/* bnx2x_cmn.c: Broadcom Everest network driver.
 *
 * Copyright (c) 2007-2013 Broadcom Corporation
 *
 * This program is free software; you can redistribute it and/or modify
 * it under the terms of the GNU General Public License as published by
 * the Free Software Foundation.
 *
 * Maintained by: Eilon Greenstein <eilong@broadcom.com>
 * Written by: Eliezer Tamir
 * Based on code from Michael Chan's bnx2 driver
 * UDP CSUM errata workaround by Arik Gendelman
 * Slowpath and fastpath rework by Vladislav Zolotarov
 * Statistics and Link management by Yitchak Gertner
 *
 */

#define pr_fmt(fmt) KBUILD_MODNAME ": " fmt

#include <linux/etherdevice.h>
#include <linux/if_vlan.h>
#include <linux/interrupt.h>
#include <linux/ip.h>
#include <net/tcp.h>
#include <net/ipv6.h>
#include <net/ip6_checksum.h>
#include <net/busy_poll.h>
#include <linux/prefetch.h>
#include "bnx2x_cmn.h"
#include "bnx2x_init.h"
#include "bnx2x_sp.h"

/**
 * bnx2x_move_fp - move content of the fastpath structure.
 *
 * @bp:         driver handle
 * @from:       source FP index
 * @to:         destination FP index
 *
 * Makes sure the contents of the bp->fp[to].napi is kept
 * intact. This is done by first copying the napi struct from
 * the target to the source, and then mem copying the entire
 * source onto the target. Update txdata pointers and related
 * content.
 */
static inline void bnx2x_move_fp(struct bnx2x *bp, int from, int to)
{
        struct bnx2x_fastpath *from_fp = &bp->fp[from];
        struct bnx2x_fastpath *to_fp = &bp->fp[to];
        struct bnx2x_sp_objs *from_sp_objs = &bp->sp_objs[from];
        struct bnx2x_sp_objs *to_sp_objs = &bp->sp_objs[to];
        struct bnx2x_fp_stats *from_fp_stats = &bp->fp_stats[from];
        struct bnx2x_fp_stats *to_fp_stats = &bp->fp_stats[to];
        int old_max_eth_txqs, new_max_eth_txqs;
        int old_txdata_index = 0, new_txdata_index = 0;
        struct bnx2x_agg_info *old_tpa_info = to_fp->tpa_info;

        /* Copy the NAPI object as it has been already initialized */
        from_fp->napi = to_fp->napi;

        /* Move bnx2x_fastpath contents */
        memcpy(to_fp, from_fp, sizeof(*to_fp));
        to_fp->index = to;

        /* Retain the tpa_info of the original `to' version as we don't want
         * 2 FPs to contain the same tpa_info pointer.
         */
        to_fp->tpa_info = old_tpa_info;

        /* move sp_objs contents as well, as their indices match fp ones */
        memcpy(to_sp_objs, from_sp_objs, sizeof(*to_sp_objs));

        /* move fp_stats contents as well, as their indices match fp ones */
        memcpy(to_fp_stats, from_fp_stats, sizeof(*to_fp_stats));

        /* Update txdata pointers in fp and move txdata content accordingly:
         * Each fp consumes 'max_cos' txdata structures, so the index should be
         * decremented by max_cos x delta.
         */

        old_max_eth_txqs = BNX2X_NUM_ETH_QUEUES(bp) * (bp)->max_cos;
        new_max_eth_txqs = (BNX2X_NUM_ETH_QUEUES(bp) - from + to) *
                                (bp)->max_cos;
        if (from == FCOE_IDX(bp)) {
                old_txdata_index = old_max_eth_txqs + FCOE_TXQ_IDX_OFFSET;
                new_txdata_index = new_max_eth_txqs + FCOE_TXQ_IDX_OFFSET;
        }

        memcpy(&bp->bnx2x_txq[new_txdata_index],
               &bp->bnx2x_txq[old_txdata_index],
               sizeof(struct bnx2x_fp_txdata));
        to_fp->txdata_ptr[0] = &bp->bnx2x_txq[new_txdata_index];
}

/**
 * bnx2x_fill_fw_str - Fill buffer with FW version string.
 *
 * @bp:        driver handle
 * @buf:       character buffer to fill with the fw name
 * @buf_len:   length of the above buffer
 *
 */
void bnx2x_fill_fw_str(struct bnx2x *bp, char *buf, size_t buf_len)
{
        if (IS_PF(bp)) {
                u8 phy_fw_ver[PHY_FW_VER_LEN];

                phy_fw_ver[0] = '\0';
                bnx2x_get_ext_phy_fw_version(&bp->link_params,
                                             phy_fw_ver, PHY_FW_VER_LEN);
                strlcpy(buf, bp->fw_ver, buf_len);
                snprintf(buf + strlen(bp->fw_ver), 32 - strlen(bp->fw_ver),
                         "bc %d.%d.%d%s%s",
                         (bp->common.bc_ver & 0xff0000) >> 16,
                         (bp->common.bc_ver & 0xff00) >> 8,
                         (bp->common.bc_ver & 0xff),
                         ((phy_fw_ver[0] != '\0') ? " phy " : ""), phy_fw_ver);
        } else {
                bnx2x_vf_fill_fw_str(bp, buf, buf_len);
        }
}

/**
 * bnx2x_shrink_eth_fp - guarantees fastpath structures stay intact
 *
 * @bp: driver handle
 * @delta:      number of eth queues which were not allocated
 */
static void bnx2x_shrink_eth_fp(struct bnx2x *bp, int delta)
{
        int i, cos, old_eth_num = BNX2X_NUM_ETH_QUEUES(bp);

        /* Queue pointer cannot be re-set on an fp-basis, as moving pointer
         * backward along the array could cause memory to be overridden
         */
        for (cos = 1; cos < bp->max_cos; cos++) {
                for (i = 0; i < old_eth_num - delta; i++) {
                        struct bnx2x_fastpath *fp = &bp->fp[i];
                        int new_idx = cos * (old_eth_num - delta) + i;

                        memcpy(&bp->bnx2x_txq[new_idx], fp->txdata_ptr[cos],
                               sizeof(struct bnx2x_fp_txdata));
                        fp->txdata_ptr[cos] = &bp->bnx2x_txq[new_idx];
                }
        }
}

int load_count[2][3] = { {0} }; /* per-path: 0-common, 1-port0, 2-port1 */

/* free skb in the packet ring at pos idx
 * return idx of last bd freed
 */
static u16 bnx2x_free_tx_pkt(struct bnx2x *bp, struct bnx2x_fp_txdata *txdata,
                             u16 idx, unsigned int *pkts_compl,
                             unsigned int *bytes_compl)
{
        struct sw_tx_bd *tx_buf = &txdata->tx_buf_ring[idx];
        struct eth_tx_start_bd *tx_start_bd;
        struct eth_tx_bd *tx_data_bd;
        struct sk_buff *skb = tx_buf->skb;
        u16 bd_idx = TX_BD(tx_buf->first_bd), new_cons;
        int nbd;

        /* prefetch skb end pointer to speedup dev_kfree_skb() */
        prefetch(&skb->end);

        DP(NETIF_MSG_TX_DONE, "fp[%d]: pkt_idx %d  buff @(%p)->skb %p\n",
           txdata->txq_index, idx, tx_buf, skb);

        /* unmap first bd */
        tx_start_bd = &txdata->tx_desc_ring[bd_idx].start_bd;
        dma_unmap_single(&bp->pdev->dev, BD_UNMAP_ADDR(tx_start_bd),
                         BD_UNMAP_LEN(tx_start_bd), DMA_TO_DEVICE);

        nbd = le16_to_cpu(tx_start_bd->nbd) - 1;
#ifdef BNX2X_STOP_ON_ERROR
        if ((nbd - 1) > (MAX_SKB_FRAGS + 2)) {
                BNX2X_ERR("BAD nbd!\n");
                bnx2x_panic();
        }
#endif
        new_cons = nbd + tx_buf->first_bd;

        /* Get the next bd */
        bd_idx = TX_BD(NEXT_TX_IDX(bd_idx));

        /* Skip a parse bd... */
        --nbd;
        bd_idx = TX_BD(NEXT_TX_IDX(bd_idx));

        /* ...and the TSO split header bd since they have no mapping */
        if (tx_buf->flags & BNX2X_TSO_SPLIT_BD) {
                --nbd;
                bd_idx = TX_BD(NEXT_TX_IDX(bd_idx));
        }

        /* now free frags */
        while (nbd > 0) {

                tx_data_bd = &txdata->tx_desc_ring[bd_idx].reg_bd;
                dma_unmap_page(&bp->pdev->dev, BD_UNMAP_ADDR(tx_data_bd),
                               BD_UNMAP_LEN(tx_data_bd), DMA_TO_DEVICE);
                if (--nbd)
                        bd_idx = TX_BD(NEXT_TX_IDX(bd_idx));
        }

        /* release skb */
        WARN_ON(!skb);
        if (likely(skb)) {
                (*pkts_compl)++;
                (*bytes_compl) += skb->len;
        }

        dev_kfree_skb_any(skb);
        tx_buf->first_bd = 0;
        tx_buf->skb = NULL;

        return new_cons;
}

int bnx2x_tx_int(struct bnx2x *bp, struct bnx2x_fp_txdata *txdata)
{
        struct netdev_queue *txq;
        u16 hw_cons, sw_cons, bd_cons = txdata->tx_bd_cons;
        unsigned int pkts_compl = 0, bytes_compl = 0;

#ifdef BNX2X_STOP_ON_ERROR
        if (unlikely(bp->panic))
                return -1;
#endif

        txq = netdev_get_tx_queue(bp->dev, txdata->txq_index);
        hw_cons = le16_to_cpu(*txdata->tx_cons_sb);
        sw_cons = txdata->tx_pkt_cons;

        while (sw_cons != hw_cons) {
                u16 pkt_cons;

                pkt_cons = TX_BD(sw_cons);

                DP(NETIF_MSG_TX_DONE,
                   "queue[%d]: hw_cons %u  sw_cons %u  pkt_cons %u\n",
                   txdata->txq_index, hw_cons, sw_cons, pkt_cons);

                bd_cons = bnx2x_free_tx_pkt(bp, txdata, pkt_cons,
                                            &pkts_compl, &bytes_compl);

                sw_cons++;
        }

        netdev_tx_completed_queue(txq, pkts_compl, bytes_compl);

        txdata->tx_pkt_cons = sw_cons;
        txdata->tx_bd_cons = bd_cons;

        /* Need to make the tx_bd_cons update visible to start_xmit()
         * before checking for netif_tx_queue_stopped().  Without the
         * memory barrier, there is a small possibility that
         * start_xmit() will miss it and cause the queue to be stopped
         * forever.
         * On the other hand we need an rmb() here to ensure the proper
         * ordering of bit testing in the following
         * netif_tx_queue_stopped(txq) call.
         */
        smp_mb();

        if (unlikely(netif_tx_queue_stopped(txq))) {
                /* Taking tx_lock() is needed to prevent re-enabling the queue
                 * while it's empty. This could have happen if rx_action() gets
                 * suspended in bnx2x_tx_int() after the condition before
                 * netif_tx_wake_queue(), while tx_action (bnx2x_start_xmit()):
                 *
                 * stops the queue->sees fresh tx_bd_cons->releases the queue->
                 * sends some packets consuming the whole queue again->
                 * stops the queue
                 */

                __netif_tx_lock(txq, smp_processor_id());

                if ((netif_tx_queue_stopped(txq)) &&
                    (bp->state == BNX2X_STATE_OPEN) &&
                    (bnx2x_tx_avail(bp, txdata) >= MAX_DESC_PER_TX_PKT))
                        netif_tx_wake_queue(txq);

                __netif_tx_unlock(txq);
        }
        return 0;
}

static inline void bnx2x_update_last_max_sge(struct bnx2x_fastpath *fp,
                                             u16 idx)
{
        u16 last_max = fp->last_max_sge;

        if (SUB_S16(idx, last_max) > 0)
                fp->last_max_sge = idx;
}

static inline void bnx2x_update_sge_prod(struct bnx2x_fastpath *fp,
                                         u16 sge_len,
                                         struct eth_end_agg_rx_cqe *cqe)
{
        struct bnx2x *bp = fp->bp;
        u16 last_max, last_elem, first_elem;
        u16 delta = 0;
        u16 i;

        if (!sge_len)
                return;

        /* First mark all used pages */
        for (i = 0; i < sge_len; i++)
                BIT_VEC64_CLEAR_BIT(fp->sge_mask,
                        RX_SGE(le16_to_cpu(cqe->sgl_or_raw_data.sgl[i])));

        DP(NETIF_MSG_RX_STATUS, "fp_cqe->sgl[%d] = %d\n",
           sge_len - 1, le16_to_cpu(cqe->sgl_or_raw_data.sgl[sge_len - 1]));

        /* Here we assume that the last SGE index is the biggest */
        prefetch((void *)(fp->sge_mask));
        bnx2x_update_last_max_sge(fp,
                le16_to_cpu(cqe->sgl_or_raw_data.sgl[sge_len - 1]));

        last_max = RX_SGE(fp->last_max_sge);
        last_elem = last_max >> BIT_VEC64_ELEM_SHIFT;
        first_elem = RX_SGE(fp->rx_sge_prod) >> BIT_VEC64_ELEM_SHIFT;

        /* If ring is not full */
        if (last_elem + 1 != first_elem)
                last_elem++;

        /* Now update the prod */
        for (i = first_elem; i != last_elem; i = NEXT_SGE_MASK_ELEM(i)) {
                if (likely(fp->sge_mask[i]))
                        break;

                fp->sge_mask[i] = BIT_VEC64_ELEM_ONE_MASK;
                delta += BIT_VEC64_ELEM_SZ;
        }

        if (delta > 0) {
                fp->rx_sge_prod += delta;
                /* clear page-end entries */
                bnx2x_clear_sge_mask_next_elems(fp);
        }

        DP(NETIF_MSG_RX_STATUS,
           "fp->last_max_sge = %d  fp->rx_sge_prod = %d\n",
           fp->last_max_sge, fp->rx_sge_prod);
}

/* Get Toeplitz hash value in the skb using the value from the
 * CQE (calculated by HW).
 */
static u32 bnx2x_get_rxhash(const struct bnx2x *bp,
                            const struct eth_fast_path_rx_cqe *cqe,
                            bool *l4_rxhash)
{
        /* Get Toeplitz hash from CQE */
        if ((bp->dev->features & NETIF_F_RXHASH) &&
            (cqe->status_flags & ETH_FAST_PATH_RX_CQE_RSS_HASH_FLG)) {
                enum eth_rss_hash_type htype;

                htype = cqe->status_flags & ETH_FAST_PATH_RX_CQE_RSS_HASH_TYPE;
                *l4_rxhash = (htype == TCP_IPV4_HASH_TYPE) ||
                             (htype == TCP_IPV6_HASH_TYPE);
                return le32_to_cpu(cqe->rss_hash_result);
        }
        *l4_rxhash = false;
        return 0;
}

static void bnx2x_tpa_start(struct bnx2x_fastpath *fp, u16 queue,
                            u16 cons, u16 prod,
                            struct eth_fast_path_rx_cqe *cqe)
{
        struct bnx2x *bp = fp->bp;
        struct sw_rx_bd *cons_rx_buf = &fp->rx_buf_ring[cons];
        struct sw_rx_bd *prod_rx_buf = &fp->rx_buf_ring[prod];
        struct eth_rx_bd *prod_bd = &fp->rx_desc_ring[prod];
        dma_addr_t mapping;
        struct bnx2x_agg_info *tpa_info = &fp->tpa_info[queue];
        struct sw_rx_bd *first_buf = &tpa_info->first_buf;

        /* print error if current state != stop */
        if (tpa_info->tpa_state != BNX2X_TPA_STOP)
                BNX2X_ERR("start of bin not in stop [%d]\n", queue);

        /* Try to map an empty data buffer from the aggregation info  */
        mapping = dma_map_single(&bp->pdev->dev,
                                 first_buf->data + NET_SKB_PAD,
                                 fp->rx_buf_size, DMA_FROM_DEVICE);
        /*
         *  ...if it fails - move the skb from the consumer to the producer
         *  and set the current aggregation state as ERROR to drop it
         *  when TPA_STOP arrives.
         */

        if (unlikely(dma_mapping_error(&bp->pdev->dev, mapping))) {
                /* Move the BD from the consumer to the producer */
                bnx2x_reuse_rx_data(fp, cons, prod);
                tpa_info->tpa_state = BNX2X_TPA_ERROR;
                return;
        }

        /* move empty data from pool to prod */
        prod_rx_buf->data = first_buf->data;
        dma_unmap_addr_set(prod_rx_buf, mapping, mapping);
        /* point prod_bd to new data */
        prod_bd->addr_hi = cpu_to_le32(U64_HI(mapping));
        prod_bd->addr_lo = cpu_to_le32(U64_LO(mapping));

        /* move partial skb from cons to pool (don't unmap yet) */
        *first_buf = *cons_rx_buf;

        /* mark bin state as START */
        tpa_info->parsing_flags =
                le16_to_cpu(cqe->pars_flags.flags);
        tpa_info->vlan_tag = le16_to_cpu(cqe->vlan_tag);
        tpa_info->tpa_state = BNX2X_TPA_START;
        tpa_info->len_on_bd = le16_to_cpu(cqe->len_on_bd);
        tpa_info->placement_offset = cqe->placement_offset;
        tpa_info->rxhash = bnx2x_get_rxhash(bp, cqe, &tpa_info->l4_rxhash);
        if (fp->mode == TPA_MODE_GRO) {
                u16 gro_size = le16_to_cpu(cqe->pkt_len_or_gro_seg_len);
                tpa_info->full_page = SGE_PAGES / gro_size * gro_size;
                tpa_info->gro_size = gro_size;
        }

#ifdef BNX2X_STOP_ON_ERROR
        fp->tpa_queue_used |= (1 << queue);
#ifdef _ASM_GENERIC_INT_L64_H
        DP(NETIF_MSG_RX_STATUS, "fp->tpa_queue_used = 0x%lx\n",
#else
        DP(NETIF_MSG_RX_STATUS, "fp->tpa_queue_used = 0x%llx\n",
#endif
           fp->tpa_queue_used);
#endif
}

/* Timestamp option length allowed for TPA aggregation:
 *
 *              nop nop kind length echo val
 */
#define TPA_TSTAMP_OPT_LEN      12
/**
 * bnx2x_set_gro_params - compute GRO values
 *
 * @skb:                packet skb
 * @parsing_flags:      parsing flags from the START CQE
 * @len_on_bd:          total length of the first packet for the
 *                      aggregation.
 * @pkt_len:            length of all segments
 *
 * Approximate value of the MSS for this aggregation calculated using
 * the first packet of it.
 * Compute number of aggregated segments, and gso_type.
 */
static void bnx2x_set_gro_params(struct sk_buff *skb, u16 parsing_flags,
                                 u16 len_on_bd, unsigned int pkt_len,
                                 u16 num_of_coalesced_segs)
{
        /* TPA aggregation won't have either IP options or TCP options
         * other than timestamp or IPv6 extension headers.
         */
        u16 hdrs_len = ETH_HLEN + sizeof(struct tcphdr);

        if (GET_FLAG(parsing_flags, PARSING_FLAGS_OVER_ETHERNET_PROTOCOL) ==
            PRS_FLAG_OVERETH_IPV6) {
                hdrs_len += sizeof(struct ipv6hdr);
                skb_shinfo(skb)->gso_type = SKB_GSO_TCPV6;
        } else {
                hdrs_len += sizeof(struct iphdr);
                skb_shinfo(skb)->gso_type = SKB_GSO_TCPV4;
        }

        /* Check if there was a TCP timestamp, if there is it's will
         * always be 12 bytes length: nop nop kind length echo val.
         *
         * Otherwise FW would close the aggregation.
         */
        if (parsing_flags & PARSING_FLAGS_TIME_STAMP_EXIST_FLAG)
                hdrs_len += TPA_TSTAMP_OPT_LEN;

        skb_shinfo(skb)->gso_size = len_on_bd - hdrs_len;

        /* tcp_gro_complete() will copy NAPI_GRO_CB(skb)->count
         * to skb_shinfo(skb)->gso_segs
         */
        NAPI_GRO_CB(skb)->count = num_of_coalesced_segs;
}

static int bnx2x_alloc_rx_sge(struct bnx2x *bp,
                              struct bnx2x_fastpath *fp, u16 index)
{
        struct page *page = alloc_pages(GFP_ATOMIC, PAGES_PER_SGE_SHIFT);
        struct sw_rx_page *sw_buf = &fp->rx_page_ring[index];
        struct eth_rx_sge *sge = &fp->rx_sge_ring[index];
        dma_addr_t mapping;

        if (unlikely(page == NULL)) {
                BNX2X_ERR("Can't alloc sge\n");
                return -ENOMEM;
        }

        mapping = dma_map_page(&bp->pdev->dev, page, 0,
                               SGE_PAGES, DMA_FROM_DEVICE);
        if (unlikely(dma_mapping_error(&bp->pdev->dev, mapping))) {
                __free_pages(page, PAGES_PER_SGE_SHIFT);
                BNX2X_ERR("Can't map sge\n");
                return -ENOMEM;
        }

        sw_buf->page = page;
        dma_unmap_addr_set(sw_buf, mapping, mapping);

        sge->addr_hi = cpu_to_le32(U64_HI(mapping));
        sge->addr_lo = cpu_to_le32(U64_LO(mapping));

        return 0;
}

static int bnx2x_fill_frag_skb(struct bnx2x *bp, struct bnx2x_fastpath *fp,
                               struct bnx2x_agg_info *tpa_info,
                               u16 pages,
                               struct sk_buff *skb,
                               struct eth_end_agg_rx_cqe *cqe,
                               u16 cqe_idx)
{
        struct sw_rx_page *rx_pg, old_rx_pg;
        u32 i, frag_len, frag_size;
        int err, j, frag_id = 0;
        u16 len_on_bd = tpa_info->len_on_bd;
        u16 full_page = 0, gro_size = 0;

        frag_size = le16_to_cpu(cqe->pkt_len) - len_on_bd;

        if (fp->mode == TPA_MODE_GRO) {
                gro_size = tpa_info->gro_size;
                full_page = tpa_info->full_page;
        }

        /* This is needed in order to enable forwarding support */
        if (frag_size)
                bnx2x_set_gro_params(skb, tpa_info->parsing_flags, len_on_bd,
                                     le16_to_cpu(cqe->pkt_len),
                                     le16_to_cpu(cqe->num_of_coalesced_segs));

#ifdef BNX2X_STOP_ON_ERROR
        if (pages > min_t(u32, 8, MAX_SKB_FRAGS) * SGE_PAGES) {
                BNX2X_ERR("SGL length is too long: %d. CQE index is %d\n",
                          pages, cqe_idx);
                BNX2X_ERR("cqe->pkt_len = %d\n", cqe->pkt_len);
                bnx2x_panic();
                return -EINVAL;
        }
#endif

        /* Run through the SGL and compose the fragmented skb */
        for (i = 0, j = 0; i < pages; i += PAGES_PER_SGE, j++) {
                u16 sge_idx = RX_SGE(le16_to_cpu(cqe->sgl_or_raw_data.sgl[j]));

                /* FW gives the indices of the SGE as if the ring is an array
                   (meaning that "next" element will consume 2 indices) */
                if (fp->mode == TPA_MODE_GRO)
                        frag_len = min_t(u32, frag_size, (u32)full_page);
                else /* LRO */
                        frag_len = min_t(u32, frag_size, (u32)SGE_PAGES);

                rx_pg = &fp->rx_page_ring[sge_idx];
                old_rx_pg = *rx_pg;

                /* If we fail to allocate a substitute page, we simply stop
                   where we are and drop the whole packet */
                err = bnx2x_alloc_rx_sge(bp, fp, sge_idx);
                if (unlikely(err)) {
                        bnx2x_fp_qstats(bp, fp)->rx_skb_alloc_failed++;
                        return err;
                }

                /* Unmap the page as we're going to pass it to the stack */
                dma_unmap_page(&bp->pdev->dev,
                               dma_unmap_addr(&old_rx_pg, mapping),
                               SGE_PAGES, DMA_FROM_DEVICE);
                /* Add one frag and update the appropriate fields in the skb */
                if (fp->mode == TPA_MODE_LRO)
                        skb_fill_page_desc(skb, j, old_rx_pg.page, 0, frag_len);
                else { /* GRO */
                        int rem;
                        int offset = 0;
                        for (rem = frag_len; rem > 0; rem -= gro_size) {
                                int len = rem > gro_size ? gro_size : rem;
                                skb_fill_page_desc(skb, frag_id++,
                                                   old_rx_pg.page, offset, len);
                                if (offset)
                                        get_page(old_rx_pg.page);
                                offset += len;
                        }
                }

                skb->data_len += frag_len;
                skb->truesize += SGE_PAGES;
                skb->len += frag_len;

                frag_size -= frag_len;
        }

        return 0;
}

static void bnx2x_frag_free(const struct bnx2x_fastpath *fp, void *data)
{
        if (fp->rx_frag_size)
                put_page(virt_to_head_page(data));
        else
                kfree(data);
}

static void *bnx2x_frag_alloc(const struct bnx2x_fastpath *fp)
{
        if (fp->rx_frag_size)
                return netdev_alloc_frag(fp->rx_frag_size);

        return kmalloc(fp->rx_buf_size + NET_SKB_PAD, GFP_ATOMIC);
}

#ifdef CONFIG_INET
static void bnx2x_gro_ip_csum(struct bnx2x *bp, struct sk_buff *skb)
{
        const struct iphdr *iph = ip_hdr(skb);
        struct tcphdr *th;

        skb_set_transport_header(skb, sizeof(struct iphdr));
        th = tcp_hdr(skb);

        th->check = ~tcp_v4_check(skb->len - skb_transport_offset(skb),
                                  iph->saddr, iph->daddr, 0);
}

static void bnx2x_gro_ipv6_csum(struct bnx2x *bp, struct sk_buff *skb)
{
        struct ipv6hdr *iph = ipv6_hdr(skb);
        struct tcphdr *th;

        skb_set_transport_header(skb, sizeof(struct ipv6hdr));
        th = tcp_hdr(skb);

        th->check = ~tcp_v6_check(skb->len - skb_transport_offset(skb),
                                  &iph->saddr, &iph->daddr, 0);
}

static void bnx2x_gro_csum(struct bnx2x *bp, struct sk_buff *skb,
                            void (*gro_func)(struct bnx2x*, struct sk_buff*))
{
        skb_set_network_header(skb, 0);
        gro_func(bp, skb);
        tcp_gro_complete(skb);
}
#endif

static void bnx2x_gro_receive(struct bnx2x *bp, struct bnx2x_fastpath *fp,
                               struct sk_buff *skb)
{
#ifdef CONFIG_INET
        if (skb_shinfo(skb)->gso_size) {
                switch (be16_to_cpu(skb->protocol)) {
                case ETH_P_IP:
                        bnx2x_gro_csum(bp, skb, bnx2x_gro_ip_csum);
                        break;
                case ETH_P_IPV6:
                        bnx2x_gro_csum(bp, skb, bnx2x_gro_ipv6_csum);
                        break;
                default:
                        BNX2X_ERR("Error: FW GRO supports only IPv4/IPv6, not 0x%04x\n",
                                  be16_to_cpu(skb->protocol));
                }
        }
#endif
        napi_gro_receive(&fp->napi, skb);
}

static void bnx2x_tpa_stop(struct bnx2x *bp, struct bnx2x_fastpath *fp,
                           struct bnx2x_agg_info *tpa_info,
                           u16 pages,
                           struct eth_end_agg_rx_cqe *cqe,
                           u16 cqe_idx)
{
        struct sw_rx_bd *rx_buf = &tpa_info->first_buf;
        u8 pad = tpa_info->placement_offset;
        u16 len = tpa_info->len_on_bd;
        struct sk_buff *skb = NULL;
        u8 *new_data, *data = rx_buf->data;
        u8 old_tpa_state = tpa_info->tpa_state;

        tpa_info->tpa_state = BNX2X_TPA_STOP;

        /* If we there was an error during the handling of the TPA_START -
         * drop this aggregation.
         */
        if (old_tpa_state == BNX2X_TPA_ERROR)
                goto drop;

        /* Try to allocate the new data */
        new_data = bnx2x_frag_alloc(fp);
        /* Unmap skb in the pool anyway, as we are going to change
           pool entry status to BNX2X_TPA_STOP even if new skb allocation
           fails. */
        dma_unmap_single(&bp->pdev->dev, dma_unmap_addr(rx_buf, mapping),
                         fp->rx_buf_size, DMA_FROM_DEVICE);
        if (likely(new_data))
                skb = build_skb(data, fp->rx_frag_size);

        if (likely(skb)) {
#ifdef BNX2X_STOP_ON_ERROR
                if (pad + len > fp->rx_buf_size) {
                        BNX2X_ERR("skb_put is about to fail...  pad %d  len %d  rx_buf_size %d\n",
                                  pad, len, fp->rx_buf_size);
                        bnx2x_panic();
                        return;
                }
#endif

                skb_reserve(skb, pad + NET_SKB_PAD);
                skb_put(skb, len);
                skb->rxhash = tpa_info->rxhash;
                skb->l4_rxhash = tpa_info->l4_rxhash;

                skb->protocol = eth_type_trans(skb, bp->dev);
                skb->ip_summed = CHECKSUM_UNNECESSARY;

                if (!bnx2x_fill_frag_skb(bp, fp, tpa_info, pages,
                                         skb, cqe, cqe_idx)) {
                        if (tpa_info->parsing_flags & PARSING_FLAGS_VLAN)
                                __vlan_hwaccel_put_tag(skb, htons(ETH_P_8021Q), tpa_info->vlan_tag);
                        bnx2x_gro_receive(bp, fp, skb);
                } else {
                        DP(NETIF_MSG_RX_STATUS,
                           "Failed to allocate new pages - dropping packet!\n");
                        dev_kfree_skb_any(skb);
                }

                /* put new data in bin */
                rx_buf->data = new_data;

                return;
        }
        bnx2x_frag_free(fp, new_data);
drop:
        /* drop the packet and keep the buffer in the bin */
        DP(NETIF_MSG_RX_STATUS,
           "Failed to allocate or map a new skb - dropping packet!\n");
        bnx2x_fp_stats(bp, fp)->eth_q_stats.rx_skb_alloc_failed++;
}

static int bnx2x_alloc_rx_data(struct bnx2x *bp,
                               struct bnx2x_fastpath *fp, u16 index)
{
        u8 *data;
        struct sw_rx_bd *rx_buf = &fp->rx_buf_ring[index];
        struct eth_rx_bd *rx_bd = &fp->rx_desc_ring[index];
        dma_addr_t mapping;

        data = bnx2x_frag_alloc(fp);
        if (unlikely(data == NULL))
                return -ENOMEM;

        mapping = dma_map_single(&bp->pdev->dev, data + NET_SKB_PAD,
                                 fp->rx_buf_size,
                                 DMA_FROM_DEVICE);
        if (unlikely(dma_mapping_error(&bp->pdev->dev, mapping))) {
                bnx2x_frag_free(fp, data);
                BNX2X_ERR("Can't map rx data\n");
                return -ENOMEM;
        }

        rx_buf->data = data;
        dma_unmap_addr_set(rx_buf, mapping, mapping);

        rx_bd->addr_hi = cpu_to_le32(U64_HI(mapping));
        rx_bd->addr_lo = cpu_to_le32(U64_LO(mapping));

        return 0;
}

static
void bnx2x_csum_validate(struct sk_buff *skb, union eth_rx_cqe *cqe,
                                 struct bnx2x_fastpath *fp,
                                 struct bnx2x_eth_q_stats *qstats)
{
        /* Do nothing if no L4 csum validation was done.
         * We do not check whether IP csum was validated. For IPv4 we assume
         * that if the card got as far as validating the L4 csum, it also
         * validated the IP csum. IPv6 has no IP csum.
         */
        if (cqe->fast_path_cqe.status_flags &
            ETH_FAST_PATH_RX_CQE_L4_XSUM_NO_VALIDATION_FLG)
                return;

        /* If L4 validation was done, check if an error was found. */

        if (cqe->fast_path_cqe.type_error_flags &
            (ETH_FAST_PATH_RX_CQE_IP_BAD_XSUM_FLG |
             ETH_FAST_PATH_RX_CQE_L4_BAD_XSUM_FLG))
                qstats->hw_csum_err++;
        else
                skb->ip_summed = CHECKSUM_UNNECESSARY;
}

int bnx2x_rx_int(struct bnx2x_fastpath *fp, int budget)
{
        struct bnx2x *bp = fp->bp;
        u16 bd_cons, bd_prod, bd_prod_fw, comp_ring_cons;
        u16 sw_comp_cons, sw_comp_prod;
        int rx_pkt = 0;
        union eth_rx_cqe *cqe;
        struct eth_fast_path_rx_cqe *cqe_fp;

#ifdef BNX2X_STOP_ON_ERROR
        if (unlikely(bp->panic))
                return 0;
#endif

        bd_cons = fp->rx_bd_cons;
        bd_prod = fp->rx_bd_prod;
        bd_prod_fw = bd_prod;
        sw_comp_cons = fp->rx_comp_cons;
        sw_comp_prod = fp->rx_comp_prod;

        comp_ring_cons = RCQ_BD(sw_comp_cons);
        cqe = &fp->rx_comp_ring[comp_ring_cons];
        cqe_fp = &cqe->fast_path_cqe;

        DP(NETIF_MSG_RX_STATUS,
           "queue[%d]: sw_comp_cons %u\n", fp->index, sw_comp_cons);

        while (BNX2X_IS_CQE_COMPLETED(cqe_fp)) {
                struct sw_rx_bd *rx_buf = NULL;
                struct sk_buff *skb;
                u8 cqe_fp_flags;
                enum eth_rx_cqe_type cqe_fp_type;
                u16 len, pad, queue;
                u8 *data;
                bool l4_rxhash;

#ifdef BNX2X_STOP_ON_ERROR
                if (unlikely(bp->panic))
                        return 0;
#endif

                bd_prod = RX_BD(bd_prod);
                bd_cons = RX_BD(bd_cons);

                cqe_fp_flags = cqe_fp->type_error_flags;
                cqe_fp_type = cqe_fp_flags & ETH_FAST_PATH_RX_CQE_TYPE;

                DP(NETIF_MSG_RX_STATUS,
                   "CQE type %x  err %x  status %x  queue %x  vlan %x  len %u\n",
                   CQE_TYPE(cqe_fp_flags),
                   cqe_fp_flags, cqe_fp->status_flags,
                   le32_to_cpu(cqe_fp->rss_hash_result),
                   le16_to_cpu(cqe_fp->vlan_tag),
                   le16_to_cpu(cqe_fp->pkt_len_or_gro_seg_len));

                /* is this a slowpath msg? */
                if (unlikely(CQE_TYPE_SLOW(cqe_fp_type))) {
                        bnx2x_sp_event(fp, cqe);
                        goto next_cqe;
                }

                rx_buf = &fp->rx_buf_ring[bd_cons];
                data = rx_buf->data;

                if (!CQE_TYPE_FAST(cqe_fp_type)) {
                        struct bnx2x_agg_info *tpa_info;
                        u16 frag_size, pages;
#ifdef BNX2X_STOP_ON_ERROR
                        /* sanity check */
                        if (fp->disable_tpa &&
                            (CQE_TYPE_START(cqe_fp_type) ||
                             CQE_TYPE_STOP(cqe_fp_type)))
                                BNX2X_ERR("START/STOP packet while disable_tpa type %x\n",
                                          CQE_TYPE(cqe_fp_type));
#endif

                        if (CQE_TYPE_START(cqe_fp_type)) {
                                u16 queue = cqe_fp->queue_index;
                                DP(NETIF_MSG_RX_STATUS,
                                   "calling tpa_start on queue %d\n",
                                   queue);

                                bnx2x_tpa_start(fp, queue,
                                                bd_cons, bd_prod,
                                                cqe_fp);

                                goto next_rx;
                        }
                        queue = cqe->end_agg_cqe.queue_index;
                        tpa_info = &fp->tpa_info[queue];
                        DP(NETIF_MSG_RX_STATUS,
                           "calling tpa_stop on queue %d\n",
                           queue);

                        frag_size = le16_to_cpu(cqe->end_agg_cqe.pkt_len) -
                                    tpa_info->len_on_bd;

                        if (fp->mode == TPA_MODE_GRO)
                                pages = (frag_size + tpa_info->full_page - 1) /
                                         tpa_info->full_page;
                        else
                                pages = SGE_PAGE_ALIGN(frag_size) >>
                                        SGE_PAGE_SHIFT;

                        bnx2x_tpa_stop(bp, fp, tpa_info, pages,
                                       &cqe->end_agg_cqe, comp_ring_cons);
#ifdef BNX2X_STOP_ON_ERROR
                        if (bp->panic)
                                return 0;
#endif

                        bnx2x_update_sge_prod(fp, pages, &cqe->end_agg_cqe);
                        goto next_cqe;
                }
                /* non TPA */
                len = le16_to_cpu(cqe_fp->pkt_len_or_gro_seg_len);
                pad = cqe_fp->placement_offset;
                dma_sync_single_for_cpu(&bp->pdev->dev,
                                        dma_unmap_addr(rx_buf, mapping),
                                        pad + RX_COPY_THRESH,
                                        DMA_FROM_DEVICE);
                pad += NET_SKB_PAD;
                prefetch(data + pad); /* speedup eth_type_trans() */
                /* is this an error packet? */
                if (unlikely(cqe_fp_flags & ETH_RX_ERROR_FALGS)) {
                        DP(NETIF_MSG_RX_ERR | NETIF_MSG_RX_STATUS,
                           "ERROR  flags %x  rx packet %u\n",
                           cqe_fp_flags, sw_comp_cons);
                        bnx2x_fp_qstats(bp, fp)->rx_err_discard_pkt++;
                        goto reuse_rx;
                }

                /* Since we don't have a jumbo ring
                 * copy small packets if mtu > 1500
                 */
                if ((bp->dev->mtu > ETH_MAX_PACKET_SIZE) &&
                    (len <= RX_COPY_THRESH)) {
                        skb = netdev_alloc_skb_ip_align(bp->dev, len);
                        if (skb == NULL) {
                                DP(NETIF_MSG_RX_ERR | NETIF_MSG_RX_STATUS,
                                   "ERROR  packet dropped because of alloc failure\n");
                                bnx2x_fp_qstats(bp, fp)->rx_skb_alloc_failed++;
                                goto reuse_rx;
                        }
                        memcpy(skb->data, data + pad, len);
                        bnx2x_reuse_rx_data(fp, bd_cons, bd_prod);
                } else {
                        if (likely(bnx2x_alloc_rx_data(bp, fp, bd_prod) == 0)) {
                                dma_unmap_single(&bp->pdev->dev,
                                                 dma_unmap_addr(rx_buf, mapping),
                                                 fp->rx_buf_size,
                                                 DMA_FROM_DEVICE);
                                skb = build_skb(data, fp->rx_frag_size);
                                if (unlikely(!skb)) {
                                        bnx2x_frag_free(fp, data);
                                        bnx2x_fp_qstats(bp, fp)->
                                                        rx_skb_alloc_failed++;
                                        goto next_rx;
                                }
                                skb_reserve(skb, pad);
                        } else {
                                DP(NETIF_MSG_RX_ERR | NETIF_MSG_RX_STATUS,
                                   "ERROR  packet dropped because of alloc failure\n");
                                bnx2x_fp_qstats(bp, fp)->rx_skb_alloc_failed++;
reuse_rx:
                                bnx2x_reuse_rx_data(fp, bd_cons, bd_prod);
                                goto next_rx;
                        }
                }

                skb_put(skb, len);
                skb->protocol = eth_type_trans(skb, bp->dev);

                /* Set Toeplitz hash for a none-LRO skb */
                skb->rxhash = bnx2x_get_rxhash(bp, cqe_fp, &l4_rxhash);
                skb->l4_rxhash = l4_rxhash;

                skb_checksum_none_assert(skb);

                if (bp->dev->features & NETIF_F_RXCSUM)
                        bnx2x_csum_validate(skb, cqe, fp,
                                            bnx2x_fp_qstats(bp, fp));

                skb_record_rx_queue(skb, fp->rx_queue);

                if (le16_to_cpu(cqe_fp->pars_flags.flags) &
                    PARSING_FLAGS_VLAN)
                        __vlan_hwaccel_put_tag(skb, htons(ETH_P_8021Q),
                                               le16_to_cpu(cqe_fp->vlan_tag));

                skb_mark_napi_id(skb, &fp->napi);

                if (bnx2x_fp_ll_polling(fp))
                        netif_receive_skb(skb);
                else
                        napi_gro_receive(&fp->napi, skb);
next_rx:
                rx_buf->data = NULL;

                bd_cons = NEXT_RX_IDX(bd_cons);
                bd_prod = NEXT_RX_IDX(bd_prod);
                bd_prod_fw = NEXT_RX_IDX(bd_prod_fw);
                rx_pkt++;
next_cqe:
                sw_comp_prod = NEXT_RCQ_IDX(sw_comp_prod);
                sw_comp_cons = NEXT_RCQ_IDX(sw_comp_cons);

                /* mark CQE as free */
                BNX2X_SEED_CQE(cqe_fp);

                if (rx_pkt == budget)
                        break;

                comp_ring_cons = RCQ_BD(sw_comp_cons);
                cqe = &fp->rx_comp_ring[comp_ring_cons];
                cqe_fp = &cqe->fast_path_cqe;
        } /* while */

        fp->rx_bd_cons = bd_cons;
        fp->rx_bd_prod = bd_prod_fw;
        fp->rx_comp_cons = sw_comp_cons;
        fp->rx_comp_prod = sw_comp_prod;

        /* Update producers */
        bnx2x_update_rx_prod(bp, fp, bd_prod_fw, sw_comp_prod,
                             fp->rx_sge_prod);

        fp->rx_pkt += rx_pkt;
        fp->rx_calls++;

        return rx_pkt;
}

static irqreturn_t bnx2x_msix_fp_int(int irq, void *fp_cookie)
{
        struct bnx2x_fastpath *fp = fp_cookie;
        struct bnx2x *bp = fp->bp;
        u8 cos;

        DP(NETIF_MSG_INTR,
           "got an MSI-X interrupt on IDX:SB [fp %d fw_sd %d igusb %d]\n",
           fp->index, fp->fw_sb_id, fp->igu_sb_id);

        bnx2x_ack_sb(bp, fp->igu_sb_id, USTORM_ID, 0, IGU_INT_DISABLE, 0);

#ifdef BNX2X_STOP_ON_ERROR
        if (unlikely(bp->panic))
                return IRQ_HANDLED;
#endif

        /* Handle Rx and Tx according to MSI-X vector */
        for_each_cos_in_tx_queue(fp, cos)
                prefetch(fp->txdata_ptr[cos]->tx_cons_sb);

        prefetch(&fp->sb_running_index[SM_RX_ID]);
        napi_schedule(&bnx2x_fp(bp, fp->index, napi));

        return IRQ_HANDLED;
}

/* HW Lock for shared dual port PHYs */
void bnx2x_acquire_phy_lock(struct bnx2x *bp)
{
        mutex_lock(&bp->port.phy_mutex);

        bnx2x_acquire_hw_lock(bp, HW_LOCK_RESOURCE_MDIO);
}

void bnx2x_release_phy_lock(struct bnx2x *bp)
{
        bnx2x_release_hw_lock(bp, HW_LOCK_RESOURCE_MDIO);

        mutex_unlock(&bp->port.phy_mutex);
}

/* calculates MF speed according to current linespeed and MF configuration */
u16 bnx2x_get_mf_speed(struct bnx2x *bp)
{
        u16 line_speed = bp->link_vars.line_speed;
        if (IS_MF(bp)) {
                u16 maxCfg = bnx2x_extract_max_cfg(bp,
                                                   bp->mf_config[BP_VN(bp)]);

                /* Calculate the current MAX line speed limit for the MF
                 * devices
                 */
                if (IS_MF_SI(bp))
                        line_speed = (line_speed * maxCfg) / 100;
                else { /* SD mode */
                        u16 vn_max_rate = maxCfg * 100;

                        if (vn_max_rate < line_speed)
                                line_speed = vn_max_rate;
                }
        }

        return line_speed;
}

/**
 * bnx2x_fill_report_data - fill link report data to report
 *
 * @bp:         driver handle
 * @data:       link state to update
 *
 * It uses a none-atomic bit operations because is called under the mutex.
 */
static void bnx2x_fill_report_data(struct bnx2x *bp,
                                   struct bnx2x_link_report_data *data)
{
        u16 line_speed = bnx2x_get_mf_speed(bp);

        memset(data, 0, sizeof(*data));

        /* Fill the report data: effective line speed */
        data->line_speed = line_speed;

        /* Link is down */
        if (!bp->link_vars.link_up || (bp->flags & MF_FUNC_DIS))
                __set_bit(BNX2X_LINK_REPORT_LINK_DOWN,
                          &data->link_report_flags);

        /* Full DUPLEX */
        if (bp->link_vars.duplex == DUPLEX_FULL)
                __set_bit(BNX2X_LINK_REPORT_FD, &data->link_report_flags);

        /* Rx Flow Control is ON */
        if (bp->link_vars.flow_ctrl & BNX2X_FLOW_CTRL_RX)
                __set_bit(BNX2X_LINK_REPORT_RX_FC_ON, &data->link_report_flags);

        /* Tx Flow Control is ON */
        if (bp->link_vars.flow_ctrl & BNX2X_FLOW_CTRL_TX)
                __set_bit(BNX2X_LINK_REPORT_TX_FC_ON, &data->link_report_flags);
}

/**
 * bnx2x_link_report - report link status to OS.
 *
 * @bp:         driver handle
 *
 * Calls the __bnx2x_link_report() under the same locking scheme
 * as a link/PHY state managing code to ensure a consistent link
 * reporting.
 */

void bnx2x_link_report(struct bnx2x *bp)
{
        bnx2x_acquire_phy_lock(bp);
        __bnx2x_link_report(bp);
        bnx2x_release_phy_lock(bp);
}

/**
 * __bnx2x_link_report - report link status to OS.
 *
 * @bp:         driver handle
 *
 * None atomic implementation.
 * Should be called under the phy_lock.
 */
void __bnx2x_link_report(struct bnx2x *bp)
{
        struct bnx2x_link_report_data cur_data;

        /* reread mf_cfg */
        if (IS_PF(bp) && !CHIP_IS_E1(bp))
                bnx2x_read_mf_cfg(bp);

        /* Read the current link report info */
        bnx2x_fill_report_data(bp, &cur_data);

        /* Don't report link down or exactly the same link status twice */
        if (!memcmp(&cur_data, &bp->last_reported_link, sizeof(cur_data)) ||
            (test_bit(BNX2X_LINK_REPORT_LINK_DOWN,
                      &bp->last_reported_link.link_report_flags) &&
             test_bit(BNX2X_LINK_REPORT_LINK_DOWN,
                      &cur_data.link_report_flags)))
                return;

        bp->link_cnt++;

        /* We are going to report a new link parameters now -
         * remember the current data for the next time.
         */
        memcpy(&bp->last_reported_link, &cur_data, sizeof(cur_data));

        if (test_bit(BNX2X_LINK_REPORT_LINK_DOWN,
                     &cur_data.link_report_flags)) {
                netif_carrier_off(bp->dev);
                netdev_err(bp->dev, "NIC Link is Down\n");
                return;
        } else {
                const char *duplex;
                const char *flow;

                netif_carrier_on(bp->dev);

                if (test_and_clear_bit(BNX2X_LINK_REPORT_FD,
                                       &cur_data.link_report_flags))
                        duplex = "full";
                else
                        duplex = "half";

                /* Handle the FC at the end so that only these flags would be
                 * possibly set. This way we may easily check if there is no FC
                 * enabled.
                 */
                if (cur_data.link_report_flags) {
                        if (test_bit(BNX2X_LINK_REPORT_RX_FC_ON,
                                     &cur_data.link_report_flags)) {
                                if (test_bit(BNX2X_LINK_REPORT_TX_FC_ON,
                                     &cur_data.link_report_flags))
                                        flow = "ON - receive & transmit";
                                else
                                        flow = "ON - receive";
                        } else {
                                flow = "ON - transmit";
                        }
                } else {
                        flow = "none";
                }
                netdev_info(bp->dev, "NIC Link is Up, %d Mbps %s duplex, Flow control: %s\n",
                            cur_data.line_speed, duplex, flow);
        }
}

static void bnx2x_set_next_page_sgl(struct bnx2x_fastpath *fp)
{
        int i;

        for (i = 1; i <= NUM_RX_SGE_PAGES; i++) {
                struct eth_rx_sge *sge;

                sge = &fp->rx_sge_ring[RX_SGE_CNT * i - 2];
                sge->addr_hi =
                        cpu_to_le32(U64_HI(fp->rx_sge_mapping +
                        BCM_PAGE_SIZE*(i % NUM_RX_SGE_PAGES)));

                sge->addr_lo =
                        cpu_to_le32(U64_LO(fp->rx_sge_mapping +
                        BCM_PAGE_SIZE*(i % NUM_RX_SGE_PAGES)));
        }
}

static void bnx2x_free_tpa_pool(struct bnx2x *bp,
                                struct bnx2x_fastpath *fp, int last)
{
        int i;

        for (i = 0; i < last; i++) {
                struct bnx2x_agg_info *tpa_info = &fp->tpa_info[i];
                struct sw_rx_bd *first_buf = &tpa_info->first_buf;
                u8 *data = first_buf->data;

                if (data == NULL) {
                        DP(NETIF_MSG_IFDOWN, "tpa bin %d empty on free\n", i);
                        continue;
                }
                if (tpa_info->tpa_state == BNX2X_TPA_START)
                        dma_unmap_single(&bp->pdev->dev,
                                         dma_unmap_addr(first_buf, mapping),
                                         fp->rx_buf_size, DMA_FROM_DEVICE);
                bnx2x_frag_free(fp, data);
                first_buf->data = NULL;
        }
}

void bnx2x_init_rx_rings_cnic(struct bnx2x *bp)
{
        int j;

        for_each_rx_queue_cnic(bp, j) {
                struct bnx2x_fastpath *fp = &bp->fp[j];

                fp->rx_bd_cons = 0;

                /* Activate BD ring */
                /* Warning!
                 * this will generate an interrupt (to the TSTORM)
                 * must only be done after chip is initialized
                 */
                bnx2x_update_rx_prod(bp, fp, fp->rx_bd_prod, fp->rx_comp_prod,
                                     fp->rx_sge_prod);
        }
}

void bnx2x_init_rx_rings(struct bnx2x *bp)
{
        int func = BP_FUNC(bp);
        u16 ring_prod;
        int i, j;

        /* Allocate TPA resources */
        for_each_eth_queue(bp, j) {
                struct bnx2x_fastpath *fp = &bp->fp[j];

                DP(NETIF_MSG_IFUP,
                   "mtu %d  rx_buf_size %d\n", bp->dev->mtu, fp->rx_buf_size);

                if (!fp->disable_tpa) {
                        /* Fill the per-aggregation pool */
                        for (i = 0; i < MAX_AGG_QS(bp); i++) {
                                struct bnx2x_agg_info *tpa_info =
                                        &fp->tpa_info[i];
                                struct sw_rx_bd *first_buf =
                                        &tpa_info->first_buf;

                                first_buf->data = bnx2x_frag_alloc(fp);
                                if (!first_buf->data) {
                                        BNX2X_ERR("Failed to allocate TPA skb pool for queue[%d] - disabling TPA on this queue!\n",
                                                  j);
                                        bnx2x_free_tpa_pool(bp, fp, i);
                                        fp->disable_tpa = 1;
                                        break;
                                }
                                dma_unmap_addr_set(first_buf, mapping, 0);
                                tpa_info->tpa_state = BNX2X_TPA_STOP;
                        }

                        /* "next page" elements initialization */
                        bnx2x_set_next_page_sgl(fp);

                        /* set SGEs bit mask */
                        bnx2x_init_sge_ring_bit_mask(fp);

                        /* Allocate SGEs and initialize the ring elements */
                        for (i = 0, ring_prod = 0;
                             i < MAX_RX_SGE_CNT*NUM_RX_SGE_PAGES; i++) {

                                if (bnx2x_alloc_rx_sge(bp, fp, ring_prod) < 0) {
                                        BNX2X_ERR("was only able to allocate %d rx sges\n",
                                                  i);
                                        BNX2X_ERR("disabling TPA for queue[%d]\n",
                                                  j);
                                        /* Cleanup already allocated elements */
                                        bnx2x_free_rx_sge_range(bp, fp,
                                                                ring_prod);
                                        bnx2x_free_tpa_pool(bp, fp,
                                                            MAX_AGG_QS(bp));
                                        fp->disable_tpa = 1;
                                        ring_prod = 0;
                                        break;
                                }
                                ring_prod = NEXT_SGE_IDX(ring_prod);
                        }

                        fp->rx_sge_prod = ring_prod;
                }
        }

        for_each_eth_queue(bp, j) {
                struct bnx2x_fastpath *fp = &bp->fp[j];

                fp->rx_bd_cons = 0;

                /* Activate BD ring */
                /* Warning!
                 * this will generate an interrupt (to the TSTORM)
                 * must only be done after chip is initialized
                 */
                bnx2x_update_rx_prod(bp, fp, fp->rx_bd_prod, fp->rx_comp_prod,
                                     fp->rx_sge_prod);

                if (j != 0)
                        continue;

                if (CHIP_IS_E1(bp)) {
                        REG_WR(bp, BAR_USTRORM_INTMEM +
                               USTORM_MEM_WORKAROUND_ADDRESS_OFFSET(func),
                               U64_LO(fp->rx_comp_mapping));
                        REG_WR(bp, BAR_USTRORM_INTMEM +
                               USTORM_MEM_WORKAROUND_ADDRESS_OFFSET(func) + 4,
                               U64_HI(fp->rx_comp_mapping));
                }
        }
}

static void bnx2x_free_tx_skbs_queue(struct bnx2x_fastpath *fp)
{
        u8 cos;
        struct bnx2x *bp = fp->bp;

        for_each_cos_in_tx_queue(fp, cos) {
                struct bnx2x_fp_txdata *txdata = fp->txdata_ptr[cos];
                unsigned pkts_compl = 0, bytes_compl = 0;

                u16 sw_prod = txdata->tx_pkt_prod;
                u16 sw_cons = txdata->tx_pkt_cons;

                while (sw_cons != sw_prod) {
                        bnx2x_free_tx_pkt(bp, txdata, TX_BD(sw_cons),
                                          &pkts_compl, &bytes_compl);
                        sw_cons++;
                }

                netdev_tx_reset_queue(
                        netdev_get_tx_queue(bp->dev,
                                            txdata->txq_index));
        }
}

static void bnx2x_free_tx_skbs_cnic(struct bnx2x *bp)
{
        int i;

        for_each_tx_queue_cnic(bp, i) {
                bnx2x_free_tx_skbs_queue(&bp->fp[i]);
        }
}

static void bnx2x_free_tx_skbs(struct bnx2x *bp)
{
        int i;

        for_each_eth_queue(bp, i) {
                bnx2x_free_tx_skbs_queue(&bp->fp[i]);
        }
}

static void bnx2x_free_rx_bds(struct bnx2x_fastpath *fp)
{
        struct bnx2x *bp = fp->bp;
        int i;

        /* ring wasn't allocated */
        if (fp->rx_buf_ring == NULL)
                return;

        for (i = 0; i < NUM_RX_BD; i++) {
                struct sw_rx_bd *rx_buf = &fp->rx_buf_ring[i];
                u8 *data = rx_buf->data;

                if (data == NULL)
                        continue;
                dma_unmap_single(&bp->pdev->dev,
                                 dma_unmap_addr(rx_buf, mapping),
                                 fp->rx_buf_size, DMA_FROM_DEVICE);

                rx_buf->data = NULL;
                bnx2x_frag_free(fp, data);
        }
}

static void bnx2x_free_rx_skbs_cnic(struct bnx2x *bp)
{
        int j;

        for_each_rx_queue_cnic(bp, j) {
                bnx2x_free_rx_bds(&bp->fp[j]);
        }
}

static void bnx2x_free_rx_skbs(struct bnx2x *bp)
{
        int j;

        for_each_eth_queue(bp, j) {
                struct bnx2x_fastpath *fp = &bp->fp[j];

                bnx2x_free_rx_bds(fp);

                if (!fp->disable_tpa)
                        bnx2x_free_tpa_pool(bp, fp, MAX_AGG_QS(bp));
        }
}

void bnx2x_free_skbs_cnic(struct bnx2x *bp)
{
        bnx2x_free_tx_skbs_cnic(bp);
        bnx2x_free_rx_skbs_cnic(bp);
}

void bnx2x_free_skbs(struct bnx2x *bp)
{
        bnx2x_free_tx_skbs(bp);
        bnx2x_free_rx_skbs(bp);
}

void bnx2x_update_max_mf_config(struct bnx2x *bp, u32 value)
{
        /* load old values */
        u32 mf_cfg = bp->mf_config[BP_VN(bp)];

        if (value != bnx2x_extract_max_cfg(bp, mf_cfg)) {
                /* leave all but MAX value */
                mf_cfg &= ~FUNC_MF_CFG_MAX_BW_MASK;

                /* set new MAX value */
                mf_cfg |= (value << FUNC_MF_CFG_MAX_BW_SHIFT)
                                & FUNC_MF_CFG_MAX_BW_MASK;

                bnx2x_fw_command(bp, DRV_MSG_CODE_SET_MF_BW, mf_cfg);
        }
}

/**
 * bnx2x_free_msix_irqs - free previously requested MSI-X IRQ vectors
 *
 * @bp:         driver handle
 * @nvecs:      number of vectors to be released
 */
static void bnx2x_free_msix_irqs(struct bnx2x *bp, int nvecs)
{
        int i, offset = 0;

        if (nvecs == offset)
                return;

        /* VFs don't have a default SB */
        if (IS_PF(bp)) {
                free_irq(bp->msix_table[offset].vector, bp->dev);
                DP(NETIF_MSG_IFDOWN, "released sp irq (%d)\n",
                   bp->msix_table[offset].vector);
                offset++;
        }

        if (CNIC_SUPPORT(bp)) {
                if (nvecs == offset)
                        return;
                offset++;
        }

        for_each_eth_queue(bp, i) {
                if (nvecs == offset)
                        return;
                DP(NETIF_MSG_IFDOWN, "about to release fp #%d->%d irq\n",
                   i, bp->msix_table[offset].vector);

                free_irq(bp->msix_table[offset++].vector, &bp->fp[i]);
        }
}

void bnx2x_free_irq(struct bnx2x *bp)
{
        if (bp->flags & USING_MSIX_FLAG &&
            !(bp->flags & USING_SINGLE_MSIX_FLAG)) {
                int nvecs = BNX2X_NUM_ETH_QUEUES(bp) + CNIC_SUPPORT(bp);

                /* vfs don't have a default status block */
                if (IS_PF(bp))
                        nvecs++;

                bnx2x_free_msix_irqs(bp, nvecs);
        } else {
                free_irq(bp->dev->irq, bp->dev);
        }
}

int bnx2x_enable_msix(struct bnx2x *bp)
{
        int msix_vec = 0, i, rc;

        /* VFs don't have a default status block */
        if (IS_PF(bp)) {
                bp->msix_table[msix_vec].entry = msix_vec;
                BNX2X_DEV_INFO("msix_table[0].entry = %d (slowpath)\n",
                               bp->msix_table[0].entry);
                msix_vec++;
        }

        /* Cnic requires an msix vector for itself */
        if (CNIC_SUPPORT(bp)) {
                bp->msix_table[msix_vec].entry = msix_vec;
                BNX2X_DEV_INFO("msix_table[%d].entry = %d (CNIC)\n",
                               msix_vec, bp->msix_table[msix_vec].entry);
                msix_vec++;
        }

        /* We need separate vectors for ETH queues only (not FCoE) */
        for_each_eth_queue(bp, i) {
                bp->msix_table[msix_vec].entry = msix_vec;
                BNX2X_DEV_INFO("msix_table[%d].entry = %d (fastpath #%u)\n",
                               msix_vec, msix_vec, i);
                msix_vec++;
        }

        DP(BNX2X_MSG_SP, "about to request enable msix with %d vectors\n",
           msix_vec);

        rc = pci_enable_msix(bp->pdev, &bp->msix_table[0], msix_vec);

        /*
         * reconfigure number of tx/rx queues according to available
         * MSI-X vectors
         */
        if (rc >= BNX2X_MIN_MSIX_VEC_CNT(bp)) {
                /* how less vectors we will have? */
                int diff = msix_vec - rc;

                BNX2X_DEV_INFO("Trying to use less MSI-X vectors: %d\n", rc);

                rc = pci_enable_msix(bp->pdev, &bp->msix_table[0], rc);

                if (rc) {
                        BNX2X_DEV_INFO("MSI-X is not attainable rc %d\n", rc);
                        goto no_msix;
                }
                /*
                 * decrease number of queues by number of unallocated entries
                 */
                bp->num_ethernet_queues -= diff;
                bp->num_queues = bp->num_ethernet_queues + bp->num_cnic_queues;

                BNX2X_DEV_INFO("New queue configuration set: %d\n",
                               bp->num_queues);
        } else if (rc > 0) {
                /* Get by with single vector */
                rc = pci_enable_msix(bp->pdev, &bp->msix_table[0], 1);
                if (rc) {
                        BNX2X_DEV_INFO("Single MSI-X is not attainable rc %d\n",
                                       rc);
                        goto no_msix;
                }

                BNX2X_DEV_INFO("Using single MSI-X vector\n");
                bp->flags |= USING_SINGLE_MSIX_FLAG;

                BNX2X_DEV_INFO("set number of queues to 1\n");
                bp->num_ethernet_queues = 1;
                bp->num_queues = bp->num_ethernet_queues + bp->num_cnic_queues;
        } else if (rc < 0) {
                BNX2X_DEV_INFO("MSI-X is not attainable  rc %d\n", rc);
                goto no_msix;
        }

        bp->flags |= USING_MSIX_FLAG;

        return 0;

no_msix:
        /* fall to INTx if not enough memory */
        if (rc == -ENOMEM)
                bp->flags |= DISABLE_MSI_FLAG;

        return rc;
}

static int bnx2x_req_msix_irqs(struct bnx2x *bp)
{
        int i, rc, offset = 0;

        /* no default status block for vf */
        if (IS_PF(bp)) {
                rc = request_irq(bp->msix_table[offset++].vector,
                                 bnx2x_msix_sp_int, 0,
                                 bp->dev->name, bp->dev);
                if (rc) {
                        BNX2X_ERR("request sp irq failed\n");
                        return -EBUSY;
                }
        }

        if (CNIC_SUPPORT(bp))
                offset++;

        for_each_eth_queue(bp, i) {
                struct bnx2x_fastpath *fp = &bp->fp[i];
                snprintf(fp->name, sizeof(fp->name), "%s-fp-%d",
                         bp->dev->name, i);

                rc = request_irq(bp->msix_table[offset].vector,
                                 bnx2x_msix_fp_int, 0, fp->name, fp);
                if (rc) {
                        BNX2X_ERR("request fp #%d irq (%d) failed  rc %d\n", i,
                              bp->msix_table[offset].vector, rc);
                        bnx2x_free_msix_irqs(bp, offset);
                        return -EBUSY;
                }

                offset++;
        }

        i = BNX2X_NUM_ETH_QUEUES(bp);
        if (IS_PF(bp)) {
                offset = 1 + CNIC_SUPPORT(bp);
                netdev_info(bp->dev,
                            "using MSI-X  IRQs: sp %d  fp[%d] %d ... fp[%d] %d\n",
                            bp->msix_table[0].vector,
                            0, bp->msix_table[offset].vector,
                            i - 1, bp->msix_table[offset + i - 1].vector);
        } else {
                offset = CNIC_SUPPORT(bp);
                netdev_info(bp->dev,
                            "using MSI-X  IRQs: fp[%d] %d ... fp[%d] %d\n",
                            0, bp->msix_table[offset].vector,
                            i - 1, bp->msix_table[offset + i - 1].vector);
        }
        return 0;
}

int bnx2x_enable_msi(struct bnx2x *bp)
{
        int rc;

        rc = pci_enable_msi(bp->pdev);
        if (rc) {
                BNX2X_DEV_INFO("MSI is not attainable\n");
                return -1;
        }
        bp->flags |= USING_MSI_FLAG;

        return 0;
}

static int bnx2x_req_irq(struct bnx2x *bp)
{
        unsigned long flags;
        unsigned int irq;

        if (bp->flags & (USING_MSI_FLAG | USING_MSIX_FLAG))
                flags = 0;
        else
                flags = IRQF_SHARED;

        if (bp->flags & USING_MSIX_FLAG)
                irq = bp->msix_table[0].vector;
        else
                irq = bp->pdev->irq;

        return request_irq(irq, bnx2x_interrupt, flags, bp->dev->name, bp->dev);
}

static int bnx2x_setup_irqs(struct bnx2x *bp)
{
        int rc = 0;
        if (bp->flags & USING_MSIX_FLAG &&
            !(bp->flags & USING_SINGLE_MSIX_FLAG)) {
                rc = bnx2x_req_msix_irqs(bp);
                if (rc)
                        return rc;
        } else {
                rc = bnx2x_req_irq(bp);
                if (rc) {
                        BNX2X_ERR("IRQ request failed  rc %d, aborting\n", rc);
                        return rc;
                }
                if (bp->flags & USING_MSI_FLAG) {
                        bp->dev->irq = bp->pdev->irq;
                        netdev_info(bp->dev, "using MSI IRQ %d\n",
                                    bp->dev->irq);
                }
                if (bp->flags & USING_MSIX_FLAG) {
                        bp->dev->irq = bp->msix_table[0].vector;
                        netdev_info(bp->dev, "using MSIX IRQ %d\n",
                                    bp->dev->irq);
                }
        }

        return 0;
}

static void bnx2x_napi_enable_cnic(struct bnx2x *bp)
{
        int i;

        for_each_rx_queue_cnic(bp, i) {
                bnx2x_fp_init_lock(&bp->fp[i]);
                napi_enable(&bnx2x_fp(bp, i, napi));
        }
}

static void bnx2x_napi_enable(struct bnx2x *bp)
{
        int i;

        for_each_eth_queue(bp, i) {
                bnx2x_fp_init_lock(&bp->fp[i]);
                napi_enable(&bnx2x_fp(bp, i, napi));
        }
}

static void bnx2x_napi_disable_cnic(struct bnx2x *bp)
{
        int i;

        local_bh_disable();
        for_each_rx_queue_cnic(bp, i) {
                napi_disable(&bnx2x_fp(bp, i, napi));
                while (!bnx2x_fp_lock_napi(&bp->fp[i]))
                        mdelay(1);
        }
        local_bh_enable();
}

static void bnx2x_napi_disable(struct bnx2x *bp)
{
        int i;

        local_bh_disable();
        for_each_eth_queue(bp, i) {
                napi_disable(&bnx2x_fp(bp, i, napi));
                while (!bnx2x_fp_lock_napi(&bp->fp[i]))
                        mdelay(1);
        }
        local_bh_enable();
}

void bnx2x_netif_start(struct bnx2x *bp)
{
        if (netif_running(bp->dev)) {
                bnx2x_napi_enable(bp);
                if (CNIC_LOADED(bp))
                        bnx2x_napi_enable_cnic(bp);
                bnx2x_int_enable(bp);
                if (bp->state == BNX2X_STATE_OPEN)
                        netif_tx_wake_all_queues(bp->dev);
        }
}

void bnx2x_netif_stop(struct bnx2x *bp, int disable_hw)
{
        bnx2x_int_disable_sync(bp, disable_hw);
        bnx2x_napi_disable(bp);
        if (CNIC_LOADED(bp))
                bnx2x_napi_disable_cnic(bp);
}

u16 bnx2x_select_queue(struct net_device *dev, struct sk_buff *skb)
{
        struct bnx2x *bp = netdev_priv(dev);

        if (CNIC_LOADED(bp) && !NO_FCOE(bp)) {
                struct ethhdr *hdr = (struct ethhdr *)skb->data;
                u16 ether_type = ntohs(hdr->h_proto);

                /* Skip VLAN tag if present */
                if (ether_type == ETH_P_8021Q) {
                        struct vlan_ethhdr *vhdr =
                                (struct vlan_ethhdr *)skb->data;

                        ether_type = ntohs(vhdr->h_vlan_encapsulated_proto);
                }

                /* If ethertype is FCoE or FIP - use FCoE ring */
                if ((ether_type == ETH_P_FCOE) || (ether_type == ETH_P_FIP))
                        return bnx2x_fcoe_tx(bp, txq_index);
        }

        /* select a non-FCoE queue */
        return __netdev_pick_tx(dev, skb) % BNX2X_NUM_ETH_QUEUES(bp);
}

void bnx2x_set_num_queues(struct bnx2x *bp)
{
        /* RSS queues */
        bp->num_ethernet_queues = bnx2x_calc_num_queues(bp);

        /* override in STORAGE SD modes */
        if (IS_MF_STORAGE_SD(bp) || IS_MF_FCOE_AFEX(bp))
                bp->num_ethernet_queues = 1;

        /* Add special queues */
        bp->num_cnic_queues = CNIC_SUPPORT(bp); /* For FCOE */
        bp->num_queues = bp->num_ethernet_queues + bp->num_cnic_queues;

        BNX2X_DEV_INFO("set number of queues to %d\n", bp->num_queues);
}

/**
 * bnx2x_set_real_num_queues - configure netdev->real_num_[tx,rx]_queues
 *
 * @bp:         Driver handle
 *
 * We currently support for at most 16 Tx queues for each CoS thus we will
 * allocate a multiple of 16 for ETH L2 rings according to the value of the
 * bp->max_cos.
 *
 * If there is an FCoE L2 queue the appropriate Tx queue will have the next
 * index after all ETH L2 indices.
 *
 * If the actual number of Tx queues (for each CoS) is less than 16 then there
 * will be the holes at the end of each group of 16 ETh L2 indices (0..15,
 * 16..31,...) with indices that are not coupled with any real Tx queue.
 *
 * The proper configuration of skb->queue_mapping is handled by
 * bnx2x_select_queue() and __skb_tx_hash().
 *
 * bnx2x_setup_tc() takes care of the proper TC mappings so that __skb_tx_hash()
 * will return a proper Tx index if TC is enabled (netdev->num_tc > 0).
 */
static int bnx2x_set_real_num_queues(struct bnx2x *bp, int include_cnic)
{
        int rc, tx, rx;

        tx = BNX2X_NUM_ETH_QUEUES(bp) * bp->max_cos;
        rx = BNX2X_NUM_ETH_QUEUES(bp);

/* account for fcoe queue */
        if (include_cnic && !NO_FCOE(bp)) {
                rx++;
                tx++;
        }

        rc = netif_set_real_num_tx_queues(bp->dev, tx);
        if (rc) {
                BNX2X_ERR("Failed to set real number of Tx queues: %d\n", rc);
                return rc;
        }
        rc = netif_set_real_num_rx_queues(bp->dev, rx);
        if (rc) {
                BNX2X_ERR("Failed to set real number of Rx queues: %d\n", rc);
                return rc;
        }

        DP(NETIF_MSG_IFUP, "Setting real num queues to (tx, rx) (%d, %d)\n",
                          tx, rx);

        return rc;
}

static void bnx2x_set_rx_buf_size(struct bnx2x *bp)
{
        int i;

        for_each_queue(bp, i) {
                struct bnx2x_fastpath *fp = &bp->fp[i];
                u32 mtu;

                /* Always use a mini-jumbo MTU for the FCoE L2 ring */
                if (IS_FCOE_IDX(i))
                        /*
                         * Although there are no IP frames expected to arrive to
                         * this ring we still want to add an
                         * IP_HEADER_ALIGNMENT_PADDING to prevent a buffer
                         * overrun attack.
                         */
                        mtu = BNX2X_FCOE_MINI_JUMBO_MTU;
                else
                        mtu = bp->dev->mtu;
                fp->rx_buf_size = BNX2X_FW_RX_ALIGN_START +
                                  IP_HEADER_ALIGNMENT_PADDING +
                                  ETH_OVREHEAD +
                                  mtu +
                                  BNX2X_FW_RX_ALIGN_END;
                /* Note : rx_buf_size doesn't take into account NET_SKB_PAD */
                if (fp->rx_buf_size + NET_SKB_PAD <= PAGE_SIZE)
                        fp->rx_frag_size = fp->rx_buf_size + NET_SKB_PAD;
                else
                        fp->rx_frag_size = 0;
        }
}

static int bnx2x_init_rss_pf(struct bnx2x *bp)
{
        int i;
        u8 num_eth_queues = BNX2X_NUM_ETH_QUEUES(bp);

        /* Prepare the initial contents for the indirection table if RSS is
         * enabled
         */
        for (i = 0; i < sizeof(bp->rss_conf_obj.ind_table); i++)
                bp->rss_conf_obj.ind_table[i] =
                        bp->fp->cl_id +
                        ethtool_rxfh_indir_default(i, num_eth_queues);

        /*
         * For 57710 and 57711 SEARCHER configuration (rss_keys) is
         * per-port, so if explicit configuration is needed , do it only
         * for a PMF.
         *
         * For 57712 and newer on the other hand it's a per-function
         * configuration.
         */
        return bnx2x_config_rss_eth(bp, bp->port.pmf || !CHIP_IS_E1x(bp));
}

int bnx2x_config_rss_pf(struct bnx2x *bp, struct bnx2x_rss_config_obj *rss_obj,
                        bool config_hash)
{
        struct bnx2x_config_rss_params params = {NULL};

        /* Although RSS is meaningless when there is a single HW queue we
         * still need it enabled in order to have HW Rx hash generated.
         *
         * if (!is_eth_multi(bp))
         *      bp->multi_mode = ETH_RSS_MODE_DISABLED;
         */

        params.rss_obj = rss_obj;

        __set_bit(RAMROD_COMP_WAIT, &params.ramrod_flags);

        __set_bit(BNX2X_RSS_MODE_REGULAR, &params.rss_flags);

        /* RSS configuration */
        __set_bit(BNX2X_RSS_IPV4, &params.rss_flags);
        __set_bit(BNX2X_RSS_IPV4_TCP, &params.rss_flags);
        __set_bit(BNX2X_RSS_IPV6, &params.rss_flags);
        __set_bit(BNX2X_RSS_IPV6_TCP, &params.rss_flags);
        if (rss_obj->udp_rss_v4)
                __set_bit(BNX2X_RSS_IPV4_UDP, &params.rss_flags);
        if (rss_obj->udp_rss_v6)

                __set_bit(BNX2X_RSS_IPV6_UDP, &params.rss_flags);

        /* Hash bits */
        params.rss_result_mask = MULTI_MASK;

        memcpy(params.ind_table, rss_obj->ind_table, sizeof(params.ind_table));

        if (config_hash) {
                /* RSS keys */
                prandom_bytes(params.rss_key, sizeof(params.rss_key));
                __set_bit(BNX2X_RSS_SET_SRCH, &params.rss_flags);
        }

        return bnx2x_config_rss(bp, &params);
}

static int bnx2x_init_hw(struct bnx2x *bp, u32 load_code)
{
        struct bnx2x_func_state_params func_params = {NULL};

        /* Prepare parameters for function state transitions */
        __set_bit(RAMROD_COMP_WAIT, &func_params.ramrod_flags);

        func_params.f_obj = &bp->func_obj;
        func_params.cmd = BNX2X_F_CMD_HW_INIT;

        func_params.params.hw_init.load_phase = load_code;

        return bnx2x_func_state_change(bp, &func_params);
}

/*
 * Cleans the object that have internal lists without sending
 * ramrods. Should be run when interrupts are disabled.
 */
void bnx2x_squeeze_objects(struct bnx2x *bp)
{
        int rc;
        unsigned long ramrod_flags = 0, vlan_mac_flags = 0;
        struct bnx2x_mcast_ramrod_params rparam = {NULL};
        struct bnx2x_vlan_mac_obj *mac_obj = &bp->sp_objs->mac_obj;

        /***************** Cleanup MACs' object first *************************/

        /* Wait for completion of requested */
        __set_bit(RAMROD_COMP_WAIT, &ramrod_flags);
        /* Perform a dry cleanup */
        __set_bit(RAMROD_DRV_CLR_ONLY, &ramrod_flags);

        /* Clean ETH primary MAC */
        __set_bit(BNX2X_ETH_MAC, &vlan_mac_flags);
        rc = mac_obj->delete_all(bp, &bp->sp_objs->mac_obj, &vlan_mac_flags,
                                 &ramrod_flags);
        if (rc != 0)
                BNX2X_ERR("Failed to clean ETH MACs: %d\n", rc);

        /* Cleanup UC list */
        vlan_mac_flags = 0;
        __set_bit(BNX2X_UC_LIST_MAC, &vlan_mac_flags);
        rc = mac_obj->delete_all(bp, mac_obj, &vlan_mac_flags,
                                 &ramrod_flags);
        if (rc != 0)
                BNX2X_ERR("Failed to clean UC list MACs: %d\n", rc);

        /***************** Now clean mcast object *****************************/
        rparam.mcast_obj = &bp->mcast_obj;
        __set_bit(RAMROD_DRV_CLR_ONLY, &rparam.ramrod_flags);

        /* Add a DEL command... */
        rc = bnx2x_config_mcast(bp, &rparam, BNX2X_MCAST_CMD_DEL);
        if (rc < 0)
                BNX2X_ERR("Failed to add a new DEL command to a multi-cast object: %d\n",
                          rc);

        /* ...and wait until all pending commands are cleared */
        rc = bnx2x_config_mcast(bp, &rparam, BNX2X_MCAST_CMD_CONT);
        while (rc != 0) {
                if (rc < 0) {
                        BNX2X_ERR("Failed to clean multi-cast object: %d\n",
                                  rc);
                        return;
                }

                rc = bnx2x_config_mcast(bp, &rparam, BNX2X_MCAST_CMD_CONT);
        }
}

#ifndef BNX2X_STOP_ON_ERROR
#define LOAD_ERROR_EXIT(bp, label) \
        do { \
                (bp)->state = BNX2X_STATE_ERROR; \
                goto label; \
        } while (0)

#define LOAD_ERROR_EXIT_CNIC(bp, label) \
        do { \
                bp->cnic_loaded = false; \
                goto label; \
        } while (0)
#else /*BNX2X_STOP_ON_ERROR*/
#define LOAD_ERROR_EXIT(bp, label) \
        do { \
                (bp)->state = BNX2X_STATE_ERROR; \
                (bp)->panic = 1; \
                return -EBUSY; \
        } while (0)
#define LOAD_ERROR_EXIT_CNIC(bp, label) \
        do { \
                bp->cnic_loaded = false; \
                (bp)->panic = 1; \
                return -EBUSY; \
        } while (0)
#endif /*BNX2X_STOP_ON_ERROR*/

static void bnx2x_free_fw_stats_mem(struct bnx2x *bp)
{
        BNX2X_PCI_FREE(bp->fw_stats, bp->fw_stats_mapping,
                       bp->fw_stats_data_sz + bp->fw_stats_req_sz);
        return;
}

static int bnx2x_alloc_fw_stats_mem(struct bnx2x *bp)
{
        int num_groups, vf_headroom = 0;
        int is_fcoe_stats = NO_FCOE(bp) ? 0 : 1;

        /* number of queues for statistics is number of eth queues + FCoE */
        u8 num_queue_stats = BNX2X_NUM_ETH_QUEUES(bp) + is_fcoe_stats;

        /* Total number of FW statistics requests =
         * 1 for port stats + 1 for PF stats + potential 2 for FCoE (fcoe proper
         * and fcoe l2 queue) stats + num of queues (which includes another 1
         * for fcoe l2 queue if applicable)
         */
        bp->fw_stats_num = 2 + is_fcoe_stats + num_queue_stats;

        /* vf stats appear in the request list, but their data is allocated by
         * the VFs themselves. We don't include them in the bp->fw_stats_num as
         * it is used to determine where to place the vf stats queries in the
         * request struct
         */
        if (IS_SRIOV(bp))
                vf_headroom = bnx2x_vf_headroom(bp);

        /* Request is built from stats_query_header and an array of
         * stats_query_cmd_group each of which contains
         * STATS_QUERY_CMD_COUNT rules. The real number or requests is
         * configured in the stats_query_header.
         */
        num_groups =
                (((bp->fw_stats_num + vf_headroom) / STATS_QUERY_CMD_COUNT) +
                 (((bp->fw_stats_num + vf_headroom) % STATS_QUERY_CMD_COUNT) ?
                 1 : 0));

        DP(BNX2X_MSG_SP, "stats fw_stats_num %d, vf headroom %d, num_groups %d\n",
           bp->fw_stats_num, vf_headroom, num_groups);
        bp->fw_stats_req_sz = sizeof(struct stats_query_header) +
                num_groups * sizeof(struct stats_query_cmd_group);

        /* Data for statistics requests + stats_counter
         * stats_counter holds per-STORM counters that are incremented
         * when STORM has finished with the current request.
         * memory for FCoE offloaded statistics are counted anyway,
         * even if they will not be sent.
         * VF stats are not accounted for here as the data of VF stats is stored
         * in memory allocated by the VF, not here.
         */
        bp->fw_stats_data_sz = sizeof(struct per_port_stats) +
                sizeof(struct per_pf_stats) +
                sizeof(struct fcoe_statistics_params) +
                sizeof(struct per_queue_stats) * num_queue_stats +
                sizeof(struct stats_counter);

        BNX2X_PCI_ALLOC(bp->fw_stats, &bp->fw_stats_mapping,
                        bp->fw_stats_data_sz + bp->fw_stats_req_sz);

        /* Set shortcuts */
        bp->fw_stats_req = (struct bnx2x_fw_stats_req *)bp->fw_stats;
        bp->fw_stats_req_mapping = bp->fw_stats_mapping;
        bp->fw_stats_data = (struct bnx2x_fw_stats_data *)
                ((u8 *)bp->fw_stats + bp->fw_stats_req_sz);
        bp->fw_stats_data_mapping = bp->fw_stats_mapping +
                bp->fw_stats_req_sz;

        DP(BNX2X_MSG_SP, "statistics request base address set to %x %x\n",
           U64_HI(bp->fw_stats_req_mapping),
           U64_LO(bp->fw_stats_req_mapping));
        DP(BNX2X_MSG_SP, "statistics data base address set to %x %x\n",
           U64_HI(bp->fw_stats_data_mapping),
           U64_LO(bp->fw_stats_data_mapping));
        return 0;

alloc_mem_err:
        bnx2x_free_fw_stats_mem(bp);
        BNX2X_ERR("Can't allocate FW stats memory\n");
        return -ENOMEM;
}

/* send load request to mcp and analyze response */
static int bnx2x_nic_load_request(struct bnx2x *bp, u32 *load_code)
{
        u32 param;

        /* init fw_seq */
        bp->fw_seq =
                (SHMEM_RD(bp, func_mb[BP_FW_MB_IDX(bp)].drv_mb_header) &
                 DRV_MSG_SEQ_NUMBER_MASK);
        BNX2X_DEV_INFO("fw_seq 0x%08x\n", bp->fw_seq);

        /* Get current FW pulse sequence */
        bp->fw_drv_pulse_wr_seq =
                (SHMEM_RD(bp, func_mb[BP_FW_MB_IDX(bp)].drv_pulse_mb) &
                 DRV_PULSE_SEQ_MASK);
        BNX2X_DEV_INFO("drv_pulse 0x%x\n", bp->fw_drv_pulse_wr_seq);

        param = DRV_MSG_CODE_LOAD_REQ_WITH_LFA;

        if (IS_MF_SD(bp) && bnx2x_port_after_undi(bp))
                param |= DRV_MSG_CODE_LOAD_REQ_FORCE_LFA;

        /* load request */
        (*load_code) = bnx2x_fw_command(bp, DRV_MSG_CODE_LOAD_REQ, param);

        /* if mcp fails to respond we must abort */
        if (!(*load_code)) {
                BNX2X_ERR("MCP response failure, aborting\n");
                return -EBUSY;
        }

        /* If mcp refused (e.g. other port is in diagnostic mode) we
         * must abort
         */
        if ((*load_code) == FW_MSG_CODE_DRV_LOAD_REFUSED) {
                BNX2X_ERR("MCP refused load request, aborting\n");
                return -EBUSY;
        }
        return 0;
}

/* check whether another PF has already loaded FW to chip. In
 * virtualized environments a pf from another VM may have already
 * initialized the device including loading FW
 */
int bnx2x_nic_load_analyze_req(struct bnx2x *bp, u32 load_code)
{
        /* is another pf loaded on this engine? */
        if (load_code != FW_MSG_CODE_DRV_LOAD_COMMON_CHIP &&
            load_code != FW_MSG_CODE_DRV_LOAD_COMMON) {
                /* build my FW version dword */
                u32 my_fw = (BCM_5710_FW_MAJOR_VERSION) +
                        (BCM_5710_FW_MINOR_VERSION << 8) +
                        (BCM_5710_FW_REVISION_VERSION << 16) +
                        (BCM_5710_FW_ENGINEERING_VERSION << 24);

                /* read loaded FW from chip */
                u32 loaded_fw = REG_RD(bp, XSEM_REG_PRAM);

                DP(BNX2X_MSG_SP, "loaded fw %x, my fw %x\n",
                   loaded_fw, my_fw);

                /* abort nic load if version mismatch */
                if (my_fw != loaded_fw) {
                        BNX2X_ERR("bnx2x with FW %x was already loaded which mismatches my %x FW. Aborting\n",
                                  loaded_fw, my_fw);
                        return -EBUSY;
                }
        }
        return 0;
}

/* returns the "mcp load_code" according to global load_count array */
static int bnx2x_nic_load_no_mcp(struct bnx2x *bp, int port)
{
        int path = BP_PATH(bp);

        DP(NETIF_MSG_IFUP, "NO MCP - load counts[%d]      %d, %d, %d\n",
           path, load_count[path][0], load_count[path][1],
           load_count[path][2]);
        load_count[path][0]++;
        load_count[path][1 + port]++;
        DP(NETIF_MSG_IFUP, "NO MCP - new load counts[%d]  %d, %d, %d\n",
           path, load_count[path][0], load_count[path][1],
           load_count[path][2]);
        if (load_count[path][0] == 1)
                return FW_MSG_CODE_DRV_LOAD_COMMON;
        else if (load_count[path][1 + port] == 1)
                return FW_MSG_CODE_DRV_LOAD_PORT;
        else
                return FW_MSG_CODE_DRV_LOAD_FUNCTION;
}

/* mark PMF if applicable */
static void bnx2x_nic_load_pmf(struct bnx2x *bp, u32 load_code)
{
        if ((load_code == FW_MSG_CODE_DRV_LOAD_COMMON) ||
            (load_code == FW_MSG_CODE_DRV_LOAD_COMMON_CHIP) ||
            (load_code == FW_MSG_CODE_DRV_LOAD_PORT)) {
                bp->port.pmf = 1;
                /* We need the barrier to ensure the ordering between the
                 * writing to bp->port.pmf here and reading it from the
                 * bnx2x_periodic_task().
                 */
                smp_mb();
        } else {
                bp->port.pmf = 0;
        }

        DP(NETIF_MSG_LINK, "pmf %d\n", bp->port.pmf);
}

static void bnx2x_nic_load_afex_dcc(struct bnx2x *bp, int load_code)
{
        if (((load_code == FW_MSG_CODE_DRV_LOAD_COMMON) ||
             (load_code == FW_MSG_CODE_DRV_LOAD_COMMON_CHIP)) &&
            (bp->common.shmem2_base)) {
                if (SHMEM2_HAS(bp, dcc_support))
                        SHMEM2_WR(bp, dcc_support,
                                  (SHMEM_DCC_SUPPORT_DISABLE_ENABLE_PF_TLV |
                                   SHMEM_DCC_SUPPORT_BANDWIDTH_ALLOCATION_TLV));
                if (SHMEM2_HAS(bp, afex_driver_support))
                        SHMEM2_WR(bp, afex_driver_support,
                                  SHMEM_AFEX_SUPPORTED_VERSION_ONE);
        }

        /* Set AFEX default VLAN tag to an invalid value */
        bp->afex_def_vlan_tag = -1;
}

/**
 * bnx2x_bz_fp - zero content of the fastpath structure.
 *
 * @bp:         driver handle
 * @index:      fastpath index to be zeroed
 *
 * Makes sure the contents of the bp->fp[index].napi is kept
 * intact.
 */
static void bnx2x_bz_fp(struct bnx2x *bp, int index)
{
        struct bnx2x_fastpath *fp = &bp->fp[index];
        int cos;
        struct napi_struct orig_napi = fp->napi;
        struct bnx2x_agg_info *orig_tpa_info = fp->tpa_info;

        /* bzero bnx2x_fastpath contents */
        if (fp->tpa_info)
                memset(fp->tpa_info, 0, ETH_MAX_AGGREGATION_QUEUES_E1H_E2 *
                       sizeof(struct bnx2x_agg_info));
        memset(fp, 0, sizeof(*fp));

        /* Restore the NAPI object as it has been already initialized */
        fp->napi = orig_napi;
        fp->tpa_info = orig_tpa_info;
        fp->bp = bp;
        fp->index = index;
        if (IS_ETH_FP(fp))
                fp->max_cos = bp->max_cos;
        else
                /* Special queues support only one CoS */
                fp->max_cos = 1;

        /* Init txdata pointers */
        if (IS_FCOE_FP(fp))
                fp->txdata_ptr[0] = &bp->bnx2x_txq[FCOE_TXQ_IDX(bp)];
        if (IS_ETH_FP(fp))
                for_each_cos_in_tx_queue(fp, cos)
                        fp->txdata_ptr[cos] = &bp->bnx2x_txq[cos *
                                BNX2X_NUM_ETH_QUEUES(bp) + index];

        /* set the tpa flag for each queue. The tpa flag determines the queue
         * minimal size so it must be set prior to queue memory allocation
         */
        fp->disable_tpa = !(bp->flags & TPA_ENABLE_FLAG ||
                                  (bp->flags & GRO_ENABLE_FLAG &&
                                   bnx2x_mtu_allows_gro(bp->dev->mtu)));
        if (bp->flags & TPA_ENABLE_FLAG)
                fp->mode = TPA_MODE_LRO;
        else if (bp->flags & GRO_ENABLE_FLAG)
                fp->mode = TPA_MODE_GRO;

        /* We don't want TPA on an FCoE L2 ring */
        if (IS_FCOE_FP(fp))
                fp->disable_tpa = 1;
}

int bnx2x_load_cnic(struct bnx2x *bp)
{
        int i, rc, port = BP_PORT(bp);

        DP(NETIF_MSG_IFUP, "Starting CNIC-related load\n");

        mutex_init(&bp->cnic_mutex);

        if (IS_PF(bp)) {
                rc = bnx2x_alloc_mem_cnic(bp);
                if (rc) {
                        BNX2X_ERR("Unable to allocate bp memory for cnic\n");
                        LOAD_ERROR_EXIT_CNIC(bp, load_error_cnic0);
                }
        }

        rc = bnx2x_alloc_fp_mem_cnic(bp);
        if (rc) {
                BNX2X_ERR("Unable to allocate memory for cnic fps\n");
                LOAD_ERROR_EXIT_CNIC(bp, load_error_cnic0);
        }

        /* Update the number of queues with the cnic queues */
        rc = bnx2x_set_real_num_queues(bp, 1);
        if (rc) {
                BNX2X_ERR("Unable to set real_num_queues including cnic\n");
                LOAD_ERROR_EXIT_CNIC(bp, load_error_cnic0);
        }

        /* Add all CNIC NAPI objects */
        bnx2x_add_all_napi_cnic(bp);
        DP(NETIF_MSG_IFUP, "cnic napi added\n");
        bnx2x_napi_enable_cnic(bp);

        rc = bnx2x_init_hw_func_cnic(bp);
        if (rc)
                LOAD_ERROR_EXIT_CNIC(bp, load_error_cnic1);

        bnx2x_nic_init_cnic(bp);

        if (IS_PF(bp)) {
                /* Enable Timer scan */
                REG_WR(bp, TM_REG_EN_LINEAR0_TIMER + port*4, 1);

                /* setup cnic queues */
                for_each_cnic_queue(bp, i) {
                        rc = bnx2x_setup_queue(bp, &bp->fp[i], 0);
                        if (rc) {
                                BNX2X_ERR("Queue setup failed\n");
                                LOAD_ERROR_EXIT(bp, load_error_cnic2);
                        }
                }
        }

        /* Initialize Rx filter. */
        netif_addr_lock_bh(bp->dev);
        bnx2x_set_rx_mode(bp->dev);
        netif_addr_unlock_bh(bp->dev);

        /* re-read iscsi info */
        bnx2x_get_iscsi_info(bp);
        bnx2x_setup_cnic_irq_info(bp);
        bnx2x_setup_cnic_info(bp);
        bp->cnic_loaded = true;
        if (bp->state == BNX2X_STATE_OPEN)
                bnx2x_cnic_notify(bp, CNIC_CTL_START_CMD);

        DP(NETIF_MSG_IFUP, "Ending successfully CNIC-related load\n");

        return 0;

#ifndef BNX2X_STOP_ON_ERROR
load_error_cnic2:
        /* Disable Timer scan */
        REG_WR(bp, TM_REG_EN_LINEAR0_TIMER + port*4, 0);

load_error_cnic1:
        bnx2x_napi_disable_cnic(bp);
        /* Update the number of queues without the cnic queues */
        rc = bnx2x_set_real_num_queues(bp, 0);
        if (rc)
                BNX2X_ERR("Unable to set real_num_queues not including cnic\n");
load_error_cnic0:
        BNX2X_ERR("CNIC-related load failed\n");
        bnx2x_free_fp_mem_cnic(bp);
        bnx2x_free_mem_cnic(bp);
        return rc;
#endif /* ! BNX2X_STOP_ON_ERROR */
}

/* must be called with rtnl_lock */
int bnx2x_nic_load(struct bnx2x *bp, int load_mode)
{
        int port = BP_PORT(bp);
        int i, rc = 0, load_code = 0;

        DP(NETIF_MSG_IFUP, "Starting NIC load\n");
        DP(NETIF_MSG_IFUP,
           "CNIC is %s\n", CNIC_ENABLED(bp) ? "enabled" : "disabled");

#ifdef BNX2X_STOP_ON_ERROR
        if (unlikely(bp->panic)) {
                BNX2X_ERR("Can't load NIC when there is panic\n");
                return -EPERM;
        }
#endif

        bp->state = BNX2X_STATE_OPENING_WAIT4_LOAD;

        /* zero the structure w/o any lock, before SP handler is initialized */
        memset(&bp->last_reported_link, 0, sizeof(bp->last_reported_link));
        __set_bit(BNX2X_LINK_REPORT_LINK_DOWN,
                &bp->last_reported_link.link_report_flags);

        if (IS_PF(bp))
                /* must be called before memory allocation and HW init */
                bnx2x_ilt_set_info(bp);

        /*
         * Zero fastpath structures preserving invariants like napi, which are
         * allocated only once, fp index, max_cos, bp pointer.
         * Also set fp->disable_tpa and txdata_ptr.
         */
        DP(NETIF_MSG_IFUP, "num queues: %d", bp->num_queues);
        for_each_queue(bp, i)
                bnx2x_bz_fp(bp, i);
        memset(bp->bnx2x_txq, 0, (BNX2X_MAX_RSS_COUNT(bp) * BNX2X_MULTI_TX_COS +
                                  bp->num_cnic_queues) *
                                  sizeof(struct bnx2x_fp_txdata));

        bp->fcoe_init = false;

        /* Set the receive queues buffer size */
        bnx2x_set_rx_buf_size(bp);

        if (IS_PF(bp)) {
                rc = bnx2x_alloc_mem(bp);
                if (rc) {
                        BNX2X_ERR("Unable to allocate bp memory\n");
                        return rc;
                }
        }

        /* Allocated memory for FW statistics  */
        if (bnx2x_alloc_fw_stats_mem(bp))
                LOAD_ERROR_EXIT(bp, load_error0);

        /* need to be done after alloc mem, since it's self adjusting to amount
         * of memory available for RSS queues
         */
        rc = bnx2x_alloc_fp_mem(bp);
        if (rc) {
                BNX2X_ERR("Unable to allocate memory for fps\n");
                LOAD_ERROR_EXIT(bp, load_error0);
        }

        /* request pf to initialize status blocks */
        if (IS_VF(bp)) {
                rc = bnx2x_vfpf_init(bp);
                if (rc)
                        LOAD_ERROR_EXIT(bp, load_error0);
        }

        /* As long as bnx2x_alloc_mem() may possibly update
         * bp->num_queues, bnx2x_set_real_num_queues() should always
         * come after it. At this stage cnic queues are not counted.
         */
        rc = bnx2x_set_real_num_queues(bp, 0);
        if (rc) {
                BNX2X_ERR("Unable to set real_num_queues\n");
                LOAD_ERROR_EXIT(bp, load_error0);
        }

        /* configure multi cos mappings in kernel.
         * this configuration may be overridden by a multi class queue
         * discipline or by a dcbx negotiation result.
         */
        bnx2x_setup_tc(bp->dev, bp->max_cos);

        /* Add all NAPI objects */
        bnx2x_add_all_napi(bp);
        DP(NETIF_MSG_IFUP, "napi added\n");
        bnx2x_napi_enable(bp);

        if (IS_PF(bp)) {
                /* set pf load just before approaching the MCP */
                bnx2x_set_pf_load(bp);

                /* if mcp exists send load request and analyze response */
                if (!BP_NOMCP(bp)) {
                        /* attempt to load pf */
                        rc = bnx2x_nic_load_request(bp, &load_code);
                        if (rc)
                                LOAD_ERROR_EXIT(bp, load_error1);

                        /* what did mcp say? */
                        rc = bnx2x_nic_load_analyze_req(bp, load_code);
                        if (rc) {
                                bnx2x_fw_command(bp, DRV_MSG_CODE_LOAD_DONE, 0);
                                LOAD_ERROR_EXIT(bp, load_error2);
                        }
                } else {
                        load_code = bnx2x_nic_load_no_mcp(bp, port);
                }

                /* mark pmf if applicable */
                bnx2x_nic_load_pmf(bp, load_code);

                /* Init Function state controlling object */
                bnx2x__init_func_obj(bp);

                /* Initialize HW */
                rc = bnx2x_init_hw(bp, load_code);
                if (rc) {
                        BNX2X_ERR("HW init failed, aborting\n");
                        bnx2x_fw_command(bp, DRV_MSG_CODE_LOAD_DONE, 0);
                        LOAD_ERROR_EXIT(bp, load_error2);
                }
        }

        bnx2x_pre_irq_nic_init(bp);

        /* Connect to IRQs */
        rc = bnx2x_setup_irqs(bp);
        if (rc) {
                BNX2X_ERR("setup irqs failed\n");
                if (IS_PF(bp))
                        bnx2x_fw_command(bp, DRV_MSG_CODE_LOAD_DONE, 0);
                LOAD_ERROR_EXIT(bp, load_error2);
        }

        /* Init per-function objects */
        if (IS_PF(bp)) {
                /* Setup NIC internals and enable interrupts */
                bnx2x_post_irq_nic_init(bp, load_code);

                bnx2x_init_bp_objs(bp);
                bnx2x_iov_nic_init(bp);

                /* Set AFEX default VLAN tag to an invalid value */
                bp->afex_def_vlan_tag = -1;
                bnx2x_nic_load_afex_dcc(bp, load_code);
                bp->state = BNX2X_STATE_OPENING_WAIT4_PORT;
                rc = bnx2x_func_start(bp);
                if (rc) {
                        BNX2X_ERR("Function start failed!\n");
                        bnx2x_fw_command(bp, DRV_MSG_CODE_LOAD_DONE, 0);

                        LOAD_ERROR_EXIT(bp, load_error3);
                }

                /* Send LOAD_DONE command to MCP */
                if (!BP_NOMCP(bp)) {
                        load_code = bnx2x_fw_command(bp,
                                                     DRV_MSG_CODE_LOAD_DONE, 0);
                        if (!load_code) {
                                BNX2X_ERR("MCP response failure, aborting\n");
                                rc = -EBUSY;
                                LOAD_ERROR_EXIT(bp, load_error3);
                        }
                }

                /* initialize FW coalescing state machines in RAM */
                bnx2x_update_coalesce(bp);

                /* setup the leading queue */
                rc = bnx2x_setup_leading(bp);
                if (rc) {
                        BNX2X_ERR("Setup leading failed!\n");
                        LOAD_ERROR_EXIT(bp, load_error3);
                }

                /* set up the rest of the queues */
                for_each_nondefault_eth_queue(bp, i) {
                        rc = bnx2x_setup_queue(bp, &bp->fp[i], 0);
                        if (rc) {
                                BNX2X_ERR("Queue setup failed\n");
                                LOAD_ERROR_EXIT(bp, load_error3);
                        }
                }

                /* setup rss */
                rc = bnx2x_init_rss_pf(bp);
                if (rc) {
                        BNX2X_ERR("PF RSS init failed\n");
                        LOAD_ERROR_EXIT(bp, load_error3);
                }

        } else { /* vf */
                for_each_eth_queue(bp, i) {
                        rc = bnx2x_vfpf_setup_q(bp, i);
                        if (rc) {
                                BNX2X_ERR("Queue setup failed\n");
                                LOAD_ERROR_EXIT(bp, load_error3);
                        }
                }
        }

        /* Now when Clients are configured we are ready to work */
        bp->state = BNX2X_STATE_OPEN;

        /* Configure a ucast MAC */
        if (IS_PF(bp))
                rc = bnx2x_set_eth_mac(bp, true);
        else /* vf */
                rc = bnx2x_vfpf_config_mac(bp, bp->dev->dev_addr, bp->fp->index,
                                           true);
        if (rc) {
                BNX2X_ERR("Setting Ethernet MAC failed\n");
                LOAD_ERROR_EXIT(bp, load_error3);
        }

        if (IS_PF(bp) && bp->pending_max) {
                bnx2x_update_max_mf_config(bp, bp->pending_max);
                bp->pending_max = 0;
        }

        if (bp->port.pmf) {
                rc = bnx2x_initial_phy_init(bp, load_mode);
                if (rc)
                        LOAD_ERROR_EXIT(bp, load_error3);
        }
        bp->link_params.feature_config_flags &= ~FEATURE_CONFIG_BOOT_FROM_SAN;

        /* Start fast path */

        /* Initialize Rx filter. */
        netif_addr_lock_bh(bp->dev);
        bnx2x_set_rx_mode(bp->dev);
        netif_addr_unlock_bh(bp->dev);

        /* Start the Tx */
        switch (load_mode) {
        case LOAD_NORMAL:
                /* Tx queue should be only re-enabled */
                netif_tx_wake_all_queues(bp->dev);
                break;

        case LOAD_OPEN:
                netif_tx_start_all_queues(bp->dev);
                smp_mb__after_clear_bit();
                break;

        case LOAD_DIAG:
        case LOAD_LOOPBACK_EXT:
                bp->state = BNX2X_STATE_DIAG;
                break;

        default:
                break;
        }

        if (bp->port.pmf)
                bnx2x_update_drv_flags(bp, 1 << DRV_FLAGS_PORT_MASK, 0);
        else
                bnx2x__link_status_update(bp);

        /* start the timer */
        mod_timer(&bp->timer, jiffies + bp->current_interval);

        if (CNIC_ENABLED(bp))
                bnx2x_load_cnic(bp);

        if (IS_PF(bp) && SHMEM2_HAS(bp, drv_capabilities_flag)) {
                /* mark driver is loaded in shmem2 */
                u32 val;
                val = SHMEM2_RD(bp, drv_capabilities_flag[BP_FW_MB_IDX(bp)]);
                SHMEM2_WR(bp, drv_capabilities_flag[BP_FW_MB_IDX(bp)],
                          val | DRV_FLAGS_CAPABILITIES_LOADED_SUPPORTED |
                          DRV_FLAGS_CAPABILITIES_LOADED_L2);
        }

        /* Wait for all pending SP commands to complete */
        if (IS_PF(bp) && !bnx2x_wait_sp_comp(bp, ~0x0UL)) {
                BNX2X_ERR("Timeout waiting for SP elements to complete\n");
                bnx2x_nic_unload(bp, UNLOAD_CLOSE, false);
                return -EBUSY;
        }

        /* If PMF - send ADMIN DCBX msg to MFW to initiate DCBX FSM */
        if (bp->port.pmf && (bp->state != BNX2X_STATE_DIAG))
                bnx2x_dcbx_init(bp, false);

        DP(NETIF_MSG_IFUP, "Ending successfully NIC load\n");

        return 0;

#ifndef BNX2X_STOP_ON_ERROR
load_error3:
        if (IS_PF(bp)) {
                bnx2x_int_disable_sync(bp, 1);

                /* Clean queueable objects */
                bnx2x_squeeze_objects(bp);
        }

        /* Free SKBs, SGEs, TPA pool and driver internals */
        bnx2x_free_skbs(bp);
        for_each_rx_queue(bp, i)
                bnx2x_free_rx_sge_range(bp, bp->fp + i, NUM_RX_SGE);

        /* Release IRQs */
        bnx2x_free_irq(bp);
load_error2:
        if (IS_PF(bp) && !BP_NOMCP(bp)) {
                bnx2x_fw_command(bp, DRV_MSG_CODE_UNLOAD_REQ_WOL_MCP, 0);
                bnx2x_fw_command(bp, DRV_MSG_CODE_UNLOAD_DONE, 0);
        }

        bp->port.pmf = 0;
load_error1:
        bnx2x_napi_disable(bp);
        bnx2x_del_all_napi(bp);

        /* clear pf_load status, as it was already set */
        if (IS_PF(bp))
                bnx2x_clear_pf_load(bp);
load_error0:
        bnx2x_free_fp_mem(bp);
        bnx2x_free_fw_stats_mem(bp);
        bnx2x_free_mem(bp);

        return rc;
#endif /* ! BNX2X_STOP_ON_ERROR */
}

int bnx2x_drain_tx_queues(struct bnx2x *bp)
{
        u8 rc = 0, cos, i;

        /* Wait until tx fastpath tasks complete */
        for_each_tx_queue(bp, i) {
                struct bnx2x_fastpath *fp = &bp->fp[i];

                for_each_cos_in_tx_queue(fp, cos)
                        rc = bnx2x_clean_tx_queue(bp, fp->txdata_ptr[cos]);
                if (rc)
                        return rc;
        }
        return 0;
}

/* must be called with rtnl_lock */
int bnx2x_nic_unload(struct bnx2x *bp, int unload_mode, bool keep_link)
{
        int i;
        bool global = false;

        DP(NETIF_MSG_IFUP, "Starting NIC unload\n");

        /* mark driver is unloaded in shmem2 */
        if (IS_PF(bp) && SHMEM2_HAS(bp, drv_capabilities_flag)) {
                u32 val;
                val = SHMEM2_RD(bp, drv_capabilities_flag[BP_FW_MB_IDX(bp)]);
                SHMEM2_WR(bp, drv_capabilities_flag[BP_FW_MB_IDX(bp)],
                          val & ~DRV_FLAGS_CAPABILITIES_LOADED_L2);
        }

        if (IS_PF(bp) && bp->recovery_state != BNX2X_RECOVERY_DONE &&
            (bp->state == BNX2X_STATE_CLOSED ||
             bp->state == BNX2X_STATE_ERROR)) {
                /* We can get here if the driver has been unloaded
                 * during parity error recovery and is either waiting for a
                 * leader to complete or for other functions to unload and
                 * then ifdown has been issued. In this case we want to
                 * unload and let other functions to complete a recovery
                 * process.
                 */
                bp->recovery_state = BNX2X_RECOVERY_DONE;
                bp->is_leader = 0;
                bnx2x_release_leader_lock(bp);
                smp_mb();

                DP(NETIF_MSG_IFDOWN, "Releasing a leadership...\n");
                BNX2X_ERR("Can't unload in closed or error state\n");
                return -EINVAL;
        }

        /* Nothing to do during unload if previous bnx2x_nic_load()
         * have not completed successfully - all resources are released.
         *
         * we can get here only after unsuccessful ndo_* callback, during which
         * dev->IFF_UP flag is still on.
         */
        if (bp->state == BNX2X_STATE_CLOSED || bp->state == BNX2X_STATE_ERROR)
                return 0;

        /* It's important to set the bp->state to the value different from
         * BNX2X_STATE_OPEN and only then stop the Tx. Otherwise bnx2x_tx_int()
         * may restart the Tx from the NAPI context (see bnx2x_tx_int()).
         */
        bp->state = BNX2X_STATE_CLOSING_WAIT4_HALT;
        smp_mb();

        /* indicate to VFs that the PF is going down */
        bnx2x_iov_channel_down(bp);

        if (CNIC_LOADED(bp))
                bnx2x_cnic_notify(bp, CNIC_CTL_STOP_CMD);

        /* Stop Tx */
        bnx2x_tx_disable(bp);
        netdev_reset_tc(bp->dev);

        bp->rx_mode = BNX2X_RX_MODE_NONE;

        del_timer_sync(&bp->timer);

        if (IS_PF(bp)) {
                /* Set ALWAYS_ALIVE bit in shmem */
                bp->fw_drv_pulse_wr_seq |= DRV_PULSE_ALWAYS_ALIVE;
                bnx2x_drv_pulse(bp);
                bnx2x_stats_handle(bp, STATS_EVENT_STOP);
                bnx2x_save_statistics(bp);
        }

        /* wait till consumers catch up with producers in all queues */
        bnx2x_drain_tx_queues(bp);

        /* if VF indicate to PF this function is going down (PF will delete sp
         * elements and clear initializations
         */
        if (IS_VF(bp))
                bnx2x_vfpf_close_vf(bp);
        else if (unload_mode != UNLOAD_RECOVERY)
                /* if this is a normal/close unload need to clean up chip*/
                bnx2x_chip_cleanup(bp, unload_mode, keep_link);
        else {
                /* Send the UNLOAD_REQUEST to the MCP */
                bnx2x_send_unload_req(bp, unload_mode);

                /* Prevent transactions to host from the functions on the
                 * engine that doesn't reset global blocks in case of global
                 * attention once global blocks are reset and gates are opened
                 * (the engine which leader will perform the recovery
                 * last).
                 */
                if (!CHIP_IS_E1x(bp))
                        bnx2x_pf_disable(bp);

                /* Disable HW interrupts, NAPI */
                bnx2x_netif_stop(bp, 1);
                /* Delete all NAPI objects */
                bnx2x_del_all_napi(bp);
                if (CNIC_LOADED(bp))
                        bnx2x_del_all_napi_cnic(bp);
                /* Release IRQs */
                bnx2x_free_irq(bp);

                /* Report UNLOAD_DONE to MCP */
                bnx2x_send_unload_done(bp, false);
        }

        /*
         * At this stage no more interrupts will arrive so we may safely clean
         * the queueable objects here in case they failed to get cleaned so far.
         */
        if (IS_PF(bp))
                bnx2x_squeeze_objects(bp);

        /* There should be no more pending SP commands at this stage */
        bp->sp_state = 0;

        bp->port.pmf = 0;

        /* Free SKBs, SGEs, TPA pool and driver internals */
        bnx2x_free_skbs(bp);
        if (CNIC_LOADED(bp))
                bnx2x_free_skbs_cnic(bp);
        for_each_rx_queue(bp, i)
                bnx2x_free_rx_sge_range(bp, bp->fp + i, NUM_RX_SGE);

        bnx2x_free_fp_mem(bp);
        if (CNIC_LOADED(bp))
                bnx2x_free_fp_mem_cnic(bp);

        if (IS_PF(bp)) {
                if (CNIC_LOADED(bp))
                        bnx2x_free_mem_cnic(bp);
        }
        bnx2x_free_mem(bp);

        bp->state = BNX2X_STATE_CLOSED;
        bp->cnic_loaded = false;

        /* Check if there are pending parity attentions. If there are - set
         * RECOVERY_IN_PROGRESS.
         */
        if (IS_PF(bp) && bnx2x_chk_parity_attn(bp, &global, false)) {
                bnx2x_set_reset_in_progress(bp);

                /* Set RESET_IS_GLOBAL if needed */
                if (global)
                        bnx2x_set_reset_global(bp);
        }

        /* The last driver must disable a "close the gate" if there is no
         * parity attention or "process kill" pending.
         */
        if (IS_PF(bp) &&
            !bnx2x_clear_pf_load(bp) &&
            bnx2x_reset_is_done(bp, BP_PATH(bp)))
                bnx2x_disable_close_the_gate(bp);

        DP(NETIF_MSG_IFUP, "Ending NIC unload\n");

        return 0;
}

int bnx2x_set_power_state(struct bnx2x *bp, pci_power_t state)
{
        u16 pmcsr;

        /* If there is no power capability, silently succeed */
        if (!bp->pm_cap) {