cachepc-linux

Fork of AMDESE/linux with modifications for CachePC side-channel attack
git clone https://git.sinitax.com/sinitax/cachepc-linux
Log | Files | Refs | README | LICENSE | sfeed.txt

ice_sriov.c (51151B)


      1// SPDX-License-Identifier: GPL-2.0
      2/* Copyright (c) 2018, Intel Corporation. */
      3
      4#include "ice.h"
      5#include "ice_vf_lib_private.h"
      6#include "ice_base.h"
      7#include "ice_lib.h"
      8#include "ice_fltr.h"
      9#include "ice_dcb_lib.h"
     10#include "ice_flow.h"
     11#include "ice_eswitch.h"
     12#include "ice_virtchnl_allowlist.h"
     13#include "ice_flex_pipe.h"
     14#include "ice_vf_vsi_vlan_ops.h"
     15#include "ice_vlan.h"
     16
     17/**
     18 * ice_free_vf_entries - Free all VF entries from the hash table
     19 * @pf: pointer to the PF structure
     20 *
     21 * Iterate over the VF hash table, removing and releasing all VF entries.
     22 * Called during VF teardown or as cleanup during failed VF initialization.
     23 */
     24static void ice_free_vf_entries(struct ice_pf *pf)
     25{
     26	struct ice_vfs *vfs = &pf->vfs;
     27	struct hlist_node *tmp;
     28	struct ice_vf *vf;
     29	unsigned int bkt;
     30
     31	/* Remove all VFs from the hash table and release their main
     32	 * reference. Once all references to the VF are dropped, ice_put_vf()
     33	 * will call ice_release_vf which will remove the VF memory.
     34	 */
     35	lockdep_assert_held(&vfs->table_lock);
     36
     37	hash_for_each_safe(vfs->table, bkt, tmp, vf, entry) {
     38		hash_del_rcu(&vf->entry);
     39		ice_put_vf(vf);
     40	}
     41}
     42
     43/**
     44 * ice_vf_vsi_release - invalidate the VF's VSI after freeing it
     45 * @vf: invalidate this VF's VSI after freeing it
     46 */
     47static void ice_vf_vsi_release(struct ice_vf *vf)
     48{
     49	struct ice_vsi *vsi = ice_get_vf_vsi(vf);
     50
     51	if (WARN_ON(!vsi))
     52		return;
     53
     54	ice_vsi_release(vsi);
     55	ice_vf_invalidate_vsi(vf);
     56}
     57
     58/**
     59 * ice_free_vf_res - Free a VF's resources
     60 * @vf: pointer to the VF info
     61 */
     62static void ice_free_vf_res(struct ice_vf *vf)
     63{
     64	struct ice_pf *pf = vf->pf;
     65	int i, last_vector_idx;
     66
     67	/* First, disable VF's configuration API to prevent OS from
     68	 * accessing the VF's VSI after it's freed or invalidated.
     69	 */
     70	clear_bit(ICE_VF_STATE_INIT, vf->vf_states);
     71	ice_vf_fdir_exit(vf);
     72	/* free VF control VSI */
     73	if (vf->ctrl_vsi_idx != ICE_NO_VSI)
     74		ice_vf_ctrl_vsi_release(vf);
     75
     76	/* free VSI and disconnect it from the parent uplink */
     77	if (vf->lan_vsi_idx != ICE_NO_VSI) {
     78		ice_vf_vsi_release(vf);
     79		vf->num_mac = 0;
     80	}
     81
     82	last_vector_idx = vf->first_vector_idx + pf->vfs.num_msix_per - 1;
     83
     84	/* clear VF MDD event information */
     85	memset(&vf->mdd_tx_events, 0, sizeof(vf->mdd_tx_events));
     86	memset(&vf->mdd_rx_events, 0, sizeof(vf->mdd_rx_events));
     87
     88	/* Disable interrupts so that VF starts in a known state */
     89	for (i = vf->first_vector_idx; i <= last_vector_idx; i++) {
     90		wr32(&pf->hw, GLINT_DYN_CTL(i), GLINT_DYN_CTL_CLEARPBA_M);
     91		ice_flush(&pf->hw);
     92	}
     93	/* reset some of the state variables keeping track of the resources */
     94	clear_bit(ICE_VF_STATE_MC_PROMISC, vf->vf_states);
     95	clear_bit(ICE_VF_STATE_UC_PROMISC, vf->vf_states);
     96}
     97
     98/**
     99 * ice_dis_vf_mappings
    100 * @vf: pointer to the VF structure
    101 */
    102static void ice_dis_vf_mappings(struct ice_vf *vf)
    103{
    104	struct ice_pf *pf = vf->pf;
    105	struct ice_vsi *vsi;
    106	struct device *dev;
    107	int first, last, v;
    108	struct ice_hw *hw;
    109
    110	hw = &pf->hw;
    111	vsi = ice_get_vf_vsi(vf);
    112	if (WARN_ON(!vsi))
    113		return;
    114
    115	dev = ice_pf_to_dev(pf);
    116	wr32(hw, VPINT_ALLOC(vf->vf_id), 0);
    117	wr32(hw, VPINT_ALLOC_PCI(vf->vf_id), 0);
    118
    119	first = vf->first_vector_idx;
    120	last = first + pf->vfs.num_msix_per - 1;
    121	for (v = first; v <= last; v++) {
    122		u32 reg;
    123
    124		reg = (((1 << GLINT_VECT2FUNC_IS_PF_S) &
    125			GLINT_VECT2FUNC_IS_PF_M) |
    126		       ((hw->pf_id << GLINT_VECT2FUNC_PF_NUM_S) &
    127			GLINT_VECT2FUNC_PF_NUM_M));
    128		wr32(hw, GLINT_VECT2FUNC(v), reg);
    129	}
    130
    131	if (vsi->tx_mapping_mode == ICE_VSI_MAP_CONTIG)
    132		wr32(hw, VPLAN_TX_QBASE(vf->vf_id), 0);
    133	else
    134		dev_err(dev, "Scattered mode for VF Tx queues is not yet implemented\n");
    135
    136	if (vsi->rx_mapping_mode == ICE_VSI_MAP_CONTIG)
    137		wr32(hw, VPLAN_RX_QBASE(vf->vf_id), 0);
    138	else
    139		dev_err(dev, "Scattered mode for VF Rx queues is not yet implemented\n");
    140}
    141
    142/**
    143 * ice_sriov_free_msix_res - Reset/free any used MSIX resources
    144 * @pf: pointer to the PF structure
    145 *
    146 * Since no MSIX entries are taken from the pf->irq_tracker then just clear
    147 * the pf->sriov_base_vector.
    148 *
    149 * Returns 0 on success, and -EINVAL on error.
    150 */
    151static int ice_sriov_free_msix_res(struct ice_pf *pf)
    152{
    153	struct ice_res_tracker *res;
    154
    155	if (!pf)
    156		return -EINVAL;
    157
    158	res = pf->irq_tracker;
    159	if (!res)
    160		return -EINVAL;
    161
    162	/* give back irq_tracker resources used */
    163	WARN_ON(pf->sriov_base_vector < res->num_entries);
    164
    165	pf->sriov_base_vector = 0;
    166
    167	return 0;
    168}
    169
    170/**
    171 * ice_free_vfs - Free all VFs
    172 * @pf: pointer to the PF structure
    173 */
    174void ice_free_vfs(struct ice_pf *pf)
    175{
    176	struct device *dev = ice_pf_to_dev(pf);
    177	struct ice_vfs *vfs = &pf->vfs;
    178	struct ice_hw *hw = &pf->hw;
    179	struct ice_vf *vf;
    180	unsigned int bkt;
    181
    182	if (!ice_has_vfs(pf))
    183		return;
    184
    185	while (test_and_set_bit(ICE_VF_DIS, pf->state))
    186		usleep_range(1000, 2000);
    187
    188	/* Disable IOV before freeing resources. This lets any VF drivers
    189	 * running in the host get themselves cleaned up before we yank
    190	 * the carpet out from underneath their feet.
    191	 */
    192	if (!pci_vfs_assigned(pf->pdev))
    193		pci_disable_sriov(pf->pdev);
    194	else
    195		dev_warn(dev, "VFs are assigned - not disabling SR-IOV\n");
    196
    197	mutex_lock(&vfs->table_lock);
    198
    199	ice_eswitch_release(pf);
    200
    201	ice_for_each_vf(pf, bkt, vf) {
    202		mutex_lock(&vf->cfg_lock);
    203
    204		ice_dis_vf_qs(vf);
    205
    206		if (test_bit(ICE_VF_STATE_INIT, vf->vf_states)) {
    207			/* disable VF qp mappings and set VF disable state */
    208			ice_dis_vf_mappings(vf);
    209			set_bit(ICE_VF_STATE_DIS, vf->vf_states);
    210			ice_free_vf_res(vf);
    211		}
    212
    213		if (!pci_vfs_assigned(pf->pdev)) {
    214			u32 reg_idx, bit_idx;
    215
    216			reg_idx = (hw->func_caps.vf_base_id + vf->vf_id) / 32;
    217			bit_idx = (hw->func_caps.vf_base_id + vf->vf_id) % 32;
    218			wr32(hw, GLGEN_VFLRSTAT(reg_idx), BIT(bit_idx));
    219		}
    220
    221		/* clear malicious info since the VF is getting released */
    222		if (ice_mbx_clear_malvf(&hw->mbx_snapshot, pf->vfs.malvfs,
    223					ICE_MAX_SRIOV_VFS, vf->vf_id))
    224			dev_dbg(dev, "failed to clear malicious VF state for VF %u\n",
    225				vf->vf_id);
    226
    227		mutex_unlock(&vf->cfg_lock);
    228	}
    229
    230	if (ice_sriov_free_msix_res(pf))
    231		dev_err(dev, "Failed to free MSIX resources used by SR-IOV\n");
    232
    233	vfs->num_qps_per = 0;
    234	ice_free_vf_entries(pf);
    235
    236	mutex_unlock(&vfs->table_lock);
    237
    238	clear_bit(ICE_VF_DIS, pf->state);
    239	clear_bit(ICE_FLAG_SRIOV_ENA, pf->flags);
    240}
    241
    242/**
    243 * ice_vf_vsi_setup - Set up a VF VSI
    244 * @vf: VF to setup VSI for
    245 *
    246 * Returns pointer to the successfully allocated VSI struct on success,
    247 * otherwise returns NULL on failure.
    248 */
    249static struct ice_vsi *ice_vf_vsi_setup(struct ice_vf *vf)
    250{
    251	struct ice_port_info *pi = ice_vf_get_port_info(vf);
    252	struct ice_pf *pf = vf->pf;
    253	struct ice_vsi *vsi;
    254
    255	vsi = ice_vsi_setup(pf, pi, ICE_VSI_VF, vf, NULL);
    256
    257	if (!vsi) {
    258		dev_err(ice_pf_to_dev(pf), "Failed to create VF VSI\n");
    259		ice_vf_invalidate_vsi(vf);
    260		return NULL;
    261	}
    262
    263	vf->lan_vsi_idx = vsi->idx;
    264	vf->lan_vsi_num = vsi->vsi_num;
    265
    266	return vsi;
    267}
    268
    269/**
    270 * ice_calc_vf_first_vector_idx - Calculate MSIX vector index in the PF space
    271 * @pf: pointer to PF structure
    272 * @vf: pointer to VF that the first MSIX vector index is being calculated for
    273 *
    274 * This returns the first MSIX vector index in PF space that is used by this VF.
    275 * This index is used when accessing PF relative registers such as
    276 * GLINT_VECT2FUNC and GLINT_DYN_CTL.
    277 * This will always be the OICR index in the AVF driver so any functionality
    278 * using vf->first_vector_idx for queue configuration will have to increment by
    279 * 1 to avoid meddling with the OICR index.
    280 */
    281static int ice_calc_vf_first_vector_idx(struct ice_pf *pf, struct ice_vf *vf)
    282{
    283	return pf->sriov_base_vector + vf->vf_id * pf->vfs.num_msix_per;
    284}
    285
    286/**
    287 * ice_ena_vf_msix_mappings - enable VF MSIX mappings in hardware
    288 * @vf: VF to enable MSIX mappings for
    289 *
    290 * Some of the registers need to be indexed/configured using hardware global
    291 * device values and other registers need 0-based values, which represent PF
    292 * based values.
    293 */
    294static void ice_ena_vf_msix_mappings(struct ice_vf *vf)
    295{
    296	int device_based_first_msix, device_based_last_msix;
    297	int pf_based_first_msix, pf_based_last_msix, v;
    298	struct ice_pf *pf = vf->pf;
    299	int device_based_vf_id;
    300	struct ice_hw *hw;
    301	u32 reg;
    302
    303	hw = &pf->hw;
    304	pf_based_first_msix = vf->first_vector_idx;
    305	pf_based_last_msix = (pf_based_first_msix + pf->vfs.num_msix_per) - 1;
    306
    307	device_based_first_msix = pf_based_first_msix +
    308		pf->hw.func_caps.common_cap.msix_vector_first_id;
    309	device_based_last_msix =
    310		(device_based_first_msix + pf->vfs.num_msix_per) - 1;
    311	device_based_vf_id = vf->vf_id + hw->func_caps.vf_base_id;
    312
    313	reg = (((device_based_first_msix << VPINT_ALLOC_FIRST_S) &
    314		VPINT_ALLOC_FIRST_M) |
    315	       ((device_based_last_msix << VPINT_ALLOC_LAST_S) &
    316		VPINT_ALLOC_LAST_M) | VPINT_ALLOC_VALID_M);
    317	wr32(hw, VPINT_ALLOC(vf->vf_id), reg);
    318
    319	reg = (((device_based_first_msix << VPINT_ALLOC_PCI_FIRST_S)
    320		 & VPINT_ALLOC_PCI_FIRST_M) |
    321	       ((device_based_last_msix << VPINT_ALLOC_PCI_LAST_S) &
    322		VPINT_ALLOC_PCI_LAST_M) | VPINT_ALLOC_PCI_VALID_M);
    323	wr32(hw, VPINT_ALLOC_PCI(vf->vf_id), reg);
    324
    325	/* map the interrupts to its functions */
    326	for (v = pf_based_first_msix; v <= pf_based_last_msix; v++) {
    327		reg = (((device_based_vf_id << GLINT_VECT2FUNC_VF_NUM_S) &
    328			GLINT_VECT2FUNC_VF_NUM_M) |
    329		       ((hw->pf_id << GLINT_VECT2FUNC_PF_NUM_S) &
    330			GLINT_VECT2FUNC_PF_NUM_M));
    331		wr32(hw, GLINT_VECT2FUNC(v), reg);
    332	}
    333
    334	/* Map mailbox interrupt to VF MSI-X vector 0 */
    335	wr32(hw, VPINT_MBX_CTL(device_based_vf_id), VPINT_MBX_CTL_CAUSE_ENA_M);
    336}
    337
    338/**
    339 * ice_ena_vf_q_mappings - enable Rx/Tx queue mappings for a VF
    340 * @vf: VF to enable the mappings for
    341 * @max_txq: max Tx queues allowed on the VF's VSI
    342 * @max_rxq: max Rx queues allowed on the VF's VSI
    343 */
    344static void ice_ena_vf_q_mappings(struct ice_vf *vf, u16 max_txq, u16 max_rxq)
    345{
    346	struct device *dev = ice_pf_to_dev(vf->pf);
    347	struct ice_vsi *vsi = ice_get_vf_vsi(vf);
    348	struct ice_hw *hw = &vf->pf->hw;
    349	u32 reg;
    350
    351	if (WARN_ON(!vsi))
    352		return;
    353
    354	/* set regardless of mapping mode */
    355	wr32(hw, VPLAN_TXQ_MAPENA(vf->vf_id), VPLAN_TXQ_MAPENA_TX_ENA_M);
    356
    357	/* VF Tx queues allocation */
    358	if (vsi->tx_mapping_mode == ICE_VSI_MAP_CONTIG) {
    359		/* set the VF PF Tx queue range
    360		 * VFNUMQ value should be set to (number of queues - 1). A value
    361		 * of 0 means 1 queue and a value of 255 means 256 queues
    362		 */
    363		reg = (((vsi->txq_map[0] << VPLAN_TX_QBASE_VFFIRSTQ_S) &
    364			VPLAN_TX_QBASE_VFFIRSTQ_M) |
    365		       (((max_txq - 1) << VPLAN_TX_QBASE_VFNUMQ_S) &
    366			VPLAN_TX_QBASE_VFNUMQ_M));
    367		wr32(hw, VPLAN_TX_QBASE(vf->vf_id), reg);
    368	} else {
    369		dev_err(dev, "Scattered mode for VF Tx queues is not yet implemented\n");
    370	}
    371
    372	/* set regardless of mapping mode */
    373	wr32(hw, VPLAN_RXQ_MAPENA(vf->vf_id), VPLAN_RXQ_MAPENA_RX_ENA_M);
    374
    375	/* VF Rx queues allocation */
    376	if (vsi->rx_mapping_mode == ICE_VSI_MAP_CONTIG) {
    377		/* set the VF PF Rx queue range
    378		 * VFNUMQ value should be set to (number of queues - 1). A value
    379		 * of 0 means 1 queue and a value of 255 means 256 queues
    380		 */
    381		reg = (((vsi->rxq_map[0] << VPLAN_RX_QBASE_VFFIRSTQ_S) &
    382			VPLAN_RX_QBASE_VFFIRSTQ_M) |
    383		       (((max_rxq - 1) << VPLAN_RX_QBASE_VFNUMQ_S) &
    384			VPLAN_RX_QBASE_VFNUMQ_M));
    385		wr32(hw, VPLAN_RX_QBASE(vf->vf_id), reg);
    386	} else {
    387		dev_err(dev, "Scattered mode for VF Rx queues is not yet implemented\n");
    388	}
    389}
    390
    391/**
    392 * ice_ena_vf_mappings - enable VF MSIX and queue mapping
    393 * @vf: pointer to the VF structure
    394 */
    395static void ice_ena_vf_mappings(struct ice_vf *vf)
    396{
    397	struct ice_vsi *vsi = ice_get_vf_vsi(vf);
    398
    399	if (WARN_ON(!vsi))
    400		return;
    401
    402	ice_ena_vf_msix_mappings(vf);
    403	ice_ena_vf_q_mappings(vf, vsi->alloc_txq, vsi->alloc_rxq);
    404}
    405
    406/**
    407 * ice_calc_vf_reg_idx - Calculate the VF's register index in the PF space
    408 * @vf: VF to calculate the register index for
    409 * @q_vector: a q_vector associated to the VF
    410 */
    411int ice_calc_vf_reg_idx(struct ice_vf *vf, struct ice_q_vector *q_vector)
    412{
    413	struct ice_pf *pf;
    414
    415	if (!vf || !q_vector)
    416		return -EINVAL;
    417
    418	pf = vf->pf;
    419
    420	/* always add one to account for the OICR being the first MSIX */
    421	return pf->sriov_base_vector + pf->vfs.num_msix_per * vf->vf_id +
    422		q_vector->v_idx + 1;
    423}
    424
    425/**
    426 * ice_get_max_valid_res_idx - Get the max valid resource index
    427 * @res: pointer to the resource to find the max valid index for
    428 *
    429 * Start from the end of the ice_res_tracker and return right when we find the
    430 * first res->list entry with the ICE_RES_VALID_BIT set. This function is only
    431 * valid for SR-IOV because it is the only consumer that manipulates the
    432 * res->end and this is always called when res->end is set to res->num_entries.
    433 */
    434static int ice_get_max_valid_res_idx(struct ice_res_tracker *res)
    435{
    436	int i;
    437
    438	if (!res)
    439		return -EINVAL;
    440
    441	for (i = res->num_entries - 1; i >= 0; i--)
    442		if (res->list[i] & ICE_RES_VALID_BIT)
    443			return i;
    444
    445	return 0;
    446}
    447
    448/**
    449 * ice_sriov_set_msix_res - Set any used MSIX resources
    450 * @pf: pointer to PF structure
    451 * @num_msix_needed: number of MSIX vectors needed for all SR-IOV VFs
    452 *
    453 * This function allows SR-IOV resources to be taken from the end of the PF's
    454 * allowed HW MSIX vectors so that the irq_tracker will not be affected. We
    455 * just set the pf->sriov_base_vector and return success.
    456 *
    457 * If there are not enough resources available, return an error. This should
    458 * always be caught by ice_set_per_vf_res().
    459 *
    460 * Return 0 on success, and -EINVAL when there are not enough MSIX vectors
    461 * in the PF's space available for SR-IOV.
    462 */
    463static int ice_sriov_set_msix_res(struct ice_pf *pf, u16 num_msix_needed)
    464{
    465	u16 total_vectors = pf->hw.func_caps.common_cap.num_msix_vectors;
    466	int vectors_used = pf->irq_tracker->num_entries;
    467	int sriov_base_vector;
    468
    469	sriov_base_vector = total_vectors - num_msix_needed;
    470
    471	/* make sure we only grab irq_tracker entries from the list end and
    472	 * that we have enough available MSIX vectors
    473	 */
    474	if (sriov_base_vector < vectors_used)
    475		return -EINVAL;
    476
    477	pf->sriov_base_vector = sriov_base_vector;
    478
    479	return 0;
    480}
    481
    482/**
    483 * ice_set_per_vf_res - check if vectors and queues are available
    484 * @pf: pointer to the PF structure
    485 * @num_vfs: the number of SR-IOV VFs being configured
    486 *
    487 * First, determine HW interrupts from common pool. If we allocate fewer VFs, we
    488 * get more vectors and can enable more queues per VF. Note that this does not
    489 * grab any vectors from the SW pool already allocated. Also note, that all
    490 * vector counts include one for each VF's miscellaneous interrupt vector
    491 * (i.e. OICR).
    492 *
    493 * Minimum VFs - 2 vectors, 1 queue pair
    494 * Small VFs - 5 vectors, 4 queue pairs
    495 * Medium VFs - 17 vectors, 16 queue pairs
    496 *
    497 * Second, determine number of queue pairs per VF by starting with a pre-defined
    498 * maximum each VF supports. If this is not possible, then we adjust based on
    499 * queue pairs available on the device.
    500 *
    501 * Lastly, set queue and MSI-X VF variables tracked by the PF so it can be used
    502 * by each VF during VF initialization and reset.
    503 */
    504static int ice_set_per_vf_res(struct ice_pf *pf, u16 num_vfs)
    505{
    506	int max_valid_res_idx = ice_get_max_valid_res_idx(pf->irq_tracker);
    507	u16 num_msix_per_vf, num_txq, num_rxq, avail_qs;
    508	int msix_avail_per_vf, msix_avail_for_sriov;
    509	struct device *dev = ice_pf_to_dev(pf);
    510	int err;
    511
    512	lockdep_assert_held(&pf->vfs.table_lock);
    513
    514	if (!num_vfs)
    515		return -EINVAL;
    516
    517	if (max_valid_res_idx < 0)
    518		return -ENOSPC;
    519
    520	/* determine MSI-X resources per VF */
    521	msix_avail_for_sriov = pf->hw.func_caps.common_cap.num_msix_vectors -
    522		pf->irq_tracker->num_entries;
    523	msix_avail_per_vf = msix_avail_for_sriov / num_vfs;
    524	if (msix_avail_per_vf >= ICE_NUM_VF_MSIX_MED) {
    525		num_msix_per_vf = ICE_NUM_VF_MSIX_MED;
    526	} else if (msix_avail_per_vf >= ICE_NUM_VF_MSIX_SMALL) {
    527		num_msix_per_vf = ICE_NUM_VF_MSIX_SMALL;
    528	} else if (msix_avail_per_vf >= ICE_NUM_VF_MSIX_MULTIQ_MIN) {
    529		num_msix_per_vf = ICE_NUM_VF_MSIX_MULTIQ_MIN;
    530	} else if (msix_avail_per_vf >= ICE_MIN_INTR_PER_VF) {
    531		num_msix_per_vf = ICE_MIN_INTR_PER_VF;
    532	} else {
    533		dev_err(dev, "Only %d MSI-X interrupts available for SR-IOV. Not enough to support minimum of %d MSI-X interrupts per VF for %d VFs\n",
    534			msix_avail_for_sriov, ICE_MIN_INTR_PER_VF,
    535			num_vfs);
    536		return -ENOSPC;
    537	}
    538
    539	num_txq = min_t(u16, num_msix_per_vf - ICE_NONQ_VECS_VF,
    540			ICE_MAX_RSS_QS_PER_VF);
    541	avail_qs = ice_get_avail_txq_count(pf) / num_vfs;
    542	if (!avail_qs)
    543		num_txq = 0;
    544	else if (num_txq > avail_qs)
    545		num_txq = rounddown_pow_of_two(avail_qs);
    546
    547	num_rxq = min_t(u16, num_msix_per_vf - ICE_NONQ_VECS_VF,
    548			ICE_MAX_RSS_QS_PER_VF);
    549	avail_qs = ice_get_avail_rxq_count(pf) / num_vfs;
    550	if (!avail_qs)
    551		num_rxq = 0;
    552	else if (num_rxq > avail_qs)
    553		num_rxq = rounddown_pow_of_two(avail_qs);
    554
    555	if (num_txq < ICE_MIN_QS_PER_VF || num_rxq < ICE_MIN_QS_PER_VF) {
    556		dev_err(dev, "Not enough queues to support minimum of %d queue pairs per VF for %d VFs\n",
    557			ICE_MIN_QS_PER_VF, num_vfs);
    558		return -ENOSPC;
    559	}
    560
    561	err = ice_sriov_set_msix_res(pf, num_msix_per_vf * num_vfs);
    562	if (err) {
    563		dev_err(dev, "Unable to set MSI-X resources for %d VFs, err %d\n",
    564			num_vfs, err);
    565		return err;
    566	}
    567
    568	/* only allow equal Tx/Rx queue count (i.e. queue pairs) */
    569	pf->vfs.num_qps_per = min_t(int, num_txq, num_rxq);
    570	pf->vfs.num_msix_per = num_msix_per_vf;
    571	dev_info(dev, "Enabling %d VFs with %d vectors and %d queues per VF\n",
    572		 num_vfs, pf->vfs.num_msix_per, pf->vfs.num_qps_per);
    573
    574	return 0;
    575}
    576
    577/**
    578 * ice_init_vf_vsi_res - initialize/setup VF VSI resources
    579 * @vf: VF to initialize/setup the VSI for
    580 *
    581 * This function creates a VSI for the VF, adds a VLAN 0 filter, and sets up the
    582 * VF VSI's broadcast filter and is only used during initial VF creation.
    583 */
    584static int ice_init_vf_vsi_res(struct ice_vf *vf)
    585{
    586	struct ice_vsi_vlan_ops *vlan_ops;
    587	struct ice_pf *pf = vf->pf;
    588	u8 broadcast[ETH_ALEN];
    589	struct ice_vsi *vsi;
    590	struct device *dev;
    591	int err;
    592
    593	vf->first_vector_idx = ice_calc_vf_first_vector_idx(pf, vf);
    594
    595	dev = ice_pf_to_dev(pf);
    596	vsi = ice_vf_vsi_setup(vf);
    597	if (!vsi)
    598		return -ENOMEM;
    599
    600	err = ice_vsi_add_vlan_zero(vsi);
    601	if (err) {
    602		dev_warn(dev, "Failed to add VLAN 0 filter for VF %d\n",
    603			 vf->vf_id);
    604		goto release_vsi;
    605	}
    606
    607	vlan_ops = ice_get_compat_vsi_vlan_ops(vsi);
    608	err = vlan_ops->ena_rx_filtering(vsi);
    609	if (err) {
    610		dev_warn(dev, "Failed to enable Rx VLAN filtering for VF %d\n",
    611			 vf->vf_id);
    612		goto release_vsi;
    613	}
    614
    615	eth_broadcast_addr(broadcast);
    616	err = ice_fltr_add_mac(vsi, broadcast, ICE_FWD_TO_VSI);
    617	if (err) {
    618		dev_err(dev, "Failed to add broadcast MAC filter for VF %d, error %d\n",
    619			vf->vf_id, err);
    620		goto release_vsi;
    621	}
    622
    623	err = ice_vsi_apply_spoofchk(vsi, vf->spoofchk);
    624	if (err) {
    625		dev_warn(dev, "Failed to initialize spoofchk setting for VF %d\n",
    626			 vf->vf_id);
    627		goto release_vsi;
    628	}
    629
    630	vf->num_mac = 1;
    631
    632	return 0;
    633
    634release_vsi:
    635	ice_vf_vsi_release(vf);
    636	return err;
    637}
    638
    639/**
    640 * ice_start_vfs - start VFs so they are ready to be used by SR-IOV
    641 * @pf: PF the VFs are associated with
    642 */
    643static int ice_start_vfs(struct ice_pf *pf)
    644{
    645	struct ice_hw *hw = &pf->hw;
    646	unsigned int bkt, it_cnt;
    647	struct ice_vf *vf;
    648	int retval;
    649
    650	lockdep_assert_held(&pf->vfs.table_lock);
    651
    652	it_cnt = 0;
    653	ice_for_each_vf(pf, bkt, vf) {
    654		vf->vf_ops->clear_reset_trigger(vf);
    655
    656		retval = ice_init_vf_vsi_res(vf);
    657		if (retval) {
    658			dev_err(ice_pf_to_dev(pf), "Failed to initialize VSI resources for VF %d, error %d\n",
    659				vf->vf_id, retval);
    660			goto teardown;
    661		}
    662
    663		set_bit(ICE_VF_STATE_INIT, vf->vf_states);
    664		ice_ena_vf_mappings(vf);
    665		wr32(hw, VFGEN_RSTAT(vf->vf_id), VIRTCHNL_VFR_VFACTIVE);
    666		it_cnt++;
    667	}
    668
    669	ice_flush(hw);
    670	return 0;
    671
    672teardown:
    673	ice_for_each_vf(pf, bkt, vf) {
    674		if (it_cnt == 0)
    675			break;
    676
    677		ice_dis_vf_mappings(vf);
    678		ice_vf_vsi_release(vf);
    679		it_cnt--;
    680	}
    681
    682	return retval;
    683}
    684
    685/**
    686 * ice_sriov_free_vf - Free VF memory after all references are dropped
    687 * @vf: pointer to VF to free
    688 *
    689 * Called by ice_put_vf through ice_release_vf once the last reference to a VF
    690 * structure has been dropped.
    691 */
    692static void ice_sriov_free_vf(struct ice_vf *vf)
    693{
    694	mutex_destroy(&vf->cfg_lock);
    695
    696	kfree_rcu(vf, rcu);
    697}
    698
    699/**
    700 * ice_sriov_clear_mbx_register - clears SRIOV VF's mailbox registers
    701 * @vf: the vf to configure
    702 */
    703static void ice_sriov_clear_mbx_register(struct ice_vf *vf)
    704{
    705	struct ice_pf *pf = vf->pf;
    706
    707	wr32(&pf->hw, VF_MBX_ARQLEN(vf->vf_id), 0);
    708	wr32(&pf->hw, VF_MBX_ATQLEN(vf->vf_id), 0);
    709}
    710
    711/**
    712 * ice_sriov_trigger_reset_register - trigger VF reset for SRIOV VF
    713 * @vf: pointer to VF structure
    714 * @is_vflr: true if reset occurred due to VFLR
    715 *
    716 * Trigger and cleanup after a VF reset for a SR-IOV VF.
    717 */
    718static void ice_sriov_trigger_reset_register(struct ice_vf *vf, bool is_vflr)
    719{
    720	struct ice_pf *pf = vf->pf;
    721	u32 reg, reg_idx, bit_idx;
    722	unsigned int vf_abs_id, i;
    723	struct device *dev;
    724	struct ice_hw *hw;
    725
    726	dev = ice_pf_to_dev(pf);
    727	hw = &pf->hw;
    728	vf_abs_id = vf->vf_id + hw->func_caps.vf_base_id;
    729
    730	/* In the case of a VFLR, HW has already reset the VF and we just need
    731	 * to clean up. Otherwise we must first trigger the reset using the
    732	 * VFRTRIG register.
    733	 */
    734	if (!is_vflr) {
    735		reg = rd32(hw, VPGEN_VFRTRIG(vf->vf_id));
    736		reg |= VPGEN_VFRTRIG_VFSWR_M;
    737		wr32(hw, VPGEN_VFRTRIG(vf->vf_id), reg);
    738	}
    739
    740	/* clear the VFLR bit in GLGEN_VFLRSTAT */
    741	reg_idx = (vf_abs_id) / 32;
    742	bit_idx = (vf_abs_id) % 32;
    743	wr32(hw, GLGEN_VFLRSTAT(reg_idx), BIT(bit_idx));
    744	ice_flush(hw);
    745
    746	wr32(hw, PF_PCI_CIAA,
    747	     VF_DEVICE_STATUS | (vf_abs_id << PF_PCI_CIAA_VF_NUM_S));
    748	for (i = 0; i < ICE_PCI_CIAD_WAIT_COUNT; i++) {
    749		reg = rd32(hw, PF_PCI_CIAD);
    750		/* no transactions pending so stop polling */
    751		if ((reg & VF_TRANS_PENDING_M) == 0)
    752			break;
    753
    754		dev_err(dev, "VF %u PCI transactions stuck\n", vf->vf_id);
    755		udelay(ICE_PCI_CIAD_WAIT_DELAY_US);
    756	}
    757}
    758
    759/**
    760 * ice_sriov_poll_reset_status - poll SRIOV VF reset status
    761 * @vf: pointer to VF structure
    762 *
    763 * Returns true when reset is successful, else returns false
    764 */
    765static bool ice_sriov_poll_reset_status(struct ice_vf *vf)
    766{
    767	struct ice_pf *pf = vf->pf;
    768	unsigned int i;
    769	u32 reg;
    770
    771	for (i = 0; i < 10; i++) {
    772		/* VF reset requires driver to first reset the VF and then
    773		 * poll the status register to make sure that the reset
    774		 * completed successfully.
    775		 */
    776		reg = rd32(&pf->hw, VPGEN_VFRSTAT(vf->vf_id));
    777		if (reg & VPGEN_VFRSTAT_VFRD_M)
    778			return true;
    779
    780		/* only sleep if the reset is not done */
    781		usleep_range(10, 20);
    782	}
    783	return false;
    784}
    785
    786/**
    787 * ice_sriov_clear_reset_trigger - enable VF to access hardware
    788 * @vf: VF to enabled hardware access for
    789 */
    790static void ice_sriov_clear_reset_trigger(struct ice_vf *vf)
    791{
    792	struct ice_hw *hw = &vf->pf->hw;
    793	u32 reg;
    794
    795	reg = rd32(hw, VPGEN_VFRTRIG(vf->vf_id));
    796	reg &= ~VPGEN_VFRTRIG_VFSWR_M;
    797	wr32(hw, VPGEN_VFRTRIG(vf->vf_id), reg);
    798	ice_flush(hw);
    799}
    800
    801/**
    802 * ice_sriov_vsi_rebuild - release and rebuild VF's VSI
    803 * @vf: VF to release and setup the VSI for
    804 *
    805 * This is only called when a single VF is being reset (i.e. VFR, VFLR, host VF
    806 * configuration change, etc.).
    807 */
    808static int ice_sriov_vsi_rebuild(struct ice_vf *vf)
    809{
    810	struct ice_pf *pf = vf->pf;
    811
    812	ice_vf_vsi_release(vf);
    813	if (!ice_vf_vsi_setup(vf)) {
    814		dev_err(ice_pf_to_dev(pf),
    815			"Failed to release and setup the VF%u's VSI\n",
    816			vf->vf_id);
    817		return -ENOMEM;
    818	}
    819
    820	return 0;
    821}
    822
    823/**
    824 * ice_sriov_post_vsi_rebuild - tasks to do after the VF's VSI have been rebuilt
    825 * @vf: VF to perform tasks on
    826 */
    827static void ice_sriov_post_vsi_rebuild(struct ice_vf *vf)
    828{
    829	ice_vf_rebuild_host_cfg(vf);
    830	ice_vf_set_initialized(vf);
    831	ice_ena_vf_mappings(vf);
    832	wr32(&vf->pf->hw, VFGEN_RSTAT(vf->vf_id), VIRTCHNL_VFR_VFACTIVE);
    833}
    834
    835static const struct ice_vf_ops ice_sriov_vf_ops = {
    836	.reset_type = ICE_VF_RESET,
    837	.free = ice_sriov_free_vf,
    838	.clear_mbx_register = ice_sriov_clear_mbx_register,
    839	.trigger_reset_register = ice_sriov_trigger_reset_register,
    840	.poll_reset_status = ice_sriov_poll_reset_status,
    841	.clear_reset_trigger = ice_sriov_clear_reset_trigger,
    842	.vsi_rebuild = ice_sriov_vsi_rebuild,
    843	.post_vsi_rebuild = ice_sriov_post_vsi_rebuild,
    844};
    845
    846/**
    847 * ice_create_vf_entries - Allocate and insert VF entries
    848 * @pf: pointer to the PF structure
    849 * @num_vfs: the number of VFs to allocate
    850 *
    851 * Allocate new VF entries and insert them into the hash table. Set some
    852 * basic default fields for initializing the new VFs.
    853 *
    854 * After this function exits, the hash table will have num_vfs entries
    855 * inserted.
    856 *
    857 * Returns 0 on success or an integer error code on failure.
    858 */
    859static int ice_create_vf_entries(struct ice_pf *pf, u16 num_vfs)
    860{
    861	struct ice_vfs *vfs = &pf->vfs;
    862	struct ice_vf *vf;
    863	u16 vf_id;
    864	int err;
    865
    866	lockdep_assert_held(&vfs->table_lock);
    867
    868	for (vf_id = 0; vf_id < num_vfs; vf_id++) {
    869		vf = kzalloc(sizeof(*vf), GFP_KERNEL);
    870		if (!vf) {
    871			err = -ENOMEM;
    872			goto err_free_entries;
    873		}
    874		kref_init(&vf->refcnt);
    875
    876		vf->pf = pf;
    877		vf->vf_id = vf_id;
    878
    879		/* set sriov vf ops for VFs created during SRIOV flow */
    880		vf->vf_ops = &ice_sriov_vf_ops;
    881
    882		vf->vf_sw_id = pf->first_sw;
    883		/* assign default capabilities */
    884		vf->spoofchk = true;
    885		vf->num_vf_qs = pf->vfs.num_qps_per;
    886		ice_vc_set_default_allowlist(vf);
    887
    888		/* ctrl_vsi_idx will be set to a valid value only when VF
    889		 * creates its first fdir rule.
    890		 */
    891		ice_vf_ctrl_invalidate_vsi(vf);
    892		ice_vf_fdir_init(vf);
    893
    894		ice_virtchnl_set_dflt_ops(vf);
    895
    896		mutex_init(&vf->cfg_lock);
    897
    898		hash_add_rcu(vfs->table, &vf->entry, vf_id);
    899	}
    900
    901	return 0;
    902
    903err_free_entries:
    904	ice_free_vf_entries(pf);
    905	return err;
    906}
    907
    908/**
    909 * ice_ena_vfs - enable VFs so they are ready to be used
    910 * @pf: pointer to the PF structure
    911 * @num_vfs: number of VFs to enable
    912 */
    913static int ice_ena_vfs(struct ice_pf *pf, u16 num_vfs)
    914{
    915	struct device *dev = ice_pf_to_dev(pf);
    916	struct ice_hw *hw = &pf->hw;
    917	int ret;
    918
    919	/* Disable global interrupt 0 so we don't try to handle the VFLR. */
    920	wr32(hw, GLINT_DYN_CTL(pf->oicr_idx),
    921	     ICE_ITR_NONE << GLINT_DYN_CTL_ITR_INDX_S);
    922	set_bit(ICE_OICR_INTR_DIS, pf->state);
    923	ice_flush(hw);
    924
    925	ret = pci_enable_sriov(pf->pdev, num_vfs);
    926	if (ret)
    927		goto err_unroll_intr;
    928
    929	mutex_lock(&pf->vfs.table_lock);
    930
    931	ret = ice_set_per_vf_res(pf, num_vfs);
    932	if (ret) {
    933		dev_err(dev, "Not enough resources for %d VFs, err %d. Try with fewer number of VFs\n",
    934			num_vfs, ret);
    935		goto err_unroll_sriov;
    936	}
    937
    938	ret = ice_create_vf_entries(pf, num_vfs);
    939	if (ret) {
    940		dev_err(dev, "Failed to allocate VF entries for %d VFs\n",
    941			num_vfs);
    942		goto err_unroll_sriov;
    943	}
    944
    945	ret = ice_start_vfs(pf);
    946	if (ret) {
    947		dev_err(dev, "Failed to start %d VFs, err %d\n", num_vfs, ret);
    948		ret = -EAGAIN;
    949		goto err_unroll_vf_entries;
    950	}
    951
    952	clear_bit(ICE_VF_DIS, pf->state);
    953
    954	ret = ice_eswitch_configure(pf);
    955	if (ret) {
    956		dev_err(dev, "Failed to configure eswitch, err %d\n", ret);
    957		goto err_unroll_sriov;
    958	}
    959
    960	/* rearm global interrupts */
    961	if (test_and_clear_bit(ICE_OICR_INTR_DIS, pf->state))
    962		ice_irq_dynamic_ena(hw, NULL, NULL);
    963
    964	mutex_unlock(&pf->vfs.table_lock);
    965
    966	return 0;
    967
    968err_unroll_vf_entries:
    969	ice_free_vf_entries(pf);
    970err_unroll_sriov:
    971	mutex_unlock(&pf->vfs.table_lock);
    972	pci_disable_sriov(pf->pdev);
    973err_unroll_intr:
    974	/* rearm interrupts here */
    975	ice_irq_dynamic_ena(hw, NULL, NULL);
    976	clear_bit(ICE_OICR_INTR_DIS, pf->state);
    977	return ret;
    978}
    979
    980/**
    981 * ice_pci_sriov_ena - Enable or change number of VFs
    982 * @pf: pointer to the PF structure
    983 * @num_vfs: number of VFs to allocate
    984 *
    985 * Returns 0 on success and negative on failure
    986 */
    987static int ice_pci_sriov_ena(struct ice_pf *pf, int num_vfs)
    988{
    989	int pre_existing_vfs = pci_num_vf(pf->pdev);
    990	struct device *dev = ice_pf_to_dev(pf);
    991	int err;
    992
    993	if (pre_existing_vfs && pre_existing_vfs != num_vfs)
    994		ice_free_vfs(pf);
    995	else if (pre_existing_vfs && pre_existing_vfs == num_vfs)
    996		return 0;
    997
    998	if (num_vfs > pf->vfs.num_supported) {
    999		dev_err(dev, "Can't enable %d VFs, max VFs supported is %d\n",
   1000			num_vfs, pf->vfs.num_supported);
   1001		return -EOPNOTSUPP;
   1002	}
   1003
   1004	dev_info(dev, "Enabling %d VFs\n", num_vfs);
   1005	err = ice_ena_vfs(pf, num_vfs);
   1006	if (err) {
   1007		dev_err(dev, "Failed to enable SR-IOV: %d\n", err);
   1008		return err;
   1009	}
   1010
   1011	set_bit(ICE_FLAG_SRIOV_ENA, pf->flags);
   1012	return 0;
   1013}
   1014
   1015/**
   1016 * ice_check_sriov_allowed - check if SR-IOV is allowed based on various checks
   1017 * @pf: PF to enabled SR-IOV on
   1018 */
   1019static int ice_check_sriov_allowed(struct ice_pf *pf)
   1020{
   1021	struct device *dev = ice_pf_to_dev(pf);
   1022
   1023	if (!test_bit(ICE_FLAG_SRIOV_CAPABLE, pf->flags)) {
   1024		dev_err(dev, "This device is not capable of SR-IOV\n");
   1025		return -EOPNOTSUPP;
   1026	}
   1027
   1028	if (ice_is_safe_mode(pf)) {
   1029		dev_err(dev, "SR-IOV cannot be configured - Device is in Safe Mode\n");
   1030		return -EOPNOTSUPP;
   1031	}
   1032
   1033	if (!ice_pf_state_is_nominal(pf)) {
   1034		dev_err(dev, "Cannot enable SR-IOV, device not ready\n");
   1035		return -EBUSY;
   1036	}
   1037
   1038	return 0;
   1039}
   1040
   1041/**
   1042 * ice_sriov_configure - Enable or change number of VFs via sysfs
   1043 * @pdev: pointer to a pci_dev structure
   1044 * @num_vfs: number of VFs to allocate or 0 to free VFs
   1045 *
   1046 * This function is called when the user updates the number of VFs in sysfs. On
   1047 * success return whatever num_vfs was set to by the caller. Return negative on
   1048 * failure.
   1049 */
   1050int ice_sriov_configure(struct pci_dev *pdev, int num_vfs)
   1051{
   1052	struct ice_pf *pf = pci_get_drvdata(pdev);
   1053	struct device *dev = ice_pf_to_dev(pf);
   1054	int err;
   1055
   1056	err = ice_check_sriov_allowed(pf);
   1057	if (err)
   1058		return err;
   1059
   1060	if (!num_vfs) {
   1061		if (!pci_vfs_assigned(pdev)) {
   1062			ice_free_vfs(pf);
   1063			ice_mbx_deinit_snapshot(&pf->hw);
   1064			if (pf->lag)
   1065				ice_enable_lag(pf->lag);
   1066			return 0;
   1067		}
   1068
   1069		dev_err(dev, "can't free VFs because some are assigned to VMs.\n");
   1070		return -EBUSY;
   1071	}
   1072
   1073	err = ice_mbx_init_snapshot(&pf->hw, num_vfs);
   1074	if (err)
   1075		return err;
   1076
   1077	err = ice_pci_sriov_ena(pf, num_vfs);
   1078	if (err) {
   1079		ice_mbx_deinit_snapshot(&pf->hw);
   1080		return err;
   1081	}
   1082
   1083	if (pf->lag)
   1084		ice_disable_lag(pf->lag);
   1085	return num_vfs;
   1086}
   1087
   1088/**
   1089 * ice_process_vflr_event - Free VF resources via IRQ calls
   1090 * @pf: pointer to the PF structure
   1091 *
   1092 * called from the VFLR IRQ handler to
   1093 * free up VF resources and state variables
   1094 */
   1095void ice_process_vflr_event(struct ice_pf *pf)
   1096{
   1097	struct ice_hw *hw = &pf->hw;
   1098	struct ice_vf *vf;
   1099	unsigned int bkt;
   1100	u32 reg;
   1101
   1102	if (!test_and_clear_bit(ICE_VFLR_EVENT_PENDING, pf->state) ||
   1103	    !ice_has_vfs(pf))
   1104		return;
   1105
   1106	mutex_lock(&pf->vfs.table_lock);
   1107	ice_for_each_vf(pf, bkt, vf) {
   1108		u32 reg_idx, bit_idx;
   1109
   1110		reg_idx = (hw->func_caps.vf_base_id + vf->vf_id) / 32;
   1111		bit_idx = (hw->func_caps.vf_base_id + vf->vf_id) % 32;
   1112		/* read GLGEN_VFLRSTAT register to find out the flr VFs */
   1113		reg = rd32(hw, GLGEN_VFLRSTAT(reg_idx));
   1114		if (reg & BIT(bit_idx))
   1115			/* GLGEN_VFLRSTAT bit will be cleared in ice_reset_vf */
   1116			ice_reset_vf(vf, ICE_VF_RESET_VFLR | ICE_VF_RESET_LOCK);
   1117	}
   1118	mutex_unlock(&pf->vfs.table_lock);
   1119}
   1120
   1121/**
   1122 * ice_get_vf_from_pfq - get the VF who owns the PF space queue passed in
   1123 * @pf: PF used to index all VFs
   1124 * @pfq: queue index relative to the PF's function space
   1125 *
   1126 * If no VF is found who owns the pfq then return NULL, otherwise return a
   1127 * pointer to the VF who owns the pfq
   1128 *
   1129 * If this function returns non-NULL, it acquires a reference count of the VF
   1130 * structure. The caller is responsible for calling ice_put_vf() to drop this
   1131 * reference.
   1132 */
   1133static struct ice_vf *ice_get_vf_from_pfq(struct ice_pf *pf, u16 pfq)
   1134{
   1135	struct ice_vf *vf;
   1136	unsigned int bkt;
   1137
   1138	rcu_read_lock();
   1139	ice_for_each_vf_rcu(pf, bkt, vf) {
   1140		struct ice_vsi *vsi;
   1141		u16 rxq_idx;
   1142
   1143		vsi = ice_get_vf_vsi(vf);
   1144		if (!vsi)
   1145			continue;
   1146
   1147		ice_for_each_rxq(vsi, rxq_idx)
   1148			if (vsi->rxq_map[rxq_idx] == pfq) {
   1149				struct ice_vf *found;
   1150
   1151				if (kref_get_unless_zero(&vf->refcnt))
   1152					found = vf;
   1153				else
   1154					found = NULL;
   1155				rcu_read_unlock();
   1156				return found;
   1157			}
   1158	}
   1159	rcu_read_unlock();
   1160
   1161	return NULL;
   1162}
   1163
   1164/**
   1165 * ice_globalq_to_pfq - convert from global queue index to PF space queue index
   1166 * @pf: PF used for conversion
   1167 * @globalq: global queue index used to convert to PF space queue index
   1168 */
   1169static u32 ice_globalq_to_pfq(struct ice_pf *pf, u32 globalq)
   1170{
   1171	return globalq - pf->hw.func_caps.common_cap.rxq_first_id;
   1172}
   1173
   1174/**
   1175 * ice_vf_lan_overflow_event - handle LAN overflow event for a VF
   1176 * @pf: PF that the LAN overflow event happened on
   1177 * @event: structure holding the event information for the LAN overflow event
   1178 *
   1179 * Determine if the LAN overflow event was caused by a VF queue. If it was not
   1180 * caused by a VF, do nothing. If a VF caused this LAN overflow event trigger a
   1181 * reset on the offending VF.
   1182 */
   1183void
   1184ice_vf_lan_overflow_event(struct ice_pf *pf, struct ice_rq_event_info *event)
   1185{
   1186	u32 gldcb_rtctq, queue;
   1187	struct ice_vf *vf;
   1188
   1189	gldcb_rtctq = le32_to_cpu(event->desc.params.lan_overflow.prtdcb_ruptq);
   1190	dev_dbg(ice_pf_to_dev(pf), "GLDCB_RTCTQ: 0x%08x\n", gldcb_rtctq);
   1191
   1192	/* event returns device global Rx queue number */
   1193	queue = (gldcb_rtctq & GLDCB_RTCTQ_RXQNUM_M) >>
   1194		GLDCB_RTCTQ_RXQNUM_S;
   1195
   1196	vf = ice_get_vf_from_pfq(pf, ice_globalq_to_pfq(pf, queue));
   1197	if (!vf)
   1198		return;
   1199
   1200	ice_reset_vf(vf, ICE_VF_RESET_NOTIFY | ICE_VF_RESET_LOCK);
   1201	ice_put_vf(vf);
   1202}
   1203
   1204/**
   1205 * ice_set_vf_spoofchk
   1206 * @netdev: network interface device structure
   1207 * @vf_id: VF identifier
   1208 * @ena: flag to enable or disable feature
   1209 *
   1210 * Enable or disable VF spoof checking
   1211 */
   1212int ice_set_vf_spoofchk(struct net_device *netdev, int vf_id, bool ena)
   1213{
   1214	struct ice_netdev_priv *np = netdev_priv(netdev);
   1215	struct ice_pf *pf = np->vsi->back;
   1216	struct ice_vsi *vf_vsi;
   1217	struct device *dev;
   1218	struct ice_vf *vf;
   1219	int ret;
   1220
   1221	dev = ice_pf_to_dev(pf);
   1222
   1223	vf = ice_get_vf_by_id(pf, vf_id);
   1224	if (!vf)
   1225		return -EINVAL;
   1226
   1227	ret = ice_check_vf_ready_for_cfg(vf);
   1228	if (ret)
   1229		goto out_put_vf;
   1230
   1231	vf_vsi = ice_get_vf_vsi(vf);
   1232	if (!vf_vsi) {
   1233		netdev_err(netdev, "VSI %d for VF %d is null\n",
   1234			   vf->lan_vsi_idx, vf->vf_id);
   1235		ret = -EINVAL;
   1236		goto out_put_vf;
   1237	}
   1238
   1239	if (vf_vsi->type != ICE_VSI_VF) {
   1240		netdev_err(netdev, "Type %d of VSI %d for VF %d is no ICE_VSI_VF\n",
   1241			   vf_vsi->type, vf_vsi->vsi_num, vf->vf_id);
   1242		ret = -ENODEV;
   1243		goto out_put_vf;
   1244	}
   1245
   1246	if (ena == vf->spoofchk) {
   1247		dev_dbg(dev, "VF spoofchk already %s\n", ena ? "ON" : "OFF");
   1248		ret = 0;
   1249		goto out_put_vf;
   1250	}
   1251
   1252	ret = ice_vsi_apply_spoofchk(vf_vsi, ena);
   1253	if (ret)
   1254		dev_err(dev, "Failed to set spoofchk %s for VF %d VSI %d\n error %d\n",
   1255			ena ? "ON" : "OFF", vf->vf_id, vf_vsi->vsi_num, ret);
   1256	else
   1257		vf->spoofchk = ena;
   1258
   1259out_put_vf:
   1260	ice_put_vf(vf);
   1261	return ret;
   1262}
   1263
   1264/**
   1265 * ice_get_vf_cfg
   1266 * @netdev: network interface device structure
   1267 * @vf_id: VF identifier
   1268 * @ivi: VF configuration structure
   1269 *
   1270 * return VF configuration
   1271 */
   1272int
   1273ice_get_vf_cfg(struct net_device *netdev, int vf_id, struct ifla_vf_info *ivi)
   1274{
   1275	struct ice_pf *pf = ice_netdev_to_pf(netdev);
   1276	struct ice_vf *vf;
   1277	int ret;
   1278
   1279	vf = ice_get_vf_by_id(pf, vf_id);
   1280	if (!vf)
   1281		return -EINVAL;
   1282
   1283	ret = ice_check_vf_ready_for_cfg(vf);
   1284	if (ret)
   1285		goto out_put_vf;
   1286
   1287	ivi->vf = vf_id;
   1288	ether_addr_copy(ivi->mac, vf->hw_lan_addr.addr);
   1289
   1290	/* VF configuration for VLAN and applicable QoS */
   1291	ivi->vlan = ice_vf_get_port_vlan_id(vf);
   1292	ivi->qos = ice_vf_get_port_vlan_prio(vf);
   1293	if (ice_vf_is_port_vlan_ena(vf))
   1294		ivi->vlan_proto = cpu_to_be16(ice_vf_get_port_vlan_tpid(vf));
   1295
   1296	ivi->trusted = vf->trusted;
   1297	ivi->spoofchk = vf->spoofchk;
   1298	if (!vf->link_forced)
   1299		ivi->linkstate = IFLA_VF_LINK_STATE_AUTO;
   1300	else if (vf->link_up)
   1301		ivi->linkstate = IFLA_VF_LINK_STATE_ENABLE;
   1302	else
   1303		ivi->linkstate = IFLA_VF_LINK_STATE_DISABLE;
   1304	ivi->max_tx_rate = vf->max_tx_rate;
   1305	ivi->min_tx_rate = vf->min_tx_rate;
   1306
   1307out_put_vf:
   1308	ice_put_vf(vf);
   1309	return ret;
   1310}
   1311
   1312/**
   1313 * ice_unicast_mac_exists - check if the unicast MAC exists on the PF's switch
   1314 * @pf: PF used to reference the switch's rules
   1315 * @umac: unicast MAC to compare against existing switch rules
   1316 *
   1317 * Return true on the first/any match, else return false
   1318 */
   1319static bool ice_unicast_mac_exists(struct ice_pf *pf, u8 *umac)
   1320{
   1321	struct ice_sw_recipe *mac_recipe_list =
   1322		&pf->hw.switch_info->recp_list[ICE_SW_LKUP_MAC];
   1323	struct ice_fltr_mgmt_list_entry *list_itr;
   1324	struct list_head *rule_head;
   1325	struct mutex *rule_lock; /* protect MAC filter list access */
   1326
   1327	rule_head = &mac_recipe_list->filt_rules;
   1328	rule_lock = &mac_recipe_list->filt_rule_lock;
   1329
   1330	mutex_lock(rule_lock);
   1331	list_for_each_entry(list_itr, rule_head, list_entry) {
   1332		u8 *existing_mac = &list_itr->fltr_info.l_data.mac.mac_addr[0];
   1333
   1334		if (ether_addr_equal(existing_mac, umac)) {
   1335			mutex_unlock(rule_lock);
   1336			return true;
   1337		}
   1338	}
   1339
   1340	mutex_unlock(rule_lock);
   1341
   1342	return false;
   1343}
   1344
   1345/**
   1346 * ice_set_vf_mac
   1347 * @netdev: network interface device structure
   1348 * @vf_id: VF identifier
   1349 * @mac: MAC address
   1350 *
   1351 * program VF MAC address
   1352 */
   1353int ice_set_vf_mac(struct net_device *netdev, int vf_id, u8 *mac)
   1354{
   1355	struct ice_pf *pf = ice_netdev_to_pf(netdev);
   1356	struct ice_vf *vf;
   1357	int ret;
   1358
   1359	if (is_multicast_ether_addr(mac)) {
   1360		netdev_err(netdev, "%pM not a valid unicast address\n", mac);
   1361		return -EINVAL;
   1362	}
   1363
   1364	vf = ice_get_vf_by_id(pf, vf_id);
   1365	if (!vf)
   1366		return -EINVAL;
   1367
   1368	/* nothing left to do, unicast MAC already set */
   1369	if (ether_addr_equal(vf->dev_lan_addr.addr, mac) &&
   1370	    ether_addr_equal(vf->hw_lan_addr.addr, mac)) {
   1371		ret = 0;
   1372		goto out_put_vf;
   1373	}
   1374
   1375	ret = ice_check_vf_ready_for_cfg(vf);
   1376	if (ret)
   1377		goto out_put_vf;
   1378
   1379	if (ice_unicast_mac_exists(pf, mac)) {
   1380		netdev_err(netdev, "Unicast MAC %pM already exists on this PF. Preventing setting VF %u unicast MAC address to %pM\n",
   1381			   mac, vf_id, mac);
   1382		ret = -EINVAL;
   1383		goto out_put_vf;
   1384	}
   1385
   1386	mutex_lock(&vf->cfg_lock);
   1387
   1388	/* VF is notified of its new MAC via the PF's response to the
   1389	 * VIRTCHNL_OP_GET_VF_RESOURCES message after the VF has been reset
   1390	 */
   1391	ether_addr_copy(vf->dev_lan_addr.addr, mac);
   1392	ether_addr_copy(vf->hw_lan_addr.addr, mac);
   1393	if (is_zero_ether_addr(mac)) {
   1394		/* VF will send VIRTCHNL_OP_ADD_ETH_ADDR message with its MAC */
   1395		vf->pf_set_mac = false;
   1396		netdev_info(netdev, "Removing MAC on VF %d. VF driver will be reinitialized\n",
   1397			    vf->vf_id);
   1398	} else {
   1399		/* PF will add MAC rule for the VF */
   1400		vf->pf_set_mac = true;
   1401		netdev_info(netdev, "Setting MAC %pM on VF %d. VF driver will be reinitialized\n",
   1402			    mac, vf_id);
   1403	}
   1404
   1405	ice_reset_vf(vf, ICE_VF_RESET_NOTIFY);
   1406	mutex_unlock(&vf->cfg_lock);
   1407
   1408out_put_vf:
   1409	ice_put_vf(vf);
   1410	return ret;
   1411}
   1412
   1413/**
   1414 * ice_set_vf_trust
   1415 * @netdev: network interface device structure
   1416 * @vf_id: VF identifier
   1417 * @trusted: Boolean value to enable/disable trusted VF
   1418 *
   1419 * Enable or disable a given VF as trusted
   1420 */
   1421int ice_set_vf_trust(struct net_device *netdev, int vf_id, bool trusted)
   1422{
   1423	struct ice_pf *pf = ice_netdev_to_pf(netdev);
   1424	struct ice_vf *vf;
   1425	int ret;
   1426
   1427	if (ice_is_eswitch_mode_switchdev(pf)) {
   1428		dev_info(ice_pf_to_dev(pf), "Trusted VF is forbidden in switchdev mode\n");
   1429		return -EOPNOTSUPP;
   1430	}
   1431
   1432	vf = ice_get_vf_by_id(pf, vf_id);
   1433	if (!vf)
   1434		return -EINVAL;
   1435
   1436	ret = ice_check_vf_ready_for_cfg(vf);
   1437	if (ret)
   1438		goto out_put_vf;
   1439
   1440	/* Check if already trusted */
   1441	if (trusted == vf->trusted) {
   1442		ret = 0;
   1443		goto out_put_vf;
   1444	}
   1445
   1446	mutex_lock(&vf->cfg_lock);
   1447
   1448	vf->trusted = trusted;
   1449	ice_reset_vf(vf, ICE_VF_RESET_NOTIFY);
   1450	dev_info(ice_pf_to_dev(pf), "VF %u is now %strusted\n",
   1451		 vf_id, trusted ? "" : "un");
   1452
   1453	mutex_unlock(&vf->cfg_lock);
   1454
   1455out_put_vf:
   1456	ice_put_vf(vf);
   1457	return ret;
   1458}
   1459
   1460/**
   1461 * ice_set_vf_link_state
   1462 * @netdev: network interface device structure
   1463 * @vf_id: VF identifier
   1464 * @link_state: required link state
   1465 *
   1466 * Set VF's link state, irrespective of physical link state status
   1467 */
   1468int ice_set_vf_link_state(struct net_device *netdev, int vf_id, int link_state)
   1469{
   1470	struct ice_pf *pf = ice_netdev_to_pf(netdev);
   1471	struct ice_vf *vf;
   1472	int ret;
   1473
   1474	vf = ice_get_vf_by_id(pf, vf_id);
   1475	if (!vf)
   1476		return -EINVAL;
   1477
   1478	ret = ice_check_vf_ready_for_cfg(vf);
   1479	if (ret)
   1480		goto out_put_vf;
   1481
   1482	switch (link_state) {
   1483	case IFLA_VF_LINK_STATE_AUTO:
   1484		vf->link_forced = false;
   1485		break;
   1486	case IFLA_VF_LINK_STATE_ENABLE:
   1487		vf->link_forced = true;
   1488		vf->link_up = true;
   1489		break;
   1490	case IFLA_VF_LINK_STATE_DISABLE:
   1491		vf->link_forced = true;
   1492		vf->link_up = false;
   1493		break;
   1494	default:
   1495		ret = -EINVAL;
   1496		goto out_put_vf;
   1497	}
   1498
   1499	ice_vc_notify_vf_link_state(vf);
   1500
   1501out_put_vf:
   1502	ice_put_vf(vf);
   1503	return ret;
   1504}
   1505
   1506/**
   1507 * ice_calc_all_vfs_min_tx_rate - calculate cumulative min Tx rate on all VFs
   1508 * @pf: PF associated with VFs
   1509 */
   1510static int ice_calc_all_vfs_min_tx_rate(struct ice_pf *pf)
   1511{
   1512	struct ice_vf *vf;
   1513	unsigned int bkt;
   1514	int rate = 0;
   1515
   1516	rcu_read_lock();
   1517	ice_for_each_vf_rcu(pf, bkt, vf)
   1518		rate += vf->min_tx_rate;
   1519	rcu_read_unlock();
   1520
   1521	return rate;
   1522}
   1523
   1524/**
   1525 * ice_min_tx_rate_oversubscribed - check if min Tx rate causes oversubscription
   1526 * @vf: VF trying to configure min_tx_rate
   1527 * @min_tx_rate: min Tx rate in Mbps
   1528 *
   1529 * Check if the min_tx_rate being passed in will cause oversubscription of total
   1530 * min_tx_rate based on the current link speed and all other VFs configured
   1531 * min_tx_rate
   1532 *
   1533 * Return true if the passed min_tx_rate would cause oversubscription, else
   1534 * return false
   1535 */
   1536static bool
   1537ice_min_tx_rate_oversubscribed(struct ice_vf *vf, int min_tx_rate)
   1538{
   1539	struct ice_vsi *vsi = ice_get_vf_vsi(vf);
   1540	int all_vfs_min_tx_rate;
   1541	int link_speed_mbps;
   1542
   1543	if (WARN_ON(!vsi))
   1544		return false;
   1545
   1546	link_speed_mbps = ice_get_link_speed_mbps(vsi);
   1547	all_vfs_min_tx_rate = ice_calc_all_vfs_min_tx_rate(vf->pf);
   1548
   1549	/* this VF's previous rate is being overwritten */
   1550	all_vfs_min_tx_rate -= vf->min_tx_rate;
   1551
   1552	if (all_vfs_min_tx_rate + min_tx_rate > link_speed_mbps) {
   1553		dev_err(ice_pf_to_dev(vf->pf), "min_tx_rate of %d Mbps on VF %u would cause oversubscription of %d Mbps based on the current link speed %d Mbps\n",
   1554			min_tx_rate, vf->vf_id,
   1555			all_vfs_min_tx_rate + min_tx_rate - link_speed_mbps,
   1556			link_speed_mbps);
   1557		return true;
   1558	}
   1559
   1560	return false;
   1561}
   1562
   1563/**
   1564 * ice_set_vf_bw - set min/max VF bandwidth
   1565 * @netdev: network interface device structure
   1566 * @vf_id: VF identifier
   1567 * @min_tx_rate: Minimum Tx rate in Mbps
   1568 * @max_tx_rate: Maximum Tx rate in Mbps
   1569 */
   1570int
   1571ice_set_vf_bw(struct net_device *netdev, int vf_id, int min_tx_rate,
   1572	      int max_tx_rate)
   1573{
   1574	struct ice_pf *pf = ice_netdev_to_pf(netdev);
   1575	struct ice_vsi *vsi;
   1576	struct device *dev;
   1577	struct ice_vf *vf;
   1578	int ret;
   1579
   1580	dev = ice_pf_to_dev(pf);
   1581
   1582	vf = ice_get_vf_by_id(pf, vf_id);
   1583	if (!vf)
   1584		return -EINVAL;
   1585
   1586	ret = ice_check_vf_ready_for_cfg(vf);
   1587	if (ret)
   1588		goto out_put_vf;
   1589
   1590	vsi = ice_get_vf_vsi(vf);
   1591	if (!vsi) {
   1592		ret = -EINVAL;
   1593		goto out_put_vf;
   1594	}
   1595
   1596	/* when max_tx_rate is zero that means no max Tx rate limiting, so only
   1597	 * check if max_tx_rate is non-zero
   1598	 */
   1599	if (max_tx_rate && min_tx_rate > max_tx_rate) {
   1600		dev_err(dev, "Cannot set min Tx rate %d Mbps greater than max Tx rate %d Mbps\n",
   1601			min_tx_rate, max_tx_rate);
   1602		ret = -EINVAL;
   1603		goto out_put_vf;
   1604	}
   1605
   1606	if (min_tx_rate && ice_is_dcb_active(pf)) {
   1607		dev_err(dev, "DCB on PF is currently enabled. VF min Tx rate limiting not allowed on this PF.\n");
   1608		ret = -EOPNOTSUPP;
   1609		goto out_put_vf;
   1610	}
   1611
   1612	if (ice_min_tx_rate_oversubscribed(vf, min_tx_rate)) {
   1613		ret = -EINVAL;
   1614		goto out_put_vf;
   1615	}
   1616
   1617	if (vf->min_tx_rate != (unsigned int)min_tx_rate) {
   1618		ret = ice_set_min_bw_limit(vsi, (u64)min_tx_rate * 1000);
   1619		if (ret) {
   1620			dev_err(dev, "Unable to set min-tx-rate for VF %d\n",
   1621				vf->vf_id);
   1622			goto out_put_vf;
   1623		}
   1624
   1625		vf->min_tx_rate = min_tx_rate;
   1626	}
   1627
   1628	if (vf->max_tx_rate != (unsigned int)max_tx_rate) {
   1629		ret = ice_set_max_bw_limit(vsi, (u64)max_tx_rate * 1000);
   1630		if (ret) {
   1631			dev_err(dev, "Unable to set max-tx-rate for VF %d\n",
   1632				vf->vf_id);
   1633			goto out_put_vf;
   1634		}
   1635
   1636		vf->max_tx_rate = max_tx_rate;
   1637	}
   1638
   1639out_put_vf:
   1640	ice_put_vf(vf);
   1641	return ret;
   1642}
   1643
   1644/**
   1645 * ice_get_vf_stats - populate some stats for the VF
   1646 * @netdev: the netdev of the PF
   1647 * @vf_id: the host OS identifier (0-255)
   1648 * @vf_stats: pointer to the OS memory to be initialized
   1649 */
   1650int ice_get_vf_stats(struct net_device *netdev, int vf_id,
   1651		     struct ifla_vf_stats *vf_stats)
   1652{
   1653	struct ice_pf *pf = ice_netdev_to_pf(netdev);
   1654	struct ice_eth_stats *stats;
   1655	struct ice_vsi *vsi;
   1656	struct ice_vf *vf;
   1657	int ret;
   1658
   1659	vf = ice_get_vf_by_id(pf, vf_id);
   1660	if (!vf)
   1661		return -EINVAL;
   1662
   1663	ret = ice_check_vf_ready_for_cfg(vf);
   1664	if (ret)
   1665		goto out_put_vf;
   1666
   1667	vsi = ice_get_vf_vsi(vf);
   1668	if (!vsi) {
   1669		ret = -EINVAL;
   1670		goto out_put_vf;
   1671	}
   1672
   1673	ice_update_eth_stats(vsi);
   1674	stats = &vsi->eth_stats;
   1675
   1676	memset(vf_stats, 0, sizeof(*vf_stats));
   1677
   1678	vf_stats->rx_packets = stats->rx_unicast + stats->rx_broadcast +
   1679		stats->rx_multicast;
   1680	vf_stats->tx_packets = stats->tx_unicast + stats->tx_broadcast +
   1681		stats->tx_multicast;
   1682	vf_stats->rx_bytes   = stats->rx_bytes;
   1683	vf_stats->tx_bytes   = stats->tx_bytes;
   1684	vf_stats->broadcast  = stats->rx_broadcast;
   1685	vf_stats->multicast  = stats->rx_multicast;
   1686	vf_stats->rx_dropped = stats->rx_discards;
   1687	vf_stats->tx_dropped = stats->tx_discards;
   1688
   1689out_put_vf:
   1690	ice_put_vf(vf);
   1691	return ret;
   1692}
   1693
   1694/**
   1695 * ice_is_supported_port_vlan_proto - make sure the vlan_proto is supported
   1696 * @hw: hardware structure used to check the VLAN mode
   1697 * @vlan_proto: VLAN TPID being checked
   1698 *
   1699 * If the device is configured in Double VLAN Mode (DVM), then both ETH_P_8021Q
   1700 * and ETH_P_8021AD are supported. If the device is configured in Single VLAN
   1701 * Mode (SVM), then only ETH_P_8021Q is supported.
   1702 */
   1703static bool
   1704ice_is_supported_port_vlan_proto(struct ice_hw *hw, u16 vlan_proto)
   1705{
   1706	bool is_supported = false;
   1707
   1708	switch (vlan_proto) {
   1709	case ETH_P_8021Q:
   1710		is_supported = true;
   1711		break;
   1712	case ETH_P_8021AD:
   1713		if (ice_is_dvm_ena(hw))
   1714			is_supported = true;
   1715		break;
   1716	}
   1717
   1718	return is_supported;
   1719}
   1720
   1721/**
   1722 * ice_set_vf_port_vlan
   1723 * @netdev: network interface device structure
   1724 * @vf_id: VF identifier
   1725 * @vlan_id: VLAN ID being set
   1726 * @qos: priority setting
   1727 * @vlan_proto: VLAN protocol
   1728 *
   1729 * program VF Port VLAN ID and/or QoS
   1730 */
   1731int
   1732ice_set_vf_port_vlan(struct net_device *netdev, int vf_id, u16 vlan_id, u8 qos,
   1733		     __be16 vlan_proto)
   1734{
   1735	struct ice_pf *pf = ice_netdev_to_pf(netdev);
   1736	u16 local_vlan_proto = ntohs(vlan_proto);
   1737	struct device *dev;
   1738	struct ice_vf *vf;
   1739	int ret;
   1740
   1741	dev = ice_pf_to_dev(pf);
   1742
   1743	if (vlan_id >= VLAN_N_VID || qos > 7) {
   1744		dev_err(dev, "Invalid Port VLAN parameters for VF %d, ID %d, QoS %d\n",
   1745			vf_id, vlan_id, qos);
   1746		return -EINVAL;
   1747	}
   1748
   1749	if (!ice_is_supported_port_vlan_proto(&pf->hw, local_vlan_proto)) {
   1750		dev_err(dev, "VF VLAN protocol 0x%04x is not supported\n",
   1751			local_vlan_proto);
   1752		return -EPROTONOSUPPORT;
   1753	}
   1754
   1755	vf = ice_get_vf_by_id(pf, vf_id);
   1756	if (!vf)
   1757		return -EINVAL;
   1758
   1759	ret = ice_check_vf_ready_for_cfg(vf);
   1760	if (ret)
   1761		goto out_put_vf;
   1762
   1763	if (ice_vf_get_port_vlan_prio(vf) == qos &&
   1764	    ice_vf_get_port_vlan_tpid(vf) == local_vlan_proto &&
   1765	    ice_vf_get_port_vlan_id(vf) == vlan_id) {
   1766		/* duplicate request, so just return success */
   1767		dev_dbg(dev, "Duplicate port VLAN %u, QoS %u, TPID 0x%04x request\n",
   1768			vlan_id, qos, local_vlan_proto);
   1769		ret = 0;
   1770		goto out_put_vf;
   1771	}
   1772
   1773	mutex_lock(&vf->cfg_lock);
   1774
   1775	vf->port_vlan_info = ICE_VLAN(local_vlan_proto, vlan_id, qos);
   1776	if (ice_vf_is_port_vlan_ena(vf))
   1777		dev_info(dev, "Setting VLAN %u, QoS %u, TPID 0x%04x on VF %d\n",
   1778			 vlan_id, qos, local_vlan_proto, vf_id);
   1779	else
   1780		dev_info(dev, "Clearing port VLAN on VF %d\n", vf_id);
   1781
   1782	ice_reset_vf(vf, ICE_VF_RESET_NOTIFY);
   1783	mutex_unlock(&vf->cfg_lock);
   1784
   1785out_put_vf:
   1786	ice_put_vf(vf);
   1787	return ret;
   1788}
   1789
   1790/**
   1791 * ice_print_vf_rx_mdd_event - print VF Rx malicious driver detect event
   1792 * @vf: pointer to the VF structure
   1793 */
   1794void ice_print_vf_rx_mdd_event(struct ice_vf *vf)
   1795{
   1796	struct ice_pf *pf = vf->pf;
   1797	struct device *dev;
   1798
   1799	dev = ice_pf_to_dev(pf);
   1800
   1801	dev_info(dev, "%d Rx Malicious Driver Detection events detected on PF %d VF %d MAC %pM. mdd-auto-reset-vfs=%s\n",
   1802		 vf->mdd_rx_events.count, pf->hw.pf_id, vf->vf_id,
   1803		 vf->dev_lan_addr.addr,
   1804		 test_bit(ICE_FLAG_MDD_AUTO_RESET_VF, pf->flags)
   1805			  ? "on" : "off");
   1806}
   1807
   1808/**
   1809 * ice_print_vfs_mdd_events - print VFs malicious driver detect event
   1810 * @pf: pointer to the PF structure
   1811 *
   1812 * Called from ice_handle_mdd_event to rate limit and print VFs MDD events.
   1813 */
   1814void ice_print_vfs_mdd_events(struct ice_pf *pf)
   1815{
   1816	struct device *dev = ice_pf_to_dev(pf);
   1817	struct ice_hw *hw = &pf->hw;
   1818	struct ice_vf *vf;
   1819	unsigned int bkt;
   1820
   1821	/* check that there are pending MDD events to print */
   1822	if (!test_and_clear_bit(ICE_MDD_VF_PRINT_PENDING, pf->state))
   1823		return;
   1824
   1825	/* VF MDD event logs are rate limited to one second intervals */
   1826	if (time_is_after_jiffies(pf->vfs.last_printed_mdd_jiffies + HZ * 1))
   1827		return;
   1828
   1829	pf->vfs.last_printed_mdd_jiffies = jiffies;
   1830
   1831	mutex_lock(&pf->vfs.table_lock);
   1832	ice_for_each_vf(pf, bkt, vf) {
   1833		/* only print Rx MDD event message if there are new events */
   1834		if (vf->mdd_rx_events.count != vf->mdd_rx_events.last_printed) {
   1835			vf->mdd_rx_events.last_printed =
   1836							vf->mdd_rx_events.count;
   1837			ice_print_vf_rx_mdd_event(vf);
   1838		}
   1839
   1840		/* only print Tx MDD event message if there are new events */
   1841		if (vf->mdd_tx_events.count != vf->mdd_tx_events.last_printed) {
   1842			vf->mdd_tx_events.last_printed =
   1843							vf->mdd_tx_events.count;
   1844
   1845			dev_info(dev, "%d Tx Malicious Driver Detection events detected on PF %d VF %d MAC %pM.\n",
   1846				 vf->mdd_tx_events.count, hw->pf_id, vf->vf_id,
   1847				 vf->dev_lan_addr.addr);
   1848		}
   1849	}
   1850	mutex_unlock(&pf->vfs.table_lock);
   1851}
   1852
   1853/**
   1854 * ice_restore_all_vfs_msi_state - restore VF MSI state after PF FLR
   1855 * @pdev: pointer to a pci_dev structure
   1856 *
   1857 * Called when recovering from a PF FLR to restore interrupt capability to
   1858 * the VFs.
   1859 */
   1860void ice_restore_all_vfs_msi_state(struct pci_dev *pdev)
   1861{
   1862	u16 vf_id;
   1863	int pos;
   1864
   1865	if (!pci_num_vf(pdev))
   1866		return;
   1867
   1868	pos = pci_find_ext_capability(pdev, PCI_EXT_CAP_ID_SRIOV);
   1869	if (pos) {
   1870		struct pci_dev *vfdev;
   1871
   1872		pci_read_config_word(pdev, pos + PCI_SRIOV_VF_DID,
   1873				     &vf_id);
   1874		vfdev = pci_get_device(pdev->vendor, vf_id, NULL);
   1875		while (vfdev) {
   1876			if (vfdev->is_virtfn && vfdev->physfn == pdev)
   1877				pci_restore_msi_state(vfdev);
   1878			vfdev = pci_get_device(pdev->vendor, vf_id,
   1879					       vfdev);
   1880		}
   1881	}
   1882}
   1883
   1884/**
   1885 * ice_is_malicious_vf - helper function to detect a malicious VF
   1886 * @pf: ptr to struct ice_pf
   1887 * @event: pointer to the AQ event
   1888 * @num_msg_proc: the number of messages processed so far
   1889 * @num_msg_pending: the number of messages peinding in admin queue
   1890 */
   1891bool
   1892ice_is_malicious_vf(struct ice_pf *pf, struct ice_rq_event_info *event,
   1893		    u16 num_msg_proc, u16 num_msg_pending)
   1894{
   1895	s16 vf_id = le16_to_cpu(event->desc.retval);
   1896	struct device *dev = ice_pf_to_dev(pf);
   1897	struct ice_mbx_data mbxdata;
   1898	bool malvf = false;
   1899	struct ice_vf *vf;
   1900	int status;
   1901
   1902	vf = ice_get_vf_by_id(pf, vf_id);
   1903	if (!vf)
   1904		return false;
   1905
   1906	if (test_bit(ICE_VF_STATE_DIS, vf->vf_states))
   1907		goto out_put_vf;
   1908
   1909	mbxdata.num_msg_proc = num_msg_proc;
   1910	mbxdata.num_pending_arq = num_msg_pending;
   1911	mbxdata.max_num_msgs_mbx = pf->hw.mailboxq.num_rq_entries;
   1912#define ICE_MBX_OVERFLOW_WATERMARK 64
   1913	mbxdata.async_watermark_val = ICE_MBX_OVERFLOW_WATERMARK;
   1914
   1915	/* check to see if we have a malicious VF */
   1916	status = ice_mbx_vf_state_handler(&pf->hw, &mbxdata, vf_id, &malvf);
   1917	if (status)
   1918		goto out_put_vf;
   1919
   1920	if (malvf) {
   1921		bool report_vf = false;
   1922
   1923		/* if the VF is malicious and we haven't let the user
   1924		 * know about it, then let them know now
   1925		 */
   1926		status = ice_mbx_report_malvf(&pf->hw, pf->vfs.malvfs,
   1927					      ICE_MAX_SRIOV_VFS, vf_id,
   1928					      &report_vf);
   1929		if (status)
   1930			dev_dbg(dev, "Error reporting malicious VF\n");
   1931
   1932		if (report_vf) {
   1933			struct ice_vsi *pf_vsi = ice_get_main_vsi(pf);
   1934
   1935			if (pf_vsi)
   1936				dev_warn(dev, "VF MAC %pM on PF MAC %pM is generating asynchronous messages and may be overflowing the PF message queue. Please see the Adapter User Guide for more information\n",
   1937					 &vf->dev_lan_addr.addr[0],
   1938					 pf_vsi->netdev->dev_addr);
   1939		}
   1940	}
   1941
   1942out_put_vf:
   1943	ice_put_vf(vf);
   1944	return malvf;
   1945}