linux/drivers/net/ethernet/broadcom/bnx2x/bnx2x_sp.h

1323 lines
33 KiB
C
Raw Normal View History

/* bnx2x_sp.h: Broadcom Everest network driver.
*
* Copyright (c) 2011-2012 Broadcom Corporation
*
* Unless you and Broadcom execute a separate written software license
* agreement governing use of this software, this software is licensed to you
* under the terms of the GNU General Public License version 2, available
* at http://www.gnu.org/licenses/old-licenses/gpl-2.0.html (the "GPL").
*
* Notwithstanding the above, under no circumstances may you combine this
* software in any way with any other Broadcom software provided under a
* license other than the GPL, without Broadcom's express prior written
* consent.
*
* Maintained by: Eilon Greenstein <eilong@broadcom.com>
* Written by: Vladislav Zolotarov
*
*/
#ifndef BNX2X_SP_VERBS
#define BNX2X_SP_VERBS
struct bnx2x;
struct eth_context;
/* Bits representing general command's configuration */
enum {
RAMROD_TX,
RAMROD_RX,
/* Wait until all pending commands complete */
RAMROD_COMP_WAIT,
/* Don't send a ramrod, only update a registry */
RAMROD_DRV_CLR_ONLY,
/* Configure HW according to the current object state */
RAMROD_RESTORE,
/* Execute the next command now */
RAMROD_EXEC,
/*
* Don't add a new command and continue execution of posponed
* commands. If not set a new command will be added to the
* pending commands list.
*/
RAMROD_CONT,
};
typedef enum {
BNX2X_OBJ_TYPE_RX,
BNX2X_OBJ_TYPE_TX,
BNX2X_OBJ_TYPE_RX_TX,
} bnx2x_obj_type;
/* Filtering states */
enum {
BNX2X_FILTER_MAC_PENDING,
BNX2X_FILTER_VLAN_PENDING,
BNX2X_FILTER_VLAN_MAC_PENDING,
BNX2X_FILTER_RX_MODE_PENDING,
BNX2X_FILTER_RX_MODE_SCHED,
BNX2X_FILTER_ISCSI_ETH_START_SCHED,
BNX2X_FILTER_ISCSI_ETH_STOP_SCHED,
BNX2X_FILTER_FCOE_ETH_START_SCHED,
BNX2X_FILTER_FCOE_ETH_STOP_SCHED,
BNX2X_FILTER_MCAST_PENDING,
BNX2X_FILTER_MCAST_SCHED,
BNX2X_FILTER_RSS_CONF_PENDING,
};
struct bnx2x_raw_obj {
u8 func_id;
/* Queue params */
u8 cl_id;
u32 cid;
/* Ramrod data buffer params */
void *rdata;
dma_addr_t rdata_mapping;
/* Ramrod state params */
int state; /* "ramrod is pending" state bit */
unsigned long *pstate; /* pointer to state buffer */
bnx2x_obj_type obj_type;
int (*wait_comp)(struct bnx2x *bp,
struct bnx2x_raw_obj *o);
bool (*check_pending)(struct bnx2x_raw_obj *o);
void (*clear_pending)(struct bnx2x_raw_obj *o);
void (*set_pending)(struct bnx2x_raw_obj *o);
};
/************************* VLAN-MAC commands related parameters ***************/
struct bnx2x_mac_ramrod_data {
u8 mac[ETH_ALEN];
};
struct bnx2x_vlan_ramrod_data {
u16 vlan;
};
struct bnx2x_vlan_mac_ramrod_data {
u8 mac[ETH_ALEN];
u16 vlan;
};
union bnx2x_classification_ramrod_data {
struct bnx2x_mac_ramrod_data mac;
struct bnx2x_vlan_ramrod_data vlan;
struct bnx2x_vlan_mac_ramrod_data vlan_mac;
};
/* VLAN_MAC commands */
enum bnx2x_vlan_mac_cmd {
BNX2X_VLAN_MAC_ADD,
BNX2X_VLAN_MAC_DEL,
BNX2X_VLAN_MAC_MOVE,
};
struct bnx2x_vlan_mac_data {
/* Requested command: BNX2X_VLAN_MAC_XX */
enum bnx2x_vlan_mac_cmd cmd;
/*
* used to contain the data related vlan_mac_flags bits from
* ramrod parameters.
*/
unsigned long vlan_mac_flags;
/* Needed for MOVE command */
struct bnx2x_vlan_mac_obj *target_obj;
union bnx2x_classification_ramrod_data u;
};
/*************************** Exe Queue obj ************************************/
union bnx2x_exe_queue_cmd_data {
struct bnx2x_vlan_mac_data vlan_mac;
struct {
/* TODO */
} mcast;
};
struct bnx2x_exeq_elem {
struct list_head link;
/* Length of this element in the exe_chunk. */
int cmd_len;
union bnx2x_exe_queue_cmd_data cmd_data;
};
union bnx2x_qable_obj;
union bnx2x_exeq_comp_elem {
union event_ring_elem *elem;
};
struct bnx2x_exe_queue_obj;
typedef int (*exe_q_validate)(struct bnx2x *bp,
union bnx2x_qable_obj *o,
struct bnx2x_exeq_elem *elem);
typedef int (*exe_q_remove)(struct bnx2x *bp,
union bnx2x_qable_obj *o,
struct bnx2x_exeq_elem *elem);
/**
* @return positive is entry was optimized, 0 - if not, negative
* in case of an error.
*/
typedef int (*exe_q_optimize)(struct bnx2x *bp,
union bnx2x_qable_obj *o,
struct bnx2x_exeq_elem *elem);
typedef int (*exe_q_execute)(struct bnx2x *bp,
union bnx2x_qable_obj *o,
struct list_head *exe_chunk,
unsigned long *ramrod_flags);
typedef struct bnx2x_exeq_elem *
(*exe_q_get)(struct bnx2x_exe_queue_obj *o,
struct bnx2x_exeq_elem *elem);
struct bnx2x_exe_queue_obj {
/*
* Commands pending for an execution.
*/
struct list_head exe_queue;
/*
* Commands pending for an completion.
*/
struct list_head pending_comp;
spinlock_t lock;
/* Maximum length of commands' list for one execution */
int exe_chunk_len;
union bnx2x_qable_obj *owner;
/****** Virtual functions ******/
/**
* Called before commands execution for commands that are really
* going to be executed (after 'optimize').
*
* Must run under exe_queue->lock
*/
exe_q_validate validate;
/**
* Called before removing pending commands, cleaning allocated
* resources (e.g., credits from validate)
*/
exe_q_remove remove;
/**
* This will try to cancel the current pending commands list
* considering the new command.
*
* Returns the number of optimized commands or a negative error code
*
* Must run under exe_queue->lock
*/
exe_q_optimize optimize;
/**
* Run the next commands chunk (owner specific).
*/
exe_q_execute execute;
/**
* Return the exe_queue element containing the specific command
* if any. Otherwise return NULL.
*/
exe_q_get get;
};
/***************** Classification verbs: Set/Del MAC/VLAN/VLAN-MAC ************/
/*
* Element in the VLAN_MAC registry list having all currenty configured
* rules.
*/
struct bnx2x_vlan_mac_registry_elem {
struct list_head link;
/*
* Used to store the cam offset used for the mac/vlan/vlan-mac.
* Relevant for 57710 and 57711 only. VLANs and MACs share the
* same CAM for these chips.
*/
int cam_offset;
/* Needed for DEL and RESTORE flows */
unsigned long vlan_mac_flags;
union bnx2x_classification_ramrod_data u;
};
/* Bits representing VLAN_MAC commands specific flags */
enum {
BNX2X_UC_LIST_MAC,
BNX2X_ETH_MAC,
BNX2X_ISCSI_ETH_MAC,
BNX2X_NETQ_ETH_MAC,
BNX2X_DONT_CONSUME_CAM_CREDIT,
BNX2X_DONT_CONSUME_CAM_CREDIT_DEST,
};
struct bnx2x_vlan_mac_ramrod_params {
/* Object to run the command from */
struct bnx2x_vlan_mac_obj *vlan_mac_obj;
/* General command flags: COMP_WAIT, etc. */
unsigned long ramrod_flags;
/* Command specific configuration request */
struct bnx2x_vlan_mac_data user_req;
};
struct bnx2x_vlan_mac_obj {
struct bnx2x_raw_obj raw;
/* Bookkeeping list: will prevent the addition of already existing
* entries.
*/
struct list_head head;
/* TODO: Add it's initialization in the init functions */
struct bnx2x_exe_queue_obj exe_queue;
/* MACs credit pool */
struct bnx2x_credit_pool_obj *macs_pool;
/* VLANs credit pool */
struct bnx2x_credit_pool_obj *vlans_pool;
/* RAMROD command to be used */
int ramrod_cmd;
/* copy first n elements onto preallocated buffer
*
* @param n number of elements to get
* @param buf buffer preallocated by caller into which elements
* will be copied. Note elements are 4-byte aligned
* so buffer size must be able to accomodate the
* aligned elements.
*
* @return number of copied bytes
*/
int (*get_n_elements)(struct bnx2x *bp, struct bnx2x_vlan_mac_obj *o,
int n, u8 *buf);
/**
* Checks if ADD-ramrod with the given params may be performed.
*
* @return zero if the element may be added
*/
int (*check_add)(struct bnx2x_vlan_mac_obj *o,
union bnx2x_classification_ramrod_data *data);
/**
* Checks if DEL-ramrod with the given params may be performed.
*
* @return true if the element may be deleted
*/
struct bnx2x_vlan_mac_registry_elem *
(*check_del)(struct bnx2x_vlan_mac_obj *o,
union bnx2x_classification_ramrod_data *data);
/**
* Checks if DEL-ramrod with the given params may be performed.
*
* @return true if the element may be deleted
*/
bool (*check_move)(struct bnx2x_vlan_mac_obj *src_o,
struct bnx2x_vlan_mac_obj *dst_o,
union bnx2x_classification_ramrod_data *data);
/**
* Update the relevant credit object(s) (consume/return
* correspondingly).
*/
bool (*get_credit)(struct bnx2x_vlan_mac_obj *o);
bool (*put_credit)(struct bnx2x_vlan_mac_obj *o);
bool (*get_cam_offset)(struct bnx2x_vlan_mac_obj *o, int *offset);
bool (*put_cam_offset)(struct bnx2x_vlan_mac_obj *o, int offset);
/**
* Configures one rule in the ramrod data buffer.
*/
void (*set_one_rule)(struct bnx2x *bp,
struct bnx2x_vlan_mac_obj *o,
struct bnx2x_exeq_elem *elem, int rule_idx,
int cam_offset);
/**
* Delete all configured elements having the given
* vlan_mac_flags specification. Assumes no pending for
* execution commands. Will schedule all all currently
* configured MACs/VLANs/VLAN-MACs matching the vlan_mac_flags
* specification for deletion and will use the given
* ramrod_flags for the last DEL operation.
*
* @param bp
* @param o
* @param ramrod_flags RAMROD_XX flags
*
* @return 0 if the last operation has completed successfully
* and there are no more elements left, positive value
* if there are pending for completion commands,
* negative value in case of failure.
*/
int (*delete_all)(struct bnx2x *bp,
struct bnx2x_vlan_mac_obj *o,
unsigned long *vlan_mac_flags,
unsigned long *ramrod_flags);
/**
* Reconfigures the next MAC/VLAN/VLAN-MAC element from the previously
* configured elements list.
*
* @param bp
* @param p Command parameters (RAMROD_COMP_WAIT bit in
* ramrod_flags is only taken into an account)
* @param ppos a pointer to the cooky that should be given back in the
* next call to make function handle the next element. If
* *ppos is set to NULL it will restart the iterator.
* If returned *ppos == NULL this means that the last
* element has been handled.
*
* @return int
*/
int (*restore)(struct bnx2x *bp,
struct bnx2x_vlan_mac_ramrod_params *p,
struct bnx2x_vlan_mac_registry_elem **ppos);
/**
* Should be called on a completion arival.
*
* @param bp
* @param o
* @param cqe Completion element we are handling
* @param ramrod_flags if RAMROD_CONT is set the next bulk of
* pending commands will be executed.
* RAMROD_DRV_CLR_ONLY and RAMROD_RESTORE
* may also be set if needed.
*
* @return 0 if there are neither pending nor waiting for
* completion commands. Positive value if there are
* pending for execution or for completion commands.
* Negative value in case of an error (including an
* error in the cqe).
*/
int (*complete)(struct bnx2x *bp, struct bnx2x_vlan_mac_obj *o,
union event_ring_elem *cqe,
unsigned long *ramrod_flags);
/**
* Wait for completion of all commands. Don't schedule new ones,
* just wait. It assumes that the completion code will schedule
* for new commands.
*/
int (*wait)(struct bnx2x *bp, struct bnx2x_vlan_mac_obj *o);
};
/** RX_MODE verbs:DROP_ALL/ACCEPT_ALL/ACCEPT_ALL_MULTI/ACCEPT_ALL_VLAN/NORMAL */
/* RX_MODE ramrod spesial flags: set in rx_mode_flags field in
* a bnx2x_rx_mode_ramrod_params.
*/
enum {
BNX2X_RX_MODE_FCOE_ETH,
BNX2X_RX_MODE_ISCSI_ETH,
};
enum {
BNX2X_ACCEPT_UNICAST,
BNX2X_ACCEPT_MULTICAST,
BNX2X_ACCEPT_ALL_UNICAST,
BNX2X_ACCEPT_ALL_MULTICAST,
BNX2X_ACCEPT_BROADCAST,
BNX2X_ACCEPT_UNMATCHED,
BNX2X_ACCEPT_ANY_VLAN
};
struct bnx2x_rx_mode_ramrod_params {
struct bnx2x_rx_mode_obj *rx_mode_obj;
unsigned long *pstate;
int state;
u8 cl_id;
u32 cid;
u8 func_id;
unsigned long ramrod_flags;
unsigned long rx_mode_flags;
/*
* rdata is either a pointer to eth_filter_rules_ramrod_data(e2) or to
* a tstorm_eth_mac_filter_config (e1x).
*/
void *rdata;
dma_addr_t rdata_mapping;
/* Rx mode settings */
unsigned long rx_accept_flags;
/* internal switching settings */
unsigned long tx_accept_flags;
};
struct bnx2x_rx_mode_obj {
int (*config_rx_mode)(struct bnx2x *bp,
struct bnx2x_rx_mode_ramrod_params *p);
int (*wait_comp)(struct bnx2x *bp,
struct bnx2x_rx_mode_ramrod_params *p);
};
/********************** Set multicast group ***********************************/
struct bnx2x_mcast_list_elem {
struct list_head link;
u8 *mac;
};
union bnx2x_mcast_config_data {
u8 *mac;
u8 bin; /* used in a RESTORE flow */
};
struct bnx2x_mcast_ramrod_params {
struct bnx2x_mcast_obj *mcast_obj;
/* Relevant options are RAMROD_COMP_WAIT and RAMROD_DRV_CLR_ONLY */
unsigned long ramrod_flags;
struct list_head mcast_list; /* list of struct bnx2x_mcast_list_elem */
/** TODO:
* - rename it to macs_num.
* - Add a new command type for handling pending commands
* (remove "zero semantics").
*
* Length of mcast_list. If zero and ADD_CONT command - post
* pending commands.
*/
int mcast_list_len;
};
enum {
BNX2X_MCAST_CMD_ADD,
BNX2X_MCAST_CMD_CONT,
BNX2X_MCAST_CMD_DEL,
BNX2X_MCAST_CMD_RESTORE,
};
struct bnx2x_mcast_obj {
struct bnx2x_raw_obj raw;
union {
struct {
#define BNX2X_MCAST_BINS_NUM 256
#define BNX2X_MCAST_VEC_SZ (BNX2X_MCAST_BINS_NUM / 64)
u64 vec[BNX2X_MCAST_VEC_SZ];
/** Number of BINs to clear. Should be updated
* immediately when a command arrives in order to
* properly create DEL commands.
*/
int num_bins_set;
} aprox_match;
struct {
struct list_head macs;
int num_macs_set;
} exact_match;
} registry;
/* Pending commands */
struct list_head pending_cmds_head;
/* A state that is set in raw.pstate, when there are pending commands */
int sched_state;
/* Maximal number of mcast MACs configured in one command */
int max_cmd_len;
/* Total number of currently pending MACs to configure: both
* in the pending commands list and in the current command.
*/
int total_pending_num;
u8 engine_id;
/**
* @param cmd command to execute (BNX2X_MCAST_CMD_X, see above)
*/
int (*config_mcast)(struct bnx2x *bp,
struct bnx2x_mcast_ramrod_params *p, int cmd);
/**
* Fills the ramrod data during the RESTORE flow.
*
* @param bp
* @param o
* @param start_idx Registry index to start from
* @param rdata_idx Index in the ramrod data to start from
*
* @return -1 if we handled the whole registry or index of the last
* handled registry element.
*/
int (*hdl_restore)(struct bnx2x *bp, struct bnx2x_mcast_obj *o,
int start_bin, int *rdata_idx);
int (*enqueue_cmd)(struct bnx2x *bp, struct bnx2x_mcast_obj *o,
struct bnx2x_mcast_ramrod_params *p, int cmd);
void (*set_one_rule)(struct bnx2x *bp,
struct bnx2x_mcast_obj *o, int idx,
union bnx2x_mcast_config_data *cfg_data, int cmd);
/** Checks if there are more mcast MACs to be set or a previous
* command is still pending.
*/
bool (*check_pending)(struct bnx2x_mcast_obj *o);
/**
* Set/Clear/Check SCHEDULED state of the object
*/
void (*set_sched)(struct bnx2x_mcast_obj *o);
void (*clear_sched)(struct bnx2x_mcast_obj *o);
bool (*check_sched)(struct bnx2x_mcast_obj *o);
/* Wait until all pending commands complete */
int (*wait_comp)(struct bnx2x *bp, struct bnx2x_mcast_obj *o);
/**
* Handle the internal object counters needed for proper
* commands handling. Checks that the provided parameters are
* feasible.
*/
int (*validate)(struct bnx2x *bp,
struct bnx2x_mcast_ramrod_params *p, int cmd);
/**
* Restore the values of internal counters in case of a failure.
*/
void (*revert)(struct bnx2x *bp,
struct bnx2x_mcast_ramrod_params *p,
int old_num_bins);
int (*get_registry_size)(struct bnx2x_mcast_obj *o);
void (*set_registry_size)(struct bnx2x_mcast_obj *o, int n);
};
/*************************** Credit handling **********************************/
struct bnx2x_credit_pool_obj {
/* Current amount of credit in the pool */
atomic_t credit;
/* Maximum allowed credit. put() will check against it. */
int pool_sz;
/*
* Allocate a pool table statically.
*
* Currently the mamimum allowed size is MAX_MAC_CREDIT_E2(272)
*
* The set bit in the table will mean that the entry is available.
*/
#define BNX2X_POOL_VEC_SIZE (MAX_MAC_CREDIT_E2 / 64)
u64 pool_mirror[BNX2X_POOL_VEC_SIZE];
/* Base pool offset (initialized differently */
int base_pool_offset;
/**
* Get the next free pool entry.
*
* @return true if there was a free entry in the pool
*/
bool (*get_entry)(struct bnx2x_credit_pool_obj *o, int *entry);
/**
* Return the entry back to the pool.
*
* @return true if entry is legal and has been successfully
* returned to the pool.
*/
bool (*put_entry)(struct bnx2x_credit_pool_obj *o, int entry);
/**
* Get the requested amount of credit from the pool.
*
* @param cnt Amount of requested credit
* @return true if the operation is successful
*/
bool (*get)(struct bnx2x_credit_pool_obj *o, int cnt);
/**
* Returns the credit to the pool.
*
* @param cnt Amount of credit to return
* @return true if the operation is successful
*/
bool (*put)(struct bnx2x_credit_pool_obj *o, int cnt);
/**
* Reads the current amount of credit.
*/
int (*check)(struct bnx2x_credit_pool_obj *o);
};
/*************************** RSS configuration ********************************/
enum {
/* RSS_MODE bits are mutually exclusive */
BNX2X_RSS_MODE_DISABLED,
BNX2X_RSS_MODE_REGULAR,
BNX2X_RSS_MODE_VLAN_PRI,
BNX2X_RSS_MODE_E1HOV_PRI,
BNX2X_RSS_MODE_IP_DSCP,
BNX2X_RSS_SET_SRCH, /* Setup searcher, E1x specific flag */
BNX2X_RSS_IPV4,
BNX2X_RSS_IPV4_TCP,
BNX2X_RSS_IPV6,
BNX2X_RSS_IPV6_TCP,
};
struct bnx2x_config_rss_params {
struct bnx2x_rss_config_obj *rss_obj;
/* may have RAMROD_COMP_WAIT set only */
unsigned long ramrod_flags;
/* BNX2X_RSS_X bits */
unsigned long rss_flags;
/* Number hash bits to take into an account */
u8 rss_result_mask;
/* Indirection table */
u8 ind_table[T_ETH_INDIRECTION_TABLE_SIZE];
/* RSS hash values */
u32 rss_key[10];
/* valid only iff BNX2X_RSS_UPDATE_TOE is set */
u16 toe_rss_bitmap;
};
struct bnx2x_rss_config_obj {
struct bnx2x_raw_obj raw;
/* RSS engine to use */
u8 engine_id;
/* Last configured indirection table */
u8 ind_table[T_ETH_INDIRECTION_TABLE_SIZE];
int (*config_rss)(struct bnx2x *bp,
struct bnx2x_config_rss_params *p);
};
/*********************** Queue state update ***********************************/
/* UPDATE command options */
enum {
BNX2X_Q_UPDATE_IN_VLAN_REM,
BNX2X_Q_UPDATE_IN_VLAN_REM_CHNG,
BNX2X_Q_UPDATE_OUT_VLAN_REM,
BNX2X_Q_UPDATE_OUT_VLAN_REM_CHNG,
BNX2X_Q_UPDATE_ANTI_SPOOF,
BNX2X_Q_UPDATE_ANTI_SPOOF_CHNG,
BNX2X_Q_UPDATE_ACTIVATE,
BNX2X_Q_UPDATE_ACTIVATE_CHNG,
BNX2X_Q_UPDATE_DEF_VLAN_EN,
BNX2X_Q_UPDATE_DEF_VLAN_EN_CHNG,
BNX2X_Q_UPDATE_SILENT_VLAN_REM_CHNG,
BNX2X_Q_UPDATE_SILENT_VLAN_REM
};
/* Allowed Queue states */
enum bnx2x_q_state {
BNX2X_Q_STATE_RESET,
BNX2X_Q_STATE_INITIALIZED,
BNX2X_Q_STATE_ACTIVE,
bnx2x: Multiple concurrent l2 traffic classes Overview: Support mapping of priorities to traffic classes and traffic classes to transmission queues ranges in the net device. The queue ranges are (count, offset) pairs relating to the txq array. This can be done via DCBX negotiation or by kernel. As a result Enhanced Transmission Selection (ETS) and Priority Flow Control (PFC) are supported between L2 network traffic classes. Mapping: This patch uses the netdev_set_num_tc, netdev_set_prio_tc_map and netdev_set_tc_queue functions to map priorities to traffic classes and traffic classes to transmission queue ranges. This mapping is performed by bnx2x_setup_tc function which is connected to the ndo_setup_tc. This function is always called at nic load where by default it maps all priorities to tc 0, and it may also be called by the kernel or by the bnx2x upon DCBX negotiation to modify the mapping. rtnl lock: When the ndo_setup_tc is called at nic load or by kernel the rtnl lock is already taken. However, when DCBX negotiation takes place the lock is not taken. The work is therefore scheduled to be handled by the sp_rtnl task. Fastpath: The fastpath structure of the bnx2x which was previously used to hold the information of one tx queue and one rx queue was redesigned to represent multiple tx queues, one for each traffic class. The transmission queue supplied in the skb by the kernel can no longer be interpreted as a straightforward index into the fastpath structure array, but it must rather be decoded to the appropriate fastpath index and the tc within that fastpath. Slowpath: The bnx2x's queue object was redesigned to accommodate multiple transmission queues. The queue object's state machine was enhanced to allow opening multiple transmission-only connections on top of the regular tx-rx connection. Firmware: This feature relies on the tx-only queue feature introduced in the bnx2x 7.0.23 firmware and the FW likewise must have the bnx2x multi cos support. Signed-off-by: Ariel Elior <ariele@broadcom.com> Signed-off-by: Eilon Greenstein <eilong@broadcom.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2011-07-14 16:31:57 +08:00
BNX2X_Q_STATE_MULTI_COS,
BNX2X_Q_STATE_MCOS_TERMINATED,
BNX2X_Q_STATE_INACTIVE,
BNX2X_Q_STATE_STOPPED,
BNX2X_Q_STATE_TERMINATED,
BNX2X_Q_STATE_FLRED,
BNX2X_Q_STATE_MAX,
};
/* Allowed commands */
enum bnx2x_queue_cmd {
BNX2X_Q_CMD_INIT,
BNX2X_Q_CMD_SETUP,
bnx2x: Multiple concurrent l2 traffic classes Overview: Support mapping of priorities to traffic classes and traffic classes to transmission queues ranges in the net device. The queue ranges are (count, offset) pairs relating to the txq array. This can be done via DCBX negotiation or by kernel. As a result Enhanced Transmission Selection (ETS) and Priority Flow Control (PFC) are supported between L2 network traffic classes. Mapping: This patch uses the netdev_set_num_tc, netdev_set_prio_tc_map and netdev_set_tc_queue functions to map priorities to traffic classes and traffic classes to transmission queue ranges. This mapping is performed by bnx2x_setup_tc function which is connected to the ndo_setup_tc. This function is always called at nic load where by default it maps all priorities to tc 0, and it may also be called by the kernel or by the bnx2x upon DCBX negotiation to modify the mapping. rtnl lock: When the ndo_setup_tc is called at nic load or by kernel the rtnl lock is already taken. However, when DCBX negotiation takes place the lock is not taken. The work is therefore scheduled to be handled by the sp_rtnl task. Fastpath: The fastpath structure of the bnx2x which was previously used to hold the information of one tx queue and one rx queue was redesigned to represent multiple tx queues, one for each traffic class. The transmission queue supplied in the skb by the kernel can no longer be interpreted as a straightforward index into the fastpath structure array, but it must rather be decoded to the appropriate fastpath index and the tc within that fastpath. Slowpath: The bnx2x's queue object was redesigned to accommodate multiple transmission queues. The queue object's state machine was enhanced to allow opening multiple transmission-only connections on top of the regular tx-rx connection. Firmware: This feature relies on the tx-only queue feature introduced in the bnx2x 7.0.23 firmware and the FW likewise must have the bnx2x multi cos support. Signed-off-by: Ariel Elior <ariele@broadcom.com> Signed-off-by: Eilon Greenstein <eilong@broadcom.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2011-07-14 16:31:57 +08:00
BNX2X_Q_CMD_SETUP_TX_ONLY,
BNX2X_Q_CMD_DEACTIVATE,
BNX2X_Q_CMD_ACTIVATE,
BNX2X_Q_CMD_UPDATE,
BNX2X_Q_CMD_UPDATE_TPA,
BNX2X_Q_CMD_HALT,
BNX2X_Q_CMD_CFC_DEL,
BNX2X_Q_CMD_TERMINATE,
BNX2X_Q_CMD_EMPTY,
BNX2X_Q_CMD_MAX,
};
/* queue SETUP + INIT flags */
enum {
BNX2X_Q_FLG_TPA,
BNX2X_Q_FLG_TPA_IPV6,
use FW 7.2.16 The patch integrates FW 7.2.16 HSI and implements driver part of GRO flow. FW 7.2.16 adds the ability to aggregate packets for GRO (and not just LRO) and also fixes some bugs. 1. Added new aggregation mode: GRO. In this mode packets are aggregated such that the original packets can be reconstructed by the OS. 2. 57712 HW bug workaround - initialized all CAM TM registers to 0x32. 3. Adding the FCoE statistics structures to the BNX2X HSI. 4. Wrong configuration of TX HW input buffer size may cause theoretical performance effect. Performed configuration fix. 5. FCOE - Arrival of packets beyond task IO size can lead to crash. Fix firmware data-in flow. 6. iSCSI - In rare cases of on-chip termination the graceful termination timer hangs, and the termination doesn't complete. Firmware fix to MSL timer tolerance. 7. iSCSI - Chip hangs when target sends FIN out-of-order or with isles open at the initiator side. Firmware implementation corrected to drop FIN received out-of-order or with isles still open. 8. iSCSI - Chip hangs when in case of retransmission not aligned to 4-bytes from the beginning of iSCSI PDU. Firmware implementation corrected to support arbitrary aligned retransmissions. 9. iSCSI - Arrival of target-initiated NOP-IN during intense ISCSI traffic might lead to crash. Firmware fix to relevant flow. Signed-off-by: Dmitry Kravkov <dmitry@broadcom.com> Signed-off-by: Eilon Greenstein <eilong@broadcom.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2012-02-20 17:59:08 +08:00
BNX2X_Q_FLG_TPA_GRO,
BNX2X_Q_FLG_STATS,
BNX2X_Q_FLG_ZERO_STATS,
BNX2X_Q_FLG_ACTIVE,
BNX2X_Q_FLG_OV,
BNX2X_Q_FLG_VLAN,
BNX2X_Q_FLG_COS,
BNX2X_Q_FLG_HC,
BNX2X_Q_FLG_HC_EN,
BNX2X_Q_FLG_DHC,
BNX2X_Q_FLG_FCOE,
BNX2X_Q_FLG_LEADING_RSS,
BNX2X_Q_FLG_MCAST,
BNX2X_Q_FLG_DEF_VLAN,
BNX2X_Q_FLG_TX_SWITCH,
BNX2X_Q_FLG_TX_SEC,
BNX2X_Q_FLG_ANTI_SPOOF,
BNX2X_Q_FLG_SILENT_VLAN_REM
};
/* Queue type options: queue type may be a compination of below. */
enum bnx2x_q_type {
/** TODO: Consider moving both these flags into the init()
* ramrod params.
*/
BNX2X_Q_TYPE_HAS_RX,
BNX2X_Q_TYPE_HAS_TX,
};
bnx2x: Multiple concurrent l2 traffic classes Overview: Support mapping of priorities to traffic classes and traffic classes to transmission queues ranges in the net device. The queue ranges are (count, offset) pairs relating to the txq array. This can be done via DCBX negotiation or by kernel. As a result Enhanced Transmission Selection (ETS) and Priority Flow Control (PFC) are supported between L2 network traffic classes. Mapping: This patch uses the netdev_set_num_tc, netdev_set_prio_tc_map and netdev_set_tc_queue functions to map priorities to traffic classes and traffic classes to transmission queue ranges. This mapping is performed by bnx2x_setup_tc function which is connected to the ndo_setup_tc. This function is always called at nic load where by default it maps all priorities to tc 0, and it may also be called by the kernel or by the bnx2x upon DCBX negotiation to modify the mapping. rtnl lock: When the ndo_setup_tc is called at nic load or by kernel the rtnl lock is already taken. However, when DCBX negotiation takes place the lock is not taken. The work is therefore scheduled to be handled by the sp_rtnl task. Fastpath: The fastpath structure of the bnx2x which was previously used to hold the information of one tx queue and one rx queue was redesigned to represent multiple tx queues, one for each traffic class. The transmission queue supplied in the skb by the kernel can no longer be interpreted as a straightforward index into the fastpath structure array, but it must rather be decoded to the appropriate fastpath index and the tc within that fastpath. Slowpath: The bnx2x's queue object was redesigned to accommodate multiple transmission queues. The queue object's state machine was enhanced to allow opening multiple transmission-only connections on top of the regular tx-rx connection. Firmware: This feature relies on the tx-only queue feature introduced in the bnx2x 7.0.23 firmware and the FW likewise must have the bnx2x multi cos support. Signed-off-by: Ariel Elior <ariele@broadcom.com> Signed-off-by: Eilon Greenstein <eilong@broadcom.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2011-07-14 16:31:57 +08:00
#define BNX2X_PRIMARY_CID_INDEX 0
#define BNX2X_MULTI_TX_COS_E1X 3 /* QM only */
bnx2x: Multiple concurrent l2 traffic classes Overview: Support mapping of priorities to traffic classes and traffic classes to transmission queues ranges in the net device. The queue ranges are (count, offset) pairs relating to the txq array. This can be done via DCBX negotiation or by kernel. As a result Enhanced Transmission Selection (ETS) and Priority Flow Control (PFC) are supported between L2 network traffic classes. Mapping: This patch uses the netdev_set_num_tc, netdev_set_prio_tc_map and netdev_set_tc_queue functions to map priorities to traffic classes and traffic classes to transmission queue ranges. This mapping is performed by bnx2x_setup_tc function which is connected to the ndo_setup_tc. This function is always called at nic load where by default it maps all priorities to tc 0, and it may also be called by the kernel or by the bnx2x upon DCBX negotiation to modify the mapping. rtnl lock: When the ndo_setup_tc is called at nic load or by kernel the rtnl lock is already taken. However, when DCBX negotiation takes place the lock is not taken. The work is therefore scheduled to be handled by the sp_rtnl task. Fastpath: The fastpath structure of the bnx2x which was previously used to hold the information of one tx queue and one rx queue was redesigned to represent multiple tx queues, one for each traffic class. The transmission queue supplied in the skb by the kernel can no longer be interpreted as a straightforward index into the fastpath structure array, but it must rather be decoded to the appropriate fastpath index and the tc within that fastpath. Slowpath: The bnx2x's queue object was redesigned to accommodate multiple transmission queues. The queue object's state machine was enhanced to allow opening multiple transmission-only connections on top of the regular tx-rx connection. Firmware: This feature relies on the tx-only queue feature introduced in the bnx2x 7.0.23 firmware and the FW likewise must have the bnx2x multi cos support. Signed-off-by: Ariel Elior <ariele@broadcom.com> Signed-off-by: Eilon Greenstein <eilong@broadcom.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2011-07-14 16:31:57 +08:00
#define BNX2X_MULTI_TX_COS_E2_E3A0 2
#define BNX2X_MULTI_TX_COS_E3B0 3
#define BNX2X_MULTI_TX_COS 3 /* Maximum possible */
bnx2x: Multiple concurrent l2 traffic classes Overview: Support mapping of priorities to traffic classes and traffic classes to transmission queues ranges in the net device. The queue ranges are (count, offset) pairs relating to the txq array. This can be done via DCBX negotiation or by kernel. As a result Enhanced Transmission Selection (ETS) and Priority Flow Control (PFC) are supported between L2 network traffic classes. Mapping: This patch uses the netdev_set_num_tc, netdev_set_prio_tc_map and netdev_set_tc_queue functions to map priorities to traffic classes and traffic classes to transmission queue ranges. This mapping is performed by bnx2x_setup_tc function which is connected to the ndo_setup_tc. This function is always called at nic load where by default it maps all priorities to tc 0, and it may also be called by the kernel or by the bnx2x upon DCBX negotiation to modify the mapping. rtnl lock: When the ndo_setup_tc is called at nic load or by kernel the rtnl lock is already taken. However, when DCBX negotiation takes place the lock is not taken. The work is therefore scheduled to be handled by the sp_rtnl task. Fastpath: The fastpath structure of the bnx2x which was previously used to hold the information of one tx queue and one rx queue was redesigned to represent multiple tx queues, one for each traffic class. The transmission queue supplied in the skb by the kernel can no longer be interpreted as a straightforward index into the fastpath structure array, but it must rather be decoded to the appropriate fastpath index and the tc within that fastpath. Slowpath: The bnx2x's queue object was redesigned to accommodate multiple transmission queues. The queue object's state machine was enhanced to allow opening multiple transmission-only connections on top of the regular tx-rx connection. Firmware: This feature relies on the tx-only queue feature introduced in the bnx2x 7.0.23 firmware and the FW likewise must have the bnx2x multi cos support. Signed-off-by: Ariel Elior <ariele@broadcom.com> Signed-off-by: Eilon Greenstein <eilong@broadcom.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2011-07-14 16:31:57 +08:00
struct bnx2x_queue_init_params {
struct {
unsigned long flags;
u16 hc_rate;
u8 fw_sb_id;
u8 sb_cq_index;
} tx;
struct {
unsigned long flags;
u16 hc_rate;
u8 fw_sb_id;
u8 sb_cq_index;
} rx;
/* CID context in the host memory */
bnx2x: Multiple concurrent l2 traffic classes Overview: Support mapping of priorities to traffic classes and traffic classes to transmission queues ranges in the net device. The queue ranges are (count, offset) pairs relating to the txq array. This can be done via DCBX negotiation or by kernel. As a result Enhanced Transmission Selection (ETS) and Priority Flow Control (PFC) are supported between L2 network traffic classes. Mapping: This patch uses the netdev_set_num_tc, netdev_set_prio_tc_map and netdev_set_tc_queue functions to map priorities to traffic classes and traffic classes to transmission queue ranges. This mapping is performed by bnx2x_setup_tc function which is connected to the ndo_setup_tc. This function is always called at nic load where by default it maps all priorities to tc 0, and it may also be called by the kernel or by the bnx2x upon DCBX negotiation to modify the mapping. rtnl lock: When the ndo_setup_tc is called at nic load or by kernel the rtnl lock is already taken. However, when DCBX negotiation takes place the lock is not taken. The work is therefore scheduled to be handled by the sp_rtnl task. Fastpath: The fastpath structure of the bnx2x which was previously used to hold the information of one tx queue and one rx queue was redesigned to represent multiple tx queues, one for each traffic class. The transmission queue supplied in the skb by the kernel can no longer be interpreted as a straightforward index into the fastpath structure array, but it must rather be decoded to the appropriate fastpath index and the tc within that fastpath. Slowpath: The bnx2x's queue object was redesigned to accommodate multiple transmission queues. The queue object's state machine was enhanced to allow opening multiple transmission-only connections on top of the regular tx-rx connection. Firmware: This feature relies on the tx-only queue feature introduced in the bnx2x 7.0.23 firmware and the FW likewise must have the bnx2x multi cos support. Signed-off-by: Ariel Elior <ariele@broadcom.com> Signed-off-by: Eilon Greenstein <eilong@broadcom.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2011-07-14 16:31:57 +08:00
struct eth_context *cxts[BNX2X_MULTI_TX_COS];
/* maximum number of cos supported by hardware */
u8 max_cos;
};
struct bnx2x_queue_terminate_params {
/* index within the tx_only cids of this queue object */
u8 cid_index;
};
struct bnx2x_queue_cfc_del_params {
/* index within the tx_only cids of this queue object */
u8 cid_index;
};
struct bnx2x_queue_update_params {
unsigned long update_flags; /* BNX2X_Q_UPDATE_XX bits */
u16 def_vlan;
u16 silent_removal_value;
u16 silent_removal_mask;
bnx2x: Multiple concurrent l2 traffic classes Overview: Support mapping of priorities to traffic classes and traffic classes to transmission queues ranges in the net device. The queue ranges are (count, offset) pairs relating to the txq array. This can be done via DCBX negotiation or by kernel. As a result Enhanced Transmission Selection (ETS) and Priority Flow Control (PFC) are supported between L2 network traffic classes. Mapping: This patch uses the netdev_set_num_tc, netdev_set_prio_tc_map and netdev_set_tc_queue functions to map priorities to traffic classes and traffic classes to transmission queue ranges. This mapping is performed by bnx2x_setup_tc function which is connected to the ndo_setup_tc. This function is always called at nic load where by default it maps all priorities to tc 0, and it may also be called by the kernel or by the bnx2x upon DCBX negotiation to modify the mapping. rtnl lock: When the ndo_setup_tc is called at nic load or by kernel the rtnl lock is already taken. However, when DCBX negotiation takes place the lock is not taken. The work is therefore scheduled to be handled by the sp_rtnl task. Fastpath: The fastpath structure of the bnx2x which was previously used to hold the information of one tx queue and one rx queue was redesigned to represent multiple tx queues, one for each traffic class. The transmission queue supplied in the skb by the kernel can no longer be interpreted as a straightforward index into the fastpath structure array, but it must rather be decoded to the appropriate fastpath index and the tc within that fastpath. Slowpath: The bnx2x's queue object was redesigned to accommodate multiple transmission queues. The queue object's state machine was enhanced to allow opening multiple transmission-only connections on top of the regular tx-rx connection. Firmware: This feature relies on the tx-only queue feature introduced in the bnx2x 7.0.23 firmware and the FW likewise must have the bnx2x multi cos support. Signed-off-by: Ariel Elior <ariele@broadcom.com> Signed-off-by: Eilon Greenstein <eilong@broadcom.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2011-07-14 16:31:57 +08:00
/* index within the tx_only cids of this queue object */
u8 cid_index;
};
struct rxq_pause_params {
u16 bd_th_lo;
u16 bd_th_hi;
u16 rcq_th_lo;
u16 rcq_th_hi;
u16 sge_th_lo; /* valid iff BNX2X_Q_FLG_TPA */
u16 sge_th_hi; /* valid iff BNX2X_Q_FLG_TPA */
u16 pri_map;
};
/* general */
struct bnx2x_general_setup_params {
/* valid iff BNX2X_Q_FLG_STATS */
u8 stat_id;
u8 spcl_id;
u16 mtu;
bnx2x: Multiple concurrent l2 traffic classes Overview: Support mapping of priorities to traffic classes and traffic classes to transmission queues ranges in the net device. The queue ranges are (count, offset) pairs relating to the txq array. This can be done via DCBX negotiation or by kernel. As a result Enhanced Transmission Selection (ETS) and Priority Flow Control (PFC) are supported between L2 network traffic classes. Mapping: This patch uses the netdev_set_num_tc, netdev_set_prio_tc_map and netdev_set_tc_queue functions to map priorities to traffic classes and traffic classes to transmission queue ranges. This mapping is performed by bnx2x_setup_tc function which is connected to the ndo_setup_tc. This function is always called at nic load where by default it maps all priorities to tc 0, and it may also be called by the kernel or by the bnx2x upon DCBX negotiation to modify the mapping. rtnl lock: When the ndo_setup_tc is called at nic load or by kernel the rtnl lock is already taken. However, when DCBX negotiation takes place the lock is not taken. The work is therefore scheduled to be handled by the sp_rtnl task. Fastpath: The fastpath structure of the bnx2x which was previously used to hold the information of one tx queue and one rx queue was redesigned to represent multiple tx queues, one for each traffic class. The transmission queue supplied in the skb by the kernel can no longer be interpreted as a straightforward index into the fastpath structure array, but it must rather be decoded to the appropriate fastpath index and the tc within that fastpath. Slowpath: The bnx2x's queue object was redesigned to accommodate multiple transmission queues. The queue object's state machine was enhanced to allow opening multiple transmission-only connections on top of the regular tx-rx connection. Firmware: This feature relies on the tx-only queue feature introduced in the bnx2x 7.0.23 firmware and the FW likewise must have the bnx2x multi cos support. Signed-off-by: Ariel Elior <ariele@broadcom.com> Signed-off-by: Eilon Greenstein <eilong@broadcom.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2011-07-14 16:31:57 +08:00
u8 cos;
};
struct bnx2x_rxq_setup_params {
/* dma */
dma_addr_t dscr_map;
dma_addr_t sge_map;
dma_addr_t rcq_map;
dma_addr_t rcq_np_map;
u16 drop_flags;
u16 buf_sz;
u8 fw_sb_id;
u8 cl_qzone_id;
/* valid iff BNX2X_Q_FLG_TPA */
u16 tpa_agg_sz;
u16 sge_buf_sz;
u8 max_sges_pkt;
u8 max_tpa_queues;
u8 rss_engine_id;
u8 cache_line_log;
u8 sb_cq_index;
/* valid iff BXN2X_Q_FLG_SILENT_VLAN_REM */
u16 silent_removal_value;
u16 silent_removal_mask;
};
struct bnx2x_txq_setup_params {
/* dma */
dma_addr_t dscr_map;
u8 fw_sb_id;
u8 sb_cq_index;
u8 cos; /* valid iff BNX2X_Q_FLG_COS */
u16 traffic_type;
/* equals to the leading rss client id, used for TX classification*/
u8 tss_leading_cl_id;
/* valid iff BNX2X_Q_FLG_DEF_VLAN */
u16 default_vlan;
};
struct bnx2x_queue_setup_params {
struct bnx2x_general_setup_params gen_params;
struct bnx2x_txq_setup_params txq_params;
bnx2x: Multiple concurrent l2 traffic classes Overview: Support mapping of priorities to traffic classes and traffic classes to transmission queues ranges in the net device. The queue ranges are (count, offset) pairs relating to the txq array. This can be done via DCBX negotiation or by kernel. As a result Enhanced Transmission Selection (ETS) and Priority Flow Control (PFC) are supported between L2 network traffic classes. Mapping: This patch uses the netdev_set_num_tc, netdev_set_prio_tc_map and netdev_set_tc_queue functions to map priorities to traffic classes and traffic classes to transmission queue ranges. This mapping is performed by bnx2x_setup_tc function which is connected to the ndo_setup_tc. This function is always called at nic load where by default it maps all priorities to tc 0, and it may also be called by the kernel or by the bnx2x upon DCBX negotiation to modify the mapping. rtnl lock: When the ndo_setup_tc is called at nic load or by kernel the rtnl lock is already taken. However, when DCBX negotiation takes place the lock is not taken. The work is therefore scheduled to be handled by the sp_rtnl task. Fastpath: The fastpath structure of the bnx2x which was previously used to hold the information of one tx queue and one rx queue was redesigned to represent multiple tx queues, one for each traffic class. The transmission queue supplied in the skb by the kernel can no longer be interpreted as a straightforward index into the fastpath structure array, but it must rather be decoded to the appropriate fastpath index and the tc within that fastpath. Slowpath: The bnx2x's queue object was redesigned to accommodate multiple transmission queues. The queue object's state machine was enhanced to allow opening multiple transmission-only connections on top of the regular tx-rx connection. Firmware: This feature relies on the tx-only queue feature introduced in the bnx2x 7.0.23 firmware and the FW likewise must have the bnx2x multi cos support. Signed-off-by: Ariel Elior <ariele@broadcom.com> Signed-off-by: Eilon Greenstein <eilong@broadcom.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2011-07-14 16:31:57 +08:00
struct bnx2x_rxq_setup_params rxq_params;
struct rxq_pause_params pause_params;
unsigned long flags;
};
bnx2x: Multiple concurrent l2 traffic classes Overview: Support mapping of priorities to traffic classes and traffic classes to transmission queues ranges in the net device. The queue ranges are (count, offset) pairs relating to the txq array. This can be done via DCBX negotiation or by kernel. As a result Enhanced Transmission Selection (ETS) and Priority Flow Control (PFC) are supported between L2 network traffic classes. Mapping: This patch uses the netdev_set_num_tc, netdev_set_prio_tc_map and netdev_set_tc_queue functions to map priorities to traffic classes and traffic classes to transmission queue ranges. This mapping is performed by bnx2x_setup_tc function which is connected to the ndo_setup_tc. This function is always called at nic load where by default it maps all priorities to tc 0, and it may also be called by the kernel or by the bnx2x upon DCBX negotiation to modify the mapping. rtnl lock: When the ndo_setup_tc is called at nic load or by kernel the rtnl lock is already taken. However, when DCBX negotiation takes place the lock is not taken. The work is therefore scheduled to be handled by the sp_rtnl task. Fastpath: The fastpath structure of the bnx2x which was previously used to hold the information of one tx queue and one rx queue was redesigned to represent multiple tx queues, one for each traffic class. The transmission queue supplied in the skb by the kernel can no longer be interpreted as a straightforward index into the fastpath structure array, but it must rather be decoded to the appropriate fastpath index and the tc within that fastpath. Slowpath: The bnx2x's queue object was redesigned to accommodate multiple transmission queues. The queue object's state machine was enhanced to allow opening multiple transmission-only connections on top of the regular tx-rx connection. Firmware: This feature relies on the tx-only queue feature introduced in the bnx2x 7.0.23 firmware and the FW likewise must have the bnx2x multi cos support. Signed-off-by: Ariel Elior <ariele@broadcom.com> Signed-off-by: Eilon Greenstein <eilong@broadcom.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2011-07-14 16:31:57 +08:00
struct bnx2x_queue_setup_tx_only_params {
struct bnx2x_general_setup_params gen_params;
struct bnx2x_txq_setup_params txq_params;
unsigned long flags;
/* index within the tx_only cids of this queue object */
u8 cid_index;
};
struct bnx2x_queue_state_params {
struct bnx2x_queue_sp_obj *q_obj;
/* Current command */
enum bnx2x_queue_cmd cmd;
/* may have RAMROD_COMP_WAIT set only */
bnx2x: Multiple concurrent l2 traffic classes Overview: Support mapping of priorities to traffic classes and traffic classes to transmission queues ranges in the net device. The queue ranges are (count, offset) pairs relating to the txq array. This can be done via DCBX negotiation or by kernel. As a result Enhanced Transmission Selection (ETS) and Priority Flow Control (PFC) are supported between L2 network traffic classes. Mapping: This patch uses the netdev_set_num_tc, netdev_set_prio_tc_map and netdev_set_tc_queue functions to map priorities to traffic classes and traffic classes to transmission queue ranges. This mapping is performed by bnx2x_setup_tc function which is connected to the ndo_setup_tc. This function is always called at nic load where by default it maps all priorities to tc 0, and it may also be called by the kernel or by the bnx2x upon DCBX negotiation to modify the mapping. rtnl lock: When the ndo_setup_tc is called at nic load or by kernel the rtnl lock is already taken. However, when DCBX negotiation takes place the lock is not taken. The work is therefore scheduled to be handled by the sp_rtnl task. Fastpath: The fastpath structure of the bnx2x which was previously used to hold the information of one tx queue and one rx queue was redesigned to represent multiple tx queues, one for each traffic class. The transmission queue supplied in the skb by the kernel can no longer be interpreted as a straightforward index into the fastpath structure array, but it must rather be decoded to the appropriate fastpath index and the tc within that fastpath. Slowpath: The bnx2x's queue object was redesigned to accommodate multiple transmission queues. The queue object's state machine was enhanced to allow opening multiple transmission-only connections on top of the regular tx-rx connection. Firmware: This feature relies on the tx-only queue feature introduced in the bnx2x 7.0.23 firmware and the FW likewise must have the bnx2x multi cos support. Signed-off-by: Ariel Elior <ariele@broadcom.com> Signed-off-by: Eilon Greenstein <eilong@broadcom.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2011-07-14 16:31:57 +08:00
unsigned long ramrod_flags;
/* Params according to the current command */
union {
bnx2x: Multiple concurrent l2 traffic classes Overview: Support mapping of priorities to traffic classes and traffic classes to transmission queues ranges in the net device. The queue ranges are (count, offset) pairs relating to the txq array. This can be done via DCBX negotiation or by kernel. As a result Enhanced Transmission Selection (ETS) and Priority Flow Control (PFC) are supported between L2 network traffic classes. Mapping: This patch uses the netdev_set_num_tc, netdev_set_prio_tc_map and netdev_set_tc_queue functions to map priorities to traffic classes and traffic classes to transmission queue ranges. This mapping is performed by bnx2x_setup_tc function which is connected to the ndo_setup_tc. This function is always called at nic load where by default it maps all priorities to tc 0, and it may also be called by the kernel or by the bnx2x upon DCBX negotiation to modify the mapping. rtnl lock: When the ndo_setup_tc is called at nic load or by kernel the rtnl lock is already taken. However, when DCBX negotiation takes place the lock is not taken. The work is therefore scheduled to be handled by the sp_rtnl task. Fastpath: The fastpath structure of the bnx2x which was previously used to hold the information of one tx queue and one rx queue was redesigned to represent multiple tx queues, one for each traffic class. The transmission queue supplied in the skb by the kernel can no longer be interpreted as a straightforward index into the fastpath structure array, but it must rather be decoded to the appropriate fastpath index and the tc within that fastpath. Slowpath: The bnx2x's queue object was redesigned to accommodate multiple transmission queues. The queue object's state machine was enhanced to allow opening multiple transmission-only connections on top of the regular tx-rx connection. Firmware: This feature relies on the tx-only queue feature introduced in the bnx2x 7.0.23 firmware and the FW likewise must have the bnx2x multi cos support. Signed-off-by: Ariel Elior <ariele@broadcom.com> Signed-off-by: Eilon Greenstein <eilong@broadcom.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2011-07-14 16:31:57 +08:00
struct bnx2x_queue_update_params update;
struct bnx2x_queue_setup_params setup;
struct bnx2x_queue_init_params init;
struct bnx2x_queue_setup_tx_only_params tx_only;
struct bnx2x_queue_terminate_params terminate;
struct bnx2x_queue_cfc_del_params cfc_del;
} params;
};
struct bnx2x_queue_sp_obj {
bnx2x: Multiple concurrent l2 traffic classes Overview: Support mapping of priorities to traffic classes and traffic classes to transmission queues ranges in the net device. The queue ranges are (count, offset) pairs relating to the txq array. This can be done via DCBX negotiation or by kernel. As a result Enhanced Transmission Selection (ETS) and Priority Flow Control (PFC) are supported between L2 network traffic classes. Mapping: This patch uses the netdev_set_num_tc, netdev_set_prio_tc_map and netdev_set_tc_queue functions to map priorities to traffic classes and traffic classes to transmission queue ranges. This mapping is performed by bnx2x_setup_tc function which is connected to the ndo_setup_tc. This function is always called at nic load where by default it maps all priorities to tc 0, and it may also be called by the kernel or by the bnx2x upon DCBX negotiation to modify the mapping. rtnl lock: When the ndo_setup_tc is called at nic load or by kernel the rtnl lock is already taken. However, when DCBX negotiation takes place the lock is not taken. The work is therefore scheduled to be handled by the sp_rtnl task. Fastpath: The fastpath structure of the bnx2x which was previously used to hold the information of one tx queue and one rx queue was redesigned to represent multiple tx queues, one for each traffic class. The transmission queue supplied in the skb by the kernel can no longer be interpreted as a straightforward index into the fastpath structure array, but it must rather be decoded to the appropriate fastpath index and the tc within that fastpath. Slowpath: The bnx2x's queue object was redesigned to accommodate multiple transmission queues. The queue object's state machine was enhanced to allow opening multiple transmission-only connections on top of the regular tx-rx connection. Firmware: This feature relies on the tx-only queue feature introduced in the bnx2x 7.0.23 firmware and the FW likewise must have the bnx2x multi cos support. Signed-off-by: Ariel Elior <ariele@broadcom.com> Signed-off-by: Eilon Greenstein <eilong@broadcom.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2011-07-14 16:31:57 +08:00
u32 cids[BNX2X_MULTI_TX_COS];
u8 cl_id;
u8 func_id;
bnx2x: Multiple concurrent l2 traffic classes Overview: Support mapping of priorities to traffic classes and traffic classes to transmission queues ranges in the net device. The queue ranges are (count, offset) pairs relating to the txq array. This can be done via DCBX negotiation or by kernel. As a result Enhanced Transmission Selection (ETS) and Priority Flow Control (PFC) are supported between L2 network traffic classes. Mapping: This patch uses the netdev_set_num_tc, netdev_set_prio_tc_map and netdev_set_tc_queue functions to map priorities to traffic classes and traffic classes to transmission queue ranges. This mapping is performed by bnx2x_setup_tc function which is connected to the ndo_setup_tc. This function is always called at nic load where by default it maps all priorities to tc 0, and it may also be called by the kernel or by the bnx2x upon DCBX negotiation to modify the mapping. rtnl lock: When the ndo_setup_tc is called at nic load or by kernel the rtnl lock is already taken. However, when DCBX negotiation takes place the lock is not taken. The work is therefore scheduled to be handled by the sp_rtnl task. Fastpath: The fastpath structure of the bnx2x which was previously used to hold the information of one tx queue and one rx queue was redesigned to represent multiple tx queues, one for each traffic class. The transmission queue supplied in the skb by the kernel can no longer be interpreted as a straightforward index into the fastpath structure array, but it must rather be decoded to the appropriate fastpath index and the tc within that fastpath. Slowpath: The bnx2x's queue object was redesigned to accommodate multiple transmission queues. The queue object's state machine was enhanced to allow opening multiple transmission-only connections on top of the regular tx-rx connection. Firmware: This feature relies on the tx-only queue feature introduced in the bnx2x 7.0.23 firmware and the FW likewise must have the bnx2x multi cos support. Signed-off-by: Ariel Elior <ariele@broadcom.com> Signed-off-by: Eilon Greenstein <eilong@broadcom.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2011-07-14 16:31:57 +08:00
/*
* number of traffic classes supported by queue.
* The primary connection of the queue suppotrs the first traffic
* class. Any further traffic class is suppoted by a tx-only
* connection.
*
* Therefore max_cos is also a number of valid entries in the cids
* array.
*/
u8 max_cos;
u8 num_tx_only, next_tx_only;
enum bnx2x_q_state state, next_state;
/* bits from enum bnx2x_q_type */
unsigned long type;
/* BNX2X_Q_CMD_XX bits. This object implements "one
* pending" paradigm but for debug and tracing purposes it's
* more convinient to have different bits for different
* commands.
*/
unsigned long pending;
/* Buffer to use as a ramrod data and its mapping */
void *rdata;
dma_addr_t rdata_mapping;
/**
* Performs one state change according to the given parameters.
*
* @return 0 in case of success and negative value otherwise.
*/
int (*send_cmd)(struct bnx2x *bp,
struct bnx2x_queue_state_params *params);
/**
* Sets the pending bit according to the requested transition.
*/
int (*set_pending)(struct bnx2x_queue_sp_obj *o,
struct bnx2x_queue_state_params *params);
/**
* Checks that the requested state transition is legal.
*/
int (*check_transition)(struct bnx2x *bp,
struct bnx2x_queue_sp_obj *o,
struct bnx2x_queue_state_params *params);
/**
* Completes the pending command.
*/
int (*complete_cmd)(struct bnx2x *bp,
struct bnx2x_queue_sp_obj *o,
enum bnx2x_queue_cmd);
int (*wait_comp)(struct bnx2x *bp,
struct bnx2x_queue_sp_obj *o,
enum bnx2x_queue_cmd cmd);
};
/********************** Function state update *********************************/
/* Allowed Function states */
enum bnx2x_func_state {
BNX2X_F_STATE_RESET,
BNX2X_F_STATE_INITIALIZED,
BNX2X_F_STATE_STARTED,
BNX2X_F_STATE_TX_STOPPED,
BNX2X_F_STATE_MAX,
};
/* Allowed Function commands */
enum bnx2x_func_cmd {
BNX2X_F_CMD_HW_INIT,
BNX2X_F_CMD_START,
BNX2X_F_CMD_STOP,
BNX2X_F_CMD_HW_RESET,
BNX2X_F_CMD_TX_STOP,
BNX2X_F_CMD_TX_START,
BNX2X_F_CMD_MAX,
};
struct bnx2x_func_hw_init_params {
/* A load phase returned by MCP.
*
* May be:
* FW_MSG_CODE_DRV_LOAD_COMMON_CHIP
* FW_MSG_CODE_DRV_LOAD_COMMON
* FW_MSG_CODE_DRV_LOAD_PORT
* FW_MSG_CODE_DRV_LOAD_FUNCTION
*/
u32 load_phase;
};
struct bnx2x_func_hw_reset_params {
/* A load phase returned by MCP.
*
* May be:
* FW_MSG_CODE_DRV_LOAD_COMMON_CHIP
* FW_MSG_CODE_DRV_LOAD_COMMON
* FW_MSG_CODE_DRV_LOAD_PORT
* FW_MSG_CODE_DRV_LOAD_FUNCTION
*/
u32 reset_phase;
};
struct bnx2x_func_start_params {
/* Multi Function mode:
* - Single Function
* - Switch Dependent
* - Switch Independent
*/
u16 mf_mode;
/* Switch Dependent mode outer VLAN tag */
u16 sd_vlan_tag;
/* Function cos mode */
u8 network_cos_mode;
};
struct bnx2x_func_tx_start_params {
struct priority_cos traffic_type_to_priority_cos[MAX_TRAFFIC_TYPES];
u8 dcb_enabled;
u8 dcb_version;
u8 dont_add_pri_0_en;
};
struct bnx2x_func_state_params {
struct bnx2x_func_sp_obj *f_obj;
/* Current command */
enum bnx2x_func_cmd cmd;
/* may have RAMROD_COMP_WAIT set only */
unsigned long ramrod_flags;
/* Params according to the current command */
union {
struct bnx2x_func_hw_init_params hw_init;
struct bnx2x_func_hw_reset_params hw_reset;
struct bnx2x_func_start_params start;
struct bnx2x_func_tx_start_params tx_start;
} params;
};
struct bnx2x_func_sp_drv_ops {
/* Init tool + runtime initialization:
* - Common Chip
* - Common (per Path)
* - Port
* - Function phases
*/
int (*init_hw_cmn_chip)(struct bnx2x *bp);
int (*init_hw_cmn)(struct bnx2x *bp);
int (*init_hw_port)(struct bnx2x *bp);
int (*init_hw_func)(struct bnx2x *bp);
/* Reset Function HW: Common, Port, Function phases. */
void (*reset_hw_cmn)(struct bnx2x *bp);
void (*reset_hw_port)(struct bnx2x *bp);
void (*reset_hw_func)(struct bnx2x *bp);
/* Init/Free GUNZIP resources */
int (*gunzip_init)(struct bnx2x *bp);
void (*gunzip_end)(struct bnx2x *bp);
/* Prepare/Release FW resources */
int (*init_fw)(struct bnx2x *bp);
void (*release_fw)(struct bnx2x *bp);
};
struct bnx2x_func_sp_obj {
enum bnx2x_func_state state, next_state;
/* BNX2X_FUNC_CMD_XX bits. This object implements "one
* pending" paradigm but for debug and tracing purposes it's
* more convinient to have different bits for different
* commands.
*/
unsigned long pending;
/* Buffer to use as a ramrod data and its mapping */
void *rdata;
dma_addr_t rdata_mapping;
/* this mutex validates that when pending flag is taken, the next
* ramrod to be sent will be the one set the pending bit
*/
struct mutex one_pending_mutex;
/* Driver interface */
struct bnx2x_func_sp_drv_ops *drv;
/**
* Performs one state change according to the given parameters.
*
* @return 0 in case of success and negative value otherwise.
*/
int (*send_cmd)(struct bnx2x *bp,
struct bnx2x_func_state_params *params);
/**
* Checks that the requested state transition is legal.
*/
int (*check_transition)(struct bnx2x *bp,
struct bnx2x_func_sp_obj *o,
struct bnx2x_func_state_params *params);
/**
* Completes the pending command.
*/
int (*complete_cmd)(struct bnx2x *bp,
struct bnx2x_func_sp_obj *o,
enum bnx2x_func_cmd cmd);
int (*wait_comp)(struct bnx2x *bp, struct bnx2x_func_sp_obj *o,
enum bnx2x_func_cmd cmd);
};
/********************** Interfaces ********************************************/
/* Queueable objects set */
union bnx2x_qable_obj {
struct bnx2x_vlan_mac_obj vlan_mac;
};
/************** Function state update *********/
void bnx2x_init_func_obj(struct bnx2x *bp,
struct bnx2x_func_sp_obj *obj,
void *rdata, dma_addr_t rdata_mapping,
struct bnx2x_func_sp_drv_ops *drv_iface);
int bnx2x_func_state_change(struct bnx2x *bp,
struct bnx2x_func_state_params *params);
enum bnx2x_func_state bnx2x_func_get_state(struct bnx2x *bp,
struct bnx2x_func_sp_obj *o);
/******************* Queue State **************/
void bnx2x_init_queue_obj(struct bnx2x *bp,
bnx2x: Multiple concurrent l2 traffic classes Overview: Support mapping of priorities to traffic classes and traffic classes to transmission queues ranges in the net device. The queue ranges are (count, offset) pairs relating to the txq array. This can be done via DCBX negotiation or by kernel. As a result Enhanced Transmission Selection (ETS) and Priority Flow Control (PFC) are supported between L2 network traffic classes. Mapping: This patch uses the netdev_set_num_tc, netdev_set_prio_tc_map and netdev_set_tc_queue functions to map priorities to traffic classes and traffic classes to transmission queue ranges. This mapping is performed by bnx2x_setup_tc function which is connected to the ndo_setup_tc. This function is always called at nic load where by default it maps all priorities to tc 0, and it may also be called by the kernel or by the bnx2x upon DCBX negotiation to modify the mapping. rtnl lock: When the ndo_setup_tc is called at nic load or by kernel the rtnl lock is already taken. However, when DCBX negotiation takes place the lock is not taken. The work is therefore scheduled to be handled by the sp_rtnl task. Fastpath: The fastpath structure of the bnx2x which was previously used to hold the information of one tx queue and one rx queue was redesigned to represent multiple tx queues, one for each traffic class. The transmission queue supplied in the skb by the kernel can no longer be interpreted as a straightforward index into the fastpath structure array, but it must rather be decoded to the appropriate fastpath index and the tc within that fastpath. Slowpath: The bnx2x's queue object was redesigned to accommodate multiple transmission queues. The queue object's state machine was enhanced to allow opening multiple transmission-only connections on top of the regular tx-rx connection. Firmware: This feature relies on the tx-only queue feature introduced in the bnx2x 7.0.23 firmware and the FW likewise must have the bnx2x multi cos support. Signed-off-by: Ariel Elior <ariele@broadcom.com> Signed-off-by: Eilon Greenstein <eilong@broadcom.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2011-07-14 16:31:57 +08:00
struct bnx2x_queue_sp_obj *obj, u8 cl_id, u32 *cids,
u8 cid_cnt, u8 func_id, void *rdata,
dma_addr_t rdata_mapping, unsigned long type);
int bnx2x_queue_state_change(struct bnx2x *bp,
struct bnx2x_queue_state_params *params);
/********************* VLAN-MAC ****************/
void bnx2x_init_mac_obj(struct bnx2x *bp,
struct bnx2x_vlan_mac_obj *mac_obj,
u8 cl_id, u32 cid, u8 func_id, void *rdata,
dma_addr_t rdata_mapping, int state,
unsigned long *pstate, bnx2x_obj_type type,
struct bnx2x_credit_pool_obj *macs_pool);
void bnx2x_init_vlan_obj(struct bnx2x *bp,
struct bnx2x_vlan_mac_obj *vlan_obj,
u8 cl_id, u32 cid, u8 func_id, void *rdata,
dma_addr_t rdata_mapping, int state,
unsigned long *pstate, bnx2x_obj_type type,
struct bnx2x_credit_pool_obj *vlans_pool);
void bnx2x_init_vlan_mac_obj(struct bnx2x *bp,
struct bnx2x_vlan_mac_obj *vlan_mac_obj,
u8 cl_id, u32 cid, u8 func_id, void *rdata,
dma_addr_t rdata_mapping, int state,
unsigned long *pstate, bnx2x_obj_type type,
struct bnx2x_credit_pool_obj *macs_pool,
struct bnx2x_credit_pool_obj *vlans_pool);
int bnx2x_config_vlan_mac(struct bnx2x *bp,
struct bnx2x_vlan_mac_ramrod_params *p);
int bnx2x_vlan_mac_move(struct bnx2x *bp,
struct bnx2x_vlan_mac_ramrod_params *p,
struct bnx2x_vlan_mac_obj *dest_o);
/********************* RX MODE ****************/
void bnx2x_init_rx_mode_obj(struct bnx2x *bp,
struct bnx2x_rx_mode_obj *o);
/**
* Send and RX_MODE ramrod according to the provided parameters.
*
* @param bp
* @param p Command parameters
*
* @return 0 - if operation was successfull and there is no pending completions,
* positive number - if there are pending completions,
* negative - if there were errors
*/
int bnx2x_config_rx_mode(struct bnx2x *bp,
struct bnx2x_rx_mode_ramrod_params *p);
/****************** MULTICASTS ****************/
void bnx2x_init_mcast_obj(struct bnx2x *bp,
struct bnx2x_mcast_obj *mcast_obj,
u8 mcast_cl_id, u32 mcast_cid, u8 func_id,
u8 engine_id, void *rdata, dma_addr_t rdata_mapping,
int state, unsigned long *pstate,
bnx2x_obj_type type);
/**
* Configure multicast MACs list. May configure a new list
* provided in p->mcast_list (BNX2X_MCAST_CMD_ADD), clean up
* (BNX2X_MCAST_CMD_DEL) or restore (BNX2X_MCAST_CMD_RESTORE) a current
* configuration, continue to execute the pending commands
* (BNX2X_MCAST_CMD_CONT).
*
* If previous command is still pending or if number of MACs to
* configure is more that maximum number of MACs in one command,
* the current command will be enqueued to the tail of the
* pending commands list.
*
* @param bp
* @param p
* @param command to execute: BNX2X_MCAST_CMD_X
*
* @return 0 is operation was sucessfull and there are no pending completions,
* negative if there were errors, positive if there are pending
* completions.
*/
int bnx2x_config_mcast(struct bnx2x *bp,
struct bnx2x_mcast_ramrod_params *p, int cmd);
/****************** CREDIT POOL ****************/
void bnx2x_init_mac_credit_pool(struct bnx2x *bp,
struct bnx2x_credit_pool_obj *p, u8 func_id,
u8 func_num);
void bnx2x_init_vlan_credit_pool(struct bnx2x *bp,
struct bnx2x_credit_pool_obj *p, u8 func_id,
u8 func_num);
/****************** RSS CONFIGURATION ****************/
void bnx2x_init_rss_config_obj(struct bnx2x *bp,
struct bnx2x_rss_config_obj *rss_obj,
u8 cl_id, u32 cid, u8 func_id, u8 engine_id,
void *rdata, dma_addr_t rdata_mapping,
int state, unsigned long *pstate,
bnx2x_obj_type type);
/**
* Updates RSS configuration according to provided parameters.
*
* @param bp
* @param p
*
* @return 0 in case of success
*/
int bnx2x_config_rss(struct bnx2x *bp,
struct bnx2x_config_rss_params *p);
/**
* Return the current ind_table configuration.
*
* @param bp
* @param ind_table buffer to fill with the current indirection
* table content. Should be at least
* T_ETH_INDIRECTION_TABLE_SIZE bytes long.
*/
void bnx2x_get_rss_ind_table(struct bnx2x_rss_config_obj *rss_obj,
u8 *ind_table);
#endif /* BNX2X_SP_VERBS */