Home » Mailing list archives » linux-netdev » 2009 » October » 28

[net-next-2.6 PATCH 04/23] igb: use packet buffer sizes from RXPBS register

Previous thread: [PATCH 4/7] mlx4: Use bitmap_find_next_zero_area by Akinobu Mita on Wednesday, October 28, 2009 - 1:43 am. (1 message)

Next thread: Re: [PATCH net-next] bnx2x: Do Tx handling in a separate tasklet. by Vladislav Zolotarov on Wednesday, October 28, 2009 - 2:54 am. (2 messages)
[view original message]
From: Jeff Kirsher
Subject: [net-next-2.6 PATCH 01/23] igb: add support for seperate tx-usecs setting in ethtool
Date: Wednesday, October 28, 2009 - 2:45 am

From: Alexander Duyck <alexander.h.duyck@intel.com>

This patch adds support for a seperate tx-usecs interrupt moderation setting
in ethtool which is supported when tx and rx interrupt vectors are sperated.

Signed-off-by: Alexander Duyck <alexander.h.duyck@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
---

 drivers/net/igb/igb.h         |    5 +++-
 drivers/net/igb/igb_ethtool.c |   49 +++++++++++++++++++++++++++++++----------
 drivers/net/igb/igb_main.c    |   22 ++++++++++++------
 3 files changed, 54 insertions(+), 22 deletions(-)

diff --git a/drivers/net/igb/igb.h b/drivers/net/igb/igb.h
index 6c35c90..bef8cdc 100644
--- a/drivers/net/igb/igb.h
+++ b/drivers/net/igb/igb.h
@@ -249,8 +249,8 @@ struct igb_adapter {
 	unsigned int total_rx_bytes;
 	unsigned int total_rx_packets;
 	/* Interrupt Throttle Rate */
-	u32 itr;
-	u32 itr_setting;
+	u32 rx_itr_setting;
+	u32 tx_itr_setting;
 	u16 tx_itr;
 	u16 rx_itr;
 
@@ -321,6 +321,7 @@ struct igb_adapter {
 #define IGB_FLAG_HAS_MSI           (1 << 0)
 #define IGB_FLAG_DCA_ENABLED       (1 << 1)
 #define IGB_FLAG_QUAD_PORT_A       (1 << 2)
+#define IGB_FLAG_QUEUE_PAIRS       (1 << 3)
 
 enum e1000_state_t {
 	__IGB_TESTING,
diff --git a/drivers/net/igb/igb_ethtool.c b/drivers/net/igb/igb_ethtool.c
index 048a615..84fe25a 100644
--- a/drivers/net/igb/igb_ethtool.c
+++ b/drivers/net/igb/igb_ethtool.c
@@ -1827,18 +1827,37 @@ static int igb_set_coalesce(struct net_device *netdev,
 	    (ec->rx_coalesce_usecs == 2))
 		return -EINVAL;
 
+	if ((ec->tx_coalesce_usecs > IGB_MAX_ITR_USECS) ||
+	    ((ec->tx_coalesce_usecs > 3) &&
+	     (ec->tx_coalesce_usecs < IGB_MIN_ITR_USECS)) ||
+	    (ec->tx_coalesce_usecs == 2))
+		return -EINVAL;
+
+	if ((adapter->flags & IGB_FLAG_QUEUE_PAIRS) && ec->tx_coalesce_usecs)
+		return -EINVAL;
+
 	/* convert to rate of irq's per second */
-	if (ec->rx_coalesce_usecs && ec->rx_coalesce_usecs <= 3) {
-		adapter->itr_setting = ...
[ message continues ]
[view original message]
From: Jeff Kirsher
Subject: [net-next-2.6 PATCH 03/23] igb: misc cleanups within igb_ethtool.c
Date: Wednesday, October 28, 2009 - 2:46 am

From: Alexander Duyck <alexander.h.duyck@intel.com>

This patch just goes thorugh and does several cleanups on igb_ethtool.c.

Signed-off-by: Alexander Duyck <alexander.h.duyck@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
---

 drivers/net/igb/igb_ethtool.c |  105 +++++++++++++++++++++++------------------
 1 files changed, 60 insertions(+), 45 deletions(-)

diff --git a/drivers/net/igb/igb_ethtool.c b/drivers/net/igb/igb_ethtool.c
index 84fe25a..d24b902 100644
--- a/drivers/net/igb/igb_ethtool.c
+++ b/drivers/net/igb/igb_ethtool.c
@@ -101,24 +101,25 @@ static const struct igb_stats igb_gstrings_stats[] = {
 };
 
 #define IGB_QUEUE_STATS_LEN \
-	(((((struct igb_adapter *)netdev_priv(netdev))->num_rx_queues)* \
+	((((struct igb_adapter *)netdev_priv(netdev))->num_rx_queues * \
 	  (sizeof(struct igb_rx_queue_stats) / sizeof(u64))) + \
-	 ((((struct igb_adapter *)netdev_priv(netdev))->num_tx_queues) * \
+	 (((struct igb_adapter *)netdev_priv(netdev))->num_tx_queues * \
 	  (sizeof(struct igb_tx_queue_stats) / sizeof(u64))))
 #define IGB_GLOBAL_STATS_LEN	\
-	sizeof(igb_gstrings_stats) / sizeof(struct igb_stats)
+	(sizeof(igb_gstrings_stats) / sizeof(struct igb_stats))
 #define IGB_STATS_LEN (IGB_GLOBAL_STATS_LEN + IGB_QUEUE_STATS_LEN)
 static const char igb_gstrings_test[][ETH_GSTRING_LEN] = {
 	"Register test  (offline)", "Eeprom test    (offline)",
 	"Interrupt test (offline)", "Loopback test  (offline)",
 	"Link test   (on/offline)"
 };
-#define IGB_TEST_LEN sizeof(igb_gstrings_test) / ETH_GSTRING_LEN
+#define IGB_TEST_LEN (sizeof(igb_gstrings_test) / ETH_GSTRING_LEN)
 
 static int igb_get_settings(struct net_device *netdev, struct ethtool_cmd *ecmd)
 {
 	struct igb_adapter *adapter = netdev_priv(netdev);
 	struct e1000_hw *hw = &adapter->hw;
+	u32 status;
 
 	if (hw->phy.media_type == e1000_media_type_copper) {
 
@@ -153,17 +154,20 @@ static int igb_get_settings(struct net_device *netdev, struct ethtool_cmd *ecmd)
 
 	ecmd->transceiver = ...
[ message continues ]
[view original message]
From: Jeff Kirsher
Subject: [net-next-2.6 PATCH 02/23] igb: cleanup some of the code related to hw timestamping
Date: Wednesday, October 28, 2009 - 2:46 am

From: Alexander Duyck <alexander.h.duyck@intel.com>

The code for the hw timestamping is a bit bulky and making some of the
functions difficult to read.  In order to clean things up a bit I am moving
the timestamping operations into seperate functions.

Signed-off-by: Alexander Duyck <alexander.h.duyck@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
---

 drivers/net/igb/e1000_82575.h   |   14 +
 drivers/net/igb/e1000_defines.h |   33 +++
 drivers/net/igb/e1000_regs.h    |   65 +-----
 drivers/net/igb/igb.h           |    1 
 drivers/net/igb/igb_main.c      |  397 ++++++++++++++++++---------------------
 5 files changed, 242 insertions(+), 268 deletions(-)

diff --git a/drivers/net/igb/e1000_82575.h b/drivers/net/igb/e1000_82575.h
index 9418683..cbe4757 100644
--- a/drivers/net/igb/e1000_82575.h
+++ b/drivers/net/igb/e1000_82575.h
@@ -66,6 +66,8 @@ extern void igb_rx_fifo_flush_82575(struct e1000_hw *hw);
     E1000_EICR_RX_QUEUE3)
 
 /* Immediate Interrupt Rx (A.K.A. Low Latency Interrupt) */
+#define E1000_IMIREXT_SIZE_BP     0x00001000  /* Packet size bypass */
+#define E1000_IMIREXT_CTRL_BP     0x00080000  /* Bypass check of ctrl bits */
 
 /* Receive Descriptor - Advanced */
 union e1000_adv_rx_desc {
@@ -98,6 +100,7 @@ union e1000_adv_rx_desc {
 
 #define E1000_RXDADV_HDRBUFLEN_MASK      0x7FE0
 #define E1000_RXDADV_HDRBUFLEN_SHIFT     5
+#define E1000_RXDADV_STAT_TS             0x10000 /* Pkt was time stamped */
 
 /* Transmit Descriptor - Advanced */
 union e1000_adv_tx_desc {
@@ -167,6 +170,17 @@ struct e1000_adv_tx_context_desc {
 #define E1000_DCA_TXCTRL_CPUID_SHIFT 24 /* Tx CPUID now in the last byte */
 #define E1000_DCA_RXCTRL_CPUID_SHIFT 24 /* Rx CPUID now in the last byte */
 
+/* ETQF register bit definitions */
+#define E1000_ETQF_FILTER_ENABLE   (1 << 26)
+#define E1000_ETQF_1588            (1 << 30)
+
+/* FTQF register bit definitions */
+#define E1000_FTQF_VF_BP               0x00008000
+#define ...
[ message continues ]
[view original message]
From: Jeff Kirsher
Subject: [net-next-2.6 PATCH 04/23] igb: use packet buffer sizes from RXPBS register
Date: Wednesday, October 28, 2009 - 2:46 am

From: Alexander Duyck <alexander.h.duyck@intel.com>

This patch changes the configuration for 82576 so that it uses the actual
value of the 82576 rx packet buffer size instead of just assuming the
value.

Signed-off-by: Alexander Duyck <alexander.h.duyck@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
---

 drivers/net/igb/e1000_82575.h |    2 ++
 drivers/net/igb/e1000_regs.h  |    2 ++
 drivers/net/igb/igb_main.c    |    3 ++-
 3 files changed, 6 insertions(+), 1 deletions(-)

diff --git a/drivers/net/igb/e1000_82575.h b/drivers/net/igb/e1000_82575.h
index cbe4757..b3808ca 100644
--- a/drivers/net/igb/e1000_82575.h
+++ b/drivers/net/igb/e1000_82575.h
@@ -228,6 +228,8 @@ struct e1000_adv_tx_context_desc {
 
 #define ALL_QUEUES   0xFFFF
 
+/* RX packet buffer size defines */
+#define E1000_RXPBS_SIZE_MASK_82576  0x0000007F
 void igb_vmdq_set_loopback_pf(struct e1000_hw *, bool);
 void igb_vmdq_set_replication_pf(struct e1000_hw *, bool);
 
diff --git a/drivers/net/igb/e1000_regs.h b/drivers/net/igb/e1000_regs.h
index 24f2c24..934e03b 100644
--- a/drivers/net/igb/e1000_regs.h
+++ b/drivers/net/igb/e1000_regs.h
@@ -102,7 +102,9 @@
 #define E1000_ETQF(_n)  (0x05CB0 + (4 * (_n))) /* EType Queue Fltr */
 
 #define E1000_RQDPC(_n) (0x0C030 + ((_n) * 0x40))
+
 /* Split and Replication RX Control - RW */
+#define E1000_RXPBS    0x02404  /* Rx Packet Buffer Size - RW */
 /*
  * Convenience macros
  *
diff --git a/drivers/net/igb/igb_main.c b/drivers/net/igb/igb_main.c
index 7e628ba..2ed2694 100644
--- a/drivers/net/igb/igb_main.c
+++ b/drivers/net/igb/igb_main.c
@@ -1215,7 +1215,8 @@ void igb_reset(struct igb_adapter *adapter)
 	 */
 	switch (mac->type) {
 	case e1000_82576:
-		pba = E1000_PBA_64K;
+		pba = rd32(E1000_RXPBS);
+		pba &= E1000_RXPBS_SIZE_MASK_82576;
 		break;
 	case e1000_82575:
 	default:

--


[ message continues ]
[view original message]
From: Jeff Kirsher
Subject: [net-next-2.6 PATCH 05/23] igb: replace the VF clear_to_send with a flags value
Date: Wednesday, October 28, 2009 - 2:46 am

From: Alexander Duyck <alexander.h.duyck@intel.com>

In order to support future features it is easiest to replace the
clear_to_send boolean with a flag value.

Signed-off-by: Alexander Duyck <alexander.h.duyck@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
---

 drivers/net/igb/igb.h      |    5 +-
 drivers/net/igb/igb_main.c |  125 +++++++++++++++++++++++---------------------
 2 files changed, 70 insertions(+), 60 deletions(-)

diff --git a/drivers/net/igb/igb.h b/drivers/net/igb/igb.h
index 1a0ae57..7801d3f 100644
--- a/drivers/net/igb/igb.h
+++ b/drivers/net/igb/igb.h
@@ -73,9 +73,12 @@ struct vf_data_storage {
 	u16 vf_mc_hashes[IGB_MAX_VF_MC_ENTRIES];
 	u16 num_vf_mc_hashes;
 	u16 vlans_enabled;
-	bool clear_to_send;
+	u32 flags;
+	unsigned long last_nack;
 };
 
+#define IGB_VF_FLAG_CTS            0x00000001 /* VF is clear to send data */
+
 /* RX descriptor control thresholds.
  * PTHRESH - MAC will consider prefetch if it has fewer than this number of
  *           descriptors available in its onboard memory.
diff --git a/drivers/net/igb/igb_main.c b/drivers/net/igb/igb_main.c
index 2ed2694..a9e8ba2 100644
--- a/drivers/net/igb/igb_main.c
+++ b/drivers/net/igb/igb_main.c
@@ -125,9 +125,8 @@ static void igb_restore_vlan(struct igb_adapter *);
 static void igb_rar_set_qsel(struct igb_adapter *, u8 *, u32 , u8);
 static void igb_ping_all_vfs(struct igb_adapter *);
 static void igb_msg_task(struct igb_adapter *);
-static int igb_rcv_msg_from_vf(struct igb_adapter *, u32);
 static void igb_vmm_control(struct igb_adapter *);
-static int igb_set_vf_mac(struct igb_adapter *adapter, int, unsigned char *);
+static int igb_set_vf_mac(struct igb_adapter *, int, unsigned char *);
 static void igb_restore_vf_multicasts(struct igb_adapter *adapter);
 
 static inline void igb_set_vmolr(struct e1000_hw *hw, int vfn)
@@ -1291,10 +1290,10 @@ void igb_reset(struct igb_adapter *adapter)
 	if (adapter->vfs_allocated_count) {
 		int i;
 		for (i = 0 ; i < ...
[ message continues ]
[view original message]
From: Jeff Kirsher
Subject: [net-next-2.6 PATCH 06/23] igb: rework use of VMOLR in regards to PF and VFs
Date: Wednesday, October 28, 2009 - 2:47 am

From: Alexander Duyck <alexander.h.duyck@intel.com>

This patch updates the use of the VMOLR to include enabling multicast
promiscous for the VFs should they attempt to send over 30 multicast
addresses or if they use the new message type to enable multicast
promiscuous.

Signed-off-by: Alexander Duyck <alexander.h.duyck@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
---

 drivers/net/igb/e1000_mbx.h |   10 +-
 drivers/net/igb/igb.h       |    2 
 drivers/net/igb/igb_main.c  |  173 +++++++++++++++++++++++++++++--------------
 3 files changed, 125 insertions(+), 60 deletions(-)

diff --git a/drivers/net/igb/e1000_mbx.h b/drivers/net/igb/e1000_mbx.h
index ebc02ea..bb112fb 100644
--- a/drivers/net/igb/e1000_mbx.h
+++ b/drivers/net/igb/e1000_mbx.h
@@ -58,10 +58,12 @@
 #define E1000_VT_MSGINFO_MASK     (0xFF << E1000_VT_MSGINFO_SHIFT)
 
 #define E1000_VF_RESET            0x01 /* VF requests reset */
-#define E1000_VF_SET_MAC_ADDR     0x02 /* VF requests PF to set MAC addr */
-#define E1000_VF_SET_MULTICAST    0x03 /* VF requests PF to set MC addr */
-#define E1000_VF_SET_VLAN         0x04 /* VF requests PF to set VLAN */
-#define E1000_VF_SET_LPE          0x05 /* VF requests PF to set VMOLR.LPE */
+#define E1000_VF_SET_MAC_ADDR     0x02 /* VF requests to set MAC addr */
+#define E1000_VF_SET_MULTICAST    0x03 /* VF requests to set MC addr */
+#define E1000_VF_SET_VLAN         0x04 /* VF requests to set VLAN */
+#define E1000_VF_SET_LPE          0x05 /* VF requests to set VMOLR.LPE */
+#define E1000_VF_SET_PROMISC      0x06 /*VF requests to clear VMOLR.ROPE/MPME*/
+#define E1000_VF_SET_PROMISC_MULTICAST    (0x02 << E1000_VT_MSGINFO_SHIFT)
 
 #define E1000_PF_CONTROL_MSG      0x0100 /* PF control message */
 
diff --git a/drivers/net/igb/igb.h b/drivers/net/igb/igb.h
index 7801d3f..c27dc1a 100644
--- a/drivers/net/igb/igb.h
+++ b/drivers/net/igb/igb.h
@@ -78,6 +78,8 @@ struct vf_data_storage {
 };
 
 #define IGB_VF_FLAG_CTS            0x00000001 /* VF ...
[ message continues ]
[view original message]
From: Jeff Kirsher
Subject: [net-next-2.6 PATCH 07/23] igb: rework handling of the vfta and vlvf registers in relation to mng_vlan
Date: Wednesday, October 28, 2009 - 2:47 am

From: Alexander Duyck <alexander.h.duyck@intel.com>

This patch corrects some errors in how vlans are being handled when vfs
start interacting with the management vlans.

Signed-off-by: Alexander Duyck <alexander.h.duyck@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
---

 drivers/net/igb/igb_main.c |   94 +++++++++++++++-----------------------------
 1 files changed, 32 insertions(+), 62 deletions(-)

diff --git a/drivers/net/igb/igb_main.c b/drivers/net/igb/igb_main.c
index 3ab2978..9170927 100644
--- a/drivers/net/igb/igb_main.c
+++ b/drivers/net/igb/igb_main.c
@@ -963,24 +963,23 @@ static void igb_irq_enable(struct igb_adapter *adapter)
 
 static void igb_update_mng_vlan(struct igb_adapter *adapter)
 {
-	struct net_device *netdev = adapter->netdev;
+	struct e1000_hw *hw = &adapter->hw;
 	u16 vid = adapter->hw.mng_cookie.vlan_id;
 	u16 old_vid = adapter->mng_vlan_id;
-	if (adapter->vlgrp) {
-		if (!vlan_group_get_device(adapter->vlgrp, vid)) {
-			if (adapter->hw.mng_cookie.status &
-				E1000_MNG_DHCP_COOKIE_STATUS_VLAN) {
-				igb_vlan_rx_add_vid(netdev, vid);
-				adapter->mng_vlan_id = vid;
-			} else
-				adapter->mng_vlan_id = IGB_MNG_VLAN_NONE;
 
-			if ((old_vid != (u16)IGB_MNG_VLAN_NONE) &&
-					(vid != old_vid) &&
-			    !vlan_group_get_device(adapter->vlgrp, old_vid))
-				igb_vlan_rx_kill_vid(netdev, old_vid);
-		} else
-			adapter->mng_vlan_id = vid;
+	if (hw->mng_cookie.status & E1000_MNG_DHCP_COOKIE_STATUS_VLAN) {
+		/* add VID to filter table */
+		igb_vfta_set(hw, vid, true);
+		adapter->mng_vlan_id = vid;
+	} else {
+		adapter->mng_vlan_id = IGB_MNG_VLAN_NONE;
+	}
+
+	if ((old_vid != (u16)IGB_MNG_VLAN_NONE) &&
+	    (vid != old_vid) &&
+	    !vlan_group_get_device(adapter->vlgrp, old_vid)) {
+		/* remove VID from filter table */
+		igb_vfta_set(hw, old_vid, false);
 	}
 }
 
@@ -1847,11 +1846,6 @@ static int igb_open(struct net_device *netdev)
 
 	/* e1000_power_up_phy(adapter); */
 
-	adapter->mng_vlan_id = ...
[ message continues ]
[view original message]
From: Jeff Kirsher
Subject: [net-next-2.6 PATCH 08/23] igb: move vf init into a seperate function
Date: Wednesday, October 28, 2009 - 2:47 am

From: Alexander Duyck <alexander.h.duyck@intel.com>

This patch moves VF initialization into a seperate function to help improve
the readability of igb_probe.

Signed-off-by: Alexander Duyck <alexander.h.duyck@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
---

 drivers/net/igb/igb_main.c |   98 +++++++++++++++++++++++++-------------------
 1 files changed, 56 insertions(+), 42 deletions(-)

diff --git a/drivers/net/igb/igb_main.c b/drivers/net/igb/igb_main.c
index 9170927..54e8f02 100644
--- a/drivers/net/igb/igb_main.c
+++ b/drivers/net/igb/igb_main.c
@@ -1409,46 +1409,6 @@ static int __devinit igb_probe(struct pci_dev *pdev,
 	if (err)
 		goto err_sw_init;
 
-#ifdef CONFIG_PCI_IOV
-	/* since iov functionality isn't critical to base device function we
-	 * can accept failure.  If it fails we don't allow iov to be enabled */
-	if (hw->mac.type == e1000_82576) {
-		/* 82576 supports a maximum of 7 VFs in addition to the PF */
-		unsigned int num_vfs = (max_vfs > 7) ? 7 : max_vfs;
-		int i;
-		unsigned char mac_addr[ETH_ALEN];
-
-		if (num_vfs) {
-			adapter->vf_data = kcalloc(num_vfs,
-						sizeof(struct vf_data_storage),
-						GFP_KERNEL);
-			if (!adapter->vf_data) {
-				dev_err(&pdev->dev,
-				        "Could not allocate VF private data - "
-					"IOV enable failed\n");
-			} else {
-				err = pci_enable_sriov(pdev, num_vfs);
-				if (!err) {
-					adapter->vfs_allocated_count = num_vfs;
-					dev_info(&pdev->dev,
-					         "%d vfs allocated\n",
-					         num_vfs);
-					for (i = 0;
-					     i < adapter->vfs_allocated_count;
-					     i++) {
-						random_ether_addr(mac_addr);
-						igb_set_vf_mac(adapter, i,
-						               mac_addr);
-					}
-				} else {
-					kfree(adapter->vf_data);
-					adapter->vf_data = NULL;
-				}
-			}
-		}
-	}
-
-#endif
 	/* setup the private structure */
 	err = igb_sw_init(adapter);
 	if (err)
@@ -1772,6 +1732,54 @@ static void __devexit igb_remove(struct pci_dev ...
[ message continues ]
[view original message]
From: Jeff Kirsher
Subject: [net-next-2.6 PATCH 09/23] igb: only process global stats in igb_update_stats
Date: Wednesday, October 28, 2009 - 2:48 am

From: Alexander Duyck <alexander.h.duyck@intel.com>

This patch moves the update of adapter->net_stats.rx/tx values out of the
interrupt routine and into igb_update_stats by just adding together the
tx/rx byte/packet counts for the rings.

Signed-off-by: Alexander Duyck <alexander.h.duyck@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
---

 drivers/net/igb/igb_main.c |   61 +++++++++++++++++++++-----------------------
 1 files changed, 29 insertions(+), 32 deletions(-)

diff --git a/drivers/net/igb/igb_main.c b/drivers/net/igb/igb_main.c
index 54e8f02..62b6876 100644
--- a/drivers/net/igb/igb_main.c
+++ b/drivers/net/igb/igb_main.c
@@ -3769,7 +3769,10 @@ void igb_update_stats(struct igb_adapter *adapter)
 	struct net_device *netdev = adapter->netdev;
 	struct e1000_hw *hw = &adapter->hw;
 	struct pci_dev *pdev = adapter->pdev;
+	u32 rnbc;
 	u16 phy_tmp;
+	int i;
+	u64 bytes, packets;
 
 #define PHY_IDLE_ERROR_COUNT_MASK 0x00FF
 
@@ -3782,6 +3785,29 @@ void igb_update_stats(struct igb_adapter *adapter)
 	if (pci_channel_offline(pdev))
 		return;
 
+	bytes = 0;
+	packets = 0;
+	for (i = 0; i < adapter->num_rx_queues; i++) {
+		u32 rqdpc_tmp = rd32(E1000_RQDPC(i)) & 0x0FFF;
+		adapter->rx_ring[i].rx_stats.drops += rqdpc_tmp;
+		netdev->stats.rx_fifo_errors += rqdpc_tmp;
+		bytes += adapter->rx_ring[i].rx_stats.bytes;
+		packets += adapter->rx_ring[i].rx_stats.packets;
+	}
+
+	netdev->stats.rx_bytes = bytes;
+	netdev->stats.rx_packets = packets;
+
+	bytes = 0;
+	packets = 0;
+	for (i = 0; i < adapter->num_tx_queues; i++) {
+		bytes += adapter->tx_ring[i].tx_stats.bytes;
+		packets += adapter->tx_ring[i].tx_stats.packets;
+	}
+	netdev->stats.tx_bytes = bytes;
+	netdev->stats.tx_packets = packets;
+
+	/* read stats registers */
 	adapter->stats.crcerrs += rd32(E1000_CRCERRS);
 	adapter->stats.gprc += rd32(E1000_GPRC);
 	adapter->stats.gorc += rd32(E1000_GORCL);
@@ -3814,7 +3840,9 @@ void igb_update_stats(struct igb_adapter *adapter)
 ...
[ message continues ]
[view original message]
From: Jeff Kirsher
Subject: [net-next-2.6 PATCH 10/23] igb: move global_quad_port_a from global into local static define
Date: Wednesday, October 28, 2009 - 2:48 am

From: Alexander Duyck <alexander.h.duyck@intel.com>

This change moves global_quad_port_a into igb_probe as a static define
since it doesn't actually need to be global.

Signed-off-by: Alexander Duyck <alexander.h.duyck@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
---

 drivers/net/igb/igb_main.c |    8 ++------
 1 files changed, 2 insertions(+), 6 deletions(-)

diff --git a/drivers/net/igb/igb_main.c b/drivers/net/igb/igb_main.c
index 62b6876..7f32211 100644
--- a/drivers/net/igb/igb_main.c
+++ b/drivers/net/igb/igb_main.c
@@ -179,8 +179,6 @@ static struct pci_driver igb_driver = {
 	.err_handler = &igb_err_handler
 };
 
-static int global_quad_port_a; /* global quad port a indication */
-
 MODULE_AUTHOR("Intel Corporation, <e1000-devel@lists.sourceforge.net>");
 MODULE_DESCRIPTION("Intel(R) Gigabit Ethernet Network Driver");
 MODULE_LICENSE("GPL");
@@ -252,12 +250,9 @@ static int __init igb_init_module(void)
 
 	printk(KERN_INFO "%s\n", igb_copyright);
 
-	global_quad_port_a = 0;
-
 #ifdef CONFIG_IGB_DCA
 	dca_register_notify(&dca_notifier);
 #endif
-
 	ret = pci_register_driver(&igb_driver);
 	return ret;
 }
@@ -1318,10 +1313,11 @@ static int __devinit igb_probe(struct pci_dev *pdev,
 	struct net_device *netdev;
 	struct igb_adapter *adapter;
 	struct e1000_hw *hw;
+	u16 eeprom_data = 0;
+	static int global_quad_port_a; /* global quad port a indication */
 	const struct e1000_info *ei = igb_info_tbl[ent->driver_data];
 	unsigned long mmio_start, mmio_len;
 	int err, pci_using_dac;
-	u16 eeprom_data = 0;
 	u16 eeprom_apme_mask = IGB_EEPROM_APME;
 	u32 part_num;
 

--


[ message continues ]
[view original message]
From: Jeff Kirsher
Subject: [net-next-2.6 PATCH 11/23] igb: make tx hang check multiqueue, check eop descriptor
Date: Wednesday, October 28, 2009 - 2:48 am

From: Alexander Duyck <alexander.h.duyck@intel.com>

This change makes the tx hang check run over all tx queues instead of just
queue 0.  Also have hang display info on EOP descriptor instead of the
descriptor at the start of the chain.

Signed-off-by: Alexander Duyck <alexander.h.duyck@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
---

 drivers/net/igb/igb_main.c |   10 ++++++----
 1 files changed, 6 insertions(+), 4 deletions(-)

diff --git a/drivers/net/igb/igb_main.c b/drivers/net/igb/igb_main.c
index 7f32211..f75f90f 100644
--- a/drivers/net/igb/igb_main.c
+++ b/drivers/net/igb/igb_main.c
@@ -3002,6 +3002,10 @@ link_up:
 		}
 	}
 
+	/* Force detection of hung controller every watchdog period */
+	for (i = 0; i < adapter->num_tx_queues; i++)
+		adapter->tx_ring[i].detect_tx_hung = true;
+
 	/* Cause software interrupt to ensure rx ring is cleaned */
 	if (adapter->msix_entries) {
 		u32 eics = 0;
@@ -3014,9 +3018,6 @@ link_up:
 		wr32(E1000_ICS, E1000_ICS_RXDMT0);
 	}
 
-	/* Force detection of hung controller every watchdog period */
-	tx_ring->detect_tx_hung = true;
-
 	/* Reset the timer */
 	if (!test_bit(__IGB_DOWN, &adapter->state))
 		mod_timer(&adapter->watchdog_timer,
@@ -3667,6 +3668,7 @@ static void igb_tx_timeout(struct net_device *netdev)
 
 	/* Do the reset outside of interrupt context */
 	adapter->tx_timeout_count++;
+
 	schedule_work(&adapter->reset_task);
 	wr32(E1000_EICS,
 	     (adapter->eims_enable_mask & ~adapter->eims_other));
@@ -4804,7 +4806,7 @@ static bool igb_clean_tx_irq(struct igb_q_vector *q_vector)
 				readl(tx_ring->tail),
 				tx_ring->next_to_use,
 				tx_ring->next_to_clean,
-				tx_ring->buffer_info[i].time_stamp,
+				tx_ring->buffer_info[eop].time_stamp,
 				eop,
 				jiffies,
 				eop_desc->wb.status);

--


[ message continues ]
[view original message]
From: Jeff Kirsher
Subject: [net-next-2.6 PATCH 12/23] igb: cleanup code related to ring resource allocation and free
Date: Wednesday, October 28, 2009 - 2:49 am

From: Alexander Duyck <alexander.h.duyck@intel.com>

This patch cleans up some of the ring alloc and free code to better handle
exceptions such as attempting to free resources on an already freed ring.

Signed-off-by: Alexander Duyck <alexander.h.duyck@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
---

 drivers/net/igb/igb_main.c |   29 +++++++++++++++++------------
 1 files changed, 17 insertions(+), 12 deletions(-)

diff --git a/drivers/net/igb/igb_main.c b/drivers/net/igb/igb_main.c
index f75f90f..e67ff0e 100644
--- a/drivers/net/igb/igb_main.c
+++ b/drivers/net/igb/igb_main.c
@@ -1952,7 +1952,8 @@ int igb_setup_tx_resources(struct igb_ring *tx_ring)
 	tx_ring->size = tx_ring->count * sizeof(union e1000_adv_tx_desc);
 	tx_ring->size = ALIGN(tx_ring->size, 4096);
 
-	tx_ring->desc = pci_alloc_consistent(pdev, tx_ring->size,
+	tx_ring->desc = pci_alloc_consistent(pdev,
+	                                     tx_ring->size,
 					     &tx_ring->dma);
 
 	if (!tx_ring->desc)
@@ -1978,13 +1979,13 @@ err:
  **/
 static int igb_setup_all_tx_resources(struct igb_adapter *adapter)
 {
+	struct pci_dev *pdev = adapter->pdev;
 	int i, err = 0;
-	int r_idx;
 
 	for (i = 0; i < adapter->num_tx_queues; i++) {
 		err = igb_setup_tx_resources(&adapter->tx_ring[i]);
 		if (err) {
-			dev_err(&adapter->pdev->dev,
+			dev_err(&pdev->dev,
 				"Allocation for Tx Queue %u failed\n", i);
 			for (i--; i >= 0; i--)
 				igb_free_tx_resources(&adapter->tx_ring[i]);
@@ -1993,7 +1994,7 @@ static int igb_setup_all_tx_resources(struct igb_adapter *adapter)
 	}
 
 	for (i = 0; i < IGB_MAX_TX_QUEUES; i++) {
-		r_idx = i % adapter->num_tx_queues;
+		int r_idx = i % adapter->num_tx_queues;
 		adapter->multi_tx_table[i] = &adapter->tx_ring[r_idx];
 	}
 	return err;
@@ -2116,6 +2117,7 @@ int igb_setup_rx_resources(struct igb_ring *rx_ring)
 
 err:
 	vfree(rx_ring->buffer_info);
+	rx_ring->buffer_info = NULL;
 	dev_err(&pdev->dev, "Unable to allocate memory for "
 ...
[ message continues ]
[view original message]
From: Jeff Kirsher
Subject: [net-next-2.6 PATCH 13/23] igb: change queue ordering for 82576 based adapters
Date: Wednesday, October 28, 2009 - 2:49 am

From: Alexander Duyck <alexander.h.duyck@intel.com>

This patch changes the queue ordering for 82576 adapters so that if VFs are
enabled the queues will first be allocated out of the PF pool.  Any
remaining queues will be allocated out of other VMDq pools.

Signed-off-by: Alexander Duyck <alexander.h.duyck@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
---

 drivers/net/igb/igb_main.c |   25 +++++++++++++------------
 1 files changed, 13 insertions(+), 12 deletions(-)

diff --git a/drivers/net/igb/igb_main.c b/drivers/net/igb/igb_main.c
index e67ff0e..52c4302 100644
--- a/drivers/net/igb/igb_main.c
+++ b/drivers/net/igb/igb_main.c
@@ -285,7 +285,7 @@ module_exit(igb_exit_module);
  **/
 static void igb_cache_ring_register(struct igb_adapter *adapter)
 {
-	int i;
+	int i = 0, j = 0;
 	u32 rbase_offset = adapter->vfs_allocated_count;
 
 	switch (adapter->hw.mac.type) {
@@ -295,19 +295,20 @@ static void igb_cache_ring_register(struct igb_adapter *adapter)
 		 * In order to avoid collision we start at the first free queue
 		 * and continue consuming queues in the same sequence
 		 */
-		for (i = 0; i < adapter->num_rx_queues; i++)
-			adapter->rx_ring[i].reg_idx = rbase_offset +
-			                              Q_IDX_82576(i);
-		for (i = 0; i < adapter->num_tx_queues; i++)
-			adapter->tx_ring[i].reg_idx = rbase_offset +
-			                              Q_IDX_82576(i);
-		break;
+		if (adapter->vfs_allocated_count) {
+			for (; i < adapter->num_rx_queues; i++)
+				adapter->rx_ring[i].reg_idx = rbase_offset +
+				                              Q_IDX_82576(i);
+			for (; j < adapter->num_tx_queues; j++)
+				adapter->tx_ring[j].reg_idx = rbase_offset +
+				                              Q_IDX_82576(j);
+		}
 	case e1000_82575:
 	default:
-		for (i = 0; i < adapter->num_rx_queues; i++)
-			adapter->rx_ring[i].reg_idx = i;
-		for (i = 0; i < adapter->num_tx_queues; i++)
-			adapter->tx_ring[i].reg_idx = i;
+		for (; i < ...
[ message continues ]
[view original message]
From: Jeff Kirsher
Subject: [net-next-2.6 PATCH 14/23] igb: cleanup interrupt enablement in regards to msix_other
Date: Wednesday, October 28, 2009 - 2:49 am

From: Alexander Duyck <alexander.h.duyck@intel.com>

This patch changes a few things so that instead of firing a link status
interrupt directly the get_link_status bit is set and the watchdog is
scheduled.  In addition the mailbox bit is now only enabled if VFs are
enabled.

Signed-off-by: Alexander Duyck <alexander.h.duyck@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
---

 drivers/net/igb/igb_main.c |   31 +++++++++++++++++++++++--------
 1 files changed, 23 insertions(+), 8 deletions(-)

diff --git a/drivers/net/igb/igb_main.c b/drivers/net/igb/igb_main.c
index 52c4302..5b0f939 100644
--- a/drivers/net/igb/igb_main.c
+++ b/drivers/net/igb/igb_main.c
@@ -919,6 +919,11 @@ static void igb_irq_disable(struct igb_adapter *adapter)
 {
 	struct e1000_hw *hw = &adapter->hw;
 
+	/*
+	 * we need to be careful when disabling interrupts.  The VFs are also
+	 * mapped into these registers and so clearing the bits can cause
+	 * issues on the VF drivers so we only need to clear what we set
+	 */
 	if (adapter->msix_entries) {
 		u32 regval = rd32(E1000_EIAM);
 		wr32(E1000_EIAM, regval & ~adapter->eims_enable_mask);
@@ -942,15 +947,17 @@ static void igb_irq_enable(struct igb_adapter *adapter)
 	struct e1000_hw *hw = &adapter->hw;
 
 	if (adapter->msix_entries) {
+		u32 ims = E1000_IMS_LSC | E1000_IMS_DOUTSYNC;
 		u32 regval = rd32(E1000_EIAC);
 		wr32(E1000_EIAC, regval | adapter->eims_enable_mask);
 		regval = rd32(E1000_EIAM);
 		wr32(E1000_EIAM, regval | adapter->eims_enable_mask);
 		wr32(E1000_EIMS, adapter->eims_enable_mask);
-		if (adapter->vfs_allocated_count)
+		if (adapter->vfs_allocated_count) {
 			wr32(E1000_MBVFIMR, 0xFF);
-		wr32(E1000_IMS, (E1000_IMS_LSC | E1000_IMS_VMMB |
-		                 E1000_IMS_DOUTSYNC));
+			ims |= E1000_IMS_VMMB;
+		}
+		wr32(E1000_IMS, ims);
 	} else {
 		wr32(E1000_IMS, IMS_ENABLE_MASK);
 		wr32(E1000_IAM, IMS_ENABLE_MASK);
@@ -1091,8 +1098,10 @@ int igb_up(struct igb_adapter *adapter)
 
 ...
[ message continues ]
[view original message]
From: Jeff Kirsher
Subject: [net-next-2.6 PATCH 15/23] igb: Remove invalid stats counters
Date: Wednesday, October 28, 2009 - 2:50 am

From: Alexander Duyck <alexander.h.duyck@intel.com>

There are several counters being used like they are static when in fact
they are clear on read.  In order to prevent the values from being
incorrect I am removing the defunct counters.

Signed-off-by: Alexander Duyck <alexander.h.duyck@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
---

 drivers/net/igb/igb.h      |    6 ------
 drivers/net/igb/igb_main.c |   18 +-----------------
 2 files changed, 1 insertions(+), 23 deletions(-)

diff --git a/drivers/net/igb/igb.h b/drivers/net/igb/igb.h
index c27dc1a..b9fcfd3 100644
--- a/drivers/net/igb/igb.h
+++ b/drivers/net/igb/igb.h
@@ -269,10 +269,6 @@ struct igb_adapter {
 	/* TX */
 	struct igb_ring *tx_ring;      /* One per active queue */
 	unsigned long tx_queue_len;
-	u32 gotc;
-	u64 gotc_old;
-	u64 tpt_old;
-	u64 colc_old;
 	u32 tx_timeout_count;
 
 	/* RX */
@@ -280,8 +276,6 @@ struct igb_adapter {
 	int num_tx_queues;
 	int num_rx_queues;
 
-	u32 gorc;
-	u64 gorc_old;
 	u32 max_frame_size;
 	u32 min_frame_size;
 
diff --git a/drivers/net/igb/igb_main.c b/drivers/net/igb/igb_main.c
index 5b0f939..cb1acca 100644
--- a/drivers/net/igb/igb_main.c
+++ b/drivers/net/igb/igb_main.c
@@ -2925,9 +2925,6 @@ static void igb_watchdog_task(struct work_struct *work)
 	int i;
 
 	link = igb_has_link(adapter);
-	if ((netif_carrier_ok(netdev)) && link)
-		goto link_up;
-
 	if (link) {
 		if (!netif_carrier_ok(netdev)) {
 			u32 ctrl;
@@ -2990,20 +2987,8 @@ static void igb_watchdog_task(struct work_struct *work)
 		}
 	}
 
-link_up:
 	igb_update_stats(adapter);
-
-	hw->mac.tx_packet_delta = adapter->stats.tpt - adapter->tpt_old;
-	adapter->tpt_old = adapter->stats.tpt;
-	hw->mac.collision_delta = adapter->stats.colc - adapter->colc_old;
-	adapter->colc_old = adapter->stats.colc;
-
-	adapter->gorc = adapter->stats.gorc - adapter->gorc_old;
-	adapter->gorc_old = adapter->stats.gorc;
-	adapter->gotc = adapter->stats.gotc - ...
[ message continues ]
[view original message]
From: Jeff Kirsher
Subject: [net-next-2.6 PATCH 16/23] igb: cleanup igb.h header whitespace and some structure formatting
Date: Wednesday, October 28, 2009 - 2:50 am

From: Alexander Duyck <alexander.h.duyck@intel.com>

This patch changes the layout of the ring and adapter structs to fill a few
holes in the structure.  It also cleans up some whitespace and formatting
issues.

Signed-off-by: Alexander Duyck <alexander.h.duyck@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
---

 drivers/net/igb/igb.h |   12 ++++++------
 1 files changed, 6 insertions(+), 6 deletions(-)

diff --git a/drivers/net/igb/igb.h b/drivers/net/igb/igb.h
index b9fcfd3..3298f5a 100644
--- a/drivers/net/igb/igb.h
+++ b/drivers/net/igb/igb.h
@@ -185,16 +185,15 @@ struct igb_ring {
 	dma_addr_t dma;                /* phys address of the ring */
 	void *desc;                    /* descriptor ring memory */
 	unsigned int size;             /* length of desc. ring in bytes */
-	unsigned int count;            /* number of desc. in the ring */
+	u16 count;                     /* number of desc. in the ring */
 	u16 next_to_use;
 	u16 next_to_clean;
+	u8 queue_index;
+	u8 reg_idx;
 	void __iomem *head;
 	void __iomem *tail;
 	struct igb_buffer *buffer_info; /* array of buffer info structs */
 
-	u8 queue_index;
-	u8 reg_idx;
-
 	unsigned int total_bytes;
 	unsigned int total_packets;
 
@@ -249,6 +248,7 @@ struct igb_adapter {
 	u32 en_mng_pt;
 	u16 link_speed;
 	u16 link_duplex;
+
 	unsigned int total_tx_bytes;
 	unsigned int total_tx_packets;
 	unsigned int total_rx_bytes;
@@ -311,8 +311,8 @@ struct igb_adapter {
 	u32 eeprom_wol;
 
 	struct igb_ring *multi_tx_table[IGB_ABS_MAX_TX_QUEUES];
-	unsigned int tx_ring_count;
-	unsigned int rx_ring_count;
+	u16 tx_ring_count;
+	u16 rx_ring_count;
 	unsigned int vfs_allocated_count;
 	struct vf_data_storage *vf_data;
 };

--


[ message continues ]
[view original message]
From: Jeff Kirsher
Subject: [net-next-2.6 PATCH 17/23] igb: cleanup igb xmit frame path
Date: Wednesday, October 28, 2009 - 2:50 am

From: Alexander Duyck <alexander.h.duyck@intel.com>

This patch cleans up the xmit frame path for igb to better handle xmit
frame errors and avoid null pointer exceptions.  It also cleans up some
whitespace issues found in the xmit frame path.

Signed-off-by: Alexander Duyck <alexander.h.duyck@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
---

 drivers/net/igb/igb_main.c |   30 +++++++++++++++---------------
 1 files changed, 15 insertions(+), 15 deletions(-)

diff --git a/drivers/net/igb/igb_main.c b/drivers/net/igb/igb_main.c
index cb1acca..8f8b7cc 100644
--- a/drivers/net/igb/igb_main.c
+++ b/drivers/net/igb/igb_main.c
@@ -3245,9 +3245,9 @@ set_itr_now:
 #define IGB_TX_FLAGS_VLAN		0x00000002
 #define IGB_TX_FLAGS_TSO		0x00000004
 #define IGB_TX_FLAGS_IPV4		0x00000008
-#define IGB_TX_FLAGS_TSTAMP             0x00000010
-#define IGB_TX_FLAGS_VLAN_MASK	0xffff0000
-#define IGB_TX_FLAGS_VLAN_SHIFT	16
+#define IGB_TX_FLAGS_TSTAMP		0x00000010
+#define IGB_TX_FLAGS_VLAN_MASK		0xffff0000
+#define IGB_TX_FLAGS_VLAN_SHIFT		        16
 
 static inline int igb_tso_adv(struct igb_ring *tx_ring,
 			      struct sk_buff *skb, u32 tx_flags, u8 *hdr_len)
@@ -3346,6 +3346,7 @@ static inline bool igb_tx_csum_adv(struct igb_ring *tx_ring,
 
 		if (tx_flags & IGB_TX_FLAGS_VLAN)
 			info |= (tx_flags & IGB_TX_FLAGS_VLAN_MASK);
+
 		info |= (skb_network_offset(skb) << E1000_ADVTXD_MACLEN_SHIFT);
 		if (skb->ip_summed == CHECKSUM_PARTIAL)
 			info |= skb_network_header_len(skb);
@@ -3462,17 +3463,17 @@ static inline int igb_tx_map_adv(struct igb_ring *tx_ring, struct sk_buff *skb,
 	tx_ring->buffer_info[i].skb = skb;
 	tx_ring->buffer_info[first].next_to_watch = i;
 
-	return count + 1;
+	return ++count;
 }
 
 static inline void igb_tx_queue_adv(struct igb_ring *tx_ring,
 				    int tx_flags, int count, u32 paylen,
 				    u8 hdr_len)
 {
-	union e1000_adv_tx_desc *tx_desc = NULL;
+	union e1000_adv_tx_desc *tx_desc;
 	struct igb_buffer *buffer_info;
 ...
[ message continues ]
[view original message]
From: Jeff Kirsher
Subject: [net-next-2.6 PATCH 18/23] igb: cleanup clean_rx_irq_adv and alloc_rx_buffers_adv
Date: Wednesday, October 28, 2009 - 2:51 am

From: Alexander Duyck <alexander.h.duyck@intel.com>

This patch cleans up some whitespace issues in clean_rx_irq_adv.  It also
adds NUMA aware page allocation and dma error handling to
alloc_rx_buffers_adv.

Signed-off-by: Alexander Duyck <alexander.h.duyck@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
---

 drivers/net/igb/igb_main.c |   24 +++++++++++++++++++-----
 1 files changed, 19 insertions(+), 5 deletions(-)

diff --git a/drivers/net/igb/igb_main.c b/drivers/net/igb/igb_main.c
index 8f8b7cc..d3e8316 100644
--- a/drivers/net/igb/igb_main.c
+++ b/drivers/net/igb/igb_main.c
@@ -4952,6 +4952,7 @@ static bool igb_clean_rx_irq_adv(struct igb_q_vector *q_vector,
 		i++;
 		if (i == rx_ring->count)
 			i = 0;
+
 		next_rxd = E1000_RX_DESC_ADV(*rx_ring, i);
 		prefetch(next_rxd);
 		next_buffer = &rx_ring->buffer_info[i];
@@ -4989,7 +4990,6 @@ static bool igb_clean_rx_irq_adv(struct igb_q_vector *q_vector,
 
 			skb->len += length;
 			skb->data_len += length;
-
 			skb->truesize += length;
 		}
 
@@ -5071,7 +5071,7 @@ void igb_alloc_rx_buffers_adv(struct igb_ring *rx_ring, int cleaned_count)
 
 		if ((bufsz < IGB_RXBUFFER_1024) && !buffer_info->page_dma) {
 			if (!buffer_info->page) {
-				buffer_info->page = alloc_page(GFP_ATOMIC);
+				buffer_info->page = netdev_alloc_page(netdev);
 				if (!buffer_info->page) {
 					rx_ring->rx_stats.alloc_failed++;
 					goto no_buffers;
@@ -5085,9 +5085,16 @@ void igb_alloc_rx_buffers_adv(struct igb_ring *rx_ring, int cleaned_count)
 					     buffer_info->page_offset,
 					     PAGE_SIZE / 2,
 					     PCI_DMA_FROMDEVICE);
+			if (pci_dma_mapping_error(rx_ring->pdev,
+			                          buffer_info->page_dma)) {
+				buffer_info->page_dma = 0;
+				rx_ring->rx_stats.alloc_failed++;
+				goto no_buffers;
+			}
 		}
 
-		if (!buffer_info->skb) {
+		skb = buffer_info->skb;
+		if (!skb) {
 			skb = netdev_alloc_skb_ip_align(netdev, bufsz);
 			if (!skb) {
 ...
[ message continues ]
[view original message]
From: Jeff Kirsher
Subject: [net-next-2.6 PATCH 19/23] igb: replace unecessary &adapter->hw with just hw where applicable
Date: Wednesday, October 28, 2009 - 2:51 am

From: Alexander Duyck <alexander.h.duyck@intel.com>

This patch just cleans up some unecessary references to the adapter->hw
member when it has already been placed in a local variable named hw.

Signed-off-by: Alexander Duyck <alexander.h.duyck@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
---

 drivers/net/igb/igb_main.c |   26 ++++++++++++--------------
 1 files changed, 12 insertions(+), 14 deletions(-)

diff --git a/drivers/net/igb/igb_main.c b/drivers/net/igb/igb_main.c
index d3e8316..b2c0c97 100644
--- a/drivers/net/igb/igb_main.c
+++ b/drivers/net/igb/igb_main.c
@@ -1107,8 +1107,8 @@ int igb_up(struct igb_adapter *adapter)
 
 void igb_down(struct igb_adapter *adapter)
 {
-	struct e1000_hw *hw = &adapter->hw;
 	struct net_device *netdev = adapter->netdev;
+	struct e1000_hw *hw = &adapter->hw;
 	u32 tctl, rctl;
 	int i;
 
@@ -1271,10 +1271,10 @@ void igb_reset(struct igb_adapter *adapter)
 	}
 
 	/* Allow time for pending master requests to run */
-	adapter->hw.mac.ops.reset_hw(&adapter->hw);
+	hw->mac.ops.reset_hw(hw);
 	wr32(E1000_WUC, 0);
 
-	if (adapter->hw.mac.ops.init_hw(&adapter->hw))
+	if (hw->mac.ops.init_hw(hw))
 		dev_err(&adapter->pdev->dev, "Hardware Error\n");
 
 	igb_update_mng_vlan(adapter);
@@ -1282,8 +1282,8 @@ void igb_reset(struct igb_adapter *adapter)
 	/* Enable h/w to recognize an 802.1Q VLAN Ethernet packet */
 	wr32(E1000_VET, ETHERNET_IEEE_VLAN_TYPE);
 
-	igb_reset_adaptive(&adapter->hw);
-	igb_get_phy_info(&adapter->hw);
+	igb_reset_adaptive(hw);
+	igb_get_phy_info(hw);
 }
 
 static const struct net_device_ops igb_netdev_ops = {
@@ -1404,8 +1404,6 @@ static int __devinit igb_probe(struct pci_dev *pdev,
 	hw->subsystem_vendor_id = pdev->subsystem_vendor;
 	hw->subsystem_device_id = pdev->subsystem_device;
 
-	/* setup the private structure */
-	hw->back = adapter;
 	/* Copy the default MAC, PHY and NVM function pointers */
 	memcpy(&hw->mac.ops, ei->mac_ops, sizeof(hw->mac.ops));
 ...
[ message continues ]
[view original message]
From: Jeff Kirsher
Subject: [net-next-2.6 PATCH 20/23] igb: add pci_dev in few spots to clean up use of dev_err/info/warn
Date: Wednesday, October 28, 2009 - 2:51 am

From: Alexander Duyck <alexander.h.duyck@intel.com>

This patch relpaces several references to adapter->pdev->dev with just
pdev->dev.  This allows for cleanup of several multiline dev_err/info
calls.

Signed-off-by: Alexander Duyck <alexander.h.duyck@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
---

 drivers/net/igb/igb_main.c |   24 ++++++++++++------------
 1 files changed, 12 insertions(+), 12 deletions(-)

diff --git a/drivers/net/igb/igb_main.c b/drivers/net/igb/igb_main.c
index b2c0c97..264ff00 100644
--- a/drivers/net/igb/igb_main.c
+++ b/drivers/net/igb/igb_main.c
@@ -1173,6 +1173,7 @@ void igb_reinit_locked(struct igb_adapter *adapter)
 
 void igb_reset(struct igb_adapter *adapter)
 {
+	struct pci_dev *pdev = adapter->pdev;
 	struct e1000_hw *hw = &adapter->hw;
 	struct e1000_mac_info *mac = &hw->mac;
 	struct e1000_fc_info *fc = &hw->fc;
@@ -1275,7 +1276,7 @@ void igb_reset(struct igb_adapter *adapter)
 	wr32(E1000_WUC, 0);
 
 	if (hw->mac.ops.init_hw(hw))
-		dev_err(&adapter->pdev->dev, "Hardware Error\n");
+		dev_err(&pdev->dev, "Hardware Error\n");
 
 	igb_update_mng_vlan(adapter);
 
@@ -3704,17 +3705,18 @@ static struct net_device_stats *igb_get_stats(struct net_device *netdev)
 static int igb_change_mtu(struct net_device *netdev, int new_mtu)
 {
 	struct igb_adapter *adapter = netdev_priv(netdev);
+	struct pci_dev *pdev = adapter->pdev;
 	int max_frame = new_mtu + ETH_HLEN + ETH_FCS_LEN;
 	u32 rx_buffer_len, i;
 
 	if ((max_frame < ETH_ZLEN + ETH_FCS_LEN) ||
 	    (max_frame > MAX_JUMBO_FRAME_SIZE)) {
-		dev_err(&adapter->pdev->dev, "Invalid MTU setting\n");
+		dev_err(&pdev->dev, "Invalid MTU setting\n");
 		return -EINVAL;
 	}
 
 	if (max_frame > MAX_STD_JUMBO_FRAME_SIZE) {
-		dev_err(&adapter->pdev->dev, "MTU > 9216 not supported.\n");
+		dev_err(&pdev->dev, "MTU > 9216 not supported.\n");
 		return -EINVAL;
 	}
 
@@ -3739,7 +3741,7 @@ static int igb_change_mtu(struct net_device *netdev, int new_mtu)
 	if ...
[ message continues ]
[view original message]
From: Jeff Kirsher
Subject: [net-next-2.6 PATCH 21/23] igb: limit minimum mtu to 68 to keep ip bound to interface
Date: Wednesday, October 28, 2009 - 2:52 am

From: Alexander Duyck <alexander.h.duyck@intel.com>

Limit the minimum mtu to 68 in order to prevent ip from being unbound from
the interface.

Signed-off-by: Alexander Duyck <alexander.h.duyck@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
---

 drivers/net/igb/igb_main.c |    3 +--
 1 files changed, 1 insertions(+), 2 deletions(-)

diff --git a/drivers/net/igb/igb_main.c b/drivers/net/igb/igb_main.c
index 264ff00..846e64f 100644
--- a/drivers/net/igb/igb_main.c
+++ b/drivers/net/igb/igb_main.c
@@ -3709,8 +3709,7 @@ static int igb_change_mtu(struct net_device *netdev, int new_mtu)
 	int max_frame = new_mtu + ETH_HLEN + ETH_FCS_LEN;
 	u32 rx_buffer_len, i;
 
-	if ((max_frame < ETH_ZLEN + ETH_FCS_LEN) ||
-	    (max_frame > MAX_JUMBO_FRAME_SIZE)) {
+	if ((new_mtu < 68) || (max_frame > MAX_JUMBO_FRAME_SIZE)) {
 		dev_err(&pdev->dev, "Invalid MTU setting\n");
 		return -EINVAL;
 	}

--


[ message continues ]
[view original message]
From: Jeff Kirsher
Subject: [net-next-2.6 PATCH 22/23] igb: open up SCTP checksum offloads to all MACs 82576 and newer
Date: Wednesday, October 28, 2009 - 2:52 am

From: Alexander Duyck <alexander.h.duyck@intel.com>

Going forward the plan is to have the MACs support SCTP checksum offloads
so change the check from == to >=.

Signed-off-by: Alexander Duyck <alexander.h.duyck@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
---

 drivers/net/igb/igb_main.c |    2 +-
 1 files changed, 1 insertions(+), 1 deletions(-)

diff --git a/drivers/net/igb/igb_main.c b/drivers/net/igb/igb_main.c
index 846e64f..1a6c074 100644
--- a/drivers/net/igb/igb_main.c
+++ b/drivers/net/igb/igb_main.c
@@ -1456,7 +1456,7 @@ static int __devinit igb_probe(struct pci_dev *pdev,
 	if (pci_using_dac)
 		netdev->features |= NETIF_F_HIGHDMA;
 
-	if (adapter->hw.mac.type == e1000_82576)
+	if (hw->mac.type >= e1000_82576)
 		netdev->features |= NETIF_F_SCTP_CSUM;
 
 	adapter->en_mng_pt = igb_enable_mng_pass_thru(hw);

--


[ message continues ]
[view original message]
From: Jeff Kirsher
Subject: [net-next-2.6 PATCH 23/23] igb: cleanup whitespace issues in igb_main.c
Date: Wednesday, October 28, 2009 - 2:52 am

From: Alexander Duyck <alexander.h.duyck@intel.com>

This patch goes through and cleans up whitespace issues in igb_main.c
to help improve readability.

Signed-off-by: Alexander Duyck <alexander.h.duyck@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
---

 drivers/net/igb/igb_main.c |   41 +++++++++++++++++++----------------------
 1 files changed, 19 insertions(+), 22 deletions(-)

diff --git a/drivers/net/igb/igb_main.c b/drivers/net/igb/igb_main.c
index 1a6c074..b044c98 100644
--- a/drivers/net/igb/igb_main.c
+++ b/drivers/net/igb/igb_main.c
@@ -1006,7 +1006,6 @@ static void igb_release_hw_control(struct igb_adapter *adapter)
 			ctrl_ext & ~E1000_CTRL_EXT_DRV_LOAD);
 }
 
-
 /**
  * igb_get_hw_control - get control of the h/w from f/w
  * @adapter: address of board private structure
@@ -1067,7 +1066,6 @@ static void igb_configure(struct igb_adapter *adapter)
  * igb_up - Open the interface and prepare it to handle traffic
  * @adapter: board private structure
  **/
-
 int igb_up(struct igb_adapter *adapter)
 {
 	struct e1000_hw *hw = &adapter->hw;
@@ -1288,7 +1286,7 @@ void igb_reset(struct igb_adapter *adapter)
 }
 
 static const struct net_device_ops igb_netdev_ops = {
-	.ndo_open 		= igb_open,
+	.ndo_open		= igb_open,
 	.ndo_stop		= igb_close,
 	.ndo_start_xmit		= igb_xmit_frame_adv,
 	.ndo_get_stats		= igb_get_stats,
@@ -1444,7 +1442,6 @@ static int __devinit igb_probe(struct pci_dev *pdev,
 	netdev->features |= NETIF_F_IPV6_CSUM;
 	netdev->features |= NETIF_F_TSO;
 	netdev->features |= NETIF_F_TSO6;
-
 	netdev->features |= NETIF_F_GRO;
 
 	netdev->vlan_features |= NETIF_F_TSO;
@@ -1569,7 +1566,6 @@ static int __devinit igb_probe(struct pci_dev *pdev,
 	}
 
 #endif
-
 	switch (hw->mac.type) {
 	case e1000_82576:
 		/*
@@ -1624,8 +1620,8 @@ static int __devinit igb_probe(struct pci_dev *pdev,
 	/* print bus type/speed/width info */
 	dev_info(&pdev->dev, "%s: (PCIe:%s:%s) %pM\n",
 		 netdev->name,
-		 ((hw->bus.speed == ...
[ message continues ]
[view original message]
From: David Miller
Subject: Re: [net-next-2.6 PATCH 01/23] igb: add support for seperate tx-usecs setting in ethtool
Date: Wednesday, October 28, 2009 - 3:39 am

All applied to net-next-2.6, but then I had to add the following
patch to kill a warning:

igb: Fix warnings in igb_set_ringparam()

drivers/net/igb/igb_ethtool.c: In function ‘igb_set_ringparam’:
drivers/net/igb/igb_ethtool.c:744: warning: comparison of distinct pointer types lacks a cast
drivers/net/igb/igb_ethtool.c:748: warning: comparison of distinct pointer types lacks a cast

Casts were to u16 on the constant, but the type of new_{r,t}x_count is
u32.  Cast to u32 instead.

Signed-off-by: David S. Miller <davem@davemloft.net>
---
 drivers/net/igb/igb_ethtool.c |    4 ++--
 1 files changed, 2 insertions(+), 2 deletions(-)

diff --git a/drivers/net/igb/igb_ethtool.c b/drivers/net/igb/igb_ethtool.c
index d24b902..90b89a8 100644
--- a/drivers/net/igb/igb_ethtool.c
+++ b/drivers/net/igb/igb_ethtool.c
@@ -741,11 +741,11 @@ static int igb_set_ringparam(struct net_device *netdev,
 		return -EINVAL;
 
 	new_rx_count = min(ring->rx_pending, (u32)IGB_MAX_RXD);
-	new_rx_count = max(new_rx_count, (u16)IGB_MIN_RXD);
+	new_rx_count = max(new_rx_count, (u32)IGB_MIN_RXD);
 	new_rx_count = ALIGN(new_rx_count, REQ_RX_DESCRIPTOR_MULTIPLE);
 
 	new_tx_count = min(ring->tx_pending, (u32)IGB_MAX_TXD);
-	new_tx_count = max(new_tx_count, (u16)IGB_MIN_TXD);
+	new_tx_count = max(new_tx_count, (u32)IGB_MIN_TXD);
 	new_tx_count = ALIGN(new_tx_count, REQ_TX_DESCRIPTOR_MULTIPLE);
 
 	if ((new_tx_count == adapter->tx_ring_count) &&
-- 
1.6.5.1


[ message continues ]
[view original message]
From: Stephen Hemminger
Subject: Re: [net-next-2.6 PATCH 01/23] igb: add support for seperate tx-usecs setting in ethtool
Date: Wednesday, October 28, 2009 - 8:42 am

On Wed, 28 Oct 2009 03:39:22 -0700 (PDT)

new_rx_count = max_t(u32, new_rx_count, IGB_MIN_RXD)
  is slightly cleaner (hides cast)

-- 
--


[ message continues ]
Previous thread: [PATCH 4/7] mlx4: Use bitmap_find_next_zero_area by Akinobu Mita on Wednesday, October 28, 2009 - 1:43 am. (1 message)

Next thread: Re: [PATCH net-next] bnx2x: Do Tx handling in a separate tasklet. by Vladislav Zolotarov on Wednesday, October 28, 2009 - 2:54 am. (2 messages)