diff options
Diffstat (limited to 'drivers/net/ethernet/broadcom/bnx2x')
18 files changed, 1105 insertions, 721 deletions
diff --git a/drivers/net/ethernet/broadcom/bnx2x/bnx2x.h b/drivers/net/ethernet/broadcom/bnx2x/bnx2x.h index 3dba2a70a00..dedbd76c033 100644 --- a/drivers/net/ethernet/broadcom/bnx2x/bnx2x.h +++ b/drivers/net/ethernet/broadcom/bnx2x/bnx2x.h @@ -34,12 +34,10 @@ #define BCM_DCBNL #endif - #include "bnx2x_hsi.h" #include "../cnic_if.h" - #define BNX2X_MIN_MSIX_VEC_CNT(bp) ((bp)->min_msix_vec_cnt) #include <linux/mdio.h> @@ -114,7 +112,6 @@ do { \ #define BNX2X_ERROR(fmt, ...) \ pr_err("[%s:%d]" fmt, __func__, __LINE__, ##__VA_ARGS__) - /* before we have a dev->name use dev_info() */ #define BNX2X_DEV_INFO(fmt, ...) \ do { \ @@ -147,7 +144,6 @@ do { \ #define U64_HI(x) ((u32)(((u64)(x)) >> 32)) #define HILO_U64(hi, lo) ((((u64)(hi)) << 32) + (lo)) - #define REG_ADDR(bp, offset) ((bp->regview) + (offset)) #define REG_RD(bp, offset) readl(REG_ADDR(bp, offset)) @@ -366,7 +362,7 @@ union db_prod { /* * Number of required SGEs is the sum of two: * 1. Number of possible opened aggregations (next packet for - * these aggregations will probably consume SGE immidiatelly) + * these aggregations will probably consume SGE immediately) * 2. Rest of BRB blocks divided by 2 (block will consume new SGE only * after placement on BD for new TPA aggregation) * @@ -387,7 +383,6 @@ union db_prod { #define BIT_VEC64_ELEM_SHIFT 6 #define BIT_VEC64_ELEM_MASK ((u64)BIT_VEC64_ELEM_SZ - 1) - #define __BIT_VEC64_SET_BIT(el, bit) \ do { \ el = ((el) | ((u64)0x1 << (bit))); \ @@ -398,7 +393,6 @@ union db_prod { el = ((el) & (~((u64)0x1 << (bit)))); \ } while (0) - #define BIT_VEC64_SET_BIT(vec64, idx) \ __BIT_VEC64_SET_BIT((vec64)[(idx) >> BIT_VEC64_ELEM_SHIFT], \ (idx) & BIT_VEC64_ELEM_MASK) @@ -419,8 +413,6 @@ union db_prod { /*******************************************************/ - - /* Number of u64 elements in SGE mask array */ #define RX_SGE_MASK_LEN (NUM_RX_SGE / BIT_VEC64_ELEM_SZ) #define RX_SGE_MASK_LEN_MASK (RX_SGE_MASK_LEN - 1) @@ -493,11 +485,26 @@ struct bnx2x_fastpath { struct bnx2x *bp; /* parent */ struct napi_struct napi; + +#ifdef CONFIG_NET_LL_RX_POLL + unsigned int state; +#define BNX2X_FP_STATE_IDLE 0 +#define BNX2X_FP_STATE_NAPI (1 << 0) /* NAPI owns this FP */ +#define BNX2X_FP_STATE_POLL (1 << 1) /* poll owns this FP */ +#define BNX2X_FP_STATE_NAPI_YIELD (1 << 2) /* NAPI yielded this FP */ +#define BNX2X_FP_STATE_POLL_YIELD (1 << 3) /* poll yielded this FP */ +#define BNX2X_FP_YIELD (BNX2X_FP_STATE_NAPI_YIELD | BNX2X_FP_STATE_POLL_YIELD) +#define BNX2X_FP_LOCKED (BNX2X_FP_STATE_NAPI | BNX2X_FP_STATE_POLL) +#define BNX2X_FP_USER_PEND (BNX2X_FP_STATE_POLL | BNX2X_FP_STATE_POLL_YIELD) + /* protect state */ + spinlock_t lock; +#endif /* CONFIG_NET_LL_RX_POLL */ + union host_hc_status_block status_blk; - /* chip independed shortcuts into sb structure */ + /* chip independent shortcuts into sb structure */ __le16 *sb_index_values; __le16 *sb_running_index; - /* chip independed shortcut into rx_prods_offset memory */ + /* chip independent shortcut into rx_prods_offset memory */ u32 ustorm_rx_prods_offset; u32 rx_buf_size; @@ -565,6 +572,116 @@ struct bnx2x_fastpath { #define bnx2x_fp_stats(bp, fp) (&((bp)->fp_stats[(fp)->index])) #define bnx2x_fp_qstats(bp, fp) (&((bp)->fp_stats[(fp)->index].eth_q_stats)) +#ifdef CONFIG_NET_LL_RX_POLL +static inline void bnx2x_fp_init_lock(struct bnx2x_fastpath *fp) +{ + spin_lock_init(&fp->lock); + fp->state = BNX2X_FP_STATE_IDLE; +} + +/* called from the device poll routine to get ownership of a FP */ +static inline bool bnx2x_fp_lock_napi(struct bnx2x_fastpath *fp) +{ + bool rc = true; + + spin_lock(&fp->lock); + if (fp->state & BNX2X_FP_LOCKED) { + WARN_ON(fp->state & BNX2X_FP_STATE_NAPI); + fp->state |= BNX2X_FP_STATE_NAPI_YIELD; + rc = false; + } else { + /* we don't care if someone yielded */ + fp->state = BNX2X_FP_STATE_NAPI; + } + spin_unlock(&fp->lock); + return rc; +} + +/* returns true is someone tried to get the FP while napi had it */ +static inline bool bnx2x_fp_unlock_napi(struct bnx2x_fastpath *fp) +{ + bool rc = false; + + spin_lock(&fp->lock); + WARN_ON(fp->state & + (BNX2X_FP_STATE_POLL | BNX2X_FP_STATE_NAPI_YIELD)); + + if (fp->state & BNX2X_FP_STATE_POLL_YIELD) + rc = true; + fp->state = BNX2X_FP_STATE_IDLE; + spin_unlock(&fp->lock); + return rc; +} + +/* called from bnx2x_low_latency_poll() */ +static inline bool bnx2x_fp_lock_poll(struct bnx2x_fastpath *fp) +{ + bool rc = true; + + spin_lock_bh(&fp->lock); + if ((fp->state & BNX2X_FP_LOCKED)) { + fp->state |= BNX2X_FP_STATE_POLL_YIELD; + rc = false; + } else { + /* preserve yield marks */ + fp->state |= BNX2X_FP_STATE_POLL; + } + spin_unlock_bh(&fp->lock); + return rc; +} + +/* returns true if someone tried to get the FP while it was locked */ +static inline bool bnx2x_fp_unlock_poll(struct bnx2x_fastpath *fp) +{ + bool rc = false; + + spin_lock_bh(&fp->lock); + WARN_ON(fp->state & BNX2X_FP_STATE_NAPI); + + if (fp->state & BNX2X_FP_STATE_POLL_YIELD) + rc = true; + fp->state = BNX2X_FP_STATE_IDLE; + spin_unlock_bh(&fp->lock); + return rc; +} + +/* true if a socket is polling, even if it did not get the lock */ +static inline bool bnx2x_fp_ll_polling(struct bnx2x_fastpath *fp) +{ + WARN_ON(!(fp->state & BNX2X_FP_LOCKED)); + return fp->state & BNX2X_FP_USER_PEND; +} +#else +static inline void bnx2x_fp_init_lock(struct bnx2x_fastpath *fp) +{ +} + +static inline bool bnx2x_fp_lock_napi(struct bnx2x_fastpath *fp) +{ + return true; +} + +static inline bool bnx2x_fp_unlock_napi(struct bnx2x_fastpath *fp) +{ + return false; +} + +static inline bool bnx2x_fp_lock_poll(struct bnx2x_fastpath *fp) +{ + return false; +} + +static inline bool bnx2x_fp_unlock_poll(struct bnx2x_fastpath *fp) +{ + return false; +} + +static inline bool bnx2x_fp_ll_polling(struct bnx2x_fastpath *fp) +{ + return false; +} +#endif /* CONFIG_NET_LL_RX_POLL */ + /* Use 2500 as a mini-jumbo MTU for FCoE */ #define BNX2X_FCOE_MINI_JUMBO_MTU 2500 @@ -580,12 +697,10 @@ struct bnx2x_fastpath { txdata_ptr[FIRST_TX_COS_INDEX] \ ->var) - #define IS_ETH_FP(fp) ((fp)->index < BNX2X_NUM_ETH_QUEUES((fp)->bp)) #define IS_FCOE_FP(fp) ((fp)->index == FCOE_IDX((fp)->bp)) #define IS_FCOE_IDX(idx) ((idx) == FCOE_IDX(bp)) - /* MC hsi */ #define MAX_FETCH_BD 13 /* HW max BDs per packet */ #define RX_COPY_THRESH 92 @@ -613,7 +728,7 @@ struct bnx2x_fastpath { * START_BD(splitted) - includes unpaged data segment for GSO * PARSING_BD - for TSO and CSUM data * PARSING_BD2 - for encapsulation data - * Frag BDs - decribes pages for frags + * Frag BDs - describes pages for frags */ #define BDS_PER_TX_PKT 4 #define MAX_BDS_PER_TX_PKT (MAX_SKB_FRAGS + BDS_PER_TX_PKT) @@ -693,12 +808,10 @@ struct bnx2x_fastpath { FW_DROP_LEVEL(bp)) #define RCQ_TH_HI(bp) (RCQ_TH_LO(bp) + DROPLESS_FC_HEADROOM) - /* This is needed for determining of last_max */ #define SUB_S16(a, b) (s16)((s16)(a) - (s16)(b)) #define SUB_S32(a, b) (s32)((s32)(a) - (s32)(b)) - #define BNX2X_SWCID_SHIFT 17 #define BNX2X_SWCID_MASK ((0x1 << BNX2X_SWCID_SHIFT) - 1) @@ -723,7 +836,6 @@ struct bnx2x_fastpath { DPM_TRIGER_TYPE); \ } while (0) - /* TX CSUM helpers */ #define SKB_CS_OFF(skb) (offsetof(struct tcphdr, check) - \ skb->csum_offset) @@ -766,7 +878,6 @@ struct bnx2x_fastpath { #define BNX2X_RX_SUM_FIX(cqe) \ BNX2X_PRS_FLAG_OVERETH_IPV4(cqe->fast_path_cqe.pars_flags.flags) - #define FP_USB_FUNC_OFF \ offsetof(struct cstorm_status_block_u, func) #define FP_CSB_FUNC_OFF \ @@ -900,14 +1011,14 @@ struct bnx2x_common { #define CHIP_IS_E3A0(bp) (CHIP_IS_E3(bp) && \ (CHIP_REV(bp) == CHIP_REV_Ax)) /* This define is used in two main places: - * 1. In the early stages of nic_load, to know if to configrue Parser / Searcher + * 1. In the early stages of nic_load, to know if to configure Parser / Searcher * to nic-only mode or to offload mode. Offload mode is configured if either the * chip is E1x (where MIC_MODE register is not applicable), or if cnic already * registered for this port (which means that the user wants storage services). * 2. During cnic-related load, to know if offload mode is already configured in - * the HW or needs to be configrued. + * the HW or needs to be configured. * Since the transition from nic-mode to offload-mode in HW causes traffic - * coruption, nic-mode is configured only in ports on which storage services + * corruption, nic-mode is configured only in ports on which storage services * where never requested. */ #define CONFIGURE_NIC_MODE(bp) (!CHIP_IS_E1x(bp) && !CNIC_ENABLED(bp)) @@ -1008,14 +1119,14 @@ extern struct workqueue_struct *bnx2x_wq; * If the maximum number of FP-SB available is X then: * a. If CNIC is supported it consumes 1 FP-SB thus the max number of * regular L2 queues is Y=X-1 - * b. in MF mode the actual number of L2 queues is Y= (X-1/MF_factor) + * b. In MF mode the actual number of L2 queues is Y= (X-1/MF_factor) * c. If the FCoE L2 queue is supported the actual number of L2 queues * is Y+1 * d. The number of irqs (MSIX vectors) is either Y+1 (one extra for * slow-path interrupts) or Y+2 if CNIC is supported (one additional * FP interrupt context for the CNIC). * e. The number of HW context (CID count) is always X or X+1 if FCoE - * L2 queue is supported. the cid for the FCoE L2 queue is always X. + * L2 queue is supported. The cid for the FCoE L2 queue is always X. */ /* fast-path interrupt contexts E1x */ @@ -1068,7 +1179,6 @@ struct bnx2x_slowpath { struct eth_classify_rules_ramrod_data e2; } mac_rdata; - union { struct tstorm_eth_mac_filter_config e1x; struct eth_filter_rules_ramrod_data e2; @@ -1119,7 +1229,6 @@ struct bnx2x_slowpath { #define bnx2x_sp_mapping(bp, var) \ (bp->slowpath_mapping + offsetof(struct bnx2x_slowpath, var)) - /* attn group wiring */ #define MAX_DYNAMIC_ATTN_GRPS 8 @@ -1221,11 +1330,11 @@ enum { BNX2X_SP_RTNL_AFEX_F_UPDATE, BNX2X_SP_RTNL_ENABLE_SRIOV, BNX2X_SP_RTNL_VFPF_MCAST, + BNX2X_SP_RTNL_VFPF_CHANNEL_DOWN, BNX2X_SP_RTNL_VFPF_STORM_RX_MODE, BNX2X_SP_RTNL_HYPERVISOR_VLAN, }; - struct bnx2x_prev_path_list { struct list_head list; u8 bus; @@ -1392,6 +1501,7 @@ struct bnx2x { #define USING_SINGLE_MSIX_FLAG (1 << 20) #define BC_SUPPORTS_DCBX_MSG_NON_PMF (1 << 21) #define IS_VF_FLAG (1 << 22) +#define INTERRUPTS_ENABLED_FLAG (1 << 23) #define BP_NOMCP(bp) ((bp)->flags & NO_MCP_FLAG) @@ -1585,7 +1695,7 @@ struct bnx2x { struct mutex cnic_mutex; struct bnx2x_vlan_mac_obj iscsi_l2_mac_obj; - /* Start index of the "special" (CNIC related) L2 cleints */ + /* Start index of the "special" (CNIC related) L2 clients */ u8 cnic_base_cl_id; int dmae_ready; @@ -1699,7 +1809,7 @@ struct bnx2x { /* operation indication for the sp_rtnl task */ unsigned long sp_rtnl_state; - /* DCBX Negotation results */ + /* DCBX Negotiation results */ struct dcbx_features dcbx_local_feat; u32 dcbx_error; @@ -1755,7 +1865,6 @@ extern int num_queues; #define FUNC_FLG_SPQ 0x0010 #define FUNC_FLG_LEADING 0x0020 /* PF only */ - struct bnx2x_func_init_params { /* dma */ dma_addr_t fw_stat_map; /* valid iff FUNC_FLG_STATS */ @@ -1853,9 +1962,6 @@ struct bnx2x_func_init_params { #define skip_queue(bp, idx) (NO_FCOE(bp) && IS_FCOE_IDX(idx)) - - - /** * bnx2x_set_mac_one - configure a single MAC address * @@ -1921,7 +2027,6 @@ u32 bnx2x_dmae_opcode(struct bnx2x *bp, u8 src_type, u8 dst_type, void bnx2x_prep_dmae_with_comp(struct bnx2x *bp, struct dmae_command *dmae, u8 src_type, u8 dst_type); int bnx2x_issue_dmae_with_comp(struct bnx2x *bp, struct dmae_command *dmae); -void bnx2x_dp_dmae(struct bnx2x *bp, struct dmae_command *dmae, int msglvl); /* FLR related routines */ u32 bnx2x_flr_clnup_poll_count(struct bnx2x *bp); @@ -1937,6 +2042,8 @@ int bnx2x_sp_post(struct bnx2x *bp, int command, int cid, void bnx2x_update_coalesce(struct bnx2x *bp); int bnx2x_get_cur_phy_idx(struct bnx2x *bp); +bool bnx2x_port_after_undi(struct bnx2x *bp); + static inline u32 reg_poll(struct bnx2x *bp, u32 reg, u32 expected, int ms, int wait) { @@ -1998,7 +2105,6 @@ void bnx2x_igu_clear_sb_gen(struct bnx2x *bp, u8 func, u8 idu_sb_id, #define UNLOAD_CLOSE 1 #define UNLOAD_RECOVERY 2 - /* DMAE command defines */ #define DMAE_TIMEOUT -1 #define DMAE_PCI_ERROR -2 /* E2 and onward */ @@ -2062,7 +2168,8 @@ void bnx2x_igu_clear_sb_gen(struct bnx2x *bp, u8 func, u8 idu_sb_id, #define DMAE_LEN32_WR_MAX(bp) (CHIP_IS_E1(bp) ? 0x400 : 0x2000) #define DMAE_COMP_VAL 0x60d0d0ae /* E2 and on - upper bit - indicates eror */ + * indicates error + */ #define MAX_DMAE_C_PER_PORT 8 #define INIT_DMAE_C(bp) (BP_PORT(bp) * MAX_DMAE_C_PER_PORT + \ @@ -2100,7 +2207,6 @@ void bnx2x_igu_clear_sb_gen(struct bnx2x *bp, u8 func, u8 idu_sb_id, #define SP_DESC_CNT (BCM_PAGE_SIZE / sizeof(struct eth_spe)) #define MAX_SP_DESC_CNT (SP_DESC_CNT - 1) - #define BNX2X_BTR 4 #define MAX_SPQ_PENDING 8 @@ -2137,6 +2243,8 @@ void bnx2x_igu_clear_sb_gen(struct bnx2x *bp, u8 func, u8 idu_sb_id, #define ATTN_HARD_WIRED_MASK 0xff00 #define ATTENTION_ID 4 +#define IS_MF_STORAGE_ONLY(bp) (IS_MF_STORAGE_SD(bp) || \ + IS_MF_FCOE_AFEX(bp)) /* stuff added to make the code fit 80Col */ @@ -2338,4 +2446,9 @@ enum { #define NUM_MACS 8 +enum bnx2x_pci_bus_speed { + BNX2X_PCI_LINK_SPEED_2500 = 2500, + BNX2X_PCI_LINK_SPEED_5000 = 5000, + BNX2X_PCI_LINK_SPEED_8000 = 8000 +}; #endif /* bnx2x.h */ diff --git a/drivers/net/ethernet/broadcom/bnx2x/bnx2x_cmn.c b/drivers/net/ethernet/broadcom/bnx2x/bnx2x_cmn.c index 638e55435b0..ec3aa1d451e 100644 --- a/drivers/net/ethernet/broadcom/bnx2x/bnx2x_cmn.c +++ b/drivers/net/ethernet/broadcom/bnx2x/bnx2x_cmn.c @@ -24,6 +24,7 @@ #include <net/tcp.h> #include <net/ipv6.h> #include <net/ip6_checksum.h> +#include <net/ll_poll.h> #include <linux/prefetch.h> #include "bnx2x_cmn.h" #include "bnx2x_init.h" @@ -124,7 +125,7 @@ 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 overriden + * 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++) { @@ -165,7 +166,6 @@ static u16 bnx2x_free_tx_pkt(struct bnx2x *bp, struct bnx2x_fp_txdata *txdata, dma_unmap_single(&bp->pdev->dev, BD_UNMAP_ADDR(tx_start_bd), BD_UNMAP_LEN(tx_start_bd), DMA_TO_DEVICE); - nbd = le16_to_cpu(tx_start_bd->nbd) - 1; #ifdef BNX2X_STOP_ON_ERROR if ((nbd - 1) > (MAX_SKB_FRAGS + 2)) { @@ -259,7 +259,7 @@ 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()): @@ -572,7 +572,7 @@ static int bnx2x_fill_frag_skb(struct bnx2x *bp, struct bnx2x_fastpath *fp, 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_PAGES, DMA_FROM_DEVICE); @@ -733,7 +733,6 @@ static void bnx2x_tpa_stop(struct bnx2x *bp, struct bnx2x_fastpath *fp, dev_kfree_skb_any(skb); } - /* put new data in bin */ rx_buf->data = new_data; @@ -805,40 +804,32 @@ 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++; - bd_cons = fp->rx_bd_cons; bd_prod = fp->rx_bd_prod; bd_prod_fw = bd_prod; sw_comp_cons = fp->rx_comp_cons; sw_comp_prod = fp->rx_comp_prod; - /* 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; @@ -850,12 +841,9 @@ int bnx2x_rx_int(struct bnx2x_fastpath *fp, int budget) 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; cqe_fp_flags = cqe_fp->type_error_flags; cqe_fp_type = cqe_fp_flags & ETH_FAST_PATH_RX_CQE_TYPE; @@ -899,7 +887,6 @@ 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]; @@ -1002,9 +989,13 @@ reuse_rx: PARSING_FLAGS_VLAN) __vlan_hwaccel_put_tag(skb, htons(ETH_P_8021Q), le16_to_cpu(cqe_fp->vlan_tag)); - napi_gro_receive(&fp->napi, skb); + skb_mark_ll(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; @@ -1016,8 +1007,15 @@ 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; @@ -1053,8 +1051,6 @@ 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_ptr[cos]->tx_cons_sb); @@ -1118,7 +1114,7 @@ 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 */ @@ -1161,7 +1157,7 @@ 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) @@ -1304,7 +1300,7 @@ void bnx2x_init_rx_rings(struct bnx2x *bp) "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]; @@ -1726,7 +1722,7 @@ static int bnx2x_req_irq(struct bnx2x *bp) return request_irq(irq, bnx2x_interrupt, flags, bp->dev->name, bp->dev); } -int bnx2x_setup_irqs(struct bnx2x *bp) +static int bnx2x_setup_irqs(struct bnx2x *bp) { int rc = 0; if (bp->flags & USING_MSIX_FLAG && @@ -1759,32 +1755,46 @@ static void bnx2x_napi_enable_cnic(struct bnx2x *bp) { int i; - for_each_rx_queue_cnic(bp, i) + for_each_rx_queue_cnic(bp, i) { + bnx2x_fp_init_lock(&bp->fp[i]); napi_enable(&bnx2x_fp(bp, i, napi)); + } } static void bnx2x_napi_enable(struct bnx2x *bp) { int i; - for_each_eth_queue(bp, i) + 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) + local_bh_disable(); + for_each_rx_queue_cnic(bp, i) { napi_disable(&bnx2x_fp(bp, i, napi)); + while (!bnx2x_fp_lock_napi(&bp->fp[i])) + mdelay(1); + } + local_bh_enable(); } static void bnx2x_napi_disable(struct bnx2x *bp) { int i; - for_each_eth_queue(bp, i) + local_bh_disable(); + for_each_eth_queue(bp, i) { napi_disable(&bnx2x_fp(bp, i, napi)); + while (!bnx2x_fp_lock_napi(&bp->fp[i])) + mdelay(1); + } + local_bh_enable(); } void bnx2x_netif_start(struct bnx2x *bp) @@ -1829,7 +1839,7 @@ u16 bnx2x_select_queue(struct net_device *dev, struct sk_buff *skb) } /* select a non-FCoE queue */ - return __skb_tx_hash(dev, skb, BNX2X_NUM_ETH_QUEUES(bp)); + return __netdev_pick_tx(dev, skb) % BNX2X_NUM_ETH_QUEUES(bp); } void bnx2x_set_num_queues(struct bnx2x *bp) @@ -1862,7 +1872,7 @@ 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(). @@ -1924,7 +1934,7 @@ 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 @@ -1937,7 +1947,7 @@ static int bnx2x_init_rss_pf(struct bnx2x *bp) int i; 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(bp->rss_conf_obj.ind_table); i++) @@ -2015,7 +2025,7 @@ 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. */ void bnx2x_squeeze_objects(struct bnx2x *bp) { @@ -2166,10 +2176,10 @@ static int bnx2x_alloc_fw_stats_mem(struct bnx2x *bp) 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", + 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", + 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; @@ -2183,6 +2193,8 @@ alloc_mem_err: /* 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) & @@ -2195,9 +2207,13 @@ static int bnx2x_nic_load_request(struct bnx2x *bp, u32 *load_code) 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, - DRV_MSG_CODE_LOAD_REQ_WITH_LFA); + (*load_code) = bnx2x_fw_command(bp, DRV_MSG_CODE_LOAD_REQ, param); /* if mcp fails to respond we must abort */ if (!(*load_code)) { @@ -2238,7 +2254,7 @@ int bnx2x_nic_load_analyze_req(struct bnx2x *bp, u32 load_code) /* abort nic load if version mismatch */ if (my_fw != loaded_fw) { - BNX2X_ERR("bnx2x with FW %x was already loaded which mismatches my %x FW. aborting\n", + BNX2X_ERR("bnx2x with FW %x was already loaded which mismatches my %x FW. Aborting\n", loaded_fw, my_fw); return -EBUSY; } @@ -2316,10 +2332,10 @@ static void bnx2x_nic_load_afex_dcc(struct bnx2x *bp, int load_code) static void bnx2x_bz_fp(struct bnx2x *bp, int index) { struct bnx2x_fastpath *fp = &bp->fp[index]; - int cos; struct napi_struct orig_napi = fp->napi; struct bnx2x_agg_info *orig_tpa_info = fp->tpa_info; + /* bzero bnx2x_fastpath contents */ if (fp->tpa_info) memset(fp->tpa_info, 0, ETH_MAX_AGGREGATION_QUEUES_E1H_E2 * @@ -2345,8 +2361,7 @@ static void bnx2x_bz_fp(struct bnx2x *bp, int index) 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 + /* 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 || @@ -2429,7 +2444,6 @@ int bnx2x_load_cnic(struct bnx2x *bp) 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; @@ -2472,6 +2486,7 @@ int bnx2x_nic_load(struct bnx2x *bp, int load_mode) bp->state = BNX2X_STATE_OPENING_WAIT4_LOAD; + /* zero the structure w/o any lock, before SP handler is initialized */ memset(&bp->last_reported_link, 0, sizeof(bp->last_reported_link)); __set_bit(BNX2X_LINK_REPORT_LINK_DOWN, &bp->last_reported_link.link_report_flags); @@ -2536,8 +2551,8 @@ int bnx2x_nic_load(struct bnx2x *bp, int load_mode) } /* 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); @@ -2696,7 +2711,7 @@ int bnx2x_nic_load(struct bnx2x *bp, int load_mode) /* 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; @@ -2841,7 +2856,7 @@ int bnx2x_nic_unload(struct bnx2x *bp, int unload_mode, bool keep_link) } /* Nothing to do during unload if previous bnx2x_nic_load() - * have not completed succesfully - all resourses are released. + * 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. @@ -2856,6 +2871,9 @@ int bnx2x_nic_unload(struct bnx2x *bp, int unload_mode, bool keep_link) 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); @@ -2890,10 +2908,9 @@ int bnx2x_nic_unload(struct bnx2x *bp, int unload_mode, bool keep_link) /* 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). */ @@ -2914,7 +2931,7 @@ int bnx2x_nic_unload(struct bnx2x *bp, int unload_mode, bool keep_link) } /* - * 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. */ if (IS_PF(bp)) @@ -2955,7 +2972,6 @@ int bnx2x_nic_unload(struct bnx2x *bp, int unload_mode, bool keep_link) bnx2x_set_reset_global(bp); } - /* The last driver must disable a "close the gate" if there is no * parity attention or "process kill" pending. */ @@ -3040,6 +3056,8 @@ 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_ptr[cos])) @@ -3049,12 +3067,15 @@ int bnx2x_poll(struct napi_struct *napi, int budget) 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))) { + 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 @@ -3096,6 +3117,32 @@ int bnx2x_poll(struct napi_struct *napi, int budget) return work_done; } +#ifdef CONFIG_NET_LL_RX_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; |