[net-next 00/14][pull request] Intel Wired LAN Driver Updates

[net-next 00/14][pull request] Intel Wired LAN Driver Updates

[Jeff Kirsher]

This series contains updates to ixgbe and igb.

The following are changes since commit db0fe0b2f6bba2fda939737d063db2ae14c58d71:
  Merge tag 'batman-adv-fix-for-davem' of git://git.open-mesh.org/linux-merge
and are available in the git repository at:
  git://git.kernel.org/pub/scm/linux/kernel/git/jkirsher/net-next master

Alexander Duyck (13):
  ixgbe: Initialize q_vector cpu and affinity masks correctly
  ixgbe: Enable jumbo frames support w/ SR-IOV
  ixgbe: Move message handling routines into their own functions
  ixgbe: Add mailbox API version negotiation support to ixgbe PF
  igb: Split Rx timestamping into two separate functions
  igb: Do not use header split, instead receive all frames into a
    single buffer
  igb: Combine post-processing of skb into a single function
  igb: Map entire page and sync half instead of mapping and unmapping
    half pages
  igb: Move rx_buffer related code in Rx cleanup path into separate
    function
  igb: Lock buffer size at 2K even on systems with larger pages
  igb: Combine q_vector and ring allocation into a single function
  igb: Move the calls to set the Tx and Rx queues into igb_open
  igb: Split igb_update_dca into separate Tx and Rx functions

Tushar Dave (1):
  igb: Correcting and improving small packet check and padding

 drivers/net/ethernet/intel/igb/e1000_82575.h   |    3 +
 drivers/net/ethernet/intel/igb/igb.h           |   70 +-
 drivers/net/ethernet/intel/igb/igb_ethtool.c   |   53 +-
 drivers/net/ethernet/intel/igb/igb_main.c      | 1242 +++++++++++++++---------
 drivers/net/ethernet/intel/igb/igb_ptp.c       |   55 +-
 drivers/net/ethernet/intel/ixgbe/ixgbe.h       |    1 +
 drivers/net/ethernet/intel/ixgbe/ixgbe_fcoe.c  |    4 +
 drivers/net/ethernet/intel/ixgbe/ixgbe_lib.c   |    7 +-
 drivers/net/ethernet/intel/ixgbe/ixgbe_main.c  |   13 +-
 drivers/net/ethernet/intel/ixgbe/ixgbe_mbx.h   |   21 +-
 drivers/net/ethernet/intel/ixgbe/ixgbe_sriov.c |  359 ++++---
 11 files changed, 1195 insertions(+), 633 deletions(-)

-- 
1.7.11.7

[net-next 01/14] ixgbe: Initialize q_vector cpu and affinity masks correctly

[Jeff Kirsher]

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

When enabling DCB the rings belonging to a q_vector on CPU 0 were not
reinitializing their DCA registers.  Upon closer inspection the issue was
that the q_vector CPU variable was left at 0 resulting in the driver not
updating the DCA registers.

In order to guarantee the DCA registers will be updated I am adding a
couple line change so that we initialize the CPU variable to -1 which will
force a DCA update the first time an interrupt fires on that q_vector.

In addition we were setting the CPU affinity hint to all CPUs when we were
not specifying a CPU.  Instead we should leave it as all zeros to avoid any
possible confusion about the fact that we shouldn't be giving a hint.

Signed-off-by: Alexander Duyck <alexander.h.duyck@intel.com>
Tested-by: Phil Schmitt <phillip.j.schmitt@intel.com>
Tested-by: Ross Brattain <ross.b.brattain@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
---
 drivers/net/ethernet/intel/ixgbe/ixgbe_lib.c | 7 +++++--
 1 file changed, 5 insertions(+), 2 deletions(-)

diff --git a/drivers/net/ethernet/intel/ixgbe/ixgbe_lib.c b/drivers/net/ethernet/intel/ixgbe/ixgbe_lib.c
index 17ecbce..5e508b6 100644
--- a/drivers/net/ethernet/intel/ixgbe/ixgbe_lib.c
+++ b/drivers/net/ethernet/intel/ixgbe/ixgbe_lib.c
@@ -802,10 +802,13 @@ static int ixgbe_alloc_q_vector(struct ixgbe_adapter *adapter,
 	/* setup affinity mask and node */
 	if (cpu != -1)
 		cpumask_set_cpu(cpu, &q_vector->affinity_mask);
-	else
-		cpumask_copy(&q_vector->affinity_mask, cpu_online_mask);
 	q_vector->numa_node = node;
 
+#ifdef CONFIG_IXGBE_DCA
+	/* initialize CPU for DCA */
+	q_vector->cpu = -1;
+
+#endif
 	/* initialize NAPI */
 	netif_napi_add(adapter->netdev, &q_vector->napi,
 		       ixgbe_poll, 64);
-- 
1.7.11.7

[net-next 02/14] ixgbe: Enable jumbo frames support w/ SR-IOV

[Jeff Kirsher]

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

This change makes it so that we can have limited support for jumbo frames
when SR-IOV is enabled.  In order to accomplish this it is necessary to
disable all VFs when the PF has jumbo frames enabled.  If the VFs then
request the same maximum frame size as the PF they will be re-enabled.  A
follow on patch will add a means of identifying when a VF can support
spanning buffers and does not need to be worried about the actual supported
max frame size.

Signed-off-by: Alexander Duyck <alexander.h.duyck@intel.com>
Tested-by: Ross Brattain <ross.b.brattain@intel.com>
Tested-by: Robert Garrett <robertx.e.garrett@intel.com>
Tested-by: Sibai Li <Sibai.li@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
---
 drivers/net/ethernet/intel/ixgbe/ixgbe_fcoe.c  |   4 +
 drivers/net/ethernet/intel/ixgbe/ixgbe_main.c  |  13 ++-
 drivers/net/ethernet/intel/ixgbe/ixgbe_sriov.c | 107 ++++++++++++++++++-------
 3 files changed, 93 insertions(+), 31 deletions(-)

diff --git a/drivers/net/ethernet/intel/ixgbe/ixgbe_fcoe.c b/drivers/net/ethernet/intel/ixgbe/ixgbe_fcoe.c
index ae73ef1..252850d 100644
--- a/drivers/net/ethernet/intel/ixgbe/ixgbe_fcoe.c
+++ b/drivers/net/ethernet/intel/ixgbe/ixgbe_fcoe.c
@@ -800,6 +800,10 @@ int ixgbe_fcoe_enable(struct net_device *netdev)
 		return -EINVAL;
 
 	e_info(drv, "Enabling FCoE offload features.\n");
+
+	if (adapter->flags & IXGBE_FLAG_SRIOV_ENABLED)
+		e_warn(probe, "Enabling FCoE on PF will disable legacy VFs\n");
+
 	if (netif_running(netdev))
 		netdev->netdev_ops->ndo_stop(netdev);
 
diff --git a/drivers/net/ethernet/intel/ixgbe/ixgbe_main.c b/drivers/net/ethernet/intel/ixgbe/ixgbe_main.c
index fa3d552..e2a6691 100644
--- a/drivers/net/ethernet/intel/ixgbe/ixgbe_main.c
+++ b/drivers/net/ethernet/intel/ixgbe/ixgbe_main.c
@@ -3263,6 +3263,11 @@ static void ixgbe_set_rx_buffer_len(struct ixgbe_adapter *adapter)
 		max_frame = IXGBE_FCOE_JUMBO_FRAME_SIZE;
 
 #endif /* IXGBE_FCOE */
+
+	/* adjust max frame to be at least the size of a standard frame */
+	if (max_frame < (ETH_FRAME_LEN + ETH_FCS_LEN))
+		max_frame = (ETH_FRAME_LEN + ETH_FCS_LEN);
+
 	mhadd = IXGBE_READ_REG(hw, IXGBE_MHADD);
 	if (max_frame != (mhadd >> IXGBE_MHADD_MFS_SHIFT)) {
 		mhadd &= ~IXGBE_MHADD_MFS_MASK;
@@ -4828,14 +4833,14 @@ static int ixgbe_change_mtu(struct net_device *netdev, int new_mtu)
 		return -EINVAL;
 
 	/*
-	 * For 82599EB we cannot allow PF to change MTU greater than 1500
-	 * in SR-IOV mode as it may cause buffer overruns in guest VFs that
-	 * don't allocate and chain buffers correctly.
+	 * For 82599EB we cannot allow legacy VFs to enable their receive
+	 * paths when MTU greater than 1500 is configured.  So display a
+	 * warning that legacy VFs will be disabled.
 	 */
 	if ((adapter->flags & IXGBE_FLAG_SRIOV_ENABLED) &&
 	    (adapter->hw.mac.type == ixgbe_mac_82599EB) &&
 	    (max_frame > MAXIMUM_ETHERNET_VLAN_SIZE))
-			return -EINVAL;
+		e_warn(probe, "Setting MTU > 1500 will disable legacy VFs\n");
 
 	e_info(probe, "changing MTU from %d to %d\n", netdev->mtu, new_mtu);
 
diff --git a/drivers/net/ethernet/intel/ixgbe/ixgbe_sriov.c b/drivers/net/ethernet/intel/ixgbe/ixgbe_sriov.c
index dce48bf..420766e 100644
--- a/drivers/net/ethernet/intel/ixgbe/ixgbe_sriov.c
+++ b/drivers/net/ethernet/intel/ixgbe/ixgbe_sriov.c
@@ -150,16 +150,6 @@ void ixgbe_enable_sriov(struct ixgbe_adapter *adapter,
 		adapter->flags2 &= ~(IXGBE_FLAG2_RSC_CAPABLE |
 				     IXGBE_FLAG2_RSC_ENABLED);
 
-#ifdef IXGBE_FCOE
-		/*
-		 * When SR-IOV is enabled 82599 cannot support jumbo frames
-		 * so we must disable FCoE because we cannot support FCoE MTU.
-		 */
-		if (adapter->hw.mac.type == ixgbe_mac_82599EB)
-			adapter->flags &= ~(IXGBE_FLAG_FCOE_ENABLED |
-					    IXGBE_FLAG_FCOE_CAPABLE);
-#endif
-
 		/* enable spoof checking for all VFs */
 		for (i = 0; i < adapter->num_vfs; i++)
 			adapter->vfinfo[i].spoofchk_enabled = true;
@@ -353,31 +343,77 @@ static int ixgbe_set_vf_vlan(struct ixgbe_adapter *adapter, int add, int vid,
 	return adapter->hw.mac.ops.set_vfta(&adapter->hw, vid, vf, (bool)add);
 }
 
-static void ixgbe_set_vf_lpe(struct ixgbe_adapter *adapter, u32 *msgbuf)
+static s32 ixgbe_set_vf_lpe(struct ixgbe_adapter *adapter, u32 *msgbuf, u32 vf)
 {
 	struct ixgbe_hw *hw = &adapter->hw;
-	int new_mtu = msgbuf[1];
+	int max_frame = msgbuf[1];
 	u32 max_frs;
-	int max_frame = new_mtu + ETH_HLEN + ETH_FCS_LEN;
 
-	/* Only X540 supports jumbo frames in IOV mode */
-	if (adapter->hw.mac.type != ixgbe_mac_X540)
-		return;
+	/*
+	 * For 82599EB we have to keep all PFs and VFs operating with
+	 * the same max_frame value in order to avoid sending an oversize
+	 * frame to a VF.  In order to guarantee this is handled correctly
+	 * for all cases we have several special exceptions to take into
+	 * account before we can enable the VF for receive
+	 */
+	if (adapter->hw.mac.type == ixgbe_mac_82599EB) {
+		struct net_device *dev = adapter->netdev;
+		int pf_max_frame = dev->mtu + ETH_HLEN;
+		u32 reg_offset, vf_shift, vfre;
+		s32 err = 0;
+
+#ifdef CONFIG_FCOE
+		if (dev->features & NETIF_F_FCOE_MTU)
+			pf_max_frame = max_t(int, pf_max_frame,
+					     IXGBE_FCOE_JUMBO_FRAME_SIZE);
+
+#endif /* CONFIG_FCOE */
+		/*
+		 * If the PF or VF are running w/ jumbo frames enabled we
+		 * need to shut down the VF Rx path as we cannot support
+		 * jumbo frames on legacy VFs
+		 */
+		if ((pf_max_frame > ETH_FRAME_LEN) ||
+		    (max_frame > (ETH_FRAME_LEN + ETH_FCS_LEN)))
+			err = -EINVAL;
+
+		/* determine VF receive enable location */
+		vf_shift = vf % 32;
+		reg_offset = vf / 32;
+
+		/* enable or disable receive depending on error */
+		vfre = IXGBE_READ_REG(hw, IXGBE_VFRE(reg_offset));
+		if (err)
+			vfre &= ~(1 << vf_shift);
+		else
+			vfre |= 1 << vf_shift;
+		IXGBE_WRITE_REG(hw, IXGBE_VFRE(reg_offset), vfre);
+
+		if (err) {
+			e_err(drv, "VF max_frame %d out of range\n", max_frame);
+			return err;
+		}
+	}
 
 	/* MTU < 68 is an error and causes problems on some kernels */
-	if ((new_mtu < 68) || (max_frame > IXGBE_MAX_JUMBO_FRAME_SIZE)) {
-		e_err(drv, "VF mtu %d out of range\n", new_mtu);
-		return;
+	if (max_frame > IXGBE_MAX_JUMBO_FRAME_SIZE) {
+		e_err(drv, "VF max_frame %d out of range\n", max_frame);
+		return -EINVAL;
 	}
 
-	max_frs = (IXGBE_READ_REG(hw, IXGBE_MAXFRS) &
-		   IXGBE_MHADD_MFS_MASK) >> IXGBE_MHADD_MFS_SHIFT;
-	if (max_frs < new_mtu) {
-		max_frs = new_mtu << IXGBE_MHADD_MFS_SHIFT;
+	/* pull current max frame size from hardware */
+	max_frs = IXGBE_READ_REG(hw, IXGBE_MAXFRS);
+	max_frs &= IXGBE_MHADD_MFS_MASK;
+	max_frs >>= IXGBE_MHADD_MFS_SHIFT;
+
+	if (max_frs < max_frame) {
+		max_frs = max_frame << IXGBE_MHADD_MFS_SHIFT;
 		IXGBE_WRITE_REG(hw, IXGBE_MAXFRS, max_frs);
 	}
 
-	e_info(hw, "VF requests change max MTU to %d\n", new_mtu);
+	e_info(hw, "VF requests change max MTU to %d\n", max_frame);
+
+	return 0;
 }
 
 static void ixgbe_set_vmolr(struct ixgbe_hw *hw, u32 vf, bool aupe)
@@ -532,11 +568,28 @@ static inline void ixgbe_vf_reset_msg(struct ixgbe_adapter *adapter, u32 vf)
 
 	/* enable transmit and receive for vf */
 	reg = IXGBE_READ_REG(hw, IXGBE_VFTE(reg_offset));
-	reg |= (reg | (1 << vf_shift));
+	reg |= 1 << vf_shift;
 	IXGBE_WRITE_REG(hw, IXGBE_VFTE(reg_offset), reg);
 
 	reg = IXGBE_READ_REG(hw, IXGBE_VFRE(reg_offset));
-	reg |= (reg | (1 << vf_shift));
+	reg |= 1 << vf_shift;
+	/*
+	 * The 82599 cannot support a mix of jumbo and non-jumbo PF/VFs.
+	 * For more info take a look at ixgbe_set_vf_lpe
+	 */
+	if (adapter->hw.mac.type == ixgbe_mac_82599EB) {
+		struct net_device *dev = adapter->netdev;
+		int pf_max_frame = dev->mtu + ETH_HLEN;
+
+#ifdef CONFIG_FCOE
+		if (dev->features & NETIF_F_FCOE_MTU)
+			pf_max_frame = max_t(int, pf_max_frame,
+					     IXGBE_FCOE_JUMBO_FRAME_SIZE);
+
+#endif /* CONFIG_FCOE */
+		if (pf_max_frame > ETH_FRAME_LEN)
+			reg &= ~(1 << vf_shift);
+	}
 	IXGBE_WRITE_REG(hw, IXGBE_VFRE(reg_offset), reg);
 
 	/* Enable counting of spoofed packets in the SSVPC register */
@@ -633,7 +686,7 @@ static int ixgbe_rcv_msg_from_vf(struct ixgbe_adapter *adapter, u32 vf)
 		                                 hash_list, vf);
 		break;
 	case IXGBE_VF_SET_LPE:
-		ixgbe_set_vf_lpe(adapter, msgbuf);
+		retval = ixgbe_set_vf_lpe(adapter, msgbuf, vf);
 		break;
 	case IXGBE_VF_SET_VLAN:
 		add = (msgbuf[0] & IXGBE_VT_MSGINFO_MASK)
-- 
1.7.11.7

[net-next 03/14] ixgbe: Move message handling routines into their own functions

[Jeff Kirsher]

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

Instead of trying to maintain one large monolithic function that handles
most of the different messages from the VF it makes sense to break the
message handling function up so that we can just go through one switch
statement and call the correct routine for a given message.

Signed-off-by: Alexander Duyck <alexander.h.duyck@intel.com>
Tested-by: Phil Schmitt <phillip.j.schmitt@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
---
 drivers/net/ethernet/intel/ixgbe/ixgbe_sriov.c | 223 ++++++++++++++-----------
 1 file changed, 130 insertions(+), 93 deletions(-)

diff --git a/drivers/net/ethernet/intel/ixgbe/ixgbe_sriov.c b/drivers/net/ethernet/intel/ixgbe/ixgbe_sriov.c
index 420766e..d641c04 100644
--- a/drivers/net/ethernet/intel/ixgbe/ixgbe_sriov.c
+++ b/drivers/net/ethernet/intel/ixgbe/ixgbe_sriov.c
@@ -255,8 +255,11 @@ void ixgbe_disable_sriov(struct ixgbe_adapter *adapter)
 }
 
 static int ixgbe_set_vf_multicasts(struct ixgbe_adapter *adapter,
-				   int entries, u16 *hash_list, u32 vf)
+				   u32 *msgbuf, u32 vf)
 {
+	int entries = (msgbuf[0] & IXGBE_VT_MSGINFO_MASK)
+		       >> IXGBE_VT_MSGINFO_SHIFT;
+	u16 *hash_list = (u16 *)&msgbuf[1];
 	struct vf_data_storage *vfinfo = &adapter->vfinfo[vf];
 	struct ixgbe_hw *hw = &adapter->hw;
 	int i;
@@ -557,20 +560,31 @@ int ixgbe_vf_configuration(struct pci_dev *pdev, unsigned int event_mask)
 	return 0;
 }
 
-static inline void ixgbe_vf_reset_msg(struct ixgbe_adapter *adapter, u32 vf)
+static int ixgbe_vf_reset_msg(struct ixgbe_adapter *adapter, u32 vf)
 {
 	struct ixgbe_hw *hw = &adapter->hw;
-	u32 reg;
+	unsigned char *vf_mac = adapter->vfinfo[vf].vf_mac_addresses;
+	u32 reg, msgbuf[4];
 	u32 reg_offset, vf_shift;
+	u8 *addr = (u8 *)(&msgbuf[1]);
+
+	e_info(probe, "VF Reset msg received from vf %d\n", vf);
+
+	/* reset the filters for the device */
+	ixgbe_vf_reset_event(adapter, vf);
+
+	/* set vf mac address */
+	ixgbe_set_vf_mac(adapter, vf, vf_mac);
 
 	vf_shift = vf % 32;
 	reg_offset = vf / 32;
 
-	/* enable transmit and receive for vf */
+	/* enable transmit for vf */
 	reg = IXGBE_READ_REG(hw, IXGBE_VFTE(reg_offset));
 	reg |= 1 << vf_shift;
 	IXGBE_WRITE_REG(hw, IXGBE_VFTE(reg_offset), reg);
 
+	/* enable receive for vf */
 	reg = IXGBE_READ_REG(hw, IXGBE_VFRE(reg_offset));
 	reg |= 1 << vf_shift;
 	/*
@@ -592,12 +606,115 @@ static inline void ixgbe_vf_reset_msg(struct ixgbe_adapter *adapter, u32 vf)
 	}
 	IXGBE_WRITE_REG(hw, IXGBE_VFRE(reg_offset), reg);
 
+	/* enable VF mailbox for further messages */
+	adapter->vfinfo[vf].clear_to_send = true;
+
 	/* Enable counting of spoofed packets in the SSVPC register */
 	reg = IXGBE_READ_REG(hw, IXGBE_VMECM(reg_offset));
 	reg |= (1 << vf_shift);
 	IXGBE_WRITE_REG(hw, IXGBE_VMECM(reg_offset), reg);
 
-	ixgbe_vf_reset_event(adapter, vf);
+	/* reply to reset with ack and vf mac address */
+	msgbuf[0] = IXGBE_VF_RESET | IXGBE_VT_MSGTYPE_ACK;
+	memcpy(addr, vf_mac, ETH_ALEN);
+
+	/*
+	 * Piggyback the multicast filter type so VF can compute the
+	 * correct vectors
+	 */
+	msgbuf[3] = hw->mac.mc_filter_type;
+	ixgbe_write_mbx(hw, msgbuf, IXGBE_VF_PERMADDR_MSG_LEN, vf);
+
+	return 0;
+}
+
+static int ixgbe_set_vf_mac_addr(struct ixgbe_adapter *adapter,
+				 u32 *msgbuf, u32 vf)
+{
+	u8 *new_mac = ((u8 *)(&msgbuf[1]));
+
+	if (!is_valid_ether_addr(new_mac)) {
+		e_warn(drv, "VF %d attempted to set invalid mac\n", vf);
+		return -1;
+	}
+
+	if (adapter->vfinfo[vf].pf_set_mac &&
+	    memcmp(adapter->vfinfo[vf].vf_mac_addresses, new_mac,
+		   ETH_ALEN)) {
+		e_warn(drv,
+		       "VF %d attempted to override administratively set MAC address\n"
+		       "Reload the VF driver to resume operations\n",
+		       vf);
+		return -1;
+	}
+
+	return ixgbe_set_vf_mac(adapter, vf, new_mac);
+}
+
+static int ixgbe_set_vf_vlan_msg(struct ixgbe_adapter *adapter,
+				 u32 *msgbuf, u32 vf)
+{
+	struct ixgbe_hw *hw = &adapter->hw;
+	int add = (msgbuf[0] & IXGBE_VT_MSGINFO_MASK) >> IXGBE_VT_MSGINFO_SHIFT;
+	int vid = (msgbuf[1] & IXGBE_VLVF_VLANID_MASK);
+	int err;
+
+	if (adapter->vfinfo[vf].pf_vlan) {
+		e_warn(drv,
+		       "VF %d attempted to override administratively set VLAN configuration\n"
+		       "Reload the VF driver to resume operations\n",
+		       vf);
+		return -1;
+	}
+
+	if (add)
+		adapter->vfinfo[vf].vlan_count++;
+	else if (adapter->vfinfo[vf].vlan_count)
+		adapter->vfinfo[vf].vlan_count--;
+
+	err = ixgbe_set_vf_vlan(adapter, add, vid, vf);
+	if (!err && adapter->vfinfo[vf].spoofchk_enabled)
+		hw->mac.ops.set_vlan_anti_spoofing(hw, true, vf);
+
+	return err;
+}
+
+static int ixgbe_set_vf_macvlan_msg(struct ixgbe_adapter *adapter,
+				    u32 *msgbuf, u32 vf)
+{
+	u8 *new_mac = ((u8 *)(&msgbuf[1]));
+	int index = (msgbuf[0] & IXGBE_VT_MSGINFO_MASK) >>
+		    IXGBE_VT_MSGINFO_SHIFT;
+	int err;
+
+	if (adapter->vfinfo[vf].pf_set_mac && index > 0) {
+		e_warn(drv,
+		       "VF %d requested MACVLAN filter but is administratively denied\n",
+		       vf);
+		return -1;
+	}
+
+	/* An non-zero index indicates the VF is setting a filter */
+	if (index) {
+		if (!is_valid_ether_addr(new_mac)) {
+			e_warn(drv, "VF %d attempted to set invalid mac\n", vf);
+			return -1;
+		}
+
+		/*
+		 * If the VF is allowed to set MAC filters then turn off
+		 * anti-spoofing to avoid false positives.
+		 */
+		if (adapter->vfinfo[vf].spoofchk_enabled)
+			ixgbe_ndo_set_vf_spoofchk(adapter->netdev, vf, false);
+	}
+
+	err = ixgbe_set_vf_macvlan(adapter, vf, index, new_mac);
+	if (err == -ENOSPC)
+		e_warn(drv,
+		       "VF %d has requested a MACVLAN filter but there is no space for it\n",
+		       vf);
+	return err;
 }
 
 static int ixgbe_rcv_msg_from_vf(struct ixgbe_adapter *adapter, u32 vf)
@@ -606,10 +723,6 @@ static int ixgbe_rcv_msg_from_vf(struct ixgbe_adapter *adapter, u32 vf)
 	u32 msgbuf[IXGBE_VFMAILBOX_SIZE];
 	struct ixgbe_hw *hw = &adapter->hw;
 	s32 retval;
-	int entries;
-	u16 *hash_list;
-	int add, vid, index;
-	u8 *new_mac;
 
 	retval = ixgbe_read_mbx(hw, msgbuf, mbx_size, vf);
 
@@ -630,33 +743,8 @@ static int ixgbe_rcv_msg_from_vf(struct ixgbe_adapter *adapter, u32 vf)
 	 * allowed to start any configuration.
 	 */
 
-	if (msgbuf[0] == IXGBE_VF_RESET) {
-		unsigned char *vf_mac = adapter->vfinfo[vf].vf_mac_addresses;
-		new_mac = (u8 *)(&msgbuf[1]);
-		e_info(probe, "VF Reset msg received from vf %d\n", vf);
-		adapter->vfinfo[vf].clear_to_send = false;
-		ixgbe_vf_reset_msg(adapter, vf);
-		adapter->vfinfo[vf].clear_to_send = true;
-
-		if (is_valid_ether_addr(new_mac) &&
-		    !adapter->vfinfo[vf].pf_set_mac)
-			ixgbe_set_vf_mac(adapter, vf, vf_mac);
-		else
-			ixgbe_set_vf_mac(adapter,
-				 vf, adapter->vfinfo[vf].vf_mac_addresses);
-
-		/* reply to reset with ack and vf mac address */
-		msgbuf[0] = IXGBE_VF_RESET | IXGBE_VT_MSGTYPE_ACK;
-		memcpy(new_mac, vf_mac, ETH_ALEN);
-		/*
-		 * Piggyback the multicast filter type so VF can compute the
-		 * correct vectors
-		 */
-		msgbuf[3] = hw->mac.mc_filter_type;
-		ixgbe_write_mbx(hw, msgbuf, IXGBE_VF_PERMADDR_MSG_LEN, vf);
-
-		return retval;
-	}
+	if (msgbuf[0] == IXGBE_VF_RESET)
+		return ixgbe_vf_reset_msg(adapter, vf);
 
 	if (!adapter->vfinfo[vf].clear_to_send) {
 		msgbuf[0] |= IXGBE_VT_MSGTYPE_NACK;
@@ -666,70 +754,19 @@ static int ixgbe_rcv_msg_from_vf(struct ixgbe_adapter *adapter, u32 vf)
 
 	switch ((msgbuf[0] & 0xFFFF)) {
 	case IXGBE_VF_SET_MAC_ADDR:
-		new_mac = ((u8 *)(&msgbuf[1]));
-		if (is_valid_ether_addr(new_mac) &&
-		    !adapter->vfinfo[vf].pf_set_mac) {
-			ixgbe_set_vf_mac(adapter, vf, new_mac);
-		} else if (memcmp(adapter->vfinfo[vf].vf_mac_addresses,
-				  new_mac, ETH_ALEN)) {
-			e_warn(drv, "VF %d attempted to override "
-			       "administratively set MAC address\nReload "
-			       "the VF driver to resume operations\n", vf);
-			retval = -1;
-		}
+		retval = ixgbe_set_vf_mac_addr(adapter, msgbuf, vf);
 		break;
 	case IXGBE_VF_SET_MULTICAST:
-		entries = (msgbuf[0] & IXGBE_VT_MSGINFO_MASK)
-		          >> IXGBE_VT_MSGINFO_SHIFT;
-		hash_list = (u16 *)&msgbuf[1];
-		retval = ixgbe_set_vf_multicasts(adapter, entries,
-		                                 hash_list, vf);
+		retval = ixgbe_set_vf_multicasts(adapter, msgbuf, vf);
+		break;
+	case IXGBE_VF_SET_VLAN:
+		retval = ixgbe_set_vf_vlan_msg(adapter, msgbuf, vf);
 		break;
 	case IXGBE_VF_SET_LPE:
 		retval = ixgbe_set_vf_lpe(adapter, msgbuf, vf);
 		break;
-	case IXGBE_VF_SET_VLAN:
-		add = (msgbuf[0] & IXGBE_VT_MSGINFO_MASK)
-		      >> IXGBE_VT_MSGINFO_SHIFT;
-		vid = (msgbuf[1] & IXGBE_VLVF_VLANID_MASK);
-		if (adapter->vfinfo[vf].pf_vlan) {
-			e_warn(drv, "VF %d attempted to override "
-			       "administratively set VLAN configuration\n"
-			       "Reload the VF driver to resume operations\n",
-			       vf);
-			retval = -1;
-		} else {
-			if (add)
-				adapter->vfinfo[vf].vlan_count++;
-			else if (adapter->vfinfo[vf].vlan_count)
-				adapter->vfinfo[vf].vlan_count--;
-			retval = ixgbe_set_vf_vlan(adapter, add, vid, vf);
-			if (!retval && adapter->vfinfo[vf].spoofchk_enabled)
-				hw->mac.ops.set_vlan_anti_spoofing(hw, true, vf);
-		}
-		break;
 	case IXGBE_VF_SET_MACVLAN:
-		index = (msgbuf[0] & IXGBE_VT_MSGINFO_MASK) >>
-			IXGBE_VT_MSGINFO_SHIFT;
-		if (adapter->vfinfo[vf].pf_set_mac && index > 0) {
-			e_warn(drv, "VF %d requested MACVLAN filter but is "
-				    "administratively denied\n", vf);
-			retval = -1;
-			break;
-		}
-		/*
-		 * If the VF is allowed to set MAC filters then turn off
-		 * anti-spoofing to avoid false positives.  An index
-		 * greater than 0 will indicate the VF is setting a
-		 * macvlan MAC filter.
-		 */
-		if (index > 0 && adapter->vfinfo[vf].spoofchk_enabled)
-			ixgbe_ndo_set_vf_spoofchk(adapter->netdev, vf, false);
-		retval = ixgbe_set_vf_macvlan(adapter, vf, index,
-					      (unsigned char *)(&msgbuf[1]));
-		if (retval == -ENOSPC)
-			e_warn(drv, "VF %d has requested a MACVLAN filter "
-				    "but there is no space for it\n", vf);
+		retval = ixgbe_set_vf_macvlan_msg(adapter, msgbuf, vf);
 		break;
 	default:
 		e_err(drv, "Unhandled Msg %8.8x\n", msgbuf[0]);
-- 
1.7.11.7

[net-next 04/14] ixgbe: Add mailbox API version negotiation support to ixgbe PF

[Jeff Kirsher]

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

This change allows us to add a mailbox versioning API.  This will allow us
to determine the features supported by the VFs from the PF.  For example we
will be implementing a version 1.1 API for the VF that will indicate that
it can support us enabling Jumbo frames as the VF will support buffer
chaining.

Signed-off-by: Alexander Duyck <alexander.h.duyck@intel.com>
Tested-by: Phil Schmitt <phillip.j.schmitt@intel.com>
Tested-by: Robert Garrett <RobertX.Garrett@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
---
 drivers/net/ethernet/intel/ixgbe/ixgbe.h       |  1 +
 drivers/net/ethernet/intel/ixgbe/ixgbe_mbx.h   | 21 ++++++++++++++--
 drivers/net/ethernet/intel/ixgbe/ixgbe_sriov.c | 33 ++++++++++++++++++++++----
 3 files changed, 48 insertions(+), 7 deletions(-)

diff --git a/drivers/net/ethernet/intel/ixgbe/ixgbe.h b/drivers/net/ethernet/intel/ixgbe/ixgbe.h
index 30efc9f..ccb8505 100644
--- a/drivers/net/ethernet/intel/ixgbe/ixgbe.h
+++ b/drivers/net/ethernet/intel/ixgbe/ixgbe.h
@@ -135,6 +135,7 @@ struct vf_data_storage {
 	u16 tx_rate;
 	u16 vlan_count;
 	u8 spoofchk_enabled;
+	unsigned int vf_api;
 };
 
 struct vf_macvlans {
diff --git a/drivers/net/ethernet/intel/ixgbe/ixgbe_mbx.h b/drivers/net/ethernet/intel/ixgbe/ixgbe_mbx.h
index 310bdd9..d4c842e 100644
--- a/drivers/net/ethernet/intel/ixgbe/ixgbe_mbx.h
+++ b/drivers/net/ethernet/intel/ixgbe/ixgbe_mbx.h
@@ -62,12 +62,29 @@
 /* bits 23:16 are used for exra info for certain messages */
 #define IXGBE_VT_MSGINFO_MASK     (0xFF << IXGBE_VT_MSGINFO_SHIFT)
 
+/* definitions to support mailbox API version negotiation */
+
+/*
+ * Each element denotes a version of the API; existing numbers may not
+ * change; any additions must go at the end
+ */
+enum ixgbe_pfvf_api_rev {
+	ixgbe_mbox_api_10,	/* API version 1.0, linux/freebsd VF driver */
+	ixgbe_mbox_api_20,	/* API version 2.0, solaris Phase1 VF driver */
+	/* This value should always be last */
+	ixgbe_mbox_api_unknown,	/* indicates that API version is not known */
+};
+
+/* mailbox API, legacy requests */
 #define IXGBE_VF_RESET            0x01 /* VF requests reset */
 #define IXGBE_VF_SET_MAC_ADDR     0x02 /* VF requests PF to set MAC addr */
 #define IXGBE_VF_SET_MULTICAST    0x03 /* VF requests PF to set MC addr */
 #define IXGBE_VF_SET_VLAN         0x04 /* VF requests PF to set VLAN */
-#define IXGBE_VF_SET_LPE          0x05 /* VF requests PF to set VMOLR.LPE */
-#define IXGBE_VF_SET_MACVLAN      0x06 /* VF requests PF for unicast filter */
+
+/* mailbox API, version 1.0 VF requests */
+#define IXGBE_VF_SET_LPE	0x05 /* VF requests PF to set VMOLR.LPE */
+#define IXGBE_VF_SET_MACVLAN	0x06 /* VF requests PF for unicast filter */
+#define IXGBE_VF_API_NEGOTIATE	0x08 /* negotiate API version */
 
 /* length of permanent address message returned from PF */
 #define IXGBE_VF_PERMADDR_MSG_LEN 4
diff --git a/drivers/net/ethernet/intel/ixgbe/ixgbe_sriov.c b/drivers/net/ethernet/intel/ixgbe/ixgbe_sriov.c
index d641c04..f563625 100644
--- a/drivers/net/ethernet/intel/ixgbe/ixgbe_sriov.c
+++ b/drivers/net/ethernet/intel/ixgbe/ixgbe_sriov.c
@@ -469,6 +469,9 @@ static inline void ixgbe_vf_reset_event(struct ixgbe_adapter *adapter, u32 vf)
 	ixgbe_set_rx_mode(adapter->netdev);
 
 	hw->mac.ops.clear_rar(hw, rar_entry);
+
+	/* reset VF api back to unknown */
+	adapter->vfinfo[vf].vf_api = ixgbe_mbox_api_10;
 }
 
 static int ixgbe_set_vf_mac(struct ixgbe_adapter *adapter,
@@ -717,6 +720,24 @@ static int ixgbe_set_vf_macvlan_msg(struct ixgbe_adapter *adapter,
 	return err;
 }
 
+static int ixgbe_negotiate_vf_api(struct ixgbe_adapter *adapter,
+				  u32 *msgbuf, u32 vf)
+{
+	int api = msgbuf[1];
+
+	switch (api) {
+	case ixgbe_mbox_api_10:
+		adapter->vfinfo[vf].vf_api = api;
+		return 0;
+	default:
+		break;
+	}
+
+	e_info(drv, "VF %d requested invalid api version %u\n", vf, api);
+
+	return -1;
+}
+
 static int ixgbe_rcv_msg_from_vf(struct ixgbe_adapter *adapter, u32 vf)
 {
 	u32 mbx_size = IXGBE_VFMAILBOX_SIZE;
@@ -738,14 +759,13 @@ static int ixgbe_rcv_msg_from_vf(struct ixgbe_adapter *adapter, u32 vf)
 	/* flush the ack before we write any messages back */
 	IXGBE_WRITE_FLUSH(hw);
 
+	if (msgbuf[0] == IXGBE_VF_RESET)
+		return ixgbe_vf_reset_msg(adapter, vf);
+
 	/*
 	 * until the vf completes a virtual function reset it should not be
 	 * allowed to start any configuration.
 	 */
-
-	if (msgbuf[0] == IXGBE_VF_RESET)
-		return ixgbe_vf_reset_msg(adapter, vf);
-
 	if (!adapter->vfinfo[vf].clear_to_send) {
 		msgbuf[0] |= IXGBE_VT_MSGTYPE_NACK;
 		ixgbe_write_mbx(hw, msgbuf, 1, vf);
@@ -768,6 +788,9 @@ static int ixgbe_rcv_msg_from_vf(struct ixgbe_adapter *adapter, u32 vf)
 	case IXGBE_VF_SET_MACVLAN:
 		retval = ixgbe_set_vf_macvlan_msg(adapter, msgbuf, vf);
 		break;
+	case IXGBE_VF_API_NEGOTIATE:
+		retval = ixgbe_negotiate_vf_api(adapter, msgbuf, vf);
+		break;
 	default:
 		e_err(drv, "Unhandled Msg %8.8x\n", msgbuf[0]);
 		retval = IXGBE_ERR_MBX;
@@ -782,7 +805,7 @@ static int ixgbe_rcv_msg_from_vf(struct ixgbe_adapter *adapter, u32 vf)
 
 	msgbuf[0] |= IXGBE_VT_MSGTYPE_CTS;
 
-	ixgbe_write_mbx(hw, msgbuf, 1, vf);
+	ixgbe_write_mbx(hw, msgbuf, mbx_size, vf);
 
 	return retval;
 }
-- 
1.7.11.7

[net-next 05/14] igb: Correcting and improving small packet check and padding

[Jeff Kirsher]

From: Tushar Dave <tushar.n.dave@intel.com>

Current implementation mess up the tail pointer. This patch sets skb->tail
correctly.
Also, the small packet check and padding is optimized by using unlikely and
calling skb_pad directly.

Signed-off-by: Tushar Dave <tushar.n.dave@intel.com>
Tested-by: Jeff Pieper <jeffrey.e.pieper@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
---
 drivers/net/ethernet/intel/igb/igb_main.c | 5 +++--
 1 file changed, 3 insertions(+), 2 deletions(-)

diff --git a/drivers/net/ethernet/intel/igb/igb_main.c b/drivers/net/ethernet/intel/igb/igb_main.c
index e1ceb37..c611cff 100644
--- a/drivers/net/ethernet/intel/igb/igb_main.c
+++ b/drivers/net/ethernet/intel/igb/igb_main.c
@@ -4467,10 +4467,11 @@ static netdev_tx_t igb_xmit_frame(struct sk_buff *skb,
 	 * The minimum packet size with TCTL.PSP set is 17 so pad the skb
 	 * in order to meet this minimum size requirement.
 	 */
-	if (skb->len < 17) {
-		if (skb_padto(skb, 17))
+	if (unlikely(skb->len < 17)) {
+		if (skb_pad(skb, 17 - skb->len))
 			return NETDEV_TX_OK;
 		skb->len = 17;
+		skb_set_tail_pointer(skb, 17);
 	}
 
 	return igb_xmit_frame_ring(skb, igb_tx_queue_mapping(adapter, skb));
-- 
1.7.11.7

[net-next 06/14] igb: Split Rx timestamping into two separate functions

[Jeff Kirsher]

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

In order to support page based receive we will need to split up the two
different types of timestamping into two separate functions.  The first one
will handle legacy timestamps with the value in the register, and the new
one will handle timestamps in the Rx buffer itself.

Signed-off-by: Alexander Duyck <alexander.h.duyck@intel.com>
Acked-by: Matthew Vick <matthew.vick@intel.com>
Tested-by: Aaron Brown <aaron.f.brown@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
---
 drivers/net/ethernet/intel/igb/igb.h     | 17 ++++++++--
 drivers/net/ethernet/intel/igb/igb_ptp.c | 55 ++++++++++++++++++++++----------
 2 files changed, 53 insertions(+), 19 deletions(-)

diff --git a/drivers/net/ethernet/intel/igb/igb.h b/drivers/net/ethernet/intel/igb/igb.h
index 8aad230..f6a1cd9 100644
--- a/drivers/net/ethernet/intel/igb/igb.h
+++ b/drivers/net/ethernet/intel/igb/igb.h
@@ -442,9 +442,22 @@ extern void igb_ptp_stop(struct igb_adapter *adapter);
 extern void igb_ptp_reset(struct igb_adapter *adapter);
 extern void igb_ptp_tx_work(struct work_struct *work);
 extern void igb_ptp_tx_hwtstamp(struct igb_adapter *adapter);
-extern void igb_ptp_rx_hwtstamp(struct igb_q_vector *q_vector,
-				union e1000_adv_rx_desc *rx_desc,
+extern void igb_ptp_rx_rgtstamp(struct igb_q_vector *q_vector,
 				struct sk_buff *skb);
+extern void igb_ptp_rx_pktstamp(struct igb_q_vector *q_vector,
+				unsigned char *va,
+				struct sk_buff *skb);
+static inline void igb_ptp_rx_hwtstamp(struct igb_q_vector *q_vector,
+				       union e1000_adv_rx_desc *rx_desc,
+				       struct sk_buff *skb)
+{
+	if (igb_test_staterr(rx_desc, E1000_RXDADV_STAT_TSIP)) {
+		igb_ptp_rx_pktstamp(q_vector, skb->data, skb);
+		skb_pull(skb, IGB_TS_HDR_LEN);
+	} else if (igb_test_staterr(rx_desc, E1000_RXDADV_STAT_TS)) {
+		igb_ptp_rx_rgtstamp(q_vector, skb);
+	}
+}
 extern int igb_ptp_hwtstamp_ioctl(struct net_device *netdev,
 				  struct ifreq *ifr, int cmd);
 #endif /* CONFIG_IGB_PTP */
diff --git a/drivers/net/ethernet/intel/igb/igb_ptp.c b/drivers/net/ethernet/intel/igb/igb_ptp.c
index ee21445..a7db4ce 100644
--- a/drivers/net/ethernet/intel/igb/igb_ptp.c
+++ b/drivers/net/ethernet/intel/igb/igb_ptp.c
@@ -441,18 +441,46 @@ void igb_ptp_tx_hwtstamp(struct igb_adapter *adapter)
 	adapter->ptp_tx_skb = NULL;
 }
 
-void igb_ptp_rx_hwtstamp(struct igb_q_vector *q_vector,
-			 union e1000_adv_rx_desc *rx_desc,
+/**
+ * igb_ptp_rx_pktstamp - retrieve Rx per packet timestamp
+ * @q_vector: Pointer to interrupt specific structure
+ * @va: Pointer to address containing Rx buffer
+ * @skb: Buffer containing timestamp and packet
+ *
+ * This function is meant to retrieve a timestamp from the first buffer of an
+ * incoming frame.  The value is stored in little endian format starting on
+ * byte 8.
+ */
+void igb_ptp_rx_pktstamp(struct igb_q_vector *q_vector,
+			 unsigned char *va,
+			 struct sk_buff *skb)
+{
+	u64 *regval = (u64 *)va;
+
+	/*
+	 * The timestamp is recorded in little endian format.
+	 * DWORD: 0        1        2        3
+	 * Field: Reserved Reserved SYSTIML  SYSTIMH
+	 */
+	igb_ptp_systim_to_hwtstamp(q_vector->adapter, skb_hwtstamps(skb),
+				   le64_to_cpu(regval[1]));
+}
+
+/**
+ * igb_ptp_rx_rgtstamp - retrieve Rx timestamp stored in register
+ * @q_vector: Pointer to interrupt specific structure
+ * @skb: Buffer containing timestamp and packet
+ *
+ * This function is meant to retrieve a timestamp from the internal registers
+ * of the adapter and store it in the skb.
+ */
+void igb_ptp_rx_rgtstamp(struct igb_q_vector *q_vector,
 			 struct sk_buff *skb)
 {
 	struct igb_adapter *adapter = q_vector->adapter;
 	struct e1000_hw *hw = &adapter->hw;
 	u64 regval;
 
-	if (!igb_test_staterr(rx_desc, E1000_RXDADV_STAT_TSIP |
-				       E1000_RXDADV_STAT_TS))
-		return;
-
 	/*
 	 * If this bit is set, then the RX registers contain the time stamp. No
 	 * other packet will be time stamped until we read these registers, so
@@ -464,18 +492,11 @@ void igb_ptp_rx_hwtstamp(struct igb_q_vector *q_vector,
 	 * If nothing went wrong, then it should have a shared tx_flags that we
 	 * can turn into a skb_shared_hwtstamps.
 	 */
-	if (igb_test_staterr(rx_desc, E1000_RXDADV_STAT_TSIP)) {
-		u32 *stamp = (u32 *)skb->data;
-		regval = le32_to_cpu(*(stamp + 2));
-		regval |= (u64)le32_to_cpu(*(stamp + 3)) << 32;
-		skb_pull(skb, IGB_TS_HDR_LEN);
-	} else {
-		if (!(rd32(E1000_TSYNCRXCTL) & E1000_TSYNCRXCTL_VALID))
-			return;
+	if (!(rd32(E1000_TSYNCRXCTL) & E1000_TSYNCRXCTL_VALID))
+		return;
 
-		regval = rd32(E1000_RXSTMPL);
-		regval |= (u64)rd32(E1000_RXSTMPH) << 32;
-	}
+	regval = rd32(E1000_RXSTMPL);
+	regval |= (u64)rd32(E1000_RXSTMPH) << 32;
 
 	igb_ptp_systim_to_hwtstamp(adapter, skb_hwtstamps(skb), regval);
 }
-- 
1.7.11.7

[net-next 07/14] igb: Do not use header split, instead receive all frames into a single buffer

[Jeff Kirsher]

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

This change makes it so that we no longer use header split.  The idea is to
reduce partial cache line writes by hardware when handling frames larger
then header size.  We can compensate for the extra overhead of having to
memcpy the header buffer by avoiding the cache misses seen by leaving an
full skb allocated and sitting on the ring.

Signed-off-by: Alexander Duyck <alexander.h.duyck@intel.com>
Tested-by: Aaron Brown <aaron.f.brown@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
---
 drivers/net/ethernet/intel/igb/igb.h         |  13 +-
 drivers/net/ethernet/intel/igb/igb_ethtool.c |  31 +-
 drivers/net/ethernet/intel/igb/igb_main.c    | 420 ++++++++++++++++++---------
 3 files changed, 312 insertions(+), 152 deletions(-)

diff --git a/drivers/net/ethernet/intel/igb/igb.h b/drivers/net/ethernet/intel/igb/igb.h
index f6a1cd9..72ab9ac 100644
--- a/drivers/net/ethernet/intel/igb/igb.h
+++ b/drivers/net/ethernet/intel/igb/igb.h
@@ -174,11 +174,9 @@ struct igb_tx_buffer {
 };
 
 struct igb_rx_buffer {
-	struct sk_buff *skb;
 	dma_addr_t dma;
 	struct page *page;
-	dma_addr_t page_dma;
-	u32 page_offset;
+	unsigned int page_offset;
 };
 
 struct igb_tx_queue_stats {
@@ -251,6 +249,7 @@ struct igb_ring {
 		};
 		/* RX */
 		struct {
+			struct sk_buff *skb;
 			struct igb_rx_queue_stats rx_stats;
 			struct u64_stats_sync rx_syncp;
 		};
@@ -451,13 +450,11 @@ static inline void igb_ptp_rx_hwtstamp(struct igb_q_vector *q_vector,
 				       union e1000_adv_rx_desc *rx_desc,
 				       struct sk_buff *skb)
 {
-	if (igb_test_staterr(rx_desc, E1000_RXDADV_STAT_TSIP)) {
-		igb_ptp_rx_pktstamp(q_vector, skb->data, skb);
-		skb_pull(skb, IGB_TS_HDR_LEN);
-	} else if (igb_test_staterr(rx_desc, E1000_RXDADV_STAT_TS)) {
+	if (igb_test_staterr(rx_desc, E1000_RXDADV_STAT_TS) &&
+	    !igb_test_staterr(rx_desc, E1000_RXDADV_STAT_TSIP))
 		igb_ptp_rx_rgtstamp(q_vector, skb);
-	}
 }
+
 extern int igb_ptp_hwtstamp_ioctl(struct net_device *netdev,
 				  struct ifreq *ifr, int cmd);
 #endif /* CONFIG_IGB_PTP */
diff --git a/drivers/net/ethernet/intel/igb/igb_ethtool.c b/drivers/net/ethernet/intel/igb/igb_ethtool.c
index 2ea0128..0faac42 100644
--- a/drivers/net/ethernet/intel/igb/igb_ethtool.c
+++ b/drivers/net/ethernet/intel/igb/igb_ethtool.c
@@ -37,6 +37,7 @@
 #include <linux/sched.h>
 #include <linux/slab.h>
 #include <linux/pm_runtime.h>
+#include <linux/highmem.h>
 
 #include "igb.h"
 
@@ -1685,16 +1686,24 @@ static void igb_create_lbtest_frame(struct sk_buff *skb,
 	memset(&skb->data[frame_size + 12], 0xAF, 1);
 }
 
-static int igb_check_lbtest_frame(struct sk_buff *skb, unsigned int frame_size)
+static int igb_check_lbtest_frame(struct igb_rx_buffer *rx_buffer,
+				  unsigned int frame_size)
 {
-	frame_size /= 2;
-	if (*(skb->data + 3) == 0xFF) {
-		if ((*(skb->data + frame_size + 10) == 0xBE) &&
-		   (*(skb->data + frame_size + 12) == 0xAF)) {
-			return 0;
-		}
-	}
-	return 13;
+	unsigned char *data;
+	bool match = true;
+
+	frame_size >>= 1;
+
+	data = kmap(rx_buffer->page) + rx_buffer->page_offset;
+
+	if (data[3] != 0xFF ||
+	    data[frame_size + 10] != 0xBE ||
+	    data[frame_size + 12] != 0xAF)
+		match = false;
+
+	kunmap(rx_buffer->page);
+
+	return match;
 }
 
 static int igb_clean_test_rings(struct igb_ring *rx_ring,
@@ -1720,12 +1729,12 @@ static int igb_clean_test_rings(struct igb_ring *rx_ring,
 		/* unmap rx buffer, will be remapped by alloc_rx_buffers */
 		dma_unmap_single(rx_ring->dev,
 				 rx_buffer_info->dma,
-				 IGB_RX_HDR_LEN,
+				 PAGE_SIZE / 2,
 				 DMA_FROM_DEVICE);
 		rx_buffer_info->dma = 0;
 
 		/* verify contents of skb */
-		if (!igb_check_lbtest_frame(rx_buffer_info->skb, size))
+		if (igb_check_lbtest_frame(rx_buffer_info, size))
 			count++;
 
 		/* unmap buffer on tx side */
diff --git a/drivers/net/ethernet/intel/igb/igb_main.c b/drivers/net/ethernet/intel/igb/igb_main.c
index c611cff..665eafa 100644
--- a/drivers/net/ethernet/intel/igb/igb_main.c
+++ b/drivers/net/ethernet/intel/igb/igb_main.c
@@ -534,25 +534,21 @@ rx_ring_summary:
 
 			if (staterr & E1000_RXD_STAT_DD) {
 				/* Descriptor Done */
-				pr_info("%s[0x%03X]     %016llX %016llX -------"
-					"--------- %p%s\n", "RWB", i,
+				pr_info("%s[0x%03X]     %016llX %016llX ---------------- %s\n",
+					"RWB", i,
 					le64_to_cpu(u0->a),
 					le64_to_cpu(u0->b),
-					buffer_info->skb, next_desc);
+					next_desc);
 			} else {
-				pr_info("%s[0x%03X]     %016llX %016llX %016llX"
-					" %p%s\n", "R  ", i,
+				pr_info("%s[0x%03X]     %016llX %016llX %016llX %s\n",
+					"R  ", i,
 					le64_to_cpu(u0->a),
 					le64_to_cpu(u0->b),
 					(u64)buffer_info->dma,
-					buffer_info->skb, next_desc);
+					next_desc);
 
 				if (netif_msg_pktdata(adapter) &&
-				    buffer_info->dma && buffer_info->skb) {
-					print_hex_dump(KERN_INFO, "",
-						  DUMP_PREFIX_ADDRESS,
-						  16, 1, buffer_info->skb->data,
-						  IGB_RX_HDR_LEN, true);
+				    buffer_info->dma && buffer_info->page) {
 					print_hex_dump(KERN_INFO, "",
 					  DUMP_PREFIX_ADDRESS,
 					  16, 1,
@@ -3111,7 +3107,7 @@ void igb_configure_rx_ring(struct igb_adapter *adapter,
 #else
 	srrctl |= (PAGE_SIZE / 2) >> E1000_SRRCTL_BSIZEPKT_SHIFT;
 #endif
-	srrctl |= E1000_SRRCTL_DESCTYPE_HDR_SPLIT_ALWAYS;
+	srrctl |= E1000_SRRCTL_DESCTYPE_ADV_ONEBUF;
 #ifdef CONFIG_IGB_PTP
 	if (hw->mac.type >= e1000_82580)
 		srrctl |= E1000_SRRCTL_TIMESTAMP;
@@ -3305,36 +3301,27 @@ static void igb_clean_rx_ring(struct igb_ring *rx_ring)
 	unsigned long size;
 	u16 i;
 
+	if (rx_ring->skb)
+		dev_kfree_skb(rx_ring->skb);
+	rx_ring->skb = NULL;
+
 	if (!rx_ring->rx_buffer_info)
 		return;
 
 	/* Free all the Rx ring sk_buffs */
 	for (i = 0; i < rx_ring->count; i++) {
 		struct igb_rx_buffer *buffer_info = &rx_ring->rx_buffer_info[i];
-		if (buffer_info->dma) {
-			dma_unmap_single(rx_ring->dev,
-			                 buffer_info->dma,
-					 IGB_RX_HDR_LEN,
-					 DMA_FROM_DEVICE);
-			buffer_info->dma = 0;
-		}
 
-		if (buffer_info->skb) {
-			dev_kfree_skb(buffer_info->skb);
-			buffer_info->skb = NULL;
-		}
-		if (buffer_info->page_dma) {
+		if (buffer_info->dma)
 			dma_unmap_page(rx_ring->dev,
-			               buffer_info->page_dma,
+				       buffer_info->dma,
 				       PAGE_SIZE / 2,
 				       DMA_FROM_DEVICE);
-			buffer_info->page_dma = 0;
-		}
-		if (buffer_info->page) {
-			put_page(buffer_info->page);
-			buffer_info->page = NULL;
-			buffer_info->page_offset = 0;
-		}
+		buffer_info->dma = 0;
+		if (buffer_info->page)
+			__free_page(buffer_info->page);
+		buffer_info->page = NULL;
+		buffer_info->page_offset = 0;
 	}
 
 	size = sizeof(struct igb_rx_buffer) * rx_ring->count;
@@ -5906,23 +5893,219 @@ static void igb_rx_vlan(struct igb_ring *ring,
 	}
 }
 
-static inline u16 igb_get_hlen(union e1000_adv_rx_desc *rx_desc)
+/**
+ * igb_get_headlen - determine size of header for LRO/GRO
+ * @data: pointer to the start of the headers
+ * @max_len: total length of section to find headers in
+ *
+ * This function is meant to determine the length of headers that will
+ * be recognized by hardware for LRO, and GRO offloads.  The main
+ * motivation of doing this is to only perform one pull for IPv4 TCP
+ * packets so that we can do basic things like calculating the gso_size
+ * based on the average data per packet.
+ **/
+static unsigned int igb_get_headlen(unsigned char *data,
+				    unsigned int max_len)
+{
+	union {
+		unsigned char *network;
+		/* l2 headers */
+		struct ethhdr *eth;
+		struct vlan_hdr *vlan;
+		/* l3 headers */
+		struct iphdr *ipv4;
+		struct ipv6hdr *ipv6;
+	} hdr;
+	__be16 protocol;
+	u8 nexthdr = 0;	/* default to not TCP */
+	u8 hlen;
+
+	/* this should never happen, but better safe than sorry */
+	if (max_len < ETH_HLEN)
+		return max_len;
+
+	/* initialize network frame pointer */
+	hdr.network = data;
+
+	/* set first protocol and move network header forward */
+	protocol = hdr.eth->h_proto;
+	hdr.network += ETH_HLEN;
+
+	/* handle any vlan tag if present */
+	if (protocol == __constant_htons(ETH_P_8021Q)) {
+		if ((hdr.network - data) > (max_len - VLAN_HLEN))
+			return max_len;
+
+		protocol = hdr.vlan->h_vlan_encapsulated_proto;
+		hdr.network += VLAN_HLEN;
+	}
+
+	/* handle L3 protocols */
+	if (protocol == __constant_htons(ETH_P_IP)) {
+		if ((hdr.network - data) > (max_len - sizeof(struct iphdr)))
+			return max_len;
+
+		/* access ihl as a u8 to avoid unaligned access on ia64 */
+		hlen = (hdr.network[0] & 0x0F) << 2;
+
+		/* verify hlen meets minimum size requirements */
+		if (hlen < sizeof(struct iphdr))
+			return hdr.network - data;
+
+		/* record next protocol */
+		nexthdr = hdr.ipv4->protocol;
+		hdr.network += hlen;
+	} else if (protocol == __constant_htons(ETH_P_IPV6)) {
+		if ((hdr.network - data) > (max_len - sizeof(struct ipv6hdr)))
+			return max_len;
+
+		/* record next protocol */
+		nexthdr = hdr.ipv6->nexthdr;
+		hdr.network += sizeof(struct ipv6hdr);
+	} else {
+		return hdr.network - data;
+	}
+
+	/* finally sort out TCP */
+	if (nexthdr == IPPROTO_TCP) {
+		if ((hdr.network - data) > (max_len - sizeof(struct tcphdr)))
+			return max_len;
+
+		/* access doff as a u8 to avoid unaligned access on ia64 */
+		hlen = (hdr.network[12] & 0xF0) >> 2;
+
+		/* verify hlen meets minimum size requirements */
+		if (hlen < sizeof(struct tcphdr))
+			return hdr.network - data;
+
+		hdr.network += hlen;
+	} else if (nexthdr == IPPROTO_UDP) {
+		if ((hdr.network - data) > (max_len - sizeof(struct udphdr)))
+			return max_len;
+
+		hdr.network += sizeof(struct udphdr);
+	}
+
+	/*
+	 * If everything has gone correctly hdr.network should be the
+	 * data section of the packet and will be the end of the header.
+	 * If not then it probably represents the end of the last recognized
+	 * header.
+	 */
+	if ((hdr.network - data) < max_len)
+		return hdr.network - data;
+	else
+		return max_len;
+}
+
+/**
+ * igb_pull_tail - igb specific version of skb_pull_tail
+ * @rx_ring: rx descriptor ring packet is being transacted on
+ * @skb: pointer to current skb being adjusted
+ *
+ * This function is an igb specific version of __pskb_pull_tail.  The
+ * main difference between this version and the original function is that
+ * this function can make several assumptions about the state of things
+ * that allow for significant optimizations versus the standard function.
+ * As a result we can do things like drop a frag and maintain an accurate
+ * truesize for the skb.
+ */
+static void igb_pull_tail(struct igb_ring *rx_ring,
+			  union e1000_adv_rx_desc *rx_desc,
+			  struct sk_buff *skb)
 {
-	/* HW will not DMA in data larger than the given buffer, even if it
-	 * parses the (NFS, of course) header to be larger.  In that case, it
-	 * fills the header buffer and spills the rest into the page.
+	struct skb_frag_struct *frag = &skb_shinfo(skb)->frags[0];
+	unsigned char *va;
+	unsigned int pull_len;
+
+	/*
+	 * it is valid to use page_address instead of kmap since we are
+	 * working with pages allocated out of the lomem pool per
+	 * alloc_page(GFP_ATOMIC)
 	 */
-	u16 hlen = (le16_to_cpu(rx_desc->wb.lower.lo_dword.hdr_info) &
-	           E1000_RXDADV_HDRBUFLEN_MASK) >> E1000_RXDADV_HDRBUFLEN_SHIFT;
-	if (hlen > IGB_RX_HDR_LEN)
-		hlen = IGB_RX_HDR_LEN;
-	return hlen;
+	va = skb_frag_address(frag);
+
+#ifdef CONFIG_IGB_PTP
+	if (igb_test_staterr(rx_desc, E1000_RXDADV_STAT_TSIP)) {
+		/* retrieve timestamp from buffer */
+		igb_ptp_rx_pktstamp(rx_ring->q_vector, va, skb);
+
+		/* update pointers to remove timestamp header */
+		skb_frag_size_sub(frag, IGB_TS_HDR_LEN);
+		frag->page_offset += IGB_TS_HDR_LEN;
+		skb->data_len -= IGB_TS_HDR_LEN;
+		skb->len -= IGB_TS_HDR_LEN;
+
+		/* move va to start of packet data */
+		va += IGB_TS_HDR_LEN;
+	}
+
+#endif
+	/*
+	 * we need the header to contain the greater of either ETH_HLEN or
+	 * 60 bytes if the skb->len is less than 60 for skb_pad.
+	 */
+	pull_len = igb_get_headlen(va, IGB_RX_HDR_LEN);
+
+	/* align pull length to size of long to optimize memcpy performance */
+	skb_copy_to_linear_data(skb, va, ALIGN(pull_len, sizeof(long)));
+
+	/* update all of the pointers */
+	skb_frag_size_sub(frag, pull_len);
+	frag->page_offset += pull_len;
+	skb->data_len -= pull_len;
+	skb->tail += pull_len;
+}
+
+/**
+ * igb_cleanup_headers - Correct corrupted or empty headers
+ * @rx_ring: rx descriptor ring packet is being transacted on
+ * @rx_desc: pointer to the EOP Rx descriptor
+ * @skb: pointer to current skb being fixed
+ *
+ * Address the case where we are pulling data in on pages only
+ * and as such no data is present in the skb header.
+ *
+ * In addition if skb is not at least 60 bytes we need to pad it so that
+ * it is large enough to qualify as a valid Ethernet frame.
+ *
+ * Returns true if an error was encountered and skb was freed.
+ **/
+static bool igb_cleanup_headers(struct igb_ring *rx_ring,
+				union e1000_adv_rx_desc *rx_desc,
+				struct sk_buff *skb)
+{
+
+	if (unlikely((igb_test_staterr(rx_desc,
+				       E1000_RXDEXT_ERR_FRAME_ERR_MASK)))) {
+		struct net_device *netdev = rx_ring->netdev;
+		if (!(netdev->features & NETIF_F_RXALL)) {
+			dev_kfree_skb_any(skb);
+			return true;
+		}
+	}
+
+	/* place header in linear portion of buffer */
+	if (skb_is_nonlinear(skb))
+		igb_pull_tail(rx_ring, rx_desc, skb);
+
+	/* if skb_pad returns an error the skb was freed */
+	if (unlikely(skb->len < 60)) {
+		int pad_len = 60 - skb->len;
+
+		if (skb_pad(skb, pad_len))
+			return true;
+		__skb_put(skb, pad_len);
+	}
+
+	return false;
 }
 
 static bool igb_clean_rx_irq(struct igb_q_vector *q_vector, int budget)
 {
 	struct igb_ring *rx_ring = q_vector->rx.ring;
 	union e1000_adv_rx_desc *rx_desc;
+	struct sk_buff *skb = rx_ring->skb;
 	const int current_node = numa_node_id();
 	unsigned int total_bytes = 0, total_packets = 0;
 	u16 cleaned_count = igb_desc_unused(rx_ring);
@@ -5932,12 +6115,9 @@ static bool igb_clean_rx_irq(struct igb_q_vector *q_vector, int budget)
 
 	while (igb_test_staterr(rx_desc, E1000_RXD_STAT_DD)) {
 		struct igb_rx_buffer *buffer_info = &rx_ring->rx_buffer_info[i];
-		struct sk_buff *skb = buffer_info->skb;
+		struct page *page;
 		union e1000_adv_rx_desc *next_rxd;
 
-		buffer_info->skb = NULL;
-		prefetch(skb->data);
-
 		i++;
 		if (i == rx_ring->count)
 			i = 0;
@@ -5952,52 +6132,57 @@ static bool igb_clean_rx_irq(struct igb_q_vector *q_vector, int budget)
 		 */
 		rmb();
 
-		if (!skb_is_nonlinear(skb)) {
-			__skb_put(skb, igb_get_hlen(rx_desc));
-			dma_unmap_single(rx_ring->dev, buffer_info->dma,
-					 IGB_RX_HDR_LEN,
-					 DMA_FROM_DEVICE);
-			buffer_info->dma = 0;
-		}
-
-		if (rx_desc->wb.upper.length) {
-			u16 length = le16_to_cpu(rx_desc->wb.upper.length);
+		page = buffer_info->page;
+		prefetchw(page);
 
-			skb_fill_page_desc(skb, skb_shinfo(skb)->nr_frags,
-						buffer_info->page,
-						buffer_info->page_offset,
-						length);
+		if (likely(!skb)) {
+			void *page_addr = page_address(page) +
+					  buffer_info->page_offset;
 
-			skb->len += length;
-			skb->data_len += length;
-			skb->truesize += PAGE_SIZE / 2;
+			/* prefetch first cache line of first page */
+			prefetch(page_addr);
+#if L1_CACHE_BYTES < 128
+			prefetch(page_addr + L1_CACHE_BYTES);
+#endif
 
-			if ((page_count(buffer_info->page) != 1) ||
-			    (page_to_nid(buffer_info->page) != current_node))
-				buffer_info->page = NULL;
-			else
-				get_page(buffer_info->page);
+			/* allocate a skb to store the frags */
+			skb = netdev_alloc_skb_ip_align(rx_ring->netdev,
+							IGB_RX_HDR_LEN);
+			if (unlikely(!skb)) {
+				rx_ring->rx_stats.alloc_failed++;
+				break;
+			}
 
-			dma_unmap_page(rx_ring->dev, buffer_info->page_dma,
-				       PAGE_SIZE / 2, DMA_FROM_DEVICE);
-			buffer_info->page_dma = 0;
+			/*
+			 * we will be copying header into skb->data in
+			 * pskb_may_pull so it is in our interest to prefetch
+			 * it now to avoid a possible cache miss
+			 */
+			prefetchw(skb->data);
 		}
 
-		if (!igb_test_staterr(rx_desc, E1000_RXD_STAT_EOP)) {
-			struct igb_rx_buffer *next_buffer;
-			next_buffer = &rx_ring->rx_buffer_info[i];
-			buffer_info->skb = next_buffer->skb;
-			buffer_info->dma = next_buffer->dma;
-			next_buffer->skb = skb;
-			next_buffer->dma = 0;
-			goto next_desc;
-		}
+		skb_add_rx_frag(skb, skb_shinfo(skb)->nr_frags, page,
+				buffer_info->page_offset,
+				le16_to_cpu(rx_desc->wb.upper.length),
+				PAGE_SIZE / 2);
 
-		if (unlikely((igb_test_staterr(rx_desc,
-					       E1000_RXDEXT_ERR_FRAME_ERR_MASK))
-			     && !(rx_ring->netdev->features & NETIF_F_RXALL))) {
-			dev_kfree_skb_any(skb);
+		if ((page_count(buffer_info->page) != 1) ||
+		    (page_to_nid(buffer_info->page) != current_node))
+			buffer_info->page = NULL;
+		else
+			get_page(buffer_info->page);
+
+		dma_unmap_page(rx_ring->dev, buffer_info->dma,
+			       PAGE_SIZE / 2, DMA_FROM_DEVICE);
+		buffer_info->dma = 0;
+
+		if (!igb_test_staterr(rx_desc, E1000_RXD_STAT_EOP))
 			goto next_desc;
+
+		/* verify the packet layout is correct */
+		if (igb_cleanup_headers(rx_ring, rx_desc, skb)) {
+			skb = NULL;
+			continue;
 		}
 
 #ifdef CONFIG_IGB_PTP
@@ -6010,10 +6195,14 @@ static bool igb_clean_rx_irq(struct igb_q_vector *q_vector, int budget)
 		total_bytes += skb->len;
 		total_packets++;
 
+		skb_record_rx_queue(skb, rx_ring->queue_index);
 		skb->protocol = eth_type_trans(skb, rx_ring->netdev);
 
 		napi_gro_receive(&q_vector->napi, skb);
 
+		/* reset skb pointer */
+		skb = NULL;
+
 		budget--;
 next_desc:
 		if (!budget)
@@ -6030,6 +6219,9 @@ next_desc:
 		rx_desc = next_rxd;
 	}
 
+	/* place incomplete frames back on ring for completion */
+	rx_ring->skb = skb;
+
 	rx_ring->next_to_clean = i;
 	u64_stats_update_begin(&rx_ring->rx_syncp);
 	rx_ring->rx_stats.packets += total_packets;
@@ -6044,70 +6236,37 @@ next_desc:
 	return !!budget;
 }
 
-static bool igb_alloc_mapped_skb(struct igb_ring *rx_ring,
-				 struct igb_rx_buffer *bi)
-{
-	struct sk_buff *skb = bi->skb;
-	dma_addr_t dma = bi->dma;
-
-	if (dma)
-		return true;
-
-	if (likely(!skb)) {
-		skb = netdev_alloc_skb_ip_align(rx_ring->netdev,
-						IGB_RX_HDR_LEN);
-		bi->skb = skb;
-		if (!skb) {
-			rx_ring->rx_stats.alloc_failed++;
-			return false;
-		}
-
-		/* initialize skb for ring */
-		skb_record_rx_queue(skb, rx_ring->queue_index);
-	}
-
-	dma = dma_map_single(rx_ring->dev, skb->data,
-			     IGB_RX_HDR_LEN, DMA_FROM_DEVICE);
-
-	if (dma_mapping_error(rx_ring->dev, dma)) {
-		rx_ring->rx_stats.alloc_failed++;
-		return false;
-	}
-
-	bi->dma = dma;
-	return true;
-}
-
 static bool igb_alloc_mapped_page(struct igb_ring *rx_ring,
 				  struct igb_rx_buffer *bi)
 {
 	struct page *page = bi->page;
-	dma_addr_t page_dma = bi->page_dma;
+	dma_addr_t dma = bi->dma;
 	unsigned int page_offset = bi->page_offset ^ (PAGE_SIZE / 2);
 
-	if (page_dma)
+	if (dma)
 		return true;
 
 	if (!page) {
-		page = __skb_alloc_page(GFP_ATOMIC, bi->skb);
-		bi->page = page;
+		page = __skb_alloc_page(GFP_ATOMIC | __GFP_COLD, NULL);
 		if (unlikely(!page)) {
 			rx_ring->rx_stats.alloc_failed++;
 			return false;
 		}
+		bi->page = page;
 	}
 
-	page_dma = dma_map_page(rx_ring->dev, page,
-				page_offset, PAGE_SIZE / 2,
-				DMA_FROM_DEVICE);
+	dma = dma_map_page(rx_ring->dev, page,
+			   page_offset, PAGE_SIZE / 2,
+			   DMA_FROM_DEVICE);
 
-	if (dma_mapping_error(rx_ring->dev, page_dma)) {
+	if (dma_mapping_error(rx_ring->dev, dma)) {
 		rx_ring->rx_stats.alloc_failed++;
 		return false;
 	}
 
-	bi->page_dma = page_dma;
+	bi->dma = dma;
 	bi->page_offset = page_offset;
+
 	return true;
 }
 
@@ -6126,17 +6285,12 @@ void igb_alloc_rx_buffers(struct igb_ring *rx_ring, u16 cleaned_count)
 	i -= rx_ring->count;
 
 	while (cleaned_count--) {
-		if (!igb_alloc_mapped_skb(rx_ring, bi))
+		if (!igb_alloc_mapped_page(rx_ring, bi))
 			break;
 
 		/* Refresh the desc even if buffer_addrs didn't change
 		 * because each write-back erases this info. */
-		rx_desc->read.hdr_addr = cpu_to_le64(bi->dma);
-
-		if (!igb_alloc_mapped_page(rx_ring, bi))
-			break;
-
-		rx_desc->read.pkt_addr = cpu_to_le64(bi->page_dma);
+		rx_desc->read.pkt_addr = cpu_to_le64(bi->dma);
 
 		rx_desc++;
 		bi++;
-- 
1.7.11.7

[net-next 08/14] igb: Combine post-processing of skb into a single function

[Jeff Kirsher]

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

This change is meant to just clean-up a number of function calls that were
made at the end of the Rx clean-up path by combining them into a single
function call.

Signed-off-by: Alexander Duyck <alexander.h.duyck@intel.com>
Tested-by: Aaron Brown <aaron.f.brown@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
---
 drivers/net/ethernet/intel/igb/igb_main.c | 69 ++++++++++++++++++++-----------
 1 file changed, 44 insertions(+), 25 deletions(-)

diff --git a/drivers/net/ethernet/intel/igb/igb_main.c b/drivers/net/ethernet/intel/igb/igb_main.c
index 665eafa..9e38f14 100644
--- a/drivers/net/ethernet/intel/igb/igb_main.c
+++ b/drivers/net/ethernet/intel/igb/igb_main.c
@@ -5877,22 +5877,6 @@ static inline void igb_rx_hash(struct igb_ring *ring,
 		skb->rxhash = le32_to_cpu(rx_desc->wb.lower.hi_dword.rss);
 }
 
-static void igb_rx_vlan(struct igb_ring *ring,
-			union e1000_adv_rx_desc *rx_desc,
-			struct sk_buff *skb)
-{
-	if (igb_test_staterr(rx_desc, E1000_RXD_STAT_VP)) {
-		u16 vid;
-		if (igb_test_staterr(rx_desc, E1000_RXDEXT_STATERR_LB) &&
-		    test_bit(IGB_RING_FLAG_RX_LB_VLAN_BSWAP, &ring->flags))
-			vid = be16_to_cpu(rx_desc->wb.upper.vlan);
-		else
-			vid = le16_to_cpu(rx_desc->wb.upper.vlan);
-
-		__vlan_hwaccel_put_tag(skb, vid);
-	}
-}
-
 /**
  * igb_get_headlen - determine size of header for LRO/GRO
  * @data: pointer to the start of the headers
@@ -6101,6 +6085,47 @@ static bool igb_cleanup_headers(struct igb_ring *rx_ring,
 	return false;
 }
 
+/**
+ * igb_process_skb_fields - Populate skb header fields from Rx descriptor
+ * @rx_ring: rx descriptor ring packet is being transacted on
+ * @rx_desc: pointer to the EOP Rx descriptor
+ * @skb: pointer to current skb being populated
+ *
+ * This function checks the ring, descriptor, and packet information in
+ * order to populate the hash, checksum, VLAN, timestamp, protocol, and
+ * other fields within the skb.
+ **/
+static void igb_process_skb_fields(struct igb_ring *rx_ring,
+				   union e1000_adv_rx_desc *rx_desc,
+				   struct sk_buff *skb)
+{
+	struct net_device *dev = rx_ring->netdev;
+
+	igb_rx_hash(rx_ring, rx_desc, skb);
+
+	igb_rx_checksum(rx_ring, rx_desc, skb);
+
+#ifdef CONFIG_IGB_PTP
+	igb_ptp_rx_hwtstamp(rx_ring->q_vector, rx_desc, skb);
+#endif /* CONFIG_IGB_PTP */
+
+	if ((dev->features & NETIF_F_HW_VLAN_RX) &&
+	    igb_test_staterr(rx_desc, E1000_RXD_STAT_VP)) {
+		u16 vid;
+		if (igb_test_staterr(rx_desc, E1000_RXDEXT_STATERR_LB) &&
+		    test_bit(IGB_RING_FLAG_RX_LB_VLAN_BSWAP, &rx_ring->flags))
+			vid = be16_to_cpu(rx_desc->wb.upper.vlan);
+		else
+			vid = le16_to_cpu(rx_desc->wb.upper.vlan);
+
+		__vlan_hwaccel_put_tag(skb, vid);
+	}
+
+	skb_record_rx_queue(skb, rx_ring->queue_index);
+
+	skb->protocol = eth_type_trans(skb, rx_ring->netdev);
+}
+
 static bool igb_clean_rx_irq(struct igb_q_vector *q_vector, int budget)
 {
 	struct igb_ring *rx_ring = q_vector->rx.ring;
@@ -6185,18 +6210,12 @@ static bool igb_clean_rx_irq(struct igb_q_vector *q_vector, int budget)
 			continue;
 		}
 
-#ifdef CONFIG_IGB_PTP
-		igb_ptp_rx_hwtstamp(q_vector, rx_desc, skb);
-#endif /* CONFIG_IGB_PTP */
-		igb_rx_hash(rx_ring, rx_desc, skb);
-		igb_rx_checksum(rx_ring, rx_desc, skb);
-		igb_rx_vlan(rx_ring, rx_desc, skb);
-
+		/* probably a little skewed due to removing CRC */
 		total_bytes += skb->len;
 		total_packets++;
 
-		skb_record_rx_queue(skb, rx_ring->queue_index);
-		skb->protocol = eth_type_trans(skb, rx_ring->netdev);
+		/* populate checksum, timestamp, VLAN, and protocol */
+		igb_process_skb_fields(rx_ring, rx_desc, skb);
 
 		napi_gro_receive(&q_vector->napi, skb);
 
-- 
1.7.11.7

[net-next 09/14] igb: Map entire page and sync half instead of mapping and unmapping half pages

[Jeff Kirsher]

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

This change makes it so that we map the entire page and just sync half of
it for the device at a time.  The advantage to this approach is that we can
avoid the locking on map/unmap seen in many IOMMU implementations.

Signed-off-by: Alexander Duyck <alexander.h.duyck@intel.com>
Tested-by: Aaron Brown <aaron.f.brown@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
---
 drivers/net/ethernet/intel/igb/igb.h         |   1 +
 drivers/net/ethernet/intel/igb/igb_ethtool.c |  26 ++--
 drivers/net/ethernet/intel/igb/igb_main.c    | 211 +++++++++++++++++++++------
 3 files changed, 181 insertions(+), 57 deletions(-)

diff --git a/drivers/net/ethernet/intel/igb/igb.h b/drivers/net/ethernet/intel/igb/igb.h
index 72ab9ac..1d15bb0 100644
--- a/drivers/net/ethernet/intel/igb/igb.h
+++ b/drivers/net/ethernet/intel/igb/igb.h
@@ -239,6 +239,7 @@ struct igb_ring {
 	/* everything past this point are written often */
 	u16 next_to_clean ____cacheline_aligned_in_smp;
 	u16 next_to_use;
+	u16 next_to_alloc;
 
 	union {
 		/* TX */
diff --git a/drivers/net/ethernet/intel/igb/igb_ethtool.c b/drivers/net/ethernet/intel/igb/igb_ethtool.c
index 0faac42..96c6df6 100644
--- a/drivers/net/ethernet/intel/igb/igb_ethtool.c
+++ b/drivers/net/ethernet/intel/igb/igb_ethtool.c
@@ -1694,7 +1694,7 @@ static int igb_check_lbtest_frame(struct igb_rx_buffer *rx_buffer,
 
 	frame_size >>= 1;
 
-	data = kmap(rx_buffer->page) + rx_buffer->page_offset;
+	data = kmap(rx_buffer->page);
 
 	if (data[3] != 0xFF ||
 	    data[frame_size + 10] != 0xBE ||
@@ -1713,9 +1713,7 @@ static int igb_clean_test_rings(struct igb_ring *rx_ring,
 	union e1000_adv_rx_desc *rx_desc;
 	struct igb_rx_buffer *rx_buffer_info;
 	struct igb_tx_buffer *tx_buffer_info;
-	struct netdev_queue *txq;
 	u16 rx_ntc, tx_ntc, count = 0;
-	unsigned int total_bytes = 0, total_packets = 0;
 
 	/* initialize next to clean and descriptor values */
 	rx_ntc = rx_ring->next_to_clean;
@@ -1726,21 +1724,24 @@ static int igb_clean_test_rings(struct igb_ring *rx_ring,
 		/* check rx buffer */
 		rx_buffer_info = &rx_ring->rx_buffer_info[rx_ntc];
 
-		/* unmap rx buffer, will be remapped by alloc_rx_buffers */
-		dma_unmap_single(rx_ring->dev,
-				 rx_buffer_info->dma,
-				 PAGE_SIZE / 2,
-				 DMA_FROM_DEVICE);
-		rx_buffer_info->dma = 0;
+		/* sync Rx buffer for CPU read */
+		dma_sync_single_for_cpu(rx_ring->dev,
+					rx_buffer_info->dma,
+					PAGE_SIZE / 2,
+					DMA_FROM_DEVICE);
 
 		/* verify contents of skb */
 		if (igb_check_lbtest_frame(rx_buffer_info, size))
 			count++;
 
+		/* sync Rx buffer for device write */
+		dma_sync_single_for_device(rx_ring->dev,
+					   rx_buffer_info->dma,
+					   PAGE_SIZE / 2,
+					   DMA_FROM_DEVICE);
+
 		/* unmap buffer on tx side */
 		tx_buffer_info = &tx_ring->tx_buffer_info[tx_ntc];
-		total_bytes += tx_buffer_info->bytecount;
-		total_packets += tx_buffer_info->gso_segs;
 		igb_unmap_and_free_tx_resource(tx_ring, tx_buffer_info);
 
 		/* increment rx/tx next to clean counters */
@@ -1755,8 +1756,7 @@ static int igb_clean_test_rings(struct igb_ring *rx_ring,
 		rx_desc = IGB_RX_DESC(rx_ring, rx_ntc);
 	}
 
-	txq = netdev_get_tx_queue(tx_ring->netdev, tx_ring->queue_index);
-	netdev_tx_completed_queue(txq, total_packets, total_bytes);
+	netdev_tx_reset_queue(txring_txq(tx_ring));
 
 	/* re-map buffers to ring, store next to clean values */
 	igb_alloc_rx_buffers(rx_ring, count);
diff --git a/drivers/net/ethernet/intel/igb/igb_main.c b/drivers/net/ethernet/intel/igb/igb_main.c
index 9e38f14..18ad18f 100644
--- a/drivers/net/ethernet/intel/igb/igb_main.c
+++ b/drivers/net/ethernet/intel/igb/igb_main.c
@@ -2785,6 +2785,7 @@ int igb_setup_rx_resources(struct igb_ring *rx_ring)
 	if (!rx_ring->desc)
 		goto err;
 
+	rx_ring->next_to_alloc = 0;
 	rx_ring->next_to_clean = 0;
 	rx_ring->next_to_use = 0;
 
@@ -3312,16 +3313,16 @@ static void igb_clean_rx_ring(struct igb_ring *rx_ring)
 	for (i = 0; i < rx_ring->count; i++) {
 		struct igb_rx_buffer *buffer_info = &rx_ring->rx_buffer_info[i];
 
-		if (buffer_info->dma)
-			dma_unmap_page(rx_ring->dev,
-				       buffer_info->dma,
-				       PAGE_SIZE / 2,
-				       DMA_FROM_DEVICE);
-		buffer_info->dma = 0;
-		if (buffer_info->page)
-			__free_page(buffer_info->page);
+		if (!buffer_info->page)
+			continue;
+
+		dma_unmap_page(rx_ring->dev,
+			       buffer_info->dma,
+			       PAGE_SIZE,
+			       DMA_FROM_DEVICE);
+		__free_page(buffer_info->page);
+
 		buffer_info->page = NULL;
-		buffer_info->page_offset = 0;
 	}
 
 	size = sizeof(struct igb_rx_buffer) * rx_ring->count;
@@ -3330,6 +3331,7 @@ static void igb_clean_rx_ring(struct igb_ring *rx_ring)
 	/* Zero out the descriptor ring */
 	memset(rx_ring->desc, 0, rx_ring->size);
 
+	rx_ring->next_to_alloc = 0;
 	rx_ring->next_to_clean = 0;
 	rx_ring->next_to_use = 0;
 }
@@ -5828,6 +5830,104 @@ static bool igb_clean_tx_irq(struct igb_q_vector *q_vector)
 	return !!budget;
 }
 
+/**
+ * igb_reuse_rx_page - page flip buffer and store it back on the ring
+ * @rx_ring: rx descriptor ring to store buffers on
+ * @old_buff: donor buffer to have page reused
+ *
+ * Synchronizes page for reuse by the adapter
+ **/
+static void igb_reuse_rx_page(struct igb_ring *rx_ring,
+			      struct igb_rx_buffer *old_buff)
+{
+	struct igb_rx_buffer *new_buff;
+	u16 nta = rx_ring->next_to_alloc;
+
+	new_buff = &rx_ring->rx_buffer_info[nta];
+
+	/* update, and store next to alloc */
+	nta++;
+	rx_ring->next_to_alloc = (nta < rx_ring->count) ? nta : 0;
+
+	/* transfer page from old buffer to new buffer */
+	memcpy(new_buff, old_buff, sizeof(struct igb_rx_buffer));
+
+	/* sync the buffer for use by the device */
+	dma_sync_single_range_for_device(rx_ring->dev, old_buff->dma,
+					 old_buff->page_offset,
+					 PAGE_SIZE / 2,
+					 DMA_FROM_DEVICE);
+}
+
+/**
+ * igb_add_rx_frag - Add contents of Rx buffer to sk_buff
+ * @rx_ring: rx descriptor ring to transact packets on
+ * @rx_buffer: buffer containing page to add
+ * @rx_desc: descriptor containing length of buffer written by hardware
+ * @skb: sk_buff to place the data into
+ *
+ * This function will add the data contained in rx_buffer->page to the skb.
+ * This is done either through a direct copy if the data in the buffer is
+ * less than the skb header size, otherwise it will just attach the page as
+ * a frag to the skb.
+ *
+ * The function will then update the page offset if necessary and return
+ * true if the buffer can be reused by the adapter.
+ **/
+static bool igb_add_rx_frag(struct igb_ring *rx_ring,
+			    struct igb_rx_buffer *rx_buffer,
+			    union e1000_adv_rx_desc *rx_desc,
+			    struct sk_buff *skb)
+{
+	struct page *page = rx_buffer->page;
+	unsigned int size = le16_to_cpu(rx_desc->wb.upper.length);
+
+	if ((size <= IGB_RX_HDR_LEN) && !skb_is_nonlinear(skb)) {
+		unsigned char *va = page_address(page) + rx_buffer->page_offset;
+
+#ifdef CONFIG_IGB_PTP
+		if (igb_test_staterr(rx_desc, E1000_RXDADV_STAT_TSIP)) {
+			igb_ptp_rx_pktstamp(rx_ring->q_vector, va, skb);
+			va += IGB_TS_HDR_LEN;
+			size -= IGB_TS_HDR_LEN;
+		}
+
+#endif
+		memcpy(__skb_put(skb, size), va, ALIGN(size, sizeof(long)));
+
+		/* we can reuse buffer as-is, just make sure it is local */
+		if (likely(page_to_nid(page) == numa_node_id()))
+			return true;
+
+		/* this page cannot be reused so discard it */
+		put_page(page);
+		return false;
+	}
+
+	skb_add_rx_frag(skb, skb_shinfo(skb)->nr_frags, page,
+			rx_buffer->page_offset, size, PAGE_SIZE / 2);
+
+	/* avoid re-using remote pages */
+	if (unlikely(page_to_nid(page) != numa_node_id()))
+		return false;
+
+	/* if we are only owner of page we can reuse it */
+	if (unlikely(page_count(page) != 1))
+		return false;
+
+	/* flip page offset to other buffer */
+	rx_buffer->page_offset ^= PAGE_SIZE / 2;
+
+	/*
+	 * since we are the only owner of the page and we need to
+	 * increment it, just set the value to 2 in order to avoid
+	 * an unnecessary locked operation
+	 */
+	atomic_set(&page->_count, 2);
+
+	return true;
+}
+
 static inline void igb_rx_checksum(struct igb_ring *ring,
 				   union e1000_adv_rx_desc *rx_desc,
 				   struct sk_buff *skb)
@@ -5985,6 +6085,7 @@ static unsigned int igb_get_headlen(unsigned char *data,
 /**
  * igb_pull_tail - igb specific version of skb_pull_tail
  * @rx_ring: rx descriptor ring packet is being transacted on
+ * @rx_desc: pointer to the EOP Rx descriptor
  * @skb: pointer to current skb being adjusted
  *
  * This function is an igb specific version of __pskb_pull_tail.  The
@@ -6131,7 +6232,6 @@ static bool igb_clean_rx_irq(struct igb_q_vector *q_vector, int budget)
 	struct igb_ring *rx_ring = q_vector->rx.ring;
 	union e1000_adv_rx_desc *rx_desc;
 	struct sk_buff *skb = rx_ring->skb;
-	const int current_node = numa_node_id();
 	unsigned int total_bytes = 0, total_packets = 0;
 	u16 cleaned_count = igb_desc_unused(rx_ring);
 	u16 i = rx_ring->next_to_clean;
@@ -6186,20 +6286,25 @@ static bool igb_clean_rx_irq(struct igb_q_vector *q_vector, int budget)
 			prefetchw(skb->data);
 		}
 
-		skb_add_rx_frag(skb, skb_shinfo(skb)->nr_frags, page,
-				buffer_info->page_offset,
-				le16_to_cpu(rx_desc->wb.upper.length),
-				PAGE_SIZE / 2);
-
-		if ((page_count(buffer_info->page) != 1) ||
-		    (page_to_nid(buffer_info->page) != current_node))
-			buffer_info->page = NULL;
-		else
-			get_page(buffer_info->page);
+		/* we are reusing so sync this buffer for CPU use */
+		dma_sync_single_range_for_cpu(rx_ring->dev,
+					      buffer_info->dma,
+					      buffer_info->page_offset,
+					      PAGE_SIZE / 2,
+					      DMA_FROM_DEVICE);
+
+		/* pull page into skb */
+		if (igb_add_rx_frag(rx_ring, buffer_info, rx_desc, skb)) {
+			/* hand second half of page back to the ring */
+			igb_reuse_rx_page(rx_ring, buffer_info);
+		} else {
+			/* we are not reusing the buffer so unmap it */
+			dma_unmap_page(rx_ring->dev, buffer_info->dma,
+				       PAGE_SIZE, DMA_FROM_DEVICE);
+		}
 
-		dma_unmap_page(rx_ring->dev, buffer_info->dma,
-			       PAGE_SIZE / 2, DMA_FROM_DEVICE);
-		buffer_info->dma = 0;
+		/* clear contents of buffer_info */
+		buffer_info->page = NULL;
 
 		if (!igb_test_staterr(rx_desc, E1000_RXD_STAT_EOP))
 			goto next_desc;
@@ -6259,32 +6364,36 @@ static bool igb_alloc_mapped_page(struct igb_ring *rx_ring,
 				  struct igb_rx_buffer *bi)
 {
 	struct page *page = bi->page;
-	dma_addr_t dma = bi->dma;
-	unsigned int page_offset = bi->page_offset ^ (PAGE_SIZE / 2);
+	dma_addr_t dma;
 
-	if (dma)
+	/* since we are recycling buffers we should seldom need to alloc */
+	if (likely(page))
 		return true;
 
-	if (!page) {
-		page = __skb_alloc_page(GFP_ATOMIC | __GFP_COLD, NULL);
-		if (unlikely(!page)) {
-			rx_ring->rx_stats.alloc_failed++;
-			return false;
-		}
-		bi->page = page;
+	/* alloc new page for storage */
+	page = __skb_alloc_page(GFP_ATOMIC | __GFP_COLD, NULL);
+	if (unlikely(!page)) {
+		rx_ring->rx_stats.alloc_failed++;
+		return false;
 	}
 
-	dma = dma_map_page(rx_ring->dev, page,
-			   page_offset, PAGE_SIZE / 2,
-			   DMA_FROM_DEVICE);
+	/* map page for use */
+	dma = dma_map_page(rx_ring->dev, page, 0, PAGE_SIZE, DMA_FROM_DEVICE);
 
+	/*
+	 * if mapping failed free memory back to system since
+	 * there isn't much point in holding memory we can't use
+	 */
 	if (dma_mapping_error(rx_ring->dev, dma)) {
+		__free_page(page);
+
 		rx_ring->rx_stats.alloc_failed++;
 		return false;
 	}
 
 	bi->dma = dma;
-	bi->page_offset = page_offset;
+	bi->page = page;
+	bi->page_offset = 0;
 
 	return true;
 }
@@ -6299,17 +6408,23 @@ void igb_alloc_rx_buffers(struct igb_ring *rx_ring, u16 cleaned_count)
 	struct igb_rx_buffer *bi;
 	u16 i = rx_ring->next_to_use;
 
+	/* nothing to do */
+	if (!cleaned_count)
+		return;
+
 	rx_desc = IGB_RX_DESC(rx_ring, i);
 	bi = &rx_ring->rx_buffer_info[i];
 	i -= rx_ring->count;
 
-	while (cleaned_count--) {
+	do {
 		if (!igb_alloc_mapped_page(rx_ring, bi))
 			break;
 
-		/* Refresh the desc even if buffer_addrs didn't change
-		 * because each write-back erases this info. */
-		rx_desc->read.pkt_addr = cpu_to_le64(bi->dma);
+		/*
+		 * Refresh the desc even if buffer_addrs didn't change
+		 * because each write-back erases this info.
+		 */
+		rx_desc->read.pkt_addr = cpu_to_le64(bi->dma + bi->page_offset);
 
 		rx_desc++;
 		bi++;
@@ -6322,17 +6437,25 @@ void igb_alloc_rx_buffers(struct igb_ring *rx_ring, u16 cleaned_count)
 
 		/* clear the hdr_addr for the next_to_use descriptor */
 		rx_desc->read.hdr_addr = 0;
-	}
+
+		cleaned_count--;
+	} while (cleaned_count);
 
 	i += rx_ring->count;
 
 	if (rx_ring->next_to_use != i) {
+		/* record the next descriptor to use */
 		rx_ring->next_to_use = i;
 
-		/* Force memory writes to complete before letting h/w
+		/* update next to alloc since we have filled the ring */
+		rx_ring->next_to_alloc = i;
+
+		/*
+		 * Force memory writes to complete before letting h/w
 		 * know there are new descriptors to fetch.  (Only
 		 * applicable for weak-ordered memory model archs,
-		 * such as IA-64). */
+		 * such as IA-64).
+		 */
 		wmb();
 		writel(i, rx_ring->tail);
 	}
-- 
1.7.11.7

[net-next 10/14] igb: Move rx_buffer related code in Rx cleanup path into separate function

[Jeff Kirsher]

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

In order to try and isolate things a bit further I am moving the code
related to retrieving data from the rx_buffer_info structure into a
separate function.

Signed-off-by: Alexander Duyck <alexander.h.duyck@intel.com>
Tested-by: Aaron Brown <aaron.f.brown@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
---
 drivers/net/ethernet/intel/igb/igb_main.c | 206 +++++++++++++++++-------------
 1 file changed, 120 insertions(+), 86 deletions(-)

diff --git a/drivers/net/ethernet/intel/igb/igb_main.c b/drivers/net/ethernet/intel/igb/igb_main.c
index 18ad18f..fa7ddec 100644
--- a/drivers/net/ethernet/intel/igb/igb_main.c
+++ b/drivers/net/ethernet/intel/igb/igb_main.c
@@ -5928,6 +5928,74 @@ static bool igb_add_rx_frag(struct igb_ring *rx_ring,
 	return true;
 }
 
+static struct sk_buff *igb_fetch_rx_buffer(struct igb_ring *rx_ring,
+					   union e1000_adv_rx_desc *rx_desc,
+					   struct sk_buff *skb)
+{
+	struct igb_rx_buffer *rx_buffer;
+	struct page *page;
+
+	rx_buffer = &rx_ring->rx_buffer_info[rx_ring->next_to_clean];
+
+	/*
+	 * This memory barrier is needed to keep us from reading
+	 * any other fields out of the rx_desc until we know the
+	 * RXD_STAT_DD bit is set
+	 */
+	rmb();
+
+	page = rx_buffer->page;
+	prefetchw(page);
+
+	if (likely(!skb)) {
+		void *page_addr = page_address(page) +
+				  rx_buffer->page_offset;
+
+		/* prefetch first cache line of first page */
+		prefetch(page_addr);
+#if L1_CACHE_BYTES < 128
+		prefetch(page_addr + L1_CACHE_BYTES);
+#endif
+
+		/* allocate a skb to store the frags */
+		skb = netdev_alloc_skb_ip_align(rx_ring->netdev,
+						IGB_RX_HDR_LEN);
+		if (unlikely(!skb)) {
+			rx_ring->rx_stats.alloc_failed++;
+			return NULL;
+		}
+
+		/*
+		 * we will be copying header into skb->data in
+		 * pskb_may_pull so it is in our interest to prefetch
+		 * it now to avoid a possible cache miss
+		 */
+		prefetchw(skb->data);
+	}
+
+	/* we are reusing so sync this buffer for CPU use */
+	dma_sync_single_range_for_cpu(rx_ring->dev,
+				      rx_buffer->dma,
+				      rx_buffer->page_offset,
+				      PAGE_SIZE / 2,
+				      DMA_FROM_DEVICE);
+
+	/* pull page into skb */
+	if (igb_add_rx_frag(rx_ring, rx_buffer, rx_desc, skb)) {
+		/* hand second half of page back to the ring */
+		igb_reuse_rx_page(rx_ring, rx_buffer);
+	} else {
+		/* we are not reusing the buffer so unmap it */
+		dma_unmap_page(rx_ring->dev, rx_buffer->dma,
+			       PAGE_SIZE, DMA_FROM_DEVICE);
+	}
+
+	/* clear contents of rx_buffer */
+	rx_buffer->page = NULL;
+
+	return skb;
+}
+
 static inline void igb_rx_checksum(struct igb_ring *ring,
 				   union e1000_adv_rx_desc *rx_desc,
 				   struct sk_buff *skb)
@@ -5978,6 +6046,34 @@ static inline void igb_rx_hash(struct igb_ring *ring,
 }
 
 /**
+ * igb_is_non_eop - process handling of non-EOP buffers
+ * @rx_ring: Rx ring being processed
+ * @rx_desc: Rx descriptor for current buffer
+ * @skb: current socket buffer containing buffer in progress
+ *
+ * This function updates next to clean.  If the buffer is an EOP buffer
+ * this function exits returning false, otherwise it will place the
+ * sk_buff in the next buffer to be chained and return true indicating
+ * that this is in fact a non-EOP buffer.
+ **/
+static bool igb_is_non_eop(struct igb_ring *rx_ring,
+			   union e1000_adv_rx_desc *rx_desc)
+{
+	u32 ntc = rx_ring->next_to_clean + 1;
+
+	/* fetch, update, and store next to clean */
+	ntc = (ntc < rx_ring->count) ? ntc : 0;
+	rx_ring->next_to_clean = ntc;
+
+	prefetch(IGB_RX_DESC(rx_ring, ntc));
+
+	if (likely(igb_test_staterr(rx_desc, E1000_RXD_STAT_EOP)))
+		return false;
+
+	return true;
+}
+
+/**
  * igb_get_headlen - determine size of header for LRO/GRO
  * @data: pointer to the start of the headers
  * @max_len: total length of section to find headers in
@@ -6227,87 +6323,39 @@ static void igb_process_skb_fields(struct igb_ring *rx_ring,
 	skb->protocol = eth_type_trans(skb, rx_ring->netdev);
 }
 
-static bool igb_clean_rx_irq(struct igb_q_vector *q_vector, int budget)
+static bool igb_clean_rx_irq(struct igb_q_vector *q_vector, const int budget)
 {
 	struct igb_ring *rx_ring = q_vector->rx.ring;
-	union e1000_adv_rx_desc *rx_desc;
 	struct sk_buff *skb = rx_ring->skb;
 	unsigned int total_bytes = 0, total_packets = 0;
 	u16 cleaned_count = igb_desc_unused(rx_ring);
-	u16 i = rx_ring->next_to_clean;
-
-	rx_desc = IGB_RX_DESC(rx_ring, i);
-
-	while (igb_test_staterr(rx_desc, E1000_RXD_STAT_DD)) {
-		struct igb_rx_buffer *buffer_info = &rx_ring->rx_buffer_info[i];
-		struct page *page;
-		union e1000_adv_rx_desc *next_rxd;
-
-		i++;
-		if (i == rx_ring->count)
-			i = 0;
-
-		next_rxd = IGB_RX_DESC(rx_ring, i);
-		prefetch(next_rxd);
 
-		/*
-		 * This memory barrier is needed to keep us from reading
-		 * any other fields out of the rx_desc until we know the
-		 * RXD_STAT_DD bit is set
-		 */
-		rmb();
-
-		page = buffer_info->page;
-		prefetchw(page);
+	do {
+		union e1000_adv_rx_desc *rx_desc;
 
-		if (likely(!skb)) {
-			void *page_addr = page_address(page) +
-					  buffer_info->page_offset;
+		/* return some buffers to hardware, one at a time is too slow */
+		if (cleaned_count >= IGB_RX_BUFFER_WRITE) {
+			igb_alloc_rx_buffers(rx_ring, cleaned_count);
+			cleaned_count = 0;
+		}
 
-			/* prefetch first cache line of first page */
-			prefetch(page_addr);
-#if L1_CACHE_BYTES < 128
-			prefetch(page_addr + L1_CACHE_BYTES);
-#endif
+		rx_desc = IGB_RX_DESC(rx_ring, rx_ring->next_to_clean);
 
-			/* allocate a skb to store the frags */
-			skb = netdev_alloc_skb_ip_align(rx_ring->netdev,
-							IGB_RX_HDR_LEN);
-			if (unlikely(!skb)) {
-				rx_ring->rx_stats.alloc_failed++;
-				break;
-			}
+		if (!igb_test_staterr(rx_desc, E1000_RXD_STAT_DD))
+			break;
 
-			/*
-			 * we will be copying header into skb->data in
-			 * pskb_may_pull so it is in our interest to prefetch
-			 * it now to avoid a possible cache miss
-			 */
-			prefetchw(skb->data);
-		}
+		/* retrieve a buffer from the ring */
+		skb = igb_fetch_rx_buffer(rx_ring, rx_desc, skb);
 
-		/* we are reusing so sync this buffer for CPU use */
-		dma_sync_single_range_for_cpu(rx_ring->dev,
-					      buffer_info->dma,
-					      buffer_info->page_offset,
-					      PAGE_SIZE / 2,
-					      DMA_FROM_DEVICE);
-
-		/* pull page into skb */
-		if (igb_add_rx_frag(rx_ring, buffer_info, rx_desc, skb)) {
-			/* hand second half of page back to the ring */
-			igb_reuse_rx_page(rx_ring, buffer_info);
-		} else {
-			/* we are not reusing the buffer so unmap it */
-			dma_unmap_page(rx_ring->dev, buffer_info->dma,
-				       PAGE_SIZE, DMA_FROM_DEVICE);
-		}
+		/* exit if we failed to retrieve a buffer */
+		if (!skb)
+			break;
 
-		/* clear contents of buffer_info */
-		buffer_info->page = NULL;
+		cleaned_count++;
 
-		if (!igb_test_staterr(rx_desc, E1000_RXD_STAT_EOP))
-			goto next_desc;
+		/* fetch next buffer in frame if non-eop */
+		if (igb_is_non_eop(rx_ring, rx_desc))
+			continue;
 
 		/* verify the packet layout is correct */
 		if (igb_cleanup_headers(rx_ring, rx_desc, skb)) {
@@ -6317,7 +6365,6 @@ static bool igb_clean_rx_irq(struct igb_q_vector *q_vector, int budget)
 
 		/* probably a little skewed due to removing CRC */
 		total_bytes += skb->len;
-		total_packets++;
 
 		/* populate checksum, timestamp, VLAN, and protocol */
 		igb_process_skb_fields(rx_ring, rx_desc, skb);
@@ -6327,26 +6374,13 @@ static bool igb_clean_rx_irq(struct igb_q_vector *q_vector, int budget)
 		/* reset skb pointer */
 		skb = NULL;
 
-		budget--;
-next_desc:
-		if (!budget)
-			break;
-
-		cleaned_count++;
-		/* return some buffers to hardware, one at a time is too slow */
-		if (cleaned_count >= IGB_RX_BUFFER_WRITE) {
-			igb_alloc_rx_buffers(rx_ring, cleaned_count);
-			cleaned_count = 0;
-		}
-
-		/* use prefetched values */
-		rx_desc = next_rxd;
-	}
+		/* update budget accounting */
+		total_packets++;
+	} while (likely(total_packets < budget));
 
 	/* place incomplete frames back on ring for completion */
 	rx_ring->skb = skb;
 
-	rx_ring->next_to_clean = i;
 	u64_stats_update_begin(&rx_ring->rx_syncp);
 	rx_ring->rx_stats.packets += total_packets;
 	rx_ring->rx_stats.bytes += total_bytes;
@@ -6357,7 +6391,7 @@ next_desc:
 	if (cleaned_count)
 		igb_alloc_rx_buffers(rx_ring, cleaned_count);
 
-	return !!budget;
+	return (total_packets < budget);
 }
 
 static bool igb_alloc_mapped_page(struct igb_ring *rx_ring,
-- 
1.7.11.7

[net-next 11/14] igb: Lock buffer size at 2K even on systems with larger pages

[Jeff Kirsher]

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

This change locks us in at 2K buffers even on a system that supports larger
frames.  The reason for this change is to make better use of pages and to
reduce the overall truesize of frames generated by igb.

Signed-off-by: Alexander Duyck <alexander.h.duyck@intel.com>
Tested-by: Aaron Brown <aaron.f.brown@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
---
 drivers/net/ethernet/intel/igb/igb.h         |  7 ++++---
 drivers/net/ethernet/intel/igb/igb_ethtool.c |  4 ++--
 drivers/net/ethernet/intel/igb/igb_main.c    | 27 +++++++++++++++++----------
 3 files changed, 23 insertions(+), 15 deletions(-)

diff --git a/drivers/net/ethernet/intel/igb/igb.h b/drivers/net/ethernet/intel/igb/igb.h
index 1d15bb0..d3fd012 100644
--- a/drivers/net/ethernet/intel/igb/igb.h
+++ b/drivers/net/ethernet/intel/igb/igb.h
@@ -132,9 +132,10 @@ struct vf_data_storage {
 #define MAXIMUM_ETHERNET_VLAN_SIZE 1522
 
 /* Supported Rx Buffer Sizes */
-#define IGB_RXBUFFER_256   256
-#define IGB_RXBUFFER_16384 16384
-#define IGB_RX_HDR_LEN     IGB_RXBUFFER_256
+#define IGB_RXBUFFER_256	256
+#define IGB_RXBUFFER_2048	2048
+#define IGB_RX_HDR_LEN		IGB_RXBUFFER_256
+#define IGB_RX_BUFSZ		IGB_RXBUFFER_2048
 
 /* How many Tx Descriptors do we need to call netif_wake_queue ? */
 #define IGB_TX_QUEUE_WAKE	16
diff --git a/drivers/net/ethernet/intel/igb/igb_ethtool.c b/drivers/net/ethernet/intel/igb/igb_ethtool.c
index 96c6df6..375c0da 100644
--- a/drivers/net/ethernet/intel/igb/igb_ethtool.c
+++ b/drivers/net/ethernet/intel/igb/igb_ethtool.c
@@ -1727,7 +1727,7 @@ static int igb_clean_test_rings(struct igb_ring *rx_ring,
 		/* sync Rx buffer for CPU read */
 		dma_sync_single_for_cpu(rx_ring->dev,
 					rx_buffer_info->dma,
-					PAGE_SIZE / 2,
+					IGB_RX_BUFSZ,
 					DMA_FROM_DEVICE);
 
 		/* verify contents of skb */
@@ -1737,7 +1737,7 @@ static int igb_clean_test_rings(struct igb_ring *rx_ring,
 		/* sync Rx buffer for device write */
 		dma_sync_single_for_device(rx_ring->dev,
 					   rx_buffer_info->dma,
-					   PAGE_SIZE / 2,
+					   IGB_RX_BUFSZ,
 					   DMA_FROM_DEVICE);
 
 		/* unmap buffer on tx side */
diff --git a/drivers/net/ethernet/intel/igb/igb_main.c b/drivers/net/ethernet/intel/igb/igb_main.c
index fa7ddec..0141ef3 100644
--- a/drivers/net/ethernet/intel/igb/igb_main.c
+++ b/drivers/net/ethernet/intel/igb/igb_main.c
@@ -554,7 +554,7 @@ rx_ring_summary:
 					  16, 1,
 					  page_address(buffer_info->page) +
 						      buffer_info->page_offset,
-					  PAGE_SIZE/2, true);
+					  IGB_RX_BUFSZ, true);
 				}
 			}
 		}
@@ -3103,11 +3103,7 @@ void igb_configure_rx_ring(struct igb_adapter *adapter,
 
 	/* set descriptor configuration */
 	srrctl = IGB_RX_HDR_LEN << E1000_SRRCTL_BSIZEHDRSIZE_SHIFT;
-#if (PAGE_SIZE / 2) > IGB_RXBUFFER_16384
-	srrctl |= IGB_RXBUFFER_16384 >> E1000_SRRCTL_BSIZEPKT_SHIFT;
-#else
-	srrctl |= (PAGE_SIZE / 2) >> E1000_SRRCTL_BSIZEPKT_SHIFT;
-#endif
+	srrctl |= IGB_RX_BUFSZ >> E1000_SRRCTL_BSIZEPKT_SHIFT;
 	srrctl |= E1000_SRRCTL_DESCTYPE_ADV_ONEBUF;
 #ifdef CONFIG_IGB_PTP
 	if (hw->mac.type >= e1000_82580)
@@ -5855,7 +5851,7 @@ static void igb_reuse_rx_page(struct igb_ring *rx_ring,
 	/* sync the buffer for use by the device */
 	dma_sync_single_range_for_device(rx_ring->dev, old_buff->dma,
 					 old_buff->page_offset,
-					 PAGE_SIZE / 2,
+					 IGB_RX_BUFSZ,
 					 DMA_FROM_DEVICE);
 }
 
@@ -5905,18 +5901,19 @@ static bool igb_add_rx_frag(struct igb_ring *rx_ring,
 	}
 
 	skb_add_rx_frag(skb, skb_shinfo(skb)->nr_frags, page,
-			rx_buffer->page_offset, size, PAGE_SIZE / 2);
+			rx_buffer->page_offset, size, IGB_RX_BUFSZ);
 
 	/* avoid re-using remote pages */
 	if (unlikely(page_to_nid(page) != numa_node_id()))
 		return false;
 
+#if (PAGE_SIZE < 8192)
 	/* if we are only owner of page we can reuse it */
 	if (unlikely(page_count(page) != 1))
 		return false;
 
 	/* flip page offset to other buffer */
-	rx_buffer->page_offset ^= PAGE_SIZE / 2;
+	rx_buffer->page_offset ^= IGB_RX_BUFSZ;
 
 	/*
 	 * since we are the only owner of the page and we need to
@@ -5924,6 +5921,16 @@ static bool igb_add_rx_frag(struct igb_ring *rx_ring,
 	 * an unnecessary locked operation
 	 */
 	atomic_set(&page->_count, 2);
+#else
+	/* move offset up to the next cache line */
+	rx_buffer->page_offset += SKB_DATA_ALIGN(size);
+
+	if (rx_buffer->page_offset > (PAGE_SIZE - IGB_RX_BUFSZ))
+		return false;
+
+	/* bump ref count on page before it is given to the stack */
+	get_page(page);
+#endif
 
 	return true;
 }
@@ -5977,7 +5984,7 @@ static struct sk_buff *igb_fetch_rx_buffer(struct igb_ring *rx_ring,
 	dma_sync_single_range_for_cpu(rx_ring->dev,
 				      rx_buffer->dma,
 				      rx_buffer->page_offset,
-				      PAGE_SIZE / 2,
+				      IGB_RX_BUFSZ,
 				      DMA_FROM_DEVICE);
 
 	/* pull page into skb */
-- 
1.7.11.7

[net-next 12/14] igb: Combine q_vector and ring allocation into a single function

[Jeff Kirsher]

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

This change combines the the allocation of q_vectors and rings into a single
function.  The advantage of this is that we are guaranteed we will avoid
overlap in the L1 cache sets.

Signed-off-by: Alexander Duyck <alexander.h.duyck@intel.com>
Tested-by: Aaron Brown <aaron.f.brown@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
---
 drivers/net/ethernet/intel/igb/igb.h      |  42 ++--
 drivers/net/ethernet/intel/igb/igb_main.c | 375 +++++++++++++++---------------
 2 files changed, 215 insertions(+), 202 deletions(-)

diff --git a/drivers/net/ethernet/intel/igb/igb.h b/drivers/net/ethernet/intel/igb/igb.h
index d3fd012..be1971b 100644
--- a/drivers/net/ethernet/intel/igb/igb.h
+++ b/drivers/net/ethernet/intel/igb/igb.h
@@ -204,22 +204,6 @@ struct igb_ring_container {
 	u8 itr;				/* current ITR setting for ring */
 };
 
-struct igb_q_vector {
-	struct igb_adapter *adapter;	/* backlink */
-	int cpu;			/* CPU for DCA */
-	u32 eims_value;			/* EIMS mask value */
-
-	struct igb_ring_container rx, tx;
-
-	struct napi_struct napi;
-
-	u16 itr_val;
-	u8 set_itr;
-	void __iomem *itr_register;
-
-	char name[IFNAMSIZ + 9];
-};
-
 struct igb_ring {
 	struct igb_q_vector *q_vector;	/* backlink to q_vector */
 	struct net_device *netdev;	/* back pointer to net_device */
@@ -231,14 +215,15 @@ struct igb_ring {
 	void *desc;			/* descriptor ring memory */
 	unsigned long flags;		/* ring specific flags */
 	void __iomem *tail;		/* pointer to ring tail register */
+	dma_addr_t dma;			/* phys address of the ring */
+	unsigned int  size;		/* length of desc. ring in bytes */
 
 	u16 count;			/* number of desc. in the ring */
 	u8 queue_index;			/* logical index of the ring*/
 	u8 reg_idx;			/* physical index of the ring */
-	u32 size;			/* length of desc. ring in bytes */
 
 	/* everything past this point are written often */
-	u16 next_to_clean ____cacheline_aligned_in_smp;
+	u16 next_to_clean;
 	u16 next_to_use;
 	u16 next_to_alloc;
 
@@ -256,8 +241,25 @@ struct igb_ring {
 			struct u64_stats_sync rx_syncp;
 		};
 	};
-	/* Items past this point are only used during ring alloc / free */
-	dma_addr_t dma;                /* phys address of the ring */
+} ____cacheline_internodealigned_in_smp;
+
+struct igb_q_vector {
+	struct igb_adapter *adapter;	/* backlink */
+	int cpu;			/* CPU for DCA */
+	u32 eims_value;			/* EIMS mask value */
+
+	u16 itr_val;
+	u8 set_itr;
+	void __iomem *itr_register;
+
+	struct igb_ring_container rx, tx;
+
+	struct napi_struct napi;
+	struct rcu_head rcu;	/* to avoid race with update stats on free */
+	char name[IFNAMSIZ + 9];
+
+	/* for dynamic allocation of rings associated with this q_vector */
+	struct igb_ring ring[0] ____cacheline_internodealigned_in_smp;
 };
 
 enum e1000_ring_flags_t {
diff --git a/drivers/net/ethernet/intel/igb/igb_main.c b/drivers/net/ethernet/intel/igb/igb_main.c
index 0141ef3..4a25b8f 100644
--- a/drivers/net/ethernet/intel/igb/igb_main.c
+++ b/drivers/net/ethernet/intel/igb/igb_main.c
@@ -652,80 +652,6 @@ static void igb_cache_ring_register(struct igb_adapter *adapter)
 	}
 }
 
-static void igb_free_queues(struct igb_adapter *adapter)
-{
-	int i;
-
-	for (i = 0; i < adapter->num_tx_queues; i++) {
-		kfree(adapter->tx_ring[i]);
-		adapter->tx_ring[i] = NULL;
-	}
-	for (i = 0; i < adapter->num_rx_queues; i++) {
-		kfree(adapter->rx_ring[i]);
-		adapter->rx_ring[i] = NULL;
-	}
-	adapter->num_rx_queues = 0;
-	adapter->num_tx_queues = 0;
-}
-
-/**
- * igb_alloc_queues - Allocate memory for all rings
- * @adapter: board private structure to initialize
- *
- * We allocate one ring per queue at run-time since we don't know the
- * number of queues at compile-time.
- **/
-static int igb_alloc_queues(struct igb_adapter *adapter)
-{
-	struct igb_ring *ring;
-	int i;
-
-	for (i = 0; i < adapter->num_tx_queues; i++) {
-		ring = kzalloc(sizeof(struct igb_ring), GFP_KERNEL);
-		if (!ring)
-			goto err;
-		ring->count = adapter->tx_ring_count;
-		ring->queue_index = i;
-		ring->dev = &adapter->pdev->dev;
-		ring->netdev = adapter->netdev;
-		/* For 82575, context index must be unique per ring. */
-		if (adapter->hw.mac.type == e1000_82575)
-			set_bit(IGB_RING_FLAG_TX_CTX_IDX, &ring->flags);
-		adapter->tx_ring[i] = ring;
-	}
-
-	for (i = 0; i < adapter->num_rx_queues; i++) {
-		ring = kzalloc(sizeof(struct igb_ring), GFP_KERNEL);
-		if (!ring)
-			goto err;
-		ring->count = adapter->rx_ring_count;
-		ring->queue_index = i;
-		ring->dev = &adapter->pdev->dev;
-		ring->netdev = adapter->netdev;
-		/* set flag indicating ring supports SCTP checksum offload */
-		if (adapter->hw.mac.type >= e1000_82576)
-			set_bit(IGB_RING_FLAG_RX_SCTP_CSUM, &ring->flags);
-
-		/*
-		 * On i350, i210, and i211, loopback VLAN packets
-		 * have the tag byte-swapped.
-		 * */
-		if (adapter->hw.mac.type >= e1000_i350)
-			set_bit(IGB_RING_FLAG_RX_LB_VLAN_BSWAP, &ring->flags);
-
-		adapter->rx_ring[i] = ring;
-	}
-
-	igb_cache_ring_register(adapter);
-
-	return 0;
-
-err:
-	igb_free_queues(adapter);
-
-	return -ENOMEM;
-}
-
 /**
  *  igb_write_ivar - configure ivar for given MSI-X vector
  *  @hw: pointer to the HW structure
@@ -956,6 +882,35 @@ static void igb_reset_interrupt_capability(struct igb_adapter *adapter)
 }
 
 /**
+ * igb_free_q_vector - Free memory allocated for specific interrupt vector
+ * @adapter: board private structure to initialize
+ * @v_idx: Index of vector to be freed
+ *
+ * This function frees the memory allocated to the q_vector.  In addition if
+ * NAPI is enabled it will delete any references to the NAPI struct prior
+ * to freeing the q_vector.
+ **/
+static void igb_free_q_vector(struct igb_adapter *adapter, int v_idx)
+{
+	struct igb_q_vector *q_vector = adapter->q_vector[v_idx];
+
+	if (q_vector->tx.ring)
+		adapter->tx_ring[q_vector->tx.ring->queue_index] = NULL;
+
+	if (q_vector->rx.ring)
+		adapter->tx_ring[q_vector->rx.ring->queue_index] = NULL;
+
+	adapter->q_vector[v_idx] = NULL;
+	netif_napi_del(&q_vector->napi);
+
+	/*
+	 * ixgbe_get_stats64() might access the rings on this vector,
+	 * we must wait a grace period before freeing it.
+	 */
+	kfree_rcu(q_vector, rcu);
+}
+
+/**
  * igb_free_q_vectors - Free memory allocated for interrupt vectors
  * @adapter: board private structure to initialize
  *
@@ -965,17 +920,14 @@ static void igb_reset_interrupt_capability(struct igb_adapter *adapter)
  **/
 static void igb_free_q_vectors(struct igb_adapter *adapter)
 {
-	int v_idx;
+	int v_idx = adapter->num_q_vectors;
 
-	for (v_idx = 0; v_idx < adapter->num_q_vectors; v_idx++) {
-		struct igb_q_vector *q_vector = adapter->q_vector[v_idx];
-		adapter->q_vector[v_idx] = NULL;
-		if (!q_vector)
-			continue;
-		netif_napi_del(&q_vector->napi);
-		kfree(q_vector);
-	}
+	adapter->num_tx_queues = 0;
+	adapter->num_rx_queues = 0;
 	adapter->num_q_vectors = 0;
+
+	while (v_idx--)
+		igb_free_q_vector(adapter, v_idx);
 }
 
 /**
@@ -986,7 +938,6 @@ static void igb_free_q_vectors(struct igb_adapter *adapter)
  */
 static void igb_clear_interrupt_scheme(struct igb_adapter *adapter)
 {
-	igb_free_queues(adapter);
 	igb_free_q_vectors(adapter);
 	igb_reset_interrupt_capability(adapter);
 }
@@ -1074,95 +1025,181 @@ out:
 	return err;
 }
 
+static void igb_add_ring(struct igb_ring *ring,
+			 struct igb_ring_container *head)
+{
+	head->ring = ring;
+	head->count++;
+}
+
 /**
- * igb_alloc_q_vectors - Allocate memory for interrupt vectors
+ * igb_alloc_q_vector - Allocate memory for a single interrupt vector
  * @adapter: board private structure to initialize
+ * @v_count: q_vectors allocated on adapter, used for ring interleaving
+ * @v_idx: index of vector in adapter struct
+ * @txr_count: total number of Tx rings to allocate
+ * @txr_idx: index of first Tx ring to allocate
+ * @rxr_count: total number of Rx rings to allocate
+ * @rxr_idx: index of first Rx ring to allocate
  *
- * We allocate one q_vector per queue interrupt.  If allocation fails we
- * return -ENOMEM.
+ * We allocate one q_vector.  If allocation fails we return -ENOMEM.
  **/
-static int igb_alloc_q_vectors(struct igb_adapter *adapter)
+static int igb_alloc_q_vector(struct igb_adapter *adapter,
+			      int v_count, int v_idx,
+			      int txr_count, int txr_idx,
+			      int rxr_count, int rxr_idx)
 {
 	struct igb_q_vector *q_vector;
-	struct e1000_hw *hw = &adapter->hw;
-	int v_idx;
+	struct igb_ring *ring;
+	int ring_count, size;
 
-	for (v_idx = 0; v_idx < adapter->num_q_vectors; v_idx++) {
-		q_vector = kzalloc(sizeof(struct igb_q_vector),
-				   GFP_KERNEL);
-		if (!q_vector)
-			goto err_out;
-		q_vector->adapter = adapter;
-		q_vector->itr_register = hw->hw_addr + E1000_EITR(0);
-		q_vector->itr_val = IGB_START_ITR;
-		netif_napi_add(adapter->netdev, &q_vector->napi, igb_poll, 64);
-		adapter->q_vector[v_idx] = q_vector;
+	/* igb only supports 1 Tx and/or 1 Rx queue per vector */
+	if (txr_count > 1 || rxr_count > 1)
+		return -ENOMEM;
+
+	ring_count = txr_count + rxr_count;
+	size = sizeof(struct igb_q_vector) +
+	       (sizeof(struct igb_ring) * ring_count);
+
+	/* allocate q_vector and rings */
+	q_vector = kzalloc(size, GFP_KERNEL);
+	if (!q_vector)
+		return -ENOMEM;
+
+	/* initialize NAPI */
+	netif_napi_add(adapter->netdev, &q_vector->napi,
+		       igb_poll, 64);
+
+	/* tie q_vector and adapter together */
+	adapter->q_vector[v_idx] = q_vector;
+	q_vector->adapter = adapter;
+
+	/* initialize work limits */
+	q_vector->tx.work_limit = adapter->tx_work_limit;
+
+	/* initialize ITR configuration */
+	q_vector->itr_register = adapter->hw.hw_addr + E1000_EITR(0);
+	q_vector->itr_val = IGB_START_ITR;
+
+	/* initialize pointer to rings */
+	ring = q_vector->ring;
+
+	if (txr_count) {
+		/* assign generic ring traits */
+		ring->dev = &adapter->pdev->dev;
+		ring->netdev = adapter->netdev;
+
+		/* configure backlink on ring */
+		ring->q_vector = q_vector;
+
+		/* update q_vector Tx values */
+		igb_add_ring(ring, &q_vector->tx);
+
+		/* For 82575, context index must be unique per ring. */
+		if (adapter->hw.mac.type == e1000_82575)
+			set_bit(IGB_RING_FLAG_TX_CTX_IDX, &ring->flags);
+
+		/* apply Tx specific ring traits */
+		ring->count = adapter->tx_ring_count;
+		ring->queue_index = txr_idx;
+
+		/* assign ring to adapter */
+		adapter->tx_ring[txr_idx] = ring;
+
+		/* push pointer to next ring */
+		ring++;
 	}
 
-	return 0;
+	if (rxr_count) {
+		/* assign generic ring traits */
+		ring->dev = &adapter->pdev->dev;
+		ring->netdev = adapter->netdev;
 
-err_out:
-	igb_free_q_vectors(adapter);
-	return -ENOMEM;
-}
+		/* configure backlink on ring */
+		ring->q_vector = q_vector;
 
-static void igb_map_rx_ring_to_vector(struct igb_adapter *adapter,
-                                      int ring_idx, int v_idx)
-{
-	struct igb_q_vector *q_vector = adapter->q_vector[v_idx];
+		/* update q_vector Rx values */
+		igb_add_ring(ring, &q_vector->rx);
 
-	q_vector->rx.ring = adapter->rx_ring[ring_idx];
-	q_vector->rx.ring->q_vector = q_vector;
-	q_vector->rx.count++;
-	q_vector->itr_val = adapter->rx_itr_setting;
-	if (q_vector->itr_val && q_vector->itr_val <= 3)
-		q_vector->itr_val = IGB_START_ITR;
-}
+		/* set flag indicating ring supports SCTP checksum offload */
+		if (adapter->hw.mac.type >= e1000_82576)
+			set_bit(IGB_RING_FLAG_RX_SCTP_CSUM, &ring->flags);
 
-static void igb_map_tx_ring_to_vector(struct igb_adapter *adapter,
-                                      int ring_idx, int v_idx)
-{
-	struct igb_q_vector *q_vector = adapter->q_vector[v_idx];
+		/*
+		 * On i350, i210, and i211, loopback VLAN packets
+		 * have the tag byte-swapped.
+		 * */
+		if (adapter->hw.mac.type >= e1000_i350)
+			set_bit(IGB_RING_FLAG_RX_LB_VLAN_BSWAP, &ring->flags);
 
-	q_vector->tx.ring = adapter->tx_ring[ring_idx];
-	q_vector->tx.ring->q_vector = q_vector;
-	q_vector->tx.count++;
-	q_vector->itr_val = adapter->tx_itr_setting;
-	q_vector->tx.work_limit = adapter->tx_work_limit;
-	if (q_vector->itr_val && q_vector->itr_val <= 3)
-		q_vector->itr_val = IGB_START_ITR;
+		/* apply Rx specific ring traits */
+		ring->count = adapter->rx_ring_count;
+		ring->queue_index = rxr_idx;
+
+		/* assign ring to adapter */
+		adapter->rx_ring[rxr_idx] = ring;
+	}
+
+	return 0;
 }
 
+
 /**
- * igb_map_ring_to_vector - maps allocated queues to vectors
+ * igb_alloc_q_vectors - Allocate memory for interrupt vectors
+ * @adapter: board private structure to initialize
  *
- * This function maps the recently allocated queues to vectors.
+ * We allocate one q_vector per queue interrupt.  If allocation fails we
+ * return -ENOMEM.
  **/
-static int igb_map_ring_to_vector(struct igb_adapter *adapter)
+static int igb_alloc_q_vectors(struct igb_adapter *adapter)
 {
-	int i;
-	int v_idx = 0;
+	int q_vectors = adapter->num_q_vectors;
+	int rxr_remaining = adapter->num_rx_queues;
+	int txr_remaining = adapter->num_tx_queues;
+	int rxr_idx = 0, txr_idx = 0, v_idx = 0;
+	int err;
 
-	if ((adapter->num_q_vectors < adapter->num_rx_queues) ||
-	    (adapter->num_q_vectors < adapter->num_tx_queues))
-		return -ENOMEM;
+	if (q_vectors >= (rxr_remaining + txr_remaining)) {
+		for (; rxr_remaining; v_idx++) {
+			err = igb_alloc_q_vector(adapter, q_vectors, v_idx,
+						 0, 0, 1, rxr_idx);
 
-	if (adapter->num_q_vectors >=
-	    (adapter->num_rx_queues + adapter->num_tx_queues)) {
-		for (i = 0; i < adapter->num_rx_queues; i++)
-			igb_map_rx_ring_to_vector(adapter, i, v_idx++);
-		for (i = 0; i < adapter->num_tx_queues; i++)
-			igb_map_tx_ring_to_vector(adapter, i, v_idx++);
-	} else {
-		for (i = 0; i < adapter->num_rx_queues; i++) {
-			if (i < adapter->num_tx_queues)
-				igb_map_tx_ring_to_vector(adapter, i, v_idx);
-			igb_map_rx_ring_to_vector(adapter, i, v_idx++);
+			if (err)
+				goto err_out;
+
+			/* update counts and index */
+			rxr_remaining--;
+			rxr_idx++;
 		}
-		for (; i < adapter->num_tx_queues; i++)
-			igb_map_tx_ring_to_vector(adapter, i, v_idx++);
 	}
+
+	for (; v_idx < q_vectors; v_idx++) {
+		int rqpv = DIV_ROUND_UP(rxr_remaining, q_vectors - v_idx);
+		int tqpv = DIV_ROUND_UP(txr_remaining, q_vectors - v_idx);
+		err = igb_alloc_q_vector(adapter, q_vectors, v_idx,
+					 tqpv, txr_idx, rqpv, rxr_idx);
+
+		if (err)
+			goto err_out;
+
+		/* update counts and index */
+		rxr_remaining -= rqpv;
+		txr_remaining -= tqpv;
+		rxr_idx++;
+		txr_idx++;
+	}
+
 	return 0;
+
+err_out:
+	adapter->num_tx_queues = 0;
+	adapter->num_rx_queues = 0;
+	adapter->num_q_vectors = 0;
+
+	while (v_idx--)
+		igb_free_q_vector(adapter, v_idx);
+
+	return -ENOMEM;
 }
 
 /**
@@ -1185,24 +1222,10 @@ static int igb_init_interrupt_scheme(struct igb_adapter *adapter)
 		goto err_alloc_q_vectors;
 	}
 
-	err = igb_alloc_queues(adapter);
-	if (err) {
-		dev_err(&pdev->dev, "Unable to allocate memory for queues\n");
-		goto err_alloc_queues;
-	}
-
-	err = igb_map_ring_to_vector(adapter);
-	if (err) {
-		dev_err(&pdev->dev, "Invalid q_vector to ring mapping\n");
-		goto err_map_queues;
-	}
-
+	igb_cache_ring_register(adapter);
 
 	return 0;
-err_map_queues:
-	igb_free_queues(adapter);
-err_alloc_queues:
-	igb_free_q_vectors(adapter);
+
 err_alloc_q_vectors:
 	igb_reset_interrupt_capability(adapter);
 	return err;
@@ -1225,11 +1248,11 @@ static int igb_request_irq(struct igb_adapter *adapter)
 		if (!err)
 			goto request_done;
 		/* fall back to MSI */
+		igb_free_all_tx_resources(adapter);
+		igb_free_all_rx_resources(adapter);
 		igb_clear_interrupt_scheme(adapter);
 		if (!pci_enable_msi(pdev))
 			adapter->flags |= IGB_FLAG_HAS_MSI;
-		igb_free_all_tx_resources(adapter);
-		igb_free_all_rx_resources(adapter);
 		adapter->num_tx_queues = 1;
 		adapter->num_rx_queues = 1;
 		adapter->num_q_vectors = 1;
@@ -1239,13 +1262,6 @@ static int igb_request_irq(struct igb_adapter *adapter)
 			        "Unable to allocate memory for vectors\n");
 			goto request_done;
 		}
-		err = igb_alloc_queues(adapter);
-		if (err) {
-			dev_err(&pdev->dev,
-			        "Unable to allocate memory for queues\n");
-			igb_free_q_vectors(adapter);
-			goto request_done;
-		}
 		igb_setup_all_tx_resources(adapter);
 		igb_setup_all_rx_resources(adapter);
 	}
@@ -2633,10 +2649,8 @@ 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 = dma_alloc_coherent(dev,
-					   tx_ring->size,
-					   &tx_ring->dma,
-					   GFP_KERNEL);
+	tx_ring->desc = dma_alloc_coherent(dev, tx_ring->size,
+					   &tx_ring->dma, GFP_KERNEL);
 	if (!tx_ring->desc)
 		goto err;
 
@@ -2773,15 +2787,12 @@ int igb_setup_rx_resources(struct igb_ring *rx_ring)
 	if (!rx_ring->rx_buffer_info)
 		goto err;
 
-
 	/* Round up to nearest 4K */
 	rx_ring->size = rx_ring->count * sizeof(union e1000_adv_rx_desc);
 	rx_ring->size = ALIGN(rx_ring->size, 4096);
 
-	rx_ring->desc = dma_alloc_coherent(dev,
-					   rx_ring->size,
-					   &rx_ring->dma,
-					   GFP_KERNEL);
+	rx_ring->desc = dma_alloc_coherent(dev, rx_ring->size,
+					   &rx_ring->dma, GFP_KERNEL);
 	if (!rx_ring->desc)
 		goto err;
 
-- 
1.7.11.7

[net-next 13/14] igb: Move the calls to set the Tx and Rx queues into igb_open

[Jeff Kirsher]

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

This change helps to address locking issues seen with
netif_set_real_num_tx_queues and netif_set_real_num_rx_queues when used in
the igb_set_interrupt_capability function.  To resolve these locking issues
I have moved the two function calls into __igb_open so that they can be
called while the RTNL lock is held.

An added advantage to this is that the number of queues is not updated
until the last possible moment so if there are any issues in allocating
MSI-X interrupts or resources for the rings we have time to change the
values prior to updating the netdev.

Signed-off-by: Alexander Duyck <alexander.h.duyck@intel.com>
Tested-by: Aaron Brown <aaron.f.brown@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
---
 drivers/net/ethernet/intel/igb/igb_main.c | 31 ++++++++++++++++++-------------
 1 file changed, 18 insertions(+), 13 deletions(-)

diff --git a/drivers/net/ethernet/intel/igb/igb_main.c b/drivers/net/ethernet/intel/igb/igb_main.c
index 4a25b8f..e7b1027 100644
--- a/drivers/net/ethernet/intel/igb/igb_main.c
+++ b/drivers/net/ethernet/intel/igb/igb_main.c
@@ -948,7 +948,7 @@ static void igb_clear_interrupt_scheme(struct igb_adapter *adapter)
  * Attempt to configure interrupts using the best available
  * capabilities of the hardware and kernel.
  **/
-static int igb_set_interrupt_capability(struct igb_adapter *adapter)
+static void igb_set_interrupt_capability(struct igb_adapter *adapter)
 {
 	int err;
 	int numvecs, i;
@@ -985,7 +985,7 @@ static int igb_set_interrupt_capability(struct igb_adapter *adapter)
 			      adapter->msix_entries,
 			      numvecs);
 	if (err == 0)
-		goto out;
+		return;
 
 	igb_reset_interrupt_capability(adapter);
 
@@ -1015,14 +1015,6 @@ msi_only:
 	adapter->num_q_vectors = 1;
 	if (!pci_enable_msi(adapter->pdev))
 		adapter->flags |= IGB_FLAG_HAS_MSI;
-out:
-	/* Notify the stack of the (possibly) reduced queue counts. */
-	rtnl_lock();
-	netif_set_real_num_tx_queues(adapter->netdev, adapter->num_tx_queues);
-	err = netif_set_real_num_rx_queues(adapter->netdev,
-		adapter->num_rx_queues);
-	rtnl_unlock();
-	return err;
 }
 
 static void igb_add_ring(struct igb_ring *ring,
@@ -1212,9 +1204,7 @@ static int igb_init_interrupt_scheme(struct igb_adapter *adapter)
 	struct pci_dev *pdev = adapter->pdev;
 	int err;
 
-	err = igb_set_interrupt_capability(adapter);
-	if (err)
-		return err;
+	igb_set_interrupt_capability(adapter);
 
 	err = igb_alloc_q_vectors(adapter);
 	if (err) {
@@ -2543,6 +2533,17 @@ static int __igb_open(struct net_device *netdev, bool resuming)
 	if (err)
 		goto err_req_irq;
 
+	/* Notify the stack of the actual queue counts. */
+	err = netif_set_real_num_tx_queues(adapter->netdev,
+					   adapter->num_tx_queues);
+	if (err)
+		goto err_set_queues;
+
+	err = netif_set_real_num_rx_queues(adapter->netdev,
+					   adapter->num_rx_queues);
+	if (err)
+		goto err_set_queues;
+
 	/* From here on the code is the same as igb_up() */
 	clear_bit(__IGB_DOWN, &adapter->state);
 
@@ -2572,6 +2573,8 @@ static int __igb_open(struct net_device *netdev, bool resuming)
 
 	return 0;
 
+err_set_queues:
+	igb_free_irq(adapter);
 err_req_irq:
 	igb_release_hw_control(adapter);
 	igb_power_down_link(adapter);
@@ -6841,7 +6844,9 @@ static int igb_resume(struct device *dev)
 	wr32(E1000_WUS, ~0);
 
 	if (netdev->flags & IFF_UP) {
+		rtnl_lock();
 		err = __igb_open(netdev, true);
+		rtnl_unlock();
 		if (err)
 			return err;
 	}
-- 
1.7.11.7

[net-next 14/14] igb: Split igb_update_dca into separate Tx and Rx functions

[Jeff Kirsher]

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

This change makes it so that igb_update_dca is broken into two halves, one
for Rx and one for Tx.  The advantage to this is primarily readability.

In addition I am enabling relaxed ordering for reads from hardware since
this is supported on all of the igb parts.

Signed-off-by: Alexander Duyck <alexander.h.duyck@intel.com>
Tested-by: Aaron Brown <aaron.f.brown@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
---
 drivers/net/ethernet/intel/igb/e1000_82575.h |  3 ++
 drivers/net/ethernet/intel/igb/igb_main.c    | 80 +++++++++++++++++-----------
 2 files changed, 52 insertions(+), 31 deletions(-)

diff --git a/drivers/net/ethernet/intel/igb/e1000_82575.h b/drivers/net/ethernet/intel/igb/e1000_82575.h
index e85c453..44b76b3 100644
--- a/drivers/net/ethernet/intel/igb/e1000_82575.h
+++ b/drivers/net/ethernet/intel/igb/e1000_82575.h
@@ -172,10 +172,13 @@ struct e1000_adv_tx_context_desc {
 #define E1000_DCA_RXCTRL_DESC_DCA_EN (1 << 5) /* DCA Rx Desc enable */
 #define E1000_DCA_RXCTRL_HEAD_DCA_EN (1 << 6) /* DCA Rx Desc header enable */
 #define E1000_DCA_RXCTRL_DATA_DCA_EN (1 << 7) /* DCA Rx Desc payload enable */
+#define E1000_DCA_RXCTRL_DESC_RRO_EN (1 << 9) /* DCA Rx rd Desc Relax Order */
 
 #define E1000_DCA_TXCTRL_CPUID_MASK 0x0000001F /* Tx CPUID Mask */
 #define E1000_DCA_TXCTRL_DESC_DCA_EN (1 << 5) /* DCA Tx Desc enable */
+#define E1000_DCA_TXCTRL_DESC_RRO_EN (1 << 9) /* Tx rd Desc Relax Order */
 #define E1000_DCA_TXCTRL_TX_WB_RO_EN (1 << 11) /* Tx Desc writeback RO bit */
+#define E1000_DCA_TXCTRL_DATA_RRO_EN (1 << 13) /* Tx rd data Relax Order */
 
 /* Additional DCA related definitions, note change in position of CPUID */
 #define E1000_DCA_TXCTRL_CPUID_MASK_82576 0xFF000000 /* Tx CPUID Mask */
diff --git a/drivers/net/ethernet/intel/igb/igb_main.c b/drivers/net/ethernet/intel/igb/igb_main.c
index e7b1027..87abb57 100644
--- a/drivers/net/ethernet/intel/igb/igb_main.c
+++ b/drivers/net/ethernet/intel/igb/igb_main.c
@@ -4851,45 +4851,63 @@ static irqreturn_t igb_msix_ring(int irq, void *data)
 }
 
 #ifdef CONFIG_IGB_DCA
+static void igb_update_tx_dca(struct igb_adapter *adapter,
+			      struct igb_ring *tx_ring,
+			      int cpu)
+{
+	struct e1000_hw *hw = &adapter->hw;
+	u32 txctrl = dca3_get_tag(tx_ring->dev, cpu);
+
+	if (hw->mac.type != e1000_82575)
+		txctrl <<= E1000_DCA_TXCTRL_CPUID_SHIFT;
+
+	/*
+	 * We can enable relaxed ordering for reads, but not writes when
+	 * DCA is enabled.  This is due to a known issue in some chipsets
+	 * which will cause the DCA tag to be cleared.
+	 */
+	txctrl |= E1000_DCA_TXCTRL_DESC_RRO_EN |
+		  E1000_DCA_TXCTRL_DATA_RRO_EN |
+		  E1000_DCA_TXCTRL_DESC_DCA_EN;
+
+	wr32(E1000_DCA_TXCTRL(tx_ring->reg_idx), txctrl);
+}
+
+static void igb_update_rx_dca(struct igb_adapter *adapter,
+			      struct igb_ring *rx_ring,
+			      int cpu)
+{
+	struct e1000_hw *hw = &adapter->hw;
+	u32 rxctrl = dca3_get_tag(&adapter->pdev->dev, cpu);
+
+	if (hw->mac.type != e1000_82575)
+		rxctrl <<= E1000_DCA_RXCTRL_CPUID_SHIFT;
+
+	/*
+	 * We can enable relaxed ordering for reads, but not writes when
+	 * DCA is enabled.  This is due to a known issue in some chipsets
+	 * which will cause the DCA tag to be cleared.
+	 */
+	rxctrl |= E1000_DCA_RXCTRL_DESC_RRO_EN |
+		  E1000_DCA_RXCTRL_DESC_DCA_EN;
+
+	wr32(E1000_DCA_RXCTRL(rx_ring->reg_idx), rxctrl);
+}
+
 static void igb_update_dca(struct igb_q_vector *q_vector)
 {
 	struct igb_adapter *adapter = q_vector->adapter;
-	struct e1000_hw *hw = &adapter->hw;
 	int cpu = get_cpu();
 
 	if (q_vector->cpu == cpu)
 		goto out_no_update;
 
-	if (q_vector->tx.ring) {
-		int q = q_vector->tx.ring->reg_idx;
-		u32 dca_txctrl = rd32(E1000_DCA_TXCTRL(q));
-		if (hw->mac.type == e1000_82575) {
-			dca_txctrl &= ~E1000_DCA_TXCTRL_CPUID_MASK;
-			dca_txctrl |= dca3_get_tag(&adapter->pdev->dev, cpu);
-		} else {
-			dca_txctrl &= ~E1000_DCA_TXCTRL_CPUID_MASK_82576;
-			dca_txctrl |= dca3_get_tag(&adapter->pdev->dev, cpu) <<
-			              E1000_DCA_TXCTRL_CPUID_SHIFT;
-		}
-		dca_txctrl |= E1000_DCA_TXCTRL_DESC_DCA_EN;
-		wr32(E1000_DCA_TXCTRL(q), dca_txctrl);
-	}
-	if (q_vector->rx.ring) {
-		int q = q_vector->rx.ring->reg_idx;
-		u32 dca_rxctrl = rd32(E1000_DCA_RXCTRL(q));
-		if (hw->mac.type == e1000_82575) {
-			dca_rxctrl &= ~E1000_DCA_RXCTRL_CPUID_MASK;
-			dca_rxctrl |= dca3_get_tag(&adapter->pdev->dev, cpu);
-		} else {
-			dca_rxctrl &= ~E1000_DCA_RXCTRL_CPUID_MASK_82576;
-			dca_rxctrl |= dca3_get_tag(&adapter->pdev->dev, cpu) <<
-			              E1000_DCA_RXCTRL_CPUID_SHIFT;
-		}
-		dca_rxctrl |= E1000_DCA_RXCTRL_DESC_DCA_EN;
-		dca_rxctrl |= E1000_DCA_RXCTRL_HEAD_DCA_EN;
-		dca_rxctrl |= E1000_DCA_RXCTRL_DATA_DCA_EN;
-		wr32(E1000_DCA_RXCTRL(q), dca_rxctrl);
-	}
+	if (q_vector->tx.ring)
+		igb_update_tx_dca(adapter, q_vector->tx.ring, cpu);
+
+	if (q_vector->rx.ring)
+		igb_update_rx_dca(adapter, q_vector->rx.ring, cpu);
+
 	q_vector->cpu = cpu;
 out_no_update:
 	put_cpu();
-- 
1.7.11.7

RE: [net-next 08/14] igb: Combine post-processing of skb into a single function

[David Laight]

> From: Alexander Duyck <alexander.h.duyck@intel.com>
> 
> This change is meant to just clean-up a number of function calls that were
> made at the end of the Rx clean-up path by combining them into a single
> function call.

Since the functions are static and only called once they
are prime candidates for inlining.
This might happen anyway (depending on the gcc version
and the direction the wind is blowing in).

Marking them with __attribute__((always_inline)) might
be needed to get them actually inlined.

That would maintain the logical separation.

	David

RE: [net-next 08/14] igb: Combine post-processing of skb into a single function

[Eric Dumazet]

On Fri, 2012-10-19 at 14:49 +0100, David Laight wrote:
> > From: Alexander Duyck <alexander.h.duyck@intel.com>
> > 
> > This change is meant to just clean-up a number of function calls that were
> > made at the end of the Rx clean-up path by combining them into a single
> > function call.
> 
> Since the functions are static and only called once they
> are prime candidates for inlining.
> This might happen anyway (depending on the gcc version
> and the direction the wind is blowing in).
> 
> Marking them with __attribute__((always_inline)) might
> be needed to get them actually inlined.
> 
> That would maintain the logical separation.

vi +712 Documentation/CodingStyle

Re: [net-next 00/14][pull request] Intel Wired LAN Driver Updates

[David Miller]

From: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Date: Fri, 19 Oct 2012 04:45:00 -0700

> This series contains updates to ixgbe and igb.
> 
> The following are changes since commit db0fe0b2f6bba2fda939737d063db2ae14c58d71:
>   Merge tag 'batman-adv-fix-for-davem' of git://git.open-mesh.org/linux-merge
> and are available in the git repository at:
>   git://git.kernel.org/pub/scm/linux/kernel/git/jkirsher/net-next master

Pulled, thanks Jeff.

Re: [net-next 00/14][pull request] Intel Wired LAN Driver Updates

[David Miller]

From: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Date: Wed, 23 Apr 2014 04:15:12 -0700

> This series contains updates to ixgbe, ixgbevf, e1000e, igb and i40e.

Pulled, thanks a lot Jeff.

[net-next 00/14][pull request] Intel Wired LAN Driver Updates

[Jeff Kirsher]

This series contains updates to ixgbe, ixgbevf, e1000e, igb and i40e.

Jacob converts the ixgbe low_water into an array which allows the
algorithm to output different values for different TCs and we can
distinguish between them.  Removes vlan_filter_disable() and
vlan_filter_enable() in ixgbe so that we can do the work directly in
set_rx_mode().  Changes the setting of multicast filters only when
the interface is not in promiscuous mode for multicast packets in
ixgbe.  Improves MAC filter handling by adding mac_table API based
on work done for igb, which includes functions to add/delete MAC
filters.

Mark changes register reads in ixgbe to an out-of-line function since
register reads are slow.

Emil provides a ixgbevf patch to update the driver description since
it supports more than just 82599 parts now.

David provides several cleanup patches for e1000e which resolve some
checkpatch issues as well as changing occurrences of returning 0 or 1 in
bool functions to returning true false or true.

Carolyn provides several cleanup patches for igb which fix checkpatch
warnings.

Mitch provides a fix for i40evf where the driver would correctly allow
the virtual function link state to be controlled by 'ip set link', but
would not report it correctly back.  This is fixed by filling out
the appropriate field in the VF info struct.

The following are changes since commit fd0d192be6e814495aec91f357b5801afc3b6262:
  Merge branch 'master' of git://git.kernel.org/pub/scm/linux/kernel/git/jkirsher/net-next
and are available in the git repository at:
  git://git.kernel.org/pub/scm/linux/kernel/git/jkirsher/net-next master

Carolyn Wyborny (3):
  igb: Cleanups for messaging
  igb: Cleanups to fix braces location warnings
  igb: Cleanups to fix incorrect indentation

David Ertman (4):
  e1000e: Cleanup return values in ethtool
  e1000e: Cleanup to fix checkpatch missing blank lines
  e1000e: Cleanup checkpatch extra space
  e1000e: Cleanup use of deprecated DEFINE_PCI_DEVICE_TABLE

Emil Tantilov (1):
  ixgbevf: remove 82599 from the module description

Jacob Keller (4):
  ixgbe: convert low_water into an array
  ixgbe: remove vlan_filter_disable and enable functions
  ixgbe: change handling of multicast filters
  ixgbe: improve mac filter handling

Mark Rustad (1):
  ixgbe: Use out-of-line function for register reads

Mitch Williams (1):
  i40e: report VF link state correctly

 drivers/net/ethernet/intel/e1000e/e1000.h          |   8 +-
 drivers/net/ethernet/intel/e1000e/ethtool.c        |  11 +-
 drivers/net/ethernet/intel/e1000e/ich8lan.c        |   1 +
 drivers/net/ethernet/intel/e1000e/netdev.c         |  19 +-
 drivers/net/ethernet/intel/e1000e/nvm.c            |   1 +
 drivers/net/ethernet/intel/e1000e/param.c          |   4 +
 drivers/net/ethernet/intel/e1000e/phy.c            |   1 +
 drivers/net/ethernet/intel/i40e/i40e_virtchnl_pf.c |   7 +
 drivers/net/ethernet/intel/igb/e1000_82575.c       |  24 +-
 drivers/net/ethernet/intel/igb/e1000_82575.h       |  22 +-
 drivers/net/ethernet/intel/igb/e1000_defines.h     |  25 +-
 drivers/net/ethernet/intel/igb/e1000_hw.h          |  50 ++--
 drivers/net/ethernet/intel/igb/e1000_mac.c         |  10 +-
 drivers/net/ethernet/intel/igb/e1000_nvm.c         |   1 +
 drivers/net/ethernet/intel/igb/e1000_nvm.h         |   2 +-
 drivers/net/ethernet/intel/igb/e1000_phy.c         |   3 +-
 drivers/net/ethernet/intel/igb/e1000_regs.h        |   7 +-
 drivers/net/ethernet/intel/igb/igb.h               |   1 +
 drivers/net/ethernet/intel/igb/igb_ethtool.c       |  24 +-
 drivers/net/ethernet/intel/igb/igb_main.c          |  74 ++---
 drivers/net/ethernet/intel/ixgbe/ixgbe.h           |  18 +-
 drivers/net/ethernet/intel/ixgbe/ixgbe_82598.c     |  27 +-
 drivers/net/ethernet/intel/ixgbe/ixgbe_common.c    |  35 ++-
 drivers/net/ethernet/intel/ixgbe/ixgbe_common.h    |  15 +-
 drivers/net/ethernet/intel/ixgbe/ixgbe_dcb_82598.c |   2 +-
 drivers/net/ethernet/intel/ixgbe/ixgbe_dcb_82599.c |   2 +-
 drivers/net/ethernet/intel/ixgbe/ixgbe_fcoe.h      |   2 -
 drivers/net/ethernet/intel/ixgbe/ixgbe_main.c      | 319 +++++++++++++++------
 drivers/net/ethernet/intel/ixgbe/ixgbe_sriov.c     |  49 ++--
 drivers/net/ethernet/intel/ixgbe/ixgbe_sriov.h     |   2 +
 drivers/net/ethernet/intel/ixgbe/ixgbe_type.h      |   2 +-
 drivers/net/ethernet/intel/ixgbevf/ixgbevf_main.c  |   2 +-
 32 files changed, 478 insertions(+), 292 deletions(-)

-- 
1.9.0

Re: [net-next 00/14][pull request] Intel Wired LAN Driver Updates

[David Miller]

From: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Date: Tue, 17 Dec 2013 22:44:28 -0800

> This series contains updates to i40e, ixgbevf, ixgbe and igb.

Pulled, thanks Jeff.

[net-next 00/14][pull request] Intel Wired LAN Driver Updates

[Jeff Kirsher]

This series contains updates to i40e, ixgbevf, ixgbe and igb.

Don provides an ixgbevf patch to add DCB configuration into the queue
setup so that we won't have to allocate queues in a separate place when
enabling DCB.

Guenter Roeck provides 2 patches for ixgbe to simplify the code by
attaching hwmon sysfs attributes to hwmon device instead of PCI device.
Also fix an issues where the temperature sensor attribute index was
being started with the value 0 and not 1 as per the hwmon API.

Carolyn provides igb patches to fix queue allocation method to
accommodate changes during runtime.  This includes changing how the
driver initializes MSIx and checks for MSIx configuration to make it
easier to reconfigure the device when queue changes happen at runtime.

Neerav and Shannon fixes i40e debugfs commands that dump hex information
by using print_hex_dump().

Shannon provides several i40e fixes which include the prevention of
null pointer exception in the dump descriptor by checking that rings
were allocated before trying to reference them.  Fixed up a couple of
scanfs to accept various base numbers instead of silently requiring hex.

Anjali fixes up i40e where the incorrect defines were being used for
misc interrupts.

Alan Cox provides a fix for i40e where we assume that the resulting
buffer is zero terminated when we then re-use it.  The sscanf is limited
to 512 bytes but needs to be 511 to allow for a terminator.

Stephen Hemminger fixes i40e by making local functions static and removes
unused code (i40e_aq_add/remove_vlan() functions).

The following are changes since commit 02d5cb5bb20b9d34db20860aad1891cd9b8e81d5:
  qeth: Accurate ethtool output
and are available in the git repository at:
  git://git.kernel.org/pub/scm/linux/kernel/git/jkirsher/net-next master

Alan Cox (1):
  i40e: Fix off by one in i40e_dbg_command_write

Anjali Singhai Jain (1):
  i40e: Fix wrong mask bits being used in misc interrupt

Carolyn Wyborny (2):
  igb: Fix queue allocation method to accomodate changing during runtime
  igb: Change to use statically allocated array for MSIx entries

Catherine Sullivan (1):
  i40e: Bump version number

Don Skidmore (1):
  ixgbevf: add DCB configuration into queue setup

Guenter Roeck (2):
  ixgbe: Convert to use devm_hwmon_device_register_with_groups
  ixgbe: Start temperature sensor attribute index with 1

Neerav Parikh (1):
  i40e: Fix dump output from debugfs calls

Shannon Nelson (4):
  i40e: prevent null pointer exception in dump descriptor
  i40e: simplify error messages for dump descriptor
  i40e: fix up scanf decoders
  i40e: more print_hex_dump use

Stephen Hemminger (1):
  i40e: make functions static and remove dead code

 drivers/net/ethernet/intel/i40e/i40e_adminq.c     |   6 +-
 drivers/net/ethernet/intel/i40e/i40e_common.c     |  80 --------
 drivers/net/ethernet/intel/i40e/i40e_debugfs.c    | 132 ++++---------
 drivers/net/ethernet/intel/i40e/i40e_main.c       |   6 +-
 drivers/net/ethernet/intel/i40e/i40e_prototype.h  |   8 -
 drivers/net/ethernet/intel/igb/igb.h              |   9 +-
 drivers/net/ethernet/intel/igb/igb_ethtool.c      |   6 +-
 drivers/net/ethernet/intel/igb/igb_main.c         |  94 +++++----
 drivers/net/ethernet/intel/ixgbe/ixgbe.h          |   8 +-
 drivers/net/ethernet/intel/ixgbe/ixgbe_sysfs.c    |  84 ++++----
 drivers/net/ethernet/intel/ixgbevf/ixgbevf.h      |   1 +
 drivers/net/ethernet/intel/ixgbevf/ixgbevf_main.c | 228 +++++++++-------------
 12 files changed, 243 insertions(+), 419 deletions(-)

-- 
1.8.3.1

Re: [net-next 00/14][pull request] Intel Wired LAN Driver Updates

[David Miller]

From: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Date: Fri,  6 Dec 2013 18:17:09 -0800

> This series contains updates to i40e only.

Pulled, thanks Jeff.

[net-next 00/14][pull request] Intel Wired LAN Driver Updates

[Jeff Kirsher]

This series contains updates to i40e only.

Jacob provides a i40e patch to get 1588 work correctly by separating
TSYNVALID and TSYNINDX fields in the receive descriptor.

Jesse provides several i40e patches, first to correct the checking
of the multi-bit state.  The hash is reported correctly in the RSS
field if and only if the filter status is 3.  Other values of the
filter status mean different things and we should not depend on a
bitwise result.  Then provides a patch to enable a couple of
workarounds based on revision ID that allow the driver to work
more fully on early hardware.

Shannon provides several i40e patches as well.  First sets the media
type in the hardware structure based on the external connection type.
Then provides a patch to only setup the rings that will be used.  Lastly
provides a fix where the TESTING state was still set when exiting the
ethtool diagnostics.

Kevin Scott provides one i40e patch to add a new flag to the i40e_add_veb()
which allows the driver to request the hardware to filter on layer 2
parameters.

Anjali provides four i40e patches, first refactors the reset code in
order to re-size queues and vectors while the interface is still up.
Then provides a patch to enable all PCTYPEs expect FCoE for RSS.  Adds
a message to notify the user of how many VFs are initialized on each
port.  Lastly adds a new variable to track the number of PF instances,
this is a global counter on purpose so that each PF loaded has a
unique ID.

Catherine bumps the driver version.

The following are changes since commit 0d74c42f788caf3cad727c61c490d9459bc8918b:
  ether_addr_equal: Optimize implementation, remove unused compare_ether_addr
and are available in the git repository at:
  git://git.kernel.org/pub/scm/linux/kernel/git/jkirsher/net-next master

Anjali Singhai Jain (4):
  i40e: refactor reset code
  i40e: Enable all PCTYPEs except FCOE for RSS.
  i40e: add num_VFs message
  i40e: Add a new variable to track number of pf instances

Catherine Sullivan (1):
  i40e: Bump version

Jacob Keller (1):
  i40e: separate TSYNVALID and TSYNINDX fields in Rx descriptor

Jeff Kirsher (1):
  i40e: whitespace

Jesse Brandeburg (3):
  i40e: check multi-bit state correctly
  i40e: get media type during link info
  i40e: enable early hardware support

Kevin Scott (1):
  i40e: Add flag for L2 VEB filtering

Shannon Nelson (3):
  i40e: only set up the rings to be used
  i40e: clear test state bit after all ethtool tests
  i40e: refactor ethtool tests

 drivers/net/ethernet/intel/i40e/i40e.h           |   3 +
 drivers/net/ethernet/intel/i40e/i40e_common.c    |  77 +++++++-
 drivers/net/ethernet/intel/i40e/i40e_ethtool.c   |  66 ++++---
 drivers/net/ethernet/intel/i40e/i40e_main.c      | 241 +++++++++++++++++------
 drivers/net/ethernet/intel/i40e/i40e_prototype.h |   3 +-
 drivers/net/ethernet/intel/i40e/i40e_register.h  |   7 +
 drivers/net/ethernet/intel/i40e/i40e_txrx.c      |  16 +-
 drivers/net/ethernet/intel/i40e/i40e_type.h      |  10 +-
 8 files changed, 316 insertions(+), 107 deletions(-)

-- 
1.8.3.1

[net-next 00/14][pull request] Intel Wired LAN Driver Updates

[Jeff Kirsher]

This series contains updates to i40e only.

Jesse provides 6 patches against i40e.  First is a patch to reduce
CPU utilization by reducing read-flush to read in the hot path.  Next
couple of patches resolve coverity issues reported by Hannes Frederic
Sowa <hannes@stressinduktion.org>.  Then Jesse refactored i40e to cleanup
functions which used cpu_to_xxx(foo) which caused a lot of line wrapping.

Mitch provides 2 i40e patches.  First fixes a panic when tx_rings[0]
are not allocated, his second patch corrects a math error when
assigning MSI-X vectors to VFs.  The vectors-per-vf value reported
by the hardware already conveniently reports one less than the actual
value.

Shannon provides 5 patches against i40e.  His first patch corrects a
number of little bugs in the error handling of irq setup, most of
which ended up panicing the kernel.  Next he fixes the overactive
IRQ issue seen in testing and allows the use of the legacy interrupt.
Shannon then provides a cleanup of the arguments declared at the
beginning of each function.  Then he provides a patch to make sure
that there are really rings and queues before trying to dump
information in them.  Lastly he simplifies the code by using an
already existing variable.

Catherine provides an i40e patch to bump the version.

The following are changes since commit 7cc7c5e54b7128195a1403747a63971c3c3f8e25:
  net: Delete trailing semi-colon from definition of netdev_WARN()
and are available in the git repository at:
  git://git.kernel.org/pub/scm/linux/kernel/git/jkirsher/net-next master

Catherine Sullivan (1):
  i40e: Bump version

Jesse Brandeburg (6):
  i40e: do not flush after re-enabling interrupts
  i40e: debugfs fixups
  i40e: clamp debugfs nvm read command
  i40e: fix use of untrusted scalar value warning
  i40e: fix sign extension issue
  i40e: refactor fdir setup function

Mitch Williams (2):
  i40e: don't free nonexistent rings
  i40e: assign correct vector to VF

Shannon Nelson (5):
  i40e: fixup legacy interrupt handling
  i40e: tweaking icr0 handling for legacy irq
  i40e: reorder block declarations in debugfs
  i40e: check vsi ptrs before dumping them
  i40e: use pf_id for pf function id in qtx_ctl

 drivers/net/ethernet/intel/i40e/i40e.h             |   1 +
 drivers/net/ethernet/intel/i40e/i40e_debugfs.c     | 135 ++++++++++++---------
 drivers/net/ethernet/intel/i40e/i40e_main.c        |  50 ++++----
 drivers/net/ethernet/intel/i40e/i40e_txrx.c        |  83 ++++++-------
 drivers/net/ethernet/intel/i40e/i40e_virtchnl_pf.c |   4 +-
 5 files changed, 146 insertions(+), 127 deletions(-)

-- 
1.8.3.1

[net-next 00/14][pull request] Intel Wired LAN Driver Updates

[Jeff Kirsher]

This series contains updates to ixgbe, igb and pci.

The ixgbe changes contains a fix to a possible divide by zero by bailing
out of the ixgbe_update_itr() function if the last interrupt timeslice is
zero.  In addition, support is added for the new OCP x520 adapter as well
as LX support for 82599 devices.  Jacob provides a patch to change
variable wol_supported to wol_enabled to better reflect what the code
is actually doing (i.e. checking if WoL is enabled).

Alex adds SRIOV helper function to pci that will determine if a PF
has any VFs that are currently assigned to a guest.

The remaining 8 patches are against igb and contain the following changes:
* implement SERDES loopback configuration for i210 devices by unsetting
  sigdetect bit, so as to fix Ethtool loopback test failure
* add support for the SMBI semaphore for I210/I211 devices
* implement the new generic pci_vfs_assigned helper function (Alex's PCI
  helper function)
* display warning when link speed is downgraded due to Smartspeed
* ensure that VLAN hardware filtering remains enabled when the device is
  in promiscuous mode and VT mode simultaneously
* cleanup dead code in igb
* bump the driver version

The following are changes since commit 95a06161e6b903ad5b96285cb57c8df3b7c8ad34:
  Merge branch 'master' of git://git.kernel.org/pub/scm/linux/kernel/git/pablo/nf-next
and are available in the git repository at:
  git://git.kernel.org/pub/scm/linux/kernel/git/jkirsher/net-next master

Akeem G. Abodunrin (1):
  igb: SERDES loopback sigdetect bit on i210 devices

Alexander Duyck (2):
  pci: Add SRIOV helper function to determine if VFs are assigned to
    guest
  igb: Use pci_vfs_assigned instead of igb_vfs_are_assigned

Carolyn Wyborny (2):
  igb: Remove id's that will not be productized for Linux.
  igb: Bump version of driver

Don Skidmore (3):
  ixgbe: fix possible divide by zero in ixgbe_update_itr
  ixgbe: add driver support for x520 OCP adapter.
  ixgbe: add SFP+ LX module support

Emil Tantilov (1):
  ixgbe: add WOL support for new subdevice ID

Greg Rose (1):
  igb: Retain HW VLAN filtering while in promiscuous + VT mode

Jacob Keller (1):
  ixgbe: rename wol_supported to more fitting wol_enabled

Koki Sanagi (1):
  igb: display a warning message when SmartSpeed works

Matthew Vick (2):
  igb: Add SMBI semaphore to I210/I211
  igb: Remove dead code path

 drivers/net/ethernet/intel/igb/e1000_82575.c     |   9 +-
 drivers/net/ethernet/intel/igb/e1000_hw.h        |   3 +-
 drivers/net/ethernet/intel/igb/e1000_i210.c      |  65 ++++++++-----
 drivers/net/ethernet/intel/igb/igb_ethtool.c     |   9 +-
 drivers/net/ethernet/intel/igb/igb_main.c        | 117 +++++++++++++++--------
 drivers/net/ethernet/intel/ixgbe/ixgbe_82599.c   |   4 +-
 drivers/net/ethernet/intel/ixgbe/ixgbe_ethtool.c |  10 +-
 drivers/net/ethernet/intel/ixgbe/ixgbe_main.c    |  16 +++-
 drivers/net/ethernet/intel/ixgbe/ixgbe_phy.c     |  21 +++-
 drivers/net/ethernet/intel/ixgbe/ixgbe_type.h    |   6 +-
 drivers/pci/iov.c                                |  41 ++++++++
 include/linux/pci.h                              |   5 +
 12 files changed, 212 insertions(+), 94 deletions(-)

-- 
1.7.11.7

Re: [net-next 00/14][pull request] Intel Wired LAN Driver Updates

[David Miller]

From: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Date: Mon, 28 Jan 2013 01:04:15 -0800

> This series contains updates to e1000e, ixgbevf, igb and igbvf.
> Majority of the patches are code cleanups of e1000e where code
> is removed (Yeah!).  The other two e1000e patches are fixes.  The
> first is to fix the maximum frame size for 82579 devices.  The second
> fix is to resolve an issue with devices other than 82579 that suffer
> from dropped transactions on platforms with deep C-states when
> jumbo frames are enabled.
> 
> The ixgbevf patch is to ensure that the driver fetches the correct,
> refreshed value for link status and speed when the values have changed.
> 
> The igb and igbvf patches are a solution to an issue Stefan Assmann
> reported, where when the PF is up and igbvf is loaded, the MAC address
> is not generated using eth_hw_addr_random().
> 
> The following are changes since commit a1b1add07fa794974573d93483d68e373edfe7bd:
>   gro: Fix kcalloc argument order
> and are available in the git repository at:
>   git://git.kernel.org/pub/scm/linux/kernel/git/jkirsher/net-next master

Pulled, thanks Jeff.

[net-next 00/14][pull request] Intel Wired LAN Driver Updates

[Jeff Kirsher]

This series contains updates to e1000e, ixgbevf, igb and igbvf.
Majority of the patches are code cleanups of e1000e where code
is removed (Yeah!).  The other two e1000e patches are fixes.  The
first is to fix the maximum frame size for 82579 devices.  The second
fix is to resolve an issue with devices other than 82579 that suffer
from dropped transactions on platforms with deep C-states when
jumbo frames are enabled.

The ixgbevf patch is to ensure that the driver fetches the correct,
refreshed value for link status and speed when the values have changed.

The igb and igbvf patches are a solution to an issue Stefan Assmann
reported, where when the PF is up and igbvf is loaded, the MAC address
is not generated using eth_hw_addr_random().

The following are changes since commit a1b1add07fa794974573d93483d68e373edfe7bd:
  gro: Fix kcalloc argument order
and are available in the git repository at:
  git://git.kernel.org/pub/scm/linux/kernel/git/jkirsher/net-next master

Bruce Allan (11):
  e1000e: cleanup: remove e1000_set_d0_lplu_state()
  e1000e: cleanup: remove e1000_force_speed_duplex()
  e1000e: cleanup: rename e1000_get_cfg_done()
  e1000e: cleanup: remove e1000_get_phy_cfg_done()
  e1000e: cleanup: remove e1000_get_cable_length()
  e1000e: cleanup: remove e1000e_commit_phy()
  e1000e: correct maximum frame size on 82579
  e1000e: adjust PM QoS request
  e1000e: cleanup: remove unused #define
  e1000e: cleanup hw.h
  e1000e: cleanup: remove comments which are no longer applicable

Greg Rose (1):
  ixgbevf: Make sure link status and speed are fetched

Mitch A Williams (2):
  igb: Don't give VFs random MAC addresses
  igbvf: be sane about random MAC addresses

 drivers/net/ethernet/intel/e1000e/80003es2lan.c |   2 +-
 drivers/net/ethernet/intel/e1000e/82571.c       |   4 +-
 drivers/net/ethernet/intel/e1000e/defines.h     |   9 +-
 drivers/net/ethernet/intel/e1000e/e1000.h       |  12 +--
 drivers/net/ethernet/intel/e1000e/ethtool.c     |   5 +-
 drivers/net/ethernet/intel/e1000e/hw.h          |  25 +++---
 drivers/net/ethernet/intel/e1000e/ich8lan.c     |   4 +-
 drivers/net/ethernet/intel/e1000e/netdev.c      |  32 +++----
 drivers/net/ethernet/intel/e1000e/phy.c         | 114 +++++-------------------
 drivers/net/ethernet/intel/igb/igb_main.c       |   6 +-
 drivers/net/ethernet/intel/igbvf/netdev.c       |  24 +++--
 drivers/net/ethernet/intel/ixgbevf/ethtool.c    |   1 +
 12 files changed, 76 insertions(+), 162 deletions(-)

-- 
1.7.11.7

Re: [net-next 00/14][pull request] Intel Wired LAN Driver Updates

[David Miller]

From: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Date: Sat, 26 Jan 2013 22:41:01 -0800

> On Sun, 2013-01-27 at 01:28 -0500, David Miller wrote:
>> From: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
>> Date: Wed, 23 Jan 2013 14:44:36 -0800
>> 
>> > This series contains updates to ixgbe and ixgbevf.  Majority of the
>> > changes are against ixgbe PTP and SR-IOV.
>> > 
>> > The following are changes since commit 93b9c1ddd3fb4a5b67d512e534b30070f9ecec28:
>> >   Merge branch 'testing' of git://git.kernel.org/pub/scm/linux/kernel/git/klassert/ipsec-next
>> > and are available in the git repository at:
>> >   git://git.kernel.org/pub/scm/linux/kernel/git/jkirsher/net-next master
>> 
>> Pulled, thanks Jeff.
> 
> Hope you are feeling better!

I'm starting to turn the corner on this flu, thanks for your concern :-)

Re: [net-next 00/14][pull request] Intel Wired LAN Driver Updates

[Jeff Kirsher]

[-- Attachment #1: Type: text/plain, Size: 651 bytes --]

On Sun, 2013-01-27 at 01:28 -0500, David Miller wrote:
> From: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
> Date: Wed, 23 Jan 2013 14:44:36 -0800
> 
> > This series contains updates to ixgbe and ixgbevf.  Majority of the
> > changes are against ixgbe PTP and SR-IOV.
> > 
> > The following are changes since commit 93b9c1ddd3fb4a5b67d512e534b30070f9ecec28:
> >   Merge branch 'testing' of git://git.kernel.org/pub/scm/linux/kernel/git/klassert/ipsec-next
> > and are available in the git repository at:
> >   git://git.kernel.org/pub/scm/linux/kernel/git/jkirsher/net-next master
> 
> Pulled, thanks Jeff.

Hope you are feeling better!

[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 836 bytes --]

Re: [net-next 00/14][pull request] Intel Wired LAN Driver Updates

[David Miller]

From: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Date: Wed, 23 Jan 2013 14:44:36 -0800

> This series contains updates to ixgbe and ixgbevf.  Majority of the
> changes are against ixgbe PTP and SR-IOV.
> 
> The following are changes since commit 93b9c1ddd3fb4a5b67d512e534b30070f9ecec28:
>   Merge branch 'testing' of git://git.kernel.org/pub/scm/linux/kernel/git/klassert/ipsec-next
> and are available in the git repository at:
>   git://git.kernel.org/pub/scm/linux/kernel/git/jkirsher/net-next master

Pulled, thanks Jeff.

[net-next 00/14][pull request] Intel Wired LAN Driver Updates

[Jeff Kirsher]

This series contains updates to ixgbe and ixgbevf.  Majority of the
changes are against ixgbe PTP and SR-IOV.

The following are changes since commit 93b9c1ddd3fb4a5b67d512e534b30070f9ecec28:
  Merge branch 'testing' of git://git.kernel.org/pub/scm/linux/kernel/git/klassert/ipsec-next
and are available in the git repository at:
  git://git.kernel.org/pub/scm/linux/kernel/git/jkirsher/net-next master

Alexander Duyck (1):
  ixgbe: Inline Rx PTP descriptor handling

Donald Dutile (1):
  ixgbe: Limit number of reported VFs to device specific value

Greg Rose (4):
  ixgbe: Make mailbox ops initialization unconditional
  ixgbe: Modularize SR-IOV enablement code
  ixgbe: Implement PCI SR-IOV sysfs callback operation
  ixgbevf: Fix link speed message to support 100Mbps

Jacob Keller (8):
  ixgbe: ethtool ixgbe_diag_test cleanup
  ixgbe: add missing supported filters to get_ts_info
  ixgbe: Update ptp_overflow check comment and jiffies
  ixgbe: Use watchdog check in favor of BPF for detecting latched
    timestamp
  ixgbe: Add ptp work item to poll for the Tx timestamp
  ixgbe: add warning when scheduling reset
  ixgbe: Fix overwriting of rx_mtrl in ixgbe_ptp_hwtstamp_ioctl
  ixgbe: only compile ixgbe_debugfs.o when enabled

 drivers/net/ethernet/intel/ixgbe/Makefile         |   3 +-
 drivers/net/ethernet/intel/ixgbe/ixgbe.h          |  34 +++-
 drivers/net/ethernet/intel/ixgbe/ixgbe_debugfs.c  |   5 -
 drivers/net/ethernet/intel/ixgbe/ixgbe_ethtool.c  |  46 +++--
 drivers/net/ethernet/intel/ixgbe/ixgbe_main.c     |  62 +++++-
 drivers/net/ethernet/intel/ixgbe/ixgbe_ptp.c      | 221 ++++++++++------------
 drivers/net/ethernet/intel/ixgbe/ixgbe_sriov.c    | 192 ++++++++++++++-----
 drivers/net/ethernet/intel/ixgbe/ixgbe_sriov.h    |   6 +-
 drivers/net/ethernet/intel/ixgbevf/ixgbevf_main.c |  19 +-
 9 files changed, 375 insertions(+), 213 deletions(-)

-- 
1.7.11.7

[net-next 00/14][pull request] Intel Wired LAN Driver Updates

[Jeff Kirsher]

This series contains updates to e1000e and igb.  Most notably is the
added timestamp support in e1000e and additional software timestamp
support in igb.  As well as, the added thermal data support and SR-IOV
configuration support in igb.

The following are changes since commit d59577b6ffd313d0ab3be39cb1ab47e29bdc9182:
  sk-filter: Add ability to lock a socket filter program
and are available in the git repository at:
  git://git.kernel.org/pub/scm/linux/kernel/git/jkirsher/net-next master

Alexander Duyck (1):
  igb: Report L4 Rx hash via skb->l4_rxhash

Bjorn Helgaas (1):
  e1000e: Use standard #defines for PCIe Capability ASPM fields

Bruce Allan (3):
  e1000e: add ethtool .get_eee/.set_eee
  e1000e: add support for hardware timestamping on some devices
  e1000e: add support for IEEE-1588 PTP

Carolyn Wyborny (3):
  igb: Add i2c interface to igb.
  igb: Add support functions to access thermal data.
  igb: Enable hwmon data output for thermal sensors via I2C.

Greg Rose (1):
  igb: Enable SR-IOV configuration via PCI sysfs interface

Matthew Vick (5):
  igb: Add support for SW timestamping
  igb: Add timeout for PTP Tx work item
  igb: Add mechanism for detecting latched hardware Rx timestamp
  igb: Use in-kernel PTP_EV_PORT #define
  igb: Free any held skb that should have been timestamped on remove

 drivers/net/ethernet/intel/Kconfig             |  14 +
 drivers/net/ethernet/intel/e1000e/82571.c      |   2 +
 drivers/net/ethernet/intel/e1000e/Makefile     |   2 +-
 drivers/net/ethernet/intel/e1000e/defines.h    |  55 +++
 drivers/net/ethernet/intel/e1000e/e1000.h      |  56 ++-
 drivers/net/ethernet/intel/e1000e/ethtool.c    | 171 ++++++-
 drivers/net/ethernet/intel/e1000e/hw.h         |  13 +
 drivers/net/ethernet/intel/e1000e/ich8lan.c    |  26 +-
 drivers/net/ethernet/intel/e1000e/netdev.c     | 487 +++++++++++++++++-
 drivers/net/ethernet/intel/e1000e/ptp.c        | 276 +++++++++++
 drivers/net/ethernet/intel/igb/Makefile        |   2 +-
 drivers/net/ethernet/intel/igb/e1000_82575.c   | 140 ++++++
 drivers/net/ethernet/intel/igb/e1000_82575.h   |  26 +-
 drivers/net/ethernet/intel/igb/e1000_defines.h |  13 +
 drivers/net/ethernet/intel/igb/e1000_hw.h      |  20 +
 drivers/net/ethernet/intel/igb/e1000_regs.h    |  16 +
 drivers/net/ethernet/intel/igb/igb.h           |  46 +-
 drivers/net/ethernet/intel/igb/igb_ethtool.c   |  11 +
 drivers/net/ethernet/intel/igb/igb_hwmon.c     | 242 +++++++++
 drivers/net/ethernet/intel/igb/igb_main.c      | 653 +++++++++++++++++++++----
 drivers/net/ethernet/intel/igb/igb_ptp.c       |  65 ++-
 21 files changed, 2191 insertions(+), 145 deletions(-)
 create mode 100644 drivers/net/ethernet/intel/e1000e/ptp.c
 create mode 100644 drivers/net/ethernet/intel/igb/igb_hwmon.c

-- 
1.7.11.7

[net-next 00/14][pull request] Intel Wired LAN Driver Updates

[Jeff Kirsher]

This series contains updates to ixgbe, ixgbevf, igbvf, igb and
networking core (bridge).  Most notably is the addition of support
for local link multicast addresses in SR-IOV mode to the networking
core.

Also note, the ixgbe patch "ixgbe: Add support for pipeline reset" is
v2 based on community feedback.

The following are changes since commit e269ed26d4c9e6da1614709c1bf4c50ff8c8c5b3:
  l2tp: session is an array not a pointer
and are available in the git repository at:
  git://git.kernel.org/pub/scm/linux/kernel/git/jkirsher/net-next master

Alexander Duyck (2):
  ixgbe: Do not decrement budget in ixgbe_clean_rx_irq
  igb: Fix sparse warning in igb_ptp_rx_pktstamp

Carolyn Wyborny (1):
  igb: Update firmware version info for ethtool output.

Don Skidmore (1):
  ixgbe: Add support for pipeline reset

Emil Tantilov (1):
  ixgbe: clean up the condition for turning on/off the laser

Greg Rose (4):
  ixgbe: Fix return value from macvlan filter function
  ixgbe: Return success or failure on VF MAC filter set
  ixgbevf: Do not forward LLDP type frames
  igbvf: Check for error on dma_map_single call

Jakub Kicinski (1):
  ixgbevf: make sure probe fails on MSI-X enable error

John Fastabend (1):
  net, ixgbe: handle link local multicast addresses in SR-IOV mode

Matthew Vick (1):
  igb: Enable auto-crossover during forced operation on 82580 and
    above.

joshua.a.hay@intel.com (2):
  ixgbe: eliminate Smatch warnings in ixgbe_debugfs.c
  ixgbe: add/udpate descriptor maps in comments

 drivers/net/ethernet/intel/igb/e1000_defines.h    |  14 +++
 drivers/net/ethernet/intel/igb/e1000_mac.c        |   4 +
 drivers/net/ethernet/intel/igb/e1000_nvm.c        |  70 +++++++++++++
 drivers/net/ethernet/intel/igb/e1000_nvm.h        |  16 +++
 drivers/net/ethernet/intel/igb/e1000_phy.c        |  29 +++---
 drivers/net/ethernet/intel/igb/igb_main.c         |  76 +++++----------
 drivers/net/ethernet/intel/igb/igb_ptp.c          |   2 +-
 drivers/net/ethernet/intel/igbvf/netdev.c         |  13 +++
 drivers/net/ethernet/intel/ixgbe/ixgbe.h          |   1 +
 drivers/net/ethernet/intel/ixgbe/ixgbe_82599.c    | 114 ++++++++++++++++------
 drivers/net/ethernet/intel/ixgbe/ixgbe_common.c   |  70 ++++++++++++-
 drivers/net/ethernet/intel/ixgbe/ixgbe_debugfs.c  |  72 +++++++-------
 drivers/net/ethernet/intel/ixgbe/ixgbe_main.c     | 103 ++++++++++++-------
 drivers/net/ethernet/intel/ixgbe/ixgbe_sriov.c    |   5 +-
 drivers/net/ethernet/intel/ixgbevf/ixgbevf_main.c |  29 +++---
 drivers/net/ethernet/intel/ixgbevf/vf.c           |   3 +
 include/linux/etherdevice.h                       |  19 ++++
 net/bridge/br_device.c                            |   2 +-
 net/bridge/br_input.c                             |  15 ---
 net/bridge/br_private.h                           |   1 -
 net/bridge/br_sysfs_br.c                          |   3 +-
 21 files changed, 460 insertions(+), 201 deletions(-)

-- 
1.7.11.7

Re: [net-next 00/14][pull request] Intel Wired LAN Driver Updates

[David Miller]

From: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Date: Mon, 12 Mar 2012 21:03:38 -0700

> This series of patches contains fixes/cleanups igb, ixgbe and net.
> Majority of the patches are against ixgbe and this series is part
> one of three to update ixgbe.
> 
> The following are changes since commit f124488e4713dc9afa2028553261b1d399286e68:
>   bnx2x: code doesn't use stats for allocating Rx BDs
> and are available in the git repository at:
>   git://git.kernel.org/pub/scm/linux/kernel/git/jkirsher/net-next master

Pulled, thanks Jeff.

[net-next 00/14][pull request] Intel Wired LAN Driver Updates

[Jeff Kirsher]

This series of patches contains fixes/cleanups igb, ixgbe and net.
Majority of the patches are against ixgbe and this series is part
one of three to update ixgbe.

The following are changes since commit f124488e4713dc9afa2028553261b1d399286e68:
  bnx2x: code doesn't use stats for allocating Rx BDs
and are available in the git repository at:
  git://git.kernel.org/pub/scm/linux/kernel/git/jkirsher/net-next master

Alexander Duyck (12):
  ixgbe: add support for byte queue limits
  net: Fix issue with netdev_tx_reset_queue not resetting queue from
    XOFF state
  net: Add memory barriers to prevent possible race in byte queue
    limits
  ixgbe: Do no clear Tx status bits since eop_desc provides enough info
  ixgbe: Reorder adapter contents for better cache utilization
  ixgbe: Address issues with Tx WHTRESH value not being set correctly
  ixgbe: Correct Adaptive Interrupt Moderation so that it will change
    values
  ixgbe: Default to queue pairs when number of queues is less than CPUs
  ixgbe: Drop unnecessary napi_schedule_prep and spare blank line from
    ixgbe_intr
  ixgbe: Allocate rings as part of the q_vector
  ixgbe: Add iterator for cycling through rings on a q_vector
  ixgbe: Simplify logic for ethtool loopback frame creation and testing

Jeff Kirsher (2):
  igb: fix ethtool offline test
  ixgbe: remove tie between NAPI work limits and interrupt moderation

 drivers/net/ethernet/intel/igb/igb_ethtool.c     |    7 +
 drivers/net/ethernet/intel/igb/igb_main.c        |    3 +-
 drivers/net/ethernet/intel/ixgbe/ixgbe.h         |  115 +++--
 drivers/net/ethernet/intel/ixgbe/ixgbe_ethtool.c |   71 ++--
 drivers/net/ethernet/intel/ixgbe/ixgbe_main.c    |  547 +++++++++++-----------
 include/linux/netdevice.h                        |   50 ++-
 6 files changed, 409 insertions(+), 384 deletions(-)

-- 
1.7.7.6