diff options
Diffstat (limited to 'drivers/net/ethernet/broadcom/bnx2x/bnx2x_cmn.c')
| -rw-r--r-- | drivers/net/ethernet/broadcom/bnx2x/bnx2x_cmn.c | 2542 |
1 files changed, 1772 insertions, 770 deletions
diff --git a/drivers/net/ethernet/broadcom/bnx2x/bnx2x_cmn.c b/drivers/net/ethernet/broadcom/bnx2x/bnx2x_cmn.c index 8098eea9704..c43e7238de2 100644 --- a/drivers/net/ethernet/broadcom/bnx2x/bnx2x_cmn.c +++ b/drivers/net/ethernet/broadcom/bnx2x/bnx2x_cmn.c @@ -1,12 +1,12 @@ /* bnx2x_cmn.c: Broadcom Everest network driver. * - * Copyright (c) 2007-2012 Broadcom Corporation + * 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> + * Maintained by: Ariel Elior <ariel.elior@qlogic.com> * Written by: Eliezer Tamir * Based on code from Michael Chan's bnx2 driver * UDP CSUM errata workaround by Arik Gendelman @@ -21,14 +21,55 @@ #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" +static void bnx2x_free_fp_mem_cnic(struct bnx2x *bp); +static int bnx2x_alloc_fp_mem_cnic(struct bnx2x *bp); +static int bnx2x_alloc_fp_mem(struct bnx2x *bp); +static int bnx2x_poll(struct napi_struct *napi, int budget); +static void bnx2x_add_all_napi_cnic(struct bnx2x *bp) +{ + int i; + + /* Add NAPI objects */ + for_each_rx_queue_cnic(bp, i) { + netif_napi_add(bp->dev, &bnx2x_fp(bp, i, napi), + bnx2x_poll, NAPI_POLL_WEIGHT); + napi_hash_add(&bnx2x_fp(bp, i, napi)); + } +} + +static void bnx2x_add_all_napi(struct bnx2x *bp) +{ + int i; + + /* Add NAPI objects */ + for_each_eth_queue(bp, i) { + netif_napi_add(bp->dev, &bnx2x_fp(bp, i, napi), + bnx2x_poll, NAPI_POLL_WEIGHT); + napi_hash_add(&bnx2x_fp(bp, i, napi)); + } +} + +static int bnx2x_calc_num_queues(struct bnx2x *bp) +{ + int nq = bnx2x_num_queues ? : netif_get_num_default_rss_queues(); + + /* Reduce memory usage in kdump environment by using only one queue */ + if (reset_devices) + nq = 1; + + nq = clamp(nq, 1, BNX2X_MAX_QUEUES(bp)); + return nq; +} /** * bnx2x_move_fp - move content of the fastpath structure. @@ -40,12 +81,20 @@ * 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 + * 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; @@ -53,9 +102,91 @@ static inline void bnx2x_move_fp(struct bnx2x *bp, int from, int to) /* 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 */ +int bnx2x_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 @@ -70,6 +201,7 @@ static u16 bnx2x_free_tx_pkt(struct bnx2x *bp, struct bnx2x_fp_txdata *txdata, struct sk_buff *skb = tx_buf->skb; u16 bd_idx = TX_BD(tx_buf->first_bd), new_cons; int nbd; + u16 split_bd_len = 0; /* prefetch skb end pointer to speedup dev_kfree_skb() */ prefetch(&skb->end); @@ -77,11 +209,7 @@ static u16 bnx2x_free_tx_pkt(struct bnx2x *bp, struct bnx2x_fp_txdata *txdata, 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 @@ -99,12 +227,25 @@ static u16 bnx2x_free_tx_pkt(struct bnx2x *bp, struct bnx2x_fp_txdata *txdata, --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_HAS_SECOND_PBD) { + /* Skip second parse bd... */ + --nbd; + bd_idx = TX_BD(NEXT_TX_IDX(bd_idx)); + } + + /* TSO headers+data bds share a common mapping. See bnx2x_tx_split() */ if (tx_buf->flags & BNX2X_TSO_SPLIT_BD) { + tx_data_bd = &txdata->tx_desc_ring[bd_idx].reg_bd; + split_bd_len = BD_UNMAP_LEN(tx_data_bd); --nbd; bd_idx = TX_BD(NEXT_TX_IDX(bd_idx)); } + /* unmap first bd */ + dma_unmap_single(&bp->pdev->dev, BD_UNMAP_ADDR(tx_start_bd), + BD_UNMAP_LEN(tx_start_bd) + split_bd_len, + DMA_TO_DEVICE); + /* now free frags */ while (nbd > 0) { @@ -154,7 +295,7 @@ int bnx2x_tx_int(struct bnx2x *bp, struct bnx2x_fp_txdata *txdata) txdata->txq_index, hw_cons, sw_cons, pkt_cons); bd_cons = bnx2x_free_tx_pkt(bp, txdata, pkt_cons, - &pkts_compl, &bytes_compl); + &pkts_compl, &bytes_compl); sw_cons++; } @@ -176,7 +317,7 @@ int bnx2x_tx_int(struct bnx2x *bp, struct bnx2x_fp_txdata *txdata) smp_mb(); if (unlikely(netif_tx_queue_stopped(txq))) { - /* Taking tx_lock() is needed to prevent reenabling the queue + /* 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()): @@ -190,7 +331,7 @@ int bnx2x_tx_int(struct bnx2x *bp, struct bnx2x_fp_txdata *txdata) if ((netif_tx_queue_stopped(txq)) && (bp->state == BNX2X_STATE_OPEN) && - (bnx2x_tx_avail(bp, txdata) >= MAX_SKB_FRAGS + 4)) + (bnx2x_tx_avail(bp, txdata) >= MAX_DESC_PER_TX_PKT)) netif_tx_wake_queue(txq); __netif_tx_unlock(txq); @@ -260,16 +401,26 @@ static inline void bnx2x_update_sge_prod(struct bnx2x_fastpath *fp, fp->last_max_sge, fp->rx_sge_prod); } -/* Set Toeplitz hash value in the skb using the value from the +/* 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) + const struct eth_fast_path_rx_cqe *cqe, + enum pkt_hash_types *rxhash_type) { - /* Set Toeplitz hash from CQE */ + /* Get Toeplitz hash from CQE */ if ((bp->dev->features & NETIF_F_RXHASH) && - (cqe->status_flags & ETH_FAST_PATH_RX_CQE_RSS_HASH_FLG)) + (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; + *rxhash_type = ((htype == TCP_IPV4_HASH_TYPE) || + (htype == TCP_IPV6_HASH_TYPE)) ? + PKT_HASH_TYPE_L4 : PKT_HASH_TYPE_L3; + return le32_to_cpu(cqe->rss_hash_result); + } + *rxhash_type = PKT_HASH_TYPE_NONE; return 0; } @@ -323,11 +474,10 @@ static void bnx2x_tpa_start(struct bnx2x_fastpath *fp, u16 queue, 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->rxhash = bnx2x_get_rxhash(bp, cqe, &tpa_info->rxhash_type); if (fp->mode == TPA_MODE_GRO) { u16 gro_size = le16_to_cpu(cqe->pkt_len_or_gro_seg_len); - tpa_info->full_page = - SGE_PAGE_SIZE * PAGES_PER_SGE / gro_size * gro_size; + tpa_info->full_page = SGE_PAGES / gro_size * gro_size; tpa_info->gro_size = gro_size; } @@ -348,31 +498,35 @@ static void bnx2x_tpa_start(struct bnx2x_fastpath *fp, u16 queue, */ #define TPA_TSTAMP_OPT_LEN 12 /** - * bnx2x_set_lro_mss - calculate the approximate value of the MSS + * bnx2x_set_gro_params - compute GRO values * - * @bp: driver handle + * @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 u16 bnx2x_set_lro_mss(struct bnx2x *bp, u16 parsing_flags, - u16 len_on_bd) +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 arrgregation won't have either IP options or TCP options + /* 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) + PRS_FLAG_OVERETH_IPV6) { hdrs_len += sizeof(struct ipv6hdr); - else /* IPv4 */ + 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. @@ -382,13 +536,18 @@ static u16 bnx2x_set_lro_mss(struct bnx2x *bp, u16 parsing_flags, if (parsing_flags & PARSING_FLAGS_TIME_STAMP_EXIST_FLAG) hdrs_len += TPA_TSTAMP_OPT_LEN; - return len_on_bd - hdrs_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) +static int bnx2x_alloc_rx_sge(struct bnx2x *bp, struct bnx2x_fastpath *fp, + u16 index, gfp_t gfp_mask) { - struct page *page = alloc_pages(GFP_ATOMIC, PAGES_PER_SGE_SHIFT); + struct page *page = alloc_pages(gfp_mask, 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; @@ -399,7 +558,7 @@ static int bnx2x_alloc_rx_sge(struct bnx2x *bp, } mapping = dma_map_page(&bp->pdev->dev, page, 0, - SGE_PAGE_SIZE*PAGES_PER_SGE, DMA_FROM_DEVICE); + 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"); @@ -436,22 +595,13 @@ static int bnx2x_fill_frag_skb(struct bnx2x *bp, struct bnx2x_fastpath *fp, } /* This is needed in order to enable forwarding support */ - if (frag_size) { - skb_shinfo(skb)->gso_size = bnx2x_set_lro_mss(bp, - tpa_info->parsing_flags, len_on_bd); - - /* set for GRO */ - if (fp->mode == TPA_MODE_GRO) - skb_shinfo(skb)->gso_type = - (GET_FLAG(tpa_info->parsing_flags, - PARSING_FLAGS_OVER_ETHERNET_PROTOCOL) == - PRS_FLAG_OVERETH_IPV6) ? - SKB_GSO_TCPV6 : SKB_GSO_TCPV4; - } - + 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_PAGE_SIZE*PAGES_PER_SGE) { + 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); @@ -469,24 +619,23 @@ static int bnx2x_fill_frag_skb(struct bnx2x *bp, struct bnx2x_fastpath *fp, 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_PAGE_SIZE * PAGES_PER_SGE)); + 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); + err = bnx2x_alloc_rx_sge(bp, fp, sge_idx, GFP_ATOMIC); if (unlikely(err)) { - fp->eth_q_stats.rx_skb_alloc_failed++; + bnx2x_fp_qstats(bp, fp)->rx_skb_alloc_failed++; return err; } - /* Unmap the page as we r going to pass it to the stack */ + /* 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_PAGE_SIZE*PAGES_PER_SGE, DMA_FROM_DEVICE); + 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); @@ -504,7 +653,7 @@ static int bnx2x_fill_frag_skb(struct bnx2x *bp, struct bnx2x_fastpath *fp, } skb->data_len += frag_len; - skb->truesize += SGE_PAGE_SIZE * PAGES_PER_SGE; + skb->truesize += SGE_PAGES; skb->len += frag_len; frag_size -= frag_len; @@ -513,6 +662,83 @@ static int bnx2x_fill_frag_skb(struct bnx2x *bp, struct bnx2x_fastpath *fp, 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, gfp_t gfp_mask) +{ + if (fp->rx_frag_size) { + /* GFP_KERNEL allocations are used only during initialization */ + if (unlikely(gfp_mask & __GFP_WAIT)) + return (void *)__get_free_page(gfp_mask); + + return netdev_alloc_frag(fp->rx_frag_size); + } + + return kmalloc(fp->rx_buf_size + NET_SKB_PAD, gfp_mask); +} + +#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 + skb_record_rx_queue(skb, fp->rx_queue); + 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, @@ -535,15 +761,14 @@ static void bnx2x_tpa_stop(struct bnx2x *bp, struct bnx2x_fastpath *fp, goto drop; /* Try to allocate the new data */ - new_data = kmalloc(fp->rx_buf_size + NET_SKB_PAD, GFP_ATOMIC); - + new_data = bnx2x_frag_alloc(fp, GFP_ATOMIC); /* 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, 0); + skb = build_skb(data, fp->rx_frag_size); if (likely(skb)) { #ifdef BNX2X_STOP_ON_ERROR @@ -557,7 +782,7 @@ static void bnx2x_tpa_stop(struct bnx2x *bp, struct bnx2x_fastpath *fp, skb_reserve(skb, pad + NET_SKB_PAD); skb_put(skb, len); - skb->rxhash = tpa_info->rxhash; + skb_set_hash(skb, tpa_info->rxhash, tpa_info->rxhash_type); skb->protocol = eth_type_trans(skb, bp->dev); skb->ip_summed = CHECKSUM_UNNECESSARY; @@ -565,37 +790,37 @@ static void bnx2x_tpa_stop(struct bnx2x *bp, struct bnx2x_fastpath *fp, 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, tpa_info->vlan_tag); - napi_gro_receive(&fp->napi, skb); + __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; } - kfree(new_data); + if (new_data) + 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"); - fp->eth_q_stats.rx_skb_alloc_failed++; + 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) +static int bnx2x_alloc_rx_data(struct bnx2x *bp, struct bnx2x_fastpath *fp, + u16 index, gfp_t gfp_mask) { 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 = kmalloc(fp->rx_buf_size + NET_SKB_PAD, GFP_ATOMIC); + data = bnx2x_frag_alloc(fp, gfp_mask); if (unlikely(data == NULL)) return -ENOMEM; @@ -603,7 +828,7 @@ static int bnx2x_alloc_rx_data(struct bnx2x *bp, fp->rx_buf_size, DMA_FROM_DEVICE); if (unlikely(dma_mapping_error(&bp->pdev->dev, mapping))) { - kfree(data); + bnx2x_frag_free(fp, data); BNX2X_ERR("Can't map rx data\n"); return -ENOMEM; } @@ -617,43 +842,45 @@ static int bnx2x_alloc_rx_data(struct bnx2x *bp, return 0; } -static void bnx2x_csum_validate(struct sk_buff *skb, union eth_rx_cqe *cqe, - struct bnx2x_fastpath *fp) +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 IP/L4 csum validation was done */ - + /* 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_IP_XSUM_NO_VALIDATION_FLG | - ETH_FAST_PATH_RX_CQE_L4_XSUM_NO_VALIDATION_FLG)) + ETH_FAST_PATH_RX_CQE_L4_XSUM_NO_VALIDATION_FLG) return; - /* If both IP/L4 validation were done, check if an error was found. */ + /* 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)) - fp->eth_q_stats.hw_csum_err++; + qstats->hw_csum_err++; else skb->ip_summed = CHECKSUM_UNNECESSARY; } -int bnx2x_rx_int(struct bnx2x_fastpath *fp, int budget) +static 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 hw_comp_cons, sw_comp_cons, sw_comp_prod; + 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 - - /* CQ "next element" is of the size of the regular element, - that's why it's ok here */ - hw_comp_cons = le16_to_cpu(*fp->rx_cons_sb); - if ((hw_comp_cons & MAX_RCQ_DESC_CNT) == MAX_RCQ_DESC_CNT) - hw_comp_cons++; + if (budget <= 0) + return rx_pkt; bd_cons = fp->rx_bd_cons; bd_prod = fp->rx_bd_prod; @@ -661,36 +888,43 @@ int bnx2x_rx_int(struct bnx2x_fastpath *fp, int budget) sw_comp_cons = fp->rx_comp_cons; sw_comp_prod = fp->rx_comp_prod; - /* Memory barrier necessary as speculative reads of the rx - * buffer can be ahead of the index in the status block - */ - rmb(); + 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]: hw_comp_cons %u sw_comp_cons %u\n", - fp->index, hw_comp_cons, sw_comp_cons); + "queue[%d]: sw_comp_cons %u\n", fp->index, sw_comp_cons); - while (sw_comp_cons != hw_comp_cons) { + while (BNX2X_IS_CQE_COMPLETED(cqe_fp)) { struct sw_rx_bd *rx_buf = NULL; struct sk_buff *skb; - union eth_rx_cqe *cqe; - struct eth_fast_path_rx_cqe *cqe_fp; u8 cqe_fp_flags; enum eth_rx_cqe_type cqe_fp_type; u16 len, pad, queue; u8 *data; + u32 rxhash; + enum pkt_hash_types rxhash_type; #ifdef BNX2X_STOP_ON_ERROR if (unlikely(bp->panic)) return 0; #endif - comp_ring_cons = RCQ_BD(sw_comp_cons); bd_prod = RX_BD(bd_prod); bd_cons = RX_BD(bd_cons); - cqe = &fp->rx_comp_ring[comp_ring_cons]; - cqe_fp = &cqe->fast_path_cqe; + /* A rmb() is required to ensure that the CQE is not read + * before it is written by the adapter DMA. PCI ordering + * rules will make sure the other fields are written before + * the marker at the end of struct eth_fast_path_rx_cqe + * but without rmb() a weakly ordered processor can process + * stale data. Without the barrier TPA state-machine might + * enter inconsistent state and kernel stack might be + * provided with incorrect packet description - these lead + * to various kernel crashed. + */ + rmb(); + cqe_fp_flags = cqe_fp->type_error_flags; cqe_fp_type = cqe_fp_flags & ETH_FAST_PATH_RX_CQE_TYPE; @@ -734,7 +968,6 @@ int bnx2x_rx_int(struct bnx2x_fastpath *fp, int budget) cqe_fp); goto next_rx; - } queue = cqe->end_agg_cqe.queue_index; tpa_info = &fp->tpa_info[queue]; @@ -776,7 +1009,7 @@ int bnx2x_rx_int(struct bnx2x_fastpath *fp, int budget) DP(NETIF_MSG_RX_ERR | NETIF_MSG_RX_STATUS, "ERROR flags %x rx packet %u\n", cqe_fp_flags, sw_comp_cons); - fp->eth_q_stats.rx_err_discard_pkt++; + bnx2x_fp_qstats(bp, fp)->rx_err_discard_pkt++; goto reuse_rx; } @@ -789,28 +1022,30 @@ int bnx2x_rx_int(struct bnx2x_fastpath *fp, int budget) if (skb == NULL) { DP(NETIF_MSG_RX_ERR | NETIF_MSG_RX_STATUS, "ERROR packet dropped because of alloc failure\n"); - fp->eth_q_stats.rx_skb_alloc_failed++; + 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)) { + if (likely(bnx2x_alloc_rx_data(bp, fp, bd_prod, + GFP_ATOMIC) == 0)) { dma_unmap_single(&bp->pdev->dev, dma_unmap_addr(rx_buf, mapping), fp->rx_buf_size, DMA_FROM_DEVICE); - skb = build_skb(data, 0); + skb = build_skb(data, fp->rx_frag_size); if (unlikely(!skb)) { - kfree(data); - fp->eth_q_stats.rx_skb_alloc_failed++; + 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"); - fp->eth_q_stats.rx_skb_alloc_failed++; + bnx2x_fp_qstats(bp, fp)->rx_skb_alloc_failed++; reuse_rx: bnx2x_reuse_rx_data(fp, bd_cons, bd_prod); goto next_rx; @@ -821,23 +1056,28 @@ reuse_rx: skb->protocol = eth_type_trans(skb, bp->dev); /* Set Toeplitz hash for a none-LRO skb */ - skb->rxhash = bnx2x_get_rxhash(bp, cqe_fp); + rxhash = bnx2x_get_rxhash(bp, cqe_fp, &rxhash_type); + skb_set_hash(skb, rxhash, rxhash_type); skb_checksum_none_assert(skb); if (bp->dev->features & NETIF_F_RXCSUM) - bnx2x_csum_validate(skb, cqe, fp); - + 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, + __vlan_hwaccel_put_tag(skb, htons(ETH_P_8021Q), le16_to_cpu(cqe_fp->vlan_tag)); - napi_gro_receive(&fp->napi, skb); + 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; @@ -849,8 +1089,15 @@ 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; @@ -877,6 +1124,7 @@ static irqreturn_t bnx2x_msix_fp_int(int irq, void *fp_cookie) 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 @@ -885,10 +1133,8 @@ static irqreturn_t bnx2x_msix_fp_int(int irq, void *fp_cookie) #endif /* Handle Rx and Tx according to MSI-X vector */ - prefetch(fp->rx_cons_sb); - for_each_cos_in_tx_queue(fp, cos) - prefetch(fp->txdata[cos].tx_cons_sb); + prefetch(fp->txdata_ptr[cos]->tx_cons_sb); prefetch(&fp->sb_running_index[SM_RX_ID]); napi_schedule(&bnx2x_fp(bp, fp->index, napi)); @@ -901,14 +1147,12 @@ void bnx2x_acquire_phy_lock(struct bnx2x *bp) { mutex_lock(&bp->port.phy_mutex); - if (bp->port.need_hw_lock) - bnx2x_acquire_hw_lock(bp, HW_LOCK_RESOURCE_MDIO); + bnx2x_acquire_hw_lock(bp, HW_LOCK_RESOURCE_MDIO); } void bnx2x_release_phy_lock(struct bnx2x *bp) { - if (bp->port.need_hw_lock) - bnx2x_release_hw_lock(bp, HW_LOCK_RESOURCE_MDIO); + bnx2x_release_hw_lock(bp, HW_LOCK_RESOURCE_MDIO); mutex_unlock(&bp->port.phy_mutex); } @@ -952,7 +1196,7 @@ static void bnx2x_fill_report_data(struct bnx2x *bp, memset(data, 0, sizeof(*data)); - /* Fill the report data: efective line speed */ + /* Fill the report data: effective line speed */ data->line_speed = line_speed; /* Link is down */ @@ -995,7 +1239,7 @@ void bnx2x_link_report(struct bnx2x *bp) * * @bp: driver handle * - * None atomic inmlementation. + * None atomic implementation. * Should be called under the phy_lock. */ void __bnx2x_link_report(struct bnx2x *bp) @@ -1003,7 +1247,7 @@ void __bnx2x_link_report(struct bnx2x *bp) struct bnx2x_link_report_data cur_data; /* reread mf_cfg */ - if (!CHIP_IS_E1(bp)) + if (IS_PF(bp) && !CHIP_IS_E1(bp)) bnx2x_read_mf_cfg(bp); /* Read the current link report info */ @@ -1100,11 +1344,30 @@ static void bnx2x_free_tpa_pool(struct bnx2x *bp, dma_unmap_single(&bp->pdev->dev, dma_unmap_addr(first_buf, mapping), fp->rx_buf_size, DMA_FROM_DEVICE); - kfree(data); + 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); @@ -1112,22 +1375,22 @@ void bnx2x_init_rx_rings(struct bnx2x *bp) int i, j; /* Allocate TPA resources */ - for_each_rx_queue(bp, j) { + 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-aggregtion pool */ + /* 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 = kmalloc(fp->rx_buf_size + NET_SKB_PAD, - GFP_ATOMIC); + first_buf->data = + bnx2x_frag_alloc(fp, GFP_KERNEL); if (!first_buf->data) { BNX2X_ERR("Failed to allocate TPA skb pool for queue[%d] - disabling TPA on this queue!\n", j); @@ -1149,7 +1412,8 @@ void bnx2x_init_rx_rings(struct bnx2x *bp) 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) { + if (bnx2x_alloc_rx_sge(bp, fp, ring_prod, + GFP_KERNEL) < 0) { BNX2X_ERR("was only able to allocate %d rx sges\n", i); BNX2X_ERR("disabling TPA for queue[%d]\n", @@ -1170,7 +1434,7 @@ void bnx2x_init_rx_rings(struct bnx2x *bp) } } - for_each_rx_queue(bp, j) { + for_each_eth_queue(bp, j) { struct bnx2x_fastpath *fp = &bp->fp[j]; fp->rx_bd_cons = 0; @@ -1197,28 +1461,45 @@ void bnx2x_init_rx_rings(struct bnx2x *bp) } } -static void bnx2x_free_tx_skbs(struct bnx2x *bp) +static void bnx2x_free_tx_skbs_queue(struct bnx2x_fastpath *fp) { - int i; u8 cos; + struct bnx2x *bp = fp->bp; - for_each_tx_queue(bp, i) { - struct bnx2x_fastpath *fp = &bp->fp[i]; - for_each_cos_in_tx_queue(fp, cos) { - struct bnx2x_fp_txdata *txdata = &fp->txdata[cos]; - unsigned pkts_compl = 0, bytes_compl = 0; + 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; + 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)); + 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]); } } @@ -1242,7 +1523,16 @@ static void bnx2x_free_rx_bds(struct bnx2x_fastpath *fp) fp->rx_buf_size, DMA_FROM_DEVICE); rx_buf->data = NULL; - kfree(data); + 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]); } } @@ -1250,7 +1540,7 @@ static void bnx2x_free_rx_skbs(struct bnx2x *bp) { int j; - for_each_rx_queue(bp, j) { + for_each_eth_queue(bp, j) { struct bnx2x_fastpath *fp = &bp->fp[j]; bnx2x_free_rx_bds(fp); @@ -1260,6 +1550,12 @@ static void bnx2x_free_rx_skbs(struct bnx2x *bp) } } +static 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); @@ -1295,15 +1591,20 @@ static void bnx2x_free_msix_irqs(struct bnx2x *bp, int nvecs) if (nvecs == offset) return; - free_irq(bp->msix_table[offset].vector, bp->dev); - DP(NETIF_MSG_IFDOWN, "released sp irq (%d)\n", - bp->msix_table[offset].vector); - offset++; -#ifdef BCM_CNIC - if (nvecs == offset) - return; - offset++; -#endif + + /* 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) @@ -1318,28 +1619,39 @@ static void bnx2x_free_msix_irqs(struct bnx2x *bp, int nvecs) void bnx2x_free_irq(struct bnx2x *bp) { if (bp->flags & USING_MSIX_FLAG && - !(bp->flags & USING_SINGLE_MSIX_FLAG)) - bnx2x_free_msix_irqs(bp, BNX2X_NUM_ETH_QUEUES(bp) + - CNIC_PRESENT + 1); - else + !(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 __devinit bnx2x_enable_msix(struct bnx2x *bp) +int bnx2x_enable_msix(struct bnx2x *bp) { - int msix_vec = 0, i, rc, req_cnt; + int msix_vec = 0, i, rc; - 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++; + /* 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++; + } -#ifdef BCM_CNIC - bp->msix_table[msix_vec].entry = msix_vec; - BNX2X_DEV_INFO("msix_table[%d].entry = %d (CNIC)\n", - bp->msix_table[msix_vec].entry, bp->msix_table[msix_vec].entry); - msix_vec++; -#endif /* We need separate vectors for ETH queues only (not FCoE) */ for_each_eth_queue(bp, i) { bp->msix_table[msix_vec].entry = msix_vec; @@ -1348,37 +1660,19 @@ int __devinit bnx2x_enable_msix(struct bnx2x *bp) msix_vec++; } - req_cnt = BNX2X_NUM_ETH_QUEUES(bp) + CNIC_PRESENT + 1; - - rc = pci_enable_msix(bp->pdev, &bp->msix_table[0], req_cnt); + DP(BNX2X_MSG_SP, "about to request enable msix with %d vectors\n", + msix_vec); + rc = pci_enable_msix_range(bp->pdev, &bp->msix_table[0], + BNX2X_MIN_MSIX_VEC_CNT(bp), msix_vec); /* * reconfigure number of tx/rx queues according to available * MSI-X vectors */ - if (rc >= BNX2X_MIN_MSIX_VEC_CNT) { - /* how less vectors we will have? */ - int diff = req_cnt - 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_queues -= diff; - - BNX2X_DEV_INFO("New queue configuration set: %d\n", - bp->num_queues); - } else if (rc > 0) { + if (rc == -ENOSPC) { /* Get by with single vector */ - rc = pci_enable_msix(bp->pdev, &bp->msix_table[0], 1); - if (rc) { + rc = pci_enable_msix_range(bp->pdev, &bp->msix_table[0], 1, 1); + if (rc < 0) { BNX2X_DEV_INFO("Single MSI-X is not attainable rc %d\n", rc); goto no_msix; @@ -1387,9 +1681,26 @@ int __devinit bnx2x_enable_msix(struct bnx2x *bp) 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); + BNX2X_DEV_INFO("MSI-X is not attainable rc %d\n", rc); goto no_msix; + } else if (rc < msix_vec) { + /* how less vectors we will have? */ + int diff = msix_vec - rc; + + BNX2X_DEV_INFO("Trying to use less MSI-X vectors: %d\n", rc); + + /* + * 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); } bp->flags |= USING_MSIX_FLAG; @@ -1408,17 +1719,20 @@ static int bnx2x_req_msix_irqs(struct bnx2x *bp) { int i, rc, offset = 0; - 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; + /* 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; + } } -#ifdef BCM_CNIC - offset++; -#endif + 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", @@ -1437,12 +1751,20 @@ static int bnx2x_req_msix_irqs(struct bnx2x *bp) } i = BNX2X_NUM_ETH_QUEUES(bp); - offset = 1 + CNIC_PRESENT; - 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); - + 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; } @@ -1487,7 +1809,6 @@ static int bnx2x_setup_irqs(struct bnx2x *bp) if (rc) return rc; } else { - bnx2x_ack_int(bp); rc = bnx2x_req_irq(bp); if (rc) { BNX2X_ERR("IRQ request failed rc %d, aborting\n", rc); @@ -1508,26 +1829,54 @@ static int bnx2x_setup_irqs(struct bnx2x *bp) 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_rx_queue(bp, 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; + + for_each_rx_queue_cnic(bp, i) { + napi_disable(&bnx2x_fp(bp, i, napi)); + while (!bnx2x_fp_ll_disable(&bp->fp[i])) + usleep_range(1000, 2000); + } } static void bnx2x_napi_disable(struct bnx2x *bp) { int i; - for_each_rx_queue(bp, i) + for_each_eth_queue(bp, i) { napi_disable(&bnx2x_fp(bp, i, napi)); + while (!bnx2x_fp_ll_disable(&bp->fp[i])) + usleep_range(1000, 2000); + } } 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); @@ -1538,14 +1887,16 @@ 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) +u16 bnx2x_select_queue(struct net_device *dev, struct sk_buff *skb, + void *accel_priv, select_queue_fallback_t fallback) { struct bnx2x *bp = netdev_priv(dev); -#ifdef BCM_CNIC - if (!NO_FCOE(bp)) { + if (CNIC_LOADED(bp) && !NO_FCOE(bp)) { struct ethhdr *hdr = (struct ethhdr *)skb->data; u16 ether_type = ntohs(hdr->h_proto); @@ -1561,24 +1912,25 @@ u16 bnx2x_select_queue(struct net_device *dev, struct sk_buff *skb) if ((ether_type == ETH_P_FCOE) || (ether_type == ETH_P_FIP)) return bnx2x_fcoe_tx(bp, txq_index); } -#endif + /* select a non-FCoE queue */ - return __skb_tx_hash(dev, skb, BNX2X_NUM_ETH_QUEUES(bp)); + return fallback(dev, skb) % BNX2X_NUM_ETH_QUEUES(bp); } - void bnx2x_set_num_queues(struct bnx2x *bp) { /* RSS queues */ - bp->num_queues = bnx2x_calc_num_queues(bp); + bp->num_ethernet_queues = bnx2x_calc_num_queues(bp); -#ifdef BCM_CNIC /* override in STORAGE SD modes */ if (IS_MF_STORAGE_SD(bp) || IS_MF_FCOE_AFEX(bp)) - bp->num_queues = 1; -#endif + bp->num_ethernet_queues = 1; + /* Add special queues */ - bp->num_queues += NON_ETH_CONTEXT_USE; + 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); } /** @@ -1595,7 +1947,7 @@ void bnx2x_set_num_queues(struct bnx2x *bp) * * 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 indicies that are not coupled with any real Tx queue. + * 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(). @@ -1603,20 +1955,18 @@ void bnx2x_set_num_queues(struct bnx2x *bp) * 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) +static int bnx2x_set_real_num_queues(struct bnx2x *bp, int include_cnic) { int rc, tx, rx; - tx = MAX_TXQS_PER_COS * bp->max_cos; + tx = BNX2X_NUM_ETH_QUEUES(bp) * bp->max_cos; rx = BNX2X_NUM_ETH_QUEUES(bp); /* account for fcoe queue */ -#ifdef BCM_CNIC - if (!NO_FCOE(bp)) { - rx += FCOE_PRESENT; - tx += FCOE_PRESENT; + if (include_cnic && !NO_FCOE(bp)) { + rx++; + tx++; } -#endif rc = netif_set_real_num_tx_queues(bp->dev, tx); if (rc) { @@ -1659,21 +2009,24 @@ static void bnx2x_set_rx_buf_size(struct bnx2x *bp) ETH_OVREHEAD + mtu + BNX2X_FW_RX_ALIGN_END; - /* Note : rx_buf_size doesnt take into account NET_SKB_PAD */ + /* 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) +static int bnx2x_init_rss(struct bnx2x *bp) { int i; - u8 ind_table[T_ETH_INDIRECTION_TABLE_SIZE] = {0}; u8 num_eth_queues = BNX2X_NUM_ETH_QUEUES(bp); - /* Prepare the initial contents fo the indirection table if RSS is + /* Prepare the initial contents for the indirection table if RSS is * enabled */ - for (i = 0; i < sizeof(ind_table); i++) - ind_table[i] = + 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); @@ -1685,15 +2038,13 @@ static int bnx2x_init_rss_pf(struct bnx2x *bp) * For 57712 and newer on the other hand it's a per-function * configuration. */ - return bnx2x_config_rss_eth(bp, ind_table, - bp->port.pmf || !CHIP_IS_E1x(bp)); + 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, - u8 *ind_table, bool config_hash) +int bnx2x_rss(struct bnx2x *bp, struct bnx2x_rss_config_obj *rss_obj, + bool config_hash, bool enable) { struct bnx2x_config_rss_params params = {NULL}; - int i; /* Although RSS is meaningless when there is a single HW queue we * still need it enabled in order to have HW Rx hash generated. @@ -1706,28 +2057,37 @@ int bnx2x_config_rss_pf(struct bnx2x *bp, struct bnx2x_rss_config_obj *rss_obj, __set_bit(RAMROD_COMP_WAIT, ¶ms.ramrod_flags); - __set_bit(BNX2X_RSS_MODE_REGULAR, ¶ms.rss_flags); - - /* RSS configuration */ - __set_bit(BNX2X_RSS_IPV4, ¶ms.rss_flags); - __set_bit(BNX2X_RSS_IPV4_TCP, ¶ms.rss_flags); - __set_bit(BNX2X_RSS_IPV6, ¶ms.rss_flags); - __set_bit(BNX2X_RSS_IPV6_TCP, ¶ms.rss_flags); + if (enable) { + __set_bit(BNX2X_RSS_MODE_REGULAR, ¶ms.rss_flags); + + /* RSS configuration */ + __set_bit(BNX2X_RSS_IPV4, ¶ms.rss_flags); + __set_bit(BNX2X_RSS_IPV4_TCP, ¶ms.rss_flags); + __set_bit(BNX2X_RSS_IPV6, ¶ms.rss_flags); + __set_bit(BNX2X_RSS_IPV6_TCP, ¶ms.rss_flags); + if (rss_obj->udp_rss_v4) + __set_bit(BNX2X_RSS_IPV4_UDP, ¶ms.rss_flags); + if (rss_obj->udp_rss_v6) + __set_bit(BNX2X_RSS_IPV6_UDP, ¶ms.rss_flags); + } else { + __set_bit(BNX2X_RSS_MODE_DISABLED, ¶ms.rss_flags); + } /* Hash bits */ params.rss_result_mask = MULTI_MASK; - memcpy(params.ind_table, ind_table, sizeof(params.ind_table)); + memcpy(params.ind_table, rss_obj->ind_table, sizeof(params.ind_table)); if (config_hash) { /* RSS keys */ - for (i = 0; i < sizeof(params.rss_key) / 4; i++) - params.rss_key[i] = random32(); - + prandom_bytes(params.rss_key, T_ETH_RSS_KEY * 4); __set_bit(BNX2X_RSS_SET_SRCH, ¶ms.rss_flags); } - return bnx2x_config_rss(bp, ¶ms); + if (IS_PF(bp)) + return bnx2x_config_rss(bp, ¶ms); + else + return bnx2x_vfpf_config_rss(bp, ¶ms); } static int bnx2x_init_hw(struct bnx2x *bp, u32 load_code) @@ -1747,14 +2107,14 @@ static int bnx2x_init_hw(struct bnx2x *bp, u32 load_code) /* * Cleans the object that have internal lists without sending - * ramrods. Should be run when interrutps are disabled. + * ramrods. Should be run when interrupts are disabled. */ -static void bnx2x_squeeze_objects(struct bnx2x *bp) +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->fp->mac_obj; + struct bnx2x_vlan_mac_obj *mac_obj = &bp->sp_objs->mac_obj; /***************** Cleanup MACs' object first *************************/ @@ -1765,7 +2125,7 @@ static void bnx2x_squeeze_objects(struct bnx2x *bp) /* Clean ETH primary MAC */ __set_bit(BNX2X_ETH_MAC, &vlan_mac_flags); - rc = mac_obj->delete_all(bp, &bp->fp->mac_obj, &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); @@ -1782,7 +2142,11 @@ static void bnx2x_squeeze_objects(struct bnx2x *bp) rparam.mcast_obj = &bp->mcast_obj; __set_bit(RAMROD_DRV_CLR_ONLY, &rparam.ramrod_flags); - /* Add a DEL command... */ + /* Add a DEL command... - Since we're doing a driver cleanup only, + * we take a lock surrounding both the initial send and the CONTs, + * as we don't want a true completion to disrupt us in the middle. + */ + netif_addr_lock_bh(bp->dev); 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", @@ -1794,11 +2158,13 @@ static void bnx2x_squeeze_objects(struct bnx2x *bp) if (rc < 0) { BNX2X_ERR("Failed to clean multi-cast object: %d\n", rc); + netif_addr_unlock_bh(bp->dev); return; } rc = bnx2x_config_mcast(bp, &rparam, BNX2X_MCAST_CMD_CONT); } + netif_addr_unlock_bh(bp->dev); } #ifndef BNX2X_STOP_ON_ERROR @@ -1807,36 +2173,245 @@ static void bnx2x_squeeze_objects(struct bnx2x *bp) (bp)->state = BNX2X_STATE_ERROR; \ goto label; \ } while (0) -#else + +#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) -#endif +#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); + + bp->fw_stats = BNX2X_PCI_ALLOC(&bp->fw_stats_mapping, + bp->fw_stats_data_sz + bp->fw_stats_req_sz); + if (!bp->fw_stats) + goto alloc_mem_err; + + /* 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_compare_fw_ver(struct bnx2x *bp, u32 load_code, bool print_err) +{ + /* 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) { + if (print_err) + BNX2X_ERR("bnx2x with FW %x was already loaded which mismatches my %x FW. Aborting\n", + loaded_fw, my_fw); + else + BNX2X_DEV_INFO("bnx2x with FW %x was already loaded which mismatches my %x FW, possibly due to MF UNDI\n", + loaded_fw, my_fw); + return -EBUSY; + } + } + return 0; +} -bool bnx2x_test_firmware_version(struct bnx2x *bp, bool is_err) +/* returns the "mcp load_code" according to global load_count array */ +static int bnx2x_nic_load_no_mcp(struct bnx2x *bp, int port) { - /* build 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); + int path = BP_PATH(bp); + + DP(NETIF_MSG_IFUP, "NO MCP - load counts[%d] %d, %d, %d\n", + path, bnx2x_load_count[path][0], bnx2x_load_count[path][1], + bnx2x_load_count[path][2]); + bnx2x_load_count[path][0]++; + bnx2x_load_count[path][1 + port]++; + DP(NETIF_MSG_IFUP, "NO MCP - new load counts[%d] %d, %d, %d\n", + path, bnx2x_load_count[path][0], bnx2x_load_count[path][1], + bnx2x_load_count[path][2]); + if (bnx2x_load_count[path][0] == 1) + return FW_MSG_CODE_DRV_LOAD_COMMON; + else if (bnx2x_load_count[path][1 + port] == 1) + return FW_MSG_CODE_DRV_LOAD_PORT; + else + return FW_MSG_CODE_DRV_LOAD_FUNCTION; +} - /* read loaded FW from chip */ - u32 loaded_fw = REG_RD(bp, XSEM_REG_PRAM); +/* 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_IFUP, "loaded fw %x, my fw %x\n", loaded_fw, my_fw); + DP(NETIF_MSG_LINK, "pmf %d\n", bp->port.pmf); +} - if (loaded_fw != my_fw) { - if (is_err) - BNX2X_ERR("bnx2x with FW %x was already loaded, which mismatches my %x FW. aborting\n", - loaded_fw, my_fw); - return false; +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); } - return true; + /* Set AFEX default VLAN tag to an invalid value */ + bp->afex_def_vlan_tag = -1; } /** @@ -1851,47 +2426,19 @@ bool bnx2x_test_firmware_version(struct bnx2x *bp, bool is_err) 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; - /* bzero bnx2x_fastpath contents */ - if (bp->stats_init) - memset(fp, 0, sizeof(*fp)); - else { - /* Keep Queue statistics */ - struct bnx2x_eth_q_stats *tmp_eth_q_stats; - struct bnx2x_eth_q_stats_old *tmp_eth_q_stats_old; - - tmp_eth_q_stats = kzalloc(sizeof(struct bnx2x_eth_q_stats), - GFP_KERNEL); - if (tmp_eth_q_stats) - memcpy(tmp_eth_q_stats, &fp->eth_q_stats, - sizeof(struct bnx2x_eth_q_stats)); - - tmp_eth_q_stats_old = - kzalloc(sizeof(struct bnx2x_eth_q_stats_old), - GFP_KERNEL); - if (tmp_eth_q_stats_old) - memcpy(tmp_eth_q_stats_old, &fp->eth_q_stats_old, - sizeof(struct bnx2x_eth_q_stats_old)); - - memset(fp, 0, sizeof(*fp)); - - if (tmp_eth_q_stats) { - memcpy(&fp->eth_q_stats, tmp_eth_q_stats, - sizeof(struct bnx2x_eth_q_stats)); - kfree(tmp_eth_q_stats); - } - - if (tmp_eth_q_stats_old) { - memcpy(&fp->eth_q_stats_old, tmp_eth_q_stats_old, - sizeof(struct bnx2x_eth_q_stats_old)); - kfree(tmp_eth_q_stats_old); - } + 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)) @@ -1900,8 +2447,15 @@ static void bnx2x_bz_fp(struct bnx2x *bp, int index) /* Special queues support only one CoS */ fp->max_cos = 1; - /* - * set the tpa flag for each queue. The tpa flag determines the queue + /* 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 || @@ -1912,20 +2466,107 @@ static void bnx2x_bz_fp(struct bnx2x *bp, int index) else if (bp->flags & GRO_ENABLE_FLAG) fp->mode = TPA_MODE_GRO; -#ifdef BCM_CNIC /* We don't want TPA on an FCoE L2 ring */ if (IS_FCOE_FP(fp)) fp->disable_tpa = 1; -#endif } +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. */ + bnx2x_set_rx_mode_inner(bp); + + /* 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 */ + if (bnx2x_set_real_num_queues(bp, 0)) + 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); - u32 load_code; - int i, rc; + 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)) { @@ -1936,246 +2577,234 @@ int bnx2x_nic_load(struct bnx2x *bp, int load_mode) bp->state = BNX2X_STATE_OPENING_WAIT4_LOAD; - /* Set the initial link reported state to link down */ - bnx2x_acquire_phy_lock(bp); + /* 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); - bnx2x_release_phy_lock(bp); - /* must be called before memory allocation and HW init */ - bnx2x_ilt_set_info(bp); + 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. + * 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 (bnx2x_alloc_mem(bp)) - return -ENOMEM; + if (IS_PF(bp)) { + rc = bnx2x_alloc_mem(bp); + if (rc) { + BNX2X_ERR("Unable to allocate bp memory\n"); + return rc; + } + } + + /* 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); + } + + /* Allocated memory for FW statistics */ + if (bnx2x_alloc_fw_stats_mem(bp)) + 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. + * come after it. At this stage cnic queues are not counted. */ - rc = bnx2x_set_real_num_queues(bp); + 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 overriden by a multi class queue discipline - * or by a dcbx negotiation result. + * 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); - /* set pf load just before approaching the MCP */ - bnx2x_set_pf_load(bp); - - /* Send LOAD_REQUEST command to MCP - * Returns the type of LOAD command: - * if it is the first port to be initialized - * common blocks should be initialized, otherwise - not - */ - if (!BP_NOMCP(bp)) { - /* 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); - - load_code = bnx2x_fw_command(bp, DRV_MSG_CODE_LOAD_REQ, 0); - if (!load_code) { - BNX2X_ERR("MCP response failure, aborting\n"); - rc = -EBUSY; - LOAD_ERROR_EXIT(bp, load_error1); - } - if (load_code == FW_MSG_CODE_DRV_LOAD_REFUSED) { - BNX2X_ERR("Driver load refused\n"); - rc = -EBUSY; /* other port in diagnostic mode */ - LOAD_ERROR_EXIT(bp, load_error1); - } - if (load_code != FW_MSG_CODE_DRV_LOAD_COMMON_CHIP && - load_code != FW_MSG_CODE_DRV_LOAD_COMMON) { - /* abort nic load if version mismatch */ - if (!bnx2x_test_firmware_version(bp, true)) { - rc = -EBUSY; + 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_compare_fw_ver(bp, load_code, true); + 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); } - } else { - 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) - load_code = FW_MSG_CODE_DRV_LOAD_COMMON; - else if (load_count[path][1 + port] == 1) - load_code = FW_MSG_CODE_DRV_LOAD_PORT; - else - load_code = FW_MSG_CODE_DRV_LOAD_FUNCTION; - } + /* mark pmf if applicable */ + bnx2x_nic_load_pmf(bp, 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_IFUP, "pmf %d\n", bp->port.pmf); - - /* Init Function state controlling object */ - bnx2x__init_func_obj(bp); + /* 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); + /* 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("IRQs setup failed\n"); - bnx2x_fw_command(bp, DRV_MSG_CODE_LOAD_DONE, 0); + 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); } - /* Setup NIC internals and enable interrupts */ - bnx2x_nic_init(bp, load_code); - /* Init per-function objects */ - bnx2x_init_bp_objs(bp); - - 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; - - 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); - } + 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); - /* 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); } + + /* 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); } -#ifdef BCM_CNIC - /* Enable Timer scan */ - REG_WR(bp, TM_REG_EN_LINEAR0_TIMER + port*4, 1); -#endif - - for_each_nondefault_queue(bp, i) { - rc = bnx2x_setup_queue(bp, &bp->fp[i], 0); + /* set up the rest of the queues */ + for_each_nondefault_eth_queue(bp, i) { + if (IS_PF(bp)) + rc = bnx2x_setup_queue(bp, &bp->fp[i], false); + else /* VF */ + rc = bnx2x_vfpf_setup_q(bp, &bp->fp[i], false); if (rc) { - BNX2X_ERR("Queue setup failed\n"); - LOAD_ERROR_EXIT(bp, load_error4); + BNX2X_ERR("Queue %d setup failed\n", i); + LOAD_ERROR_EXIT(bp, load_error3); } } - rc = bnx2x_init_rss_pf(bp); + /* setup rss */ + rc = bnx2x_init_rss(bp); if (rc) { BNX2X_ERR("PF RSS init failed\n"); - LOAD_ERROR_EXIT(bp, load_error4); + 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 */ - rc = bnx2x_set_eth_mac(bp, true); + 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_error4); + LOAD_ERROR_EXIT(bp, load_error3); } - if (bp->pending_max) { + if (IS_PF(bp) && bp->pending_max) { bnx2x_update_max_mf_config(bp, bp->pending_max); bp->pending_max = 0; } - if (bp->port.pmf) - bnx2x_initial_phy_init(bp, load_mode); + 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); + bnx2x_set_rx_mode_inner(bp); /* Start the Tx */ switch (load_mode) { case LOAD_NORMAL: - /* Tx queue should be only reenabled */ + /* 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(); + smp_mb__after_atomic(); break; case LOAD_DIAG: + case LOAD_LOOPBACK_EXT: bp->state = BNX2X_STATE_DIAG; break; @@ -2184,23 +2813,21 @@ int bnx2x_nic_load(struct bnx2x *bp, int load_mode) } if (bp->port.pmf) - bnx2x_update_drv_flags(bp, 1 << DRV_FLAGS_DCB_CONFIGURED, 0); + 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); -#ifdef BCM_CNIC - /* re-read iscsi info */ - bnx2x_get_iscsi_info(bp); - bnx2x_setup_cnic_irq_info(bp); - if (bp->state == BNX2X_STATE_OPEN) - bnx2x_cnic_notify(bp, CNIC_CTL_START_CMD); -#endif + if (CNIC_ENABLED(bp)) + bnx2x_load_cnic(bp); + + if (IS_PF(bp)) + bnx2x_schedule_sp_rtnl(bp, BNX2X_SP_RTNL_GET_DRV_VERSION, 0); - /* mark driver is loaded in shmem2 */ - if (SHMEM2_HAS(bp, drv_capabilities_flag)) { + 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)], @@ -2209,26 +2836,28 @@ int bnx2x_nic_load(struct bnx2x *bp, int load_mode) } /* Wait for all pending SP commands to complete */ - if (!bnx2x_wait_sp_comp(bp, ~0x0UL)) { + 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); + bnx2x_nic_unload(bp, UNLOAD_CLOSE, false); return -EBUSY; } - bnx2x_dcbx_init(bp); + /* 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_error4: -#ifdef BCM_CNIC - /* Disable Timer scan */ - REG_WR(bp, TM_REG_EN_LINEAR0_TIMER + port*4, 0); -#endif load_error3: - bnx2x_int_disable_sync(bp, 1); + if (IS_PF(bp)) { + bnx2x_int_disable_sync(bp, 1); - /* Clean queueable objects */ - bnx2x_squeeze_objects(bp); + /* Clean queueable objects */ + bnx2x_squeeze_objects(bp); + } /* Free SKBs, SGEs, TPA pool and driver internals */ bnx2x_free_skbs(bp); @@ -2238,7 +2867,7 @@ load_error3: /* Release IRQs */ bnx2x_free_irq(bp); load_error2: - if (!BP_NOMCP(bp)) { + 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); } @@ -2246,31 +2875,55 @@ load_error2: bp->port.pmf = 0; load_error1: bnx2x_napi_disable(bp); + bnx2x_del_all_napi(bp); + /* clear pf_load status, as it was already set */ - bnx2x_clear_pf_load(bp); + if (IS_PF(bp)) + bnx2x_clear_pf_load(bp); load_error0: + bnx2x_free_fw_stats_mem(bp); + bnx2x_free_fp_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) +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 (SHMEM2_HAS(bp, drv_capabilities_flag)) { + 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 ((bp->state == BNX2X_STATE_CLOSED) || - (bp->state == BNX2X_STATE_ERROR)) { + 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 @@ -2288,44 +2941,62 @@ int bnx2x_nic_unload(struct bnx2x *bp, int unload_mode) return -EINVAL; } - /* - * It's important to set the bp->state to the value different from + /* 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); - -#ifdef BCM_CNIC - bnx2x_cnic_notify(bp, CNIC_CTL_STOP_CMD); -#endif + netdev_reset_tc(bp->dev); bp->rx_mode = BNX2X_RX_MODE_NONE; del_timer_sync(&bp->timer); - /* Set ALWAYS_ALIVE bit in shmem */ - bp->fw_drv_pulse_wr_seq |= DRV_PULSE_ALWAYS_ALIVE; - - bnx2x_drv_pulse(bp); + 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); + } - 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); - /* Cleanup the chip if needed */ - if (unload_mode != UNLOAD_RECOVERY) - bnx2x_chip_cleanup(bp, unload_mode); + /* 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 + /* Prevent transactions to host from the functions on the * engine that doesn't reset global blocks in case of global - * attention once gloabl blocks are reset and gates are opened + * attention once global blocks are reset and gates are opened * (the engine which leader will perform the recovery * last). */ @@ -2334,38 +3005,61 @@ int bnx2x_nic_unload(struct bnx2x *bp, int unload_mode) /* 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); + bnx2x_send_unload_done(bp, false); } /* - * At this stage no more interrupts will arrive so we may safly clean + * 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. */ - bnx2x_squeeze_objects(bp); + 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; + /* clear pending work in rtnl task */ + bp->sp_rtnl_state = 0; + smp_mb(); + /* 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; + + /* Clear driver version indication in shmem */ + if (IS_PF(bp)) + bnx2x_update_mng_version(bp); /* Check if there are pending parity attentions. If there are - set * RECOVERY_IN_PROGRESS. */ - if (bnx2x_chk_parity_attn(bp, &global, false)) { + if (IS_PF(bp) && bnx2x_chk_parity_attn(bp, &global, false)) { bnx2x_set_reset_in_progress(bp); /* Set RESET_IS_GLOBAL if needed */ @@ -2373,13 +3067,16 @@ int bnx2x_nic_unload(struct bnx2x *bp, int unload_mode) 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 (!bnx2x_clear_pf_load(bp) && bnx2x_reset_is_done(bp, BP_PATH(bp))) + 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; } @@ -2388,16 +3085,16 @@ 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) { + if (!bp->pdev->pm_cap) { BNX2X_DEV_INFO("No power capability. Breaking.\n"); return 0; } - pci_read_config_word(bp->pdev, bp->pm_cap + PCI_PM_CTRL, &pmcsr); + pci_read_config_word(bp->pdev, bp->pdev->pm_cap + PCI_PM_CTRL, &pmcsr); switch (state) { case PCI_D0: - pci_write_config_word(bp->pdev, bp->pm_cap + PCI_PM_CTRL, + pci_write_config_word(bp->pdev, bp->pdev->pm_cap + PCI_PM_CTRL, ((pmcsr & ~PCI_PM_CTRL_STATE_MASK) | PCI_PM_CTRL_PME_STATUS)); @@ -2421,7 +3118,7 @@ int bnx2x_set_power_state(struct bnx2x *bp, pci_power_t state) if (bp->wol) pmcsr |= PCI_PM_CTRL_PME_ENABLE; - pci_write_config_word(bp->pdev, bp->pm_cap + PCI_PM_CTRL, + pci_write_config_word(bp->pdev, bp->pdev->pm_cap + PCI_PM_CTRL, pmcsr); /* No more memory access after this point until @@ -2439,7 +3136,7 @@ int bnx2x_set_power_state(struct bnx2x *bp, pci_power_t state) /* * net_device service functions */ -int bnx2x_poll(struct napi_struct *napi, int budget) +static int bnx2x_poll(struct napi_struct *napi, int budget) { int work_done = 0; u8 cos; @@ -2454,23 +3151,27 @@ int bnx2x_poll(struct napi_struct *napi, int budget) return 0; } #endif + if (!bnx2x_fp_lock_napi(fp)) + return work_done; for_each_cos_in_tx_queue(fp, cos) - if (bnx2x_tx_queue_has_work(&fp->txdata[cos])) - bnx2x_tx_int(bp, &fp->txdata[cos]); - + if (bnx2x_tx_queue_has_work(fp->txdata_ptr[cos])) + bnx2x_tx_int(bp, fp->txdata_ptr[cos]); if (bnx2x_has_rx_work(fp)) { work_done += bnx2x_rx_int(fp, budget - work_done); /* must not complete if we consumed full budget */ - if (work_done >= budget) + if (work_done >= budget) { + bnx2x_fp_unlock_napi(fp); break; + } } /* Fall out from the NAPI loop if needed */ - if (!(bnx2x_has_rx_work(fp) || bnx2x_has_tx_work(fp))) { -#ifdef BCM_CNIC + if (!bnx2x_fp_unlock_napi(fp) && + !(bnx2x_has_rx_work(fp) || bnx2x_has_tx_work(fp))) { + /* No need to update SB for FCoE L2 ring as long as * it's connected to the default SB and the SB * has been updated when NAPI was scheduled. @@ -2479,8 +3180,6 @@ int bnx2x_poll(struct napi_struct *napi, int budget) napi_complete(napi); break; } -#endif - bnx2x_update_fpsb_idx(fp); /* bnx2x_has_rx_work() reads the status block, * thus we need to ensure that status block indices @@ -2513,15 +3212,41 @@ int bnx2x_poll(struct napi_struct *napi, int budget) return work_done; } +#ifdef CONFIG_NET_RX_BUSY_POLL +/* must be called with local_bh_disable()d */ +int bnx2x_low_latency_recv(struct napi_struct *napi) +{ + struct bnx2x_fastpath *fp = container_of(napi, struct bnx2x_fastpath, + napi); + struct bnx2x *bp = fp->bp; + int found = 0; + + if ((bp->state == BNX2X_STATE_CLOSED) || + (bp->state == BNX2X_STATE_ERROR) || + (bp->flags & (TPA_ENABLE_FLAG | GRO_ENABLE_FLAG))) + return LL_FLUSH_FAILED; + + if (!bnx2x_fp_lock_poll(fp)) + return LL_FLUSH_BUSY; + + if (bnx2x_has_rx_work(fp)) + found = bnx2x_rx_int(fp, 4); + + bnx2x_fp_unlock_poll(fp); + + return found; +} +#endif + /* we split the first BD into headers and data BDs * to ease the pain of our fellow microcode engineers * we use one mapping for both BDs */ -static noinline u16 bnx2x_tx_split(struct bnx2x *bp, - struct bnx2x_fp_txdata *txdata, - struct sw_tx_bd *tx_buf, - struct eth_tx_start_bd **tx_bd, u16 hlen, - u16 bd_prod, int nbd) +static u16 bnx2x_tx_split(struct bnx2x *bp, + struct bnx2x_fp_txdata *txdata, + struct sw_tx_bd *tx_buf, + struct eth_tx_start_bd **tx_bd, u16 hlen, + u16 bd_prod) { struct eth_tx_start_bd *h_tx_bd = *tx_bd; struct eth_tx_bd *d_tx_bd; @@ -2529,11 +3254,10 @@ static noinline u16 bnx2x_tx_split(struct bnx2x *bp, int old_len = le16_to_cpu(h_tx_bd->nbytes); /* first fix first BD */ - h_tx_bd->nbd = cpu_to_le16(nbd); h_tx_bd->nbytes = cpu_to_le16(hlen); - DP(NETIF_MSG_TX_QUEUED, "TSO split header size is %d (%x:%x) nbd %d\n", - h_tx_bd->nbytes, h_tx_bd->addr_hi, h_tx_bd->addr_lo, h_tx_bd->nbd); + DP(NETIF_MSG_TX_QUEUED, "TSO split header size is %d (%x:%x)\n", + h_tx_bd->nbytes, h_tx_bd->addr_hi, h_tx_bd->addr_lo); /* now get a new data BD * (after the pbd) and fill it */ @@ -2560,43 +3284,66 @@ static noinline u16 bnx2x_tx_split(struct bnx2x *bp, return bd_prod; } -static inline u16 bnx2x_csum_fix(unsigned char *t_header, u16 csum, s8 fix) +#define bswab32(b32) ((__force __le32) swab32((__force __u32) (b32))) +#define bswab16(b16) ((__force __le16) swab16((__force __u16) (b16))) +static __le16 bnx2x_csum_fix(unsigned char *t_header, u16 csum, s8 fix) { + __sum16 tsum = (__force __sum16) csum; + if (fix > 0) - csum = (u16) ~csum_fold(csum_sub(csum, - csum_partial(t_header - fix, fix, 0))); + tsum = ~csum_fold(csum_sub((__force __wsum) csum, + csum_partial(t_header - fix, fix, 0))); else if (fix < 0) - csum = (u16) ~csum_fold(csum_add(csum, - csum_partial(t_header, -fix, 0))); + tsum = ~csum_fold(csum_add((__force __wsum) csum, + csum_partial(t_header, -fix, 0))); - return swab16(csum); + return bswab16(tsum); } -static inline u32 bnx2x_xmit_type(struct bnx2x *bp, struct sk_buff *skb) +static u32 bnx2x_xmit_type(struct bnx2x *bp, struct sk_buff *skb) { u32 rc; + __u8 prot = 0; + __be16 protocol; if (skb->ip_summed != CHECKSUM_PARTIAL) - rc = XMIT_PLAIN; + return XMIT_PLAIN; - else { - if (vlan_get_protocol(skb) == htons(ETH_P_IPV6)) { - rc = XMIT_CSUM_V6; - if (ipv6_hdr(skb)->nexthdr == IPPROTO_TCP) - rc |= XMIT_CSUM_TCP; + protocol = vlan_get_protocol(skb); + if (protocol == htons(ETH_P_IPV6)) { + rc = XMIT_CSUM_V6; + prot = ipv6_hdr(skb)->nexthdr; + } else { + rc = XMIT_CSUM_V4; + prot = ip_hdr(skb)->protocol; + } + if (!CHIP_IS_E1x(bp) && skb->encapsulation) { + if (inner_ip_hdr(skb)->version == 6) { + rc |= XMIT_CSUM_ENC_V6; + if (inner_ipv6_hdr(skb)->nexthdr == IPPROTO_TCP) + rc |= XMIT_CSUM_TCP; } else { - rc = XMIT_CSUM_V4; - if (ip_hdr(skb)->protocol == IPPROTO_TCP) + rc |= XMIT_CSUM_ENC_V4; + if (inner_ip_hdr(skb)->protocol == IPPROTO_TCP) rc |= XMIT_CSUM_TCP; } } - - if (skb_is_gso_v6(skb)) - rc |= XMIT_GSO_V6 | XMIT_CSUM_TCP | XMIT_CSUM_V6; - else if (skb_is_gso(skb)) - rc |= XMIT_GSO_V4 | XMIT_CSUM_V4 | XMIT_CSUM_TCP; + if (prot == IPPROTO_TCP) + rc |= XMIT_CSUM_TCP; + + if (skb_is_gso(skb)) { + if (skb_is_gso_v6(skb)) { + rc |= (XMIT_GSO_V6 | XMIT_CSUM_TCP); + if (rc & XMIT_CSUM_ENC) + rc |= XMIT_GSO_ENC_V6; + } else { + rc |= (XMIT_GSO_V4 | XMIT_CSUM_TCP); + if (rc & XMIT_CSUM_ENC) + rc |= XMIT_GSO_ENC_V4; + } + } return rc; } @@ -2681,14 +3428,23 @@ exit_lbl: } #endif -static inline void bnx2x_set_pbd_gso_e2(struct sk_buff *skb, u32 *parsing_data, - u32 xmit_type) +static void bnx2x_set_pbd_gso_e2(struct sk_buff *skb, u32 *parsing_data, + u32 xmit_type) { + struct ipv6hdr *ipv6; + *parsing_data |= (skb_shinfo(skb)->gso_size << ETH_TX_PARSE_BD_E2_LSO_MSS_SHIFT) & ETH_TX_PARSE_BD_E2_LSO_MSS; - if ((xmit_type & XMIT_GSO_V6) && - (ipv6_hdr(skb)->nexthdr == NEXTHDR_IPV6)) + + if (xmit_type & XMIT_GSO_ENC_V6) + ipv6 = inner_ipv6_hdr(skb); + else if (xmit_type & XMIT_GSO_V6) + ipv6 = ipv6_hdr(skb); + else + ipv6 = NULL; + + if (ipv6 && ipv6->nexthdr == NEXTHDR_IPV6) *parsing_data |= ETH_TX_PARSE_BD_E2_IPV6_WITH_EXT_HDR; } @@ -2699,28 +3455,67 @@ static inline void bnx2x_set_pbd_gso_e2(struct sk_buff *skb, u32 *parsing_data, * @pbd: parse BD * @xmit_type: xmit flags */ -static inline void bnx2x_set_pbd_gso(struct sk_buff *skb, - struct eth_tx_parse_bd_e1x *pbd, - u32 xmit_type) +static void bnx2x_set_pbd_gso(struct sk_buff *skb, + struct eth_tx_parse_bd_e1x *pbd, + struct eth_tx_start_bd *tx_start_bd, + u32 xmit_type) { pbd->lso_mss = cpu_to_le16(skb_shinfo(skb)->gso_size); - pbd->tcp_send_seq = swab32(tcp_hdr(skb)->seq); - pbd->tcp_flags = pbd_tcp_flags(skb); + pbd->tcp_send_seq = bswab32(tcp_hdr(skb)->seq); + pbd->tcp_flags = pbd_tcp_flags(tcp_hdr(skb)); if (xmit_type & XMIT_GSO_V4) { - pbd->ip_id = swab16(ip_hdr(skb)->id); + pbd->ip_id = bswab16(ip_hdr(skb)->id); pbd->tcp_pseudo_csum = - swab16(~csum_tcpudp_magic(ip_hdr(skb)->saddr, - ip_hdr(skb)->daddr, - 0, IPPROTO_TCP, 0)); + bswab16(~csum_tcpudp_magic(ip_hdr(skb)->saddr, + ip_hdr(skb)->daddr, + 0, IPPROTO_TCP, 0)); - } else + /* GSO on 57710/57711 needs FW to calculate IP checksum */ + tx_start_bd->bd_flags.as_bitfield |= ETH_TX_BD_FLAGS_IP_CSUM; + } else { pbd->tcp_pseudo_csum = - swab16(~csum_ipv6_magic(&ipv6_hdr(skb)->saddr, - &ipv6_hdr(skb)->daddr, - 0, IPPROTO_TCP, 0)); + bswab16(~csum_ipv6_magic(&ipv6_hdr(skb)->saddr, + &ipv6_hdr(skb)->daddr, + 0, IPPROTO_TCP, 0)); + } - pbd->global_data |= ETH_TX_PARSE_BD_E1X_PSEUDO_CS_WITHOUT_LEN; + pbd->global_data |= + cpu_to_le16(ETH_TX_PARSE_BD_E1X_PSEUDO_CS_WITHOUT_LEN); +} + +/** + * bnx2x_set_pbd_csum_enc - update PBD with checksum and return header length + * + * @bp: driver handle + * @skb: packet skb + * @parsing_data: data to be updated + * @xmit_type: xmit flags + * + * 57712/578xx related, when skb has encapsulation + */ +static u8 bnx2x_set_pbd_csum_enc(struct bnx2x *bp, struct sk_buff *skb, + u32 *parsing_data, u32 xmit_type) +{ + *parsing_data |= + ((((u8 *)skb_inner_transport_header(skb) - skb->data) >> 1) << + ETH_TX_PARSE_BD_E2_L4_HDR_START_OFFSET_W_SHIFT) & + ETH_TX_PARSE_BD_E2_L4_HDR_START_OFFSET_W; + + if (xmit_type & XMIT_CSUM_TCP) { + *parsing_data |= ((inner_tcp_hdrlen(skb) / 4) << + ETH_TX_PARSE_BD_E2_TCP_HDR_LENGTH_DW_SHIFT) & + ETH_TX_PARSE_BD_E2_TCP_HDR_LENGTH_DW; + + return skb_inner_transport_header(skb) + + inner_tcp_hdrlen(skb) - skb->data; + } + + /* We support checksum offload for TCP and UDP only. + * No need to pass the UDP header length - it's a constant. + */ + return skb_inner_transport_header(skb) + + sizeof(struct udphdr) - skb->data; } /** @@ -2731,15 +3526,15 @@ static inline void bnx2x_set_pbd_gso(struct sk_buff *skb, * @parsing_data: data to be updated * @xmit_type: xmit flags * - * 57712 related + * 57712/578xx related */ -static inline u8 bnx2x_set_pbd_csum_e2(struct bnx2x *bp, struct sk_buff *skb, - u32 *parsing_data, u32 xmit_type) +static u8 bnx2x_set_pbd_csum_e2(struct bnx2x *bp, struct sk_buff *skb, + u32 *parsing_data, u32 xmit_type) { *parsing_data |= - ((((u8 *)skb_transport_header(skb) - skb->data) >> 1) << - ETH_TX_PARSE_BD_E2_TCP_HDR_START_OFFSET_W_SHIFT) & - ETH_TX_PARSE_BD_E2_TCP_HDR_START_OFFSET_W; + ((((u8 *)skb_transport_header(skb) - skb->data) >> 1) << + ETH_TX_PARSE_BD_E2_L4_HDR_START_OFFSET_W_SHIFT) & + ETH_TX_PARSE_BD_E2_L4_HDR_START_OFFSET_W; if (xmit_type & XMIT_CSUM_TCP) { *parsing_data |= ((tcp_hdrlen(skb) / 4) << @@ -2747,25 +3542,22 @@ static inline u8 bnx2x_set_pbd_csum_e2(struct bnx2x *bp, struct sk_buff *skb, ETH_TX_PARSE_BD_E2_TCP_HDR_LENGTH_DW; return skb_transport_header(skb) + tcp_hdrlen(skb) - skb->data; - } else - /* We support checksum offload for TCP and UDP only. - * No need to pass the UDP header length - it's a constant. - */ - return skb_transport_header(skb) + - sizeof(struct udphdr) - skb->data; + } + /* We support checksum offload for TCP and UDP only. + * No need to pass the UDP header length - it's a constant. + */ + return skb_transport_header(skb) + sizeof(struct udphdr) - skb->data; } -static inline void bnx2x_set_sbd_csum(struct bnx2x *bp, struct sk_buff *skb, - struct eth_tx_start_bd *tx_start_bd, u32 xmit_type) +/* set FW indication according to inner or outer protocols if tunneled */ +static void bnx2x_set_sbd_csum(struct bnx2x *bp, struct sk_buff *skb, + struct eth_tx_start_bd *tx_start_bd, + u32 xmit_type) { tx_start_bd->bd_flags.as_bitfield |= ETH_TX_BD_FLAGS_L4_CSUM; - if (xmit_type & XMIT_CSUM_V4) - tx_start_bd->bd_flags.as_bitfield |= - ETH_TX_BD_FLAGS_IP_CSUM; - else - tx_start_bd->bd_flags.as_bitfield |= - ETH_TX_BD_FLAGS_IPV6; + if (xmit_type & (XMIT_CSUM_ENC_V6 | XMIT_CSUM_V6)) + tx_start_bd->bd_flags.as_bitfield |= ETH_TX_BD_FLAGS_IPV6; if (!(xmit_type & XMIT_CSUM_TCP)) tx_start_bd->bd_flags.as_bitfield |= ETH_TX_BD_FLAGS_IS_UDP; @@ -2779,16 +3571,17 @@ static inline void bnx2x_set_sbd_csum(struct bnx2x *bp, struct sk_buff *skb, * @pbd: parse BD to be updated * @xmit_type: xmit flags */ -static inline u8 bnx2x_set_pbd_csum(struct bnx2x *bp, struct sk_buff *skb, - struct eth_tx_parse_bd_e1x *pbd, - u32 xmit_type) +static u8 bnx2x_set_pbd_csum(struct bnx2x *bp, struct sk_buff *skb, + struct eth_tx_parse_bd_e1x *pbd, + u32 xmit_type) { u8 hlen = (skb_network_header(skb) - skb->data) >> 1; /* for now NS flag is not used in Linux */ pbd->global_data = - (hlen | ((skb->protocol == cpu_to_be16(ETH_P_8021Q)) << - ETH_TX_PARSE_BD_E1X_LLC_SNAP_EN_SHIFT)); + cpu_to_le16(hlen | + ((skb->protocol == cpu_to_be16(ETH_P_8021Q)) << + ETH_TX_PARSE_BD_E1X_LLC_SNAP_EN_SHIFT)); pbd->ip_hlen_w = (skb_transport_header(skb) - skb_network_header(skb)) >> 1; @@ -2805,7 +3598,7 @@ static inline u8 bnx2x_set_pbd_csum(struct bnx2x *bp, struct sk_buff *skb, hlen = hlen*2; if (xmit_type & XMIT_CSUM_TCP) { - pbd->tcp_pseudo_csum = swab16(tcp_hdr(skb)->check); + pbd->tcp_pseudo_csum = bswab16(tcp_hdr(skb)->check); } else { s8 fix = SKB_CS_OFF(skb); /* signed! */ @@ -2826,6 +3619,77 @@ static inline u8 bnx2x_set_pbd_csum(struct bnx2x *bp, struct sk_buff *skb, return hlen; } +static void bnx2x_update_pbds_gso_enc(struct sk_buff *skb, + struct eth_tx_parse_bd_e2 *pbd_e2, + struct eth_tx_parse_2nd_bd *pbd2, + u16 *global_data, + u32 xmit_type) +{ + u16 hlen_w = 0; + u8 outerip_off, outerip_len = 0; + + /* from outer IP to transport */ + hlen_w = (skb_inner_transport_header(skb) - + skb_network_header(skb)) >> 1; + + /* transport len */ + hlen_w += inner_tcp_hdrlen(skb) >> 1; + + pbd2->fw_ip_hdr_to_payload_w = hlen_w; + + /* outer IP header info */ + if (xmit_type & XMIT_CSUM_V4) { + struct iphdr *iph = ip_hdr(skb); + u32 csum = (__force u32)(~iph->check) - + (__force u32)iph->tot_len - + (__force u32)iph->frag_off; + + pbd2->fw_ip_csum_wo_len_flags_frag = + bswab16(csum_fold((__force __wsum)csum)); + } else { + pbd2->fw_ip_hdr_to_payload_w = + hlen_w - ((sizeof(struct ipv6hdr)) >> 1); + } + + pbd2->tcp_send_seq = bswab32(inner_tcp_hdr(skb)->seq); + + pbd2->tcp_flags = pbd_tcp_flags(inner_tcp_hdr(skb)); + + if (xmit_type & XMIT_GSO_V4) { + pbd2->hw_ip_id = bswab16(inner_ip_hdr(skb)->id); + + pbd_e2->data.tunnel_data.pseudo_csum = + bswab16(~csum_tcpudp_magic( + inner_ip_hdr(skb)->saddr, + inner_ip_hdr(skb)->daddr, + 0, IPPROTO_TCP, 0)); + + outerip_len = ip_hdr(skb)->ihl << 1; + } else { + pbd_e2->data.tunnel_data.pseudo_csum = + bswab16(~csum_ipv6_magic( + &inner_ipv6_hdr(skb)->saddr, + &inner_ipv6_hdr(skb)->daddr, + 0, IPPROTO_TCP, 0)); + } + + outerip_off = (skb_network_header(skb) - skb->data) >> 1; + + *global_data |= + outerip_off | + (!!(xmit_type & XMIT_CSUM_V6) << + ETH_TX_PARSE_2ND_BD_IP_HDR_TYPE_OUTER_SHIFT) | + (outerip_len << + ETH_TX_PARSE_2ND_BD_IP_HDR_LEN_OUTER_W_SHIFT) | + ((skb->protocol == cpu_to_be16(ETH_P_8021Q)) << + ETH_TX_PARSE_2ND_BD_LLC_SNAP_EN_SHIFT); + + if (ip_hdr(skb)->protocol == IPPROTO_UDP) { + SET_FLAG(*global_data, ETH_TX_PARSE_2ND_BD_TUNNEL_UDP_EXIST, 1); + pbd2->tunnel_udp_hdr_start_w = skb_transport_offset(skb) >> 1; + } +} + /* called with netif_tx_lock * bnx2x_tx_int() runs without netif_tx_lock unless it needs to call * netif_wake_queue() @@ -2834,7 +3698,6 @@ netdev_tx_t bnx2x_start_xmit(struct sk_buff *skb, struct net_device *dev) { struct bnx2x *bp = netdev_priv(dev); - struct bnx2x_fastpath *fp; struct netdev_queue *txq; struct bnx2x_fp_txdata *txdata; struct sw_tx_bd *tx_buf; @@ -2842,9 +3705,10 @@ netdev_tx_t bnx2x_start_xmit(struct sk_buff *skb, struct net_device *dev) struct eth_tx_bd *tx_data_bd, *total_pkt_bd = NULL; struct eth_tx_parse_bd_e1x *pbd_e1x = NULL; struct eth_tx_parse_bd_e2 *pbd_e2 = NULL; + struct eth_tx_parse_2nd_bd *pbd2 = NULL; u32 pbd_e2_parsing_data = 0; u16 pkt_prod, bd_prod; - int nbd, txq_index, fp_index, txdata_index; + int nbd, txq_index; dma_addr_t mapping; u32 xmit_type = bnx2x_xmit_type(bp, skb); int i; @@ -2861,50 +3725,43 @@ netdev_tx_t bnx2x_start_xmit(struct sk_buff *skb, struct net_device *dev) txq_index = skb_get_queue_mapping(skb); txq = netdev_get_tx_queue(dev, txq_index); - BUG_ON(txq_index >= MAX_ETH_TXQ_IDX(bp) + FCOE_PRESENT); - - /* decode the fastpath index and the cos index from the txq */ - fp_index = TXQ_TO_FP(txq_index); - txdata_index = TXQ_TO_COS(txq_index); + BUG_ON(txq_index >= MAX_ETH_TXQ_IDX(bp) + (CNIC_LOADED(bp) ? 1 : 0)); -#ifdef BCM_CNIC - /* - * Override the above for the FCoE queue: - * - FCoE fp entry is right after the ETH entries. - * - FCoE L2 queue uses bp->txdata[0] only. - */ - if (unlikely(!NO_FCOE(bp) && (txq_index == - bnx2x_fcoe_tx(bp, txq_index)))) { - fp_index = FCOE_IDX; - txdata_index = 0; - } -#endif + txdata = &bp->bnx2x_txq[txq_index]; /* enable this debug print to view the transmission queue being used DP(NETIF_MSG_TX_QUEUED, "indices: txq %d, fp %d, txdata %d\n", txq_index, fp_index, txdata_index); */ - /* locate the fastpath and the txdata */ - fp = &bp->fp[fp_index]; - txdata = &fp->txdata[txdata_index]; - - /* enable this debug print to view the tranmission details + /* enable this debug print to view the transmission details DP(NETIF_MSG_TX_QUEUED, "transmitting packet cid %d fp index %d txdata_index %d tx_data ptr %p fp pointer %p\n", txdata->cid, fp_index, txdata_index, txdata, fp); */ if (unlikely(bnx2x_tx_avail(bp, txdata) < - (skb_shinfo(skb)->nr_frags + 3))) { - fp->eth_q_stats.driver_xoff++; + skb_shinfo(skb)->nr_frags + + BDS_PER_TX_PKT + + NEXT_CNT_PER_TX_PKT(MAX_BDS_PER_TX_PKT))) { + /* Handle special storage cases separately */ + if (txdata->tx_ring_size == 0) { + struct bnx2x_eth_q_stats *q_stats = + bnx2x_fp_qstats(bp, txdata->parent_fp); + q_stats->driver_filtered_tx_pkt++; + dev_kfree_skb(skb); + return NETDEV_TX_OK; + } + bnx2x_fp_qstats(bp, txdata->parent_fp)->driver_xoff++; netif_tx_stop_queue(txq); BNX2X_ERR("BUG! Tx ring full when queue awake!\n"); + return NETDEV_TX_BUSY; } DP(NETIF_MSG_TX_QUEUED, - "queue[%d]: SKB: summed %x protocol %x protocol(%x,%x) gso type %x xmit_type %x\n", + "queue[%d]: SKB: summed %x protocol %x protocol(%x,%x) gso type %x xmit_type %x len %d\n", txq_index, skb->ip_summed, skb->protocol, ipv6_hdr(skb)->nexthdr, - ip_hdr(skb)->protocol, skb_shinfo(skb)->gso_type, xmit_type); + ip_hdr(skb)->protocol, skb_shinfo(skb)->gso_type, xmit_type, + skb->len); eth = (struct ethhdr *)skb->data; @@ -2916,7 +3773,7 @@ netdev_tx_t bnx2x_start_xmit(struct sk_buff *skb, struct net_device *dev) mac_type = MULTICAST_ADDRESS; } -#if (MAX_SKB_FRAGS >= MAX_FETCH_BD - 3) +#if (MAX_SKB_FRAGS >= MAX_FETCH_BD - BDS_PER_TX_PKT) /* First, check if we need to linearize the skb (due to FW restrictions). No need to check fragmentation if page size > 8K (there will be no violation to FW restrictions) */ @@ -2964,11 +3821,9 @@ netdev_tx_t bnx2x_start_xmit(struct sk_buff *skb, struct net_device *dev) first_bd = tx_start_bd; tx_start_bd->bd_flags.as_bitfield = ETH_TX_BD_FLAGS_START_BD; - SET_FLAG(tx_start_bd->general_data, ETH_TX_START_BD_ETH_ADDR_TYPE, - mac_type); - /* header nbd */ - SET_FLAG(tx_start_bd->general_data, ETH_TX_START_BD_HDR_NBDS, 1); + /* header nbd: indirectly zero other flags! */ + tx_start_bd->general_data = 1 << ETH_TX_START_BD_HDR_NBDS_SHIFT; /* remember the first BD of the packet */ tx_buf->first_bd = txdata->tx_bd_prod; @@ -2984,8 +3839,19 @@ netdev_tx_t bnx2x_start_xmit(struct sk_buff *skb, struct net_device *dev) cpu_to_le16(vlan_tx_tag_get(skb)); tx_start_bd->bd_flags.as_bitfield |= (X_ETH_OUTBAND_VLAN << ETH_TX_BD_FLAGS_VLAN_MODE_SHIFT); - } else - tx_start_bd->vlan_or_ethertype = cpu_to_le16(pkt_prod); + } else { + /* when transmitting in a vf, start bd must hold the ethertype + * for fw to enforce it + */ + if (IS_VF(bp)) + tx_start_bd->vlan_or_ethertype = + cpu_to_le16(ntohs(eth->h_proto)); + else + /* used by FW for packet accounting */ + tx_start_bd->vlan_or_ethertype = cpu_to_le16(pkt_prod); + } + + nbd = 2; /* start_bd + pbd + frags (updated when pages are mapped) */ /* turn on parsing and get a BD */ bd_prod = TX_BD(NEXT_TX_IDX(bd_prod)); @@ -2996,45 +3862,96 @@ netdev_tx_t bnx2x_start_xmit(struct sk_buff *skb, struct net_device *dev) if (!CHIP_IS_E1x(bp)) { pbd_e2 = &txdata->tx_desc_ring[bd_prod].parse_bd_e2; memset(pbd_e2, 0, sizeof(struct eth_tx_parse_bd_e2)); - /* Set PBD in checksum offload case */ - if (xmit_type & XMIT_CSUM) + + if (xmit_type & XMIT_CSUM_ENC) { + u16 global_data = 0; + + /* Set PBD in enc checksum offload case */ + hlen = bnx2x_set_pbd_csum_enc(bp, skb, + &pbd_e2_parsing_data, + xmit_type); + + /* turn on 2nd parsing and get a BD */ + bd_prod = TX_BD(NEXT_TX_IDX(bd_prod)); + + pbd2 = &txdata->tx_desc_ring[bd_prod].parse_2nd_bd; + + memset(pbd2, 0, sizeof(*pbd2)); + + pbd_e2->data.tunnel_data.ip_hdr_start_inner_w = + (skb_inner_network_header(skb) - + skb->data) >> 1; + + if (xmit_type & XMIT_GSO_ENC) + bnx2x_update_pbds_gso_enc(skb, pbd_e2, pbd2, + &global_data, + xmit_type); + + pbd2->global_data = cpu_to_le16(global_data); + + /* add addition parse BD indication to start BD */ + SET_FLAG(tx_start_bd->general_data, + ETH_TX_START_BD_PARSE_NBDS, 1); + /* set encapsulation flag in start BD */ + SET_FLAG(tx_start_bd->general_data, + ETH_TX_START_BD_TUNNEL_EXIST, 1); + + tx_buf->flags |= BNX2X_HAS_SECOND_PBD; + + nbd++; + } else if (xmit_type & XMIT_CSUM) { + /* Set PBD in checksum offload case w/o encapsulation */ hlen = bnx2x_set_pbd_csum_e2(bp, skb, &pbd_e2_parsing_data, xmit_type); - if (IS_MF_SI(bp)) { - /* - * fill in the MAC addresses in the PBD - for local - * switching - */ - bnx2x_set_fw_mac_addr(&pbd_e2->src_mac_addr_hi, - &pbd_e2->src_mac_addr_mid, - &pbd_e2->src_mac_addr_lo, + } + + /* Add the macs to the parsing BD if this is a vf or if + * Tx Switching is enabled. + */ + if (IS_VF(bp)) { + /* override GRE parameters in BD */ + bnx2x_set_fw_mac_addr(&pbd_e2->data.mac_addr.src_hi, + &pbd_e2->data.mac_addr.src_mid, + &pbd_e2->data.mac_addr.src_lo, eth->h_source); - bnx2x_set_fw_mac_addr(&pbd_e2->dst_mac_addr_hi, - &pbd_e2->dst_mac_addr_mid, - &pbd_e2->dst_mac_addr_lo, + + bnx2x_set_fw_mac_addr(&pbd_e2->data.mac_addr.dst_hi, + &pbd_e2->data.mac_addr.dst_mid, + &pbd_e2->data.mac_addr.dst_lo, + eth->h_dest); + } else if (bp->flags & TX_SWITCHING) { + bnx2x_set_fw_mac_addr(&pbd_e2->data.mac_addr.dst_hi, + &pbd_e2->data.mac_addr.dst_mid, + &pbd_e2->data.mac_addr.dst_lo, eth->h_dest); } + + SET_FLAG(pbd_e2_parsing_data, + ETH_TX_PARSE_BD_E2_ETH_ADDR_TYPE, mac_type); } else { + u16 global_data = 0; pbd_e1x = &txdata->tx_desc_ring[bd_prod].parse_bd_e1x; memset(pbd_e1x, 0, sizeof(struct eth_tx_parse_bd_e1x)); /* Set PBD in checksum offload case */ if (xmit_type & XMIT_CSUM) hlen = bnx2x_set_pbd_csum(bp, skb, pbd_e1x, xmit_type); + SET_FLAG(global_data, + ETH_TX_PARSE_BD_E1X_ETH_ADDR_TYPE, mac_type); + pbd_e1x->global_data |= cpu_to_le16(global_data); } /* Setup the data pointer of the first BD of the packet */ tx_start_bd->addr_hi = cpu_to_le32(U64_HI(mapping)); tx_start_bd->addr_lo = cpu_to_le32(U64_LO(mapping)); - nbd = 2; /* start_bd + pbd + frags (updated when pages are mapped) */ tx_start_bd->nbytes = cpu_to_le16(skb_headlen(skb)); pkt_size = tx_start_bd->nbytes; DP(NETIF_MSG_TX_QUEUED, - "first bd @%p addr (%x:%x) nbd %d nbytes %d flags %x vlan %x\n", + "first bd @%p addr (%x:%x) nbytes %d flags %x vlan %x\n", tx_start_bd, tx_start_bd->addr_hi, tx_start_bd->addr_lo, - le16_to_cpu(tx_start_bd->nbd), le16_to_cpu(tx_start_bd->nbytes), + le16_to_cpu(tx_start_bd->nbytes), tx_start_bd->bd_flags.as_bitfield, le16_to_cpu(tx_start_bd->vlan_or_ethertype)); @@ -3047,15 +3964,17 @@ netdev_tx_t bnx2x_start_xmit(struct sk_buff *skb, struct net_device *dev) tx_start_bd->bd_flags.as_bitfield |= ETH_TX_BD_FLAGS_SW_LSO; - if (unlikely(skb_headlen(skb) > hlen)) + if (unlikely(skb_headlen(skb) > hlen)) { + nbd++; bd_prod = bnx2x_tx_split(bp, txdata, tx_buf, &tx_start_bd, hlen, - bd_prod, ++nbd); + bd_prod); + } if (!CHIP_IS_E1x(bp)) bnx2x_set_pbd_gso_e2(skb, &pbd_e2_parsing_data, xmit_type); else - bnx2x_set_pbd_gso(skb, pbd_e1x, xmit_type); + bnx2x_set_pbd_gso(skb, pbd_e1x, first_bd, xmit_type); } /* Set the PBD's parsing_data field if not zero @@ -3140,9 +4059,13 @@ netdev_tx_t bnx2x_start_xmit(struct sk_buff *skb, struct net_device *dev) if (pbd_e2) DP(NETIF_MSG_TX_QUEUED, "PBD (E2) @%p dst %x %x %x src %x %x %x parsing_data %x\n", - pbd_e2, pbd_e2->dst_mac_addr_hi, pbd_e2->dst_mac_addr_mid, - pbd_e2->dst_mac_addr_lo, pbd_e2->src_mac_addr_hi, - pbd_e2->src_mac_addr_mid, pbd_e2->src_mac_addr_lo, + pbd_e2, + pbd_e2->data.mac_addr.dst_hi, + pbd_e2->data.mac_addr.dst_mid, + pbd_e2->data.mac_addr.dst_lo, + pbd_e2->data.mac_addr.src_hi, + pbd_e2->data.mac_addr.src_mid, + pbd_e2->data.mac_addr.src_lo, pbd_e2->parsing_data); DP(NETIF_MSG_TX_QUEUED, "doorbell: nbd %d bd %u\n", nbd, bd_prod); @@ -3169,7 +4092,7 @@ netdev_tx_t bnx2x_start_xmit(struct sk_buff *skb, struct net_device *dev) txdata->tx_bd_prod += nbd; - if (unlikely(bnx2x_tx_avail(bp, txdata) < MAX_SKB_FRAGS + 4)) { + if (unlikely(bnx2x_tx_avail(bp, txdata) < MAX_DESC_PER_TX_PKT)) { netif_tx_stop_queue(txq); /* paired memory barrier is in bnx2x_tx_int(), we have to keep @@ -3177,8 +4100,8 @@ netdev_tx_t bnx2x_start_xmit(struct sk_buff *skb, struct net_device *dev) * fp->bd_tx_cons */ smp_mb(); - fp->eth_q_stats.driver_xoff++; - if (bnx2x_tx_avail(bp, txdata) >= MAX_SKB_FRAGS + 4) + bnx2x_fp_qstats(bp, txdata->parent_fp)->driver_xoff++; + if (bnx2x_tx_avail(bp, txdata) >= MAX_DESC_PER_TX_PKT) netif_tx_wake_queue(txq); } txdata->tx_pkt++; @@ -3202,7 +4125,7 @@ int bnx2x_setup_tc(struct net_device *dev, u8 num_tc) /* setup tc must be called under rtnl lock */ ASSERT_RTNL(); - /* no traffic classes requested. aborting */ + /* no traffic classes requested. Aborting */ if (!num_tc) { netdev_reset_tc(dev); return 0; @@ -3210,7 +4133,7 @@ int bnx2x_setup_tc(struct net_device *dev, u8 num_tc) /* requested to support too many traffic classes */ if (num_tc > bp->max_cos) { - BNX2X_ERR("support for too many traffic classes requested: %d. max supported is %d\n", + BNX2X_ERR("support for too many traffic classes requested: %d. Max supported is %d\n", num_tc, bp->max_cos); return -EINVAL; } @@ -3229,8 +4152,7 @@ int bnx2x_setup_tc(struct net_device *dev, u8 num_tc) prio, bp->prio_to_cos[prio]); } - - /* Use this configuration to diffrentiate tc0 from other COSes + /* Use this configuration to differentiate tc0 from other COSes This can be used for ets or pfc, and save the effort of setting up a multio class queue disc or negotiating DCBX with a switch netdev_set_prio_tc_map(dev, 0, 0); @@ -3243,7 +4165,7 @@ int bnx2x_setup_tc(struct net_device *dev, u8 num_tc) /* configure traffic class to transmission queue mapping */ for (cos = 0; cos < bp->max_cos; cos++) { count = BNX2X_NUM_ETH_QUEUES(bp); - offset = cos * MAX_TXQS_PER_COS; + offset = cos * BNX2X_NUM_NON_CNIC_QUEUES(bp); netdev_set_tc_queue(dev, cos, count, offset); DP(BNX2X_MSG_SP | NETIF_MSG_IFUP, "mapping tc %d to offset %d count %d\n", @@ -3265,13 +4187,11 @@ int bnx2x_change_mac_addr(struct net_device *dev, void *p) return -EINVAL; } -#ifdef BCM_CNIC if ((IS_MF_STORAGE_SD(bp) || IS_MF_FCOE_AFEX(bp)) && !is_zero_ether_addr(addr->sa_data)) { BNX2X_ERR("Can't configure non-zero address on iSCSI or FCoE functions in MF-SD mode\n"); return -EINVAL; } -#endif if (netif_running(dev)) { rc = bnx2x_set_eth_mac(bp, false); @@ -3279,7 +4199,6 @@ int bnx2x_change_mac_addr(struct net_device *dev, void *p) return rc; } - dev->addr_assign_type &= ~NET_ADDR_RANDOM; memcpy(dev->dev_addr, addr->sa_data, dev->addr_len); if (netif_running(dev)) @@ -3295,13 +4214,11 @@ static void bnx2x_free_fp_mem_at(struct bnx2x *bp, int fp_index) u8 cos; /* Common */ -#ifdef BCM_CNIC + if (IS_FCOE_IDX(fp_index)) { memset(sb, 0, sizeof(union host_hc_status_block)); fp->status_blk_mapping = 0; - } else { -#endif /* status blocks */ if (!CHIP_IS_E1x(bp)) BNX2X_PCI_FREE(sb->e2_sb, @@ -3313,9 +4230,8 @@ static void bnx2x_free_fp_mem_at(struct bnx2x *bp, int fp_index) bnx2x_fp(bp, fp_index, status_blk_mapping), sizeof(struct host_hc_status_block_e1x)); -#ifdef BCM_CNIC } -#endif + /* Rx */ if (!skip_rx_queue(bp, fp_index)) { bnx2x_free_rx_bds(fp); @@ -3342,7 +4258,7 @@ static void bnx2x_free_fp_mem_at(struct bnx2x *bp, int fp_index) if (!skip_tx_queue(bp, fp_index)) { /* fastpath tx rings: tx_buf tx_desc */ for_each_cos_in_tx_queue(fp, cos) { - struct bnx2x_fp_txdata *txdata = &fp->txdata[cos]; + struct bnx2x_fp_txdata *txdata = fp->txdata_ptr[cos]; DP(NETIF_MSG_IFDOWN, "freeing tx memory of fp %d cos %d cid %d\n", @@ -3357,10 +4273,17 @@ static void bnx2x_free_fp_mem_at(struct bnx2x *bp, int fp_index) /* end of fastpath */ } +static void bnx2x_free_fp_mem_cnic(struct bnx2x *bp) +{ + int i; + for_each_cnic_queue(bp, i) + bnx2x_free_fp_mem_at(bp, i); +} + void bnx2x_free_fp_mem(struct bnx2x *bp) { int i; - for_each_queue(bp, i) + for_each_eth_queue(bp, i) bnx2x_free_fp_mem_at(bp, i); } @@ -3395,7 +4318,7 @@ static int bnx2x_alloc_rx_bds(struct bnx2x_fastpath *fp, * fp->eth_q_stats.rx_skb_alloc_failed = 0 */ for (i = 0; i < rx_ring_size; i++) { - if (bnx2x_alloc_rx_data(bp, fp, ring_prod) < 0) { + if (bnx2x_alloc_rx_data(bp, fp, ring_prod, GFP_KERNEL) < 0) { failure_cnt++; continue; } @@ -3414,7 +4337,7 @@ static int bnx2x_alloc_rx_bds(struct bnx2x_fastpath *fp, cqe_ring_prod); fp->rx_pkt = fp->rx_calls = 0; - fp->eth_q_stats.rx_skb_alloc_failed += failure_cnt; + bnx2x_fp_stats(bp, fp)->eth_q_stats.rx_skb_alloc_failed += failure_cnt; return i - failure_cnt; } @@ -3445,23 +4368,23 @@ static int bnx2x_alloc_fp_mem_at(struct bnx2x *bp, int index) u8 cos; int rx_ring_size = 0; -#ifdef BCM_CNIC if (!bp->rx_ring_size && (IS_MF_STORAGE_SD(bp) || IS_MF_FCOE_AFEX(bp))) { rx_ring_size = MIN_RX_SIZE_NONTPA; bp->rx_ring_size = rx_ring_size; - } else -#endif - if (!bp->rx_ring_size) { - u32 cfg = SHMEM_RD(bp, - dev_info.port_hw_config[BP_PORT(bp)].default_cfg); - + } else if (!bp->rx_ring_size) { rx_ring_size = MAX_RX_AVAIL/BNX2X_NUM_RX_QUEUES(bp); - /* Dercease ring size for 1G functions */ - if ((cfg & PORT_HW_CFG_NET_SERDES_IF_MASK) == - PORT_HW_CFG_NET_SERDES_IF_SGMII) - rx_ring_size /= 10; + if (CHIP_IS_E3(bp)) { + u32 cfg = SHMEM_RD(bp, + dev_info.port_hw_config[BP_PORT(bp)]. + default_cfg); + + /* Decrease ring size for 1G functions */ + if ((cfg & PORT_HW_CFG_NET_SERDES_IF_MASK) == + PORT_HW_CFG_NET_SERDES_IF_SGMII) + rx_ring_size /= 10; + } /* allocate at least number of buffers required by FW */ rx_ring_size = max_t(int, bp->disable_tpa ? MIN_RX_SIZE_NONTPA : @@ -3471,23 +4394,25 @@ static int bnx2x_alloc_fp_mem_at(struct bnx2x *bp, int index) } else /* if rx_ring_size specified - use it */ rx_ring_size = bp->rx_ring_size; + DP(BNX2X_MSG_SP, "calculated rx_ring_size %d\n", rx_ring_size); + /* Common */ sb = &bnx2x_fp(bp, index, status_blk); -#ifdef BCM_CNIC + if (!IS_FCOE_IDX(index)) { -#endif /* status blocks */ - if (!CHIP_IS_E1x(bp)) - BNX2X_PCI_ALLOC(sb->e2_sb, - &bnx2x_fp(bp, index, status_blk_mapping), - sizeof(struct host_hc_status_block_e2)); - else - BNX2X_PCI_ALLOC(sb->e1x_sb, - &bnx2x_fp(bp, index, status_blk_mapping), - sizeof(struct host_hc_status_block_e1x)); -#ifdef BCM_CNIC + if (!CHIP_IS_E1x(bp)) { + sb->e2_sb = BNX2X_PCI_ALLOC(&bnx2x_fp(bp, index, status_blk_mapping), + sizeof(struct host_hc_status_block_e2)); + if (!sb->e2_sb) + goto alloc_mem_err; + } else { + sb->e1x_sb = BNX2X_PCI_ALLOC(&bnx2x_fp(bp, index, status_blk_mapping), + sizeof(struct host_hc_status_block_e1x)); + if (!sb->e1x_sb) + goto alloc_mem_err; + } } -#endif /* FCoE Queue uses Default SB and doesn't ACK the SB, thus no need to * set shortcuts for it. @@ -3499,40 +4424,55 @@ static int bnx2x_alloc_fp_mem_at(struct bnx2x *bp, int index) if (!skip_tx_queue(bp, index)) { /* fastpath tx rings: tx_buf tx_desc */ for_each_cos_in_tx_queue(fp, cos) { - struct bnx2x_fp_txdata *txdata = &fp->txdata[cos]; + struct bnx2x_fp_txdata *txdata = fp->txdata_ptr[cos]; DP(NETIF_MSG_IFUP, "allocating tx memory of fp %d cos %d\n", index, cos); - BNX2X_ALLOC(txdata->tx_buf_ring, - sizeof(struct sw_tx_bd) * NUM_TX_BD); - BNX2X_PCI_ALLOC(txdata->tx_desc_ring, - &txdata->tx_desc_mapping, - sizeof(union eth_tx_bd_types) * NUM_TX_BD); + txdata->tx_buf_ring = kcalloc(NUM_TX_BD, + sizeof(struct sw_tx_bd), + GFP_KERNEL); + if (!txdata->tx_buf_ring) + goto alloc_mem_err; + txdata->tx_desc_ring = BNX2X_PCI_ALLOC(&txdata->tx_desc_mapping, + sizeof(union eth_tx_bd_types) * NUM_TX_BD); + if (!txdata->tx_desc_ring) + goto alloc_mem_err; } } /* Rx */ if (!skip_rx_queue(bp, index)) { /* fastpath rx rings: rx_buf rx_desc rx_comp */ - BNX2X_ALLOC(bnx2x_fp(bp, index, rx_buf_ring), - sizeof(struct sw_rx_bd) * NUM_RX_BD); - BNX2X_PCI_ALLOC(bnx2x_fp(bp, index, rx_desc_ring), - &bnx2x_fp(bp, index, rx_desc_mapping), - sizeof(struct eth_rx_bd) * NUM_RX_BD); + bnx2x_fp(bp, index, rx_buf_ring) = + kcalloc(NUM_RX_BD, sizeof(struct sw_rx_bd), GFP_KERNEL); + if (!bnx2x_fp(bp, index, rx_buf_ring)) + goto alloc_mem_err; + bnx2x_fp(bp, index, rx_desc_ring) = + BNX2X_PCI_ALLOC(&bnx2x_fp(bp, index, rx_desc_mapping), + sizeof(struct eth_rx_bd) * NUM_RX_BD); + if (!bnx2x_fp(bp, index, rx_desc_ring)) + goto alloc_mem_err; - BNX2X_PCI_ALLOC(bnx2x_fp(bp, index, rx_comp_ring), - &bnx2x_fp(bp, index, rx_comp_mapping), - sizeof(struct eth_fast_path_rx_cqe) * - NUM_RCQ_BD); + /* Seed all CQEs by 1s */ + bnx2x_fp(bp, index, rx_comp_ring) = + BNX2X_PCI_FALLOC(&bnx2x_fp(bp, index, rx_comp_mapping), + sizeof(struct eth_fast_path_rx_cqe) * NUM_RCQ_BD); + if (!bnx2x_fp(bp, index, rx_comp_ring)) + goto alloc_mem_err; /* SGE ring */ - BNX2X_ALLOC(bnx2x_fp(bp, index, rx_page_ring), - sizeof(struct sw_rx_page) * NUM_RX_SGE); - BNX2X_PCI_ALLOC(bnx2x_fp(bp, index, rx_sge_ring), - &bnx2x_fp(bp, index, rx_sge_mapping), - BCM_PAGE_SIZE * NUM_RX_SGE_PAGES); + bnx2x_fp(bp, index, rx_page_ring) = + kcalloc(NUM_RX_SGE, sizeof(struct sw_rx_page), + GFP_KERNEL); + if (!bnx2x_fp(bp, index, rx_page_ring)) + goto alloc_mem_err; + bnx2x_fp(bp, index, rx_sge_ring) = + BNX2X_PCI_ALLOC(&bnx2x_fp(bp, index, rx_sge_mapping), + BCM_PAGE_SIZE * NUM_RX_SGE_PAGES); + if (!bnx2x_fp(bp, index, rx_sge_ring)) + goto alloc_mem_err; /* RX BD ring */ bnx2x_set_next_page_rx_bd(fp); @@ -3564,31 +4504,31 @@ alloc_mem_err: return 0; } -int bnx2x_alloc_fp_mem(struct bnx2x *bp) +static int bnx2x_alloc_fp_mem_cnic(struct bnx2x *bp) +{ + if (!NO_FCOE(bp)) + /* FCoE */ + if (bnx2x_alloc_fp_mem_at(bp, FCOE_IDX(bp))) + /* we will fail load process instead of mark + * NO_FCOE_FLAG + */ + return -ENOMEM; + + return 0; +} + +static int bnx2x_alloc_fp_mem(struct bnx2x *bp) { int i; - /** - * 1. Allocate FP for leading - fatal if error - * 2. {CNIC} Allocate FCoE FP - fatal if error - * 3. {CNIC} Allocate OOO + FWD - disable OOO if error - * 4. Allocate RSS - fix number of queues if error + /* 1. Allocate FP for leading - fatal if error + * 2. Allocate RSS - fix number of queues if error */ /* leading */ if (bnx2x_alloc_fp_mem_at(bp, 0)) return -ENOMEM; -#ifdef BCM_CNIC - if (!NO_FCOE(bp)) - /* FCoE */ - if (bnx2x_alloc_fp_mem_at(bp, FCOE_IDX)) - /* we will fail load process instead of mark - * NO_FCOE_FLAG - */ - return -ENOMEM; -#endif - /* RSS */ for_each_nondefault_eth_queue(bp, i) if (bnx2x_alloc_fp_mem_at(bp, i)) @@ -3599,17 +4539,18 @@ int bnx2x_alloc_fp_mem(struct bnx2x *bp) int delta = BNX2X_NUM_ETH_QUEUES(bp) - i; WARN_ON(delta < 0); -#ifdef BCM_CNIC - /** - * move non eth FPs next to last eth FP - * must be done in that order - * FCOE_IDX < FWD_IDX < OOO_IDX - */ + bnx2x_shrink_eth_fp(bp, delta); + if (CNIC_SUPPORT(bp)) + /* move non eth FPs next to last eth FP + * must be done in that order + * FCOE_IDX < FWD_IDX < OOO_IDX + */ - /* move FCoE fp even NO_FCOE_FLAG is on */ - bnx2x_move_fp(bp, FCOE_IDX, FCOE_IDX - delta); -#endif - bp->num_queues -= delta; + /* move FCoE fp even NO_FCOE_FLAG is on */ + bnx2x_move_fp(bp, FCOE_IDX(bp), FCOE_IDX(bp) - delta); + bp->num_ethernet_queues -= delta; + bp->num_queues = bp->num_ethernet_queues + + bp->num_cnic_queues; BNX2X_ERR("Adjusted num of queues from %d to %d\n", bp->num_queues + delta, bp->num_queues); } @@ -3619,31 +4560,76 @@ int bnx2x_alloc_fp_mem(struct bnx2x *bp) void bnx2x_free_mem_bp(struct bnx2x *bp) { + int i; + + for (i = 0; i < bp->fp_array_size; i++) + kfree(bp->fp[i].tpa_info); kfree(bp->fp); + kfree(bp->sp_objs); + kfree(bp->fp_stats); + kfree(bp->bnx2x_txq); kfree(bp->msix_table); kfree(bp->ilt); } -int __devinit bnx2x_alloc_mem_bp(struct bnx2x *bp) +int bnx2x_alloc_mem_bp(struct bnx2x *bp) { struct bnx2x_fastpath *fp; struct msix_entry *tbl; struct bnx2x_ilt *ilt; int msix_table_size = 0; + int fp_array_size, txq_array_size; + int i; /* * The biggest MSI-X table we might need is as a maximum number of fast - * path IGU SBs plus default SB (for PF). + * path IGU SBs plus default SB (for PF only). */ - msix_table_size = bp->igu_sb_cnt + 1; + msix_table_size = bp->igu_sb_cnt; + if (IS_PF(bp)) + msix_table_size++; + BNX2X_DEV_INFO("msix_table_size %d\n", msix_table_size); /* fp array: RSS plus CNIC related L2 queues */ - fp = kcalloc(BNX2X_MAX_RSS_COUNT(bp) + NON_ETH_CONTEXT_USE, - sizeof(*fp), GFP_KERNEL); + fp_array_size = BNX2X_MAX_RSS_COUNT(bp) + CNIC_SUPPORT(bp); + bp->fp_array_size = fp_array_size; + BNX2X_DEV_INFO("fp_array_size %d\n", bp->fp_array_size); + + fp = kcalloc(bp->fp_array_size, sizeof(*fp), GFP_KERNEL); if (!fp) goto alloc_err; + for (i = 0; i < bp->fp_array_size; i++) { + fp[i].tpa_info = + kcalloc(ETH_MAX_AGGREGATION_QUEUES_E1H_E2, + sizeof(struct bnx2x_agg_info), GFP_KERNEL); + if (!(fp[i].tpa_info)) + goto alloc_err; + } + bp->fp = fp; + /* allocate sp objs */ + bp->sp_objs = kcalloc(bp->fp_array_size, sizeof(struct bnx2x_sp_objs), + GFP_KERNEL); + if (!bp->sp_objs) + goto alloc_err; + + /* allocate fp_stats */ + bp->fp_stats = kcalloc(bp->fp_array_size, sizeof(struct bnx2x_fp_stats), + GFP_KERNEL); + if (!bp->fp_stats) + goto alloc_err; + + /* Allocate memory for the transmission queues array */ + txq_array_size = + BNX2X_MAX_RSS_COUNT(bp) * BNX2X_MULTI_TX_COS + CNIC_SUPPORT(bp); + BNX2X_DEV_INFO("txq_array_size %d", txq_array_size); + + bp->bnx2x_txq = kcalloc(txq_array_size, sizeof(struct bnx2x_fp_txdata), + GFP_KERNEL); + if (!bp->bnx2x_txq) + goto alloc_err; + /* msix table */ tbl = kcalloc(msix_table_size, sizeof(*tbl), GFP_KERNEL); if (!tbl) @@ -3660,7 +4646,6 @@ int __devinit bnx2x_alloc_mem_bp(struct bnx2x *bp) alloc_err: bnx2x_free_mem_bp(bp); return -ENOMEM; - } int bnx2x_reload_if_running(struct net_device *dev) @@ -3670,7 +4655,7 @@ int bnx2x_reload_if_running(struct net_device *dev) if (unlikely(!netif_running(dev))) return 0; - bnx2x_nic_unload(bp, UNLOAD_NORMAL); + bnx2x_nic_unload(bp, UNLOAD_NORMAL, true); return bnx2x_nic_load(bp, LOAD_NORMAL); } @@ -3702,13 +4687,12 @@ int bnx2x_get_cur_phy_idx(struct bnx2x *bp) } return sel_phy_idx; - } int bnx2x_get_link_cfg_idx(struct bnx2x *bp) { u32 sel_phy_idx = bnx2x_get_cur_phy_idx(bp); /* - * The selected actived PHY is always after swapping (in case PHY + * The selected activated PHY is always after swapping (in case PHY * swapping is enabled). So when swapping is enabled, we need to reverse * the configuration */ @@ -3723,7 +4707,7 @@ int bnx2x_get_link_cfg_idx(struct bnx2x *bp) return LINK_CONFIG_IDX(sel_phy_idx); } -#if defined(NETDEV_FCOE_WWNN) && defined(BCM_CNIC) +#ifdef NETDEV_FCOE_WWNN int bnx2x_fcoe_get_wwn(struct net_device *dev, u64 *wwn, int type) { struct bnx2x *bp = netdev_priv(dev); @@ -3790,6 +4774,7 @@ int bnx2x_set_features(struct net_device *dev, netdev_features_t features) { struct bnx2x *bp = netdev_priv(dev); u32 flags = bp->flags; + u32 changes; bool bnx2x_reload = false; if (features & NETIF_F_LRO) @@ -3814,10 +4799,16 @@ int bnx2x_set_features(struct net_device *dev, netdev_features_t features) } } - if (flags ^ bp->flags) { - bp->flags = flags; + changes = flags ^ bp->flags; + + /* if GRO is changed while LRO is enabled, don't force a reload */ + if ((changes & GRO_ENABLE_FLAG) && (flags & TPA_ENABLE_FLAG)) + changes &= ~GRO_ENABLE_FLAG; + + if (changes) bnx2x_reload = true; - } + + bp->flags = flags; if (bnx2x_reload) { if (bp->recovery_state == BNX2X_RECOVERY_DONE) @@ -3837,12 +4828,8 @@ void bnx2x_tx_timeout(struct net_device *dev) bnx2x_panic(); #endif - smp_mb__before_clear_bit(); - set_bit(BNX2X_SP_RTNL_TX_TIMEOUT, &bp->sp_rtnl_state); - smp_mb__after_clear_bit(); - /* This allows the netif to be shutdown gracefully before resetting */ - schedule_delayed_work(&bp->sp_rtnl_task, 0); + bnx2x_schedule_sp_rtnl(bp, BNX2X_SP_RTNL_TX_TIMEOUT, 0); } int bnx2x_suspend(struct pci_dev *pdev, pm_message_t state) @@ -3867,7 +4854,7 @@ int bnx2x_suspend(struct pci_dev *pdev, pm_message_t state) netif_device_detach(dev); - bnx2x_nic_unload(bp, UNLOAD_CLOSE); + bnx2x_nic_unload(bp, UNLOAD_CLOSE, false); bnx2x_set_power_state(bp, pci_choose_state(pdev, state)); @@ -3912,10 +4899,14 @@ int bnx2x_resume(struct pci_dev *pdev) return rc; } - void bnx2x_set_ctx_validation(struct bnx2x *bp, struct eth_context *cxt, u32 cid) { + if (!cxt) { + BNX2X_ERR("bad context pointer %p\n", cxt); + return; + } + /* ustorm cxt validation */ cxt->ustorm_ag_context.cdu_usage = CDU_RSRVD_VALUE_TYPE_A(HW_CID(bp, cid), @@ -3930,7 +4921,6 @@ static void storm_memset_hc_timeout(struct bnx2x *bp, u8 port, u8 fw_sb_id, u8 sb_index, u8 ticks) { - u32 addr = BAR_CSTRORM_INTMEM + CSTORM_STATUS_BLOCK_DATA_TIMEOUT_OFFSET(fw_sb_id, sb_index); REG_WR8(bp, addr, ticks); @@ -3946,11 +4936,11 @@ static void storm_memset_hc_disable(struct bnx2x *bp, u8 port, u32 enable_flag = disable ? 0 : (1 << HC_INDEX_DATA_HC_ENABLED_SHIFT); u32 addr = BAR_CSTRORM_INTMEM + CSTORM_STATUS_BLOCK_DATA_FLAGS_OFFSET(fw_sb_id, sb_index); - u16 flags = REG_RD16(bp, addr); + u8 flags = REG_RD8(bp, addr); /* clear and set */ flags &= ~HC_INDEX_DATA_HC_ENABLED; flags |= enable_flag; - REG_WR16(bp, addr, flags); + REG_WR8(bp, addr, flags); DP(NETIF_MSG_IFUP, "port %x fw_sb_id %d sb_index %d disable %d\n", port, fw_sb_id, sb_index, disable); @@ -3967,3 +4957,15 @@ void bnx2x_update_coalesce_sb_index(struct bnx2x *bp, u8 fw_sb_id, disable = disable ? 1 : (usec ? 0 : 1); storm_memset_hc_disable(bp, port, fw_sb_id, sb_index, disable); } + +void bnx2x_schedule_sp_rtnl(struct bnx2x *bp, enum sp_rtnl_flag flag, + u32 verbose) +{ + smp_mb__before_atomic(); + set_bit(flag, &bp->sp_rtnl_state); + smp_mb__after_atomic(); + DP((BNX2X_MSG_SP | verbose), "Scheduling sp_rtnl task [Flag: %d]\n", + flag); + schedule_delayed_work(&bp->sp_rtnl_task, 0); +} +EXPORT_SYMBOL(bnx2x_schedule_sp_rtnl); |