| lists.openwall.net | lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC | |
|
Open Source and information security mailing list archives
| ||
|
Date: Fri, 18 Jul 2008 14:20:48 -0700
From: "Benjamin Li" <benli@...adcom.com>
To: netdev@...r.kernel.org
cc: benli@...adcom.com, "Michael Chan" <mchan@...adcom.com>
Subject: [PATCH 2/4] bnx2: Add TX multiqueue support.
Signed-off-by: Benjamin Li <benli@...adcom.com>
Signed-off-by: Michael Chan <mchan@...adcom.com>
---
drivers/net/bnx2.c | 53 ++++++++++++++++++++++++++++++++++-----------------
drivers/net/bnx2.h | 3 ++
2 files changed, 38 insertions(+), 18 deletions(-)
diff --git a/drivers/net/bnx2.c b/drivers/net/bnx2.c
index 9ce30fc..dc6f261 100644
--- a/drivers/net/bnx2.c
+++ b/drivers/net/bnx2.c
@@ -47,6 +47,7 @@
#include <linux/prefetch.h>
#include <linux/cache.h>
#include <linux/zlib.h>
+#include <linux/log2.h>
#include "bnx2.h"
#include "bnx2_fw.h"
@@ -492,7 +493,7 @@ bnx2_netif_start(struct bnx2 *bp)
{
if (atomic_dec_and_test(&bp->intr_sem)) {
if (netif_running(bp->dev)) {
- netif_wake_queue(bp->dev);
+ netif_tx_wake_all_queues(bp->dev);
bnx2_napi_enable(bp);
bnx2_enable_int(bp);
}
@@ -2584,7 +2585,11 @@ bnx2_tx_int(struct bnx2 *bp, struct bnx2_napi *bnapi, int budget)
{
struct bnx2_tx_ring_info *txr = &bnapi->tx_ring;
u16 hw_cons, sw_cons, sw_ring_cons;
- int tx_pkt = 0;
+ int tx_pkt = 0, index;
+ struct netdev_queue *txq;
+
+ index = (bnapi - bp->bnx2_napi);
+ txq = netdev_get_tx_queue(bp->dev, index);
hw_cons = bnx2_get_hw_tx_cons(bnapi);
sw_cons = txr->tx_cons;
@@ -2644,21 +2649,23 @@ bnx2_tx_int(struct bnx2 *bp, struct bnx2_napi *bnapi, int budget)
txr->hw_tx_cons = hw_cons;
txr->tx_cons = sw_cons;
+
/* Need to make the tx_cons update visible to bnx2_start_xmit()
- * before checking for netif_queue_stopped(). Without the
+ * before checking for netif_tx_queue_stopped(). Without the
* memory barrier, there is a small possibility that bnx2_start_xmit()
* will miss it and cause the queue to be stopped forever.
*/
smp_mb();
- if (unlikely(netif_queue_stopped(bp->dev)) &&
+ if (unlikely(netif_tx_queue_stopped(txq)) &&
(bnx2_tx_avail(bp, txr) > bp->tx_wake_thresh)) {
- netif_tx_lock(bp->dev);
- if ((netif_queue_stopped(bp->dev)) &&
+ __netif_tx_lock(txq, smp_processor_id());
+ if ((netif_tx_queue_stopped(txq)) &&
(bnx2_tx_avail(bp, txr) > bp->tx_wake_thresh))
- netif_wake_queue(bp->dev);
- netif_tx_unlock(bp->dev);
+ netif_tx_wake_queue(txq);
+ __netif_tx_unlock(txq);
}
+
return tx_pkt;
}
@@ -5766,7 +5773,7 @@ static void
bnx2_setup_int_mode(struct bnx2 *bp, int dis_msi)
{
int cpus = num_online_cpus();
- int msix_vecs = min(cpus + 1, RX_MAX_RSS_RINGS);
+ int msix_vecs = min(cpus + 1, RX_MAX_RINGS);
bp->irq_tbl[0].handler = bnx2_interrupt;
strcpy(bp->irq_tbl[0].name, bp->dev->name);
@@ -5789,7 +5796,10 @@ bnx2_setup_int_mode(struct bnx2 *bp, int dis_msi)
bp->irq_tbl[0].vector = bp->pdev->irq;
}
}
- bp->num_tx_rings = 1;
+
+ bp->num_tx_rings = rounddown_pow_of_two(bp->irq_nvecs);
+ bp->dev->real_num_tx_queues = bp->num_tx_rings;
+
bp->num_rx_rings = bp->irq_nvecs;
}
@@ -5858,7 +5868,7 @@ bnx2_open(struct net_device *dev)
else if (bp->flags & BNX2_FLAG_USING_MSIX)
printk(KERN_INFO PFX "%s: using MSIX\n", dev->name);
- netif_start_queue(dev);
+ netif_tx_start_all_queues(dev);
return 0;
@@ -5927,12 +5937,19 @@ bnx2_start_xmit(struct sk_buff *skb, struct net_device *dev)
u32 len, vlan_tag_flags, last_frag, mss;
u16 prod, ring_prod;
int i;
- struct bnx2_napi *bnapi = &bp->bnx2_napi[0];
- struct bnx2_tx_ring_info *txr = &bnapi->tx_ring;
+ struct bnx2_napi *bnapi;
+ struct bnx2_tx_ring_info *txr;
+ struct netdev_queue *txq;
+
+ /* Determine which tx ring we will be placed on */
+ i = skb_get_queue_mapping(skb);
+ bnapi = &bp->bnx2_napi[i];
+ txr = &bnapi->tx_ring;
+ txq = netdev_get_tx_queue(dev, i);
if (unlikely(bnx2_tx_avail(bp, txr) <
(skb_shinfo(skb)->nr_frags + 1))) {
- netif_stop_queue(dev);
+ netif_tx_stop_queue(txq);
printk(KERN_ERR PFX "%s: BUG! Tx ring full when queue awake!\n",
dev->name);
@@ -6047,9 +6064,9 @@ bnx2_start_xmit(struct sk_buff *skb, struct net_device *dev)
dev->trans_start = jiffies;
if (unlikely(bnx2_tx_avail(bp, txr) <= MAX_SKB_FRAGS)) {
- netif_stop_queue(dev);
+ netif_tx_stop_queue(txq);
if (bnx2_tx_avail(bp, txr) > bp->tx_wake_thresh)
- netif_wake_queue(dev);
+ netif_tx_wake_queue(txq);
}
return NETDEV_TX_OK;
@@ -7294,7 +7311,7 @@ bnx2_init_board(struct pci_dev *pdev, struct net_device *dev)
INIT_WORK(&bp->reset_task, bnx2_reset_task);
dev->base_addr = dev->mem_start = pci_resource_start(pdev, 0);
- mem_len = MB_GET_CID_ADDR(TX_TSS_CID + 1);
+ mem_len = MB_GET_CID_ADDR(TX_TSS_CID + TX_MAX_TSS_RINGS);
dev->mem_end = dev->mem_start + mem_len;
dev->irq = pdev->irq;
@@ -7647,7 +7664,7 @@ bnx2_init_one(struct pci_dev *pdev, const struct pci_device_id *ent)
printk(KERN_INFO "%s", version);
/* dev zeroed in init_etherdev */
- dev = alloc_etherdev(sizeof(*bp));
+ dev = alloc_etherdev_mq(sizeof(*bp), TX_MAX_RINGS);
if (!dev)
return -ENOMEM;
diff --git a/drivers/net/bnx2.h b/drivers/net/bnx2.h
index bb7b5d5..c3c579f 100644
--- a/drivers/net/bnx2.h
+++ b/drivers/net/bnx2.h
@@ -6509,6 +6509,9 @@ struct l2_fhdr {
#define RX_CID 0
#define RX_RSS_CID 4
#define RX_MAX_RSS_RINGS 7
+#define RX_MAX_RINGS (RX_MAX_RSS_RINGS + 1)
+#define TX_MAX_TSS_RINGS 7
+#define TX_MAX_RINGS (TX_MAX_TSS_RINGS + 1)
#define MB_TX_CID_ADDR MB_GET_CID_ADDR(TX_CID)
#define MB_RX_CID_ADDR MB_GET_CID_ADDR(RX_CID)
--
1.5.6.1
--
To unsubscribe from this list: send the line "unsubscribe netdev" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Powered by blists - more mailing lists