2026-01-29 22:25:33 +08:00

3123 lines
83 KiB
C

// SPDX-License-Identifier: GPL-2.0
/* Huawei HiNIC PCI Express Linux driver
* Copyright(c) 2017 Huawei Technologies Co., Ltd
*
* This program is free software; you can redistribute it and/or modify it
* under the terms and conditions of the GNU General Public License,
* version 2, as published by the Free Software Foundation.
*
* This program is distributed in the hope it will be useful, but WITHOUT
* ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
* FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License
* for more details.
*
*/
#define pr_fmt(fmt) KBUILD_MODNAME ": [NIC]" fmt
#include <linux/kernel.h>
#include <linux/pci.h>
#include <linux/device.h>
#include <linux/module.h>
#include <linux/moduleparam.h>
#include <linux/types.h>
#include <linux/errno.h>
#include <linux/interrupt.h>
#include <linux/etherdevice.h>
#include <linux/netdevice.h>
#include <linux/if_vlan.h>
#include <linux/ethtool.h>
#include <linux/dcbnl.h>
#include <linux/tcp.h>
#include <linux/ip.h>
#include <linux/debugfs.h>
#include "ossl_knl.h"
#include "hinic_hw_mgmt.h"
#include "hinic_hw.h"
#include "hinic_dbg.h"
#include "hinic_nic_cfg.h"
#include "hinic_nic_dev.h"
#include "hinic_tx.h"
#include "hinic_rx.h"
#include "hinic_qp.h"
#include "hinic_dcb.h"
#include "hinic_lld.h"
#include "hinic_sriov.h"
#include "hinic_pci_id_tbl.h"
static u16 num_qps;
module_param(num_qps, ushort, 0444);
MODULE_PARM_DESC(num_qps, "Number of Queue Pairs (default unset)");
static u16 ovs_num_qps = 16;
module_param(ovs_num_qps, ushort, 0444);
MODULE_PARM_DESC(ovs_num_qps, "Number of Queue Pairs in ovs mode (default=16)");
#define DEFAULT_POLL_WEIGHT 64
static unsigned int poll_weight = DEFAULT_POLL_WEIGHT;
module_param(poll_weight, uint, 0444);
MODULE_PARM_DESC(poll_weight, "Number packets for NAPI budget (default=64)");
#define HINIC_DEAULT_TXRX_MSIX_PENDING_LIMIT 2
#define HINIC_DEAULT_TXRX_MSIX_COALESC_TIMER_CFG 32
#define HINIC_DEAULT_TXRX_MSIX_RESEND_TIMER_CFG 7
/* suit for sdi3.0 vm mode, change this define for test best performance */
#define SDI_VM_PENDING_LIMT 2
#define SDI_VM_COALESCE_TIMER_CFG 16
#define SDI_VM_RX_PKT_RATE_HIGH 1000000
#define SDI_VM_RX_PKT_RATE_LOW 30000
#define SDI_VM_RX_USECS_HIGH 56
#define SDI_VM_RX_PENDING_LIMT_HIGH 20
#define SDI_VM_RX_USECS_LOW 16
#define SDI_VM_RX_PENDING_LIMT_LOW 2
/* if qp_coalesc_use_drv_params_switch !=0, use user setting params */
static unsigned char qp_coalesc_use_drv_params_switch;
module_param(qp_coalesc_use_drv_params_switch, byte, 0444);
MODULE_PARM_DESC(qp_coalesc_use_drv_params_switch, "QP MSI-X Interrupt coalescing parameter switch (default=0, not use drv parameter)");
static unsigned char qp_pending_limit = HINIC_DEAULT_TXRX_MSIX_PENDING_LIMIT;
module_param(qp_pending_limit, byte, 0444);
MODULE_PARM_DESC(qp_pending_limit, "QP MSI-X Interrupt coalescing parameter pending_limit (default=2)");
static unsigned char qp_coalesc_timer_cfg =
HINIC_DEAULT_TXRX_MSIX_COALESC_TIMER_CFG;
module_param(qp_coalesc_timer_cfg, byte, 0444);
MODULE_PARM_DESC(qp_coalesc_timer_cfg, "QP MSI-X Interrupt coalescing parameter coalesc_timer_cfg (default=32)");
/* For arm64 server, the best known configuration of lro max wqe number
* is 4 (8K), for x86_64 server, it is 8 (16K). You can also
* configure these values by hinicadm.
*/
static unsigned char set_max_wqe_num;
module_param(set_max_wqe_num, byte, 0444);
MODULE_PARM_DESC(set_max_wqe_num, "Set lro max wqe number, valid range is 1 - 32, default is 4(arm) / 8(x86)");
#define DEFAULT_RX_BUFF_LEN 2
u16 rx_buff = DEFAULT_RX_BUFF_LEN;
module_param(rx_buff, ushort, 0444);
MODULE_PARM_DESC(rx_buff, "Set rx_buff size, buffer len must be 2^n. 2 - 16, default is 2KB");
static u32 set_lro_timer;
module_param(set_lro_timer, uint, 0444);
MODULE_PARM_DESC(set_lro_timer, "Set lro timer in micro second, valid range is 1 - 1024, default is 16");
static unsigned char set_link_status_follow = HINIC_LINK_FOLLOW_STATUS_MAX;
module_param(set_link_status_follow, byte, 0444);
MODULE_PARM_DESC(set_link_status_follow, "Set link status follow port status. 0 - default, 1 - follow, 2 - separate, other - unset. (default unset)");
static unsigned int lro_replenish_thld = 256;
module_param(lro_replenish_thld, uint, 0444);
MODULE_PARM_DESC(lro_replenish_thld, "Number wqe for lro replenish buffer (default=256)");
static bool l2nic_interrupt_switch = true;
module_param(l2nic_interrupt_switch, bool, 0644);
MODULE_PARM_DESC(l2nic_interrupt_switch, "Control whether execute l2nic io interrupt switch or not, default is true");
static unsigned char lro_en_status = HINIC_LRO_STATUS_UNSET;
module_param(lro_en_status, byte, 0444);
MODULE_PARM_DESC(lro_en_status, "lro enable status. 0 - disable, 1 - enable, other - unset. (default unset)");
static unsigned char qp_pending_limit_low = HINIC_RX_PENDING_LIMIT_LOW;
module_param(qp_pending_limit_low, byte, 0444);
MODULE_PARM_DESC(qp_pending_limit_low, "MSI-X adaptive low coalesce pending limit, range is 0 - 255");
static unsigned char qp_coalesc_timer_low = HINIC_RX_COAL_TIME_LOW;
module_param(qp_coalesc_timer_low, byte, 0444);
MODULE_PARM_DESC(qp_coalesc_timer_low, "MSI-X adaptive low coalesce time, range is 0 - 255");
static unsigned char qp_pending_limit_high = HINIC_RX_PENDING_LIMIT_HIGH;
module_param(qp_pending_limit_high, byte, 0444);
MODULE_PARM_DESC(qp_pending_limit_high, "MSI-X adaptive high coalesce pending limit, range is 0 - 255");
static unsigned char qp_coalesc_timer_high = HINIC_RX_COAL_TIME_HIGH;
module_param(qp_coalesc_timer_high, byte, 0444);
MODULE_PARM_DESC(qp_coalesc_timer_high, "MSI-X adaptive high coalesce time, range is 0 - 255");
#define HINIC_NIC_DEV_WQ_NAME "hinic_nic_dev_wq"
#define DEFAULT_MSG_ENABLE (NETIF_MSG_DRV | NETIF_MSG_LINK | \
NETIF_MSG_RX_ERR)
#define QID_MASKED(q_id, nic_dev) ((q_id) & ((nic_dev)->num_qps - 1))
#define VLAN_BITMAP_BYTE_SIZE(nic_dev) (sizeof(*(nic_dev)->vlan_bitmap))
#define VLAN_BITMAP_BITS_SIZE(nic_dev) (VLAN_BITMAP_BYTE_SIZE(nic_dev) * 8)
#define VLAN_NUM_BITMAPS(nic_dev) (VLAN_N_VID / \
VLAN_BITMAP_BITS_SIZE(nic_dev))
#define VLAN_BITMAP_SIZE(nic_dev) (VLAN_N_VID / \
VLAN_BITMAP_BYTE_SIZE(nic_dev))
#define VID_LINE(nic_dev, vid) ((vid) / VLAN_BITMAP_BITS_SIZE(nic_dev))
#define VID_COL(nic_dev, vid) ((vid) & (VLAN_BITMAP_BITS_SIZE(nic_dev) - 1))
enum hinic_rx_mod {
HINIC_RX_MODE_UC = 1 << 0,
HINIC_RX_MODE_MC = 1 << 1,
HINIC_RX_MODE_BC = 1 << 2,
HINIC_RX_MODE_MC_ALL = 1 << 3,
HINIC_RX_MODE_PROMISC = 1 << 4,
};
enum hinic_rx_buff_len {
RX_BUFF_VALID_2KB = 2,
RX_BUFF_VALID_4KB = 4,
RX_BUFF_VALID_8KB = 8,
RX_BUFF_VALID_16KB = 16,
};
#define HINIC_AVG_PKT_SMALL 256U
#define HINIC_MODERATONE_DELAY HZ
#define CONVERT_UNIT 1024
#define HINIC_MAX_VLAN_DEPTH_OFFLOAD_SUPPORT 2
#define HINIC_VLAN_CLEAR_OFFLOAD (NETIF_F_IP_CSUM | NETIF_F_IPV6_CSUM | \
NETIF_F_SCTP_CRC | NETIF_F_RXCSUM | \
NETIF_F_ALL_TSO)
int hinic_netdev_event(struct notifier_block *notifier,
unsigned long event, void *ptr)
{
struct net_device *ndev = netdev_notifier_info_to_dev(ptr);
struct net_device *real_dev, *ret;
struct hinic_nic_dev *nic_dev;
u16 vlan_depth;
if (!is_vlan_dev(ndev))
return NOTIFY_DONE;
dev_hold(ndev);
switch (event) {
case NETDEV_REGISTER:
real_dev = vlan_dev_real_dev(ndev);
nic_dev = hinic_get_uld_dev_by_ifname(real_dev->name,
SERVICE_T_NIC);
if (!nic_dev)
goto out;
vlan_depth = 1;
ret = vlan_dev_priv(ndev)->real_dev;
while (is_vlan_dev(ret)) {
ret = vlan_dev_priv(ret)->real_dev;
vlan_depth++;
}
if (vlan_depth == HINIC_MAX_VLAN_DEPTH_OFFLOAD_SUPPORT) {
ndev->vlan_features &= (~HINIC_VLAN_CLEAR_OFFLOAD);
} else if (vlan_depth > HINIC_MAX_VLAN_DEPTH_OFFLOAD_SUPPORT) {
ndev->hw_features &= (~HINIC_VLAN_CLEAR_OFFLOAD);
ndev->features &= (~HINIC_VLAN_CLEAR_OFFLOAD);
}
break;
default:
break;
};
out:
dev_put(ndev);
return NOTIFY_DONE;
}
/* used for netdev notifier register/unregister */
DEFINE_MUTEX(g_hinic_netdev_notifiers_mutex);
static int hinic_netdev_notifiers_ref_cnt;
static struct notifier_block hinic_netdev_notifier = {
.notifier_call = hinic_netdev_event,
};
static void hinic_register_notifier(struct hinic_nic_dev *nic_dev)
{
int err;
mutex_lock(&g_hinic_netdev_notifiers_mutex);
hinic_netdev_notifiers_ref_cnt++;
if (hinic_netdev_notifiers_ref_cnt == 1) {
err = register_netdevice_notifier(&hinic_netdev_notifier);
if (err) {
hinic_info(nic_dev, drv, "Register netdevice notifier failed, err: %d\n",
err);
hinic_netdev_notifiers_ref_cnt--;
}
}
mutex_unlock(&g_hinic_netdev_notifiers_mutex);
}
static void hinic_unregister_notifier(struct hinic_nic_dev *nic_dev)
{
mutex_lock(&g_hinic_netdev_notifiers_mutex);
if (hinic_netdev_notifiers_ref_cnt == 1)
unregister_netdevice_notifier(&hinic_netdev_notifier);
if (hinic_netdev_notifiers_ref_cnt)
hinic_netdev_notifiers_ref_cnt--;
mutex_unlock(&g_hinic_netdev_notifiers_mutex);
}
void hinic_link_status_change(struct hinic_nic_dev *nic_dev, bool status)
{
struct net_device *netdev = nic_dev->netdev;
if (!test_bit(HINIC_INTF_UP, &nic_dev->flags) ||
test_bit(HINIC_LP_TEST, &nic_dev->flags))
return;
if (status) {
if (netif_carrier_ok(netdev))
return;
nic_dev->link_status = status;
netif_carrier_on(netdev);
nicif_info(nic_dev, link, netdev, "Link is up\n");
} else {
if (!netif_carrier_ok(netdev))
return;
nic_dev->link_status = status;
netif_carrier_off(netdev);
nicif_info(nic_dev, link, netdev, "Link is down\n");
}
}
static void hinic_heart_lost(struct hinic_nic_dev *nic_dev)
{
nic_dev->heart_status = false;
}
static int hinic_setup_qps_resources(struct hinic_nic_dev *nic_dev)
{
struct net_device *netdev = nic_dev->netdev;
int err;
err = hinic_setup_all_tx_resources(netdev);
if (err) {
nicif_err(nic_dev, drv, netdev,
"Failed to create Tx queues\n");
return err;
}
err = hinic_setup_all_rx_resources(netdev, nic_dev->qps_irq_info);
if (err) {
nicif_err(nic_dev, drv, netdev,
"Failed to create Rx queues\n");
goto create_rxqs_err;
}
return 0;
create_rxqs_err:
hinic_free_all_tx_resources(netdev);
return err;
}
static int hinic_configure(struct hinic_nic_dev *nic_dev)
{
struct net_device *netdev = nic_dev->netdev;
int err;
/* rx rss init */
err = hinic_rx_configure(netdev);
if (err) {
nicif_err(nic_dev, drv, netdev, "Failed to configure rx\n");
return err;
}
return 0;
}
static void hinic_remove_configure(struct hinic_nic_dev *nic_dev)
{
hinic_rx_remove_configure(nic_dev->netdev);
}
static void hinic_setup_dcb_qps(struct hinic_nic_dev *nic_dev, u16 max_qps)
{
struct net_device *netdev = nic_dev->netdev;
u16 num_rss;
u8 num_tcs;
u8 i;
if (!test_bit(HINIC_DCB_ENABLE, &nic_dev->flags) ||
!test_bit(HINIC_RSS_ENABLE, &nic_dev->flags))
return;
num_tcs = (u8)netdev_get_num_tc(netdev);
/* For now, we don't support to change num_tcs */
if (num_tcs != nic_dev->max_cos || max_qps < num_tcs) {
nicif_err(nic_dev, drv, netdev, "Invalid num_tcs: %d or num_qps: %d, disable DCB\n",
num_tcs, max_qps);
netdev_reset_tc(netdev);
clear_bit(HINIC_DCB_ENABLE, &nic_dev->flags);
/* if we can't enable rss or get enough num_qps,
* need to sync default configure to hw
*/
hinic_configure_dcb(netdev);
} else {
/* We bind sq with cos but not tc */
num_rss = (u16)(max_qps / nic_dev->max_cos);
num_rss = min_t(u16, num_rss, nic_dev->rss_limit);
for (i = 0; i < nic_dev->max_cos; i++)
netdev_set_tc_queue(netdev, i, num_rss,
(u16)(num_rss * i));
nic_dev->num_rss = num_rss;
nic_dev->num_qps = (u16)(num_tcs * num_rss);
}
}
/* determin num_qps from rss_tmpl_id/irq_num/dcb_en */
static int hinic_setup_num_qps(struct hinic_nic_dev *nic_dev)
{
struct net_device *netdev = nic_dev->netdev;
u32 irq_size;
u16 resp_irq_num, i;
int err;
if (test_bit(HINIC_RSS_ENABLE, &nic_dev->flags)) {
nic_dev->num_rss = nic_dev->rss_limit;
nic_dev->num_qps = nic_dev->rss_limit;
} else {
nic_dev->num_rss = 0;
nic_dev->num_qps = 1;
}
hinic_setup_dcb_qps(nic_dev, nic_dev->max_qps);
irq_size = sizeof(*nic_dev->qps_irq_info) * nic_dev->num_qps;
if (!irq_size) {
nicif_err(nic_dev, drv, netdev, "Cannot allocate zero size entries\n");
return -EINVAL;
}
nic_dev->qps_irq_info = kzalloc(irq_size, GFP_KERNEL);
if (!nic_dev->qps_irq_info)
return -ENOMEM;
err = hinic_alloc_irqs(nic_dev->hwdev, SERVICE_T_NIC, nic_dev->num_qps,
nic_dev->qps_irq_info, &resp_irq_num);
if (err) {
nicif_err(nic_dev, drv, netdev, "Failed to alloc irqs\n");
kfree(nic_dev->qps_irq_info);
return err;
}
/* available irq number is less than rq numbers, adjust rq numbers */
if (resp_irq_num < nic_dev->num_qps) {
nic_dev->num_qps = resp_irq_num;
nic_dev->num_rss = nic_dev->num_qps;
hinic_setup_dcb_qps(nic_dev, nic_dev->num_qps);
nicif_warn(nic_dev, drv, netdev,
"Can not get enough irqs, adjust num_qps to %d\n",
nic_dev->num_qps);
/* after adjust num_qps, free the remaind irq */
for (i = nic_dev->num_qps; i < resp_irq_num; i++)
hinic_free_irq(nic_dev->hwdev, SERVICE_T_NIC,
nic_dev->qps_irq_info[i].irq_id);
}
nicif_info(nic_dev, drv, netdev, "Finally num_qps: %d, num_rss: %d\n",
nic_dev->num_qps, nic_dev->num_rss);
return 0;
}
static void hinic_destroy_num_qps(struct hinic_nic_dev *nic_dev)
{
u16 i;
for (i = 0; i < nic_dev->num_qps; i++)
hinic_free_irq(nic_dev->hwdev, SERVICE_T_NIC,
nic_dev->qps_irq_info[i].irq_id);
kfree(nic_dev->qps_irq_info);
}
static int hinic_poll(struct napi_struct *napi, int budget)
{
int tx_pkts, rx_pkts;
struct hinic_irq *irq_cfg = container_of(napi, struct hinic_irq, napi);
struct hinic_nic_dev *nic_dev = netdev_priv(irq_cfg->netdev);
rx_pkts = hinic_rx_poll(irq_cfg->rxq, budget);
tx_pkts = hinic_tx_poll(irq_cfg->txq, budget);
if (tx_pkts >= budget || rx_pkts >= budget)
return budget;
set_bit(HINIC_RESEND_ON, &irq_cfg->intr_flag);
rx_pkts += hinic_rx_poll(irq_cfg->rxq, budget - rx_pkts);
tx_pkts += hinic_tx_poll(irq_cfg->txq, budget - tx_pkts);
if (rx_pkts >= budget || tx_pkts >= budget) {
clear_bit(HINIC_RESEND_ON, &irq_cfg->intr_flag);
return budget;
}
napi_complete(napi);
if (!test_and_set_bit(HINIC_INTR_ON, &irq_cfg->intr_flag)) {
if (!HINIC_FUNC_IS_VF(nic_dev->hwdev))
hinic_set_msix_state(nic_dev->hwdev,
irq_cfg->msix_entry_idx,
HINIC_MSIX_ENABLE);
else if (!nic_dev->in_vm &&
(hinic_get_func_mode(nic_dev->hwdev) ==
FUNC_MOD_NORMAL_HOST))
enable_irq(irq_cfg->irq_id);
}
return max(tx_pkts, rx_pkts);
}
static void qp_add_napi(struct hinic_irq *irq_cfg)
{
struct hinic_nic_dev *nic_dev = netdev_priv(irq_cfg->netdev);
netif_napi_add_tx_weight(nic_dev->netdev, &irq_cfg->napi,
hinic_poll, nic_dev->poll_weight);
napi_enable(&irq_cfg->napi);
}
static void qp_del_napi(struct hinic_irq *irq_cfg)
{
napi_disable(&irq_cfg->napi);
netif_napi_del(&irq_cfg->napi);
}
static irqreturn_t qp_irq(int irq, void *data)
{
struct hinic_irq *irq_cfg = (struct hinic_irq *)data;
struct hinic_nic_dev *nic_dev = netdev_priv(irq_cfg->netdev);
u16 msix_entry_idx = irq_cfg->msix_entry_idx;
if (napi_schedule_prep(&irq_cfg->napi)) {
if (l2nic_interrupt_switch) {
/* Disable the interrupt until napi will be completed */
if (!HINIC_FUNC_IS_VF(nic_dev->hwdev)) {
hinic_set_msix_state(nic_dev->hwdev,
msix_entry_idx,
HINIC_MSIX_DISABLE);
} else if (!nic_dev->in_vm &&
(hinic_get_func_mode(nic_dev->hwdev) ==
FUNC_MOD_NORMAL_HOST)) {
disable_irq_nosync(irq_cfg->irq_id);
}
clear_bit(HINIC_INTR_ON, &irq_cfg->intr_flag);
}
hinic_misx_intr_clear_resend_bit(nic_dev->hwdev,
msix_entry_idx, 1);
clear_bit(HINIC_RESEND_ON, &irq_cfg->intr_flag);
__napi_schedule(&irq_cfg->napi);
} else if (!test_bit(HINIC_RESEND_ON, &irq_cfg->intr_flag)) {
hinic_misx_intr_clear_resend_bit(nic_dev->hwdev, msix_entry_idx,
1);
}
return IRQ_HANDLED;
}
static int hinic_request_irq(struct hinic_irq *irq_cfg, u16 q_id)
{
struct hinic_nic_dev *nic_dev = netdev_priv(irq_cfg->netdev);
struct nic_interrupt_info info = {0};
int err;
qp_add_napi(irq_cfg);
info.msix_index = irq_cfg->msix_entry_idx;
info.lli_set = 0;
info.interrupt_coalesc_set = 1;
info.pending_limt = nic_dev->intr_coalesce[q_id].pending_limt;
info.coalesc_timer_cfg =
nic_dev->intr_coalesce[q_id].coalesce_timer_cfg;
info.resend_timer_cfg = nic_dev->intr_coalesce[q_id].resend_timer_cfg;
nic_dev->rxqs[q_id].last_coalesc_timer_cfg =
nic_dev->intr_coalesce[q_id].coalesce_timer_cfg;
nic_dev->rxqs[q_id].last_pending_limt =
nic_dev->intr_coalesce[q_id].pending_limt;
err = hinic_set_interrupt_cfg(nic_dev->hwdev, info);
if (err) {
nicif_err(nic_dev, drv, irq_cfg->netdev,
"Failed to set RX interrupt coalescing attribute\n");
qp_del_napi(irq_cfg);
return err;
}
err = request_irq(irq_cfg->irq_id, &qp_irq, 0,
irq_cfg->irq_name, irq_cfg);
if (err) {
nicif_err(nic_dev, drv, irq_cfg->netdev, "Failed to request Rx irq\n");
qp_del_napi(irq_cfg);
return err;
}
/* assign the mask for this irq */
irq_set_affinity_hint(irq_cfg->irq_id, &irq_cfg->affinity_mask);
return 0;
}
static int set_interrupt_moder(struct hinic_nic_dev *nic_dev, u16 q_id,
u8 coalesc_timer_cfg, u8 pending_limt)
{
struct nic_interrupt_info interrupt_info = {0};
int err;
if (coalesc_timer_cfg == nic_dev->rxqs[q_id].last_coalesc_timer_cfg &&
pending_limt == nic_dev->rxqs[q_id].last_pending_limt)
return 0;
/* netdev not running or qp not in using,
* don't need to set coalesce to hw
*/
if (!test_bit(HINIC_INTF_UP, &nic_dev->flags) ||
q_id >= nic_dev->num_qps)
return 0;
interrupt_info.lli_set = 0;
interrupt_info.interrupt_coalesc_set = 1;
interrupt_info.coalesc_timer_cfg = coalesc_timer_cfg;
interrupt_info.pending_limt = pending_limt;
interrupt_info.msix_index = nic_dev->irq_cfg[q_id].msix_entry_idx;
interrupt_info.resend_timer_cfg =
nic_dev->intr_coalesce[q_id].resend_timer_cfg;
err = hinic_set_interrupt_cfg(nic_dev->hwdev, interrupt_info);
if (err) {
nicif_err(nic_dev, drv, nic_dev->netdev,
"Failed modifying moderation for Queue: %d\n", q_id);
} else {
nic_dev->rxqs[q_id].last_coalesc_timer_cfg = coalesc_timer_cfg;
nic_dev->rxqs[q_id].last_pending_limt = pending_limt;
}
return err;
}
static void __calc_coal_para(struct hinic_nic_dev *nic_dev,
struct hinic_intr_coal_info *q_coal, u64 rate,
u8 *coalesc_timer_cfg, u8 *pending_limt)
{
if (rate < q_coal->pkt_rate_low) {
*coalesc_timer_cfg = q_coal->rx_usecs_low;
*pending_limt = q_coal->rx_pending_limt_low;
} else if (rate > q_coal->pkt_rate_high) {
*coalesc_timer_cfg = q_coal->rx_usecs_high;
*pending_limt = q_coal->rx_pending_limt_high;
} else {
*coalesc_timer_cfg =
(u8)((rate - q_coal->pkt_rate_low) *
(q_coal->rx_usecs_high -
q_coal->rx_usecs_low) /
(q_coal->pkt_rate_high -
q_coal->pkt_rate_low) +
q_coal->rx_usecs_low);
if (nic_dev->in_vm)
*pending_limt = (u8)((rate - q_coal->pkt_rate_low) *
(q_coal->rx_pending_limt_high -
q_coal->rx_pending_limt_low) /
(q_coal->pkt_rate_high -
q_coal->pkt_rate_low) +
q_coal->rx_pending_limt_low);
else
*pending_limt = q_coal->rx_pending_limt_low;
}
}
static void update_queue_coal(struct hinic_nic_dev *nic_dev, u16 qid,
u64 rate, u64 avg_pkt_size, u64 tx_rate)
{
struct hinic_intr_coal_info *q_coal;
u8 coalesc_timer_cfg, pending_limt;
q_coal = &nic_dev->intr_coalesce[qid];
if ((rate > HINIC_RX_RATE_THRESH &&
avg_pkt_size > HINIC_AVG_PKT_SMALL) ||
(nic_dev->in_vm && rate > HINIC_RX_RATE_THRESH)) {
__calc_coal_para(nic_dev, q_coal, rate,
&coalesc_timer_cfg, &pending_limt);
} else {
coalesc_timer_cfg = HINIC_LOWEST_LATENCY;
pending_limt = q_coal->rx_pending_limt_low;
}
set_interrupt_moder(nic_dev, qid, coalesc_timer_cfg,
pending_limt);
}
#define SDI_VM_PPS_3W 30000
#define SDI_VM_PPS_5W 50000
#define SDI_VM_BPS_100MB 12500000
#define SDI_VM_BPS_1GB 125000000
static void update_queue_coal_sdi_vm(struct hinic_nic_dev *nic_dev,
u16 qid, u64 rx_pps, u64 rx_bps,
u64 tx_pps, u64 tx_bps)
{
struct hinic_intr_coal_info *q_coal = NULL;
u8 coalesc_timer_cfg, pending_limt;
q_coal = &nic_dev->intr_coalesce[qid];
if (qp_coalesc_use_drv_params_switch == 0) {
if (rx_pps < SDI_VM_PPS_3W &&
tx_pps < SDI_VM_PPS_3W &&
rx_bps < SDI_VM_BPS_100MB &&
tx_bps < SDI_VM_BPS_100MB) {
set_interrupt_moder(nic_dev, qid, 0, 0);
} else if (tx_pps > SDI_VM_PPS_3W &&
tx_pps < SDI_VM_PPS_5W &&
tx_bps > SDI_VM_BPS_1GB) {
set_interrupt_moder(nic_dev, qid, 7, 7);
} else {
__calc_coal_para(nic_dev, q_coal, rx_pps,
&coalesc_timer_cfg,
&pending_limt);
set_interrupt_moder(nic_dev, qid,
coalesc_timer_cfg,
pending_limt);
}
} else {
__calc_coal_para(nic_dev, q_coal, rx_pps,
&coalesc_timer_cfg,
&pending_limt);
set_interrupt_moder(nic_dev, qid, coalesc_timer_cfg,
pending_limt);
}
}
static void hinic_auto_moderation_work(struct work_struct *work)
{
struct delayed_work *delay = to_delayed_work(work);
struct hinic_nic_dev *nic_dev = container_of(delay,
struct hinic_nic_dev,
moderation_task);
unsigned long period = (unsigned long)(jiffies -
nic_dev->last_moder_jiffies);
u64 rx_packets, rx_bytes, rx_pkt_diff, rate, avg_pkt_size;
u64 tx_packets, tx_bytes, tx_pkt_diff, tx_rate, rx_bps, tx_bps;
u16 qid;
if (!test_bit(HINIC_INTF_UP, &nic_dev->flags))
return;
queue_delayed_work(nic_dev->workq, &nic_dev->moderation_task,
HINIC_MODERATONE_DELAY);
if (!nic_dev->adaptive_rx_coal || !period)
return;
for (qid = 0; qid < nic_dev->num_qps; qid++) {
rx_packets = nic_dev->rxqs[qid].rxq_stats.packets;
rx_bytes = nic_dev->rxqs[qid].rxq_stats.bytes;
tx_packets = nic_dev->txqs[qid].txq_stats.packets;
tx_bytes = nic_dev->txqs[qid].txq_stats.bytes;
rx_pkt_diff =
rx_packets - nic_dev->rxqs[qid].last_moder_packets;
avg_pkt_size = rx_pkt_diff ?
((unsigned long)(rx_bytes -
nic_dev->rxqs[qid].last_moder_bytes)) /
rx_pkt_diff : 0;
rate = rx_pkt_diff * HZ / period;
tx_pkt_diff =
tx_packets - nic_dev->txqs[qid].last_moder_packets;
tx_rate = tx_pkt_diff * HZ / period;
rx_bps = (unsigned long)(rx_bytes -
nic_dev->rxqs[qid].last_moder_bytes)
* HZ / period;
tx_bps = (unsigned long)(tx_bytes -
nic_dev->txqs[qid].last_moder_bytes)
* HZ / period;
if ((nic_dev->is_vm_slave && nic_dev->in_vm) ||
nic_dev->is_bm_slave) {
update_queue_coal_sdi_vm(nic_dev, qid, rate, rx_bps,
tx_rate, tx_bps);
} else {
update_queue_coal(nic_dev, qid, rate, avg_pkt_size,
tx_rate);
}
nic_dev->rxqs[qid].last_moder_packets = rx_packets;
nic_dev->rxqs[qid].last_moder_bytes = rx_bytes;
nic_dev->txqs[qid].last_moder_packets = tx_packets;
nic_dev->txqs[qid].last_moder_bytes = tx_bytes;
}
nic_dev->last_moder_jiffies = jiffies;
}
static void hinic_release_irq(struct hinic_irq *irq_cfg)
{
irq_set_affinity_hint(irq_cfg->irq_id, NULL);
synchronize_irq(irq_cfg->irq_id);
free_irq(irq_cfg->irq_id, irq_cfg);
qp_del_napi(irq_cfg);
}
static int hinic_qps_irq_init(struct hinic_nic_dev *nic_dev)
{
struct pci_dev *pdev = nic_dev->pdev;
struct irq_info *qp_irq_info;
struct hinic_irq *irq_cfg;
u16 q_id, i;
u32 local_cpu;
int err;
nic_dev->irq_cfg = kcalloc(nic_dev->num_qps, sizeof(*nic_dev->irq_cfg),
GFP_KERNEL);
if (!nic_dev->irq_cfg)
return -ENOMEM;
for (q_id = 0; q_id < nic_dev->num_qps; q_id++) {
qp_irq_info = &nic_dev->qps_irq_info[q_id];
irq_cfg = &nic_dev->irq_cfg[q_id];
irq_cfg->irq_id = qp_irq_info->irq_id;
irq_cfg->msix_entry_idx = qp_irq_info->msix_entry_idx;
irq_cfg->netdev = nic_dev->netdev;
irq_cfg->txq = &nic_dev->txqs[q_id];
irq_cfg->rxq = &nic_dev->rxqs[q_id];
nic_dev->rxqs[q_id].irq_cfg = irq_cfg;
if (nic_dev->force_affinity) {
irq_cfg->affinity_mask = nic_dev->affinity_mask;
} else {
local_cpu =
cpumask_local_spread(q_id,
dev_to_node(&pdev->dev));
cpumask_set_cpu(local_cpu, &irq_cfg->affinity_mask);
}
err = snprintf(irq_cfg->irq_name, sizeof(irq_cfg->irq_name),
"%s_qp%d", nic_dev->netdev->name, q_id);
if (err <= 0 || err >= (int)sizeof(irq_cfg->irq_name)) {
nic_err(&pdev->dev,
"Failed snprintf irq_name, function return(%d) and dest_len(%d)\n",
err, (int)sizeof(irq_cfg->irq_name));
goto req_tx_irq_err;
}
set_bit(HINIC_INTR_ON, &irq_cfg->intr_flag);
err = hinic_request_irq(irq_cfg, q_id);
if (err) {
nicif_err(nic_dev, drv, nic_dev->netdev, "Failed to request Rx irq\n");
goto req_tx_irq_err;
}
hinic_set_msix_state(nic_dev->hwdev,
irq_cfg->msix_entry_idx,
HINIC_MSIX_ENABLE);
}
INIT_DELAYED_WORK(&nic_dev->moderation_task,
hinic_auto_moderation_work);
return 0;
req_tx_irq_err:
for (i = 0; i < q_id; i++) {
hinic_set_msix_state(nic_dev->hwdev,
nic_dev->irq_cfg[i].msix_entry_idx,
HINIC_MSIX_DISABLE);
hinic_release_irq(&nic_dev->irq_cfg[i]);
}
kfree(nic_dev->irq_cfg);
return err;
}
static void hinic_qps_irq_deinit(struct hinic_nic_dev *nic_dev)
{
u16 q_id;
for (q_id = 0; q_id < nic_dev->num_qps; q_id++) {
hinic_set_msix_state(nic_dev->hwdev,
nic_dev->irq_cfg[q_id].msix_entry_idx,
HINIC_MSIX_DISABLE);
hinic_release_irq(&nic_dev->irq_cfg[q_id]);
}
kfree(nic_dev->irq_cfg);
}
int hinic_force_port_disable(struct hinic_nic_dev *nic_dev)
{
int err;
down(&nic_dev->port_state_sem);
err = hinic_set_port_enable(nic_dev->hwdev, false);
if (!err)
nic_dev->force_port_disable = true;
up(&nic_dev->port_state_sem);
return err;
}
int hinic_force_set_port_state(struct hinic_nic_dev *nic_dev, bool enable)
{
int err = 0;
down(&nic_dev->port_state_sem);
nic_dev->force_port_disable = false;
err = hinic_set_port_enable(nic_dev->hwdev, enable);
up(&nic_dev->port_state_sem);
return err;
}
int hinic_maybe_set_port_state(struct hinic_nic_dev *nic_dev, bool enable)
{
int err;
down(&nic_dev->port_state_sem);
/* Do nothing when force disable
* Port will disable when call force port disable
* and should not enable port when in force mode
*/
if (nic_dev->force_port_disable) {
up(&nic_dev->port_state_sem);
return 0;
}
err = hinic_set_port_enable(nic_dev->hwdev, enable);
up(&nic_dev->port_state_sem);
return err;
}
static void hinic_print_link_message(struct hinic_nic_dev *nic_dev,
u8 link_status)
{
if (nic_dev->link_status == link_status)
return;
nic_dev->link_status = link_status;
nicif_info(nic_dev, link, nic_dev->netdev, "Link is %s\n",
(link_status ? "up" : "down"));
}
int hinic_open(struct net_device *netdev)
{
struct hinic_nic_dev *nic_dev = netdev_priv(netdev);
u8 link_status = 0;
int err;
if (test_bit(HINIC_INTF_UP, &nic_dev->flags)) {
nicif_info(nic_dev, drv, netdev, "Netdev already open, do nothing\n");
return 0;
}
err = hinic_setup_num_qps(nic_dev);
if (err) {
nicif_err(nic_dev, drv, netdev, "Failed to setup num_qps\n");
return err;
}
err = hinic_create_qps(nic_dev->hwdev, nic_dev->num_qps,
nic_dev->sq_depth, nic_dev->rq_depth,
nic_dev->qps_irq_info, HINIC_MAX_SQ_BUFDESCS);
if (err) {
nicif_err(nic_dev, drv, netdev, "Failed to create queue pairs\n");
goto create_qps_err;
}
err = hinic_setup_qps_resources(nic_dev);
if (err)
goto setup_qps_resources_err;
err = hinic_init_qp_ctxts(nic_dev->hwdev);
if (err) {
nicif_err(nic_dev, drv, netdev, "Failed to init qp ctxts\n");
goto init_qp_ctxts_err;
}
err = hinic_set_port_mtu(nic_dev->hwdev, netdev->mtu);
if (err) {
nicif_err(nic_dev, drv, netdev, "Failed to set mtu\n");
goto mtu_err;
}
err = hinic_configure(nic_dev);
if (err) {
nicif_err(nic_dev, drv, netdev, "Failed to configure txrx\n");
goto cfg_err;
}
err = hinic_qps_irq_init(nic_dev);
if (err) {
nicif_err(nic_dev, drv, netdev, "Failed to qps irq init\n");
goto qps_irqs_init_err;
}
err = hinic_set_vport_enable(nic_dev->hwdev, true);
if (err) {
nicif_err(nic_dev, drv, netdev, "Failed to enable vport\n");
goto vport_enable_err;
}
err = hinic_maybe_set_port_state(nic_dev, true);
if (err) {
nicif_err(nic_dev, drv, netdev, "Failed to enable port\n");
goto port_enable_err;
}
set_bit(HINIC_INTF_UP, &nic_dev->flags);
netif_set_real_num_tx_queues(netdev, nic_dev->num_qps);
netif_set_real_num_rx_queues(netdev, nic_dev->num_qps);
netif_tx_wake_all_queues(netdev);
queue_delayed_work(nic_dev->workq, &nic_dev->moderation_task,
HINIC_MODERATONE_DELAY);
err = hinic_get_link_state(nic_dev->hwdev, &link_status);
if (!err && link_status) {
hinic_update_pf_bw(nic_dev->hwdev);
netif_carrier_on(netdev);
}
hinic_print_link_message(nic_dev, link_status);
if (!HINIC_FUNC_IS_VF(nic_dev->hwdev))
hinic_notify_all_vfs_link_changed(nic_dev->hwdev, link_status);
nicif_info(nic_dev, drv, nic_dev->netdev, "Netdev is up\n");
return 0;
port_enable_err:
hinic_set_vport_enable(nic_dev->hwdev, false);
vport_enable_err:
hinic_flush_sq_res(nic_dev->hwdev);
/* After set vport disable 100ms, no packets will be send to host */
msleep(100);
hinic_qps_irq_deinit(nic_dev);
qps_irqs_init_err:
hinic_remove_configure(nic_dev);
cfg_err:
mtu_err:
hinic_free_qp_ctxts(nic_dev->hwdev);
init_qp_ctxts_err:
hinic_free_all_rx_resources(netdev);
hinic_free_all_tx_resources(netdev);
setup_qps_resources_err:
hinic_free_qps(nic_dev->hwdev);
create_qps_err:
hinic_destroy_num_qps(nic_dev);
return err;
}
int hinic_close(struct net_device *netdev)
{
struct hinic_nic_dev *nic_dev = netdev_priv(netdev);
if (!test_and_clear_bit(HINIC_INTF_UP, &nic_dev->flags)) {
nicif_info(nic_dev, drv, netdev, "Netdev already close, do nothing\n");
return 0;
}
netif_carrier_off(netdev);
netif_tx_disable(netdev);
cancel_delayed_work_sync(&nic_dev->moderation_task);
if (hinic_get_chip_present_flag(nic_dev->hwdev)) {
if (!HINIC_FUNC_IS_VF(nic_dev->hwdev))
hinic_notify_all_vfs_link_changed(nic_dev->hwdev, 0);
hinic_maybe_set_port_state(nic_dev, false);
hinic_set_vport_enable(nic_dev->hwdev, false);
hinic_flush_txqs(netdev);
hinic_flush_sq_res(nic_dev->hwdev);
/* After set vport disable 100ms,
* no packets will be send to host
*/
msleep(100);
}
hinic_qps_irq_deinit(nic_dev);
hinic_remove_configure(nic_dev);
if (hinic_get_chip_present_flag(nic_dev->hwdev))
hinic_free_qp_ctxts(nic_dev->hwdev);
mutex_lock(&nic_dev->nic_mutex);
hinic_free_all_rx_resources(netdev);
hinic_free_all_tx_resources(netdev);
hinic_free_qps(nic_dev->hwdev);
hinic_destroy_num_qps(nic_dev);
mutex_unlock(&nic_dev->nic_mutex);
nicif_info(nic_dev, drv, nic_dev->netdev, "Netdev is down\n");
return 0;
}
static inline u32 calc_toeplitz_rss(u32 sip, u32 dip, u32 sport, u32 dport,
const u32 *rss_key)
{
u32 i, port, rss = 0;
port = (sport << 16) | dport;
/* The key - SIP, DIP, SPORT, DPORT */
for (i = 0; i < 32; i++)
if (sip & ((u32)1 << (u32)(31 - i)))
rss ^= (rss_key[0] << i) |
(u32)((u64)rss_key[1] >> (32 - i));
for (i = 0; i < 32; i++)
if (dip & ((u32)1 << (u32)(31 - i)))
rss ^= (rss_key[1] << i) |
(u32)((u64)rss_key[2] >> (32 - i));
for (i = 0; i < 32; i++)
if (port & ((u32)1 << (u32)(31 - i)))
rss ^= (rss_key[2] << i) |
(u32)((u64)rss_key[3] >> (32 - i));
return rss;
}
static u16 select_queue_by_toeplitz(struct net_device *dev,
struct sk_buff *skb,
unsigned int num_tx_queues)
{
struct hinic_nic_dev *nic_dev = netdev_priv(dev);
struct tcphdr *tcphdr;
struct iphdr *iphdr;
u32 hash = 0;
if (skb_rx_queue_recorded(skb)) {
hash = skb_get_rx_queue(skb);
while (unlikely(hash >= num_tx_queues))
hash -= num_tx_queues;
return (u16)hash;
}
/*lint -save -e778*/
if (vlan_get_protocol(skb) == htons(ETH_P_IP)) {
iphdr = ip_hdr(skb);
if (iphdr->protocol == IPPROTO_UDP ||
iphdr->protocol == IPPROTO_TCP) {
tcphdr = tcp_hdr(skb);
hash = calc_toeplitz_rss(ntohl(iphdr->daddr),
ntohl(iphdr->saddr),
ntohs(tcphdr->dest),
ntohs(tcphdr->source),
nic_dev->rss_hkey_user_be);
}
}
/*lint -restore*/
return (u16)nic_dev->rss_indir_user[hash & 0xFF];
}
static u16 hinic_select_queue(struct net_device *netdev, struct sk_buff *skb,
struct net_device *sb_dev)
{
struct hinic_nic_dev *nic_dev = netdev_priv(netdev);
skb->priority = skb->vlan_tci >> VLAN_PRIO_SHIFT;
if (netdev_get_num_tc(netdev) || !nic_dev->rss_hkey_user_be)
goto fallback;
if (nic_dev->rss_hash_engine == HINIC_RSS_HASH_ENGINE_TYPE_TOEP &&
test_bit(HINIC_SAME_RXTX, &nic_dev->flags))
return select_queue_by_toeplitz(netdev, skb,
netdev->real_num_tx_queues);
fallback:
return netdev_pick_tx(netdev, skb, NULL);
}
static void hinic_get_stats64(struct net_device *netdev,
struct rtnl_link_stats64 *stats)
{
struct hinic_nic_dev *nic_dev = netdev_priv(netdev);
struct hinic_txq_stats *txq_stats;
struct hinic_rxq_stats *rxq_stats;
struct hinic_txq *txq;
struct hinic_rxq *rxq;
u64 bytes, packets, dropped, errors;
unsigned int start;
int i;
bytes = 0;
packets = 0;
dropped = 0;
for (i = 0; i < nic_dev->max_qps; i++) {
if (!nic_dev->txqs)
break;
txq = &nic_dev->txqs[i];
txq_stats = &txq->txq_stats;
do {
start = u64_stats_fetch_begin(&txq_stats->syncp);
bytes += txq_stats->bytes;
packets += txq_stats->packets;
dropped += txq_stats->dropped;
} while (u64_stats_fetch_retry(&txq_stats->syncp, start));
}
stats->tx_packets = packets;
stats->tx_bytes = bytes;
stats->tx_dropped = dropped;
bytes = 0;
packets = 0;
errors = 0;
dropped = 0;
for (i = 0; i < nic_dev->max_qps; i++) {
if (!nic_dev->rxqs)
break;
rxq = &nic_dev->rxqs[i];
rxq_stats = &rxq->rxq_stats;
do {
start = u64_stats_fetch_begin(&rxq_stats->syncp);
bytes += rxq_stats->bytes;
packets += rxq_stats->packets;
errors += rxq_stats->csum_errors +
rxq_stats->other_errors;
dropped += rxq_stats->dropped;
} while (u64_stats_fetch_retry(&rxq_stats->syncp, start));
}
stats->rx_packets = packets;
stats->rx_bytes = bytes;
stats->rx_errors = errors;
stats->rx_dropped = dropped;
}
static void hinic_tx_timeout(struct net_device *netdev, unsigned int txqueue)
{
struct hinic_nic_dev *nic_dev = netdev_priv(netdev);
u16 msix_idx;
u8 q_id;
HINIC_NIC_STATS_INC(nic_dev, netdev_tx_timeout);
nicif_err(nic_dev, drv, netdev, "Tx timeout\n");
for (q_id = 0; q_id < nic_dev->num_qps; q_id++) {
if (!netif_xmit_stopped(netdev_get_tx_queue(netdev, q_id)))
continue;
msix_idx = nic_dev->irq_cfg[q_id].msix_entry_idx;
nicif_info(nic_dev, drv, netdev,
"txq%d: sw_pi: %d, hw_ci: %d, sw_ci: %d, napi->state: 0x%lx, msix mask: %d, intr_flag: 0x%lx\n",
q_id, hinic_dbg_get_sq_pi(nic_dev->hwdev, q_id),
hinic_get_sq_hw_ci(nic_dev->hwdev, q_id),
hinic_get_sq_local_ci(nic_dev->hwdev, q_id),
nic_dev->irq_cfg[q_id].napi.state,
hinic_get_msix_state(nic_dev->hwdev, msix_idx),
nic_dev->irq_cfg[q_id].intr_flag);
}
}
static int hinic_change_mtu(struct net_device *netdev, int new_mtu)
{
struct hinic_nic_dev *nic_dev = netdev_priv(netdev);
u32 mtu = (u32)new_mtu;
int err = 0;
err = hinic_set_port_mtu(nic_dev->hwdev, mtu);
if (err) {
nicif_err(nic_dev, drv, netdev, "Failed to change port mtu to %d\n",
new_mtu);
} else {
nicif_info(nic_dev, drv, nic_dev->netdev, "Change mtu from %d to %d\n",
netdev->mtu, new_mtu);
netdev->mtu = mtu;
}
return err;
}
static int hinic_set_mac_addr(struct net_device *netdev, void *addr)
{
struct hinic_nic_dev *nic_dev = netdev_priv(netdev);
struct sockaddr *saddr = addr;
u16 func_id;
int err;
if (!FUNC_SUPPORT_CHANGE_MAC(nic_dev->hwdev)) {
nicif_warn(nic_dev, drv, netdev,
"Current function don't support to set mac\n");
return -EOPNOTSUPP;
}
if (!is_valid_ether_addr(saddr->sa_data))
return -EADDRNOTAVAIL;
if (ether_addr_equal(netdev->dev_addr, saddr->sa_data)) {
nicif_info(nic_dev, drv, netdev,
"Already using mac address %pM\n",
saddr->sa_data);
return 0;
}
err = hinic_global_func_id_get(nic_dev->hwdev, &func_id);
if (err)
return err;
err = hinic_update_mac(nic_dev->hwdev, netdev->dev_addr, saddr->sa_data,
0, func_id);
if (err) {
nicif_err(nic_dev, drv, netdev, "Failed to update mac, err: %d\n",
err);
return err == HINIC_PF_SET_VF_ALREADY ? -EPERM : err;
}
eth_hw_addr_set(netdev, saddr->sa_data);
nicif_info(nic_dev, drv, netdev, "Set new mac address %pM\n",
saddr->sa_data);
return 0;
}
static int
hinic_vlan_rx_add_vid(struct net_device *netdev,
__always_unused __be16 proto,
u16 vid)
{
struct hinic_nic_dev *nic_dev = netdev_priv(netdev);
unsigned long *vlan_bitmap = nic_dev->vlan_bitmap;
u16 func_id;
u32 col, line;
int err;
col = VID_COL(nic_dev, vid);
line = VID_LINE(nic_dev, vid);
err = hinic_global_func_id_get(nic_dev->hwdev, &func_id);
if (err)
goto end;
err = hinic_add_vlan(nic_dev->hwdev, vid, func_id);
if (err) {
nicif_err(nic_dev, drv, netdev, "Failed to add vlan%d\n", vid);
goto end;
}
set_bit(col, &vlan_bitmap[line]);
nicif_info(nic_dev, drv, netdev, "Add vlan %d\n", vid);
end:
return err;
}
static int
hinic_vlan_rx_kill_vid(struct net_device *netdev,
__always_unused __be16 proto,
u16 vid)
{
struct hinic_nic_dev *nic_dev = netdev_priv(netdev);
unsigned long *vlan_bitmap = nic_dev->vlan_bitmap;
u16 func_id;
int err, col, line;
col = VID_COL(nic_dev, vid);
line = VID_LINE(nic_dev, vid);
err = hinic_global_func_id_get(nic_dev->hwdev, &func_id);
if (err)
goto end;
err = hinic_del_vlan(nic_dev->hwdev, vid, func_id);
if (err) {
nicif_err(nic_dev, drv, netdev, "Failed to delete vlan\n");
goto end;
}
clear_bit(col, &vlan_bitmap[line]);
nicif_info(nic_dev, drv, netdev, "Remove vlan %d\n", vid);
end:
return err;
}
#define FEATURES_OP_STR(op) ((op) ? "Enable" : "Disable")
static int set_feature_tso(struct hinic_nic_dev *nic_dev,
netdev_features_t wanted_features,
netdev_features_t features,
netdev_features_t *failed_features)
{
netdev_features_t changed = wanted_features ^ features;
bool en = !!(wanted_features & NETIF_F_TSO);
int err;
if (!(changed & NETIF_F_TSO))
return 0;
err = hinic_set_tx_tso(nic_dev->hwdev, en);
if (err) {
hinic_err(nic_dev, drv, "%s tso failed\n", FEATURES_OP_STR(en));
*failed_features |= NETIF_F_TSO;
} else {
hinic_info(nic_dev, drv, "%s tso success\n",
FEATURES_OP_STR(en));
}
return err;
}
static int set_feature_cvlan(struct hinic_nic_dev *nic_dev,
netdev_features_t wanted_features,
netdev_features_t features,
netdev_features_t *failed_features)
{
netdev_features_t changed = wanted_features ^ features;
netdev_features_t vlan_feature = NETIF_F_HW_VLAN_CTAG_RX;
bool en = !!(wanted_features & vlan_feature);
int err;
if (!(changed & vlan_feature))
return 0;
err = hinic_set_rx_vlan_offload(nic_dev->hwdev, en);
if (err) {
hinic_err(nic_dev, drv, "%s rxvlan failed\n",
FEATURES_OP_STR(en));
*failed_features |= vlan_feature;
} else {
hinic_info(nic_dev, drv, "%s rxvlan success\n",
FEATURES_OP_STR(en));
}
return err;
}
static int set_feature_rxcsum(struct hinic_nic_dev *nic_dev,
netdev_features_t wanted_features,
netdev_features_t features,
netdev_features_t *failed_features)
{
netdev_features_t changed = wanted_features ^ features;
bool en = !!(wanted_features & NETIF_F_RXCSUM);
int err;
if (!(changed & NETIF_F_RXCSUM))
return 0;
/* hw should always enable rx csum */
err = hinic_set_rx_csum_offload(nic_dev->hwdev,
HINIC_RX_CSUM_OFFLOAD_EN);
if (err) {
hinic_err(nic_dev, drv, "%s rx csum failed\n",
FEATURES_OP_STR(en));
*failed_features |= NETIF_F_RXCSUM;
} else {
hinic_info(nic_dev, drv, "%s rx csum success\n",
FEATURES_OP_STR(en));
}
return err;
}
static int set_feature_lro(struct hinic_nic_dev *nic_dev,
netdev_features_t wanted_features,
netdev_features_t features,
netdev_features_t *failed_features)
{
netdev_features_t changed = wanted_features ^ features;
bool en = !!(wanted_features & NETIF_F_LRO);
u32 lro_timer, lro_buf_size;
int err;
if (!(changed & NETIF_F_LRO))
return 0;
lro_timer = nic_dev->adaptive_cfg.lro.timer;
lro_buf_size = nic_dev->adaptive_cfg.lro.buffer_size;
err = hinic_set_rx_lro_state(nic_dev->hwdev, en, lro_timer,
lro_buf_size / nic_dev->rx_buff_len);
if (err) {
hinic_err(nic_dev, drv, "%s lro failed\n", FEATURES_OP_STR(en));
*failed_features |= NETIF_F_LRO;
} else {
hinic_info(nic_dev, drv, "%s lro success\n",
FEATURES_OP_STR(en));
}
return err;
}
static int set_features(struct hinic_nic_dev *nic_dev,
netdev_features_t pre_features,
netdev_features_t features)
{
netdev_features_t failed_features = 0;
u32 err;
err = (u32)set_feature_tso(nic_dev, features, pre_features,
&failed_features);
err |= (u32)set_feature_cvlan(nic_dev, features, pre_features,
&failed_features);
err |= (u32)set_feature_rxcsum(nic_dev, features, pre_features,
&failed_features);
err |= (u32)set_feature_lro(nic_dev, features, pre_features,
&failed_features);
if (err) {
nic_dev->netdev->features = features ^ failed_features;
return -EIO;
}
return 0;
}
static int hinic_set_features(struct net_device *netdev,
netdev_features_t features)
{
struct hinic_nic_dev *nic_dev = netdev_priv(netdev);
return set_features(nic_dev, nic_dev->netdev->features,
features);
}
static netdev_features_t hinic_fix_features(struct net_device *netdev,
netdev_features_t features)
{
/* If Rx checksum is disabled, then LRO should also be disabled */
if (!(features & NETIF_F_RXCSUM))
features &= ~NETIF_F_LRO;
return features;
}
static int hinic_set_default_hw_feature(struct hinic_nic_dev *nic_dev)
{
int err;
if (!HINIC_FUNC_IS_VF(nic_dev->hwdev)) {
if (FUNC_SUPPORT_DCB(nic_dev->hwdev)) {
err = hinic_dcb_reset_hw_config(nic_dev);
if (err) {
nic_err(&nic_dev->pdev->dev, "Failed to reset hw dcb configuration\n");
return -EFAULT;
}
}
if (FUNC_SUPPORT_PORT_SETTING(nic_dev->hwdev)) {
err = hinic_reset_port_link_cfg(nic_dev->hwdev);
if (err)
return -EFAULT;
}
hinic_set_anti_attack(nic_dev->hwdev, true);
if (set_link_status_follow < HINIC_LINK_FOLLOW_STATUS_MAX &&
FUNC_SUPPORT_PORT_SETTING(nic_dev->hwdev)) {
err = hinic_set_link_status_follow(nic_dev->hwdev,
set_link_status_follow);
if (err == HINIC_MGMT_CMD_UNSUPPORTED)
nic_warn(&nic_dev->pdev->dev,
"Current version of firmware don't support to set link status follow port status\n");
}
}
/* enable all hw features in netdev->features */
return set_features(nic_dev, ~nic_dev->netdev->features,
nic_dev->netdev->features);
}
static int hinic_setup_tc_mqprio(struct net_device *dev,
struct tc_mqprio_qopt *mqprio)
{
mqprio->hw = TC_MQPRIO_HW_OFFLOAD_TCS;
return hinic_setup_tc(dev, mqprio->num_tc);
}
static int __hinic_setup_tc(struct net_device *dev, enum tc_setup_type type,
void *type_data)
{
switch (type) {
case TC_SETUP_QDISC_MQPRIO:
return hinic_setup_tc_mqprio(dev, type_data);
default:
return -EOPNOTSUPP;
}
}
#ifdef CONFIG_NET_POLL_CONTROLLER
static void hinic_netpoll(struct net_device *netdev)
{
struct hinic_nic_dev *nic_dev = netdev_priv(netdev);
u16 i;
for (i = 0; i < nic_dev->num_qps; i++)
napi_schedule(&nic_dev->irq_cfg[i].napi);
}
#endif /* CONFIG_NET_POLL_CONTROLLER */
static int hinic_uc_sync(struct net_device *netdev, u8 *addr)
{
struct hinic_nic_dev *nic_dev = netdev_priv(netdev);
u16 func_id;
int err;
err = hinic_global_func_id_get(nic_dev->hwdev, &func_id);
if (err)
return err;
err = hinic_set_mac(nic_dev->hwdev, addr, 0, func_id);
return err;
}
static int hinic_uc_unsync(struct net_device *netdev, u8 *addr)
{
struct hinic_nic_dev *nic_dev = netdev_priv(netdev);
u16 func_id;
int err;
/* The addr is in use */
if (ether_addr_equal(addr, netdev->dev_addr))
return 0;
err = hinic_global_func_id_get(nic_dev->hwdev, &func_id);
if (err)
return err;
err = hinic_del_mac(nic_dev->hwdev, addr, 0, func_id);
return err;
}
static void hinic_clean_mac_list_filter(struct hinic_nic_dev *nic_dev)
{
struct net_device *netdev = nic_dev->netdev;
struct hinic_mac_filter *f, *ftmp;
list_for_each_entry_safe(f, ftmp, &nic_dev->uc_filter_list, list) {
if (f->state == HINIC_MAC_HW_SYNCED)
hinic_uc_unsync(netdev, f->addr);
list_del(&f->list);
kfree(f);
}
list_for_each_entry_safe(f, ftmp, &nic_dev->mc_filter_list, list) {
if (f->state == HINIC_MAC_HW_SYNCED)
hinic_uc_unsync(netdev, f->addr);
list_del(&f->list);
kfree(f);
}
}
static struct hinic_mac_filter *hinic_find_mac(struct list_head *filter_list,
u8 *addr)
{
struct hinic_mac_filter *f;
list_for_each_entry(f, filter_list, list) {
if (ether_addr_equal(addr, f->addr))
return f;
}
return NULL;
}
static struct hinic_mac_filter
*hinic_add_filter(struct hinic_nic_dev *nic_dev,
struct list_head *mac_filter_list, u8 *addr)
{
struct hinic_mac_filter *f;
f = kzalloc(sizeof(*f), GFP_ATOMIC);
if (!f)
goto out;
memcpy(f->addr, addr, ETH_ALEN);
INIT_LIST_HEAD(&f->list);
list_add_tail(&f->list, mac_filter_list);
f->state = HINIC_MAC_WAIT_HW_SYNC;
set_bit(HINIC_MAC_FILTER_CHANGED, &nic_dev->flags);
out:
return f;
}
static void hinic_del_filter(struct hinic_nic_dev *nic_dev,
struct hinic_mac_filter *f)
{
set_bit(HINIC_MAC_FILTER_CHANGED, &nic_dev->flags);
if (f->state == HINIC_MAC_WAIT_HW_SYNC) {
/* have not added to hw, delete it directly */
list_del(&f->list);
kfree(f);
return;
}
f->state = HINIC_MAC_WAIT_HW_UNSYNC;
}
static struct hinic_mac_filter
*hinic_mac_filter_entry_clone(struct hinic_mac_filter *src)
{
struct hinic_mac_filter *f;
f = kzalloc(sizeof(*f), GFP_ATOMIC);
if (!f)
return NULL;
*f = *src;
INIT_LIST_HEAD(&f->list);
return f;
}
static void hinic_undo_del_filter_entries(struct list_head *filter_list,
struct list_head *from)
{
struct hinic_mac_filter *f, *ftmp;
list_for_each_entry_safe(f, ftmp, from, list) {
if (hinic_find_mac(filter_list, f->addr))
continue;
if (f->state == HINIC_MAC_HW_SYNCED)
f->state = HINIC_MAC_WAIT_HW_UNSYNC;
list_move_tail(&f->list, filter_list);
}
}
static void hinic_undo_add_filter_entries(struct list_head *filter_list,
struct list_head *from)
{
struct hinic_mac_filter *f, *ftmp, *tmp;
list_for_each_entry_safe(f, ftmp, from, list) {
tmp = hinic_find_mac(filter_list, f->addr);
if (tmp && tmp->state == HINIC_MAC_HW_SYNCED)
tmp->state = HINIC_MAC_WAIT_HW_SYNC;
}
}
static void hinic_cleanup_filter_list(struct list_head *head)
{
struct hinic_mac_filter *f, *ftmp;
list_for_each_entry_safe(f, ftmp, head, list) {
list_del(&f->list);
kfree(f);
}
}
static int hinic_mac_filter_sync_hw(struct hinic_nic_dev *nic_dev,
struct list_head *del_list,
struct list_head *add_list)
{
struct net_device *netdev = nic_dev->netdev;
struct hinic_mac_filter *f, *ftmp;
int err = 0, add_count = 0;
if (!list_empty(del_list)) {
list_for_each_entry_safe(f, ftmp, del_list, list) {
err = hinic_uc_unsync(netdev, f->addr);
if (err) { /* ignore errors when delete mac */
nic_err(&nic_dev->pdev->dev, "Failed to delete mac\n");
}
list_del(&f->list);
kfree(f);
}
}
if (!list_empty(add_list)) {
list_for_each_entry_safe(f, ftmp, add_list, list) {
err = hinic_uc_sync(netdev, f->addr);
if (err) {
nic_err(&nic_dev->pdev->dev, "Failed to add mac\n");
return err;
}
add_count++;
list_del(&f->list);
kfree(f);
}
}
return add_count;
}
static int hinic_mac_filter_sync(struct hinic_nic_dev *nic_dev,
struct list_head *mac_filter_list, bool uc)
{
struct net_device *netdev = nic_dev->netdev;
struct list_head tmp_del_list, tmp_add_list;
struct hinic_mac_filter *f, *ftmp, *fclone;
int err = 0, add_count = 0;
INIT_LIST_HEAD(&tmp_del_list);
INIT_LIST_HEAD(&tmp_add_list);
list_for_each_entry_safe(f, ftmp, mac_filter_list, list) {
if (f->state != HINIC_MAC_WAIT_HW_UNSYNC)
continue;
f->state = HINIC_MAC_HW_UNSYNCED;
list_move_tail(&f->list, &tmp_del_list);
}
list_for_each_entry_safe(f, ftmp, mac_filter_list, list) {
if (f->state != HINIC_MAC_WAIT_HW_SYNC)
continue;
fclone = hinic_mac_filter_entry_clone(f);
if (!fclone) {
err = -ENOMEM;
break;
}
f->state = HINIC_MAC_HW_SYNCED;
list_add_tail(&fclone->list, &tmp_add_list);
}
if (err) {
hinic_undo_del_filter_entries(mac_filter_list, &tmp_del_list);
hinic_undo_add_filter_entries(mac_filter_list, &tmp_add_list);
nicif_err(nic_dev, drv, netdev, "Failed to clone mac_filter_entry\n");
}
if (err) {
hinic_cleanup_filter_list(&tmp_del_list);
hinic_cleanup_filter_list(&tmp_add_list);
return -ENOMEM;
}
add_count =
hinic_mac_filter_sync_hw(nic_dev, &tmp_del_list, &tmp_add_list);
if (list_empty(&tmp_add_list))
return add_count;
/* there are errors when add mac to hw, delete all mac in hw */
hinic_undo_add_filter_entries(mac_filter_list, &tmp_add_list);
/* VF don't support to enter promisc mode,
* so we can't delete any other uc mac
*/
if (!HINIC_FUNC_IS_VF(nic_dev->hwdev) || !uc) {
list_for_each_entry_safe(f, ftmp, mac_filter_list, list) {
if (f->state != HINIC_MAC_HW_SYNCED)
continue;
fclone = hinic_mac_filter_entry_clone(f);
if (!fclone)
break;
f->state = HINIC_MAC_WAIT_HW_SYNC;
list_add_tail(&fclone->list, &tmp_del_list);
}
}
hinic_cleanup_filter_list(&tmp_add_list);
hinic_mac_filter_sync_hw(nic_dev, &tmp_del_list, &tmp_add_list);
/* need to enter promisc/allmulti mode */
return -ENOMEM;
}
static void hinic_mac_filter_sync_all(struct hinic_nic_dev *nic_dev)
{
struct net_device *netdev = nic_dev->netdev;
int add_count;
if (test_bit(HINIC_MAC_FILTER_CHANGED, &nic_dev->flags)) {
clear_bit(HINIC_MAC_FILTER_CHANGED, &nic_dev->flags);
add_count = hinic_mac_filter_sync(nic_dev,
&nic_dev->uc_filter_list,
true);
if (add_count < 0 && !HINIC_FUNC_IS_VF(nic_dev->hwdev)) {
set_bit(HINIC_PROMISC_FORCE_ON, &nic_dev->rx_mod_state);
nicif_info(nic_dev, drv, netdev, "Promisc mode forced on\n");
} else if (add_count) {
clear_bit(HINIC_PROMISC_FORCE_ON,
&nic_dev->rx_mod_state);
}
add_count = hinic_mac_filter_sync(nic_dev,
&nic_dev->mc_filter_list,
false);
if (add_count < 0) {
set_bit(HINIC_ALLMULTI_FORCE_ON,
&nic_dev->rx_mod_state);
nicif_info(nic_dev, drv, netdev, "All multicast mode forced on\n");
} else if (add_count) {
clear_bit(HINIC_ALLMULTI_FORCE_ON,
&nic_dev->rx_mod_state);
}
}
}
#define HINIC_DEFAULT_RX_MODE (HINIC_RX_MODE_UC | HINIC_RX_MODE_MC | \
HINIC_RX_MODE_BC)
static void hinic_update_mac_filter(struct hinic_nic_dev *nic_dev,
struct netdev_hw_addr_list *src_list,
struct list_head *filter_list)
{
struct netdev_hw_addr *ha;
struct hinic_mac_filter *f, *ftmp, *filter;
/* add addr if not already in the filter list */
netif_addr_lock_bh(nic_dev->netdev);
netdev_hw_addr_list_for_each(ha, src_list) {
filter = hinic_find_mac(filter_list, ha->addr);
if (!filter)
hinic_add_filter(nic_dev, filter_list, ha->addr);
else if (filter->state == HINIC_MAC_WAIT_HW_UNSYNC)
filter->state = HINIC_MAC_HW_SYNCED;
}
netif_addr_unlock_bh(nic_dev->netdev);
/* delete addr if not in netdev list */
list_for_each_entry_safe(f, ftmp, filter_list, list) {
bool found = false;
netif_addr_lock_bh(nic_dev->netdev);
netdev_hw_addr_list_for_each(ha, src_list)
if (ether_addr_equal(ha->addr, f->addr)) {
found = true;
break;
}
netif_addr_unlock_bh(nic_dev->netdev);
if (found)
continue;
hinic_del_filter(nic_dev, f);
}
}
static void __update_mac_filter(struct hinic_nic_dev *nic_dev)
{
struct net_device *netdev = nic_dev->netdev;
if (test_and_clear_bit(HINIC_UPDATE_MAC_FILTER, &nic_dev->flags)) {
hinic_update_mac_filter(nic_dev, &netdev->uc,
&nic_dev->uc_filter_list);
hinic_update_mac_filter(nic_dev, &netdev->mc,
&nic_dev->mc_filter_list);
}
}
static void hinic_set_rx_mode_work(struct work_struct *work)
{
struct hinic_nic_dev *nic_dev =
container_of(work, struct hinic_nic_dev, rx_mode_work);
struct net_device *netdev = nic_dev->netdev;
int promisc_en = 0, allmulti_en = 0;
int err = 0;
__update_mac_filter(nic_dev);
hinic_mac_filter_sync_all(nic_dev);
/* VF don't support to enter promisc mode */
if (!HINIC_FUNC_IS_VF(nic_dev->hwdev)) {
promisc_en = !!(netdev->flags & IFF_PROMISC) ||
test_bit(HINIC_PROMISC_FORCE_ON,
&nic_dev->rx_mod_state);
}
allmulti_en = !!(netdev->flags & IFF_ALLMULTI) ||
test_bit(HINIC_ALLMULTI_FORCE_ON, &nic_dev->rx_mod_state);
if (promisc_en !=
test_bit(HINIC_HW_PROMISC_ON, &nic_dev->rx_mod_state) ||
allmulti_en !=
test_bit(HINIC_HW_ALLMULTI_ON, &nic_dev->rx_mod_state)) {
enum hinic_rx_mod rx_mod = HINIC_DEFAULT_RX_MODE;
rx_mod |= (promisc_en ? HINIC_RX_MODE_PROMISC : 0);
rx_mod |= (allmulti_en ? HINIC_RX_MODE_MC_ALL : 0);
/* FOR DEBUG */
if (promisc_en !=
test_bit(HINIC_HW_PROMISC_ON, &nic_dev->rx_mod_state))
nicif_info(nic_dev, drv, netdev,
"%s promisc mode\n",
promisc_en ? "Enter" : "Left");
if (allmulti_en !=
test_bit(HINIC_HW_ALLMULTI_ON, &nic_dev->rx_mod_state))
nicif_info(nic_dev, drv, netdev,
"%s all_multi mode\n",
allmulti_en ? "Enter" : "Left");
err = hinic_set_rx_mode(nic_dev->hwdev, rx_mod);
if (!err) {
promisc_en ?
set_bit(HINIC_HW_PROMISC_ON, &nic_dev->rx_mod_state) :
clear_bit(HINIC_HW_PROMISC_ON, &nic_dev->rx_mod_state);
allmulti_en ?
set_bit(HINIC_HW_ALLMULTI_ON, &nic_dev->rx_mod_state) :
clear_bit(HINIC_HW_ALLMULTI_ON, &nic_dev->rx_mod_state);
} else {
nicif_err(nic_dev, drv, netdev, "Failed to set rx_mode\n");
}
}
}
static void hinic_nic_set_rx_mode(struct net_device *netdev)
{
struct hinic_nic_dev *nic_dev = netdev_priv(netdev);
if (netdev_uc_count(netdev) != nic_dev->netdev_uc_cnt ||
netdev_mc_count(netdev) != nic_dev->netdev_mc_cnt) {
set_bit(HINIC_UPDATE_MAC_FILTER, &nic_dev->flags);
nic_dev->netdev_uc_cnt = netdev_uc_count(netdev);
nic_dev->netdev_mc_cnt = netdev_mc_count(netdev);
}
if (FUNC_SUPPORT_RX_MODE(nic_dev->hwdev))
queue_work(nic_dev->workq, &nic_dev->rx_mode_work);
}
static const struct net_device_ops hinic_netdev_ops = {
.ndo_open = hinic_open,
.ndo_stop = hinic_close,
.ndo_start_xmit = hinic_xmit_frame,
.ndo_get_stats64 = hinic_get_stats64,
.ndo_tx_timeout = hinic_tx_timeout,
.ndo_select_queue = hinic_select_queue,
.ndo_change_mtu = hinic_change_mtu,
.ndo_set_mac_address = hinic_set_mac_addr,
.ndo_validate_addr = eth_validate_addr,
.ndo_vlan_rx_add_vid = hinic_vlan_rx_add_vid,
.ndo_vlan_rx_kill_vid = hinic_vlan_rx_kill_vid,
.ndo_set_vf_mac = hinic_ndo_set_vf_mac,
.ndo_set_vf_vlan = hinic_ndo_set_vf_vlan,
.ndo_set_vf_rate = hinic_ndo_set_vf_bw,
.ndo_set_vf_spoofchk = hinic_ndo_set_vf_spoofchk,
.ndo_set_vf_trust = hinic_ndo_set_vf_trust,
.ndo_get_vf_config = hinic_ndo_get_vf_config,
.ndo_setup_tc = __hinic_setup_tc,
#ifdef CONFIG_NET_POLL_CONTROLLER
.ndo_poll_controller = hinic_netpoll,
#endif /* CONFIG_NET_POLL_CONTROLLER */
.ndo_set_rx_mode = hinic_nic_set_rx_mode,
.ndo_set_vf_link_state = hinic_ndo_set_vf_link_state,
.ndo_fix_features = hinic_fix_features,
.ndo_set_features = hinic_set_features,
};
static const struct net_device_ops hinicvf_netdev_ops = {
.ndo_open = hinic_open,
.ndo_stop = hinic_close,
.ndo_start_xmit = hinic_xmit_frame,
.ndo_get_stats64 = hinic_get_stats64,
.ndo_tx_timeout = hinic_tx_timeout,
.ndo_select_queue = hinic_select_queue,
.ndo_change_mtu = hinic_change_mtu,
.ndo_set_mac_address = hinic_set_mac_addr,
.ndo_validate_addr = eth_validate_addr,
.ndo_vlan_rx_add_vid = hinic_vlan_rx_add_vid,
.ndo_vlan_rx_kill_vid = hinic_vlan_rx_kill_vid,
#ifdef CONFIG_NET_POLL_CONTROLLER
.ndo_poll_controller = hinic_netpoll,
#endif /* CONFIG_NET_POLL_CONTROLLER */
.ndo_set_rx_mode = hinic_nic_set_rx_mode,
.ndo_fix_features = hinic_fix_features,
.ndo_set_features = hinic_set_features,
};
static void netdev_feature_init(struct net_device *netdev)
{
struct hinic_nic_dev *nic_dev = netdev_priv(netdev);
netdev_features_t hw_features;
netdev->features = NETIF_F_SG | NETIF_F_HIGHDMA |
NETIF_F_IP_CSUM | NETIF_F_IPV6_CSUM |
NETIF_F_TSO |
NETIF_F_TSO6 | NETIF_F_RXCSUM;
if (FUNC_SUPPORT_SCTP_CRC(nic_dev->hwdev))
netdev->features |= NETIF_F_SCTP_CRC;
netdev->vlan_features = netdev->features;
if (FUNC_SUPPORT_ENCAP_TSO_CSUM(nic_dev->hwdev))
netdev->features |= NETIF_F_GSO_UDP_TUNNEL |
NETIF_F_GSO_UDP_TUNNEL_CSUM;
if (FUNC_SUPPORT_HW_VLAN(nic_dev->hwdev)) {
netdev->features |= NETIF_F_HW_VLAN_CTAG_TX;
netdev->features |= NETIF_F_HW_VLAN_CTAG_RX;
}
/* copy netdev features into list of user selectable features */
hw_features = netdev->hw_features;
hw_features |= netdev->features;
if (FUNC_SUPPORT_LRO(nic_dev->hwdev)) {
/* LRO is disable in default, only set hw features */
hw_features |= NETIF_F_LRO;
/* Enable LRO */
if (nic_dev->adaptive_cfg.lro.enable &&
!HINIC_FUNC_IS_VF(nic_dev->hwdev))
netdev->features |= NETIF_F_LRO;
}
netdev->hw_features = hw_features;
/* Set after hw_features because this could not be part of
* hw_features
*/
netdev->features |= NETIF_F_HW_VLAN_CTAG_FILTER;
netdev->priv_flags |= IFF_UNICAST_FLT;
if (FUNC_SUPPORT_ENCAP_TSO_CSUM(nic_dev->hwdev)) {
netdev->hw_enc_features |= NETIF_F_IP_CSUM | NETIF_F_IPV6_CSUM
| NETIF_F_SCTP_CRC | NETIF_F_SG;
netdev->hw_enc_features |= NETIF_F_TSO | NETIF_F_TSO6
| NETIF_F_TSO_ECN
| NETIF_F_GSO_UDP_TUNNEL_CSUM
| NETIF_F_GSO_UDP_TUNNEL;
}
}
#define MOD_PARA_VALIDATE_NUM_QPS(nic_dev, num_qps, out_qps) { \
if ((num_qps) > (nic_dev)->max_qps) \
nic_warn(&(nic_dev)->pdev->dev, \
"Module Parameter %s value %d is out of range, "\
"Maximum value for the device: %d, using %d\n",\
#num_qps, num_qps, (nic_dev)->max_qps, \
(nic_dev)->max_qps); \
if (!(num_qps) || (num_qps) > (nic_dev)->max_qps) \
out_qps = (nic_dev)->max_qps; \
else \
out_qps = num_qps; \
}
static void hinic_try_to_enable_rss(struct hinic_nic_dev *nic_dev)
{
u8 prio_tc[HINIC_DCB_UP_MAX] = {0};
int i, node, err = 0;
u16 num_cpus = 0;
enum hinic_service_mode service_mode =
hinic_get_service_mode(nic_dev->hwdev);
nic_dev->max_qps = hinic_func_max_nic_qnum(nic_dev->hwdev);
if (nic_dev->max_qps <= 1) {
clear_bit(HINIC_RSS_ENABLE, &nic_dev->flags);
nic_dev->rss_limit = nic_dev->max_qps;
nic_dev->num_qps = nic_dev->max_qps;
nic_dev->num_rss = nic_dev->max_qps;
return;
}
err = hinic_rss_template_alloc(nic_dev->hwdev, &nic_dev->rss_tmpl_idx);
if (err) {
if (err == -ENOSPC)
nic_warn(&nic_dev->pdev->dev,
"Failed to alloc tmpl_idx for rss, table is full\n");
else
nic_err(&nic_dev->pdev->dev,
"Failed to alloc tmpl_idx for rss, can't enable rss for this function\n");
clear_bit(HINIC_RSS_ENABLE, &nic_dev->flags);
nic_dev->max_qps = 1;
nic_dev->rss_limit = nic_dev->max_qps;
nic_dev->num_qps = nic_dev->max_qps;
nic_dev->num_rss = nic_dev->max_qps;
return;
}
set_bit(HINIC_RSS_ENABLE, &nic_dev->flags);
nic_dev->max_qps = hinic_func_max_nic_qnum(nic_dev->hwdev);
MOD_PARA_VALIDATE_NUM_QPS(nic_dev, num_qps, nic_dev->num_qps);
/* To reduce memory footprint in ovs mode.
* VF can't get board info correctly with early pf driver.
*/
if ((hinic_get_func_mode(nic_dev->hwdev) == FUNC_MOD_NORMAL_HOST) &&
service_mode == HINIC_WORK_MODE_OVS &&
hinic_func_type(nic_dev->hwdev) != TYPE_VF)
MOD_PARA_VALIDATE_NUM_QPS(nic_dev, ovs_num_qps,
nic_dev->num_qps);
for (i = 0; i < (int)num_online_cpus(); i++) {
node = (int)cpu_to_node(i);
if (node == dev_to_node(&nic_dev->pdev->dev))
num_cpus++;
}
if (!num_cpus)
num_cpus = (u16)num_online_cpus();
nic_dev->num_qps = min_t(u16, nic_dev->num_qps, num_cpus);
nic_dev->rss_limit = nic_dev->num_qps;
nic_dev->num_rss = nic_dev->num_qps;
hinic_init_rss_parameters(nic_dev->netdev);
hinic_set_hw_rss_parameters(nic_dev->netdev, 0, 0, prio_tc);
}
static int hinic_sw_init(struct hinic_nic_dev *adapter)
{
struct net_device *netdev = adapter->netdev;
u8 mac[ETH_ALEN];
u16 func_id;
int err = 0;
sema_init(&adapter->port_state_sem, 1);
err = hinic_dcb_init(adapter);
if (err) {
nic_err(&adapter->pdev->dev, "Failed to init dcb\n");
return -EFAULT;
}
if (HINIC_FUNC_IS_VF(adapter->hwdev)) {
err = hinic_sq_cos_mapping(netdev);
if (err) {
nic_err(&adapter->pdev->dev, "Failed to set sq_cos_mapping\n");
return -EFAULT;
}
}
adapter->sq_depth = HINIC_SQ_DEPTH;
adapter->rq_depth = HINIC_RQ_DEPTH;
hinic_try_to_enable_rss(adapter);
err = hinic_get_default_mac(adapter->hwdev, mac);
if (err) {
nic_err(&adapter->pdev->dev, "Failed to get MAC address\n");
goto get_mac_err;
}
eth_hw_addr_set(netdev, mac);
if (!is_valid_ether_addr(netdev->dev_addr)) {
if (!HINIC_FUNC_IS_VF(adapter->hwdev)) {
nic_err(&adapter->pdev->dev, "Invalid MAC address\n");
err = -EIO;
goto err_mac;
}
nic_info(&adapter->pdev->dev, "Invalid MAC address %pM, using random\n",
netdev->dev_addr);
eth_hw_addr_random(netdev);
}
err = hinic_global_func_id_get(adapter->hwdev, &func_id);
if (err)
goto func_id_err;
err = hinic_set_mac(adapter->hwdev, netdev->dev_addr, 0, func_id);
/* When this is VF driver, we must consider that PF has already set VF
* MAC, and we can't consider this condition is error status during
* driver probe procedure.
*/
if (err && err != HINIC_PF_SET_VF_ALREADY) {
nic_err(&adapter->pdev->dev, "Failed to set default MAC\n");
goto set_mac_err;
}
/* MTU range: 256 - 9600 */
netdev->min_mtu = HINIC_MIN_MTU_SIZE;
netdev->max_mtu = HINIC_MAX_JUMBO_FRAME_SIZE;
return 0;
set_mac_err:
func_id_err:
err_mac:
get_mac_err:
if (test_bit(HINIC_RSS_ENABLE, &adapter->flags))
hinic_rss_template_free(adapter->hwdev, adapter->rss_tmpl_idx);
return err;
}
static void hinic_assign_netdev_ops(struct hinic_nic_dev *adapter)
{
if (!HINIC_FUNC_IS_VF(adapter->hwdev)) {
adapter->netdev->netdev_ops = &hinic_netdev_ops;
if (FUNC_SUPPORT_DCB(adapter->hwdev))
adapter->netdev->dcbnl_ops = &hinic_dcbnl_ops;
hinic_set_ethtool_ops(adapter->netdev);
} else {
adapter->netdev->netdev_ops = &hinicvf_netdev_ops;
hinicvf_set_ethtool_ops(adapter->netdev);
}
adapter->netdev->watchdog_timeo = 5 * HZ;
}
#define HINIC_DFT_PG_10GE_TXRX_MSIX_PENDING_LIMIT 1
#define HINIC_DFT_PG_10GE_TXRX_MSIX_COALESC_TIMER 1
#define HINIC_DFT_PG_25GE_TXRX_MSIX_PENDING_LIMIT 2
#define HINIC_DFT_PG_25GE_TXRX_MSIX_COALESC_TIMER 2
#define HINIC_DFT_PG_ARM_25GE_TXRX_MSIX_COALESC_TIMER 3
#define HINIC_DFT_PG_100GE_TXRX_MSIX_PENDING_LIMIT 2
#define HINIC_DFT_PG_100GE_TXRX_MSIX_COALESC_TIMER 2
#define HINIC_DFT_PG_ARM_100GE_TXRX_MSIX_COALESC_TIMER 3
static void update_queue_coal_param(struct hinic_nic_dev *nic_dev,
struct pci_device_id *id, u16 qid)
{
struct hinic_intr_coal_info *info = NULL;
info = &nic_dev->intr_coalesce[qid];
if (!nic_dev->intr_coal_set_flag) {
switch (id->driver_data) {
case HINIC_BOARD_PG_TP_10GE:
info->pending_limt =
HINIC_DFT_PG_10GE_TXRX_MSIX_PENDING_LIMIT;
info->coalesce_timer_cfg =
HINIC_DFT_PG_10GE_TXRX_MSIX_COALESC_TIMER;
break;
case HINIC_BOARD_PG_SM_25GE:
info->pending_limt =
HINIC_DFT_PG_25GE_TXRX_MSIX_PENDING_LIMIT;
info->coalesce_timer_cfg =
HINIC_DFT_PG_ARM_25GE_TXRX_MSIX_COALESC_TIMER;
break;
case HINIC_BOARD_PG_100GE:
info->pending_limt =
HINIC_DFT_PG_100GE_TXRX_MSIX_PENDING_LIMIT;
info->coalesce_timer_cfg =
HINIC_DFT_PG_ARM_100GE_TXRX_MSIX_COALESC_TIMER;
break;
default:
info->pending_limt = qp_pending_limit;
info->coalesce_timer_cfg = qp_coalesc_timer_cfg;
break;
}
}
info->resend_timer_cfg = HINIC_DEAULT_TXRX_MSIX_RESEND_TIMER_CFG;
info->pkt_rate_high = HINIC_RX_RATE_HIGH;
info->rx_usecs_high = qp_coalesc_timer_high;
info->rx_pending_limt_high = qp_pending_limit_high;
info->pkt_rate_low = HINIC_RX_RATE_LOW;
info->rx_usecs_low = qp_coalesc_timer_low;
info->rx_pending_limt_low = qp_pending_limit_low;
if (nic_dev->in_vm) {
if (qp_pending_limit_high == HINIC_RX_PENDING_LIMIT_HIGH)
qp_pending_limit_high = HINIC_RX_PENDING_LIMIT_HIGH_VM;
info->pkt_rate_low = HINIC_RX_RATE_LOW_VM;
info->rx_pending_limt_high = qp_pending_limit_high;
}
/* suit for sdi3.0 vm mode vf drv or bm mode pf/vf drv */
if ((nic_dev->is_vm_slave && nic_dev->in_vm) ||
nic_dev->is_bm_slave) {
info->pkt_rate_high = SDI_VM_RX_PKT_RATE_HIGH;
info->pkt_rate_low = SDI_VM_RX_PKT_RATE_LOW;
if (qp_coalesc_use_drv_params_switch == 0) {
/* if arm server, maybe need to change this value
* again
*/
info->pending_limt = SDI_VM_PENDING_LIMT;
info->coalesce_timer_cfg = SDI_VM_COALESCE_TIMER_CFG;
info->rx_usecs_high = SDI_VM_RX_USECS_HIGH;
info->rx_pending_limt_high =
SDI_VM_RX_PENDING_LIMT_HIGH;
info->rx_usecs_low = SDI_VM_RX_USECS_LOW;
info->rx_pending_limt_low = SDI_VM_RX_PENDING_LIMT_LOW;
} else {
info->rx_usecs_high = qp_coalesc_timer_high;
info->rx_pending_limt_high = qp_pending_limit_high;
info->rx_usecs_low = qp_coalesc_timer_low;
info->rx_pending_limt_low = qp_pending_limit_low;
}
}
}
static void init_intr_coal_param(struct hinic_nic_dev *nic_dev)
{
struct pci_device_id *id;
u16 i;
id = hinic_get_pci_device_id(nic_dev->pdev);
switch (id->driver_data) {
case HINIC_BOARD_10GE:
case HINIC_BOARD_PG_TP_10GE:
nic_dev->his_link_speed = SPEED_10000;
break;
case HINIC_BOARD_25GE:
case HINIC_BOARD_PG_SM_25GE:
nic_dev->his_link_speed = SPEED_25000;
break;
case HINIC_BOARD_40GE:
nic_dev->his_link_speed = SPEED_40000;
break;
case HINIC_BOARD_100GE:
case HINIC_BOARD_PG_100GE:
nic_dev->his_link_speed = SPEED_100000;
break;
default:
break;
}
for (i = 0; i < nic_dev->max_qps; i++)
update_queue_coal_param(nic_dev, id, i);
}
static int hinic_init_intr_coalesce(struct hinic_nic_dev *nic_dev)
{
u64 size;
if (qp_pending_limit != HINIC_DEAULT_TXRX_MSIX_PENDING_LIMIT ||
qp_coalesc_timer_cfg != HINIC_DEAULT_TXRX_MSIX_COALESC_TIMER_CFG)
nic_dev->intr_coal_set_flag = 1;
else
nic_dev->intr_coal_set_flag = 0;
size = sizeof(*nic_dev->intr_coalesce) * nic_dev->max_qps;
if (!size) {
nic_err(&nic_dev->pdev->dev, "Cannot allocate zero size intr coalesce\n");
return -EINVAL;
}
nic_dev->intr_coalesce = kzalloc(size, GFP_KERNEL);
if (!nic_dev->intr_coalesce)
return -ENOMEM;
init_intr_coal_param(nic_dev);
if (test_bit(HINIC_INTR_ADAPT, &nic_dev->flags))
nic_dev->adaptive_rx_coal = 1;
else
nic_dev->adaptive_rx_coal = 0;
return 0;
}
static void hinic_free_intr_coalesce(struct hinic_nic_dev *nic_dev)
{
kfree(nic_dev->intr_coalesce);
}
static int hinic_alloc_qps(struct hinic_nic_dev *nic_dev)
{
struct net_device *netdev = nic_dev->netdev;
int err;
err = hinic_alloc_txqs(netdev);
if (err) {
nic_err(&nic_dev->pdev->dev, "Failed to alloc txqs\n");
return err;
}
err = hinic_alloc_rxqs(netdev);
if (err) {
nic_err(&nic_dev->pdev->dev, "Failed to alloc rxqs\n");
goto alloc_rxqs_err;
}
err = hinic_init_intr_coalesce(nic_dev);
if (err) {
nic_err(&nic_dev->pdev->dev, "Failed to init_intr_coalesce\n");
goto init_intr_err;
}
return 0;
init_intr_err:
hinic_free_rxqs(netdev);
alloc_rxqs_err:
hinic_free_txqs(netdev);
return err;
}
static void hinic_destroy_qps(struct hinic_nic_dev *nic_dev)
{
hinic_free_intr_coalesce(nic_dev);
hinic_free_rxqs(nic_dev->netdev);
hinic_free_txqs(nic_dev->netdev);
}
static int hinic_validate_parameters(struct hinic_lld_dev *lld_dev)
{
struct pci_dev *pdev = lld_dev->pdev;
/* Check poll_weight value, default poll_weight is 64.
* The poll_weight isn't more than max queue depth,
* so the valid value range is 1~4096.
*/
if (!poll_weight) {
nic_warn(&pdev->dev, "Module Parameter poll_weight can not be 0, resetting to %d\n",
DEFAULT_POLL_WEIGHT);
poll_weight = DEFAULT_POLL_WEIGHT;
}
if (poll_weight > HINIC_MAX_QUEUE_DEPTH) {
nic_warn(&pdev->dev, "Module Parameter poll_weight value %u is out of 1~%d, resetting to max value %d\n",
poll_weight, HINIC_MAX_QUEUE_DEPTH,
HINIC_MAX_QUEUE_DEPTH);
poll_weight = HINIC_MAX_QUEUE_DEPTH;
}
/* check rx_buff value, default rx_buff is 2KB.
* Invalid rx_buff include 2KB/4KB/8KB/16KB.
*/
if (rx_buff != RX_BUFF_VALID_2KB && rx_buff != RX_BUFF_VALID_4KB &&
rx_buff != RX_BUFF_VALID_8KB && rx_buff != RX_BUFF_VALID_16KB) {
nic_warn(&pdev->dev, "Module Parameter rx_buff value %d is out of range, must be 2^n. Valid range is 2 - 16, resetting to %dKB",
rx_buff, DEFAULT_RX_BUFF_LEN);
rx_buff = DEFAULT_RX_BUFF_LEN;
}
if (qp_coalesc_timer_high <= qp_coalesc_timer_low) {
nic_warn(&pdev->dev, "Module Parameter qp_coalesc_timer_high: %d, qp_coalesc_timer_low: %d is invalid, resetting to default\n",
qp_coalesc_timer_high, qp_coalesc_timer_low);
qp_coalesc_timer_high = HINIC_RX_COAL_TIME_HIGH;
qp_coalesc_timer_low = HINIC_RX_COAL_TIME_LOW;
}
if (qp_pending_limit_high <= qp_pending_limit_low) {
nic_warn(&pdev->dev, "Module Parameter qp_pending_limit_high: %d, qp_pending_limit_low: %d is invalid, resetting to default\n",
qp_pending_limit_high, qp_pending_limit_low);
qp_pending_limit_high = HINIC_RX_PENDING_LIMIT_HIGH;
qp_pending_limit_low = HINIC_RX_PENDING_LIMIT_LOW;
}
return 0;
}
static void check_lro_module_param(struct hinic_nic_dev *nic_dev)
{
struct hinic_lro_cfg *lro = &nic_dev->adaptive_cfg.lro;
/* Use module parameters first. */
if (set_lro_timer != 0 &&
set_lro_timer >= HINIC_LRO_RX_TIMER_LOWER &&
set_lro_timer <= HINIC_LRO_RX_TIMER_UPPER)
lro->timer = set_lro_timer;
/* Use module parameters first. */
if (set_max_wqe_num != 0 &&
set_max_wqe_num <= HINIC_LRO_MAX_WQE_NUM_UPPER &&
set_max_wqe_num >= HINIC_LRO_MAX_WQE_NUM_LOWER)
lro->buffer_size = set_max_wqe_num * nic_dev->rx_buff_len;
}
static void decide_rss_cfg(struct hinic_nic_dev *nic_dev)
{
struct hinic_environment_info *info = &nic_dev->env_info;
switch (info->cpu) {
case HINIC_CPU_ARM_GENERIC:
set_bit(HINIC_SAME_RXTX, &nic_dev->flags);
break;
case HINIC_CPU_X86_GENERIC:
clear_bit(HINIC_SAME_RXTX, &nic_dev->flags);
break;
default:
clear_bit(HINIC_SAME_RXTX, &nic_dev->flags);
break;
}
}
static void decide_lro_cfg(struct hinic_nic_dev *nic_dev)
{
struct hinic_environment_info *info = &nic_dev->env_info;
struct hinic_lro_cfg *lro = &nic_dev->adaptive_cfg.lro;
if (lro_en_status < HINIC_LRO_STATUS_UNSET) {
lro->enable = lro_en_status;
} else {
/* LRO will be opened in all Huawei OS */
switch (info->os) {
case HINIC_OS_HUAWEI:
lro->enable = 1;
break;
case HINIC_OS_NON_HUAWEI:
lro->enable = 0;
break;
default:
lro->enable = 0;
break;
}
}
switch (info->board) {
case HINIC_BOARD_25GE:
lro->timer = HINIC_LRO_RX_TIMER_DEFAULT_25GE;
break;
case HINIC_BOARD_100GE:
lro->timer = HINIC_LRO_RX_TIMER_DEFAULT_100GE;
break;
case HINIC_BOARD_PG_TP_10GE:
lro->timer = HINIC_LRO_RX_TIMER_DEFAULT_PG_10GE;
break;
case HINIC_BOARD_PG_SM_25GE:
lro->timer = HINIC_LRO_RX_TIMER_DEFAULT;
break;
case HINIC_BOARD_PG_100GE:
lro->timer = HINIC_LRO_RX_TIMER_DEFAULT_PG_100GE;
break;
default:
lro->timer = HINIC_LRO_RX_TIMER_DEFAULT;
break;
}
/* Use module parameters first. */
switch (info->cpu) {
case HINIC_CPU_ARM_GENERIC:
lro->buffer_size =
HINIC_LRO_MAX_WQE_NUM_DEFAULT_ARM *
nic_dev->rx_buff_len;
break;
case HINIC_CPU_X86_GENERIC:
lro->buffer_size =
HINIC_LRO_MAX_WQE_NUM_DEFAULT_X86 *
nic_dev->rx_buff_len;
break;
default:
lro->buffer_size =
HINIC_LRO_MAX_WQE_NUM_DEFAULT *
nic_dev->rx_buff_len;
break;
}
/* lro buffer_size need modify according board type */
switch (info->board) {
case HINIC_BOARD_PG_TP_10GE:
case HINIC_BOARD_PG_SM_25GE:
case HINIC_BOARD_PG_100GE:
lro->buffer_size =
HINIC_LRO_WQE_NUM_PANGEA_DEFAULT * nic_dev->rx_buff_len;
break;
default:
break;
}
check_lro_module_param(nic_dev);
nic_info(&nic_dev->pdev->dev,
"LRO default configuration: enable %u, timer %u, buffer size %u\n",
lro->enable, lro->timer, lro->buffer_size);
}
static void decide_intr_cfg(struct hinic_nic_dev *nic_dev)
{
struct pci_device_id *id;
id = hinic_get_pci_device_id(nic_dev->pdev);
switch (id->driver_data) {
case HINIC_BOARD_PG_TP_10GE:
case HINIC_BOARD_PG_SM_25GE:
case HINIC_BOARD_PG_100GE:
clear_bit(HINIC_INTR_ADAPT, &nic_dev->flags);
break;
default:
set_bit(HINIC_INTR_ADAPT, &nic_dev->flags);
break;
}
}
static void adaptive_configuration_init(struct hinic_nic_dev *nic_dev)
{
struct pci_device_id *id;
id = hinic_get_pci_device_id(nic_dev->pdev);
if (id)
nic_dev->env_info.board = id->driver_data;
else
nic_dev->env_info.board = HINIC_BOARD_UNKNOWN;
nic_dev->env_info.os = HINIC_OS_HUAWEI;
#if defined(__aarch64__)
nic_dev->env_info.cpu = HINIC_CPU_ARM_GENERIC;
#elif defined(__x86_64__)
nic_dev->env_info.cpu = HINIC_CPU_X86_GENERIC;
#else
nic_dev->env_info.cpu = HINIC_CPU_UNKNOWN;
#endif
nic_info(&nic_dev->pdev->dev,
"Board type %u, OS type %u, CPU type %u\n",
nic_dev->env_info.board, nic_dev->env_info.os,
nic_dev->env_info.cpu);
decide_lro_cfg(nic_dev);
decide_rss_cfg(nic_dev);
decide_intr_cfg(nic_dev);
}
static int nic_probe(struct hinic_lld_dev *lld_dev, void **uld_dev,
char *uld_dev_name)
{
struct pci_dev *pdev = lld_dev->pdev;
struct hinic_nic_dev *nic_dev;
struct net_device *netdev;
u16 max_qps;
u32 page_num;
int err;
/* *uld_dev should always no be NULL */
*uld_dev = lld_dev;
if (!hinic_support_nic(lld_dev->hwdev, NULL)) {
nic_info(&pdev->dev, "Hw don't support nic\n");
return 0;
}
err = hinic_validate_parameters(lld_dev);
if (err)
return -EINVAL;
max_qps = hinic_func_max_nic_qnum(lld_dev->hwdev);
netdev = alloc_etherdev_mq(sizeof(*nic_dev), max_qps);
if (!netdev) {
nic_err(&pdev->dev, "Failed to allocate ETH device\n");
return -ENOMEM;
}
SET_NETDEV_DEV(netdev, &pdev->dev);
nic_dev = (struct hinic_nic_dev *)netdev_priv(netdev);
nic_dev->hwdev = lld_dev->hwdev;
nic_dev->pdev = pdev;
nic_dev->poll_weight = (int)poll_weight;
nic_dev->msg_enable = DEFAULT_MSG_ENABLE;
nic_dev->heart_status = true;
nic_dev->in_vm = !hinic_is_in_host();
nic_dev->is_vm_slave = is_multi_vm_slave(lld_dev->hwdev);
nic_dev->is_bm_slave = is_multi_bm_slave(lld_dev->hwdev);
nic_dev->lro_replenish_thld = lro_replenish_thld;
nic_dev->rx_buff_len = (u16)(rx_buff * CONVERT_UNIT);
page_num = (RX_BUFF_NUM_PER_PAGE * nic_dev->rx_buff_len) / PAGE_SIZE;
nic_dev->page_order = page_num > 0 ? ilog2(page_num) : 0;
mutex_init(&nic_dev->nic_mutex);
adaptive_configuration_init(nic_dev);
nic_dev->vlan_bitmap = kzalloc(VLAN_BITMAP_SIZE(nic_dev), GFP_KERNEL);
if (!nic_dev->vlan_bitmap) {
err = -ENOMEM;
goto vlan_bitmap_err;
}
nic_dev->netdev = netdev;
hinic_assign_netdev_ops(nic_dev);
netdev_feature_init(netdev);
/* get nic cap from hw */
hinic_support_nic(lld_dev->hwdev, &nic_dev->nic_cap);
err = hinic_init_nic_hwdev(nic_dev->hwdev, nic_dev->rx_buff_len);
if (err) {
nic_err(&pdev->dev, "Failed to init nic hwdev\n");
goto init_nic_hwdev_err;
}
err = hinic_set_super_cqe_state(nic_dev->hwdev, true);
if (err) {
nic_err(&pdev->dev, "Failed to set super cqe\n");
goto set_supper_cqe_err;
}
err = hinic_sw_init(nic_dev);
if (err)
goto sw_init_err;
err = hinic_alloc_qps(nic_dev);
if (err) {
nic_err(&pdev->dev, "Failed to alloc qps\n");
goto alloc_qps_err;
}
nic_dev->workq = create_singlethread_workqueue(HINIC_NIC_DEV_WQ_NAME);
if (!nic_dev->workq) {
nic_err(&pdev->dev, "Failed to initialize AEQ workqueue\n");
err = -ENOMEM;
goto create_workq_err;
}
INIT_LIST_HEAD(&nic_dev->uc_filter_list);
INIT_LIST_HEAD(&nic_dev->mc_filter_list);
INIT_WORK(&nic_dev->rx_mode_work, hinic_set_rx_mode_work);
err = hinic_set_default_hw_feature(nic_dev);
if (err)
goto set_features_err;
hinic_register_notifier(nic_dev);
err = register_netdev(netdev);
if (err) {
nic_err(&pdev->dev, "Failed to register netdev\n");
err = -ENOMEM;
goto netdev_err;
}
netif_carrier_off(netdev);
*uld_dev = nic_dev;
nicif_info(nic_dev, probe, netdev, "Register netdev succeed\n");
return 0;
netdev_err:
hinic_unregister_notifier(nic_dev);
set_features_err:
destroy_workqueue(nic_dev->workq);
create_workq_err:
hinic_destroy_qps(nic_dev);
alloc_qps_err:
hinic_del_mac(nic_dev->hwdev, netdev->dev_addr, 0,
hinic_global_func_id_hw(nic_dev->hwdev));
sw_init_err:
(void)hinic_set_super_cqe_state(nic_dev->hwdev, false);
set_supper_cqe_err:
hinic_free_nic_hwdev(nic_dev->hwdev);
init_nic_hwdev_err:
kfree(nic_dev->vlan_bitmap);
vlan_bitmap_err:
free_netdev(netdev);
return err;
}
static void nic_remove(struct hinic_lld_dev *lld_dev, void *adapter)
{
struct hinic_nic_dev *nic_dev = adapter;
struct net_device *netdev;
if (!nic_dev || !hinic_support_nic(lld_dev->hwdev, NULL))
return;
netdev = nic_dev->netdev;
unregister_netdev(netdev);
hinic_unregister_notifier(nic_dev);
cancel_work_sync(&nic_dev->rx_mode_work);
destroy_workqueue(nic_dev->workq);
hinic_destroy_qps(nic_dev);
hinic_clean_mac_list_filter(nic_dev);
hinic_del_mac(nic_dev->hwdev, netdev->dev_addr, 0,
hinic_global_func_id_hw(nic_dev->hwdev));
if (test_bit(HINIC_RSS_ENABLE, &nic_dev->flags))
hinic_rss_template_free(nic_dev->hwdev, nic_dev->rss_tmpl_idx);
(void)hinic_set_super_cqe_state(nic_dev->hwdev, false);
hinic_free_nic_hwdev(nic_dev->hwdev);
kfree(nic_dev->vlan_bitmap);
free_netdev(netdev);
}
int hinic_disable_func_rss(struct hinic_nic_dev *nic_dev)
{
struct net_device *netdev = nic_dev->netdev;
int err, err_netdev = 0;
nicif_info(nic_dev, drv, netdev, "Start to disable RSS\n");
if (!test_bit(HINIC_RSS_ENABLE, &nic_dev->flags)) {
nicif_info(nic_dev, drv, netdev, "RSS not enabled, do nothing\n");
return 0;
}
if (netif_running(netdev)) {
err_netdev = hinic_close(netdev);
if (err_netdev) {
nicif_err(nic_dev, drv, netdev,
"Failed to close netdev\n");
return -EFAULT;
}
}
/* free rss template */
err = hinic_rss_template_free(nic_dev->hwdev, nic_dev->rss_tmpl_idx);
if (err) {
nicif_err(nic_dev, drv, netdev, "Failed to free RSS template\n");
} else {
nicif_info(nic_dev, drv, netdev, "Success to free RSS template\n");
clear_bit(HINIC_RSS_ENABLE, &nic_dev->flags);
}
if (netif_running(netdev)) {
err_netdev = hinic_open(netdev);
if (err_netdev)
nicif_err(nic_dev, drv, netdev,
"Failed to open netdev\n");
}
return err ? err : err_netdev;
}
int hinic_enable_func_rss(struct hinic_nic_dev *nic_dev)
{
struct net_device *netdev = nic_dev->netdev;
int err, err_netdev = 0;
nicif_info(nic_dev, drv, netdev, "Start to enable RSS\n");
if (test_bit(HINIC_RSS_ENABLE, &nic_dev->flags)) {
nicif_info(nic_dev, drv, netdev, "RSS already enabled, do nothing\n");
return 0;
}
if (netif_running(netdev)) {
err_netdev = hinic_close(netdev);
if (err_netdev) {
nicif_err(nic_dev, drv, netdev,
"Failed to close netdev\n");
return -EFAULT;
}
}
err = hinic_rss_template_alloc(nic_dev->hwdev, &nic_dev->rss_tmpl_idx);
if (err) {
if (err == -ENOSPC)
nicif_warn(nic_dev, drv, netdev,
"Failed to alloc RSS template, table is full\n");
else
nicif_err(nic_dev, drv, netdev,
"Failed to alloc RSS template\n");
} else {
set_bit(HINIC_RSS_ENABLE, &nic_dev->flags);
nicif_info(nic_dev, drv, netdev, "Success to alloc RSS template\n");
}
if (netif_running(netdev)) {
err_netdev = hinic_open(netdev);
if (err_netdev)
nicif_err(nic_dev, drv, netdev,
"Failed to open netdev\n");
}
return err ? err : err_netdev;
}
static const char *hinic_module_link_err[LINK_ERR_NUM] = {
"Unrecognized module",
};
static void hinic_port_module_event_handler(struct hinic_nic_dev *nic_dev,
struct hinic_event_info *event)
{
enum port_module_event_type type = event->module_event.type;
enum link_err_type err_type = event->module_event.err_type;
switch (type) {
case HINIC_PORT_MODULE_CABLE_PLUGGED:
case HINIC_PORT_MODULE_CABLE_UNPLUGGED:
nicif_info(nic_dev, link, nic_dev->netdev,
"Port module event: Cable %s\n",
type == HINIC_PORT_MODULE_CABLE_PLUGGED ?
"plugged" : "unplugged");
break;
case HINIC_PORT_MODULE_LINK_ERR:
if (err_type >= LINK_ERR_NUM) {
nicif_info(nic_dev, link, nic_dev->netdev,
"Link failed, Unknown error type: 0x%x\n",
err_type);
} else {
nicif_info(nic_dev, link, nic_dev->netdev,
"Link failed, error type: 0x%x: %s\n",
err_type, hinic_module_link_err[err_type]);
}
break;
default:
nicif_err(nic_dev, link, nic_dev->netdev,
"Unknown port module type %d\n", type);
break;
}
}
static void hinic_intr_coalesc_change(struct hinic_nic_dev *nic_dev,
struct hinic_event_info *event)
{
u32 hw_to_os_speed[LINK_SPEED_LEVELS] = {SPEED_10, SPEED_100,
SPEED_1000, SPEED_10000,
SPEED_25000, SPEED_40000,
SPEED_100000};
u8 qid, coalesc_timer_cfg, pending_limt;
struct pci_device_id *id;
u32 speed;
int err;
if (nic_dev->adaptive_rx_coal)
return;
speed = hw_to_os_speed[event->link_info.speed];
if (speed == nic_dev->his_link_speed)
return;
id = hinic_get_pci_device_id(nic_dev->pdev);
switch (id->driver_data) {
case HINIC_BOARD_PG_TP_10GE:
return;
case HINIC_BOARD_PG_SM_25GE:
if (speed == SPEED_10000) {
pending_limt =
HINIC_DFT_PG_10GE_TXRX_MSIX_PENDING_LIMIT;
coalesc_timer_cfg =
HINIC_DFT_PG_10GE_TXRX_MSIX_COALESC_TIMER;
} else if (speed == SPEED_25000) {
pending_limt =
HINIC_DFT_PG_25GE_TXRX_MSIX_PENDING_LIMIT;
coalesc_timer_cfg =
HINIC_DFT_PG_ARM_25GE_TXRX_MSIX_COALESC_TIMER;
} else {
pending_limt =
HINIC_DFT_PG_25GE_TXRX_MSIX_PENDING_LIMIT;
coalesc_timer_cfg =
HINIC_DFT_PG_25GE_TXRX_MSIX_COALESC_TIMER;
}
break;
case HINIC_BOARD_PG_100GE:
return;
default:
return;
}
for (qid = 0; qid < nic_dev->num_qps; qid++) {
if (!nic_dev->intr_coalesce[qid].user_set_intr_coal_flag) {
err = set_interrupt_moder(nic_dev, qid,
coalesc_timer_cfg,
pending_limt);
if (!err) {
nic_dev->intr_coalesce[qid].pending_limt =
pending_limt;
nic_dev->intr_coalesce[qid].coalesce_timer_cfg =
coalesc_timer_cfg;
}
}
}
nic_dev->his_link_speed = speed;
}
void nic_event(struct hinic_lld_dev *lld_dev, void *adapter,
struct hinic_event_info *event)
{
struct hinic_nic_dev *nic_dev = adapter;
struct net_device *netdev;
enum hinic_event_type type;
if (!nic_dev || !event || !hinic_support_nic(lld_dev->hwdev, NULL))
return;
netdev = nic_dev->netdev;
type = event->type;
switch (type) {
case HINIC_EVENT_LINK_DOWN:
hinic_link_status_change(nic_dev, false);
break;
case HINIC_EVENT_LINK_UP:
hinic_link_status_change(nic_dev, true);
hinic_intr_coalesc_change(nic_dev, event);
break;
case HINIC_EVENT_HEART_LOST:
hinic_heart_lost(nic_dev);
hinic_link_status_change(nic_dev, false);
break;
case HINIC_EVENT_FAULT:
if (event->info.fault_level == FAULT_LEVEL_SERIOUS_FLR &&
event->info.event.chip.func_id ==
hinic_global_func_id(lld_dev->hwdev))
hinic_link_status_change(nic_dev, false);
break;
case HINIC_EVENT_DCB_STATE_CHANGE:
if (nic_dev->default_cos_id == event->dcb_state.default_cos)
break;
/* PF notify to vf, don't need to handle this event */
if (!HINIC_FUNC_IS_VF(nic_dev->hwdev))
break;
nicif_info(nic_dev, drv, netdev, "Change default cos %d to %d\n",
nic_dev->default_cos_id,
event->dcb_state.default_cos);
nic_dev->default_cos_id = event->dcb_state.default_cos;
hinic_set_sq_default_cos(netdev, nic_dev->default_cos_id);
break;
case HINIC_EVENT_PORT_MODULE_EVENT:
hinic_port_module_event_handler(nic_dev, event);
break;
case HINIC_EVENT_MGMT_WATCHDOG_EVENT:
hinic_link_status_change(nic_dev, false);
break;
default:
break;
}
}
struct hinic_uld_info nic_uld_info = {
.probe = nic_probe,
.remove = nic_remove,
.suspend = NULL,
.resume = NULL,
.event = nic_event,
.ioctl = nic_ioctl,
}; /*lint -e766*/