From 90fb51358a264f2f7e0cabdca6fc229bbc40cd91 Mon Sep 17 00:00:00 2001
From: Auke Kok <auke\-jan.h.kok@intel.com>
Date: Wed, 1 Nov 2006 08:47:30 -0800
Subject: e1000: whitespace changes, comments, typos

Small whitespace changes, comment changes, typo fixes.

Signed-off-by: Auke Kok <auke-jan.h.kok@intel.com>
---
 drivers/net/e1000/e1000_main.c | 101 +++++++++++++++++++++--------------------
 1 file changed, 51 insertions(+), 50 deletions(-)

(limited to 'drivers/net/e1000/e1000_main.c')

diff --git a/drivers/net/e1000/e1000_main.c b/drivers/net/e1000/e1000_main.c
index 726ec5e88ab..7b1c092d921 100644
--- a/drivers/net/e1000/e1000_main.c
+++ b/drivers/net/e1000/e1000_main.c
@@ -375,7 +375,7 @@ e1000_update_mng_vlan(struct e1000_adapter *adapter)
  * e1000_release_hw_control resets {CTRL_EXT|FWSM}:DRV_LOAD bit.
  * For ASF and Pass Through versions of f/w this means that the
  * driver is no longer loaded. For AMT version (only with 82573) i
- * of the f/w this means that the netowrk i/f is closed.
+ * of the f/w this means that the network i/f is closed.
  *
  **/
 
@@ -416,7 +416,7 @@ e1000_release_hw_control(struct e1000_adapter *adapter)
  * e1000_get_hw_control sets {CTRL_EXT|FWSM}:DRV_LOAD bit.
  * For ASF and Pass Through versions of f/w this means that
  * the driver is loaded. For AMT version (only with 82573)
- * of the f/w this means that the netowrk i/f is open.
+ * of the f/w this means that the network i/f is open.
  *
  **/
 
@@ -426,6 +426,7 @@ e1000_get_hw_control(struct e1000_adapter *adapter)
 	uint32_t ctrl_ext;
 	uint32_t swsm;
 	uint32_t extcnf;
+
 	/* Let firmware know the driver has taken over */
 	switch (adapter->hw.mac_type) {
 	case e1000_82571:
@@ -1279,12 +1280,10 @@ e1000_open(struct net_device *netdev)
 		return -EBUSY;
 
 	/* allocate transmit descriptors */
-
 	if ((err = e1000_setup_all_tx_resources(adapter)))
 		goto err_setup_tx;
 
 	/* allocate receive descriptors */
-
 	if ((err = e1000_setup_all_rx_resources(adapter)))
 		goto err_setup_rx;
 
@@ -1569,6 +1568,8 @@ e1000_configure_tx(struct e1000_adapter *adapter)
 
 	if (hw->mac_type == e1000_82571 || hw->mac_type == e1000_82572) {
 		tarc = E1000_READ_REG(hw, TARC0);
+		/* set the speed mode bit, we'll clear it if we're not at
+		 * gigabit link later */
 		tarc |= (1 << 21);
 		E1000_WRITE_REG(hw, TARC0, tarc);
 	} else if (hw->mac_type == e1000_80003es2lan) {
@@ -2418,6 +2419,7 @@ e1000_watchdog(unsigned long data)
 		DPRINTK(LINK, INFO,
 			"Gigabit has been disabled, downgrading speed\n");
 	}
+
 	if (adapter->hw.mac_type == e1000_82573) {
 		e1000_enable_tx_pkt_filtering(&adapter->hw);
 		if (adapter->mng_vlan_id != adapter->hw.mng_cookie.vlan_id)
@@ -2462,13 +2464,12 @@ e1000_watchdog(unsigned long data)
 			if ((adapter->hw.mac_type == e1000_82571 ||
 			     adapter->hw.mac_type == e1000_82572) &&
 			    txb2b == 0) {
-#define SPEED_MODE_BIT (1 << 21)
 				uint32_t tarc0;
 				tarc0 = E1000_READ_REG(&adapter->hw, TARC0);
-				tarc0 &= ~SPEED_MODE_BIT;
+				tarc0 &= ~(1 << 21);
 				E1000_WRITE_REG(&adapter->hw, TARC0, tarc0);
 			}
-				
+
 #ifdef NETIF_F_TSO
 			/* disable TSO for pcie and 10/100 speeds, to avoid
 			 * some hardware issues */
@@ -3010,9 +3011,9 @@ e1000_xmit_frame(struct sk_buff *skb, struct net_device *netdev)
 		max_per_txd = min(mss << 2, max_per_txd);
 		max_txd_pwr = fls(max_per_txd) - 1;
 
-	/* TSO Workaround for 82571/2/3 Controllers -- if skb->data
-	 * points to just header, pull a few bytes of payload from
-	 * frags into skb->data */
+		/* TSO Workaround for 82571/2/3 Controllers -- if skb->data
+		* points to just header, pull a few bytes of payload from
+		* frags into skb->data */
 		hdr_len = ((skb->h.raw - skb->data) + (skb->h.th->doff << 2));
 		if (skb->data_len && (hdr_len == (skb->len - skb->data_len))) {
 			switch (adapter->hw.mac_type) {
@@ -3316,12 +3317,12 @@ e1000_update_stats(struct e1000_adapter *adapter)
 	adapter->stats.roc += E1000_READ_REG(hw, ROC);
 
 	if (adapter->hw.mac_type != e1000_ich8lan) {
-	adapter->stats.prc64 += E1000_READ_REG(hw, PRC64);
-	adapter->stats.prc127 += E1000_READ_REG(hw, PRC127);
-	adapter->stats.prc255 += E1000_READ_REG(hw, PRC255);
-	adapter->stats.prc511 += E1000_READ_REG(hw, PRC511);
-	adapter->stats.prc1023 += E1000_READ_REG(hw, PRC1023);
-	adapter->stats.prc1522 += E1000_READ_REG(hw, PRC1522);
+		adapter->stats.prc64 += E1000_READ_REG(hw, PRC64);
+		adapter->stats.prc127 += E1000_READ_REG(hw, PRC127);
+		adapter->stats.prc255 += E1000_READ_REG(hw, PRC255);
+		adapter->stats.prc511 += E1000_READ_REG(hw, PRC511);
+		adapter->stats.prc1023 += E1000_READ_REG(hw, PRC1023);
+		adapter->stats.prc1522 += E1000_READ_REG(hw, PRC1522);
 	}
 
 	adapter->stats.symerrs += E1000_READ_REG(hw, SYMERRS);
@@ -3352,12 +3353,12 @@ e1000_update_stats(struct e1000_adapter *adapter)
 	adapter->stats.tpr += E1000_READ_REG(hw, TPR);
 
 	if (adapter->hw.mac_type != e1000_ich8lan) {
-	adapter->stats.ptc64 += E1000_READ_REG(hw, PTC64);
-	adapter->stats.ptc127 += E1000_READ_REG(hw, PTC127);
-	adapter->stats.ptc255 += E1000_READ_REG(hw, PTC255);
-	adapter->stats.ptc511 += E1000_READ_REG(hw, PTC511);
-	adapter->stats.ptc1023 += E1000_READ_REG(hw, PTC1023);
-	adapter->stats.ptc1522 += E1000_READ_REG(hw, PTC1522);
+		adapter->stats.ptc64 += E1000_READ_REG(hw, PTC64);
+		adapter->stats.ptc127 += E1000_READ_REG(hw, PTC127);
+		adapter->stats.ptc255 += E1000_READ_REG(hw, PTC255);
+		adapter->stats.ptc511 += E1000_READ_REG(hw, PTC511);
+		adapter->stats.ptc1023 += E1000_READ_REG(hw, PTC1023);
+		adapter->stats.ptc1522 += E1000_READ_REG(hw, PTC1522);
 	}
 
 	adapter->stats.mptc += E1000_READ_REG(hw, MPTC);
@@ -3383,18 +3384,17 @@ e1000_update_stats(struct e1000_adapter *adapter)
 		adapter->stats.icrxoc += E1000_READ_REG(hw, ICRXOC);
 
 		if (adapter->hw.mac_type != e1000_ich8lan) {
-		adapter->stats.icrxptc += E1000_READ_REG(hw, ICRXPTC);
-		adapter->stats.icrxatc += E1000_READ_REG(hw, ICRXATC);
-		adapter->stats.ictxptc += E1000_READ_REG(hw, ICTXPTC);
-		adapter->stats.ictxatc += E1000_READ_REG(hw, ICTXATC);
-		adapter->stats.ictxqec += E1000_READ_REG(hw, ICTXQEC);
-		adapter->stats.ictxqmtc += E1000_READ_REG(hw, ICTXQMTC);
-		adapter->stats.icrxdmtc += E1000_READ_REG(hw, ICRXDMTC);
+			adapter->stats.icrxptc += E1000_READ_REG(hw, ICRXPTC);
+			adapter->stats.icrxatc += E1000_READ_REG(hw, ICRXATC);
+			adapter->stats.ictxptc += E1000_READ_REG(hw, ICTXPTC);
+			adapter->stats.ictxatc += E1000_READ_REG(hw, ICTXATC);
+			adapter->stats.ictxqec += E1000_READ_REG(hw, ICTXQEC);
+			adapter->stats.ictxqmtc += E1000_READ_REG(hw, ICTXQMTC);
+			adapter->stats.icrxdmtc += E1000_READ_REG(hw, ICRXDMTC);
 		}
 	}
 
 	/* Fill out the OS statistics structure */
-
 	adapter->net_stats.rx_packets = adapter->stats.gprc;
 	adapter->net_stats.tx_packets = adapter->stats.gptc;
 	adapter->net_stats.rx_bytes = adapter->stats.gorcl;
@@ -3426,7 +3426,6 @@ e1000_update_stats(struct e1000_adapter *adapter)
 	/* Tx Dropped needs to be maintained elsewhere */
 
 	/* Phy Stats */
-
 	if (hw->media_type == e1000_media_type_copper) {
 		if ((adapter->link_speed == SPEED_1000) &&
 		   (!e1000_read_phy_reg(hw, PHY_1000T_STATUS, &phy_tmp))) {
@@ -3502,6 +3501,8 @@ e1000_intr(int irq, void *data)
 	if (likely(netif_rx_schedule_prep(netdev)))
 		__netif_rx_schedule(netdev);
 	else
+		/* this really should not happen! if it does it is basically a
+		 * bug, but not a hard error, so enable ints and continue */
 		e1000_irq_enable(adapter);
 #else
 	/* Writing IMC and IMS is needed for 82547.
@@ -3528,7 +3529,6 @@ e1000_intr(int irq, void *data)
 		e1000_irq_enable(adapter);
 
 #endif
-
 	return IRQ_HANDLED;
 }
 
@@ -3615,7 +3615,6 @@ e1000_clean_tx_irq(struct e1000_adapter *adapter,
 			if (unlikely(++i == tx_ring->count)) i = 0;
 		}
 
-
 		eop = tx_ring->buffer_info[i].next_to_watch;
 		eop_desc = E1000_TX_DESC(*tx_ring, eop);
 #ifdef CONFIG_E1000_NAPI
@@ -3760,6 +3759,7 @@ e1000_clean_rx_irq(struct e1000_adapter *adapter,
 	while (rx_desc->status & E1000_RXD_STAT_DD) {
 		struct sk_buff *skb;
 		u8 status;
+
 #ifdef CONFIG_E1000_NAPI
 		if (*work_done >= work_to_do)
 			break;
@@ -3999,7 +3999,7 @@ e1000_clean_rx_irq_ps(struct e1000_adapter *adapter,
 			goto copydone;
 		} /* if */
 		}
-		
+
 		for (j = 0; j < adapter->rx_ps_pages; j++) {
 			if (!(length= le16_to_cpu(rx_desc->wb.upper.length[j])))
 				break;
@@ -4234,7 +4234,7 @@ e1000_alloc_rx_buffers_ps(struct e1000_adapter *adapter,
 		}
 
 		skb = netdev_alloc_skb(netdev,
-				       adapter->rx_ps_bsize0 + NET_IP_ALIGN);
+		                       adapter->rx_ps_bsize0 + NET_IP_ALIGN);
 
 		if (unlikely(!skb)) {
 			adapter->alloc_rx_buff_failed++;
@@ -4511,7 +4511,6 @@ e1000_read_pcie_cap_reg(struct e1000_hw *hw, uint32_t reg, uint16_t *value)
     return E1000_SUCCESS;
 }
 
-
 void
 e1000_io_write(struct e1000_hw *hw, unsigned long port, uint32_t value)
 {
@@ -4534,12 +4533,12 @@ e1000_vlan_rx_register(struct net_device *netdev, struct vlan_group *grp)
 		E1000_WRITE_REG(&adapter->hw, CTRL, ctrl);
 
 		if (adapter->hw.mac_type != e1000_ich8lan) {
-		/* enable VLAN receive filtering */
-		rctl = E1000_READ_REG(&adapter->hw, RCTL);
-		rctl |= E1000_RCTL_VFE;
-		rctl &= ~E1000_RCTL_CFIEN;
-		E1000_WRITE_REG(&adapter->hw, RCTL, rctl);
-		e1000_update_mng_vlan(adapter);
+			/* enable VLAN receive filtering */
+			rctl = E1000_READ_REG(&adapter->hw, RCTL);
+			rctl |= E1000_RCTL_VFE;
+			rctl &= ~E1000_RCTL_CFIEN;
+			E1000_WRITE_REG(&adapter->hw, RCTL, rctl);
+			e1000_update_mng_vlan(adapter);
 		}
 	} else {
 		/* disable VLAN tag insert/strip */
@@ -4548,14 +4547,16 @@ e1000_vlan_rx_register(struct net_device *netdev, struct vlan_group *grp)
 		E1000_WRITE_REG(&adapter->hw, CTRL, ctrl);
 
 		if (adapter->hw.mac_type != e1000_ich8lan) {
-		/* disable VLAN filtering */
-		rctl = E1000_READ_REG(&adapter->hw, RCTL);
-		rctl &= ~E1000_RCTL_VFE;
-		E1000_WRITE_REG(&adapter->hw, RCTL, rctl);
-		if (adapter->mng_vlan_id != (uint16_t)E1000_MNG_VLAN_NONE) {
-			e1000_vlan_rx_kill_vid(netdev, adapter->mng_vlan_id);
-			adapter->mng_vlan_id = E1000_MNG_VLAN_NONE;
-		}
+			/* disable VLAN filtering */
+			rctl = E1000_READ_REG(&adapter->hw, RCTL);
+			rctl &= ~E1000_RCTL_VFE;
+			E1000_WRITE_REG(&adapter->hw, RCTL, rctl);
+			if (adapter->mng_vlan_id !=
+			    (uint16_t)E1000_MNG_VLAN_NONE) {
+				e1000_vlan_rx_kill_vid(netdev,
+				                       adapter->mng_vlan_id);
+				adapter->mng_vlan_id = E1000_MNG_VLAN_NONE;
+			}
 		}
 	}
 
-- 
cgit v1.2.3


From b00dae7cce465323850c3e1fd3ac8b2d9229735c Mon Sep 17 00:00:00 2001
From: Auke Kok <auke\-jan.h.kok@intel.com>
Date: Wed, 1 Nov 2006 08:47:33 -0800
Subject: e1000: Remove DISABLE_MULR debug code

Remove debugging code disabling MULR (multiple reads). It's not usable
for a wide audience and there are no known problems with MULR right
now.

Signed-off-by: Auke Kok <auke-jan.h.kok@intel.com>
---
 drivers/net/e1000/e1000_main.c | 8 --------
 1 file changed, 8 deletions(-)

(limited to 'drivers/net/e1000/e1000_main.c')

diff --git a/drivers/net/e1000/e1000_main.c b/drivers/net/e1000/e1000_main.c
index 7b1c092d921..35d6d080522 100644
--- a/drivers/net/e1000/e1000_main.c
+++ b/drivers/net/e1000/e1000_main.c
@@ -602,9 +602,6 @@ void
 e1000_reset(struct e1000_adapter *adapter)
 {
 	uint32_t pba, manc;
-#ifdef DISABLE_MULR
-	uint32_t tctl;
-#endif
 	uint16_t fc_high_water_mark = E1000_FC_HIGH_DIFF;
 
 	/* Repartition Pba for greater than 9k mtu
@@ -671,12 +668,7 @@ e1000_reset(struct e1000_adapter *adapter)
 	e1000_reset_hw(&adapter->hw);
 	if (adapter->hw.mac_type >= e1000_82544)
 		E1000_WRITE_REG(&adapter->hw, WUC, 0);
-#ifdef DISABLE_MULR
-	/* disable Multiple Reads in Transmit Control Register for debugging */
-	tctl = E1000_READ_REG(hw, TCTL);
-	E1000_WRITE_REG(hw, TCTL, tctl & ~E1000_TCTL_MULR);
 
-#endif
 	if (e1000_init_hw(&adapter->hw))
 		DPRINTK(PROBE, ERR, "Hardware Error\n");
 	e1000_update_mng_vlan(adapter);
-- 
cgit v1.2.3


From 87ca4e5b8d729fc157a0a599d78ccab245fc0602 Mon Sep 17 00:00:00 2001
From: Auke Kok <auke\-jan.h.kok@intel.com>
Date: Wed, 1 Nov 2006 08:47:36 -0800
Subject: e1000: FIX: enable hw TSO for IPV6

Enable TSO for IPV6. All e1000 hardware supports it. This reduces CPU
utilizations by 50% when transmitting IPv6 frames.

Fix symbol naming enabling ipv6 TSO. Turn off TSO6 for 10/100.

Signed-off-by: Auke Kok <auke-jan.h.kok@intel.com>
---
 drivers/net/e1000/e1000.h         |  3 +++
 drivers/net/e1000/e1000_ethtool.c |  7 +++++++
 drivers/net/e1000/e1000_main.c    | 19 ++++++++++++++-----
 3 files changed, 24 insertions(+), 5 deletions(-)

(limited to 'drivers/net/e1000/e1000_main.c')

diff --git a/drivers/net/e1000/e1000.h b/drivers/net/e1000/e1000.h
index 7ecce438d25..3f0be02e85e 100644
--- a/drivers/net/e1000/e1000.h
+++ b/drivers/net/e1000/e1000.h
@@ -59,6 +59,9 @@
 #include <linux/capability.h>
 #include <linux/in.h>
 #include <linux/ip.h>
+#ifdef NETIF_F_TSO6
+#include <linux/ipv6.h>
+#endif
 #include <linux/tcp.h>
 #include <linux/udp.h>
 #include <net/pkt_sched.h>
diff --git a/drivers/net/e1000/e1000_ethtool.c b/drivers/net/e1000/e1000_ethtool.c
index 82d2c78e169..b9c0927d79b 100644
--- a/drivers/net/e1000/e1000_ethtool.c
+++ b/drivers/net/e1000/e1000_ethtool.c
@@ -348,6 +348,13 @@ e1000_set_tso(struct net_device *netdev, uint32_t data)
 	else
 		netdev->features &= ~NETIF_F_TSO;
 
+#ifdef NETIF_F_TSO6
+	if (data)
+		netdev->features |= NETIF_F_TSO6;
+	else
+		netdev->features &= ~NETIF_F_TSO6;
+#endif
+
 	DPRINTK(PROBE, INFO, "TSO is %s\n", data ? "Enabled" : "Disabled");
 	adapter->tso_force = TRUE;
 	return 0;
diff --git a/drivers/net/e1000/e1000_main.c b/drivers/net/e1000/e1000_main.c
index 35d6d080522..38ee39e570e 100644
--- a/drivers/net/e1000/e1000_main.c
+++ b/drivers/net/e1000/e1000_main.c
@@ -844,9 +844,9 @@ e1000_probe(struct pci_dev *pdev,
 	   (adapter->hw.mac_type != e1000_82547))
 		netdev->features |= NETIF_F_TSO;
 
-#ifdef NETIF_F_TSO_IPV6
+#ifdef NETIF_F_TSO6
 	if (adapter->hw.mac_type > e1000_82547_rev_2)
-		netdev->features |= NETIF_F_TSO_IPV6;
+		netdev->features |= NETIF_F_TSO6;
 #endif
 #endif
 	if (pci_using_dac)
@@ -1814,8 +1814,11 @@ e1000_setup_rctl(struct e1000_adapter *adapter)
 		/* Configure extra packet-split registers */
 		rfctl = E1000_READ_REG(&adapter->hw, RFCTL);
 		rfctl |= E1000_RFCTL_EXTEN;
-		/* disable IPv6 packet split support */
-		rfctl |= E1000_RFCTL_IPV6_DIS;
+		/* disable packet split support for IPv6 extension headers,
+		 * because some malformed IPv6 headers can hang the RX */
+		rfctl |= (E1000_RFCTL_IPV6_EX_DIS |
+		          E1000_RFCTL_NEW_IPV6_EXT_DIS);
+
 		E1000_WRITE_REG(&adapter->hw, RFCTL, rfctl);
 
 		rctl |= E1000_RCTL_DTYP_PS;
@@ -2473,9 +2476,15 @@ e1000_watchdog(unsigned long data)
 					DPRINTK(PROBE,INFO,
 				        "10/100 speed: disabling TSO\n");
 					netdev->features &= ~NETIF_F_TSO;
+#ifdef NETIF_F_TSO6
+					netdev->features &= ~NETIF_F_TSO6;
+#endif
 					break;
 				case SPEED_1000:
 					netdev->features |= NETIF_F_TSO;
+#ifdef NETIF_F_TSO6
+					netdev->features |= NETIF_F_TSO6;
+#endif
 					break;
 				default:
 					/* oops */
@@ -2610,7 +2619,7 @@ e1000_tso(struct e1000_adapter *adapter, struct e1000_tx_ring *tx_ring,
 						   0);
 			cmd_length = E1000_TXD_CMD_IP;
 			ipcse = skb->h.raw - skb->data - 1;
-#ifdef NETIF_F_TSO_IPV6
+#ifdef NETIF_F_TSO6
 		} else if (skb->protocol == htons(ETH_P_IPV6)) {
 			skb->nh.ipv6h->payload_len = 0;
 			skb->h.th->check =
-- 
cgit v1.2.3


From 21c4d5e07859a6fc0f62be37da15b161e142c8d1 Mon Sep 17 00:00:00 2001
From: Auke Kok <auke\-jan.h.kok@intel.com>
Date: Wed, 1 Nov 2006 08:47:39 -0800
Subject: e1000: Enble early receive (ERT) on 82573

Enable early receives on 82573 for jumbo frame performance. Jumbo's
are only supported on 82573L with ASPM disabled.

Signed-off-by: Auke Kok <auke-jan.h.kok@intel.com>
---
 drivers/net/e1000/e1000_main.c | 6 ++++++
 1 file changed, 6 insertions(+)

(limited to 'drivers/net/e1000/e1000_main.c')

diff --git a/drivers/net/e1000/e1000_main.c b/drivers/net/e1000/e1000_main.c
index 38ee39e570e..93b97c6b8a6 100644
--- a/drivers/net/e1000/e1000_main.c
+++ b/drivers/net/e1000/e1000_main.c
@@ -1934,6 +1934,12 @@ e1000_configure_rx(struct e1000_adapter *adapter)
 		E1000_WRITE_REG(hw, RXCSUM, rxcsum);
 	}
 
+	/* enable early receives on 82573, only takes effect if using > 2048
+	 * byte total frame size.  for example only for jumbo frames */
+#define E1000_ERT_2048 0x100
+	if (hw->mac_type == e1000_82573)
+		E1000_WRITE_REG(hw, ERT, E1000_ERT_2048);
+
 	/* Enable Receives */
 	E1000_WRITE_REG(hw, RCTL, rctl);
 }
-- 
cgit v1.2.3


From 2ce9047f5d8464039da8ff986e71be5546e229c0 Mon Sep 17 00:00:00 2001
From: Jesse Brandeburg <jesse.brandeburg@intel.com>
Date: Wed, 1 Nov 2006 08:47:42 -0800
Subject: e1000: add mmiowb() for IA64 to sync tail writes

IA64 SMP systems were seeing TX issues with multiple cpu's attempting
to write tail registers unordered. This mmiowb() fixes the issue.

Signed-off-by: Auke Kok <auke-jan.h.kok@intel.com>
Signed-off-by: Jesse Brandeburg <jesse.brandeburg@intel.com>
---
 drivers/net/e1000/e1000_main.c | 3 +++
 1 file changed, 3 insertions(+)

(limited to 'drivers/net/e1000/e1000_main.c')

diff --git a/drivers/net/e1000/e1000_main.c b/drivers/net/e1000/e1000_main.c
index 93b97c6b8a6..681b738c6da 100644
--- a/drivers/net/e1000/e1000_main.c
+++ b/drivers/net/e1000/e1000_main.c
@@ -2867,6 +2867,9 @@ e1000_tx_queue(struct e1000_adapter *adapter, struct e1000_tx_ring *tx_ring,
 
 	tx_ring->next_to_use = i;
 	writel(i, adapter->hw.hw_addr + tx_ring->tdt);
+	/* we need this if more than one processor can write to our tail
+	 * at a time, it syncronizes IO on IA64/Altix systems */
+	mmiowb();
 }
 
 /**
-- 
cgit v1.2.3


From 996695de21b9b301ebb32379e2950fc2142df3a7 Mon Sep 17 00:00:00 2001
From: Auke Kok <auke\-jan.h.kok@intel.com>
Date: Wed, 1 Nov 2006 08:47:50 -0800
Subject: e1000: simplify skb_put call.

Simplify two calls to skb_put by removing one call to it.

Signed-off-by: Auke Kok <auke-jan.h.kok@intel.com>
---
 drivers/net/e1000/e1000_main.c | 7 +++----
 1 file changed, 3 insertions(+), 4 deletions(-)

(limited to 'drivers/net/e1000/e1000_main.c')

diff --git a/drivers/net/e1000/e1000_main.c b/drivers/net/e1000/e1000_main.c
index 681b738c6da..44ba52229c7 100644
--- a/drivers/net/e1000/e1000_main.c
+++ b/drivers/net/e1000/e1000_main.c
@@ -3842,12 +3842,11 @@ e1000_clean_rx_irq(struct e1000_adapter *adapter,
 				/* save the skb in buffer_info as good */
 				buffer_info->skb = skb;
 				skb = new_skb;
-				skb_put(skb, length);
 			}
-		} else
-			skb_put(skb, length);
-
+			/* else just continue with the old one */
+		}
 		/* end copybreak code */
+		skb_put(skb, length);
 
 		/* Receive Checksum Offload */
 		e1000_rx_checksum(adapter,
-- 
cgit v1.2.3


From a9ebadd640927ac6529d904b4131b17e8019d199 Mon Sep 17 00:00:00 2001
From: Jesse Brandeburg <jesse.brandeburg@intel.com>
Date: Wed, 1 Nov 2006 08:47:53 -0800
Subject: e1000: Remove unneeded and unwanted memsets

This memsetting was added in a paranoid rage debugging TX hangs, but
are no longer of importance. We can beef up the performance quite a
bit removing them. Make sure to fill in next_to_watch to allow this.

Signed-off-by: Auke Kok <auke-jan.h.kok@intel.com>
Signed-off-by: Jesse Brandeburg <jesse.brandeburg@intel.com>
---
 drivers/net/e1000/e1000_main.c | 13 ++++++++++---
 1 file changed, 10 insertions(+), 3 deletions(-)

(limited to 'drivers/net/e1000/e1000_main.c')

diff --git a/drivers/net/e1000/e1000_main.c b/drivers/net/e1000/e1000_main.c
index 44ba52229c7..9d82bbb1a82 100644
--- a/drivers/net/e1000/e1000_main.c
+++ b/drivers/net/e1000/e1000_main.c
@@ -1993,10 +1993,13 @@ e1000_unmap_and_free_tx_resource(struct e1000_adapter *adapter,
 				buffer_info->dma,
 				buffer_info->length,
 				PCI_DMA_TODEVICE);
+		buffer_info->dma = 0;
 	}
-	if (buffer_info->skb)
+	if (buffer_info->skb) {
 		dev_kfree_skb_any(buffer_info->skb);
-	memset(buffer_info, 0, sizeof(struct e1000_buffer));
+		buffer_info->skb = NULL;
+	}
+	/* buffer_info must be completely set up in the transmit path */
 }
 
 /**
@@ -2661,6 +2664,7 @@ e1000_tso(struct e1000_adapter *adapter, struct e1000_tx_ring *tx_ring,
 		context_desc->cmd_and_length = cpu_to_le32(cmd_length);
 
 		buffer_info->time_stamp = jiffies;
+		buffer_info->next_to_watch = i;
 
 		if (++i == tx_ring->count) i = 0;
 		tx_ring->next_to_use = i;
@@ -2695,6 +2699,7 @@ e1000_tx_csum(struct e1000_adapter *adapter, struct e1000_tx_ring *tx_ring,
 		context_desc->cmd_and_length = cpu_to_le32(E1000_TXD_CMD_DEXT);
 
 		buffer_info->time_stamp = jiffies;
+		buffer_info->next_to_watch = i;
 
 		if (unlikely(++i == tx_ring->count)) i = 0;
 		tx_ring->next_to_use = i;
@@ -2763,6 +2768,7 @@ e1000_tx_map(struct e1000_adapter *adapter, struct e1000_tx_ring *tx_ring,
 				size,
 				PCI_DMA_TODEVICE);
 		buffer_info->time_stamp = jiffies;
+		buffer_info->next_to_watch = i;
 
 		len -= size;
 		offset += size;
@@ -2802,6 +2808,7 @@ e1000_tx_map(struct e1000_adapter *adapter, struct e1000_tx_ring *tx_ring,
 					size,
 					PCI_DMA_TODEVICE);
 			buffer_info->time_stamp = jiffies;
+			buffer_info->next_to_watch = i;
 
 			len -= size;
 			offset += size;
@@ -3620,7 +3627,7 @@ e1000_clean_tx_irq(struct e1000_adapter *adapter,
 			cleaned = (i == eop);
 
 			e1000_unmap_and_free_tx_resource(adapter, buffer_info);
-			memset(tx_desc, 0, sizeof(struct e1000_tx_desc));
+			tx_desc->upper.data = 0;
 
 			if (unlikely(++i == tx_ring->count)) i = 0;
 		}
-- 
cgit v1.2.3


From fc2307d00c15385fbdbb5928a8517e5f63c3d068 Mon Sep 17 00:00:00 2001
From: Auke Kok <auke\-jan.h.kok@intel.com>
Date: Wed, 1 Nov 2006 08:47:56 -0800
Subject: e1000: New hardware support

Add support for a Low Profile quad-port PCI-E adapter and 2 variants
of the ICH8 systems' onboard NIC's.

Signed-off-by: Auke Kok <auke-jan.h.kok@intel.com>
---
 drivers/net/e1000/e1000_ethtool.c | 1 +
 drivers/net/e1000/e1000_hw.c      | 3 +++
 drivers/net/e1000/e1000_hw.h      | 3 +++
 drivers/net/e1000/e1000_main.c    | 4 ++++
 4 files changed, 11 insertions(+)

(limited to 'drivers/net/e1000/e1000_main.c')

diff --git a/drivers/net/e1000/e1000_ethtool.c b/drivers/net/e1000/e1000_ethtool.c
index b9c0927d79b..dbac71ba71b 100644
--- a/drivers/net/e1000/e1000_ethtool.c
+++ b/drivers/net/e1000/e1000_ethtool.c
@@ -1725,6 +1725,7 @@ static int e1000_wol_exclusion(struct e1000_adapter *adapter, struct ethtool_wol
 		retval = 0;
 		break;
 	case E1000_DEV_ID_82571EB_QUAD_COPPER:
+	case E1000_DEV_ID_82571EB_QUAD_COPPER_LOWPROFILE:
 	case E1000_DEV_ID_82546GB_QUAD_COPPER_KSP3:
 		/* quad port adapters only support WoL on port A */
 		if (!adapter->quad_port_a) {
diff --git a/drivers/net/e1000/e1000_hw.c b/drivers/net/e1000/e1000_hw.c
index 0e2ccf50d4d..0bb9dd82d63 100644
--- a/drivers/net/e1000/e1000_hw.c
+++ b/drivers/net/e1000/e1000_hw.c
@@ -385,6 +385,7 @@ e1000_set_mac_type(struct e1000_hw *hw)
     case E1000_DEV_ID_82571EB_FIBER:
     case E1000_DEV_ID_82571EB_SERDES:
     case E1000_DEV_ID_82571EB_QUAD_COPPER:
+    case E1000_DEV_ID_82571EB_QUAD_COPPER_LOWPROFILE:
             hw->mac_type = e1000_82571;
         break;
     case E1000_DEV_ID_82572EI_COPPER:
@@ -408,6 +409,8 @@ e1000_set_mac_type(struct e1000_hw *hw)
     case E1000_DEV_ID_ICH8_IGP_AMT:
     case E1000_DEV_ID_ICH8_IGP_C:
     case E1000_DEV_ID_ICH8_IFE:
+    case E1000_DEV_ID_ICH8_IFE_GT:
+    case E1000_DEV_ID_ICH8_IFE_G:
     case E1000_DEV_ID_ICH8_IGP_M:
         hw->mac_type = e1000_ich8lan;
         break;
diff --git a/drivers/net/e1000/e1000_hw.h b/drivers/net/e1000/e1000_hw.h
index f247f264a57..93c1e260ee6 100644
--- a/drivers/net/e1000/e1000_hw.h
+++ b/drivers/net/e1000/e1000_hw.h
@@ -475,6 +475,7 @@ int32_t e1000_check_phy_reset_block(struct e1000_hw *hw);
 #define E1000_DEV_ID_82571EB_FIBER       0x105F
 #define E1000_DEV_ID_82571EB_SERDES      0x1060
 #define E1000_DEV_ID_82571EB_QUAD_COPPER 0x10A4
+#define E1000_DEV_ID_82571EB_QUAD_COPPER_LOWPROFILE  0x10BC
 #define E1000_DEV_ID_82572EI_COPPER      0x107D
 #define E1000_DEV_ID_82572EI_FIBER       0x107E
 #define E1000_DEV_ID_82572EI_SERDES      0x107F
@@ -492,6 +493,8 @@ int32_t e1000_check_phy_reset_block(struct e1000_hw *hw);
 #define E1000_DEV_ID_ICH8_IGP_AMT        0x104A
 #define E1000_DEV_ID_ICH8_IGP_C          0x104B
 #define E1000_DEV_ID_ICH8_IFE            0x104C
+#define E1000_DEV_ID_ICH8_IFE_GT         0x10C4
+#define E1000_DEV_ID_ICH8_IFE_G          0x10C5
 #define E1000_DEV_ID_ICH8_IGP_M          0x104D
 
 
diff --git a/drivers/net/e1000/e1000_main.c b/drivers/net/e1000/e1000_main.c
index 9d82bbb1a82..96e44a82544 100644
--- a/drivers/net/e1000/e1000_main.c
+++ b/drivers/net/e1000/e1000_main.c
@@ -103,6 +103,9 @@ static struct pci_device_id e1000_pci_tbl[] = {
 	INTEL_E1000_ETHERNET_DEVICE(0x10B9),
 	INTEL_E1000_ETHERNET_DEVICE(0x10BA),
 	INTEL_E1000_ETHERNET_DEVICE(0x10BB),
+	INTEL_E1000_ETHERNET_DEVICE(0x10BC),
+	INTEL_E1000_ETHERNET_DEVICE(0x10C4),
+	INTEL_E1000_ETHERNET_DEVICE(0x10C5),
 	/* required last entry */
 	{0,}
 };
@@ -961,6 +964,7 @@ e1000_probe(struct pci_dev *pdev,
 		break;
 	case E1000_DEV_ID_82546GB_QUAD_COPPER_KSP3:
 	case E1000_DEV_ID_82571EB_QUAD_COPPER:
+	case E1000_DEV_ID_82571EB_QUAD_COPPER_LOWPROFILE:
 		/* if quad port adapter, disable WoL on all but port A */
 		if (global_quad_port_a != 0)
 			adapter->eeprom_wol = 0;
-- 
cgit v1.2.3


From fcfb1224250d7877b6a6c6a947986e08b2160fb6 Mon Sep 17 00:00:00 2001
From: Jesse Brandeburg <jesse.brandeburg@intel.com>
Date: Wed, 1 Nov 2006 08:47:59 -0800
Subject: e1000: add queue restart counter

Add a netif_wake/start_queue counter to the ethtool statistics to indicated
to the user that their transmit ring could be too small for their workload.

Signed-off-by: Jesse brandeburg <jesse.brandeburg@intel.com>
Cc: Jamal Hadi <hadi@cyberus.ca>
Signed-off-by: Auke Kok <auke-jan.h.kok@intel.com>
---
 drivers/net/e1000/e1000.h         | 1 +
 drivers/net/e1000/e1000_ethtool.c | 1 +
 drivers/net/e1000/e1000_main.c    | 5 ++++-
 3 files changed, 6 insertions(+), 1 deletion(-)

(limited to 'drivers/net/e1000/e1000_main.c')

diff --git a/drivers/net/e1000/e1000.h b/drivers/net/e1000/e1000.h
index 3f0be02e85e..896ea8a7238 100644
--- a/drivers/net/e1000/e1000.h
+++ b/drivers/net/e1000/e1000.h
@@ -265,6 +265,7 @@ struct e1000_adapter {
 
 	/* TX */
 	struct e1000_tx_ring *tx_ring;      /* One per active queue */
+	unsigned int restart_queue;
 	unsigned long tx_queue_len;
 	uint32_t txd_cmd;
 	uint32_t tx_int_delay;
diff --git a/drivers/net/e1000/e1000_ethtool.c b/drivers/net/e1000/e1000_ethtool.c
index dbac71ba71b..da459f7177c 100644
--- a/drivers/net/e1000/e1000_ethtool.c
+++ b/drivers/net/e1000/e1000_ethtool.c
@@ -85,6 +85,7 @@ static const struct e1000_stats e1000_gstrings_stats[] = {
 	{ "tx_single_coll_ok", E1000_STAT(stats.scc) },
 	{ "tx_multi_coll_ok", E1000_STAT(stats.mcc) },
 	{ "tx_timeout_count", E1000_STAT(tx_timeout_count) },
+	{ "tx_restart_queue", E1000_STAT(restart_queue) },
 	{ "rx_long_length_errors", E1000_STAT(stats.roc) },
 	{ "rx_short_length_errors", E1000_STAT(stats.ruc) },
 	{ "rx_align_errors", E1000_STAT(stats.algnerrc) },
diff --git a/drivers/net/e1000/e1000_main.c b/drivers/net/e1000/e1000_main.c
index 96e44a82544..705e654c368 100644
--- a/drivers/net/e1000/e1000_main.c
+++ b/drivers/net/e1000/e1000_main.c
@@ -2974,6 +2974,7 @@ static int __e1000_maybe_stop_tx(struct net_device *netdev, int size)
 
 	/* A reprieve! */
 	netif_start_queue(netdev);
+	++adapter->restart_queue;
 	return 0;
 }
 
@@ -3654,8 +3655,10 @@ e1000_clean_tx_irq(struct e1000_adapter *adapter,
 		 * sees the new next_to_clean.
 		 */
 		smp_mb();
-		if (netif_queue_stopped(netdev))
+		if (netif_queue_stopped(netdev)) {
 			netif_wake_queue(netdev);
+			++adapter->restart_queue;
+		}
 	}
 
 	if (adapter->detect_tx_hung) {
-- 
cgit v1.2.3


From 6a042dab19567fc888d5b87ce6ab68ac02aea1dc Mon Sep 17 00:00:00 2001
From: Jesse Brandeburg <jesse.brandeburg@intel.com>
Date: Wed, 1 Nov 2006 08:48:04 -0800
Subject: e1000: Only set IDE for tx when we are using TIDV/TADV

Spec fix: don't set IDE unless we are actually setting the tx
int delay time.

Signed-off-by: Auke Kok <auke-jan.h.kok@intel.com>
Signed-off-by: Jesse Brandeburg <jesse.brandeburg@intel.com>
---
 drivers/net/e1000/e1000_main.c | 7 +++++--
 1 file changed, 5 insertions(+), 2 deletions(-)

(limited to 'drivers/net/e1000/e1000_main.c')

diff --git a/drivers/net/e1000/e1000_main.c b/drivers/net/e1000/e1000_main.c
index 705e654c368..fe22ae8f0a3 100644
--- a/drivers/net/e1000/e1000_main.c
+++ b/drivers/net/e1000/e1000_main.c
@@ -1580,8 +1580,11 @@ e1000_configure_tx(struct e1000_adapter *adapter)
 	e1000_config_collision_dist(hw);
 
 	/* Setup Transmit Descriptor Settings for eop descriptor */
-	adapter->txd_cmd = E1000_TXD_CMD_IDE | E1000_TXD_CMD_EOP |
-		E1000_TXD_CMD_IFCS;
+	adapter->txd_cmd = E1000_TXD_CMD_EOP | E1000_TXD_CMD_IFCS;
+
+	/* only set IDE if we are delaying interrupts using the timers */
+	if (adapter->tx_int_delay)
+		adapter->txd_cmd |= E1000_TXD_CMD_IDE;
 
 	if (hw->mac_type < e1000_82543)
 		adapter->txd_cmd |= E1000_TXD_CMD_RPS;
-- 
cgit v1.2.3


From 9ac98284428961bd5be285a6cc1f5e6f5b6644aa Mon Sep 17 00:00:00 2001
From: Jesse Brandeburg <jesse.brandeburg@intel.com>
Date: Wed, 1 Nov 2006 08:48:10 -0800
Subject: e1000: add dynamic generic MSI interrupt routine

Add a generic MSI interrupt routine that is IO read-free, speeding up
MSI interrupt handling.

Signed-off-by: Jesse Brandeburg <jesse.brandeburg@intel.com>
Signed-off-by: Auke Kok <auke-jan.h.kok@intel.com>
---
 drivers/net/e1000/e1000_main.c | 90 +++++++++++++++++++++++++++++++++++++++++-
 1 file changed, 88 insertions(+), 2 deletions(-)

(limited to 'drivers/net/e1000/e1000_main.c')

diff --git a/drivers/net/e1000/e1000_main.c b/drivers/net/e1000/e1000_main.c
index fe22ae8f0a3..35e4e32c770 100644
--- a/drivers/net/e1000/e1000_main.c
+++ b/drivers/net/e1000/e1000_main.c
@@ -157,6 +157,9 @@ static struct net_device_stats * e1000_get_stats(struct net_device *netdev);
 static int e1000_change_mtu(struct net_device *netdev, int new_mtu);
 static int e1000_set_mac(struct net_device *netdev, void *p);
 static irqreturn_t e1000_intr(int irq, void *data);
+#ifdef CONFIG_PCI_MSI
+static irqreturn_t e1000_intr_msi(int irq, void *data);
+#endif
 static boolean_t e1000_clean_tx_irq(struct e1000_adapter *adapter,
                                     struct e1000_tx_ring *tx_ring);
 #ifdef CONFIG_E1000_NAPI
@@ -288,7 +291,7 @@ static int e1000_request_irq(struct e1000_adapter *adapter)
 
 	flags = IRQF_SHARED;
 #ifdef CONFIG_PCI_MSI
-	if (adapter->hw.mac_type > e1000_82547_rev_2) {
+	if (adapter->hw.mac_type >= e1000_82571) {
 		adapter->have_msi = TRUE;
 		if ((err = pci_enable_msi(adapter->pdev))) {
 			DPRINTK(PROBE, ERR,
@@ -296,8 +299,14 @@ static int e1000_request_irq(struct e1000_adapter *adapter)
 			adapter->have_msi = FALSE;
 		}
 	}
-	if (adapter->have_msi)
+	if (adapter->have_msi) {
 		flags &= ~IRQF_SHARED;
+		err = request_irq(adapter->pdev->irq, &e1000_intr_msi, flags,
+		                  netdev->name, netdev);
+		if (err)
+			DPRINTK(PROBE, ERR,
+			       "Unable to allocate interrupt Error: %d\n", err);
+	} else
 #endif
 	if ((err = request_irq(adapter->pdev->irq, &e1000_intr, flags,
 	                       netdev->name, netdev)))
@@ -3466,6 +3475,83 @@ e1000_update_stats(struct e1000_adapter *adapter)
 
 	spin_unlock_irqrestore(&adapter->stats_lock, flags);
 }
+#ifdef CONFIG_PCI_MSI
+
+/**
+ * e1000_intr_msi - Interrupt Handler
+ * @irq: interrupt number
+ * @data: pointer to a network interface device structure
+ **/
+
+static
+irqreturn_t e1000_intr_msi(int irq, void *data)
+{
+	struct net_device *netdev = data;
+	struct e1000_adapter *adapter = netdev_priv(netdev);
+	struct e1000_hw *hw = &adapter->hw;
+#ifndef CONFIG_E1000_NAPI
+	int i;
+#endif
+
+	/* this code avoids the read of ICR but has to get 1000 interrupts
+	 * at every link change event before it will notice the change */
+	if (++adapter->detect_link >= 1000) {
+		uint32_t icr = E1000_READ_REG(hw, ICR);
+#ifdef CONFIG_E1000_NAPI
+		/* read ICR disables interrupts using IAM, so keep up with our
+		 * enable/disable accounting */
+		atomic_inc(&adapter->irq_sem);
+#endif
+		adapter->detect_link = 0;
+		if ((icr & (E1000_ICR_RXSEQ | E1000_ICR_LSC)) &&
+		    (icr & E1000_ICR_INT_ASSERTED)) {
+			hw->get_link_status = 1;
+			/* 80003ES2LAN workaround--
+			* For packet buffer work-around on link down event;
+			* disable receives here in the ISR and
+			* reset adapter in watchdog
+			*/
+			if (netif_carrier_ok(netdev) &&
+			    (adapter->hw.mac_type == e1000_80003es2lan)) {
+				/* disable receives */
+				uint32_t rctl = E1000_READ_REG(hw, RCTL);
+				E1000_WRITE_REG(hw, RCTL, rctl & ~E1000_RCTL_EN);
+			}
+			/* guard against interrupt when we're going down */
+			if (!test_bit(__E1000_DOWN, &adapter->flags))
+				mod_timer(&adapter->watchdog_timer,
+				          jiffies + 1);
+		}
+	} else {
+		E1000_WRITE_REG(hw, ICR, (0xffffffff & ~(E1000_ICR_RXSEQ |
+		                                         E1000_ICR_LSC)));
+		/* bummer we have to flush here, but things break otherwise as
+		 * some event appears to be lost or delayed and throughput
+		 * drops.  In almost all tests this flush is un-necessary */
+		E1000_WRITE_FLUSH(hw);
+#ifdef CONFIG_E1000_NAPI
+		/* Interrupt Auto-Mask (IAM)...upon writing ICR, interrupts are
+		 * masked.  No need for the IMC write, but it does mean we
+		 * should account for it ASAP. */
+		atomic_inc(&adapter->irq_sem);
+#endif
+	}
+
+#ifdef CONFIG_E1000_NAPI
+	if (likely(netif_rx_schedule_prep(netdev)))
+		__netif_rx_schedule(netdev);
+	else
+		e1000_irq_enable(adapter);
+#else
+	for (i = 0; i < E1000_MAX_INTR; i++)
+		if (unlikely(!adapter->clean_rx(adapter, adapter->rx_ring) &
+		   !e1000_clean_tx_irq(adapter, adapter->tx_ring)))
+			break;
+#endif
+
+	return IRQ_HANDLED;
+}
+#endif
 
 /**
  * e1000_intr - Interrupt Handler
-- 
cgit v1.2.3


From 835bb1298311f372a3387fb40b952b18d90aa9f8 Mon Sep 17 00:00:00 2001
From: Jesse Brandeburg <jesse.brandeburg@intel.com>
Date: Wed, 1 Nov 2006 08:48:13 -0800
Subject: e1000: add dynamic itr modes

Add a new dynamic itr algorithm, with 2 modes, and make it the default
operation mode. This greatly reduces latency and increases small packet
performance, at the "cost" of some CPU utilization. Bulk traffic
throughput is unaffected.

The driver can limit the amount of interrupts per second that the
adapter will generate for incoming packets. It does this by writing a
value to the adapter that is based on the maximum amount of interrupts
that the adapter will generate per second.

Setting InterruptThrottleRate to a value greater or equal to 100 will
program the adapter to send out a maximum of that many interrupts per
second, even if more packets have come in. This reduces interrupt
load on the system and can lower CPU utilization under heavy load,
but will increase latency as packets are not processed as quickly.

The default behaviour of the driver previously assumed a static
InterruptThrottleRate value of 8000, providing a good fallback value
for all traffic types,but lacking in small packet performance and
latency. The hardware can handle many more small packets per second
however, and for this reason an adaptive interrupt moderation algorithm
was implemented.

Since 7.3.x, the driver has two adaptive modes (setting 1 or 3) in
which it dynamically adjusts the InterruptThrottleRate value based on
the traffic that it receives. After determining the type of incoming
traffic in the last timeframe, it will adjust the InterruptThrottleRate
to an appropriate value for that traffic.

The algorithm classifies the incoming traffic every interval into
classes.  Once the class is determined, the InterruptThrottleRate
value is adjusted to suit that traffic type the best. There are
three classes defined: "Bulk traffic", for large amounts of packets
of normal size; "Low latency", for small amounts of traffic and/or
a significant percentage of small packets; and "Lowest latency",
for almost completely small packets or minimal traffic.

In dynamic conservative mode, the InterruptThrottleRate value is
set to 4000 for traffic that falls in class "Bulk traffic". If
traffic falls in the "Low latency" or "Lowest latency" class, the
InterruptThrottleRate is increased stepwise to 20000. This default
mode is suitable for most applications.

For situations where low latency is vital such as cluster or
grid computing, the algorithm can reduce latency even more when
InterruptThrottleRate is set to mode 1. In this mode, which operates
the same as mode 3, the InterruptThrottleRate will be increased
stepwise to 70000 for traffic in class "Lowest latency".

Setting InterruptThrottleRate to 0 turns off any interrupt moderation
and may improve small packet latency, but is generally not suitable
for bulk throughput traffic.

Signed-off-by: Jesse Brandeburg <jesse.brandeburg@intel.com>
Cc: Rick Jones <rick.jones2@hp.com>
Signed-off-by: Auke Kok <auke-jan.h.kok@intel.com>
---
 drivers/net/e1000/e1000.h       |  13 ++-
 drivers/net/e1000/e1000_main.c  | 226 ++++++++++++++++++++++++++++++++++------
 drivers/net/e1000/e1000_param.c |  28 +++--
 3 files changed, 228 insertions(+), 39 deletions(-)

(limited to 'drivers/net/e1000/e1000_main.c')

diff --git a/drivers/net/e1000/e1000.h b/drivers/net/e1000/e1000.h
index 896ea8a7238..f091042b146 100644
--- a/drivers/net/e1000/e1000.h
+++ b/drivers/net/e1000/e1000.h
@@ -257,6 +257,17 @@ struct e1000_adapter {
 	spinlock_t tx_queue_lock;
 #endif
 	atomic_t irq_sem;
+	unsigned int detect_link;
+	unsigned int total_tx_bytes;
+	unsigned int total_tx_packets;
+	unsigned int total_rx_bytes;
+	unsigned int total_rx_packets;
+	/* Interrupt Throttle Rate */
+	uint32_t itr;
+	uint32_t itr_setting;
+	uint16_t tx_itr;
+	uint16_t rx_itr;
+
 	struct work_struct reset_task;
 	uint8_t fc_autoneg;
 
@@ -314,8 +325,6 @@ struct e1000_adapter {
 	uint64_t gorcl_old;
 	uint16_t rx_ps_bsize0;
 
-	/* Interrupt Throttle Rate */
-	uint32_t itr;
 
 	/* OS defined structs */
 	struct net_device *netdev;
diff --git a/drivers/net/e1000/e1000_main.c b/drivers/net/e1000/e1000_main.c
index 35e4e32c770..56be5c89363 100644
--- a/drivers/net/e1000/e1000_main.c
+++ b/drivers/net/e1000/e1000_main.c
@@ -1897,7 +1897,7 @@ e1000_configure_rx(struct e1000_adapter *adapter)
 
 	if (hw->mac_type >= e1000_82540) {
 		E1000_WRITE_REG(hw, RADV, adapter->rx_abs_int_delay);
-		if (adapter->itr > 1)
+		if (adapter->itr_setting != 0)
 			E1000_WRITE_REG(hw, ITR,
 				1000000000 / (adapter->itr * 256));
 	}
@@ -1907,11 +1907,11 @@ e1000_configure_rx(struct e1000_adapter *adapter)
 		/* Reset delay timers after every interrupt */
 		ctrl_ext |= E1000_CTRL_EXT_INT_TIMER_CLR;
 #ifdef CONFIG_E1000_NAPI
-		/* Auto-Mask interrupts upon ICR read. */
+		/* Auto-Mask interrupts upon ICR access */
 		ctrl_ext |= E1000_CTRL_EXT_IAME;
+		E1000_WRITE_REG(hw, IAM, 0xffffffff);
 #endif
 		E1000_WRITE_REG(hw, CTRL_EXT, ctrl_ext);
-		E1000_WRITE_REG(hw, IAM, ~0);
 		E1000_WRITE_FLUSH(hw);
 	}
 
@@ -2576,19 +2576,6 @@ e1000_watchdog(unsigned long data)
 		}
 	}
 
-	/* Dynamic mode for Interrupt Throttle Rate (ITR) */
-	if (adapter->hw.mac_type >= e1000_82540 && adapter->itr == 1) {
-		/* Symmetric Tx/Rx gets a reduced ITR=2000; Total
-		 * asymmetrical Tx or Rx gets ITR=8000; everyone
-		 * else is between 2000-8000. */
-		uint32_t goc = (adapter->gotcl + adapter->gorcl) / 10000;
-		uint32_t dif = (adapter->gotcl > adapter->gorcl ?
-			adapter->gotcl - adapter->gorcl :
-			adapter->gorcl - adapter->gotcl) / 10000;
-		uint32_t itr = goc > 0 ? (dif * 6000 / goc + 2000) : 8000;
-		E1000_WRITE_REG(&adapter->hw, ITR, 1000000000 / (itr * 256));
-	}
-
 	/* Cause software interrupt to ensure rx ring is cleaned */
 	E1000_WRITE_REG(&adapter->hw, ICS, E1000_ICS_RXDMT0);
 
@@ -2604,6 +2591,135 @@ e1000_watchdog(unsigned long data)
 	mod_timer(&adapter->watchdog_timer, jiffies + 2 * HZ);
 }
 
+enum latency_range {
+	lowest_latency = 0,
+	low_latency = 1,
+	bulk_latency = 2,
+	latency_invalid = 255
+};
+
+/**
+ * e1000_update_itr - update the dynamic ITR value based on statistics
+ *      Stores a new ITR value based on packets and byte
+ *      counts during the last interrupt.  The advantage of per interrupt
+ *      computation is faster updates and more accurate ITR for the current
+ *      traffic pattern.  Constants in this function were computed
+ *      based on theoretical maximum wire speed and thresholds were set based
+ *      on testing data as well as attempting to minimize response time
+ *      while increasing bulk throughput.
+ *      this functionality is controlled by the InterruptThrottleRate module
+ *      parameter (see e1000_param.c)
+ * @adapter: pointer to adapter
+ * @itr_setting: current adapter->itr
+ * @packets: the number of packets during this measurement interval
+ * @bytes: the number of bytes during this measurement interval
+ **/
+static unsigned int e1000_update_itr(struct e1000_adapter *adapter,
+                                   uint16_t itr_setting,
+                                   int packets,
+                                   int bytes)
+{
+	unsigned int retval = itr_setting;
+	struct e1000_hw *hw = &adapter->hw;
+
+	if (unlikely(hw->mac_type < e1000_82540))
+		goto update_itr_done;
+
+	if (packets == 0)
+		goto update_itr_done;
+
+
+	switch (itr_setting) {
+	case lowest_latency:
+		if ((packets < 5) && (bytes > 512))
+			retval = low_latency;
+		break;
+	case low_latency:  /* 50 usec aka 20000 ints/s */
+		if (bytes > 10000) {
+			if ((packets < 10) ||
+			     ((bytes/packets) > 1200))
+				retval = bulk_latency;
+			else if ((packets > 35))
+				retval = lowest_latency;
+		} else if (packets <= 2 && bytes < 512)
+			retval = lowest_latency;
+		break;
+	case bulk_latency: /* 250 usec aka 4000 ints/s */
+		if (bytes > 25000) {
+			if (packets > 35)
+				retval = low_latency;
+		} else {
+			if (bytes < 6000)
+				retval = low_latency;
+		}
+		break;
+	}
+
+update_itr_done:
+	return retval;
+}
+
+static void e1000_set_itr(struct e1000_adapter *adapter)
+{
+	struct e1000_hw *hw = &adapter->hw;
+	uint16_t current_itr;
+	uint32_t new_itr = adapter->itr;
+
+	if (unlikely(hw->mac_type < e1000_82540))
+		return;
+
+	/* for non-gigabit speeds, just fix the interrupt rate at 4000 */
+	if (unlikely(adapter->link_speed != SPEED_1000)) {
+		current_itr = 0;
+		new_itr = 4000;
+		goto set_itr_now;
+	}
+
+	adapter->tx_itr = e1000_update_itr(adapter,
+	                            adapter->tx_itr,
+	                            adapter->total_tx_packets,
+	                            adapter->total_tx_bytes);
+	adapter->rx_itr = e1000_update_itr(adapter,
+	                            adapter->rx_itr,
+	                            adapter->total_rx_packets,
+	                            adapter->total_rx_bytes);
+
+	current_itr = max(adapter->rx_itr, adapter->tx_itr);
+
+	/* conservative mode eliminates the lowest_latency setting */
+	if (current_itr == lowest_latency && (adapter->itr_setting == 3))
+		current_itr = low_latency;
+
+	switch (current_itr) {
+	/* counts and packets in update_itr are dependent on these numbers */
+	case lowest_latency:
+		new_itr = 70000;
+		break;
+	case low_latency:
+		new_itr = 20000; /* aka hwitr = ~200 */
+		break;
+	case bulk_latency:
+		new_itr = 4000;
+		break;
+	default:
+		break;
+	}
+
+set_itr_now:
+	if (new_itr != adapter->itr) {
+		/* this attempts to bias the interrupt rate towards Bulk
+		 * by adding intermediate steps when interrupt rate is
+		 * increasing */
+		new_itr = new_itr > adapter->itr ?
+		             min(adapter->itr + (new_itr >> 2), new_itr) :
+		             new_itr;
+		adapter->itr = new_itr;
+		E1000_WRITE_REG(hw, ITR, 1000000000 / (new_itr * 256));
+	}
+
+	return;
+}
+
 #define E1000_TX_FLAGS_CSUM		0x00000001
 #define E1000_TX_FLAGS_VLAN		0x00000002
 #define E1000_TX_FLAGS_TSO		0x00000004
@@ -3538,15 +3654,27 @@ irqreturn_t e1000_intr_msi(int irq, void *data)
 	}
 
 #ifdef CONFIG_E1000_NAPI
-	if (likely(netif_rx_schedule_prep(netdev)))
+	if (likely(netif_rx_schedule_prep(netdev))) {
+		adapter->total_tx_bytes = 0;
+		adapter->total_tx_packets = 0;
+		adapter->total_rx_bytes = 0;
+		adapter->total_rx_packets = 0;
 		__netif_rx_schedule(netdev);
-	else
+	} else
 		e1000_irq_enable(adapter);
 #else
+	adapter->total_tx_bytes = 0;
+	adapter->total_rx_bytes = 0;
+	adapter->total_tx_packets = 0;
+	adapter->total_rx_packets = 0;
+
 	for (i = 0; i < E1000_MAX_INTR; i++)
 		if (unlikely(!adapter->clean_rx(adapter, adapter->rx_ring) &
 		   !e1000_clean_tx_irq(adapter, adapter->tx_ring)))
 			break;
+
+	if (likely(adapter->itr_setting & 3))
+		e1000_set_itr(adapter);
 #endif
 
 	return IRQ_HANDLED;
@@ -3568,7 +3696,17 @@ e1000_intr(int irq, void *data)
 	uint32_t rctl, icr = E1000_READ_REG(hw, ICR);
 #ifndef CONFIG_E1000_NAPI
 	int i;
-#else
+#endif
+	if (unlikely(!icr))
+		return IRQ_NONE;  /* Not our interrupt */
+
+#ifdef CONFIG_E1000_NAPI
+	/* IMS will not auto-mask if INT_ASSERTED is not set, and if it is
+	 * not set, then the adapter didn't send an interrupt */
+	if (unlikely(hw->mac_type >= e1000_82571 &&
+	             !(icr & E1000_ICR_INT_ASSERTED)))
+		return IRQ_NONE;
+
 	/* Interrupt Auto-Mask...upon reading ICR,
 	 * interrupts are masked.  No need for the
 	 * IMC write, but it does mean we should
@@ -3577,14 +3715,6 @@ e1000_intr(int irq, void *data)
 		atomic_inc(&adapter->irq_sem);
 #endif
 
-	if (unlikely(!icr)) {
-#ifdef CONFIG_E1000_NAPI
-		if (hw->mac_type >= e1000_82571)
-			e1000_irq_enable(adapter);
-#endif
-		return IRQ_NONE;  /* Not our interrupt */
-	}
-
 	if (unlikely(icr & (E1000_ICR_RXSEQ | E1000_ICR_LSC))) {
 		hw->get_link_status = 1;
 		/* 80003ES2LAN workaround--
@@ -3605,13 +3735,18 @@ e1000_intr(int irq, void *data)
 
 #ifdef CONFIG_E1000_NAPI
 	if (unlikely(hw->mac_type < e1000_82571)) {
+		/* disable interrupts, without the synchronize_irq bit */
 		atomic_inc(&adapter->irq_sem);
 		E1000_WRITE_REG(hw, IMC, ~0);
 		E1000_WRITE_FLUSH(hw);
 	}
-	if (likely(netif_rx_schedule_prep(netdev)))
+	if (likely(netif_rx_schedule_prep(netdev))) {
+		adapter->total_tx_bytes = 0;
+		adapter->total_tx_packets = 0;
+		adapter->total_rx_bytes = 0;
+		adapter->total_rx_packets = 0;
 		__netif_rx_schedule(netdev);
-	else
+	} else
 		/* this really should not happen! if it does it is basically a
 		 * bug, but not a hard error, so enable ints and continue */
 		e1000_irq_enable(adapter);
@@ -3631,11 +3766,19 @@ e1000_intr(int irq, void *data)
 		E1000_WRITE_REG(hw, IMC, ~0);
 	}
 
+	adapter->total_tx_bytes = 0;
+	adapter->total_rx_bytes = 0;
+	adapter->total_tx_packets = 0;
+	adapter->total_rx_packets = 0;
+
 	for (i = 0; i < E1000_MAX_INTR; i++)
 		if (unlikely(!adapter->clean_rx(adapter, adapter->rx_ring) &
 		   !e1000_clean_tx_irq(adapter, adapter->tx_ring)))
 			break;
 
+	if (likely(adapter->itr_setting & 3))
+		e1000_set_itr(adapter);
+
 	if (hw->mac_type == e1000_82547 || hw->mac_type == e1000_82547_rev_2)
 		e1000_irq_enable(adapter);
 
@@ -3683,6 +3826,8 @@ e1000_clean(struct net_device *poll_dev, int *budget)
 	if ((!tx_cleaned && (work_done == 0)) ||
 	   !netif_running(poll_dev)) {
 quit_polling:
+		if (likely(adapter->itr_setting & 3))
+			e1000_set_itr(adapter);
 		netif_rx_complete(poll_dev);
 		e1000_irq_enable(adapter);
 		return 0;
@@ -3709,6 +3854,7 @@ e1000_clean_tx_irq(struct e1000_adapter *adapter,
 	unsigned int count = 0;
 #endif
 	boolean_t cleaned = FALSE;
+	unsigned int total_tx_bytes=0, total_tx_packets=0;
 
 	i = tx_ring->next_to_clean;
 	eop = tx_ring->buffer_info[i].next_to_watch;
@@ -3720,6 +3866,13 @@ e1000_clean_tx_irq(struct e1000_adapter *adapter,
 			buffer_info = &tx_ring->buffer_info[i];
 			cleaned = (i == eop);
 
+			if (cleaned) {
+				/* this packet count is wrong for TSO but has a
+				 * tendency to make dynamic ITR change more
+				 * towards bulk */
+				total_tx_packets++;
+				total_tx_bytes += buffer_info->skb->len;
+			}
 			e1000_unmap_and_free_tx_resource(adapter, buffer_info);
 			tx_desc->upper.data = 0;
 
@@ -3785,6 +3938,8 @@ e1000_clean_tx_irq(struct e1000_adapter *adapter,
 			netif_stop_queue(netdev);
 		}
 	}
+	adapter->total_tx_bytes += total_tx_bytes;
+	adapter->total_tx_packets += total_tx_packets;
 	return cleaned;
 }
 
@@ -3864,6 +4019,7 @@ e1000_clean_rx_irq(struct e1000_adapter *adapter,
 	unsigned int i;
 	int cleaned_count = 0;
 	boolean_t cleaned = FALSE;
+	unsigned int total_rx_bytes=0, total_rx_packets=0;
 
 	i = rx_ring->next_to_clean;
 	rx_desc = E1000_RX_DESC(*rx_ring, i);
@@ -3930,6 +4086,10 @@ e1000_clean_rx_irq(struct e1000_adapter *adapter,
 		 * done after the TBI_ACCEPT workaround above */
 		length -= 4;
 
+		/* probably a little skewed due to removing CRC */
+		total_rx_bytes += length;
+		total_rx_packets++;
+
 		/* code added for copybreak, this should improve
 		 * performance for small packets with large amounts
 		 * of reassembly being done in the stack */
@@ -3998,6 +4158,8 @@ next_desc:
 	if (cleaned_count)
 		adapter->alloc_rx_buf(adapter, rx_ring, cleaned_count);
 
+	adapter->total_rx_packets += total_rx_packets;
+	adapter->total_rx_bytes += total_rx_bytes;
 	return cleaned;
 }
 
@@ -4027,6 +4189,7 @@ e1000_clean_rx_irq_ps(struct e1000_adapter *adapter,
 	uint32_t length, staterr;
 	int cleaned_count = 0;
 	boolean_t cleaned = FALSE;
+	unsigned int total_rx_bytes=0, total_rx_packets=0;
 
 	i = rx_ring->next_to_clean;
 	rx_desc = E1000_RX_DESC_PS(*rx_ring, i);
@@ -4131,6 +4294,9 @@ e1000_clean_rx_irq_ps(struct e1000_adapter *adapter,
 		pskb_trim(skb, skb->len - 4);
 
 copydone:
+		total_rx_bytes += skb->len;
+		total_rx_packets++;
+
 		e1000_rx_checksum(adapter, staterr,
 				  le16_to_cpu(rx_desc->wb.lower.hi_dword.csum_ip.csum), skb);
 		skb->protocol = eth_type_trans(skb, netdev);
@@ -4179,6 +4345,8 @@ next_desc:
 	if (cleaned_count)
 		adapter->alloc_rx_buf(adapter, rx_ring, cleaned_count);
 
+	adapter->total_rx_packets += total_rx_packets;
+	adapter->total_rx_bytes += total_rx_bytes;
 	return cleaned;
 }
 
diff --git a/drivers/net/e1000/e1000_param.c b/drivers/net/e1000/e1000_param.c
index e4f88920b46..cbfcd7f2889 100644
--- a/drivers/net/e1000/e1000_param.c
+++ b/drivers/net/e1000/e1000_param.c
@@ -139,7 +139,7 @@ E1000_PARAM(XsumRX, "Disable or enable Receive Checksum offload");
  * Valid Range: 0-65535
  */
 E1000_PARAM(TxIntDelay, "Transmit Interrupt Delay");
-#define DEFAULT_TIDV                  64
+#define DEFAULT_TIDV                   8
 #define MAX_TXDELAY               0xFFFF
 #define MIN_TXDELAY                    0
 
@@ -148,7 +148,7 @@ E1000_PARAM(TxIntDelay, "Transmit Interrupt Delay");
  * Valid Range: 0-65535
  */
 E1000_PARAM(TxAbsIntDelay, "Transmit Absolute Interrupt Delay");
-#define DEFAULT_TADV                  64
+#define DEFAULT_TADV                  32
 #define MAX_TXABSDELAY            0xFFFF
 #define MIN_TXABSDELAY                 0
 
@@ -167,16 +167,16 @@ E1000_PARAM(RxIntDelay, "Receive Interrupt Delay");
  * Valid Range: 0-65535
  */
 E1000_PARAM(RxAbsIntDelay, "Receive Absolute Interrupt Delay");
-#define DEFAULT_RADV                 128
+#define DEFAULT_RADV                   8
 #define MAX_RXABSDELAY            0xFFFF
 #define MIN_RXABSDELAY                 0
 
 /* Interrupt Throttle Rate (interrupts/sec)
  *
- * Valid Range: 100-100000 (0=off, 1=dynamic)
+ * Valid Range: 100-100000 (0=off, 1=dynamic, 3=dynamic conservative)
  */
 E1000_PARAM(InterruptThrottleRate, "Interrupt Throttling Rate");
-#define DEFAULT_ITR                 8000
+#define DEFAULT_ITR                    3
 #define MAX_ITR                   100000
 #define MIN_ITR                      100
 
@@ -472,15 +472,27 @@ e1000_check_options(struct e1000_adapter *adapter)
 				break;
 			case 1:
 				DPRINTK(PROBE, INFO, "%s set to dynamic mode\n",
-				        opt.name);
+					opt.name);
+				adapter->itr_setting = adapter->itr;
+				adapter->itr = 20000;
+				break;
+			case 3:
+				DPRINTK(PROBE, INFO,
+				        "%s set to dynamic conservative mode\n",
+					opt.name);
+				adapter->itr_setting = adapter->itr;
+				adapter->itr = 20000;
 				break;
 			default:
 				e1000_validate_option(&adapter->itr, &opt,
-				                      adapter);
+				        adapter);
+				/* save the setting, because the dynamic bits change itr */
+				adapter->itr_setting = adapter->itr;
 				break;
 			}
 		} else {
-			adapter->itr = opt.def;
+			adapter->itr_setting = opt.def;
+			adapter->itr = 20000;
 		}
 	}
 	{ /* Smart Power Down */
-- 
cgit v1.2.3


From 25006ac61e514628b9d0f78fce0bed155f4f109c Mon Sep 17 00:00:00 2001
From: Auke Kok <auke\-jan.h.kok@intel.com>
Date: Wed, 1 Nov 2006 08:48:15 -0800
Subject: e1000: increment version to 7.3.15-k2

Signed-off-by: Auke Kok <auke-jan.h.kok@intel.com>
---
 drivers/net/e1000/e1000_main.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

(limited to 'drivers/net/e1000/e1000_main.c')

diff --git a/drivers/net/e1000/e1000_main.c b/drivers/net/e1000/e1000_main.c
index 56be5c89363..a2f1464ba49 100644
--- a/drivers/net/e1000/e1000_main.c
+++ b/drivers/net/e1000/e1000_main.c
@@ -35,7 +35,7 @@ static char e1000_driver_string[] = "Intel(R) PRO/1000 Network Driver";
 #else
 #define DRIVERNAPI "-NAPI"
 #endif
-#define DRV_VERSION "7.2.9-k4"DRIVERNAPI
+#define DRV_VERSION "7.3.15-k2"DRIVERNAPI
 char e1000_driver_version[] = DRV_VERSION;
 static char e1000_copyright[] = "Copyright (c) 1999-2006 Intel Corporation.";
 
-- 
cgit v1.2.3


From d0bb53e102e10cc90de50953531a163d95da1e07 Mon Sep 17 00:00:00 2001
From: Andrew Morton <akpm@osdl.org>
Date: Tue, 14 Nov 2006 10:35:03 -0500
Subject: e1000 linkage fix

ia64:

 drivers/built-in.o(.text+0xd9a72): In function `e1000_xmit_frame':
 : undefined reference to `csum_ipv6_magic'

Cc: Auke Kok <auke-jan.h.kok@intel.com>
Cc: "David S. Miller" <davem@davemloft.net>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
---
 drivers/net/e1000/e1000_main.c | 1 +
 1 file changed, 1 insertion(+)

(limited to 'drivers/net/e1000/e1000_main.c')

diff --git a/drivers/net/e1000/e1000_main.c b/drivers/net/e1000/e1000_main.c
index a2f1464ba49..7a0828869ec 100644
--- a/drivers/net/e1000/e1000_main.c
+++ b/drivers/net/e1000/e1000_main.c
@@ -27,6 +27,7 @@
 *******************************************************************************/
 
 #include "e1000.h"
+#include <net/ip6_checksum.h>
 
 char e1000_driver_name[] = "e1000";
 static char e1000_driver_string[] = "Intel(R) PRO/1000 Network Driver";
-- 
cgit v1.2.3