#include <linux/time.h>
#include <linux/ethtool.h>
#include <linux/mii.h>
-#ifdef NETIF_F_HW_VLAN_TX
- #include <linux/if_vlan.h>
-#endif
+#include <linux/if_vlan.h>
#include <net/ip.h>
#include <net/tcp.h>
#include <net/checksum.h>
module_param(use_multi, int, 0);
MODULE_PARM_DESC(use_multi, "use per-CPU queues");
#endif
+static struct workqueue_struct *bnx2x_wq;
enum bnx2x_board_type {
BCM57710 = 0,
synchronize_irq(bp->pdev->irq);
/* make sure sp_task is not running */
- cancel_work_sync(&bp->sp_task);
+ cancel_delayed_work(&bp->sp_task);
+ flush_workqueue(bnx2x_wq);
}
/* fast path */
return;
pci_unmap_page(bp->pdev, pci_unmap_addr(sw_buf, mapping),
- BCM_PAGE_SIZE*PAGES_PER_SGE, PCI_DMA_FROMDEVICE);
+ SGE_PAGE_SIZE*PAGES_PER_SGE, PCI_DMA_FROMDEVICE);
__free_pages(page, PAGES_PER_SGE_SHIFT);
sw_buf->page = NULL;
if (unlikely(page == NULL))
return -ENOMEM;
- mapping = pci_map_page(bp->pdev, page, 0, BCM_PAGE_SIZE*PAGES_PER_SGE,
+ mapping = pci_map_page(bp->pdev, page, 0, SGE_PAGE_SIZE*PAGES_PER_SGE,
PCI_DMA_FROMDEVICE);
if (unlikely(dma_mapping_error(&bp->pdev->dev, mapping))) {
__free_pages(page, PAGES_PER_SGE_SHIFT);
struct eth_fast_path_rx_cqe *fp_cqe)
{
struct bnx2x *bp = fp->bp;
- u16 sge_len = BCM_PAGE_ALIGN(le16_to_cpu(fp_cqe->pkt_len) -
+ u16 sge_len = SGE_PAGE_ALIGN(le16_to_cpu(fp_cqe->pkt_len) -
le16_to_cpu(fp_cqe->len_on_bd)) >>
- BCM_PAGE_SHIFT;
+ SGE_PAGE_SHIFT;
u16 last_max, last_elem, first_elem;
u16 delta = 0;
u16 i;
u16 cqe_idx)
{
struct sw_rx_page *rx_pg, old_rx_pg;
- struct page *sge;
u16 len_on_bd = le16_to_cpu(fp_cqe->len_on_bd);
u32 i, frag_len, frag_size, pages;
int err;
int j;
frag_size = le16_to_cpu(fp_cqe->pkt_len) - len_on_bd;
- pages = BCM_PAGE_ALIGN(frag_size) >> BCM_PAGE_SHIFT;
+ pages = SGE_PAGE_ALIGN(frag_size) >> SGE_PAGE_SHIFT;
/* This is needed in order to enable forwarding support */
if (frag_size)
- skb_shinfo(skb)->gso_size = min((u32)BCM_PAGE_SIZE,
+ skb_shinfo(skb)->gso_size = min((u32)SGE_PAGE_SIZE,
max(frag_size, (u32)len_on_bd));
#ifdef BNX2X_STOP_ON_ERROR
- if (pages > 8*PAGES_PER_SGE) {
+ if (pages >
+ min((u32)8, (u32)MAX_SKB_FRAGS) * SGE_PAGE_SIZE * PAGES_PER_SGE) {
BNX2X_ERR("SGL length is too long: %d. CQE index is %d\n",
pages, cqe_idx);
BNX2X_ERR("fp_cqe->pkt_len = %d fp_cqe->len_on_bd = %d\n",
/* FW gives the indices of the SGE as if the ring is an array
(meaning that "next" element will consume 2 indices) */
- frag_len = min(frag_size, (u32)(BCM_PAGE_SIZE*PAGES_PER_SGE));
+ frag_len = min(frag_size, (u32)(SGE_PAGE_SIZE*PAGES_PER_SGE));
rx_pg = &fp->rx_page_ring[sge_idx];
- sge = rx_pg->page;
old_rx_pg = *rx_pg;
/* If we fail to allocate a substitute page, we simply stop
/* Unmap the page as we r going to pass it to the stack */
pci_unmap_page(bp->pdev, pci_unmap_addr(&old_rx_pg, mapping),
- BCM_PAGE_SIZE*PAGES_PER_SGE, PCI_DMA_FROMDEVICE);
+ SGE_PAGE_SIZE*PAGES_PER_SGE, PCI_DMA_FROMDEVICE);
/* Add one frag and update the appropriate fields in the skb */
skb_fill_page_desc(skb, j, old_rx_pg.page, 0, frag_len);
if (likely(new_skb)) {
/* fix ip xsum and give it to the stack */
/* (no need to map the new skb) */
+#ifdef BCM_VLAN
+ int is_vlan_cqe =
+ (le16_to_cpu(cqe->fast_path_cqe.pars_flags.flags) &
+ PARSING_FLAGS_VLAN);
+ int is_not_hwaccel_vlan_cqe =
+ (is_vlan_cqe && (!(bp->flags & HW_VLAN_RX_FLAG)));
+#endif
prefetch(skb);
prefetch(((char *)(skb)) + 128);
struct iphdr *iph;
iph = (struct iphdr *)skb->data;
+#ifdef BCM_VLAN
+ /* If there is no Rx VLAN offloading -
+ take VLAN tag into an account */
+ if (unlikely(is_not_hwaccel_vlan_cqe))
+ iph = (struct iphdr *)((u8 *)iph + VLAN_HLEN);
+#endif
iph->check = 0;
iph->check = ip_fast_csum((u8 *)iph, iph->ihl);
}
if (!bnx2x_fill_frag_skb(bp, fp, skb,
&cqe->fast_path_cqe, cqe_idx)) {
#ifdef BCM_VLAN
- if ((bp->vlgrp != NULL) &&
- (le16_to_cpu(cqe->fast_path_cqe.pars_flags.flags) &
- PARSING_FLAGS_VLAN))
+ if ((bp->vlgrp != NULL) && is_vlan_cqe &&
+ (!is_not_hwaccel_vlan_cqe))
vlan_hwaccel_receive_skb(skb, bp->vlgrp,
le16_to_cpu(cqe->fast_path_cqe.
vlan_tag));
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(struct tstorm_eth_rx_producers)/4; i++)
REG_WR(bp, BAR_TSTRORM_INTMEM +
TSTORM_RX_PRODS_OFFSET(BP_PORT(bp), FP_CL_ID(fp)) + i*4,
((u32 *)&rx_prods)[i]);
+ mmiowb(); /* keep prod updates ordered */
+
DP(NETIF_MSG_RX_STATUS,
"Wrote: bd_prod %u cqe_prod %u sge_prod %u\n",
bd_prod, rx_comp_prod, rx_sge_prod);
DP(NETIF_MSG_RX_STATUS, "CQE type %x err %x status %x"
" queue %x vlan %x len %u\n", CQE_TYPE(cqe_fp_flags),
cqe_fp_flags, cqe->fast_path_cqe.status_flags,
- cqe->fast_path_cqe.rss_hash_result,
+ le32_to_cpu(cqe->fast_path_cqe.rss_hash_result),
le16_to_cpu(cqe->fast_path_cqe.vlan_tag),
le16_to_cpu(cqe->fast_path_cqe.pkt_len));
}
#ifdef BCM_VLAN
- if ((bp->vlgrp != NULL) &&
+ if ((bp->vlgrp != NULL) && (bp->flags & HW_VLAN_RX_FLAG) &&
(le16_to_cpu(cqe->fast_path_cqe.pars_flags.flags) &
PARSING_FLAGS_VLAN))
vlan_hwaccel_receive_skb(skb, bp->vlgrp,
/* Update producers */
bnx2x_update_rx_prod(bp, fp, bd_prod_fw, sw_comp_prod,
fp->rx_sge_prod);
- mmiowb(); /* keep prod updates ordered */
fp->rx_pkt += rx_pkt;
fp->rx_calls++;
if (unlikely(status & 0x1)) {
- schedule_work(&bp->sp_task);
+ queue_delayed_work(bnx2x_wq, &bp->sp_task, 0);
status &= ~0x1;
if (!status)
static void bnx2x_calc_fc_adv(struct bnx2x *bp)
{
- switch (bp->link_vars.ieee_fc) {
+ switch (bp->link_vars.ieee_fc &
+ MDIO_COMBO_IEEE0_AUTO_NEG_ADV_PAUSE_MASK) {
case MDIO_COMBO_IEEE0_AUTO_NEG_ADV_PAUSE_NONE:
bp->port.advertising &= ~(ADVERTISED_Asym_Pause |
ADVERTISED_Pause);
rc = bnx2x_phy_init(&bp->link_params, &bp->link_vars);
bnx2x_release_phy_lock(bp);
+ bnx2x_calc_fc_adv(bp);
+
if (bp->link_vars.link_up)
bnx2x_link_report(bp);
- bnx2x_calc_fc_adv(bp);
return rc;
}
/* Make sure that we are synced with the current statistics */
bnx2x_stats_handle(bp, STATS_EVENT_STOP);
- bnx2x_acquire_phy_lock(bp);
bnx2x_link_update(&bp->link_params, &bp->link_vars);
- bnx2x_release_phy_lock(bp);
if (bp->link_vars.link_up) {
if (asserted & ATTN_HARD_WIRED_MASK) {
if (asserted & ATTN_NIG_FOR_FUNC) {
+ bnx2x_acquire_phy_lock(bp);
+
/* save nig interrupt mask */
bp->nig_mask = REG_RD(bp, nig_int_mask_addr);
REG_WR(bp, nig_int_mask_addr, 0);
REG_WR(bp, hc_addr, asserted);
/* now set back the mask */
- if (asserted & ATTN_NIG_FOR_FUNC)
+ if (asserted & ATTN_NIG_FOR_FUNC) {
REG_WR(bp, nig_int_mask_addr, bp->nig_mask);
+ bnx2x_release_phy_lock(bp);
+ }
}
static inline void bnx2x_attn_int_deasserted0(struct bnx2x *bp, u32 attn)
static void bnx2x_attn_int(struct bnx2x *bp)
{
/* read local copy of bits */
- u32 attn_bits = bp->def_status_blk->atten_status_block.attn_bits;
- u32 attn_ack = bp->def_status_blk->atten_status_block.attn_bits_ack;
+ u32 attn_bits = le32_to_cpu(bp->def_status_blk->atten_status_block.
+ attn_bits);
+ u32 attn_ack = le32_to_cpu(bp->def_status_blk->atten_status_block.
+ attn_bits_ack);
u32 attn_state = bp->attn_state;
/* look for changed bits */
static void bnx2x_sp_task(struct work_struct *work)
{
- struct bnx2x *bp = container_of(work, struct bnx2x, sp_task);
+ struct bnx2x *bp = container_of(work, struct bnx2x, sp_task.work);
u16 status;
if (status & 0x2)
bp->stats_pending = 0;
- bnx2x_ack_sb(bp, DEF_SB_ID, ATTENTION_ID, bp->def_att_idx,
+ bnx2x_ack_sb(bp, DEF_SB_ID, ATTENTION_ID, le16_to_cpu(bp->def_att_idx),
IGU_INT_NOP, 1);
bnx2x_ack_sb(bp, DEF_SB_ID, USTORM_ID, le16_to_cpu(bp->def_u_idx),
IGU_INT_NOP, 1);
return IRQ_HANDLED;
#endif
- schedule_work(&bp->sp_task);
+ queue_delayed_work(bnx2x_wq, &bp->sp_task, 0);
return IRQ_HANDLED;
}
#define ADD_64(s_hi, a_hi, s_lo, a_lo) \
do { \
s_lo += a_lo; \
- s_hi += a_hi + (s_lo < a_lo) ? 1 : 0; \
+ s_hi += a_hi + ((s_lo < a_lo) ? 1 : 0); \
} while (0)
/* difference = minuend - subtrahend */
static void bnx2x_init_ind_table(struct bnx2x *bp)
{
- int port = BP_PORT(bp);
+ int func = BP_FUNC(bp);
int i;
if (!is_multi(bp))
DP(NETIF_MSG_IFUP, "Initializing indirection table\n");
for (i = 0; i < TSTORM_INDIRECTION_TABLE_SIZE; i++)
REG_WR8(bp, BAR_TSTRORM_INTMEM +
- TSTORM_INDIRECTION_TABLE_OFFSET(port) + i,
- i % bp->num_queues);
-
- REG_WR(bp, PRS_REG_A_PRSU_20, 0xf);
+ TSTORM_INDIRECTION_TABLE_OFFSET(func) + i,
+ BP_CL_ID(bp) + (i % bp->num_queues));
}
static void bnx2x_set_client_config(struct bnx2x *bp)
int port = BP_PORT(bp);
int i;
- tstorm_client.mtu = bp->dev->mtu + ETH_OVREHEAD;
+ tstorm_client.mtu = bp->dev->mtu;
tstorm_client.statistics_counter_id = BP_CL_ID(bp);
tstorm_client.config_flags =
TSTORM_ETH_CLIENT_CONFIG_STATSITICS_ENABLE;
#ifdef BCM_VLAN
- if (bp->rx_mode && bp->vlgrp) {
+ if (bp->rx_mode && bp->vlgrp && (bp->flags & HW_VLAN_RX_FLAG)) {
tstorm_client.config_flags |=
TSTORM_ETH_CLIENT_CONFIG_VLAN_REMOVAL_ENABLE;
DP(NETIF_MSG_IFUP, "vlan removal enabled\n");
if (bp->flags & TPA_ENABLE_FLAG) {
tstorm_client.max_sges_for_packet =
- BCM_PAGE_ALIGN(tstorm_client.mtu) >> BCM_PAGE_SHIFT;
+ SGE_PAGE_ALIGN(tstorm_client.mtu) >> SGE_PAGE_SHIFT;
tstorm_client.max_sges_for_packet =
((tstorm_client.max_sges_for_packet +
PAGES_PER_SGE - 1) & (~(PAGES_PER_SGE - 1))) >>
bp->e1hov);
}
- /* Init CQ ring mapping and aggregation size */
- max_agg_size = min((u32)(bp->rx_buf_size +
- 8*BCM_PAGE_SIZE*PAGES_PER_SGE),
- (u32)0xffff);
+ /* Init CQ ring mapping and aggregation size, the FW limit is 8 frags */
+ max_agg_size =
+ min((u32)(min((u32)8, (u32)MAX_SKB_FRAGS) *
+ SGE_PAGE_SIZE * PAGES_PER_SGE),
+ (u32)0xffff);
for_each_queue(bp, i) {
struct bnx2x_fastpath *fp = &bp->fp[i];
bnx2x_init_context(bp);
bnx2x_init_internal(bp, load_code);
bnx2x_init_ind_table(bp);
+ bnx2x_stats_init(bp);
+
+ /* At this point, we are ready for interrupts */
+ atomic_set(&bp->intr_sem, 0);
+
+ /* flush all before enabling interrupts */
+ mb();
+ mmiowb();
+
bnx2x_int_enable(bp);
}
REG_WR(bp, PXP2_REG_RQ_SRC_ENDIAN_M, 1);
REG_WR(bp, PXP2_REG_RQ_CDU_ENDIAN_M, 1);
REG_WR(bp, PXP2_REG_RQ_DBG_ENDIAN_M, 1);
- REG_WR(bp, PXP2_REG_RQ_HC_ENDIAN_M, 1);
/* REG_WR(bp, PXP2_REG_RD_PBF_SWAP_MODE, 1); */
REG_WR(bp, PXP2_REG_RD_QM_SWAP_MODE, 1);
}
bnx2x_init_block(bp, PRS_COMMON_START, PRS_COMMON_END);
+ REG_WR(bp, PRS_REG_A_PRSU_20, 0xf);
/* set NIC mode */
REG_WR(bp, PRS_REG_NIC_MODE, 1);
if (CHIP_IS_E1H(bp))
}
}
- bnx2x_stats_init(bp);
-
bp->state = BNX2X_STATE_OPENING_WAIT4_PORT;
- /* Enable Rx interrupt handling before sending the ramrod
- as it's completed on Rx FP queue */
- bnx2x_napi_enable(bp);
-
- /* Enable interrupt handling */
- atomic_set(&bp->intr_sem, 0);
-
rc = bnx2x_setup_leading(bp);
if (rc) {
BNX2X_ERR("Setup leading failed!\n");
mutex_init(&bp->port.phy_mutex);
- INIT_WORK(&bp->sp_task, bnx2x_sp_task);
+ INIT_DELAYED_WORK(&bp->sp_task, bnx2x_sp_task);
INIT_WORK(&bp->reset_task, bnx2x_reset_task);
rc = bnx2x_get_hwinfo(bp);
tx_bd->general_data = ((UNICAST_ADDRESS <<
ETH_TX_BD_ETH_ADDR_TYPE_SHIFT) | 1);
+ wmb();
+
fp->hw_tx_prods->bds_prod =
cpu_to_le16(le16_to_cpu(fp->hw_tx_prods->bds_prod) + 1);
mb(); /* FW restriction: must not reorder writing nbd and packets */
/* Update producers */
bnx2x_update_rx_prod(bp, fp, fp->rx_bd_prod, fp->rx_comp_prod,
fp->rx_sge_prod);
- mmiowb(); /* keep prod updates ordered */
test_loopback_exit:
bp->link_params.loopback_mode = LOOPBACK_NONE;
"sending pkt %u @%p next_idx %u bd %u @%p\n",
pkt_prod, tx_buf, fp->tx_pkt_prod, bd_prod, tx_bd);
- if ((bp->vlgrp != NULL) && vlan_tx_tag_present(skb)) {
+#ifdef BCM_VLAN
+ if ((bp->vlgrp != NULL) && vlan_tx_tag_present(skb) &&
+ (bp->flags & HW_VLAN_TX_FLAG)) {
tx_bd->vlan = cpu_to_le16(vlan_tx_tag_get(skb));
tx_bd->bd_flags.as_bitfield |= ETH_TX_BD_FLAGS_VLAN_TAG;
vlan_off += 4;
} else
+#endif
tx_bd->vlan = cpu_to_le16(pkt_prod);
if (xmit_type) {
DP(NETIF_MSG_TX_QUEUED, "doorbell: nbd %d bd %u\n", nbd, bd_prod);
+ /*
+ * Make sure that the BD data is updated before updating the producer
+ * since FW might read the BD 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 packets must have BDs.
+ */
+ wmb();
+
fp->hw_tx_prods->bds_prod =
cpu_to_le16(le16_to_cpu(fp->hw_tx_prods->bds_prod) + nbd);
mb(); /* FW restriction: must not reorder writing nbd and packets */
dev->trans_start = jiffies;
if (unlikely(bnx2x_tx_avail(fp) < MAX_SKB_FRAGS + 3)) {
+ /* We want bnx2x_tx_int to "see" the updated tx_bd_prod
+ if we put Tx into XOFF state. */
+ smp_mb();
netif_stop_queue(dev);
bp->eth_stats.driver_xoff++;
if (bnx2x_tx_avail(fp) >= MAX_SKB_FRAGS + 3)
struct bnx2x *bp = netdev_priv(dev);
bp->vlgrp = vlgrp;
+
+ /* Set flags according to the required capabilities */
+ bp->flags &= ~(HW_VLAN_RX_FLAG | HW_VLAN_TX_FLAG);
+
+ if (dev->features & NETIF_F_HW_VLAN_TX)
+ bp->flags |= HW_VLAN_TX_FLAG;
+
+ if (dev->features & NETIF_F_HW_VLAN_RX)
+ bp->flags |= HW_VLAN_RX_FLAG;
+
if (netif_running(dev))
bnx2x_set_client_config(bp);
}
dev->features |= NETIF_F_HIGHDMA;
#ifdef BCM_VLAN
dev->features |= (NETIF_F_HW_VLAN_TX | NETIF_F_HW_VLAN_RX);
+ bp->flags |= (HW_VLAN_RX_FLAG | HW_VLAN_TX_FLAG);
#endif
dev->features |= (NETIF_F_TSO | NETIF_F_TSO_ECN);
dev->features |= NETIF_F_TSO6;
static int __init bnx2x_init(void)
{
+ bnx2x_wq = create_singlethread_workqueue("bnx2x");
+ if (bnx2x_wq == NULL) {
+ printk(KERN_ERR PFX "Cannot create workqueue\n");
+ return -ENOMEM;
+ }
+
return pci_register_driver(&bnx2x_pci_driver);
}
static void __exit bnx2x_cleanup(void)
{
pci_unregister_driver(&bnx2x_pci_driver);
+
+ destroy_workqueue(bnx2x_wq);
}
module_init(bnx2x_init);
projects / linux-2.6-omap-h63xx.git / blobdiff