linux/drivers/net/ethernet/broadcom/bnx2x/bnx2x_cmn.h

1402 lines
34 KiB
C
Raw Normal View History

/* bnx2x_cmn.h: Broadcom Everest network driver.
*
* Copyright (c) 2007-2012 Broadcom Corporation
*
* This program is free software; you can redistribute it and/or modify
* it under the terms of the GNU General Public License as published by
* the Free Software Foundation.
*
* Maintained by: Eilon Greenstein <eilong@broadcom.com>
* Written by: Eliezer Tamir
* Based on code from Michael Chan's bnx2 driver
* UDP CSUM errata workaround by Arik Gendelman
* Slowpath and fastpath rework by Vladislav Zolotarov
* Statistics and Link management by Yitchak Gertner
*
*/
#ifndef BNX2X_CMN_H
#define BNX2X_CMN_H
#include <linux/types.h>
#include <linux/pci.h>
#include <linux/netdevice.h>
#include <linux/etherdevice.h>
#include "bnx2x.h"
/* This is used as a replacement for an MCP if it's not present */
extern int load_count[2][3]; /* per-path: 0-common, 1-port0, 2-port1 */
extern int num_queues;
extern int int_mode;
/************************ Macros ********************************/
#define BNX2X_PCI_FREE(x, y, size) \
do { \
if (x) { \
dma_free_coherent(&bp->pdev->dev, size, (void *)x, y); \
x = NULL; \
y = 0; \
} \
} while (0)
#define BNX2X_FREE(x) \
do { \
if (x) { \
kfree((void *)x); \
x = NULL; \
} \
} while (0)
#define BNX2X_PCI_ALLOC(x, y, size) \
do { \
x = dma_alloc_coherent(&bp->pdev->dev, size, y, GFP_KERNEL); \
if (x == NULL) \
goto alloc_mem_err; \
memset((void *)x, 0, size); \
} while (0)
#define BNX2X_ALLOC(x, size) \
do { \
x = kzalloc(size, GFP_KERNEL); \
if (x == NULL) \
goto alloc_mem_err; \
} while (0)
/*********************** Interfaces ****************************
* Functions that need to be implemented by each driver version
*/
/* Init */
/**
* bnx2x_send_unload_req - request unload mode from the MCP.
*
* @bp: driver handle
* @unload_mode: requested function's unload mode
*
* Return unload mode returned by the MCP: COMMON, PORT or FUNC.
*/
u32 bnx2x_send_unload_req(struct bnx2x *bp, int unload_mode);
/**
* bnx2x_send_unload_done - send UNLOAD_DONE command to the MCP.
*
* @bp: driver handle
* @keep_link: true iff link should be kept up
*/
void bnx2x_send_unload_done(struct bnx2x *bp, bool keep_link);
/**
* bnx2x_config_rss_pf - configure RSS parameters in a PF.
*
* @bp: driver handle
* @rss_obj: RSS object to use
* @ind_table: indirection table to configure
* @config_hash: re-configure RSS hash keys configuration
*/
int bnx2x_config_rss_pf(struct bnx2x *bp, struct bnx2x_rss_config_obj *rss_obj,
bool config_hash);
/**
* bnx2x__init_func_obj - init function object
*
* @bp: driver handle
*
* Initializes the Function Object with the appropriate
* parameters which include a function slow path driver
* interface.
*/
void bnx2x__init_func_obj(struct bnx2x *bp);
/**
* bnx2x_setup_queue - setup eth queue.
*
* @bp: driver handle
* @fp: pointer to the fastpath structure
* @leading: boolean
*
*/
int bnx2x_setup_queue(struct bnx2x *bp, struct bnx2x_fastpath *fp,
bool leading);
/**
* bnx2x_setup_leading - bring up a leading eth queue.
*
* @bp: driver handle
*/
int bnx2x_setup_leading(struct bnx2x *bp);
/**
* bnx2x_fw_command - send the MCP a request
*
* @bp: driver handle
* @command: request
* @param: request's parameter
*
* block until there is a reply
*/
u32 bnx2x_fw_command(struct bnx2x *bp, u32 command, u32 param);
/**
* bnx2x_initial_phy_init - initialize link parameters structure variables.
*
* @bp: driver handle
* @load_mode: current mode
*/
int bnx2x_initial_phy_init(struct bnx2x *bp, int load_mode);
/**
* bnx2x_link_set - configure hw according to link parameters structure.
*
* @bp: driver handle
*/
void bnx2x_link_set(struct bnx2x *bp);
/**
* bnx2x_force_link_reset - Forces link reset, and put the PHY
* in reset as well.
*
* @bp: driver handle
*/
void bnx2x_force_link_reset(struct bnx2x *bp);
/**
* bnx2x_link_test - query link status.
*
* @bp: driver handle
* @is_serdes: bool
*
* Returns 0 if link is UP.
*/
u8 bnx2x_link_test(struct bnx2x *bp, u8 is_serdes);
/**
* bnx2x_drv_pulse - write driver pulse to shmem
*
* @bp: driver handle
*
* writes the value in bp->fw_drv_pulse_wr_seq to drv_pulse mbox
* in the shmem.
*/
void bnx2x_drv_pulse(struct bnx2x *bp);
/**
* bnx2x_igu_ack_sb - update IGU with current SB value
*
* @bp: driver handle
* @igu_sb_id: SB id
* @segment: SB segment
* @index: SB index
* @op: SB operation
* @update: is HW update required
*/
void bnx2x_igu_ack_sb(struct bnx2x *bp, u8 igu_sb_id, u8 segment,
u16 index, u8 op, u8 update);
/* Disable transactions from chip to host */
void bnx2x_pf_disable(struct bnx2x *bp);
/**
* bnx2x__link_status_update - handles link status change.
*
* @bp: driver handle
*/
void bnx2x__link_status_update(struct bnx2x *bp);
/**
* bnx2x_link_report - report link status to upper layer.
*
* @bp: driver handle
*/
void bnx2x_link_report(struct bnx2x *bp);
/* None-atomic version of bnx2x_link_report() */
void __bnx2x_link_report(struct bnx2x *bp);
/**
* bnx2x_get_mf_speed - calculate MF speed.
*
* @bp: driver handle
*
* Takes into account current linespeed and MF configuration.
*/
u16 bnx2x_get_mf_speed(struct bnx2x *bp);
/**
* bnx2x_msix_sp_int - MSI-X slowpath interrupt handler
*
* @irq: irq number
* @dev_instance: private instance
*/
irqreturn_t bnx2x_msix_sp_int(int irq, void *dev_instance);
/**
* bnx2x_interrupt - non MSI-X interrupt handler
*
* @irq: irq number
* @dev_instance: private instance
*/
irqreturn_t bnx2x_interrupt(int irq, void *dev_instance);
/**
* bnx2x_cnic_notify - send command to cnic driver
*
* @bp: driver handle
* @cmd: command
*/
int bnx2x_cnic_notify(struct bnx2x *bp, int cmd);
/**
* bnx2x_setup_cnic_irq_info - provides cnic with IRQ information
*
* @bp: driver handle
*/
void bnx2x_setup_cnic_irq_info(struct bnx2x *bp);
/**
* bnx2x_setup_cnic_info - provides cnic with updated info
*
* @bp: driver handle
*/
void bnx2x_setup_cnic_info(struct bnx2x *bp);
/**
* bnx2x_int_enable - enable HW interrupts.
*
* @bp: driver handle
*/
void bnx2x_int_enable(struct bnx2x *bp);
/**
* bnx2x_int_disable_sync - disable interrupts.
*
* @bp: driver handle
* @disable_hw: true, disable HW interrupts.
*
* This function ensures that there are no
* ISRs or SP DPCs (sp_task) are running after it returns.
*/
void bnx2x_int_disable_sync(struct bnx2x *bp, int disable_hw);
/**
* bnx2x_nic_init_cnic - init driver internals for cnic.
*
* @bp: driver handle
* @load_code: COMMON, PORT or FUNCTION
*
* Initializes:
* - rings
* - status blocks
* - etc.
*/
void bnx2x_nic_init_cnic(struct bnx2x *bp);
/**
* bnx2x_nic_init - init driver internals.
*
* @bp: driver handle
*
* Initializes:
* - rings
* - status blocks
* - etc.
*/
void bnx2x_nic_init(struct bnx2x *bp, u32 load_code);
/**
* bnx2x_alloc_mem_cnic - allocate driver's memory for cnic.
*
* @bp: driver handle
*/
int bnx2x_alloc_mem_cnic(struct bnx2x *bp);
/**
* bnx2x_alloc_mem - allocate driver's memory.
*
* @bp: driver handle
*/
int bnx2x_alloc_mem(struct bnx2x *bp);
/**
* bnx2x_free_mem_cnic - release driver's memory for cnic.
*
* @bp: driver handle
*/
void bnx2x_free_mem_cnic(struct bnx2x *bp);
/**
* bnx2x_free_mem - release driver's memory.
*
* @bp: driver handle
*/
void bnx2x_free_mem(struct bnx2x *bp);
/**
* bnx2x_set_num_queues - set number of queues according to mode.
*
* @bp: driver handle
*/
void bnx2x_set_num_queues(struct bnx2x *bp);
/**
* bnx2x_chip_cleanup - cleanup chip internals.
*
* @bp: driver handle
* @unload_mode: COMMON, PORT, FUNCTION
* @keep_link: true iff link should be kept up.
*
* - Cleanup MAC configuration.
* - Closes clients.
* - etc.
*/
void bnx2x_chip_cleanup(struct bnx2x *bp, int unload_mode, bool keep_link);
/**
* bnx2x_acquire_hw_lock - acquire HW lock.
*
* @bp: driver handle
* @resource: resource bit which was locked
*/
int bnx2x_acquire_hw_lock(struct bnx2x *bp, u32 resource);
/**
* bnx2x_release_hw_lock - release HW lock.
*
* @bp: driver handle
* @resource: resource bit which was locked
*/
int bnx2x_release_hw_lock(struct bnx2x *bp, u32 resource);
/**
* bnx2x_release_leader_lock - release recovery leader lock
*
* @bp: driver handle
*/
int bnx2x_release_leader_lock(struct bnx2x *bp);
/**
* bnx2x_set_eth_mac - configure eth MAC address in the HW
*
* @bp: driver handle
* @set: set or clear
*
* Configures according to the value in netdev->dev_addr.
*/
int bnx2x_set_eth_mac(struct bnx2x *bp, bool set);
/**
* bnx2x_set_rx_mode - set MAC filtering configurations.
*
* @dev: netdevice
*
* called with netif_tx_lock from dev_mcast.c
* If bp->state is OPEN, should be called with
* netif_addr_lock_bh()
*/
void bnx2x_set_rx_mode(struct net_device *dev);
/**
* bnx2x_set_storm_rx_mode - configure MAC filtering rules in a FW.
*
* @bp: driver handle
*
* If bp->state is OPEN, should be called with
* netif_addr_lock_bh().
*/
void bnx2x_set_storm_rx_mode(struct bnx2x *bp);
/**
* bnx2x_set_q_rx_mode - configures rx_mode for a single queue.
*
* @bp: driver handle
* @cl_id: client id
* @rx_mode_flags: rx mode configuration
* @rx_accept_flags: rx accept configuration
* @tx_accept_flags: tx accept configuration (tx switch)
* @ramrod_flags: ramrod configuration
*/
void bnx2x_set_q_rx_mode(struct bnx2x *bp, u8 cl_id,
unsigned long rx_mode_flags,
unsigned long rx_accept_flags,
unsigned long tx_accept_flags,
unsigned long ramrod_flags);
/* Parity errors related */
void bnx2x_set_pf_load(struct bnx2x *bp);
bool bnx2x_clear_pf_load(struct bnx2x *bp);
bool bnx2x_chk_parity_attn(struct bnx2x *bp, bool *global, bool print);
bool bnx2x_reset_is_done(struct bnx2x *bp, int engine);
void bnx2x_set_reset_in_progress(struct bnx2x *bp);
void bnx2x_set_reset_global(struct bnx2x *bp);
void bnx2x_disable_close_the_gate(struct bnx2x *bp);
int bnx2x_init_hw_func_cnic(struct bnx2x *bp);
/**
* bnx2x_sp_event - handle ramrods completion.
*
* @fp: fastpath handle for the event
* @rr_cqe: eth_rx_cqe
*/
void bnx2x_sp_event(struct bnx2x_fastpath *fp, union eth_rx_cqe *rr_cqe);
/**
* bnx2x_ilt_set_info - prepare ILT configurations.
*
* @bp: driver handle
*/
void bnx2x_ilt_set_info(struct bnx2x *bp);
/**
* bnx2x_ilt_set_cnic_info - prepare ILT configurations for SRC
* and TM.
*
* @bp: driver handle
*/
void bnx2x_ilt_set_info_cnic(struct bnx2x *bp);
/**
* bnx2x_dcbx_init - initialize dcbx protocol.
*
* @bp: driver handle
*/
void bnx2x_dcbx_init(struct bnx2x *bp, bool update_shmem);
/**
* bnx2x_set_power_state - set power state to the requested value.
*
* @bp: driver handle
* @state: required state D0 or D3hot
*
* Currently only D0 and D3hot are supported.
*/
int bnx2x_set_power_state(struct bnx2x *bp, pci_power_t state);
/**
* bnx2x_update_max_mf_config - update MAX part of MF configuration in HW.
*
* @bp: driver handle
* @value: new value
*/
void bnx2x_update_max_mf_config(struct bnx2x *bp, u32 value);
/* Error handling */
void bnx2x_panic_dump(struct bnx2x *bp);
void bnx2x_fw_dump_lvl(struct bnx2x *bp, const char *lvl);
/* validate currect fw is loaded */
bool bnx2x_test_firmware_version(struct bnx2x *bp, bool is_err);
/* dev_close main block */
int bnx2x_nic_unload(struct bnx2x *bp, int unload_mode, bool keep_link);
/* dev_open main block */
int bnx2x_nic_load(struct bnx2x *bp, int load_mode);
/* hard_xmit callback */
netdev_tx_t bnx2x_start_xmit(struct sk_buff *skb, struct net_device *dev);
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 08:31:57 +00:00
/* setup_tc callback */
int bnx2x_setup_tc(struct net_device *dev, u8 num_tc);
/* select_queue callback */
u16 bnx2x_select_queue(struct net_device *dev, struct sk_buff *skb);
static inline void bnx2x_update_rx_prod(struct bnx2x *bp,
struct bnx2x_fastpath *fp,
u16 bd_prod, u16 rx_comp_prod,
u16 rx_sge_prod)
{
struct ustorm_eth_rx_producers rx_prods = {0};
u32 i;
/* Update producers */
rx_prods.bd_prod = bd_prod;
rx_prods.cqe_prod = rx_comp_prod;
rx_prods.sge_prod = rx_sge_prod;
/* Make sure that the BD and SGE data is updated before updating the
* producers since FW might read the BD/SGE right after the producer
* is updated.
* This is only applicable for weak-ordered memory model archs such
* as IA-64. The following barrier is also mandatory since FW will
* assumes BDs must have buffers.
*/
wmb();
for (i = 0; i < sizeof(rx_prods)/4; i++)
REG_WR(bp, fp->ustorm_rx_prods_offset + i*4,
((u32 *)&rx_prods)[i]);
mmiowb(); /* keep prod updates ordered */
DP(NETIF_MSG_RX_STATUS,
"queue[%d]: wrote bd_prod %u cqe_prod %u sge_prod %u\n",
fp->index, bd_prod, rx_comp_prod, rx_sge_prod);
}
/* reload helper */
int bnx2x_reload_if_running(struct net_device *dev);
int bnx2x_change_mac_addr(struct net_device *dev, void *p);
/* NAPI poll Rx part */
int bnx2x_rx_int(struct bnx2x_fastpath *fp, int budget);
/* NAPI poll Tx part */
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 08:31:57 +00:00
int bnx2x_tx_int(struct bnx2x *bp, struct bnx2x_fp_txdata *txdata);
/* suspend/resume callbacks */
int bnx2x_suspend(struct pci_dev *pdev, pm_message_t state);
int bnx2x_resume(struct pci_dev *pdev);
/* Release IRQ vectors */
void bnx2x_free_irq(struct bnx2x *bp);
void bnx2x_free_fp_mem_cnic(struct bnx2x *bp);
void bnx2x_free_fp_mem(struct bnx2x *bp);
int bnx2x_alloc_fp_mem_cnic(struct bnx2x *bp);
int bnx2x_alloc_fp_mem(struct bnx2x *bp);
void bnx2x_init_rx_rings(struct bnx2x *bp);
void bnx2x_init_rx_rings_cnic(struct bnx2x *bp);
void bnx2x_free_skbs_cnic(struct bnx2x *bp);
void bnx2x_free_skbs(struct bnx2x *bp);
void bnx2x_netif_stop(struct bnx2x *bp, int disable_hw);
void bnx2x_netif_start(struct bnx2x *bp);
int bnx2x_load_cnic(struct bnx2x *bp);
/**
* bnx2x_enable_msix - set msix configuration.
*
* @bp: driver handle
*
* fills msix_table, requests vectors, updates num_queues
* according to number of available vectors.
*/
int bnx2x_enable_msix(struct bnx2x *bp);
/**
* bnx2x_enable_msi - request msi mode from OS, updated internals accordingly
*
* @bp: driver handle
*/
int bnx2x_enable_msi(struct bnx2x *bp);
/**
* bnx2x_poll - NAPI callback
*
* @napi: napi structure
* @budget:
*
*/
int bnx2x_poll(struct napi_struct *napi, int budget);
/**
* bnx2x_alloc_mem_bp - allocate memories outsize main driver structure
*
* @bp: driver handle
*/
int bnx2x_alloc_mem_bp(struct bnx2x *bp);
/**
* bnx2x_free_mem_bp - release memories outsize main driver structure
*
* @bp: driver handle
*/
void bnx2x_free_mem_bp(struct bnx2x *bp);
/**
* bnx2x_change_mtu - change mtu netdev callback
*
* @dev: net device
* @new_mtu: requested mtu
*
*/
int bnx2x_change_mtu(struct net_device *dev, int new_mtu);
#ifdef NETDEV_FCOE_WWNN
/**
* bnx2x_fcoe_get_wwn - return the requested WWN value for this port
*
* @dev: net_device
* @wwn: output buffer
* @type: WWN type: NETDEV_FCOE_WWNN (node) or NETDEV_FCOE_WWPN (port)
*
*/
int bnx2x_fcoe_get_wwn(struct net_device *dev, u64 *wwn, int type);
#endif
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 09:59:08 +00:00
netdev_features_t bnx2x_fix_features(struct net_device *dev,
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 09:59:08 +00:00
netdev_features_t features);
int bnx2x_set_features(struct net_device *dev, netdev_features_t features);
/**
* bnx2x_tx_timeout - tx timeout netdev callback
*
* @dev: net device
*/
void bnx2x_tx_timeout(struct net_device *dev);
/*********************** Inlines **********************************/
/*********************** Fast path ********************************/
static inline void bnx2x_update_fpsb_idx(struct bnx2x_fastpath *fp)
{
barrier(); /* status block is written to by the chip */
fp->fp_hc_idx = fp->sb_running_index[SM_RX_ID];
}
static inline void bnx2x_igu_ack_sb_gen(struct bnx2x *bp, u8 igu_sb_id,
u8 segment, u16 index, u8 op,
u8 update, u32 igu_addr)
{
struct igu_regular cmd_data = {0};
cmd_data.sb_id_and_flags =
((index << IGU_REGULAR_SB_INDEX_SHIFT) |
(segment << IGU_REGULAR_SEGMENT_ACCESS_SHIFT) |
(update << IGU_REGULAR_BUPDATE_SHIFT) |
(op << IGU_REGULAR_ENABLE_INT_SHIFT));
DP(NETIF_MSG_INTR, "write 0x%08x to IGU addr 0x%x\n",
cmd_data.sb_id_and_flags, igu_addr);
REG_WR(bp, igu_addr, cmd_data.sb_id_and_flags);
/* Make sure that ACK is written */
mmiowb();
barrier();
}
static inline void bnx2x_hc_ack_sb(struct bnx2x *bp, u8 sb_id,
u8 storm, u16 index, u8 op, u8 update)
{
u32 hc_addr = (HC_REG_COMMAND_REG + BP_PORT(bp)*32 +
COMMAND_REG_INT_ACK);
struct igu_ack_register igu_ack;
igu_ack.status_block_index = index;
igu_ack.sb_id_and_flags =
((sb_id << IGU_ACK_REGISTER_STATUS_BLOCK_ID_SHIFT) |
(storm << IGU_ACK_REGISTER_STORM_ID_SHIFT) |
(update << IGU_ACK_REGISTER_UPDATE_INDEX_SHIFT) |
(op << IGU_ACK_REGISTER_INTERRUPT_MODE_SHIFT));
REG_WR(bp, hc_addr, (*(u32 *)&igu_ack));
/* Make sure that ACK is written */
mmiowb();
barrier();
}
static inline void bnx2x_ack_sb(struct bnx2x *bp, u8 igu_sb_id, u8 storm,
u16 index, u8 op, u8 update)
{
if (bp->common.int_block == INT_BLOCK_HC)
bnx2x_hc_ack_sb(bp, igu_sb_id, storm, index, op, update);
else {
u8 segment;
if (CHIP_INT_MODE_IS_BC(bp))
segment = storm;
else if (igu_sb_id != bp->igu_dsb_id)
segment = IGU_SEG_ACCESS_DEF;
else if (storm == ATTENTION_ID)
segment = IGU_SEG_ACCESS_ATTN;
else
segment = IGU_SEG_ACCESS_DEF;
bnx2x_igu_ack_sb(bp, igu_sb_id, segment, index, op, update);
}
}
static inline u16 bnx2x_hc_ack_int(struct bnx2x *bp)
{
u32 hc_addr = (HC_REG_COMMAND_REG + BP_PORT(bp)*32 +
COMMAND_REG_SIMD_MASK);
u32 result = REG_RD(bp, hc_addr);
barrier();
return result;
}
static inline u16 bnx2x_igu_ack_int(struct bnx2x *bp)
{
u32 igu_addr = (BAR_IGU_INTMEM + IGU_REG_SISR_MDPC_WMASK_LSB_UPPER*8);
u32 result = REG_RD(bp, igu_addr);
DP(NETIF_MSG_INTR, "read 0x%08x from IGU addr 0x%x\n",
result, igu_addr);
barrier();
return result;
}
static inline u16 bnx2x_ack_int(struct bnx2x *bp)
{
barrier();
if (bp->common.int_block == INT_BLOCK_HC)
return bnx2x_hc_ack_int(bp);
else
return bnx2x_igu_ack_int(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 08:31:57 +00:00
static inline int bnx2x_has_tx_work_unload(struct bnx2x_fp_txdata *txdata)
{
/* Tell compiler that consumer and producer can change */
barrier();
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 08:31:57 +00:00
return txdata->tx_pkt_prod != txdata->tx_pkt_cons;
}
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 08:31:57 +00:00
static inline u16 bnx2x_tx_avail(struct bnx2x *bp,
struct bnx2x_fp_txdata *txdata)
{
s16 used;
u16 prod;
u16 cons;
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 08:31:57 +00:00
prod = txdata->tx_bd_prod;
cons = txdata->tx_bd_cons;
used = SUB_S16(prod, cons);
#ifdef BNX2X_STOP_ON_ERROR
WARN_ON(used < 0);
WARN_ON(used > txdata->tx_ring_size);
WARN_ON((txdata->tx_ring_size - used) > MAX_TX_AVAIL);
#endif
return (s16)(txdata->tx_ring_size) - used;
}
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 08:31:57 +00:00
static inline int bnx2x_tx_queue_has_work(struct bnx2x_fp_txdata *txdata)
{
u16 hw_cons;
/* Tell compiler that status block fields can change */
barrier();
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 08:31:57 +00:00
hw_cons = le16_to_cpu(*txdata->tx_cons_sb);
return hw_cons != txdata->tx_pkt_cons;
}
static inline bool bnx2x_has_tx_work(struct bnx2x_fastpath *fp)
{
u8 cos;
for_each_cos_in_tx_queue(fp, cos)
if (bnx2x_tx_queue_has_work(fp->txdata_ptr[cos]))
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 08:31:57 +00:00
return true;
return false;
}
static inline int bnx2x_has_rx_work(struct bnx2x_fastpath *fp)
{
u16 rx_cons_sb;
/* Tell compiler that status block fields can change */
barrier();
rx_cons_sb = le16_to_cpu(*fp->rx_cons_sb);
if ((rx_cons_sb & MAX_RCQ_DESC_CNT) == MAX_RCQ_DESC_CNT)
rx_cons_sb++;
return (fp->rx_comp_cons != rx_cons_sb);
}
/**
* bnx2x_tx_disable - disables tx from stack point of view
*
* @bp: driver handle
*/
static inline void bnx2x_tx_disable(struct bnx2x *bp)
{
netif_tx_disable(bp->dev);
netif_carrier_off(bp->dev);
}
static inline void bnx2x_free_rx_sge(struct bnx2x *bp,
struct bnx2x_fastpath *fp, u16 index)
{
struct sw_rx_page *sw_buf = &fp->rx_page_ring[index];
struct page *page = sw_buf->page;
struct eth_rx_sge *sge = &fp->rx_sge_ring[index];
/* Skip "next page" elements */
if (!page)
return;
dma_unmap_page(&bp->pdev->dev, dma_unmap_addr(sw_buf, mapping),
SGE_PAGE_SIZE*PAGES_PER_SGE, DMA_FROM_DEVICE);
__free_pages(page, PAGES_PER_SGE_SHIFT);
sw_buf->page = NULL;
sge->addr_hi = 0;
sge->addr_lo = 0;
}
static inline void bnx2x_add_all_napi_cnic(struct bnx2x *bp)
{
int i;
/* Add NAPI objects */
for_each_rx_queue_cnic(bp, i)
netif_napi_add(bp->dev, &bnx2x_fp(bp, i, napi),
bnx2x_poll, BNX2X_NAPI_WEIGHT);
}
static inline void bnx2x_add_all_napi(struct bnx2x *bp)
{
int i;
/* Add NAPI objects */
for_each_eth_queue(bp, i)
netif_napi_add(bp->dev, &bnx2x_fp(bp, i, napi),
bnx2x_poll, BNX2X_NAPI_WEIGHT);
}
static inline void bnx2x_del_all_napi_cnic(struct bnx2x *bp)
{
int i;
for_each_rx_queue_cnic(bp, i)
netif_napi_del(&bnx2x_fp(bp, i, napi));
}
static inline void bnx2x_del_all_napi(struct bnx2x *bp)
{
int i;
for_each_eth_queue(bp, i)
netif_napi_del(&bnx2x_fp(bp, i, napi));
}
int bnx2x_set_int_mode(struct bnx2x *bp);
static inline void bnx2x_disable_msi(struct bnx2x *bp)
{
if (bp->flags & USING_MSIX_FLAG) {
pci_disable_msix(bp->pdev);
bp->flags &= ~(USING_MSIX_FLAG | USING_SINGLE_MSIX_FLAG);
} else if (bp->flags & USING_MSI_FLAG) {
pci_disable_msi(bp->pdev);
bp->flags &= ~USING_MSI_FLAG;
}
}
static inline int bnx2x_calc_num_queues(struct bnx2x *bp)
{
return num_queues ?
min_t(int, num_queues, BNX2X_MAX_QUEUES(bp)) :
min_t(int, netif_get_num_default_rss_queues(),
BNX2X_MAX_QUEUES(bp));
}
static inline void bnx2x_clear_sge_mask_next_elems(struct bnx2x_fastpath *fp)
{
int i, j;
for (i = 1; i <= NUM_RX_SGE_PAGES; i++) {
int idx = RX_SGE_CNT * i - 1;
for (j = 0; j < 2; j++) {
BIT_VEC64_CLEAR_BIT(fp->sge_mask, idx);
idx--;
}
}
}
static inline void bnx2x_init_sge_ring_bit_mask(struct bnx2x_fastpath *fp)
{
/* Set the mask to all 1-s: it's faster to compare to 0 than to 0xf-s */
memset(fp->sge_mask, 0xff, sizeof(fp->sge_mask));
/* Clear the two last indices in the page to 1:
these are the indices that correspond to the "next" element,
hence will never be indicated and should be removed from
the calculations. */
bnx2x_clear_sge_mask_next_elems(fp);
}
bnx2x: uses build_skb() in receive path bnx2x uses following formula to compute its rx_buf_sz : dev->mtu + 2*L1_CACHE_BYTES + 14 + 8 + 8 + 2 Then core network adds NET_SKB_PAD and SKB_DATA_ALIGN(sizeof(struct skb_shared_info)) Final allocated size for skb head on x86_64 (L1_CACHE_BYTES = 64, MTU=1500) : 2112 bytes : SLUB/SLAB round this to 4096 bytes. Since skb truesize is then bigger than SK_MEM_QUANTUM, we have lot of false sharing because of mem_reclaim in UDP stack. One possible way to half truesize is to reduce the need by 64 bytes (2112 -> 2048 bytes) Instead of allocating a full cache line at the end of packet for alignment, we can use the fact that skb_shared_info sits at the end of skb->head, and we can use this room, if we convert bnx2x to new build_skb() infrastructure. skb_shared_info will be initialized after hardware finished its transfert, so we can eventually overwrite the final padding. Using build_skb() also reduces cache line misses in the driver, since we use cache hot skb instead of cold ones. Number of in-flight sk_buff structures is lower, they are recycled while still hot. Performance results : (820.000 pps on a rx UDP monothread benchmark, instead of 720.000 pps) Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com> CC: Eilon Greenstein <eilong@broadcom.com> CC: Ben Hutchings <bhutchings@solarflare.com> CC: Tom Herbert <therbert@google.com> CC: Jamal Hadi Salim <hadi@mojatatu.com> CC: Stephen Hemminger <shemminger@vyatta.com> CC: Thomas Graf <tgraf@infradead.org> CC: Herbert Xu <herbert@gondor.apana.org.au> CC: Jeff Kirsher <jeffrey.t.kirsher@intel.com> Acked-by: Eilon Greenstein <eilong@broadcom.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2011-11-14 06:05:34 +00:00
/* note that we are not allocating a new buffer,
* we are just moving one from cons to prod
* we are not creating a new mapping,
* so there is no need to check for dma_mapping_error().
*/
bnx2x: uses build_skb() in receive path bnx2x uses following formula to compute its rx_buf_sz : dev->mtu + 2*L1_CACHE_BYTES + 14 + 8 + 8 + 2 Then core network adds NET_SKB_PAD and SKB_DATA_ALIGN(sizeof(struct skb_shared_info)) Final allocated size for skb head on x86_64 (L1_CACHE_BYTES = 64, MTU=1500) : 2112 bytes : SLUB/SLAB round this to 4096 bytes. Since skb truesize is then bigger than SK_MEM_QUANTUM, we have lot of false sharing because of mem_reclaim in UDP stack. One possible way to half truesize is to reduce the need by 64 bytes (2112 -> 2048 bytes) Instead of allocating a full cache line at the end of packet for alignment, we can use the fact that skb_shared_info sits at the end of skb->head, and we can use this room, if we convert bnx2x to new build_skb() infrastructure. skb_shared_info will be initialized after hardware finished its transfert, so we can eventually overwrite the final padding. Using build_skb() also reduces cache line misses in the driver, since we use cache hot skb instead of cold ones. Number of in-flight sk_buff structures is lower, they are recycled while still hot. Performance results : (820.000 pps on a rx UDP monothread benchmark, instead of 720.000 pps) Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com> CC: Eilon Greenstein <eilong@broadcom.com> CC: Ben Hutchings <bhutchings@solarflare.com> CC: Tom Herbert <therbert@google.com> CC: Jamal Hadi Salim <hadi@mojatatu.com> CC: Stephen Hemminger <shemminger@vyatta.com> CC: Thomas Graf <tgraf@infradead.org> CC: Herbert Xu <herbert@gondor.apana.org.au> CC: Jeff Kirsher <jeffrey.t.kirsher@intel.com> Acked-by: Eilon Greenstein <eilong@broadcom.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2011-11-14 06:05:34 +00:00
static inline void bnx2x_reuse_rx_data(struct bnx2x_fastpath *fp,
u16 cons, u16 prod)
{
struct sw_rx_bd *cons_rx_buf = &fp->rx_buf_ring[cons];
struct sw_rx_bd *prod_rx_buf = &fp->rx_buf_ring[prod];
struct eth_rx_bd *cons_bd = &fp->rx_desc_ring[cons];
struct eth_rx_bd *prod_bd = &fp->rx_desc_ring[prod];
dma_unmap_addr_set(prod_rx_buf, mapping,
dma_unmap_addr(cons_rx_buf, mapping));
bnx2x: uses build_skb() in receive path bnx2x uses following formula to compute its rx_buf_sz : dev->mtu + 2*L1_CACHE_BYTES + 14 + 8 + 8 + 2 Then core network adds NET_SKB_PAD and SKB_DATA_ALIGN(sizeof(struct skb_shared_info)) Final allocated size for skb head on x86_64 (L1_CACHE_BYTES = 64, MTU=1500) : 2112 bytes : SLUB/SLAB round this to 4096 bytes. Since skb truesize is then bigger than SK_MEM_QUANTUM, we have lot of false sharing because of mem_reclaim in UDP stack. One possible way to half truesize is to reduce the need by 64 bytes (2112 -> 2048 bytes) Instead of allocating a full cache line at the end of packet for alignment, we can use the fact that skb_shared_info sits at the end of skb->head, and we can use this room, if we convert bnx2x to new build_skb() infrastructure. skb_shared_info will be initialized after hardware finished its transfert, so we can eventually overwrite the final padding. Using build_skb() also reduces cache line misses in the driver, since we use cache hot skb instead of cold ones. Number of in-flight sk_buff structures is lower, they are recycled while still hot. Performance results : (820.000 pps on a rx UDP monothread benchmark, instead of 720.000 pps) Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com> CC: Eilon Greenstein <eilong@broadcom.com> CC: Ben Hutchings <bhutchings@solarflare.com> CC: Tom Herbert <therbert@google.com> CC: Jamal Hadi Salim <hadi@mojatatu.com> CC: Stephen Hemminger <shemminger@vyatta.com> CC: Thomas Graf <tgraf@infradead.org> CC: Herbert Xu <herbert@gondor.apana.org.au> CC: Jeff Kirsher <jeffrey.t.kirsher@intel.com> Acked-by: Eilon Greenstein <eilong@broadcom.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2011-11-14 06:05:34 +00:00
prod_rx_buf->data = cons_rx_buf->data;
*prod_bd = *cons_bd;
}
/************************* Init ******************************************/
/* returns func by VN for current port */
static inline int func_by_vn(struct bnx2x *bp, int vn)
{
return 2 * vn + BP_PORT(bp);
}
static inline int bnx2x_config_rss_eth(struct bnx2x *bp, bool config_hash)
{
return bnx2x_config_rss_pf(bp, &bp->rss_conf_obj, config_hash);
}
/**
* bnx2x_func_start - init function
*
* @bp: driver handle
*
* Must be called before sending CLIENT_SETUP for the first client.
*/
static inline int bnx2x_func_start(struct bnx2x *bp)
{
struct bnx2x_func_state_params func_params = {NULL};
struct bnx2x_func_start_params *start_params =
&func_params.params.start;
/* Prepare parameters for function state transitions */
__set_bit(RAMROD_COMP_WAIT, &func_params.ramrod_flags);
func_params.f_obj = &bp->func_obj;
func_params.cmd = BNX2X_F_CMD_START;
/* Function parameters */
start_params->mf_mode = bp->mf_mode;
start_params->sd_vlan_tag = bp->mf_ov;
if (CHIP_IS_E2(bp) || CHIP_IS_E3(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 08:31:57 +00:00
start_params->network_cos_mode = STATIC_COS;
else /* CHIP_IS_E1X */
start_params->network_cos_mode = FW_WRR;
return bnx2x_func_state_change(bp, &func_params);
}
/**
* bnx2x_set_fw_mac_addr - fill in a MAC address in FW format
*
* @fw_hi: pointer to upper part
* @fw_mid: pointer to middle part
* @fw_lo: pointer to lower part
* @mac: pointer to MAC address
*/
static inline void bnx2x_set_fw_mac_addr(u16 *fw_hi, u16 *fw_mid, u16 *fw_lo,
u8 *mac)
{
((u8 *)fw_hi)[0] = mac[1];
((u8 *)fw_hi)[1] = mac[0];
((u8 *)fw_mid)[0] = mac[3];
((u8 *)fw_mid)[1] = mac[2];
((u8 *)fw_lo)[0] = mac[5];
((u8 *)fw_lo)[1] = mac[4];
}
static inline void bnx2x_free_rx_sge_range(struct bnx2x *bp,
struct bnx2x_fastpath *fp, int last)
{
int i;
if (fp->disable_tpa)
return;
for (i = 0; i < last; i++)
bnx2x_free_rx_sge(bp, fp, i);
}
static inline void bnx2x_set_next_page_rx_bd(struct bnx2x_fastpath *fp)
{
int i;
for (i = 1; i <= NUM_RX_RINGS; i++) {
struct eth_rx_bd *rx_bd;
rx_bd = &fp->rx_desc_ring[RX_DESC_CNT * i - 2];
rx_bd->addr_hi =
cpu_to_le32(U64_HI(fp->rx_desc_mapping +
BCM_PAGE_SIZE*(i % NUM_RX_RINGS)));
rx_bd->addr_lo =
cpu_to_le32(U64_LO(fp->rx_desc_mapping +
BCM_PAGE_SIZE*(i % NUM_RX_RINGS)));
}
}
/* Statistics ID are global per chip/path, while Client IDs for E1x are per
* port.
*/
static inline u8 bnx2x_stats_id(struct bnx2x_fastpath *fp)
{
struct bnx2x *bp = fp->bp;
if (!CHIP_IS_E1x(bp)) {
/* there are special statistics counters for FCoE 136..140 */
if (IS_FCOE_FP(fp))
return bp->cnic_base_cl_id + (bp->pf_num >> 1);
return fp->cl_id;
}
return fp->cl_id + BP_PORT(bp) * FP_SB_MAX_E1x;
}
static inline void bnx2x_init_vlan_mac_fp_objs(struct bnx2x_fastpath *fp,
bnx2x_obj_type obj_type)
{
struct bnx2x *bp = fp->bp;
/* Configure classification DBs */
bnx2x_init_mac_obj(bp, &bnx2x_sp_obj(bp, fp).mac_obj, fp->cl_id,
fp->cid, BP_FUNC(bp), bnx2x_sp(bp, mac_rdata),
bnx2x_sp_mapping(bp, mac_rdata),
BNX2X_FILTER_MAC_PENDING,
&bp->sp_state, obj_type,
&bp->macs_pool);
}
/**
* bnx2x_get_path_func_num - get number of active functions
*
* @bp: driver handle
*
* Calculates the number of active (not hidden) functions on the
* current path.
*/
static inline u8 bnx2x_get_path_func_num(struct bnx2x *bp)
{
u8 func_num = 0, i;
/* 57710 has only one function per-port */
if (CHIP_IS_E1(bp))
return 1;
/* Calculate a number of functions enabled on the current
* PATH/PORT.
*/
if (CHIP_REV_IS_SLOW(bp)) {
if (IS_MF(bp))
func_num = 4;
else
func_num = 2;
} else {
for (i = 0; i < E1H_FUNC_MAX / 2; i++) {
u32 func_config =
MF_CFG_RD(bp,
func_mf_config[BP_PORT(bp) + 2 * i].
config);
func_num +=
((func_config & FUNC_MF_CFG_FUNC_HIDE) ? 0 : 1);
}
}
WARN_ON(!func_num);
return func_num;
}
static inline void bnx2x_init_bp_objs(struct bnx2x *bp)
{
/* RX_MODE controlling object */
bnx2x_init_rx_mode_obj(bp, &bp->rx_mode_obj);
/* multicast configuration controlling object */
bnx2x_init_mcast_obj(bp, &bp->mcast_obj, bp->fp->cl_id, bp->fp->cid,
BP_FUNC(bp), BP_FUNC(bp),
bnx2x_sp(bp, mcast_rdata),
bnx2x_sp_mapping(bp, mcast_rdata),
BNX2X_FILTER_MCAST_PENDING, &bp->sp_state,
BNX2X_OBJ_TYPE_RX);
/* Setup CAM credit pools */
bnx2x_init_mac_credit_pool(bp, &bp->macs_pool, BP_FUNC(bp),
bnx2x_get_path_func_num(bp));
/* RSS configuration object */
bnx2x_init_rss_config_obj(bp, &bp->rss_conf_obj, bp->fp->cl_id,
bp->fp->cid, BP_FUNC(bp), BP_FUNC(bp),
bnx2x_sp(bp, rss_rdata),
bnx2x_sp_mapping(bp, rss_rdata),
BNX2X_FILTER_RSS_CONF_PENDING, &bp->sp_state,
BNX2X_OBJ_TYPE_RX);
}
static inline u8 bnx2x_fp_qzone_id(struct bnx2x_fastpath *fp)
{
if (CHIP_IS_E1x(fp->bp))
return fp->cl_id + BP_PORT(fp->bp) * ETH_MAX_RX_CLIENTS_E1H;
else
return fp->cl_id;
}
static inline u32 bnx2x_rx_ustorm_prods_offset(struct bnx2x_fastpath *fp)
{
struct bnx2x *bp = fp->bp;
u32 offset = BAR_USTRORM_INTMEM;
if (IS_VF(bp))
return PXP_VF_ADDR_USDM_QUEUES_START +
bp->acquire_resp.resc.hw_qid[fp->index] *
sizeof(struct ustorm_queue_zone_data);
else if (!CHIP_IS_E1x(bp))
offset += USTORM_RX_PRODS_E2_OFFSET(fp->cl_qzone_id);
else
offset += USTORM_RX_PRODS_E1X_OFFSET(BP_PORT(bp), fp->cl_id);
return offset;
}
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 08:31:57 +00:00
static inline void bnx2x_init_txdata(struct bnx2x *bp,
struct bnx2x_fp_txdata *txdata, u32 cid,
int txq_index, __le16 *tx_cons_sb,
struct bnx2x_fastpath *fp)
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 08:31:57 +00:00
{
txdata->cid = cid;
txdata->txq_index = txq_index;
txdata->tx_cons_sb = tx_cons_sb;
txdata->parent_fp = fp;
txdata->tx_ring_size = IS_FCOE_FP(fp) ? MAX_TX_AVAIL : bp->tx_ring_size;
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 08:31:57 +00:00
DP(NETIF_MSG_IFUP, "created tx data cid %d, txq %d\n",
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 08:31:57 +00:00
txdata->cid, txdata->txq_index);
}
static inline u8 bnx2x_cnic_eth_cl_id(struct bnx2x *bp, u8 cl_idx)
{
return bp->cnic_base_cl_id + cl_idx +
(bp->pf_num >> 1) * BNX2X_MAX_CNIC_ETH_CL_ID_IDX;
}
static inline u8 bnx2x_cnic_fw_sb_id(struct bnx2x *bp)
{
/* the 'first' id is allocated for the cnic */
return bp->base_fw_ndsb;
}
static inline u8 bnx2x_cnic_igu_sb_id(struct bnx2x *bp)
{
return bp->igu_base_sb;
}
static inline void bnx2x_init_fcoe_fp(struct bnx2x *bp)
{
struct bnx2x_fastpath *fp = bnx2x_fcoe_fp(bp);
unsigned long q_type = 0;
bnx2x_fcoe(bp, rx_queue) = BNX2X_NUM_ETH_QUEUES(bp);
bnx2x_fcoe(bp, cl_id) = bnx2x_cnic_eth_cl_id(bp,
BNX2X_FCOE_ETH_CL_ID_IDX);
bnx2x_fcoe(bp, cid) = BNX2X_FCOE_ETH_CID(bp);
bnx2x_fcoe(bp, fw_sb_id) = DEF_SB_ID;
bnx2x_fcoe(bp, igu_sb_id) = bp->igu_dsb_id;
bnx2x_fcoe(bp, rx_cons_sb) = BNX2X_FCOE_L2_RX_INDEX;
bnx2x_init_txdata(bp, bnx2x_fcoe(bp, txdata_ptr[0]),
fp->cid, FCOE_TXQ_IDX(bp), BNX2X_FCOE_L2_TX_INDEX,
fp);
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 08:31:57 +00:00
DP(NETIF_MSG_IFUP, "created fcoe tx data (fp index %d)\n", fp->index);
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 08:31:57 +00:00
/* qZone id equals to FW (per path) client id */
bnx2x_fcoe(bp, cl_qzone_id) = bnx2x_fp_qzone_id(fp);
/* init shortcut */
bnx2x_fcoe(bp, ustorm_rx_prods_offset) =
bnx2x_rx_ustorm_prods_offset(fp);
/* Configure Queue State object */
__set_bit(BNX2X_Q_TYPE_HAS_RX, &q_type);
__set_bit(BNX2X_Q_TYPE_HAS_TX, &q_type);
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 08:31:57 +00:00
/* No multi-CoS for FCoE L2 client */
BUG_ON(fp->max_cos != 1);
bnx2x_init_queue_obj(bp, &bnx2x_sp_obj(bp, fp).q_obj, fp->cl_id,
&fp->cid, 1, BP_FUNC(bp), bnx2x_sp(bp, q_rdata),
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 08:31:57 +00:00
bnx2x_sp_mapping(bp, q_rdata), q_type);
DP(NETIF_MSG_IFUP,
"queue[%d]: bnx2x_init_sb(%p,%p) cl_id %d fw_sb %d igu_sb %d\n",
fp->index, bp, fp->status_blk.e2_sb, fp->cl_id, fp->fw_sb_id,
fp->igu_sb_id);
}
static inline int bnx2x_clean_tx_queue(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 08:31:57 +00:00
struct bnx2x_fp_txdata *txdata)
{
int cnt = 1000;
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 08:31:57 +00:00
while (bnx2x_has_tx_work_unload(txdata)) {
if (!cnt) {
BNX2X_ERR("timeout waiting for queue[%d]: txdata->tx_pkt_prod(%d) != txdata->tx_pkt_cons(%d)\n",
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 08:31:57 +00:00
txdata->txq_index, txdata->tx_pkt_prod,
txdata->tx_pkt_cons);
#ifdef BNX2X_STOP_ON_ERROR
bnx2x_panic();
return -EBUSY;
#else
break;
#endif
}
cnt--;
usleep_range(1000, 1000);
}
return 0;
}
int bnx2x_get_link_cfg_idx(struct bnx2x *bp);
static inline void __storm_memset_struct(struct bnx2x *bp,
u32 addr, size_t size, u32 *data)
{
int i;
for (i = 0; i < size/4; i++)
REG_WR(bp, addr + (i * 4), data[i]);
}
/**
* bnx2x_wait_sp_comp - wait for the outstanding SP commands.
*
* @bp: driver handle
* @mask: bits that need to be cleared
*/
static inline bool bnx2x_wait_sp_comp(struct bnx2x *bp, unsigned long mask)
{
int tout = 5000; /* Wait for 5 secs tops */
while (tout--) {
smp_mb();
netif_addr_lock_bh(bp->dev);
if (!(bp->sp_state & mask)) {
netif_addr_unlock_bh(bp->dev);
return true;
}
netif_addr_unlock_bh(bp->dev);
usleep_range(1000, 1000);
}
smp_mb();
netif_addr_lock_bh(bp->dev);
if (bp->sp_state & mask) {
BNX2X_ERR("Filtering completion timed out. sp_state 0x%lx, mask 0x%lx\n",
bp->sp_state, mask);
netif_addr_unlock_bh(bp->dev);
return false;
}
netif_addr_unlock_bh(bp->dev);
return true;
}
/**
* bnx2x_set_ctx_validation - set CDU context validation values
*
* @bp: driver handle
* @cxt: context of the connection on the host memory
* @cid: SW CID of the connection to be configured
*/
void bnx2x_set_ctx_validation(struct bnx2x *bp, struct eth_context *cxt,
u32 cid);
void bnx2x_update_coalesce_sb_index(struct bnx2x *bp, u8 fw_sb_id,
u8 sb_index, u8 disable, u16 usec);
void bnx2x_acquire_phy_lock(struct bnx2x *bp);
void bnx2x_release_phy_lock(struct bnx2x *bp);
/**
* bnx2x_extract_max_cfg - extract MAX BW part from MF configuration.
*
* @bp: driver handle
* @mf_cfg: MF configuration
*
*/
static inline u16 bnx2x_extract_max_cfg(struct bnx2x *bp, u32 mf_cfg)
{
u16 max_cfg = (mf_cfg & FUNC_MF_CFG_MAX_BW_MASK) >>
FUNC_MF_CFG_MAX_BW_SHIFT;
if (!max_cfg) {
DP(NETIF_MSG_IFUP | BNX2X_MSG_ETHTOOL,
"Max BW configured to 0 - using 100 instead\n");
max_cfg = 100;
}
return max_cfg;
}
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 09:59:08 +00:00
/* checks if HW supports GRO for given MTU */
static inline bool bnx2x_mtu_allows_gro(int mtu)
{
/* gro frags per page */
int fpp = SGE_PAGE_SIZE / (mtu - ETH_MAX_TPA_HEADER_SIZE);
/*
* 1. number of frags should not grow above MAX_SKB_FRAGS
* 2. frag must fit the page
*/
return mtu <= SGE_PAGE_SIZE && (U_ETH_SGL_SIZE * fpp) <= MAX_SKB_FRAGS;
}
/**
* bnx2x_get_iscsi_info - update iSCSI params according to licensing info.
*
* @bp: driver handle
*
*/
void bnx2x_get_iscsi_info(struct bnx2x *bp);
/**
* bnx2x_link_sync_notify - send notification to other functions.
*
* @bp: driver handle
*
*/
static inline void bnx2x_link_sync_notify(struct bnx2x *bp)
{
int func;
int vn;
/* Set the attention towards other drivers on the same port */
for (vn = VN_0; vn < BP_MAX_VN_NUM(bp); vn++) {
if (vn == BP_VN(bp))
continue;
func = func_by_vn(bp, vn);
REG_WR(bp, MISC_REG_AEU_GENERAL_ATTN_0 +
(LINK_SYNC_ATTENTION_BIT_FUNC_0 + func)*4, 1);
}
}
/**
* bnx2x_update_drv_flags - update flags in shmem
*
* @bp: driver handle
* @flags: flags to update
* @set: set or clear
*
*/
static inline void bnx2x_update_drv_flags(struct bnx2x *bp, u32 flags, u32 set)
{
if (SHMEM2_HAS(bp, drv_flags)) {
u32 drv_flags;
bnx2x_acquire_hw_lock(bp, HW_LOCK_RESOURCE_DRV_FLAGS);
drv_flags = SHMEM2_RD(bp, drv_flags);
if (set)
SET_FLAGS(drv_flags, flags);
else
RESET_FLAGS(drv_flags, flags);
SHMEM2_WR(bp, drv_flags, drv_flags);
DP(NETIF_MSG_IFUP, "drv_flags 0x%08x\n", drv_flags);
bnx2x_release_hw_lock(bp, HW_LOCK_RESOURCE_DRV_FLAGS);
}
}
static inline bool bnx2x_is_valid_ether_addr(struct bnx2x *bp, u8 *addr)
{
if (is_valid_ether_addr(addr) ||
(is_zero_ether_addr(addr) &&
(IS_MF_STORAGE_SD(bp) || IS_MF_FCOE_AFEX(bp))))
return true;
return false;
}
#endif /* BNX2X_CMN_H */