
6496 	default:
6497 		return 0;
6498 	}
6499 }
6500 
6501 /**
6502  * ice_pcie_lnk_width_to_int - Convert driver pci-e width enum value to
6503  * a 32-bit number.
6504  * @width: enum value to convert
6505  */
6506 static int
6507 ice_pcie_lnk_width_to_int(enum ice_pcie_link_width width)
6508 {
6509 	switch (width) {
6510 	case ice_pcie_lnk_x1:
6511 		return (1);
6512 	case ice_pcie_lnk_x2:
6513 		return (2);
6514 	case ice_pcie_lnk_x4:
6515 		return (4);
6516 	case ice_pcie_lnk_x8:
6517 		return (8);
6518 	case ice_pcie_lnk_x12:
6519 		return (12);
6520 	case ice_pcie_lnk_x16:
6521 		return (16);
6522 	case ice_pcie_lnk_x32:
6523 		return (32);
6524 	case ice_pcie_lnk_width_resrv:
6525 	case ice_pcie_lnk_width_unknown:
6526 	default:
6527 		return (0);
6528 	}
6529 }
6530 
6531 /**
6532  * ice_pcie_bandwidth_check - Check if PCI-E bandwidth is sufficient for
6533  * full-speed device operation.
6534  * @sc: adapter softc
6535  *
6536  * Returns 0 if sufficient; 1 if not.
6537  */
6538 static uint8_t
6539 ice_pcie_bandwidth_check(struct ice_softc *sc)
6540 {
6541 	struct ice_hw *hw = &sc->hw;
6542 	int num_ports, pcie_width;
6543 	u64 pcie_speed, port_speed;
6544 
6545 	MPASS(hw->port_info);
6546 
6547 	num_ports = bitcount32(hw->func_caps.common_cap.valid_functions);
6548 	port_speed = ice_phy_types_to_max_rate(hw->port_info);
6549 	pcie_speed = ice_pcie_bus_speed_to_rate(hw->bus.speed);
6550 	pcie_width = ice_pcie_lnk_width_to_int(hw->bus.width);
6551 
6552 	/*
6553 	 * If 2x100, clamp ports to 1 -- 2nd port is intended for
6554 	 * failover.
6555 	 */
6556 	if (port_speed == IF_Gbps(100))
6557 		num_ports = 1;
6558 
6559 	return !!((num_ports * port_speed) > pcie_speed * pcie_width);
6560 }
6561 
6562 /**
6563  * ice_print_bus_link_data - Print PCI-E bandwidth information
6564  * @dev: device to print string for
6565  * @hw: hw struct with PCI-e link information
6566  */
6567 static void
6568 ice_print_bus_link_data(device_t dev, struct ice_hw *hw)
6569 {
6570         device_printf(dev, "PCI Express Bus: Speed %s %s\n",
6571             ((hw->bus.speed == ice_pcie_speed_16_0GT) ? "16.0GT/s" :
6572             (hw->bus.speed == ice_pcie_speed_8_0GT) ? "8.0GT/s" :
6573             (hw->bus.speed == ice_pcie_speed_5_0GT) ? "5.0GT/s" :
6574             (hw->bus.speed == ice_pcie_speed_2_5GT) ? "2.5GT/s" : "Unknown"),
6575             (hw->bus.width == ice_pcie_lnk_x32) ? "Width x32" :
6576             (hw->bus.width == ice_pcie_lnk_x16) ? "Width x16" :
6577             (hw->bus.width == ice_pcie_lnk_x12) ? "Width x12" :
6578             (hw->bus.width == ice_pcie_lnk_x8) ? "Width x8" :
6579             (hw->bus.width == ice_pcie_lnk_x4) ? "Width x4" :
6580             (hw->bus.width == ice_pcie_lnk_x2) ? "Width x2" :
6581             (hw->bus.width == ice_pcie_lnk_x1) ? "Width x1" : "Width Unknown");
6582 }
6583 
6584 /**
6585  * ice_set_pci_link_status_data - store PCI bus info
6586  * @hw: pointer to hardware structure
6587  * @link_status: the link status word from PCI config space
6588  *
6589  * Stores the PCI bus info (speed, width, type) within the ice_hw structure
6590  **/
6591 static void
6592 ice_set_pci_link_status_data(struct ice_hw *hw, u16 link_status)
6593 {
6594 	u16 reg;
6595 
6596 	hw->bus.type = ice_bus_pci_express;
6597 
6598 	reg = (link_status & PCIEM_LINK_STA_WIDTH) >> 4;
6599 
6600 	switch (reg) {
6601 	case ice_pcie_lnk_x1:
6602 	case ice_pcie_lnk_x2:
6603 	case ice_pcie_lnk_x4:
6604 	case ice_pcie_lnk_x8:
6605 	case ice_pcie_lnk_x12:
6606 	case ice_pcie_lnk_x16:
6607 	case ice_pcie_lnk_x32:
6608 		hw->bus.width = (enum ice_pcie_link_width)reg;
6609 		break;
6610 	default:
6611 		hw->bus.width = ice_pcie_lnk_width_unknown;
6612 		break;
6613 	}
6614 
6615 	reg = (link_status & PCIEM_LINK_STA_SPEED) + 0x13;
6616 
6617 	switch (reg) {
6618 	case ice_pcie_speed_2_5GT:
6619 	case ice_pcie_speed_5_0GT:
6620 	case ice_pcie_speed_8_0GT:
6621 	case ice_pcie_speed_16_0GT:
6622 		hw->bus.speed = (enum ice_pcie_bus_speed)reg;
6623 		break;
6624 	default:
6625 		hw->bus.speed = ice_pcie_speed_unknown;
6626 		break;
6627 	}
6628 }
6629 
6630 /**
6631  * ice_init_link_events - Initialize Link Status Events mask
6632  * @sc: the device softc
6633  *
6634  * Initialize the Link Status Events mask to disable notification of link
6635  * events we don't care about in software. Also request that link status
6636  * events be enabled.
6637  */
6638 int
6639 ice_init_link_events(struct ice_softc *sc)
6640 {
6641 	struct ice_hw *hw = &sc->hw;
6642 	enum ice_status status;
6643 	u16 wanted_events;
6644 
6645 	/* Set the bits for the events that we want to be notified by */
6646 	wanted_events = (ICE_AQ_LINK_EVENT_UPDOWN |
6647 			 ICE_AQ_LINK_EVENT_MEDIA_NA |
6648 			 ICE_AQ_LINK_EVENT_MODULE_QUAL_FAIL);
6649 
6650 	/* request that every event except the wanted events be masked */
6651 	status = ice_aq_set_event_mask(hw, hw->port_info->lport, ~wanted_events, NULL);
6652 	if (status) {
6653 		device_printf(sc->dev,
6654 			      "Failed to set link status event mask, err %s aq_err %s\n",
6655 			      ice_status_str(status), ice_aq_str(hw->adminq.sq_last_status));
6656 		return (EIO);
6657 	}
6658 
6659 	/* Request link info with the LSE bit set to enable link status events */
6660 	status = ice_aq_get_link_info(hw->port_info, true, NULL, NULL);
6661 	if (status) {
6662 		device_printf(sc->dev,
6663 			      "Failed to enable link status events, err %s aq_err %s\n",
6664 			      ice_status_str(status), ice_aq_str(hw->adminq.sq_last_status));
6665 		return (EIO);
6666 	}
6667 
6668 	return (0);
6669 }
6670 
6671 /**
6672  * ice_handle_mdd_event - Handle possibly malicious events
6673  * @sc: the device softc
6674  *
6675  * Called by the admin task if an MDD detection interrupt is triggered.
6676  * Identifies possibly malicious events coming from VFs. Also triggers for
6677  * similar incorrect behavior from the PF as well.
6678  */
6679 void
6680 ice_handle_mdd_event(struct ice_softc *sc)
6681 {
6682 	struct ice_hw *hw = &sc->hw;
6683 	bool mdd_detected = false, request_reinit = false;
6684 	device_t dev = sc->dev;
6685 	u32 reg;
6686 
6687 	if (!ice_testandclear_state(&sc->state, ICE_STATE_MDD_PENDING))
6688 		return;
6689 
6690 	reg = rd32(hw, GL_MDET_TX_TCLAN);
6691 	if (reg & GL_MDET_TX_TCLAN_VALID_M) {
6692 		u8 pf_num  = (reg & GL_MDET_TX_TCLAN_PF_NUM_M) >> GL_MDET_TX_TCLAN_PF_NUM_S;
6693 		u16 vf_num = (reg & GL_MDET_TX_TCLAN_VF_NUM_M) >> GL_MDET_TX_TCLAN_VF_NUM_S;
6694 		u8 event   = (reg & GL_MDET_TX_TCLAN_MAL_TYPE_M) >> GL_MDET_TX_TCLAN_MAL_TYPE_S;
6695 		u16 queue  = (reg & GL_MDET_TX_TCLAN_QNUM_M) >> GL_MDET_TX_TCLAN_QNUM_S;
6696 
6697 		device_printf(dev, "Malicious Driver Detection Tx Descriptor check event '%s' on Tx queue %u PF# %u VF# %u\n",
6698 			      ice_mdd_tx_tclan_str(event), queue, pf_num, vf_num);
6699 
6700 		/* Only clear this event if it matches this PF, that way other
6701 		 * PFs can read the event and determine VF and queue number.
6702 		 */
6703 		if (pf_num == hw->pf_id)
6704 			wr32(hw, GL_MDET_TX_TCLAN, 0xffffffff);
6705 
6706 		mdd_detected = true;
6707 	}
6708 
6709 	/* Determine what triggered the MDD event */
6710 	reg = rd32(hw, GL_MDET_TX_PQM);
6711 	if (reg & GL_MDET_TX_PQM_VALID_M) {
6712 		u8 pf_num  = (reg & GL_MDET_TX_PQM_PF_NUM_M) >> GL_MDET_TX_PQM_PF_NUM_S;
6713 		u16 vf_num = (reg & GL_MDET_TX_PQM_VF_NUM_M) >> GL_MDET_TX_PQM_VF_NUM_S;
6714 		u8 event   = (reg & GL_MDET_TX_PQM_MAL_TYPE_M) >> GL_MDET_TX_PQM_MAL_TYPE_S;
6715 		u16 queue  = (reg & GL_MDET_TX_PQM_QNUM_M) >> GL_MDET_TX_PQM_QNUM_S;
6716 
6717 		device_printf(dev, "Malicious Driver Detection Tx Quanta check event '%s' on Tx queue %u PF# %u VF# %u\n",
6718 			      ice_mdd_tx_pqm_str(event), queue, pf_num, vf_num);
6719 
6720 		/* Only clear this event if it matches this PF, that way other
6721 		 * PFs can read the event and determine VF and queue number.
6722 		 */
6723 		if (pf_num == hw->pf_id)
6724 			wr32(hw, GL_MDET_TX_PQM, 0xffffffff);
6725 
6726 		mdd_detected = true;
6727 	}
6728 
6729 	reg = rd32(hw, GL_MDET_RX);
6730 	if (reg & GL_MDET_RX_VALID_M) {
6731 		u8 pf_num  = (reg & GL_MDET_RX_PF_NUM_M) >> GL_MDET_RX_PF_NUM_S;
6732 		u16 vf_num = (reg & GL_MDET_RX_VF_NUM_M) >> GL_MDET_RX_VF_NUM_S;
6733 		u8 event   = (reg & GL_MDET_RX_MAL_TYPE_M) >> GL_MDET_RX_MAL_TYPE_S;
6734 		u16 queue  = (reg & GL_MDET_RX_QNUM_M) >> GL_MDET_RX_QNUM_S;
6735 
6736 		device_printf(dev, "Malicious Driver Detection Rx event '%s' on Rx queue %u PF# %u VF# %u\n",
6737 			      ice_mdd_rx_str(event), queue, pf_num, vf_num);
6738 
6739 		/* Only clear this event if it matches this PF, that way other
6740 		 * PFs can read the event and determine VF and queue number.
6741 		 */
6742 		if (pf_num == hw->pf_id)
6743 			wr32(hw, GL_MDET_RX, 0xffffffff);
6744 
6745 		mdd_detected = true;
6746 	}
6747 
6748 	/* Now, confirm that this event actually affects this PF, by checking
6749 	 * the PF registers.
6750 	 */
6751 	if (mdd_detected) {
6752 		reg = rd32(hw, PF_MDET_TX_TCLAN);
6753 		if (reg & PF_MDET_TX_TCLAN_VALID_M) {
6754 			wr32(hw, PF_MDET_TX_TCLAN, 0xffff);
6755 			sc->soft_stats.tx_mdd_count++;
6756 			request_reinit = true;
6757 		}
6758 
6759 		reg = rd32(hw, PF_MDET_TX_PQM);
6760 		if (reg & PF_MDET_TX_PQM_VALID_M) {
6761 			wr32(hw, PF_MDET_TX_PQM, 0xffff);
6762 			sc->soft_stats.tx_mdd_count++;
6763 			request_reinit = true;
6764 		}
6765 
6766 		reg = rd32(hw, PF_MDET_RX);
6767 		if (reg & PF_MDET_RX_VALID_M) {
6768 			wr32(hw, PF_MDET_RX, 0xffff);
6769 			sc->soft_stats.rx_mdd_count++;
6770 			request_reinit = true;
6771 		}
6772 	}
6773 
6774 	/* TODO: Implement logic to detect and handle events caused by VFs. */
6775 
6776 	/* request that the upper stack re-initialize the Tx/Rx queues */
6777 	if (request_reinit)
6778 		ice_request_stack_reinit(sc);
6779 
6780 	ice_flush(hw);
6781 }
6782 
6783 /**
6784  * ice_init_dcb_setup - Initialize DCB settings for HW
6785  * @sc: the device softc
6786  *
6787  * This needs to be called after the fw_lldp_agent sysctl is added, since that
6788  * can update the device's LLDP agent status if a tunable value is set.
6789  *
6790  * Get and store the initial state of DCB settings on driver load. Print out
6791  * informational messages as well.
6792  */
6793 void
6794 ice_init_dcb_setup(struct ice_softc *sc)
6795 {
6796 	struct ice_hw *hw = &sc->hw;
6797 	device_t dev = sc->dev;
6798 	bool dcbx_agent_status;
6799 	enum ice_status status;
6800 
6801 	/* Don't do anything if DCB isn't supported */
6802 	if (!hw->func_caps.common_cap.dcb) {
6803 		device_printf(dev, "%s: No DCB support\n",
6804 		    __func__);
6805 		return;
6806 	}
6807 
6808 	hw->port_info->qos_cfg.dcbx_status = ice_get_dcbx_status(hw);
6809 	if (hw->port_info->qos_cfg.dcbx_status != ICE_DCBX_STATUS_DONE &&
6810 	    hw->port_info->qos_cfg.dcbx_status != ICE_DCBX_STATUS_IN_PROGRESS) {
6811 		/*
6812 		 * Start DCBX agent, but not LLDP. The return value isn't
6813 		 * checked here because a more detailed dcbx agent status is
6814 		 * retrieved and checked in ice_init_dcb() and below.
6815 		 */
6816 		ice_aq_start_stop_dcbx(hw, true, &dcbx_agent_status, NULL);
6817 	}
6818 
6819 	/* This sets hw->port_info->qos_cfg.is_sw_lldp */
6820 	status = ice_init_dcb(hw, true);
6821 
6822 	/* If there is an error, then FW LLDP is not in a usable state */
6823 	if (status != 0 && status != ICE_ERR_NOT_READY) {
6824 		/* Don't print an error message if the return code from the AQ
6825 		 * cmd performed in ice_init_dcb() is is EPERM; that means the
6826 		 * FW LLDP engine is disabled, and that is a valid state.
6827 		 */
6828 		if (!(status == ICE_ERR_AQ_ERROR &&
6829 		      hw->adminq.sq_last_status == ICE_AQ_RC_EPERM)) {
6830 			device_printf(dev, "DCB init failed, err %s aq_err %s\n",
6831 				      ice_status_str(status),
6832 				      ice_aq_str(hw->adminq.sq_last_status));
6833 		}
6834 		hw->port_info->qos_cfg.dcbx_status = ICE_DCBX_STATUS_NOT_STARTED;
6835 	}
6836 
6837 	switch (hw->port_info->qos_cfg.dcbx_status) {
6838 	case ICE_DCBX_STATUS_DIS:
6839 		ice_debug(hw, ICE_DBG_DCB, "DCBX disabled\n");
6840 		break;
6841 	case ICE_DCBX_STATUS_NOT_STARTED:
6842 		ice_debug(hw, ICE_DBG_DCB, "DCBX not started\n");
6843 		break;
6844 	case ICE_DCBX_STATUS_MULTIPLE_PEERS:
6845 		ice_debug(hw, ICE_DBG_DCB, "DCBX detected multiple peers\n");
6846 		break;
6847 	default:
6848 		break;
6849 	}
6850 
6851 	/* LLDP disabled in FW */
6852 	if (hw->port_info->qos_cfg.is_sw_lldp) {
6853 		ice_add_rx_lldp_filter(sc);
6854 		device_printf(dev, "Firmware LLDP agent disabled\n");
6855 	}
6856 }
6857 
6858 /**
6859  * ice_handle_mib_change_event - helper function to log LLDP MIB change events
6860  * @sc: device softc
6861  * @event: event received on a control queue
6862  *
6863  * Prints out the type of an LLDP MIB change event in a DCB debug message.
6864  *
6865  * XXX: Should be extended to do more if the driver decides to notify other SW
6866  * of LLDP MIB changes, or needs to extract info from the MIB.
6867  */
6868 static void
6869 ice_handle_mib_change_event(struct ice_softc *sc, struct ice_rq_event_info *event)
6870 {
6871 	struct ice_aqc_lldp_get_mib *params =
6872 	    (struct ice_aqc_lldp_get_mib *)&event->desc.params.lldp_get_mib;
6873 	u8 mib_type, bridge_type, tx_status;
6874 
6875 	/* XXX: To get the contents of the MIB that caused the event, set the
6876 	 * ICE_DBG_AQ debug mask and read that output
6877 	 */
6878 	static const char* mib_type_strings[] = {
6879 	    "Local MIB",
6880 	    "Remote MIB",
6881 	    "Reserved",
6882 	    "Reserved"
6883 	};
6884 	static const char* bridge_type_strings[] = {
6885 	    "Nearest Bridge",
6886 	    "Non-TPMR Bridge",
6887 	    "Reserved",
6888 	    "Reserved"
6889 	};
6890 	static const char* tx_status_strings[] = {
6891 	    "Port's TX active",
6892 	    "Port's TX suspended and drained",
6893 	    "Reserved",
6894 	    "Port's TX suspended and srained; blocked TC pipe flushed"
6895 	};
6896 
6897 	mib_type = (params->type & ICE_AQ_LLDP_MIB_TYPE_M) >>
6898 	    ICE_AQ_LLDP_MIB_TYPE_S;
6899 	bridge_type = (params->type & ICE_AQ_LLDP_BRID_TYPE_M) >>
6900 	    ICE_AQ_LLDP_BRID_TYPE_S;
6901 	tx_status = (params->type & ICE_AQ_LLDP_TX_M) >>
6902 	    ICE_AQ_LLDP_TX_S;
6903 
6904 	ice_debug(&sc->hw, ICE_DBG_DCB, "LLDP MIB Change Event (%s, %s, %s)\n",
6905 	    mib_type_strings[mib_type], bridge_type_strings[bridge_type],
6906 	    tx_status_strings[tx_status]);
6907 }
6908 
6909 /**
6910  * ice_send_version - Send driver version to firmware
6911  * @sc: the device private softc
6912  *
6913  * Send the driver version to the firmware. This must be called as early as
6914  * possible after ice_init_hw().
6915  */
6916 int
6917 ice_send_version(struct ice_softc *sc)
6918 {
6919 	struct ice_driver_ver driver_version = {0};
6920 	struct ice_hw *hw = &sc->hw;
6921 	device_t dev = sc->dev;
6922 	enum ice_status status;
6923 
6924 	driver_version.major_ver = ice_major_version;
6925 	driver_version.minor_ver = ice_minor_version;
6926 	driver_version.build_ver = ice_patch_version;
6927 	driver_version.subbuild_ver = ice_rc_version;
6928 
6929 	strlcpy((char *)driver_version.driver_string, ice_driver_version,
6930 		sizeof(driver_version.driver_string));
6931 
6932 	status = ice_aq_send_driver_ver(hw, &driver_version, NULL);
6933 	if (status) {
6934 		device_printf(dev, "Unable to send driver version to firmware, err %s aq_err %s\n",
6935 			      ice_status_str(status), ice_aq_str(hw->adminq.sq_last_status));
6936 		return (EIO);
6937 	}
6938 
6939 	return (0);
6940 }
6941 
6942 /**
6943  * ice_handle_lan_overflow_event - helper function to log LAN overflow events
6944  * @sc: device softc
6945  * @event: event received on a control queue
6946  *
6947  * Prints out a message when a LAN overflow event is detected on a receive
6948  * queue.
6949  */
6950 static void
6951 ice_handle_lan_overflow_event(struct ice_softc *sc, struct ice_rq_event_info *event)
6952 {
6953 	struct ice_aqc_event_lan_overflow *params =
6954 	    (struct ice_aqc_event_lan_overflow *)&event->desc.params.lan_overflow;
6955 	struct ice_hw *hw = &sc->hw;
6956 
6957 	ice_debug(hw, ICE_DBG_DCB, "LAN overflow event detected, prtdcb_ruptq=0x%08x, qtx_ctl=0x%08x\n",
6958 		  LE32_TO_CPU(params->prtdcb_ruptq),
6959 		  LE32_TO_CPU(params->qtx_ctl));
6960 }
6961 
6962 /**
6963  * ice_add_ethertype_to_list - Add an Ethertype filter to a filter list
6964  * @vsi: the VSI to target packets to
6965  * @list: the list to add the filter to
6966  * @ethertype: the Ethertype to filter on
6967  * @direction: The direction of the filter (Tx or Rx)
6968  * @action: the action to take
6969  *
6970  * Add an Ethertype filter to a filter list. Used to forward a series of
6971  * filters to the firmware for configuring the switch.
6972  *
6973  * Returns 0 on success, and an error code on failure.
6974  */
6975 static int
6976 ice_add_ethertype_to_list(struct ice_vsi *vsi, struct ice_list_head *list,
6977 			  u16 ethertype, u16 direction,
6978 			  enum ice_sw_fwd_act_type action)
6979 {
6980 	struct ice_fltr_list_entry *entry;
6981 
6982 	MPASS((direction == ICE_FLTR_TX) || (direction == ICE_FLTR_RX));
6983 
6984 	entry = (__typeof(entry))malloc(sizeof(*entry), M_ICE, M_NOWAIT|M_ZERO);
6985 	if (!entry)
6986 		return (ENOMEM);
6987 
6988 	entry->fltr_info.flag = direction;
6989 	entry->fltr_info.src_id = ICE_SRC_ID_VSI;
6990 	entry->fltr_info.lkup_type = ICE_SW_LKUP_ETHERTYPE;
6991 	entry->fltr_info.fltr_act = action;
6992 	entry->fltr_info.vsi_handle = vsi->idx;
6993 	entry->fltr_info.l_data.ethertype_mac.ethertype = ethertype;
6994 
6995 	LIST_ADD(&entry->list_entry, list);
6996 
6997 	return 0;
6998 }
6999 
7000 #define ETHERTYPE_PAUSE_FRAMES 0x8808
7001 #define ETHERTYPE_LLDP_FRAMES 0x88cc
7002 
7003 /**
7004  * ice_cfg_pf_ethertype_filters - Configure switch to drop ethertypes
7005  * @sc: the device private softc
7006  *
7007  * Configure the switch to drop PAUSE frames and LLDP frames transmitted from
7008  * the host. This prevents malicious VFs from sending these frames and being
7009  * able to control or configure the network.
7010  */
7011 int
7012 ice_cfg_pf_ethertype_filters(struct ice_softc *sc)
7013 {
7014 	struct ice_list_head ethertype_list;
7015 	struct ice_vsi *vsi = &sc->pf_vsi;
7016 	struct ice_hw *hw = &sc->hw;
7017 	device_t dev = sc->dev;
7018 	enum ice_status status;
7019 	int err = 0;
7020 
7021 	INIT_LIST_HEAD(&ethertype_list);
7022 
7023 	/*
7024 	 * Note that the switch filters will ignore the VSI index for the drop
7025 	 * action, so we only need to program drop filters once for the main
7026 	 * VSI.
7027 	 */
7028 
7029 	/* Configure switch to drop all Tx pause frames coming from any VSI. */
7030 	if (sc->enable_tx_fc_filter) {
7031 		err = ice_add_ethertype_to_list(vsi, &ethertype_list,
7032 						ETHERTYPE_PAUSE_FRAMES,
7033 						ICE_FLTR_TX, ICE_DROP_PACKET);
7034 		if (err)
7035 			goto free_ethertype_list;
7036 	}
7037 
7038 	/* Configure switch to drop LLDP frames coming from any VSI */
7039 	if (sc->enable_tx_lldp_filter) {
7040 		err = ice_add_ethertype_to_list(vsi, &ethertype_list,
7041 						ETHERTYPE_LLDP_FRAMES,
7042 						ICE_FLTR_TX, ICE_DROP_PACKET);
7043 		if (err)
7044 			goto free_ethertype_list;
7045 	}
7046 
7047 	status = ice_add_eth_mac(hw, &ethertype_list);
7048 	if (status) {
7049 		device_printf(dev,
7050 			      "Failed to add Tx Ethertype filters, err %s aq_err %s\n",
7051 			      ice_status_str(status),
7052 			      ice_aq_str(hw->adminq.sq_last_status));
7053 		err = (EIO);
7054 	}
7055 
7056 free_ethertype_list:
7057 	ice_free_fltr_list(&ethertype_list);
7058 	return err;
7059 }
7060 
7061 /**
7062  * ice_add_rx_lldp_filter - add ethertype filter for Rx LLDP frames
7063  * @sc: the device private structure
7064  *
7065  * Add a switch ethertype filter which forwards the LLDP frames to the main PF
7066  * VSI. Called when the fw_lldp_agent is disabled, to allow the LLDP frames to
7067  * be forwarded to the stack.
7068  */
7069 static void
7070 ice_add_rx_lldp_filter(struct ice_softc *sc)
7071 {
7072 	struct ice_list_head ethertype_list;
7073 	struct ice_vsi *vsi = &sc->pf_vsi;
7074 	struct ice_hw *hw = &sc->hw;
7075 	device_t dev = sc->dev;
7076 	enum ice_status status;
7077 	int err;
7078 	u16 vsi_num;
7079 
7080 	/*
7081 	 * If FW is new enough, use a direct AQ command to perform the filter
7082 	 * addition.
7083 	 */
7084 	if (ice_fw_supports_lldp_fltr_ctrl(hw)) {
7085 		vsi_num = ice_get_hw_vsi_num(hw, vsi->idx);
7086 		status = ice_lldp_fltr_add_remove(hw, vsi_num, true);
7087 		if (status) {
7088 			device_printf(dev,
7089 			    "Failed to add Rx LLDP filter, err %s aq_err %s\n",
7090 			    ice_status_str(status),
7091 			    ice_aq_str(hw->adminq.sq_last_status));
7092 		} else
7093 			ice_set_state(&sc->state,
7094 			    ICE_STATE_LLDP_RX_FLTR_FROM_DRIVER);
7095 		return;
7096 	}
7097 
7098 	INIT_LIST_HEAD(&ethertype_list);
7099 
7100 	/* Forward Rx LLDP frames to the stack */
7101 	err = ice_add_ethertype_to_list(vsi, &ethertype_list,
7102 					ETHERTYPE_LLDP_FRAMES,
7103 					ICE_FLTR_RX, ICE_FWD_TO_VSI);
7104 	if (err) {
7105 		device_printf(dev,
7106 			      "Failed to add Rx LLDP filter, err %s\n",
7107 			      ice_err_str(err));
7108 		goto free_ethertype_list;
7109 	}
7110 
7111 	status = ice_add_eth_mac(hw, &ethertype_list);
7112 	if (status && status != ICE_ERR_ALREADY_EXISTS) {
7113 		device_printf(dev,
7114 			      "Failed to add Rx LLDP filter, err %s aq_err %s\n",
7115 			      ice_status_str(status),
7116 			      ice_aq_str(hw->adminq.sq_last_status));
7117 	} else {
7118 		/*
7119 		 * If status == ICE_ERR_ALREADY_EXISTS, we won't treat an
7120 		 * already existing filter as an error case.
7121 		 */
7122 		ice_set_state(&sc->state, ICE_STATE_LLDP_RX_FLTR_FROM_DRIVER);
7123 	}
7124 
7125 free_ethertype_list:
7126 	ice_free_fltr_list(&ethertype_list);
7127 }
7128 
7129 /**
7130  * ice_del_rx_lldp_filter - Remove ethertype filter for Rx LLDP frames
7131  * @sc: the device private structure
7132  *
7133  * Remove the switch filter forwarding LLDP frames to the main PF VSI, called
7134  * when the firmware LLDP agent is enabled, to stop routing LLDP frames to the
7135  * stack.
7136  */
7137 static void
7138 ice_del_rx_lldp_filter(struct ice_softc *sc)
7139 {
7140 	struct ice_list_head ethertype_list;
7141 	struct ice_vsi *vsi = &sc->pf_vsi;
7142 	struct ice_hw *hw = &sc->hw;
7143 	device_t dev = sc->dev;
7144 	enum ice_status status;
7145 	int err;
7146 	u16 vsi_num;
7147 
7148 	/*
7149 	 * Only in the scenario where the driver added the filter during
7150 	 * this session (while the driver was loaded) would we be able to
7151 	 * delete this filter.
7152 	 */
7153 	if (!ice_test_state(&sc->state, ICE_STATE_LLDP_RX_FLTR_FROM_DRIVER))
7154 		return;
7155 
7156 	/*
7157 	 * If FW is new enough, use a direct AQ command to perform the filter
7158 	 * removal.
7159 	 */
7160 	if (ice_fw_supports_lldp_fltr_ctrl(hw)) {
7161 		vsi_num = ice_get_hw_vsi_num(hw, vsi->idx);
7162 		status = ice_lldp_fltr_add_remove(hw, vsi_num, false);
7163 		if (status) {
7164 			device_printf(dev,
7165 			    "Failed to remove Rx LLDP filter, err %s aq_err %s\n",
7166 			    ice_status_str(status),
7167 			    ice_aq_str(hw->adminq.sq_last_status));
7168 		}
7169 		return;
7170 	}
7171 
7172 	INIT_LIST_HEAD(&ethertype_list);
7173 
7174 	/* Remove filter forwarding Rx LLDP frames to the stack */
7175 	err = ice_add_ethertype_to_list(vsi, &ethertype_list,
7176 					ETHERTYPE_LLDP_FRAMES,
7177 					ICE_FLTR_RX, ICE_FWD_TO_VSI);
7178 	if (err) {
7179 		device_printf(dev,
7180 			      "Failed to remove Rx LLDP filter, err %s\n",
7181 			      ice_err_str(err));
7182 		goto free_ethertype_list;
7183 	}
7184 
7185 	status = ice_remove_eth_mac(hw, &ethertype_list);
7186 	if (status == ICE_ERR_DOES_NOT_EXIST) {
7187 		; /* Don't complain if we try to remove a filter that doesn't exist */
7188 	} else if (status) {
7189 		device_printf(dev,
7190 			      "Failed to remove Rx LLDP filter, err %s aq_err %s\n",
7191 			      ice_status_str(status),
7192 			      ice_aq_str(hw->adminq.sq_last_status));
7193 	}
7194 
7195 free_ethertype_list:
7196 	ice_free_fltr_list(&ethertype_list);
7197 }
7198 
7199 /**
7200  * ice_init_link_configuration -- Setup link in different ways depending
7201  * on whether media is available or not.
7202  * @sc: device private structure
7203  *
7204  * Called at the end of the attach process to either set default link
7205  * parameters if there is media available, or force HW link down and
7206  * set a state bit if there is no media.
7207  */
7208 void
7209 ice_init_link_configuration(struct ice_softc *sc)
7210 {
7211 	struct ice_port_info *pi = sc->hw.port_info;
7212 	struct ice_hw *hw = &sc->hw;
7213 	device_t dev = sc->dev;
7214 	enum ice_status status;
7215 
7216 	pi->phy.get_link_info = true;
7217 	status = ice_get_link_status(pi, &sc->link_up);
7218 	if (status != ICE_SUCCESS) {
7219 		device_printf(dev,
7220 		    "%s: ice_get_link_status failed; status %s, aq_err %s\n",
7221 		    __func__, ice_status_str(status),
7222 		    ice_aq_str(hw->adminq.sq_last_status));
7223 		return;
7224 	}
7225 
7226 	if (pi->phy.link_info.link_info & ICE_AQ_MEDIA_AVAILABLE) {
7227 		ice_clear_state(&sc->state, ICE_STATE_NO_MEDIA);
7228 		/* Apply default link settings */
7229 		ice_apply_saved_phy_cfg(sc, ICE_APPLY_LS_FEC_FC);
7230 	} else {
7231 		 /* Set link down, and poll for media available in timer. This prevents the
7232 		  * driver from receiving spurious link-related events.
7233 		  */
7234 		ice_set_state(&sc->state, ICE_STATE_NO_MEDIA);
7235 		status = ice_aq_set_link_restart_an(pi, false, NULL);
7236 		if (status != ICE_SUCCESS)
7237 			device_printf(dev,
7238 			    "%s: ice_aq_set_link_restart_an: status %s, aq_err %s\n",
7239 			    __func__, ice_status_str(status),
7240 			    ice_aq_str(hw->adminq.sq_last_status));
7241 	}
7242 }
7243 
7244 /**
7245  * ice_apply_saved_phy_req_to_cfg -- Write saved user PHY settings to cfg data
7246  * @sc: device private structure
7247  * @cfg: new PHY config data to be modified
7248  *
7249  * Applies user settings for advertised speeds to the PHY type fields in the
7250  * supplied PHY config struct. It uses the data from pcaps to check if the
7251  * saved settings are invalid and uses the pcaps data instead if they are
7252  * invalid.
7253  */
7254 static int
7255 ice_apply_saved_phy_req_to_cfg(struct ice_softc *sc,
7256 			       struct ice_aqc_set_phy_cfg_data *cfg)
7257 {
7258 	struct ice_phy_data phy_data = { 0 };
7259 	struct ice_port_info *pi = sc->hw.port_info;
7260 	u64 phy_low = 0, phy_high = 0;
7261 	u16 link_speeds;
7262 	int ret;
7263 
7264 	link_speeds = pi->phy.curr_user_speed_req;
7265 
7266 	if (ice_is_bit_set(sc->feat_en, ICE_FEATURE_LINK_MGMT_VER_2)) {
7267 		memset(&phy_data, 0, sizeof(phy_data));
7268 		phy_data.report_mode = ICE_AQC_REPORT_DFLT_CFG;
7269 		phy_data.user_speeds_orig = link_speeds;
7270 		ret = ice_intersect_phy_types_and_speeds(sc, &phy_data);
7271 		if (ret != 0) {
7272 			/* Error message already printed within function */
7273 			return (ret);
7274 		}
7275 		phy_low = phy_data.phy_low_intr;
7276 		phy_high = phy_data.phy_high_intr;
7277 
7278 		if (link_speeds == 0 || phy_data.user_speeds_intr)
7279 			goto finalize_link_speed;
7280 		if (ice_is_bit_set(sc->feat_en, ICE_FEATURE_LENIENT_LINK_MODE)) {
7281 			memset(&phy_data, 0, sizeof(phy_data));
7282 			phy_data.report_mode = ICE_AQC_REPORT_TOPO_CAP_NO_MEDIA;
7283 			phy_data.user_speeds_orig = link_speeds;
7284 			ret = ice_intersect_phy_types_and_speeds(sc, &phy_data);
7285 			if (ret != 0) {
7286 				/* Error message already printed within function */
7287 				return (ret);
7288 			}
7289 			phy_low = phy_data.phy_low_intr;
7290 			phy_high = phy_data.phy_high_intr;
7291 
7292 			if (!phy_data.user_speeds_intr) {
7293 				phy_low = phy_data.phy_low_orig;
7294 				phy_high = phy_data.phy_high_orig;
7295 			}
7296 			goto finalize_link_speed;
7297 		}
7298 		/* If we're here, then it means the benefits of Version 2
7299 		 * link management aren't utilized.  We fall through to
7300 		 * handling Strict Link Mode the same as Version 1 link
7301 		 * management.
7302 		 */
7303 	}
7304 
7305 	memset(&phy_data, 0, sizeof(phy_data));
7306 	if ((link_speeds == 0) &&
7307 	    (sc->ldo_tlv.phy_type_low || sc->ldo_tlv.phy_type_high))
7308 		phy_data.report_mode = ICE_AQC_REPORT_TOPO_CAP_NO_MEDIA;
7309 	else
7310 		phy_data.report_mode = ICE_AQC_REPORT_TOPO_CAP_MEDIA;
7311 	phy_data.user_speeds_orig = link_speeds;
7312 	ret = ice_intersect_phy_types_and_speeds(sc, &phy_data);
7313 	if (ret != 0) {
7314 		/* Error message already printed within function */
7315 		return (ret);
7316 	}
7317 	phy_low = phy_data.phy_low_intr;
7318 	phy_high = phy_data.phy_high_intr;
7319 
7320 	if (!ice_is_bit_set(sc->feat_en, ICE_FEATURE_LENIENT_LINK_MODE)) {
7321 		if (phy_low == 0 && phy_high == 0) {
7322 			device_printf(sc->dev,
7323 			    "The selected speed is not supported by the current media. Please select a link speed that is supported by the current media.\n");
7324 			return (EINVAL);
7325 		}
7326 	} else {
7327 		if (link_speeds == 0) {
7328 			if (sc->ldo_tlv.phy_type_low & phy_low ||
7329 			    sc->ldo_tlv.phy_type_high & phy_high) {
7330 				phy_low &= sc->ldo_tlv.phy_type_low;
7331 				phy_high &= sc->ldo_tlv.phy_type_high;
7332 			}
7333 		} else if (phy_low == 0 && phy_high == 0) {
7334 			memset(&phy_data, 0, sizeof(phy_data));
7335 			phy_data.report_mode = ICE_AQC_REPORT_TOPO_CAP_NO_MEDIA;
7336 			phy_data.user_speeds_orig = link_speeds;
7337 			ret = ice_intersect_phy_types_and_speeds(sc, &phy_data);
7338 			if (ret != 0) {
7339 				/* Error message already printed within function */
7340 				return (ret);
7341 			}
7342 			phy_low = phy_data.phy_low_intr;
7343 			phy_high = phy_data.phy_high_intr;
7344 
7345 			if (!phy_data.user_speeds_intr) {
7346 				phy_low = phy_data.phy_low_orig;
7347 				phy_high = phy_data.phy_high_orig;
7348 			}
7349 		}
7350 	}
7351 
7352 finalize_link_speed:
7353 
7354 	/* Cache new user settings for speeds */
7355 	pi->phy.curr_user_speed_req = phy_data.user_speeds_intr;
7356 	cfg->phy_type_low = htole64(phy_low);
7357 	cfg->phy_type_high = htole64(phy_high);
7358 
7359 	return (ret);
7360 }
7361 
7362 /**
7363  * ice_apply_saved_fec_req_to_cfg -- Write saved user FEC mode to cfg data
7364  * @sc: device private structure
7365  * @cfg: new PHY config data to be modified
7366  *
7367  * Applies user setting for FEC mode to PHY config struct. It uses the data
7368  * from pcaps to check if the saved settings are invalid and uses the pcaps
7369  * data instead if they are invalid.
7370  */
7371 static int
7372 ice_apply_saved_fec_req_to_cfg(struct ice_softc *sc,
7373 			       struct ice_aqc_set_phy_cfg_data *cfg)
7374 {
7375 	struct ice_port_info *pi = sc->hw.port_info;
7376 	enum ice_status status;
7377 
7378 	cfg->caps &= ~ICE_AQC_PHY_EN_AUTO_FEC;
7379 	status = ice_cfg_phy_fec(pi, cfg, pi->phy.curr_user_fec_req);
7380 	if (status)
7381 		return (EIO);
7382 
7383 	return (0);
7384 }
7385 
7386 /**
7387  * ice_apply_saved_fc_req_to_cfg -- Write saved user flow control mode to cfg data
7388  * @pi: port info struct
7389  * @cfg: new PHY config data to be modified
7390  *
7391  * Applies user setting for flow control mode to PHY config struct. There are
7392  * no invalid flow control mode settings; if there are, then this function
7393  * treats them like "ICE_FC_NONE".
7394  */
7395 static void
7396 ice_apply_saved_fc_req_to_cfg(struct ice_port_info *pi,
7397 			      struct ice_aqc_set_phy_cfg_data *cfg)
7398 {
7399 	cfg->caps &= ~(ICE_AQ_PHY_ENA_TX_PAUSE_ABILITY |
7400 		       ICE_AQ_PHY_ENA_RX_PAUSE_ABILITY);
7401 
7402 	switch (pi->phy.curr_user_fc_req) {
7403 	case ICE_FC_FULL:
7404 		cfg->caps |= ICE_AQ_PHY_ENA_TX_PAUSE_ABILITY |
7405 			     ICE_AQ_PHY_ENA_RX_PAUSE_ABILITY;
7406 		break;
7407 	case ICE_FC_RX_PAUSE:
7408 		cfg->caps |= ICE_AQ_PHY_ENA_RX_PAUSE_ABILITY;
7409 		break;
7410 	case ICE_FC_TX_PAUSE:
7411 		cfg->caps |= ICE_AQ_PHY_ENA_TX_PAUSE_ABILITY;
7412 		break;
7413 	default:
7414 		/* ICE_FC_NONE */
7415 		break;
7416 	}
7417 }
7418 
7419 /**
7420  * ice_apply_saved_phy_cfg -- Re-apply user PHY config settings
7421  * @sc: device private structure
7422  * @settings: which settings to apply
7423  *
7424  * Applies user settings for advertised speeds, FEC mode, and flow
7425  * control mode to a PHY config struct; it uses the data from pcaps
7426  * to check if the saved settings are invalid and uses the pcaps
7427  * data instead if they are invalid.
7428  *
7429  * For things like sysctls where only one setting needs to be
7430  * updated, the bitmap allows the caller to specify which setting
7431  * to update.
7432  */
7433 int
7434 ice_apply_saved_phy_cfg(struct ice_softc *sc, u8 settings)
7435 {
7436 	struct ice_aqc_set_phy_cfg_data cfg = { 0 };
7437 	struct ice_port_info *pi = sc->hw.port_info;
7438 	struct ice_aqc_get_phy_caps_data pcaps = { 0 };
7439 	struct ice_hw *hw = &sc->hw;
7440 	device_t dev = sc->dev;
7441 	u64 phy_low, phy_high;
7442 	enum ice_status status;
7443 	enum ice_fec_mode dflt_fec_mode;
7444 	enum ice_fc_mode dflt_fc_mode;
7445 	u16 dflt_user_speed;
7446 
7447 	if (!settings || settings > ICE_APPLY_LS_FEC_FC) {
7448 		ice_debug(hw, ICE_DBG_LINK, "Settings out-of-bounds: %u\n",
7449 		    settings);
7450 	}
7451 
7452 	status = ice_aq_get_phy_caps(pi, false, ICE_AQC_REPORT_ACTIVE_CFG,
7453 				     &pcaps, NULL);
7454 	if (status != ICE_SUCCESS) {
7455 		device_printf(dev,
7456 		    "%s: ice_aq_get_phy_caps (ACTIVE) failed; status %s, aq_err %s\n",
7457 		    __func__, ice_status_str(status),
7458 		    ice_aq_str(hw->adminq.sq_last_status));
7459 		return (EIO);
7460 	}
7461 
7462 	phy_low = le64toh(pcaps.phy_type_low);
7463 	phy_high = le64toh(pcaps.phy_type_high);
7464 
7465 	/* Save off initial config parameters */
7466 	dflt_user_speed = ice_aq_phy_types_to_link_speeds(phy_low, phy_high);
7467 	dflt_fec_mode = ice_caps_to_fec_mode(pcaps.caps, pcaps.link_fec_options);
7468 	dflt_fc_mode = ice_caps_to_fc_mode(pcaps.caps);
7469 
7470 	/* Setup new PHY config */
7471 	ice_copy_phy_caps_to_cfg(pi, &pcaps, &cfg);
7472 
7473 	/* On error, restore active configuration values */
7474 	if ((settings & ICE_APPLY_LS) &&
7475 	    ice_apply_saved_phy_req_to_cfg(sc, &cfg)) {
7476 		pi->phy.curr_user_speed_req = dflt_user_speed;
7477 		cfg.phy_type_low = pcaps.phy_type_low;
7478 		cfg.phy_type_high = pcaps.phy_type_high;
7479 	}
7480 	if ((settings & ICE_APPLY_FEC) &&
7481 	    ice_apply_saved_fec_req_to_cfg(sc, &cfg)) {
7482 		pi->phy.curr_user_fec_req = dflt_fec_mode;
7483 	}
7484 	if (settings & ICE_APPLY_FC) {
7485 		/* No real error indicators for this process,
7486 		 * so we'll just have to assume it works. */
7487 		ice_apply_saved_fc_req_to_cfg(pi, &cfg);
7488 	}
7489 
7490 	/* Enable link and re-negotiate it */
7491 	cfg.caps |= ICE_AQ_PHY_ENA_AUTO_LINK_UPDT | ICE_AQ_PHY_ENA_LINK;
7492 
7493 	status = ice_aq_set_phy_cfg(hw, pi, &cfg, NULL);
7494 	if (status != ICE_SUCCESS) {
7495 		/* Don't indicate failure if there's no media in the port.
7496 		 * The settings have been saved and will apply when media
7497 		 * is inserted.
7498 		 */
7499 		if ((status == ICE_ERR_AQ_ERROR) &&
7500 		    (hw->adminq.sq_last_status == ICE_AQ_RC_EBUSY)) {
7501 			device_printf(dev,
7502 			    "%s: Setting will be applied when media is inserted\n",
7503 			    __func__);
7504 			return (0);
7505 		} else {
7506 			device_printf(dev,
7507 			    "%s: ice_aq_set_phy_cfg failed; status %s, aq_err %s\n",
7508 			    __func__, ice_status_str(status),
7509 			    ice_aq_str(hw->adminq.sq_last_status));
7510 			return (EIO);
7511 		}
7512 	}
7513 
7514 	return (0);
7515 }
7516 
7517 /**
7518  * ice_print_ldo_tlv - Print out LDO TLV information
7519  * @sc: device private structure
7520  * @tlv: LDO TLV information from the adapter NVM
7521  *
7522  * Dump out the information in tlv to the kernel message buffer; intended for
7523  * debugging purposes.
7524  */
7525 static void
7526 ice_print_ldo_tlv(struct ice_softc *sc, struct ice_link_default_override_tlv *tlv)
7527 {
7528 	device_t dev = sc->dev;
7529 
7530 	device_printf(dev, "TLV: -options     0x%02x\n", tlv->options);
7531 	device_printf(dev, "     -phy_config  0x%02x\n", tlv->phy_config);
7532 	device_printf(dev, "     -fec_options 0x%02x\n", tlv->fec_options);
7533 	device_printf(dev, "     -phy_high    0x%016llx\n",
7534 	    (unsigned long long)tlv->phy_type_high);
7535 	device_printf(dev, "     -phy_low     0x%016llx\n",
7536 	    (unsigned long long)tlv->phy_type_low);
7537 }
7538 
7539 /**
7540  * ice_set_link_management_mode -- Strict or lenient link management
7541  * @sc: device private structure
7542  *
7543  * Some NVMs give the adapter the option to advertise a superset of link
7544  * configurations.  This checks to see if that option is enabled.
7545  * Further, the NVM could also provide a specific set of configurations
7546  * to try; these are cached in the driver's private structure if they
7547  * are available.
7548  */
7549 void
7550 ice_set_link_management_mode(struct ice_softc *sc)
7551 {
7552 	struct ice_port_info *pi = sc->hw.port_info;
7553 	device_t dev = sc->dev;
7554 	struct ice_link_default_override_tlv tlv = { 0 };
7555 	enum ice_status status;
7556 
7557 	/* Port must be in strict mode if FW version is below a certain
7558 	 * version. (i.e. Don't set lenient mode features)
7559 	 */
7560 	if (!(ice_fw_supports_link_override(&sc->hw)))
7561 		return;
7562 
7563 	status = ice_get_link_default_override(&tlv, pi);
7564 	if (status != ICE_SUCCESS) {
7565 		device_printf(dev,
7566 		    "%s: ice_get_link_default_override failed; status %s, aq_err %s\n",
7567 		    __func__, ice_status_str(status),
7568 		    ice_aq_str(sc->hw.adminq.sq_last_status));
7569 		return;
7570 	}
7571 
7572 	if (sc->hw.debug_mask & ICE_DBG_LINK)
7573 		ice_print_ldo_tlv(sc, &tlv);
7574 
7575 	/* Set lenient link mode */
7576 	if (ice_is_bit_set(sc->feat_cap, ICE_FEATURE_LENIENT_LINK_MODE) &&
7577 	    (!(tlv.options & ICE_LINK_OVERRIDE_STRICT_MODE)))
7578 		ice_set_bit(ICE_FEATURE_LENIENT_LINK_MODE, sc->feat_en);
7579 
7580 	/* FW supports reporting a default configuration */
7581 	if (ice_is_bit_set(sc->feat_cap, ICE_FEATURE_LINK_MGMT_VER_2) &&
7582 	    ice_fw_supports_report_dflt_cfg(&sc->hw)) {
7583 		ice_set_bit(ICE_FEATURE_LINK_MGMT_VER_2, sc->feat_en);
7584 		/* Knowing we're at a high enough firmware revision to
7585 		 * support this link management configuration, we don't
7586 		 * need to check/support earlier versions.
7587 		 */
7588 		return;
7589 	}
7590 
7591 	/* Default overrides only work if in lenient link mode */
7592 	if (ice_is_bit_set(sc->feat_cap, ICE_FEATURE_LINK_MGMT_VER_1) &&
7593 	    ice_is_bit_set(sc->feat_en, ICE_FEATURE_LENIENT_LINK_MODE) &&
7594 	    (tlv.options & ICE_LINK_OVERRIDE_EN))
7595 		ice_set_bit(ICE_FEATURE_LINK_MGMT_VER_1, sc->feat_en);
7596 
7597 	/* Cache the LDO TLV structure in the driver, since it
7598 	 * won't change during the driver's lifetime.
7599 	 */
7600 	sc->ldo_tlv = tlv;
7601 }
7602 
7603 /**
7604  * ice_init_saved_phy_cfg -- Set cached user PHY cfg settings with NVM defaults
7605  * @sc: device private structure
7606  *
7607  * This should be called before the tunables for these link settings
7608  * (e.g. advertise_speed) are added -- so that these defaults don't overwrite
7609  * the cached values that the sysctl handlers will write.
7610  *
7611  * This also needs to be called before ice_init_link_configuration, to ensure
7612  * that there are sane values that can be written if there is media available
7613  * in the port.
7614  */
7615 void
7616 ice_init_saved_phy_cfg(struct ice_softc *sc)
7617 {
7618 	struct ice_port_info *pi = sc->hw.port_info;
7619 	struct ice_aqc_get_phy_caps_data pcaps = { 0 };
7620 	struct ice_hw *hw = &sc->hw;
7621 	device_t dev = sc->dev;
7622 	enum ice_status status;
7623 	u64 phy_low, phy_high;
7624 	u8 report_mode = ICE_AQC_REPORT_TOPO_CAP_MEDIA;
7625 
7626 	if (ice_is_bit_set(sc->feat_en, ICE_FEATURE_LINK_MGMT_VER_2))
7627 		report_mode = ICE_AQC_REPORT_DFLT_CFG;
7628 	status = ice_aq_get_phy_caps(pi, false, report_mode, &pcaps, NULL);
7629 	if (status != ICE_SUCCESS) {
7630 		device_printf(dev,
7631 		    "%s: ice_aq_get_phy_caps (%s) failed; status %s, aq_err %s\n",
7632 		    __func__,
7633 		    report_mode == ICE_AQC_REPORT_DFLT_CFG ? "DFLT" : "w/MEDIA",
7634 		    ice_status_str(status),
7635 		    ice_aq_str(hw->adminq.sq_last_status));
7636 		return;
7637 	}
7638 
7639 	phy_low = le64toh(pcaps.phy_type_low);
7640 	phy_high = le64toh(pcaps.phy_type_high);
7641 
7642 	/* Save off initial config parameters */
7643 	pi->phy.curr_user_speed_req =
7644 	   ice_aq_phy_types_to_link_speeds(phy_low, phy_high);
7645 	pi->phy.curr_user_fec_req = ice_caps_to_fec_mode(pcaps.caps,
7646 	    pcaps.link_fec_options);
7647 	pi->phy.curr_user_fc_req = ice_caps_to_fc_mode(pcaps.caps);
7648 }
7649 
7650 /**
7651  * ice_module_init - Driver callback to handle module load
7652  *
7653  * Callback for handling module load events. This function should initialize
7654  * any data structures that are used for the life of the device driver.
7655  */
7656 static int
7657 ice_module_init(void)
7658 {
7659 	return (0);
7660 }
7661 
7662 /**
7663  * ice_module_exit - Driver callback to handle module exit
7664  *
7665  * Callback for handling module unload events. This function should release
7666  * any resources initialized during ice_module_init.
7667  *
7668  * If this function returns non-zero, the module will not be unloaded. It
7669  * should only return such a value if the module cannot be unloaded at all,
7670  * such as due to outstanding memory references that cannot be revoked.
7671  */
7672 static int
7673 ice_module_exit(void)
7674 {
7675 	return (0);
7676 }
7677 
7678 /**
7679  * ice_module_event_handler - Callback for module events
7680  * @mod: unused module_t parameter
7681  * @what: the event requested
7682  * @arg: unused event argument
7683  *
7684  * Callback used to handle module events from the stack. Used to allow the
7685  * driver to define custom behavior that should happen at module load and
7686  * unload.
7687  */
7688 int
7689 ice_module_event_handler(module_t __unused mod, int what, void __unused *arg)
7690 {
7691 	switch (what) {
7692 	case MOD_LOAD:
7693 		return ice_module_init();
7694 	case MOD_UNLOAD:
7695 		return ice_module_exit();
7696 	default:
7697 		/* TODO: do we need to handle MOD_QUIESCE and MOD_SHUTDOWN? */
7698 		return (EOPNOTSUPP);
7699 	}
7700 }
7701 
7702 /**
7703  * ice_handle_nvm_access_ioctl - Handle an NVM access ioctl request
7704  * @sc: the device private softc
7705  * @ifd: ifdrv ioctl request pointer
7706  */
7707 int
7708 ice_handle_nvm_access_ioctl(struct ice_softc *sc, struct ifdrv *ifd)
7709 {
7710 	union ice_nvm_access_data *data;
7711 	struct ice_nvm_access_cmd *cmd;
7712 	size_t ifd_len = ifd->ifd_len, malloc_len;
7713 	struct ice_hw *hw = &sc->hw;
7714 	device_t dev = sc->dev;
7715 	enum ice_status status;
7716 	u8 *nvm_buffer;
7717 	int err;
7718 
7719 	/*
7720 	 * ifioctl forwards SIOCxDRVSPEC to iflib without performing
7721 	 * a privilege check. In turn, iflib forwards the ioctl to the driver
7722 	 * without performing a privilege check. Perform one here to ensure
7723 	 * that non-privileged threads cannot access this interface.
7724 	 */
7725 	err = priv_check(curthread, PRIV_DRIVER);
7726 	if (err)
7727 		return (err);
7728 
7729 	if (ifd_len < sizeof(struct ice_nvm_access_cmd)) {
7730 		device_printf(dev, "%s: ifdrv length is too small. Got %zu, but expected %zu\n",
7731 			      __func__, ifd_len, sizeof(struct ice_nvm_access_cmd));
7732 		return (EINVAL);
7733 	}
7734 
7735 	if (ifd->ifd_data == NULL) {
7736 		device_printf(dev, "%s: ifd data buffer not present.\n",
7737 			      __func__);
7738 		return (EINVAL);
7739 	}
7740 
7741 	/*
7742 	 * If everything works correctly, ice_handle_nvm_access should not
7743 	 * modify data past the size of the ioctl length. However, it could
7744 	 * lead to memory corruption if it did. Make sure to allocate at least
7745 	 * enough space for the command and data regardless. This
7746 	 * ensures that any access to the data union will not access invalid
7747 	 * memory.
7748 	 */
7749 	malloc_len = max(ifd_len, sizeof(*data) + sizeof(*cmd));
7750 
7751 	nvm_buffer = (u8 *)malloc(malloc_len, M_ICE, M_ZERO | M_WAITOK);
7752 	if (!nvm_buffer)
7753 		return (ENOMEM);
7754 
7755 	/* Copy the NVM access command and data in from user space */
7756 	/* coverity[tainted_data_argument] */
7757 	err = copyin(ifd->ifd_data, nvm_buffer, ifd_len);
7758 	if (err) {
7759 		device_printf(dev, "%s: Copying request from user space failed, err %s\n",
7760 			      __func__, ice_err_str(err));
7761 		goto cleanup_free_nvm_buffer;
7762 	}
7763 
7764 	/*
7765 	 * The NVM command structure is immediately followed by data which
7766 	 * varies in size based on the command.
7767 	 */
7768 	cmd = (struct ice_nvm_access_cmd *)nvm_buffer;
7769 	data = (union ice_nvm_access_data *)(nvm_buffer + sizeof(struct ice_nvm_access_cmd));
7770 
7771 	/* Handle the NVM access request */
7772 	status = ice_handle_nvm_access(hw, cmd, data);
7773 	if (status)
7774 		ice_debug(hw, ICE_DBG_NVM,
7775 			  "NVM access request failed, err %s\n",
7776 			  ice_status_str(status));
7777 
7778 	/* Copy the possibly modified contents of the handled request out */
7779 	err = copyout(nvm_buffer, ifd->ifd_data, ifd_len);
7780 	if (err) {
7781 		device_printf(dev, "%s: Copying response back to user space failed, err %s\n",
7782 			      __func__, ice_err_str(err));
7783 		goto cleanup_free_nvm_buffer;
7784 	}
7785 
7786 	/* Convert private status to an error code for proper ioctl response */
7787 	switch (status) {
7788 	case ICE_SUCCESS:
7789 		err = (0);
7790 		break;
7791 	case ICE_ERR_NO_MEMORY:
7792 		err = (ENOMEM);
7793 		break;
7794 	case ICE_ERR_OUT_OF_RANGE:
7795 		err = (ENOTTY);
7796 		break;
7797 	case ICE_ERR_PARAM:
7798 	default:
7799 		err = (EINVAL);
7800 		break;
7801 	}
7802 
7803 cleanup_free_nvm_buffer:
7804 	free(nvm_buffer, M_ICE);
7805 	return err;
7806 }
7807 
7808 /**
7809  * ice_read_sff_eeprom - Read data from SFF eeprom
7810  * @sc: device softc
7811  * @dev_addr: I2C device address (typically 0xA0 or 0xA2)
7812  * @offset: offset into the eeprom
7813  * @data: pointer to data buffer to store read data in
7814  * @length: length to read; max length is 16
7815  *
7816  * Read from the SFF eeprom in the module for this PF's port. For more details
7817  * on the contents of an SFF eeprom, refer to SFF-8724 (SFP), SFF-8636 (QSFP),
7818  * and SFF-8024 (both).
7819  */
7820 int
7821 ice_read_sff_eeprom(struct ice_softc *sc, u16 dev_addr, u16 offset, u8* data, u16 length)
7822 {
7823 	struct ice_hw *hw = &sc->hw;
7824 	int ret = 0, retries = 0;
7825 	enum ice_status status;
7826 
7827 	if (length > 16)
7828 		return (EINVAL);
7829 
7830 	if (ice_test_state(&sc->state, ICE_STATE_RECOVERY_MODE))
7831 		return (ENOSYS);
7832 
7833 	if (ice_test_state(&sc->state, ICE_STATE_NO_MEDIA))
7834 		return (ENXIO);
7835 
7836 	do {
7837 		status = ice_aq_sff_eeprom(hw, 0, dev_addr,
7838 					   offset, 0, 0, data, length,
7839 					   false, NULL);
7840 		if (!status) {
7841 			ret = 0;
7842 			break;
7843 		}
7844 		if (status == ICE_ERR_AQ_ERROR &&
7845 		    hw->adminq.sq_last_status == ICE_AQ_RC_EBUSY) {
7846 			ret = EBUSY;
7847 			continue;
7848 		}
7849 		if (status == ICE_ERR_AQ_ERROR &&
7850 		    hw->adminq.sq_last_status == ICE_AQ_RC_EACCES) {
7851 			/* FW says I2C access isn't supported */
7852 			ret = EACCES;
7853 			break;
7854 		}
7855 		if (status == ICE_ERR_AQ_ERROR &&
7856 		    hw->adminq.sq_last_status == ICE_AQ_RC_EPERM) {
7857 			device_printf(sc->dev,
7858 				  "%s: Module pointer location specified in command does not permit the required operation.\n",
7859 				  __func__);
7860 			ret = EPERM;
7861 			break;
7862 		} else {
7863 			device_printf(sc->dev,
7864 				  "%s: Error reading I2C data: err %s aq_err %s\n",
7865 				  __func__, ice_status_str(status),
7866 				  ice_aq_str(hw->adminq.sq_last_status));
7867 			ret = EIO;
7868 			break;
7869 		}
7870 	} while (retries++ < ICE_I2C_MAX_RETRIES);
7871 
7872 	if (ret == EBUSY)
7873 		device_printf(sc->dev,
7874 			  "%s: Error reading I2C data after %d retries\n",
7875 			  __func__, ICE_I2C_MAX_RETRIES);
7876 
7877 	return (ret);
7878 }
7879 
7880 /**
7881  * ice_handle_i2c_req - Driver independent I2C request handler
7882  * @sc: device softc
7883  * @req: The I2C parameters to use
7884  *
7885  * Read from the port's I2C eeprom using the parameters from the ioctl.
7886  */
7887 int
7888 ice_handle_i2c_req(struct ice_softc *sc, struct ifi2creq *req)
7889 {
7890 	return ice_read_sff_eeprom(sc, req->dev_addr, req->offset, req->data, req->len);
7891 }
7892 
7893 /**
7894  * ice_sysctl_read_i2c_diag_data - Read some module diagnostic data via i2c
7895  * @oidp: sysctl oid structure
7896  * @arg1: pointer to private data structure
7897  * @arg2: unused
7898  * @req: sysctl request pointer
7899  *
7900  * Read 8 bytes of diagnostic data from the SFF eeprom in the (Q)SFP module
7901  * inserted into the port.
7902  *
7903  *             | SFP A2  | QSFP Lower Page
7904  * ------------|---------|----------------
7905  * Temperature | 96-97	 | 22-23
7906  * Vcc         | 98-99   | 26-27
7907  * TX power    | 102-103 | 34-35..40-41
7908  * RX power    | 104-105 | 50-51..56-57
7909  */
7910 static int
7911 ice_sysctl_read_i2c_diag_data(SYSCTL_HANDLER_ARGS)
7912 {
7913 	struct ice_softc *sc = (struct ice_softc *)arg1;
7914 	device_t dev = sc->dev;
7915 	struct sbuf *sbuf;
7916 	int ret;
7917 	u8 data[16];
7918 
7919 	UNREFERENCED_PARAMETER(arg2);
7920 	UNREFERENCED_PARAMETER(oidp);
7921 
7922 	if (ice_driver_is_detaching(sc))
7923 		return (ESHUTDOWN);
7924 
7925 	if (req->oldptr == NULL) {
7926 		ret = SYSCTL_OUT(req, 0, 128);
7927 		return (ret);
7928 	}
7929 
7930 	ret = ice_read_sff_eeprom(sc, 0xA0, 0, data, 1);
7931 	if (ret)
7932 		return (ret);
7933 
7934 	/* 0x3 for SFP; 0xD/0x11 for QSFP+/QSFP28 */
7935 	if (data[0] == 0x3) {
7936 		/*
7937 		 * Check for:
7938 		 * - Internally calibrated data
7939 		 * - Diagnostic monitoring is implemented
7940 		 */
7941 		ice_read_sff_eeprom(sc, 0xA0, 92, data, 1);
7942 		if (!(data[0] & 0x60)) {
7943 			device_printf(dev, "Module doesn't support diagnostics: 0xA0[92] = %02X\n", data[0]);
7944 			return (ENODEV);
7945 		}
7946 
7947 		sbuf = sbuf_new_for_sysctl(NULL, NULL, 128, req);
7948 
7949 		ice_read_sff_eeprom(sc, 0xA2, 96, data, 4);
7950 		for (int i = 0; i < 4; i++)
7951 			sbuf_printf(sbuf, "%02X ", data[i]);
7952 
7953 		ice_read_sff_eeprom(sc, 0xA2, 102, data, 4);
7954 		for (int i = 0; i < 4; i++)
7955 			sbuf_printf(sbuf, "%02X ", data[i]);
7956 	} else if (data[0] == 0xD || data[0] == 0x11) {
7957 		/*
7958 		 * QSFP+ modules are always internally calibrated, and must indicate
7959 		 * what types of diagnostic monitoring are implemented
7960 		 */
7961 		sbuf = sbuf_new_for_sysctl(NULL, NULL, 128, req);
7962 
7963 		ice_read_sff_eeprom(sc, 0xA0, 22, data, 2);
7964 		for (int i = 0; i < 2; i++)
7965 			sbuf_printf(sbuf, "%02X ", data[i]);
7966 
7967 		ice_read_sff_eeprom(sc, 0xA0, 26, data, 2);
7968 		for (int i = 0; i < 2; i++)
7969 			sbuf_printf(sbuf, "%02X ", data[i]);
7970 
7971 		ice_read_sff_eeprom(sc, 0xA0, 34, data, 2);
7972 		for (int i = 0; i < 2; i++)
7973 			sbuf_printf(sbuf, "%02X ", data[i]);
7974 
7975 		ice_read_sff_eeprom(sc, 0xA0, 50, data, 2);
7976 		for (int i = 0; i < 2; i++)
7977 			sbuf_printf(sbuf, "%02X ", data[i]);
7978 	} else {
7979 		device_printf(dev, "Module is not SFP/SFP+/SFP28/QSFP+ (%02X)\n", data[0]);
7980 		return (ENODEV);
7981 	}
7982 
7983 	sbuf_finish(sbuf);
7984 	sbuf_delete(sbuf);
7985 
7986 	return (0);
7987 }
7988 
7989 /**
7990  * ice_alloc_intr_tracking - Setup interrupt tracking structures
7991  * @sc: device softc structure
7992  *
7993  * Sets up the resource manager for keeping track of interrupt allocations,
7994  * and initializes the tracking maps for the PF's interrupt allocations.
7995  *
7996  * Unlike the scheme for queues, this is done in one step since both the
7997  * manager and the maps both have the same lifetime.
7998  *
7999  * @returns 0 on success, or an error code on failure.
8000  */
8001 int
8002 ice_alloc_intr_tracking(struct ice_softc *sc)
8003 {
8004 	struct ice_hw *hw = &sc->hw;
8005 	device_t dev = sc->dev;
8006 	int err;
8007 
8008 	/* Initialize the interrupt allocation manager */
8009 	err = ice_resmgr_init_contig_only(&sc->imgr,
8010 	    hw->func_caps.common_cap.num_msix_vectors);
8011 	if (err) {
8012 		device_printf(dev, "Unable to initialize PF interrupt manager: %s\n",
8013 			      ice_err_str(err));
8014 		return (err);
8015 	}
8016 
8017 	/* Allocate PF interrupt mapping storage */
8018 	if (!(sc->pf_imap =
8019 	      (u16 *)malloc(sizeof(u16) * hw->func_caps.common_cap.num_msix_vectors,
8020 	      M_ICE, M_NOWAIT))) {
8021 		device_printf(dev, "Unable to allocate PF imap memory\n");
8022 		err = ENOMEM;
8023 		goto free_imgr;
8024 	}
8025 	for (u32 i = 0; i < hw->func_caps.common_cap.num_msix_vectors; i++) {
8026 		sc->pf_imap[i] = ICE_INVALID_RES_IDX;
8027 	}
8028 
8029 	return (0);
8030 
8031 free_imgr:
8032 	ice_resmgr_destroy(&sc->imgr);
8033 	return (err);
8034 }
8035 
8036 /**
8037  * ice_free_intr_tracking - Free PF interrupt tracking structures
8038  * @sc: device softc structure
8039  *
8040  * Frees the interrupt resource allocation manager and the PF's owned maps.
8041  *
8042  * VF maps are released when the owning VF's are destroyed, which should always
8043  * happen before this function is called.
8044  */
8045 void
8046 ice_free_intr_tracking(struct ice_softc *sc)
8047 {
8048 	if (sc->pf_imap) {
8049 		ice_resmgr_release_map(&sc->imgr, sc->pf_imap,
8050 				       sc->lan_vectors);
8051 		free(sc->pf_imap, M_ICE);
8052 		sc->pf_imap = NULL;
8053 	}
8054 
8055 	ice_resmgr_destroy(&sc->imgr);
8056 }
8057 
8058 /**
8059  * ice_apply_supported_speed_filter - Mask off unsupported speeds
8060  * @report_speeds: bit-field for the desired link speeds
8061  *
8062  * Given a bitmap of the desired lenient mode link speeds,
8063  * this function will mask off the speeds that are not currently
8064  * supported by the device.
8065  */
8066 static u16
8067 ice_apply_supported_speed_filter(u16 report_speeds)
8068 {
8069 	u16 speed_mask;
8070 
8071 	/* We won't offer anything lower than 1G for any part,
8072 	 * but we also won't offer anything under 25G for 100G
8073 	 * parts or under 10G for 50G parts.
8074 	 */
8075 	speed_mask = ~((u16)ICE_AQ_LINK_SPEED_1000MB - 1);
8076 	if (report_speeds & ICE_AQ_LINK_SPEED_50GB)
8077 		speed_mask = ~((u16)ICE_AQ_LINK_SPEED_10GB - 1);
8078 	if (report_speeds & ICE_AQ_LINK_SPEED_100GB)
8079 		speed_mask = ~((u16)ICE_AQ_LINK_SPEED_25GB - 1);
8080 	return (report_speeds & speed_mask);
8081 }
8082 
8083 /**
8084  * ice_init_health_events - Enable FW health event reporting
8085  * @sc: device softc
8086  *
8087  * Will try to enable firmware health event reporting, but shouldn't
8088  * cause any grief (to the caller) if this fails.
8089  */
8090 void
8091 ice_init_health_events(struct ice_softc *sc)
8092 {
8093 	enum ice_status status;
8094 	u8 health_mask;
8095 
8096 	if ((!ice_is_bit_set(sc->feat_cap, ICE_FEATURE_HEALTH_STATUS)) ||
8097 	    (!sc->enable_health_events))
8098 		return;
8099 
8100 	health_mask = ICE_AQC_HEALTH_STATUS_SET_PF_SPECIFIC_MASK |
8101 		      ICE_AQC_HEALTH_STATUS_SET_GLOBAL_MASK;
8102 
8103 	status = ice_aq_set_health_status_config(&sc->hw, health_mask, NULL);
8104 	if (status)
8105 		device_printf(sc->dev,
8106 		    "Failed to enable firmware health events, err %s aq_err %s\n",
8107 		    ice_status_str(status),
8108 		    ice_aq_str(sc->hw.adminq.sq_last_status));
8109 	else
8110 		ice_set_bit(ICE_FEATURE_HEALTH_STATUS, sc->feat_en);
8111 }
8112 
8113 /**
8114  * ice_print_health_status_string - Print message for given FW health event
8115  * @dev: the PCIe device
8116  * @elem: health status element containing status code
8117  *
8118  * A rather large list of possible health status codes and their associated
8119  * messages.
8120  */
8121 static void
8122 ice_print_health_status_string(device_t dev,
8123 			       struct ice_aqc_health_status_elem *elem)
8124 {
8125 	u16 status_code = le16toh(elem->health_status_code);
8126 
8127 	switch (status_code) {
8128 	case ICE_AQC_HEALTH_STATUS_INFO_RECOVERY:
8129 		device_printf(dev, "The device is in firmware recovery mode.\n");
8130 		device_printf(dev, "Possible Solution: Update to the latest NVM image.\n");
8131 		break;
8132 	case ICE_AQC_HEALTH_STATUS_ERR_FLASH_ACCESS:
8133 		device_printf(dev, "The flash chip cannot be accessed.\n");
8134 		device_printf(dev, "Possible Solution: If issue persists, call customer support.\n");
8135 		break;
8136 	case ICE_AQC_HEALTH_STATUS_ERR_NVM_AUTH:
8137 		device_printf(dev, "NVM authentication failed.\n");
8138 		device_printf(dev, "Possible Solution: Update to the latest NVM image.\n");
8139 		break;
8140 	case ICE_AQC_HEALTH_STATUS_ERR_OROM_AUTH:
8141 		device_printf(dev, "Option ROM authentication failed.\n");
8142 		device_printf(dev, "Possible Solution: Update to the latest NVM image.\n");
8143 		break;
8144 	case ICE_AQC_HEALTH_STATUS_ERR_DDP_AUTH:
8145 		device_printf(dev, "DDP package failed.\n");
8146 		device_printf(dev, "Possible Solution: Update to latest base driver and DDP package.\n");
8147 		break;
8148 	case ICE_AQC_HEALTH_STATUS_ERR_NVM_COMPAT:
8149 		device_printf(dev, "NVM image is incompatible.\n");
8150 		device_printf(dev, "Possible Solution: Update to the latest NVM image.\n");
8151 		break;
8152 	case ICE_AQC_HEALTH_STATUS_ERR_OROM_COMPAT:
8153 		device_printf(dev, "Option ROM is incompatible.\n");
8154 		device_printf(dev, "Possible Solution: Update to the latest NVM image.\n");
8155 		break;
8156 	case ICE_AQC_HEALTH_STATUS_ERR_DCB_MIB:
8157 		device_printf(dev, "Supplied MIB file is invalid. DCB reverted to default configuration.\n");
8158 		device_printf(dev, "Possible Solution: Disable FW-LLDP and check DCBx system configuration.\n");
8159 		break;
8160 	case ICE_AQC_HEALTH_STATUS_ERR_UNKNOWN_MOD_STRICT:
8161 		device_printf(dev, "An unsupported module was detected.\n");
8162 		device_printf(dev, "Possible Solution 1: Check your cable connection.\n");
8163 		device_printf(dev, "Possible Solution 2: Change or replace the module or cable.\n");
8164 		break;
8165 	case ICE_AQC_HEALTH_STATUS_ERR_MOD_TYPE:
8166 		device_printf(dev, "Module type is not supported.\n");
8167 		device_printf(dev, "Possible Solution: Change or replace the module or cable.\n");
8168 		break;
8169 	case ICE_AQC_HEALTH_STATUS_ERR_MOD_QUAL:
8170 		device_printf(dev, "Module is not qualified.\n");
8171 		device_printf(dev, "Possible Solution 1: Check your cable connection.\n");
8172 		device_printf(dev, "Possible Solution 2: Change or replace the module or cable.\n");
8173 		device_printf(dev, "Possible Solution 3: Manually set speed and duplex.\n");
8174 		break;
8175 	case ICE_AQC_HEALTH_STATUS_ERR_MOD_COMM:
8176 		device_printf(dev, "Device cannot communicate with the module.\n");
8177 		device_printf(dev, "Possible Solution 1: Check your cable connection.\n");
8178 		device_printf(dev, "Possible Solution 2: Change or replace the module or cable.\n");
8179 		device_printf(dev, "Possible Solution 3: Manually set speed and duplex.\n");
8180 		break;
8181 	case ICE_AQC_HEALTH_STATUS_ERR_MOD_CONFLICT:
8182 		device_printf(dev, "Unresolved module conflict.\n");
8183 		device_printf(dev, "Possible Solution 1: Manually set speed/duplex or use Intel(R) Ethernet Port Configuration Tool to change the port option.\n");
8184 		device_printf(dev, "Possible Solution 2: If the problem persists, use a cable/module that is found in the supported modules and cables list for this device.\n");
8185 		break;
8186 	case ICE_AQC_HEALTH_STATUS_ERR_MOD_NOT_PRESENT:
8187 		device_printf(dev, "Module is not present.\n");
8188 		device_printf(dev, "Possible Solution 1: Check that the module is inserted correctly.\n");
8189 		device_printf(dev, "Possible Solution 2: If the problem persists, use a cable/module that is found in the supported modules and cables list for this device.\n");
8190 		break;
8191 	case ICE_AQC_HEALTH_STATUS_INFO_MOD_UNDERUTILIZED:
8192 		device_printf(dev, "Underutilized module.\n");
8193 		device_printf(dev, "Possible Solution 1: Change or replace the module or cable.\n");
8194 		device_printf(dev, "Possible Solution 2: Use Intel(R) Ethernet Port Configuration Tool to change the port option.\n");
8195 		break;
8196 	case ICE_AQC_HEALTH_STATUS_ERR_UNKNOWN_MOD_LENIENT:
8197 		device_printf(dev, "An unsupported module was detected.\n");
8198 		device_printf(dev, "Possible Solution 1: Check your cable connection.\n");
8199 		device_printf(dev, "Possible Solution 2: Change or replace the module or cable.\n");
8200 		device_printf(dev, "Possible Solution 3: Manually set speed and duplex.\n");
8201 		break;
8202 	case ICE_AQC_HEALTH_STATUS_ERR_INVALID_LINK_CFG:
8203 		device_printf(dev, "Invalid link configuration.\n");
8204 		break;
8205 	case ICE_AQC_HEALTH_STATUS_ERR_PORT_ACCESS:
8206 		device_printf(dev, "Port hardware access error.\n");
8207 		device_printf(dev, "Possible Solution: Update to the latest NVM image.\n");
8208 		break;
8209 	case ICE_AQC_HEALTH_STATUS_ERR_PORT_UNREACHABLE:
8210 		device_printf(dev, "A port is unreachable.\n");
8211 		device_printf(dev, "Possible Solution 1: Use Intel(R) Ethernet Port Configuration Tool to change the port option.\n");
8212 		device_printf(dev, "Possible Solution 2: Update to the latest NVM image.\n");
8213 		break;
8214 	case ICE_AQC_HEALTH_STATUS_INFO_PORT_SPEED_MOD_LIMITED:
8215 		device_printf(dev, "Port speed is limited due to module.\n");
8216 		device_printf(dev, "Possible Solution: Change the module or use Intel(R) Ethernet Port Configuration Tool to configure the port option to match the current module speed.\n");
8217 		break;
8218 	case ICE_AQC_HEALTH_STATUS_ERR_PARALLEL_FAULT:
8219 		device_printf(dev, "A parallel fault was detected.\n");
8220 		device_printf(dev, "Possible Solution: Check link partner connection and configuration.\n");
8221 		break;
8222 	case ICE_AQC_HEALTH_STATUS_INFO_PORT_SPEED_PHY_LIMITED:
8223 		device_printf(dev, "Port speed is limited by PHY capabilities.\n");
8224 		device_printf(dev, "Possible Solution 1: Change the module to align to port option.\n");
8225 		device_printf(dev, "Possible Solution 2: Use Intel(R) Ethernet Port Configuration Tool to change the port option.\n");
8226 		break;
8227 	case ICE_AQC_HEALTH_STATUS_ERR_NETLIST_TOPO:
8228 		device_printf(dev, "LOM topology netlist is corrupted.\n");
8229 		device_printf(dev, "Possible Solution: Update to the latest NVM image.\n");
8230 		break;
8231 	case ICE_AQC_HEALTH_STATUS_ERR_NETLIST:
8232 		device_printf(dev, "Unrecoverable netlist error.\n");
8233 		device_printf(dev, "Possible Solution: Update to the latest NVM image.\n");
8234 		break;
8235 	case ICE_AQC_HEALTH_STATUS_ERR_TOPO_CONFLICT:
8236 		device_printf(dev, "Port topology conflict.\n");
8237 		device_printf(dev, "Possible Solution 1: Use Intel(R) Ethernet Port Configuration Tool to change the port option.\n");
8238 		device_printf(dev, "Possible Solution 2: Update to the latest NVM image.\n");
8239 		break;
8240 	case ICE_AQC_HEALTH_STATUS_ERR_LINK_HW_ACCESS:
8241 		device_printf(dev, "Unrecoverable hardware access error.\n");
8242 		device_printf(dev, "Possible Solution: Update to the latest NVM image.\n");
8243 		break;
8244 	case ICE_AQC_HEALTH_STATUS_ERR_LINK_RUNTIME:
8245 		device_printf(dev, "Unrecoverable runtime error.\n");
8246 		device_printf(dev, "Possible Solution: Update to the latest NVM image.\n");
8247 		break;
8248 	case ICE_AQC_HEALTH_STATUS_ERR_DNL_INIT:
8249 		device_printf(dev, "Link management engine failed to initialize.\n");
8250 		device_printf(dev, "Possible Solution: Update to the latest NVM image.\n");
8251 		break;
8252 	default:
8253 		break;
8254 	}
8255 }
8256 
8257 /**
8258  * ice_handle_health_status_event - helper function to output health status
8259  * @sc: device softc structure
8260  * @event: event received on a control queue
8261  *
8262  * Prints out the appropriate string based on the given Health Status Event
8263  * code.
8264  */
8265 static void
8266 ice_handle_health_status_event(struct ice_softc *sc,
8267 			       struct ice_rq_event_info *event)
8268 {
8269 	struct ice_aqc_health_status_elem *health_info;
8270 	u16 status_count;
8271 	int i;
8272 
8273 	if (!ice_is_bit_set(sc->feat_en, ICE_FEATURE_HEALTH_STATUS))
8274 		return;
8275 
8276 	health_info = (struct ice_aqc_health_status_elem *)event->msg_buf;
8277 	status_count = le16toh(event->desc.params.get_health_status.health_status_count);
8278 
8279 	if (status_count > (event->buf_len / sizeof(*health_info))) {
8280 		device_printf(sc->dev, "Received a health status event with invalid event count\n");
8281 		return;
8282 	}
8283 
8284 	for (i = 0; i < status_count; i++) {
8285 		ice_print_health_status_string(sc->dev, health_info);
8286 		health_info++;
8287 	}
8288 }
8289 
8290 /**
8291  * ice_set_default_local_lldp_mib - Set Local LLDP MIB to default settings
8292  * @sc: device softc structure
8293  *
8294  * This function needs to be called after link up; it makes sure the FW
8295  * has certain PFC/DCB settings. This is intended to workaround a FW behavior
8296  * where these settings seem to be cleared on link up.
8297  */
8298 void
8299 ice_set_default_local_lldp_mib(struct ice_softc *sc)
8300 {
8301 	struct ice_dcbx_cfg *dcbcfg;
8302 	struct ice_hw *hw = &sc->hw;
8303 	struct ice_port_info *pi;
8304 	device_t dev = sc->dev;
8305 	enum ice_status status;
8306 
8307 	pi = hw->port_info;
8308 	dcbcfg = &pi->qos_cfg.local_dcbx_cfg;
8309 
8310 	/* This value is only 3 bits; 8 TCs maps to 0 */
8311 	u8 maxtcs = hw->func_caps.common_cap.maxtc & ICE_IEEE_ETS_MAXTC_M;
8312 
8313 	/**
8314 	 * Setup the default settings used by the driver for the Set Local
8315 	 * LLDP MIB Admin Queue command (0x0A08). (1TC w/ 100% BW, ETS, no
8316 	 * PFC).
8317 	 */
8318 	memset(dcbcfg, 0, sizeof(*dcbcfg));
8319 	dcbcfg->etscfg.willing = 1;
8320 	dcbcfg->etscfg.tcbwtable[0] = 100;
8321 	dcbcfg->etscfg.maxtcs = maxtcs;
8322 	dcbcfg->etsrec.willing = 1;
8323 	dcbcfg->etsrec.tcbwtable[0] = 100;
8324 	dcbcfg->etsrec.maxtcs = maxtcs;
8325 	dcbcfg->pfc.willing = 1;
8326 	dcbcfg->pfc.pfccap = maxtcs;
8327 
8328 	status = ice_set_dcb_cfg(pi);
8329 
8330 	if (status)
8331 		device_printf(dev,
8332 		    "Error setting Local LLDP MIB: %s aq_err %s\n",
8333 		    ice_status_str(status),
8334 		    ice_aq_str(hw->adminq.sq_last_status));
8335 }
8336