2009-03-11 02:10:33

by Jeff Kirsher

Subject: [net-next PATCH 1/2] igbvf: add new driver to support 82576 virtual functions

From: Alexander Duyck <[email protected]>

This adds an igbvf driver to handle virtual functions provided
by the igb driver.

Signed-off-by: Alexander Duyck <[email protected]>
Signed-off-by: Jeff Kirsher <[email protected]>
---

drivers/net/Kconfig | 21
drivers/net/Makefile | 1
drivers/net/igbvf/Makefile | 38 +
drivers/net/igbvf/defines.h | 125 ++
drivers/net/igbvf/ethtool.c | 556 ++++++++
drivers/net/igbvf/igbvf.h | 333 +++++
drivers/net/igbvf/mbx.c | 350 +++++
drivers/net/igbvf/mbx.h | 75 +
drivers/net/igbvf/netdev.c | 2901 +++++++++++++++++++++++++++++++++++++++++++
drivers/net/igbvf/regs.h | 108 ++
drivers/net/igbvf/vf.c | 398 ++++++
drivers/net/igbvf/vf.h | 265 ++++
12 files changed, 5171 insertions(+), 0 deletions(-)
create mode 100644 drivers/net/igbvf/Makefile
create mode 100644 drivers/net/igbvf/defines.h
create mode 100644 drivers/net/igbvf/ethtool.c
create mode 100644 drivers/net/igbvf/igbvf.h
create mode 100644 drivers/net/igbvf/mbx.c
create mode 100644 drivers/net/igbvf/mbx.h
create mode 100644 drivers/net/igbvf/netdev.c
create mode 100644 drivers/net/igbvf/regs.h
create mode 100644 drivers/net/igbvf/vf.c
create mode 100644 drivers/net/igbvf/vf.h

diff --git a/drivers/net/Kconfig b/drivers/net/Kconfig
index 45403e6..e7f0713 100644
--- a/drivers/net/Kconfig
+++ b/drivers/net/Kconfig
@@ -2028,6 +2028,27 @@ config IGB_DCA
driver. DCA is a method for warming the CPU cache before data
is used, with the intent of lessening the impact of cache misses.

+config IGBVF
+ tristate "Intel(R) 82576 Virtual Function Ethernet support"
+ depends on PCI
+ ---help---
+ This driver supports Intel(R) 82576 virtual functions. For more
+ information on how to identify your adapter, go to the Adapter &
+ Driver ID Guide at:
+
+ <http://support.intel.com/support/network/adapter/pro100/21397.htm>
+
+ For general information and support, go to the Intel support
+ website at:
+
+ <http://support.intel.com>
+
+ More specific information on configuring the driver is in
+ <file:Documentation/networking/e1000.txt>.
+
+ To compile this driver as a module, choose M here. The module
+ will be called igbvf.
+
source "drivers/net/ixp2000/Kconfig"

config MYRI_SBUS
diff --git a/drivers/net/Makefile b/drivers/net/Makefile
index 3f8cb31..123d90b 100644
--- a/drivers/net/Makefile
+++ b/drivers/net/Makefile
@@ -6,6 +6,7 @@ obj-$(CONFIG_E1000) += e1000/
obj-$(CONFIG_E1000E) += e1000e/
obj-$(CONFIG_IBM_NEW_EMAC) += ibm_newemac/
obj-$(CONFIG_IGB) += igb/
+obj-$(CONFIG_IGBVF) += igbvf/
obj-$(CONFIG_IXGBE) += ixgbe/
obj-$(CONFIG_IXGB) += ixgb/
obj-$(CONFIG_IP1000) += ipg.o
diff --git a/drivers/net/igbvf/Makefile b/drivers/net/igbvf/Makefile
new file mode 100644
index 0000000..c2f150d
--- /dev/null
+++ b/drivers/net/igbvf/Makefile
@@ -0,0 +1,38 @@
+################################################################################
+#
+# Intel(R) 82576 Virtual Function Linux driver
+# Copyright(c) 2009 Intel Corporation.
+#
+# This program is free software; you can redistribute it and/or modify it
+# under the terms and conditions of the GNU General Public License,
+# version 2, as published by the Free Software Foundation.
+#
+# This program is distributed in the hope it will be useful, but WITHOUT
+# ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+# FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for
+# more details.
+#
+# You should have received a copy of the GNU General Public License along with
+# this program; if not, write to the Free Software Foundation, Inc.,
+# 51 Franklin St - Fifth Floor, Boston, MA 02110-1301 USA.
+#
+# The full GNU General Public License is included in this distribution in
+# the file called "COPYING".
+#
+# Contact Information:
+# e1000-devel Mailing List <[email protected]>
+# Intel Corporation, 5200 N.E. Elam Young Parkway, Hillsboro, OR 97124-6497
+#
+################################################################################
+
+#
+# Makefile for the Intel(R) 82576 VF ethernet driver
+#
+
+obj-$(CONFIG_IGBVF) += igbvf.o
+
+igbvf-objs := vf.o \
+ mbx.o \
+ ethtool.o \
+ netdev.o
+
diff --git a/drivers/net/igbvf/defines.h b/drivers/net/igbvf/defines.h
new file mode 100644
index 0000000..88a4753
--- /dev/null
+++ b/drivers/net/igbvf/defines.h
@@ -0,0 +1,125 @@
+/*******************************************************************************
+
+ Intel(R) 82576 Virtual Function Linux driver
+ Copyright(c) 1999 - 2009 Intel Corporation.
+
+ This program is free software; you can redistribute it and/or modify it
+ under the terms and conditions of the GNU General Public License,
+ version 2, as published by the Free Software Foundation.
+
+ This program is distributed in the hope it will be useful, but WITHOUT
+ ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for
+ more details.
+
+ You should have received a copy of the GNU General Public License along with
+ this program; if not, write to the Free Software Foundation, Inc.,
+ 51 Franklin St - Fifth Floor, Boston, MA 02110-1301 USA.
+
+ The full GNU General Public License is included in this distribution in
+ the file called "COPYING".
+
+ Contact Information:
+ e1000-devel Mailing List <[email protected]>
+ Intel Corporation, 5200 N.E. Elam Young Parkway, Hillsboro, OR 97124-6497
+
+*******************************************************************************/
+
+#ifndef _E1000_DEFINES_H_
+#define _E1000_DEFINES_H_
+
+/* Number of Transmit and Receive Descriptors must be a multiple of 8 */
+#define REQ_TX_DESCRIPTOR_MULTIPLE 8
+#define REQ_RX_DESCRIPTOR_MULTIPLE 8
+
+/* IVAR valid bit */
+#define E1000_IVAR_VALID 0x80
+
+/* Receive Descriptor bit definitions */
+#define E1000_RXD_STAT_DD 0x01 /* Descriptor Done */
+#define E1000_RXD_STAT_EOP 0x02 /* End of Packet */
+#define E1000_RXD_STAT_IXSM 0x04 /* Ignore checksum */
+#define E1000_RXD_STAT_VP 0x08 /* IEEE VLAN Packet */
+#define E1000_RXD_STAT_UDPCS 0x10 /* UDP xsum calculated */
+#define E1000_RXD_STAT_TCPCS 0x20 /* TCP xsum calculated */
+#define E1000_RXD_STAT_IPCS 0x40 /* IP xsum calculated */
+#define E1000_RXD_ERR_SE 0x02 /* Symbol Error */
+#define E1000_RXD_SPC_VLAN_MASK 0x0FFF /* VLAN ID is in lower 12 bits */
+
+#define E1000_RXDEXT_STATERR_CE 0x01000000
+#define E1000_RXDEXT_STATERR_SE 0x02000000
+#define E1000_RXDEXT_STATERR_SEQ 0x04000000
+#define E1000_RXDEXT_STATERR_CXE 0x10000000
+#define E1000_RXDEXT_STATERR_TCPE 0x20000000
+#define E1000_RXDEXT_STATERR_IPE 0x40000000
+#define E1000_RXDEXT_STATERR_RXE 0x80000000
+
+
+/* Same mask, but for extended and packet split descriptors */
+#define E1000_RXDEXT_ERR_FRAME_ERR_MASK ( \
+ E1000_RXDEXT_STATERR_CE | \
+ E1000_RXDEXT_STATERR_SE | \
+ E1000_RXDEXT_STATERR_SEQ | \
+ E1000_RXDEXT_STATERR_CXE | \
+ E1000_RXDEXT_STATERR_RXE)
+
+/* Device Control */
+#define E1000_CTRL_RST 0x04000000 /* Global reset */
+
+/* Device Status */
+#define E1000_STATUS_FD 0x00000001 /* Full duplex.0=half,1=full */
+#define E1000_STATUS_LU 0x00000002 /* Link up.0=no,1=link */
+#define E1000_STATUS_TXOFF 0x00000010 /* transmission paused */
+#define E1000_STATUS_SPEED_10 0x00000000 /* Speed 10Mb/s */
+#define E1000_STATUS_SPEED_100 0x00000040 /* Speed 100Mb/s */
+#define E1000_STATUS_SPEED_1000 0x00000080 /* Speed 1000Mb/s */
+
+#define SPEED_10 10
+#define SPEED_100 100
+#define SPEED_1000 1000
+#define HALF_DUPLEX 1
+#define FULL_DUPLEX 2
+
+/* Transmit Descriptor bit definitions */
+#define E1000_TXD_POPTS_IXSM 0x01 /* Insert IP checksum */
+#define E1000_TXD_POPTS_TXSM 0x02 /* Insert TCP/UDP checksum */
+#define E1000_TXD_CMD_DEXT 0x20000000 /* Descriptor extension (0 = legacy) */
+#define E1000_TXD_STAT_DD 0x00000001 /* Descriptor Done */
+
+#define MAX_JUMBO_FRAME_SIZE 0x3F00
+
+/* 802.1q VLAN Packet Size */
+#define VLAN_TAG_SIZE 4 /* 802.3ac tag (not DMA'd) */
+
+/* Error Codes */
+#define E1000_SUCCESS 0
+#define E1000_ERR_CONFIG 3
+#define E1000_ERR_MAC_INIT 5
+#define E1000_ERR_MBX 15
+
+#ifndef ETH_ADDR_LEN
+#define ETH_ADDR_LEN 6
+#endif
+
+/* SRRCTL bit definitions */
+#define E1000_SRRCTL_BSIZEPKT_SHIFT 10 /* Shift _right_ */
+#define E1000_SRRCTL_BSIZEHDRSIZE_MASK 0x00000F00
+#define E1000_SRRCTL_BSIZEHDRSIZE_SHIFT 2 /* Shift _left_ */
+#define E1000_SRRCTL_DESCTYPE_ADV_ONEBUF 0x02000000
+#define E1000_SRRCTL_DESCTYPE_HDR_SPLIT_ALWAYS 0x0A000000
+#define E1000_SRRCTL_DESCTYPE_MASK 0x0E000000
+#define E1000_SRRCTL_DROP_EN 0x80000000
+
+#define E1000_SRRCTL_BSIZEPKT_MASK 0x0000007F
+#define E1000_SRRCTL_BSIZEHDR_MASK 0x00003F00
+
+/* Additional Descriptor Control definitions */
+#define E1000_TXDCTL_QUEUE_ENABLE 0x02000000 /* Enable specific Tx Queue */
+#define E1000_RXDCTL_QUEUE_ENABLE 0x02000000 /* Enable specific Rx Queue */
+
+/* Direct Cache Access (DCA) definitions */
+#define E1000_DCA_TXCTRL_TX_WB_RO_EN (1 << 11) /* Tx Desc writeback RO bit */
+
+#define E1000_VF_INIT_TIMEOUT 200 /* Number of retries to clear RSTI */
+
+#endif /* _E1000_DEFINES_H_ */
diff --git a/drivers/net/igbvf/ethtool.c b/drivers/net/igbvf/ethtool.c
new file mode 100644
index 0000000..ea4d29f
--- /dev/null
+++ b/drivers/net/igbvf/ethtool.c
@@ -0,0 +1,556 @@
+/*******************************************************************************
+
+ Intel(R) 82576 Virtual Function Linux driver
+ Copyright(c) 2009 Intel Corporation.
+
+ This program is free software; you can redistribute it and/or modify it
+ under the terms and conditions of the GNU General Public License,
+ version 2, as published by the Free Software Foundation.
+
+ This program is distributed in the hope it will be useful, but WITHOUT
+ ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for
+ more details.
+
+ You should have received a copy of the GNU General Public License along with
+ this program; if not, write to the Free Software Foundation, Inc.,
+ 51 Franklin St - Fifth Floor, Boston, MA 02110-1301 USA.
+
+ The full GNU General Public License is included in this distribution in
+ the file called "COPYING".
+
+ Contact Information:
+ e1000-devel Mailing List <[email protected]>
+ Intel Corporation, 5200 N.E. Elam Young Parkway, Hillsboro, OR 97124-6497
+
+*******************************************************************************/
+
+/* ethtool support for igbvf */
+
+#include <linux/netdevice.h>
+#include <linux/ethtool.h>
+#include <linux/pci.h>
+#include <linux/delay.h>
+
+#include "igbvf.h"
+#include <linux/if_vlan.h>
+
+
+struct igbvf_stats {
+ char stat_string[ETH_GSTRING_LEN];
+ int sizeof_stat;
+ int stat_offset;
+ int base_stat_offset;
+};
+
+#define IGBVF_STAT(m, b) sizeof(((struct igbvf_adapter *)0)->m), \
+ offsetof(struct igbvf_adapter, m), \
+ offsetof(struct igbvf_adapter, b)
+
+static const struct igbvf_stats igbvf_gstrings_stats[] = {
+ { "rx_packets", IGBVF_STAT(stats.gprc, stats.base_gprc) },
+ { "tx_packets", IGBVF_STAT(stats.gptc, stats.base_gptc) },
+ { "rx_bytes", IGBVF_STAT(stats.gorc, stats.base_gorc) },
+ { "tx_bytes", IGBVF_STAT(stats.gotc, stats.base_gotc) },
+ { "multicast", IGBVF_STAT(stats.mprc, stats.base_mprc) },
+ { "lbrx_bytes", IGBVF_STAT(stats.gorlbc, stats.base_gorlbc) },
+ { "lbrx_packets", IGBVF_STAT(stats.gprlbc, stats.base_gprlbc) },
+ { "tx_restart_queue", IGBVF_STAT(restart_queue, zero_base) },
+ { "rx_long_byte_count", IGBVF_STAT(stats.gorc, stats.base_gorc) },
+ { "rx_csum_offload_good", IGBVF_STAT(hw_csum_good, zero_base) },
+ { "rx_csum_offload_errors", IGBVF_STAT(hw_csum_err, zero_base) },
+ { "rx_header_split", IGBVF_STAT(rx_hdr_split, zero_base) },
+ { "alloc_rx_buff_failed", IGBVF_STAT(alloc_rx_buff_failed, zero_base) },
+};
+
+#define IGBVF_GLOBAL_STATS_LEN \
+ (sizeof(igbvf_gstrings_stats) / sizeof(struct igbvf_stats))
+
+#define IGBVF_STATS_LEN (IGBVF_GLOBAL_STATS_LEN)
+
+static const char igbvf_gstrings_test[][ETH_GSTRING_LEN] = {
+ "Link test (on/offline)"
+};
+
+#define IGBVF_TEST_LEN ARRAY_SIZE(igbvf_gstrings_test)
+
+static int igbvf_get_settings(struct net_device *netdev,
+ struct ethtool_cmd *ecmd)
+{
+ struct igbvf_adapter *adapter = netdev_priv(netdev);
+ struct e1000_hw *hw = &adapter->hw;
+ u32 status;
+
+ ecmd->supported = SUPPORTED_1000baseT_Full;
+
+ ecmd->advertising = ADVERTISED_1000baseT_Full;
+
+ ecmd->port = -1;
+ ecmd->transceiver = XCVR_DUMMY1;
+
+ status = er32(STATUS);
+ if (status & E1000_STATUS_LU) {
+ if (status & E1000_STATUS_SPEED_1000)
+ ecmd->speed = 1000;
+ else if (status & E1000_STATUS_SPEED_100)
+ ecmd->speed = 100;
+ else
+ ecmd->speed = 10;
+
+ if (status & E1000_STATUS_FD)
+ ecmd->duplex = DUPLEX_FULL;
+ else
+ ecmd->duplex = DUPLEX_HALF;
+ } else {
+ ecmd->speed = -1;
+ ecmd->duplex = -1;
+ }
+
+ ecmd->autoneg = AUTONEG_DISABLE;
+
+ return 0;
+}
+
+static u32 igbvf_get_link(struct net_device *netdev)
+{
+ return netif_carrier_ok(netdev);
+}
+
+static int igbvf_set_settings(struct net_device *netdev,
+ struct ethtool_cmd *ecmd)
+{
+ return -EOPNOTSUPP;
+}
+
+static void igbvf_get_pauseparam(struct net_device *netdev,
+ struct ethtool_pauseparam *pause)
+{
+ return;
+}
+
+static int igbvf_set_pauseparam(struct net_device *netdev,
+ struct ethtool_pauseparam *pause)
+{
+ return -EOPNOTSUPP;
+}
+
+static u32 igbvf_get_tx_csum(struct net_device *netdev)
+{
+ return ((netdev->features & NETIF_F_IP_CSUM) != 0);
+}
+
+static int igbvf_set_tx_csum(struct net_device *netdev, u32 data)
+{
+ if (data)
+ netdev->features |= (NETIF_F_IP_CSUM | NETIF_F_IPV6_CSUM);
+ else
+ netdev->features &= ~(NETIF_F_IP_CSUM | NETIF_F_IPV6_CSUM);
+ return 0;
+}
+
+static int igbvf_set_tso(struct net_device *netdev, u32 data)
+{
+ struct igbvf_adapter *adapter = netdev_priv(netdev);
+ int i;
+ struct net_device *v_netdev;
+
+ if (data) {
+ netdev->features |= NETIF_F_TSO;
+ netdev->features |= NETIF_F_TSO6;
+ } else {
+ netdev->features &= ~NETIF_F_TSO;
+ netdev->features &= ~NETIF_F_TSO6;
+ /* disable TSO on all VLANs if they're present */
+ if (!adapter->vlgrp)
+ goto tso_out;
+ for (i = 0; i < VLAN_GROUP_ARRAY_LEN; i++) {
+ v_netdev = vlan_group_get_device(adapter->vlgrp, i);
+ if (!v_netdev)
+ continue;
+
+ v_netdev->features &= ~NETIF_F_TSO;
+ v_netdev->features &= ~NETIF_F_TSO6;
+ vlan_group_set_device(adapter->vlgrp, i, v_netdev);
+ }
+ }
+
+tso_out:
+ dev_info(&adapter->pdev->dev, "TSO is %s\n",
+ data ? "Enabled" : "Disabled");
+ adapter->flags |= FLAG_TSO_FORCE;
+ return 0;
+}
+
+static u32 igbvf_get_msglevel(struct net_device *netdev)
+{
+ struct igbvf_adapter *adapter = netdev_priv(netdev);
+ return adapter->msg_enable;
+}
+
+static void igbvf_set_msglevel(struct net_device *netdev, u32 data)
+{
+ struct igbvf_adapter *adapter = netdev_priv(netdev);
+ adapter->msg_enable = data;
+}
+
+static int igbvf_get_regs_len(struct net_device *netdev)
+{
+#define IGBVF_REGS_LEN 8
+ return IGBVF_REGS_LEN * sizeof(u32);
+}
+
+static void igbvf_get_regs(struct net_device *netdev,
+ struct ethtool_regs *regs, void *p)
+{
+ struct igbvf_adapter *adapter = netdev_priv(netdev);
+ struct e1000_hw *hw = &adapter->hw;
+ u32 *regs_buff = p;
+ u8 revision_id;
+
+ memset(p, 0, IGBVF_REGS_LEN * sizeof(u32));
+
+ pci_read_config_byte(adapter->pdev, PCI_REVISION_ID, &revision_id);
+
+ regs->version = (1 << 24) | (revision_id << 16) | adapter->pdev->device;
+
+ regs_buff[0] = er32(CTRL);
+ regs_buff[1] = er32(STATUS);
+
+ regs_buff[2] = er32(RDLEN(0));
+ regs_buff[3] = er32(RDH(0));
+ regs_buff[4] = er32(RDT(0));
+
+ regs_buff[5] = er32(TDLEN(0));
+ regs_buff[6] = er32(TDH(0));
+ regs_buff[7] = er32(TDT(0));
+}
+
+static int igbvf_get_eeprom_len(struct net_device *netdev)
+{
+ return 0;
+}
+
+static int igbvf_get_eeprom(struct net_device *netdev,
+ struct ethtool_eeprom *eeprom, u8 *bytes)
+{
+ return -EOPNOTSUPP;
+}
+
+static int igbvf_set_eeprom(struct net_device *netdev,
+ struct ethtool_eeprom *eeprom, u8 *bytes)
+{
+ return -EOPNOTSUPP;
+}
+
+static void igbvf_get_drvinfo(struct net_device *netdev,
+ struct ethtool_drvinfo *drvinfo)
+{
+ struct igbvf_adapter *adapter = netdev_priv(netdev);
+ char firmware_version[32] = "N/A";
+
+ strncpy(drvinfo->driver, igbvf_driver_name, 32);
+ strncpy(drvinfo->version, igbvf_driver_version, 32);
+ strncpy(drvinfo->fw_version, firmware_version, 32);
+ strncpy(drvinfo->bus_info, pci_name(adapter->pdev), 32);
+ drvinfo->regdump_len = igbvf_get_regs_len(netdev);
+ drvinfo->eedump_len = igbvf_get_eeprom_len(netdev);
+}
+
+static void igbvf_get_ringparam(struct net_device *netdev,
+ struct ethtool_ringparam *ring)
+{
+ struct igbvf_adapter *adapter = netdev_priv(netdev);
+ struct igbvf_ring *tx_ring = adapter->tx_ring;
+ struct igbvf_ring *rx_ring = adapter->rx_ring;
+
+ ring->rx_max_pending = IGBVF_MAX_RXD;
+ ring->tx_max_pending = IGBVF_MAX_TXD;
+ ring->rx_mini_max_pending = 0;
+ ring->rx_jumbo_max_pending = 0;
+ ring->rx_pending = rx_ring->count;
+ ring->tx_pending = tx_ring->count;
+ ring->rx_mini_pending = 0;
+ ring->rx_jumbo_pending = 0;
+}
+
+static int igbvf_set_ringparam(struct net_device *netdev,
+ struct ethtool_ringparam *ring)
+{
+ struct igbvf_adapter *adapter = netdev_priv(netdev);
+ struct igbvf_ring *tx_ring, *tx_old;
+ struct igbvf_ring *rx_ring, *rx_old;
+ int err;
+
+ if ((ring->rx_mini_pending) || (ring->rx_jumbo_pending))
+ return -EINVAL;
+
+ while (test_and_set_bit(__IGBVF_RESETTING, &adapter->state))
+ msleep(1);
+
+ if (netif_running(adapter->netdev))
+ igbvf_down(adapter);
+
+ tx_old = adapter->tx_ring;
+ rx_old = adapter->rx_ring;
+
+ err = -ENOMEM;
+ tx_ring = kzalloc(sizeof(struct igbvf_ring), GFP_KERNEL);
+ if (!tx_ring)
+ goto err_alloc_tx;
+ /*
+ * use a memcpy to save any previously configured
+ * items like napi structs from having to be
+ * reinitialized
+ */
+ memcpy(tx_ring, tx_old, sizeof(struct igbvf_ring));
+
+ rx_ring = kzalloc(sizeof(struct igbvf_ring), GFP_KERNEL);
+ if (!rx_ring)
+ goto err_alloc_rx;
+ memcpy(rx_ring, rx_old, sizeof(struct igbvf_ring));
+
+ adapter->tx_ring = tx_ring;
+ adapter->rx_ring = rx_ring;
+
+ rx_ring->count = max(ring->rx_pending, (u32)IGBVF_MIN_RXD);
+ rx_ring->count = min(rx_ring->count, (u32)(IGBVF_MAX_RXD));
+ rx_ring->count = ALIGN(rx_ring->count, REQ_RX_DESCRIPTOR_MULTIPLE);
+
+ tx_ring->count = max(ring->tx_pending, (u32)IGBVF_MIN_TXD);
+ tx_ring->count = min(tx_ring->count, (u32)(IGBVF_MAX_TXD));
+ tx_ring->count = ALIGN(tx_ring->count, REQ_TX_DESCRIPTOR_MULTIPLE);
+
+ if (netif_running(adapter->netdev)) {
+ /* Try to get new resources before deleting old */
+ err = igbvf_setup_rx_resources(adapter);
+ if (err)
+ goto err_setup_rx;
+ err = igbvf_setup_tx_resources(adapter);
+ if (err)
+ goto err_setup_tx;
+
+ /*
+ * restore the old in order to free it,
+ * then add in the new
+ */
+ adapter->rx_ring = rx_old;
+ adapter->tx_ring = tx_old;
+ igbvf_free_rx_resources(adapter);
+ igbvf_free_tx_resources(adapter);
+ kfree(tx_old);
+ kfree(rx_old);
+ adapter->rx_ring = rx_ring;
+ adapter->tx_ring = tx_ring;
+ err = igbvf_up(adapter);
+ if (err)
+ goto err_setup;
+ }
+
+ clear_bit(__IGBVF_RESETTING, &adapter->state);
+ return 0;
+err_setup_tx:
+ igbvf_free_rx_resources(adapter);
+err_setup_rx:
+ adapter->rx_ring = rx_old;
+ adapter->tx_ring = tx_old;
+ kfree(rx_ring);
+err_alloc_rx:
+ kfree(tx_ring);
+err_alloc_tx:
+ igbvf_up(adapter);
+err_setup:
+ clear_bit(__IGBVF_RESETTING, &adapter->state);
+ return err;
+}
+
+static int igbvf_link_test(struct igbvf_adapter *adapter, u64 *data)
+{
+ struct e1000_hw *hw = &adapter->hw;
+ *data = 0;
+
+ hw->mac.ops.check_for_link(hw);
+
+ if (!(er32(STATUS) & E1000_STATUS_LU))
+ *data = 1;
+
+ return *data;
+}
+
+static int igbvf_get_self_test_count(struct net_device *netdev)
+{
+ return IGBVF_TEST_LEN;
+}
+
+static int igbvf_get_stats_count(struct net_device *netdev)
+{
+ return IGBVF_STATS_LEN;
+}
+
+static void igbvf_diag_test(struct net_device *netdev,
+ struct ethtool_test *eth_test, u64 *data)
+{
+ struct igbvf_adapter *adapter = netdev_priv(netdev);
+
+ set_bit(__IGBVF_TESTING, &adapter->state);
+
+ /*
+ * Link test performed before hardware reset so autoneg doesn't
+ * interfere with test result
+ */
+ if (igbvf_link_test(adapter, &data[0]))
+ eth_test->flags |= ETH_TEST_FL_FAILED;
+
+ clear_bit(__IGBVF_TESTING, &adapter->state);
+ msleep_interruptible(4 * 1000);
+}
+
+static void igbvf_get_wol(struct net_device *netdev,
+ struct ethtool_wolinfo *wol)
+{
+ wol->supported = 0;
+ wol->wolopts = 0;
+
+ return;
+}
+
+static int igbvf_set_wol(struct net_device *netdev,
+ struct ethtool_wolinfo *wol)
+{
+ return -EOPNOTSUPP;
+}
+
+static int igbvf_phys_id(struct net_device *netdev, u32 data)
+{
+ return 0;
+}
+
+static int igbvf_get_coalesce(struct net_device *netdev,
+ struct ethtool_coalesce *ec)
+{
+ struct igbvf_adapter *adapter = netdev_priv(netdev);
+
+ if (adapter->itr_setting <= 3)
+ ec->rx_coalesce_usecs = adapter->itr_setting;
+ else
+ ec->rx_coalesce_usecs = adapter->itr_setting >> 2;
+
+ return 0;
+}
+
+static int igbvf_set_coalesce(struct net_device *netdev,
+ struct ethtool_coalesce *ec)
+{
+ struct igbvf_adapter *adapter = netdev_priv(netdev);
+ struct e1000_hw *hw = &adapter->hw;
+
+ if ((ec->rx_coalesce_usecs > IGBVF_MAX_ITR_USECS) ||
+ ((ec->rx_coalesce_usecs > 3) &&
+ (ec->rx_coalesce_usecs < IGBVF_MIN_ITR_USECS)) ||
+ (ec->rx_coalesce_usecs == 2))
+ return -EINVAL;
+
+ /* convert to rate of irq's per second */
+ if (ec->rx_coalesce_usecs && ec->rx_coalesce_usecs <= 3) {
+ adapter->itr = IGBVF_START_ITR;
+ adapter->itr_setting = ec->rx_coalesce_usecs;
+ } else {
+ adapter->itr = ec->rx_coalesce_usecs << 2;
+ adapter->itr_setting = adapter->itr;
+ }
+
+ writel(adapter->itr,
+ hw->hw_addr + adapter->rx_ring[0].itr_register);
+
+ return 0;
+}
+
+static int igbvf_nway_reset(struct net_device *netdev)
+{
+ struct igbvf_adapter *adapter = netdev_priv(netdev);
+ if (netif_running(netdev))
+ igbvf_reinit_locked(adapter);
+ return 0;
+}
+
+
+static void igbvf_get_ethtool_stats(struct net_device *netdev,
+ struct ethtool_stats *stats,
+ u64 *data)
+{
+ struct igbvf_adapter *adapter = netdev_priv(netdev);
+ int i;
+
+ igbvf_update_stats(adapter);
+ for (i = 0; i < IGBVF_GLOBAL_STATS_LEN; i++) {
+ char *p = (char *)adapter +
+ igbvf_gstrings_stats[i].stat_offset;
+ char *b = (char *)adapter +
+ igbvf_gstrings_stats[i].base_stat_offset;
+ data[i] = ((igbvf_gstrings_stats[i].sizeof_stat ==
+ sizeof(u64)) ? *(u64 *)p : *(u32 *)p) -
+ ((igbvf_gstrings_stats[i].sizeof_stat ==
+ sizeof(u64)) ? *(u64 *)b : *(u32 *)b);
+ }
+
+}
+
+static void igbvf_get_strings(struct net_device *netdev, u32 stringset,
+ u8 *data)
+{
+ u8 *p = data;
+ int i;
+
+ switch (stringset) {
+ case ETH_SS_TEST:
+ memcpy(data, *igbvf_gstrings_test, sizeof(igbvf_gstrings_test));
+ break;
+ case ETH_SS_STATS:
+ for (i = 0; i < IGBVF_GLOBAL_STATS_LEN; i++) {
+ memcpy(p, igbvf_gstrings_stats[i].stat_string,
+ ETH_GSTRING_LEN);
+ p += ETH_GSTRING_LEN;
+ }
+ break;
+ }
+}
+
+static const struct ethtool_ops igbvf_ethtool_ops = {
+ .get_settings = igbvf_get_settings,
+ .set_settings = igbvf_set_settings,
+ .get_drvinfo = igbvf_get_drvinfo,
+ .get_regs_len = igbvf_get_regs_len,
+ .get_regs = igbvf_get_regs,
+ .get_wol = igbvf_get_wol,
+ .set_wol = igbvf_set_wol,
+ .get_msglevel = igbvf_get_msglevel,
+ .set_msglevel = igbvf_set_msglevel,
+ .nway_reset = igbvf_nway_reset,
+ .get_link = igbvf_get_link,
+ .get_eeprom_len = igbvf_get_eeprom_len,
+ .get_eeprom = igbvf_get_eeprom,
+ .set_eeprom = igbvf_set_eeprom,
+ .get_ringparam = igbvf_get_ringparam,
+ .set_ringparam = igbvf_set_ringparam,
+ .get_pauseparam = igbvf_get_pauseparam,
+ .set_pauseparam = igbvf_set_pauseparam,
+ .get_tx_csum = igbvf_get_tx_csum,
+ .set_tx_csum = igbvf_set_tx_csum,
+ .get_sg = ethtool_op_get_sg,
+ .set_sg = ethtool_op_set_sg,
+ .get_tso = ethtool_op_get_tso,
+ .set_tso = igbvf_set_tso,
+ .self_test = igbvf_diag_test,
+ .get_strings = igbvf_get_strings,
+ .phys_id = igbvf_phys_id,
+ .get_ethtool_stats = igbvf_get_ethtool_stats,
+ .self_test_count = igbvf_get_self_test_count,
+ .get_stats_count = igbvf_get_stats_count,
+ .get_coalesce = igbvf_get_coalesce,
+ .set_coalesce = igbvf_set_coalesce,
+};
+
+void igbvf_set_ethtool_ops(struct net_device *netdev)
+{
+ /* have to "undeclare" const on this struct to remove warnings */
+ SET_ETHTOOL_OPS(netdev, (struct ethtool_ops *)&igbvf_ethtool_ops);
+}
diff --git a/drivers/net/igbvf/igbvf.h b/drivers/net/igbvf/igbvf.h
new file mode 100644
index 0000000..b0d5326
--- /dev/null
+++ b/drivers/net/igbvf/igbvf.h
@@ -0,0 +1,333 @@
+/*******************************************************************************
+
+ Intel(R) 82576 Virtual Function Linux driver
+ Copyright(c) 2009 Intel Corporation.
+
+ This program is free software; you can redistribute it and/or modify it
+ under the terms and conditions of the GNU General Public License,
+ version 2, as published by the Free Software Foundation.
+
+ This program is distributed in the hope it will be useful, but WITHOUT
+ ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for
+ more details.
+
+ You should have received a copy of the GNU General Public License along with
+ this program; if not, write to the Free Software Foundation, Inc.,
+ 51 Franklin St - Fifth Floor, Boston, MA 02110-1301 USA.
+
+ The full GNU General Public License is included in this distribution in
+ the file called "COPYING".
+
+ Contact Information:
+ e1000-devel Mailing List <[email protected]>
+ Intel Corporation, 5200 N.E. Elam Young Parkway, Hillsboro, OR 97124-6497
+
+*******************************************************************************/
+
+/* Linux PRO/1000 Ethernet Driver main header file */
+
+#ifndef _IGBVF_H_
+#define _IGBVF_H_
+
+#include <linux/types.h>
+#include <linux/timer.h>
+#include <linux/io.h>
+#include <linux/netdevice.h>
+
+
+#include "vf.h"
+
+/* Forward declarations */
+struct igbvf_info;
+struct igbvf_adapter;
+
+/* Interrupt defines */
+#define IGBVF_START_ITR 648 /* ~6000 ints/sec */
+
+/* Interrupt modes, as used by the IntMode paramter */
+#define IGBVF_INT_MODE_LEGACY 0
+#define IGBVF_INT_MODE_MSI 1
+#define IGBVF_INT_MODE_MSIX 2
+
+/* Tx/Rx descriptor defines */
+#define IGBVF_DEFAULT_TXD 256
+#define IGBVF_MAX_TXD 4096
+#define IGBVF_MIN_TXD 80
+
+#define IGBVF_DEFAULT_RXD 256
+#define IGBVF_MAX_RXD 4096
+#define IGBVF_MIN_RXD 80
+
+#define IGBVF_MIN_ITR_USECS 10 /* 100000 irq/sec */
+#define IGBVF_MAX_ITR_USECS 10000 /* 100 irq/sec */
+
+/* RX descriptor control thresholds.
+ * PTHRESH - MAC will consider prefetch if it has fewer than this number of
+ * descriptors available in its onboard memory.
+ * Setting this to 0 disables RX descriptor prefetch.
+ * HTHRESH - MAC will only prefetch if there are at least this many descriptors
+ * available in host memory.
+ * If PTHRESH is 0, this should also be 0.
+ * WTHRESH - RX descriptor writeback threshold - MAC will delay writing back
+ * descriptors until either it has this many to write back, or the
+ * ITR timer expires.
+ */
+#define IGBVF_RX_PTHRESH 16
+#define IGBVF_RX_HTHRESH 8
+#define IGBVF_RX_WTHRESH 1
+
+/* this is the size past which hardware will drop packets when setting LPE=0 */
+#define MAXIMUM_ETHERNET_VLAN_SIZE 1522
+
+#define IGBVF_FC_PAUSE_TIME 0x0680 /* 858 usec */
+
+/* How many Tx Descriptors do we need to call netif_wake_queue ? */
+#define IGBVF_TX_QUEUE_WAKE 32
+/* How many Rx Buffers do we bundle into one write to the hardware ? */
+#define IGBVF_RX_BUFFER_WRITE 16 /* Must be power of 2 */
+
+#define AUTO_ALL_MODES 0
+#define IGBVF_EEPROM_APME 0x0400
+
+#define IGBVF_MNG_VLAN_NONE (-1)
+
+/* Number of packet split data buffers (not including the header buffer) */
+#define PS_PAGE_BUFFERS (MAX_PS_BUFFERS - 1)
+
+enum igbvf_boards {
+ board_vf,
+};
+
+struct igbvf_queue_stats {
+ u64 packets;
+ u64 bytes;
+};
+
+/*
+ * wrappers around a pointer to a socket buffer,
+ * so a DMA handle can be stored along with the buffer
+ */
+struct igbvf_buffer {
+ dma_addr_t dma;
+ struct sk_buff *skb;
+ union {
+ /* Tx */
+ struct {
+ unsigned long time_stamp;
+ u16 length;
+ u16 next_to_watch;
+ };
+ /* Rx */
+ struct {
+ struct page *page;
+ u64 page_dma;
+ unsigned int page_offset;
+ };
+ };
+ struct page *page;
+};
+
+struct igbvf_ring {
+ struct igbvf_adapter *adapter; /* backlink */
+ void *desc; /* pointer to ring memory */
+ dma_addr_t dma; /* phys address of ring */
+ unsigned int size; /* length of ring in bytes */
+ unsigned int count; /* number of desc. in ring */
+
+ u16 next_to_use;
+ u16 next_to_clean;
+
+ u16 head;
+ u16 tail;
+
+ /* array of buffer information structs */
+ struct igbvf_buffer *buffer_info;
+ struct napi_struct napi;
+
+ char name[IFNAMSIZ + 5];
+ u32 eims_value;
+ u32 itr_val;
+ u16 itr_register;
+ int set_itr;
+
+ struct sk_buff *rx_skb_top;
+
+ struct igbvf_queue_stats stats;
+};
+
+/* board specific private data structure */
+struct igbvf_adapter {
+ struct timer_list watchdog_timer;
+ struct timer_list blink_timer;
+
+ struct work_struct reset_task;
+ struct work_struct watchdog_task;
+
+ const struct igbvf_info *ei;
+
+ struct vlan_group *vlgrp;
+ u32 bd_number;
+ u32 rx_buffer_len;
+ u32 polling_interval;
+ u16 mng_vlan_id;
+ u16 link_speed;
+ u16 link_duplex;
+
+ spinlock_t tx_queue_lock; /* prevent concurrent tail updates */
+
+ /* track device up/down/testing state */
+ unsigned long state;
+
+ /* Interrupt Throttle Rate */
+ u32 itr;
+ u32 itr_setting;
+ u16 tx_itr;
+ u16 rx_itr;
+
+ /*
+ * Tx
+ */
+ struct igbvf_ring *tx_ring /* One per active queue */
+ ____cacheline_aligned_in_smp;
+
+ unsigned long tx_queue_len;
+ unsigned int restart_queue;
+ u32 txd_cmd;
+
+ bool detect_tx_hung;
+ u8 tx_timeout_factor;
+
+ u32 tx_int_delay;
+ u32 tx_abs_int_delay;
+
+ unsigned int total_tx_bytes;
+ unsigned int total_tx_packets;
+ unsigned int total_rx_bytes;
+ unsigned int total_rx_packets;
+
+ /* Tx stats */
+ u32 tx_timeout_count;
+ u32 tx_fifo_head;
+ u32 tx_head_addr;
+ u32 tx_fifo_size;
+ u32 tx_dma_failed;
+
+ /*
+ * Rx
+ */
+ struct igbvf_ring *rx_ring;
+
+ u32 rx_int_delay;
+ u32 rx_abs_int_delay;
+
+ /* Rx stats */
+ u64 hw_csum_err;
+ u64 hw_csum_good;
+ u64 rx_hdr_split;
+ u32 alloc_rx_buff_failed;
+ u32 rx_dma_failed;
+
+ unsigned int rx_ps_hdr_size;
+ u32 max_frame_size;
+ u32 min_frame_size;
+
+ /* OS defined structs */
+ struct net_device *netdev;
+ struct pci_dev *pdev;
+ struct net_device_stats net_stats;
+ spinlock_t stats_lock; /* prevent concurrent stats updates */
+
+ /* structs defined in e1000_hw.h */
+ struct e1000_hw hw;
+
+ /* The VF counters don't clear on read so we have to get a base
+ * count on driver start up and always subtract that base on
+ * on the first update, thus the flag..
+ */
+ struct e1000_vf_stats stats;
+ u64 zero_base;
+
+ struct igbvf_ring test_tx_ring;
+ struct igbvf_ring test_rx_ring;
+ u32 test_icr;
+
+ u32 msg_enable;
+ struct msix_entry *msix_entries;
+ int int_mode;
+ u32 eims_enable_mask;
+ u32 eims_other;
+ u32 int_counter0;
+ u32 int_counter1;
+
+ u32 eeprom_wol;
+ u32 wol;
+ u32 pba;
+
+ bool fc_autoneg;
+
+ unsigned long led_status;
+
+ unsigned int flags;
+};
+
+struct igbvf_info {
+ enum e1000_mac_type mac;
+ unsigned int flags;
+ u32 pba;
+ void (*init_ops)(struct e1000_hw *);
+ s32 (*get_variants)(struct igbvf_adapter *);
+};
+
+/* hardware capability, feature, and workaround flags */
+#define FLAG_HAS_HW_VLAN_FILTER (1 << 0)
+#define FLAG_HAS_JUMBO_FRAMES (1 << 1)
+#define FLAG_MSI_ENABLED (1 << 2)
+#define FLAG_RX_CSUM_ENABLED (1 << 3)
+#define FLAG_TSO_FORCE (1 << 4)
+
+#define IGBVF_DESC_UNUSED(R) \
+ ((((R)->next_to_clean > (R)->next_to_use) ? 0 : (R)->count) + \
+ (R)->next_to_clean - (R)->next_to_use - 1)
+
+#define IGBVF_RX_DESC_ADV(R, i) \
+ (&(((union e1000_adv_rx_desc *)((R).desc))[i]))
+#define IGBVF_TX_DESC_ADV(R, i) \
+ (&(((union e1000_adv_tx_desc *)((R).desc))[i]))
+#define IGBVF_TX_CTXTDESC_ADV(R, i) \
+ (&(((struct e1000_adv_tx_context_desc *)((R).desc))[i]))
+
+enum igbvf_state_t {
+ __IGBVF_TESTING,
+ __IGBVF_RESETTING,
+ __IGBVF_DOWN
+};
+
+enum latency_range {
+ lowest_latency = 0,
+ low_latency = 1,
+ bulk_latency = 2,
+ latency_invalid = 255
+};
+
+extern char igbvf_driver_name[];
+extern const char igbvf_driver_version[];
+
+extern void igbvf_check_options(struct igbvf_adapter *adapter);
+extern void igbvf_set_ethtool_ops(struct net_device *netdev);
+
+extern int igbvf_up(struct igbvf_adapter *adapter);
+extern void igbvf_down(struct igbvf_adapter *adapter);
+extern void igbvf_reinit_locked(struct igbvf_adapter *adapter);
+extern void igbvf_reset(struct igbvf_adapter *adapter);
+extern int igbvf_setup_rx_resources(struct igbvf_adapter *adapter);
+extern int igbvf_setup_tx_resources(struct igbvf_adapter *adapter);
+extern void igbvf_free_rx_resources(struct igbvf_adapter *adapter);
+extern void igbvf_free_tx_resources(struct igbvf_adapter *adapter);
+extern void igbvf_update_stats(struct igbvf_adapter *adapter);
+extern void igbvf_set_interrupt_capability(struct igbvf_adapter *adapter);
+extern void igbvf_reset_interrupt_capability(struct igbvf_adapter *adapter);
+
+extern unsigned int copybreak;
+
+#endif /* _IGBVF_H_ */
diff --git a/drivers/net/igbvf/mbx.c b/drivers/net/igbvf/mbx.c
new file mode 100644
index 0000000..819a8ec
--- /dev/null
+++ b/drivers/net/igbvf/mbx.c
@@ -0,0 +1,350 @@
+/*******************************************************************************
+
+ Intel(R) 82576 Virtual Function Linux driver
+ Copyright(c) 2009 Intel Corporation.
+
+ This program is free software; you can redistribute it and/or modify it
+ under the terms and conditions of the GNU General Public License,
+ version 2, as published by the Free Software Foundation.
+
+ This program is distributed in the hope it will be useful, but WITHOUT
+ ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for
+ more details.
+
+ You should have received a copy of the GNU General Public License along with
+ this program; if not, write to the Free Software Foundation, Inc.,
+ 51 Franklin St - Fifth Floor, Boston, MA 02110-1301 USA.
+
+ The full GNU General Public License is included in this distribution in
+ the file called "COPYING".
+
+ Contact Information:
+ e1000-devel Mailing List <[email protected]>
+ Intel Corporation, 5200 N.E. Elam Young Parkway, Hillsboro, OR 97124-6497
+
+*******************************************************************************/
+
+#include "mbx.h"
+
+/**
+ * e1000_poll_for_msg - Wait for message notification
+ * @hw: pointer to the HW structure
+ *
+ * returns SUCCESS if it successfully received a message notification
+ **/
+static s32 e1000_poll_for_msg(struct e1000_hw *hw)
+{
+ struct e1000_mbx_info *mbx = &hw->mbx;
+ int countdown = mbx->timeout;
+
+ if (!mbx->ops.check_for_msg)
+ goto out;
+
+ while (countdown && mbx->ops.check_for_msg(hw)) {
+ countdown--;
+ udelay(mbx->usec_delay);
+ }
+
+ /* if we failed, all future posted messages fail until reset */
+ if (!countdown)
+ mbx->timeout = 0;
+out:
+ return countdown ? E1000_SUCCESS : -E1000_ERR_MBX;
+}
+
+/**
+ * e1000_poll_for_ack - Wait for message acknowledgement
+ * @hw: pointer to the HW structure
+ *
+ * returns SUCCESS if it successfully received a message acknowledgement
+ **/
+static s32 e1000_poll_for_ack(struct e1000_hw *hw)
+{
+ struct e1000_mbx_info *mbx = &hw->mbx;
+ int countdown = mbx->timeout;
+
+ if (!mbx->ops.check_for_ack)
+ goto out;
+
+ while (countdown && mbx->ops.check_for_ack(hw)) {
+ countdown--;
+ udelay(mbx->usec_delay);
+ }
+
+ /* if we failed, all future posted messages fail until reset */
+ if (!countdown)
+ mbx->timeout = 0;
+out:
+ return countdown ? E1000_SUCCESS : -E1000_ERR_MBX;
+}
+
+/**
+ * e1000_read_posted_mbx - Wait for message notification and receive message
+ * @hw: pointer to the HW structure
+ * @msg: The message buffer
+ * @size: Length of buffer
+ *
+ * returns SUCCESS if it successfully received a message notification and
+ * copied it into the receive buffer.
+ **/
+static s32 e1000_read_posted_mbx(struct e1000_hw *hw, u32 *msg, u16 size)
+{
+ struct e1000_mbx_info *mbx = &hw->mbx;
+ s32 ret_val = -E1000_ERR_MBX;
+
+ if (!mbx->ops.read)
+ goto out;
+
+ ret_val = e1000_poll_for_msg(hw);
+
+ /* if ack received read message, otherwise we timed out */
+ if (!ret_val)
+ ret_val = mbx->ops.read(hw, msg, size);
+out:
+ return ret_val;
+}
+
+/**
+ * e1000_write_posted_mbx - Write a message to the mailbox, wait for ack
+ * @hw: pointer to the HW structure
+ * @msg: The message buffer
+ * @size: Length of buffer
+ *
+ * returns SUCCESS if it successfully copied message into the buffer and
+ * received an ack to that message within delay * timeout period
+ **/
+static s32 e1000_write_posted_mbx(struct e1000_hw *hw, u32 *msg, u16 size)
+{
+ struct e1000_mbx_info *mbx = &hw->mbx;
+ s32 ret_val = -E1000_ERR_MBX;
+
+ /* exit if we either can't write or there isn't a defined timeout */
+ if (!mbx->ops.write || !mbx->timeout)
+ goto out;
+
+ /* send msg*/
+ ret_val = mbx->ops.write(hw, msg, size);
+
+ /* if msg sent wait until we receive an ack */
+ if (!ret_val)
+ ret_val = e1000_poll_for_ack(hw);
+out:
+ return ret_val;
+}
+
+/**
+ * e1000_read_v2p_mailbox - read v2p mailbox
+ * @hw: pointer to the HW structure
+ *
+ * This function is used to read the v2p mailbox without losing the read to
+ * clear status bits.
+ **/
+static u32 e1000_read_v2p_mailbox(struct e1000_hw *hw)
+{
+ u32 v2p_mailbox = er32(V2PMAILBOX(0));
+
+ v2p_mailbox |= hw->dev_spec.vf.v2p_mailbox;
+ hw->dev_spec.vf.v2p_mailbox |= v2p_mailbox & E1000_V2PMAILBOX_R2C_BITS;
+
+ return v2p_mailbox;
+}
+
+/**
+ * e1000_check_for_bit_vf - Determine if a status bit was set
+ * @hw: pointer to the HW structure
+ * @mask: bitmask for bits to be tested and cleared
+ *
+ * This function is used to check for the read to clear bits within
+ * the V2P mailbox.
+ **/
+static s32 e1000_check_for_bit_vf(struct e1000_hw *hw, u32 mask)
+{
+ u32 v2p_mailbox = e1000_read_v2p_mailbox(hw);
+ s32 ret_val = -E1000_ERR_MBX;
+
+ if (v2p_mailbox & mask)
+ ret_val = E1000_SUCCESS;
+
+ hw->dev_spec.vf.v2p_mailbox &= ~mask;
+
+ return ret_val;
+}
+
+/**
+ * e1000_check_for_msg_vf - checks to see if the PF has sent mail
+ * @hw: pointer to the HW structure
+ *
+ * returns SUCCESS if the PF has set the Status bit or else ERR_MBX
+ **/
+static s32 e1000_check_for_msg_vf(struct e1000_hw *hw)
+{
+ s32 ret_val = -E1000_ERR_MBX;
+
+ if (!e1000_check_for_bit_vf(hw, E1000_V2PMAILBOX_PFSTS)) {
+ ret_val = E1000_SUCCESS;
+ hw->mbx.stats.reqs++;
+ }
+
+ return ret_val;
+}
+
+/**
+ * e1000_check_for_ack_vf - checks to see if the PF has ACK'd
+ * @hw: pointer to the HW structure
+ *
+ * returns SUCCESS if the PF has set the ACK bit or else ERR_MBX
+ **/
+static s32 e1000_check_for_ack_vf(struct e1000_hw *hw)
+{
+ s32 ret_val = -E1000_ERR_MBX;
+
+ if (!e1000_check_for_bit_vf(hw, E1000_V2PMAILBOX_PFACK)) {
+ ret_val = E1000_SUCCESS;
+ hw->mbx.stats.acks++;
+ }
+
+ return ret_val;
+}
+
+/**
+ * e1000_check_for_rst_vf - checks to see if the PF has reset
+ * @hw: pointer to the HW structure
+ *
+ * returns true if the PF has set the reset done bit or else false
+ **/
+static s32 e1000_check_for_rst_vf(struct e1000_hw *hw)
+{
+ s32 ret_val = -E1000_ERR_MBX;
+
+ if (!e1000_check_for_bit_vf(hw, (E1000_V2PMAILBOX_RSTD |
+ E1000_V2PMAILBOX_RSTI))) {
+ ret_val = E1000_SUCCESS;
+ hw->mbx.stats.rsts++;
+ }
+
+ return ret_val;
+}
+
+/**
+ * e1000_obtain_mbx_lock_vf - obtain mailbox lock
+ * @hw: pointer to the HW structure
+ *
+ * return SUCCESS if we obtained the mailbox lock
+ **/
+static s32 e1000_obtain_mbx_lock_vf(struct e1000_hw *hw)
+{
+ s32 ret_val = -E1000_ERR_MBX;
+
+ /* Take ownership of the buffer */
+ ew32(V2PMAILBOX(0), E1000_V2PMAILBOX_VFU);
+
+ /* reserve mailbox for vf use */
+ if (e1000_read_v2p_mailbox(hw) & E1000_V2PMAILBOX_VFU)
+ ret_val = E1000_SUCCESS;
+
+ return ret_val;
+}
+
+/**
+ * e1000_write_mbx_vf - Write a message to the mailbox
+ * @hw: pointer to the HW structure
+ * @msg: The message buffer
+ * @size: Length of buffer
+ *
+ * returns SUCCESS if it successfully copied message into the buffer
+ **/
+static s32 e1000_write_mbx_vf(struct e1000_hw *hw, u32 *msg, u16 size)
+{
+ s32 err;
+ u16 i;
+
+ /* lock the mailbox to prevent pf/vf race condition */
+ err = e1000_obtain_mbx_lock_vf(hw);
+ if (err)
+ goto out_no_write;
+
+ /* flush any ack or msg as we are going to overwrite mailbox */
+ e1000_check_for_ack_vf(hw);
+ e1000_check_for_msg_vf(hw);
+
+ /* copy the caller specified message to the mailbox memory buffer */
+ for (i = 0; i < size; i++)
+ array_ew32(VMBMEM(0), i, msg[i]);
+
+ /* update stats */
+ hw->mbx.stats.msgs_tx++;
+
+ /* Drop VFU and interrupt the PF to tell it a message has been sent */
+ ew32(V2PMAILBOX(0), E1000_V2PMAILBOX_REQ);
+
+out_no_write:
+ return err;
+}
+
+/**
+ * e1000_read_mbx_vf - Reads a message from the inbox intended for vf
+ * @hw: pointer to the HW structure
+ * @msg: The message buffer
+ * @size: Length of buffer
+ *
+ * returns SUCCESS if it successfuly read message from buffer
+ **/
+static s32 e1000_read_mbx_vf(struct e1000_hw *hw, u32 *msg, u16 size)
+{
+ s32 err;
+ u16 i;
+
+ /* lock the mailbox to prevent pf/vf race condition */
+ err = e1000_obtain_mbx_lock_vf(hw);
+ if (err)
+ goto out_no_read;
+
+ /* copy the message from the mailbox memory buffer */
+ for (i = 0; i < size; i++)
+ msg[i] = array_er32(VMBMEM(0), i);
+
+ /* Acknowledge receipt and release mailbox, then we're done */
+ ew32(V2PMAILBOX(0), E1000_V2PMAILBOX_ACK);
+
+ /* update stats */
+ hw->mbx.stats.msgs_rx++;
+
+out_no_read:
+ return err;
+}
+
+/**
+ * e1000_init_mbx_params_vf - set initial values for vf mailbox
+ * @hw: pointer to the HW structure
+ *
+ * Initializes the hw->mbx struct to correct values for vf mailbox
+ */
+s32 e1000_init_mbx_params_vf(struct e1000_hw *hw)
+{
+ struct e1000_mbx_info *mbx = &hw->mbx;
+
+ /* start mailbox as timed out and let the reset_hw call set the timeout
+ * value to being communications */
+ mbx->timeout = 0;
+ mbx->usec_delay = E1000_VF_MBX_INIT_DELAY;
+
+ mbx->size = E1000_VFMAILBOX_SIZE;
+
+ mbx->ops.read = e1000_read_mbx_vf;
+ mbx->ops.write = e1000_write_mbx_vf;
+ mbx->ops.read_posted = e1000_read_posted_mbx;
+ mbx->ops.write_posted = e1000_write_posted_mbx;
+ mbx->ops.check_for_msg = e1000_check_for_msg_vf;
+ mbx->ops.check_for_ack = e1000_check_for_ack_vf;
+ mbx->ops.check_for_rst = e1000_check_for_rst_vf;
+
+ mbx->stats.msgs_tx = 0;
+ mbx->stats.msgs_rx = 0;
+ mbx->stats.reqs = 0;
+ mbx->stats.acks = 0;
+ mbx->stats.rsts = 0;
+
+ return E1000_SUCCESS;
+}
+
diff --git a/drivers/net/igbvf/mbx.h b/drivers/net/igbvf/mbx.h
new file mode 100644
index 0000000..4938609
--- /dev/null
+++ b/drivers/net/igbvf/mbx.h
@@ -0,0 +1,75 @@
+/*******************************************************************************
+
+ Intel(R) 82576 Virtual Function Linux driver
+ Copyright(c) 1999 - 2009 Intel Corporation.
+
+ This program is free software; you can redistribute it and/or modify it
+ under the terms and conditions of the GNU General Public License,
+ version 2, as published by the Free Software Foundation.
+
+ This program is distributed in the hope it will be useful, but WITHOUT
+ ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for
+ more details.
+
+ You should have received a copy of the GNU General Public License along with
+ this program; if not, write to the Free Software Foundation, Inc.,
+ 51 Franklin St - Fifth Floor, Boston, MA 02110-1301 USA.
+
+ The full GNU General Public License is included in this distribution in
+ the file called "COPYING".
+
+ Contact Information:
+ e1000-devel Mailing List <[email protected]>
+ Intel Corporation, 5200 N.E. Elam Young Parkway, Hillsboro, OR 97124-6497
+
+*******************************************************************************/
+
+#ifndef _E1000_MBX_H_
+#define _E1000_MBX_H_
+
+#include "vf.h"
+
+#define E1000_V2PMAILBOX_REQ 0x00000001 /* Request for PF Ready bit */
+#define E1000_V2PMAILBOX_ACK 0x00000002 /* Ack PF message received */
+#define E1000_V2PMAILBOX_VFU 0x00000004 /* VF owns the mailbox buffer */
+#define E1000_V2PMAILBOX_PFU 0x00000008 /* PF owns the mailbox buffer */
+#define E1000_V2PMAILBOX_PFSTS 0x00000010 /* PF wrote a message in the MB */
+#define E1000_V2PMAILBOX_PFACK 0x00000020 /* PF ack the previous VF msg */
+#define E1000_V2PMAILBOX_RSTI 0x00000040 /* PF has reset indication */
+#define E1000_V2PMAILBOX_RSTD 0x00000080 /* PF has indicated reset done */
+#define E1000_V2PMAILBOX_R2C_BITS 0x000000B0 /* All read to clear bits */
+
+#define E1000_VFMAILBOX_SIZE 16 /* 16 32 bit words - 64 bytes */
+
+/* If it's a E1000_VF_* msg then it originates in the VF and is sent to the
+ * PF. The reverse is true if it is E1000_PF_*.
+ * Message ACK's are the value or'd with 0xF0000000
+ */
+#define E1000_VT_MSGTYPE_ACK 0x80000000 /* Messages below or'd with
+ * this are the ACK */
+#define E1000_VT_MSGTYPE_NACK 0x40000000 /* Messages below or'd with
+ * this are the NACK */
+#define E1000_VT_MSGTYPE_CTS 0x20000000 /* Indicates that VF is still
+ clear to send requests */
+
+/* We have a total wait time of 1s for vf mailbox posted messages */
+#define E1000_VF_MBX_INIT_TIMEOUT 2000 /* retry count for mailbox timeout */
+#define E1000_VF_MBX_INIT_DELAY 500 /* usec delay between retries */
+
+#define E1000_VT_MSGINFO_SHIFT 16
+/* bits 23:16 are used for exra info for certain messages */
+#define E1000_VT_MSGINFO_MASK (0xFF << E1000_VT_MSGINFO_SHIFT)
+
+#define E1000_VF_RESET 0x01 /* VF requests reset */
+#define E1000_VF_SET_MAC_ADDR 0x02 /* VF requests PF to set MAC addr */
+#define E1000_VF_SET_MULTICAST 0x03 /* VF requests PF to set MC addr */
+#define E1000_VF_SET_VLAN 0x04 /* VF requests PF to set VLAN */
+#define E1000_VF_SET_LPE 0x05 /* VF requests PF to set VMOLR.LPE */
+
+#define E1000_PF_CONTROL_MSG 0x0100 /* PF control message */
+
+void e1000_init_mbx_ops_generic(struct e1000_hw *hw);
+s32 e1000_init_mbx_params_vf(struct e1000_hw *);
+
+#endif /* _E1000_MBX_H_ */
diff --git a/drivers/net/igbvf/netdev.c b/drivers/net/igbvf/netdev.c
new file mode 100644
index 0000000..54457e0
--- /dev/null
+++ b/drivers/net/igbvf/netdev.c
@@ -0,0 +1,2901 @@
+/*******************************************************************************
+
+ Intel(R) 82576 Virtual Function Linux driver
+ Copyright(c) 2009 Intel Corporation.
+
+ This program is free software; you can redistribute it and/or modify it
+ under the terms and conditions of the GNU General Public License,
+ version 2, as published by the Free Software Foundation.
+
+ This program is distributed in the hope it will be useful, but WITHOUT
+ ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for
+ more details.
+
+ You should have received a copy of the GNU General Public License along with
+ this program; if not, write to the Free Software Foundation, Inc.,
+ 51 Franklin St - Fifth Floor, Boston, MA 02110-1301 USA.
+
+ The full GNU General Public License is included in this distribution in
+ the file called "COPYING".
+
+ Contact Information:
+ e1000-devel Mailing List <[email protected]>
+ Intel Corporation, 5200 N.E. Elam Young Parkway, Hillsboro, OR 97124-6497
+
+*******************************************************************************/
+
+#include <linux/module.h>
+#include <linux/types.h>
+#include <linux/init.h>
+#include <linux/pci.h>
+#include <linux/vmalloc.h>
+#include <linux/pagemap.h>
+#include <linux/delay.h>
+#include <linux/netdevice.h>
+#include <linux/tcp.h>
+#include <linux/ipv6.h>
+#include <net/checksum.h>
+#include <net/ip6_checksum.h>
+#include <linux/mii.h>
+#include <linux/ethtool.h>
+#include <linux/if_vlan.h>
+#include <linux/pm_qos_params.h>
+
+#include "igbvf.h"
+
+#define DRV_VERSION "1.0.0-k0"
+char igbvf_driver_name[] = "igbvf";
+const char igbvf_driver_version[] = DRV_VERSION;
+static const char igbvf_driver_string[] =
+ "Intel(R) Virtual Function Network Driver";
+static const char igbvf_copyright[] = "Copyright (c) 2009 Intel Corporation.";
+
+static int igbvf_poll(struct napi_struct *napi, int budget);
+
+static struct igbvf_info igbvf_vf_info = {
+ .mac = e1000_vfadapt,
+ .flags = FLAG_HAS_JUMBO_FRAMES
+ | FLAG_RX_CSUM_ENABLED,
+ .pba = 10,
+ .init_ops = e1000_init_function_pointers_vf,
+};
+
+static const struct igbvf_info *igbvf_info_tbl[] = {
+ [board_vf] = &igbvf_vf_info,
+};
+
+/**
+ * igbvf_desc_unused - calculate if we have unused descriptors
+ **/
+static int igbvf_desc_unused(struct igbvf_ring *ring)
+{
+ if (ring->next_to_clean > ring->next_to_use)
+ return ring->next_to_clean - ring->next_to_use - 1;
+
+ return ring->count + ring->next_to_clean - ring->next_to_use - 1;
+}
+
+/**
+ * igbvf_receive_skb - helper function to handle Rx indications
+ * @adapter: board private structure
+ * @status: descriptor status field as written by hardware
+ * @vlan: descriptor vlan field as written by hardware (no le/be conversion)
+ * @skb: pointer to sk_buff to be indicated to stack
+ **/
+static void igbvf_receive_skb(struct igbvf_adapter *adapter,
+ struct net_device *netdev,
+ struct sk_buff *skb,
+ u32 status, u16 vlan)
+{
+ if (adapter->vlgrp && (status & E1000_RXD_STAT_VP))
+ vlan_hwaccel_receive_skb(skb, adapter->vlgrp,
+ le16_to_cpu(vlan) &
+ E1000_RXD_SPC_VLAN_MASK);
+ else
+ netif_receive_skb(skb);
+
+ netdev->last_rx = jiffies;
+}
+
+static inline void igbvf_rx_checksum_adv(struct igbvf_adapter *adapter,
+ u32 status_err, struct sk_buff *skb)
+{
+ skb->ip_summed = CHECKSUM_NONE;
+
+ /* Ignore Checksum bit is set or checksum is disabled through ethtool */
+ if ((status_err & E1000_RXD_STAT_IXSM))
+ return;
+ /* TCP/UDP checksum error bit is set */
+ if (status_err &
+ (E1000_RXDEXT_STATERR_TCPE | E1000_RXDEXT_STATERR_IPE)) {
+ /* let the stack verify checksum errors */
+ adapter->hw_csum_err++;
+ return;
+ }
+ /* It must be a TCP or UDP packet with a valid checksum */
+ if (status_err & (E1000_RXD_STAT_TCPCS | E1000_RXD_STAT_UDPCS))
+ skb->ip_summed = CHECKSUM_UNNECESSARY;
+
+ adapter->hw_csum_good++;
+}
+
+/**
+ * igbvf_alloc_rx_buffers - Replace used receive buffers; packet split
+ * @rx_ring: address of ring structure to repopulate
+ * @cleaned_count: number of buffers to repopulate
+ **/
+static void igbvf_alloc_rx_buffers(struct igbvf_ring *rx_ring,
+ int cleaned_count)
+{
+ struct igbvf_adapter *adapter = rx_ring->adapter;
+ struct net_device *netdev = adapter->netdev;
+ struct pci_dev *pdev = adapter->pdev;
+ union e1000_adv_rx_desc *rx_desc;
+ struct igbvf_buffer *buffer_info;
+ struct sk_buff *skb;
+ unsigned int i;
+ int bufsz;
+
+ i = rx_ring->next_to_use;
+ buffer_info = &rx_ring->buffer_info[i];
+
+ if (adapter->rx_ps_hdr_size)
+ bufsz = adapter->rx_ps_hdr_size;
+ else
+ bufsz = adapter->rx_buffer_len;
+ bufsz += NET_IP_ALIGN;
+
+ while (cleaned_count--) {
+ rx_desc = IGBVF_RX_DESC_ADV(*rx_ring, i);
+
+ if (adapter->rx_ps_hdr_size && !buffer_info->page_dma) {
+ if (!buffer_info->page) {
+ buffer_info->page = alloc_page(GFP_ATOMIC);
+ if (!buffer_info->page) {
+ adapter->alloc_rx_buff_failed++;
+ goto no_buffers;
+ }
+ buffer_info->page_offset = 0;
+ } else {
+ buffer_info->page_offset ^= PAGE_SIZE / 2;
+ }
+ buffer_info->page_dma =
+ pci_map_page(pdev, buffer_info->page,
+ buffer_info->page_offset,
+ PAGE_SIZE / 2,
+ PCI_DMA_FROMDEVICE);
+ }
+
+ if (!buffer_info->skb) {
+ skb = netdev_alloc_skb(netdev, bufsz);
+ if (!skb) {
+ adapter->alloc_rx_buff_failed++;
+ goto no_buffers;
+ }
+
+ /* Make buffer alignment 2 beyond a 16 byte boundary
+ * this will result in a 16 byte aligned IP header after
+ * the 14 byte MAC header is removed
+ */
+ skb_reserve(skb, NET_IP_ALIGN);
+
+ buffer_info->skb = skb;
+ buffer_info->dma = pci_map_single(pdev, skb->data,
+ bufsz,
+ PCI_DMA_FROMDEVICE);
+ }
+ /* Refresh the desc even if buffer_addrs didn't change because
+ * each write-back erases this info. */
+ if (adapter->rx_ps_hdr_size) {
+ rx_desc->read.pkt_addr =
+ cpu_to_le64(buffer_info->page_dma);
+ rx_desc->read.hdr_addr = cpu_to_le64(buffer_info->dma);
+ } else {
+ rx_desc->read.pkt_addr =
+ cpu_to_le64(buffer_info->dma);
+ rx_desc->read.hdr_addr = 0;
+ }
+
+ i++;
+ if (i == rx_ring->count)
+ i = 0;
+ buffer_info = &rx_ring->buffer_info[i];
+ }
+
+no_buffers:
+ if (rx_ring->next_to_use != i) {
+ rx_ring->next_to_use = i;
+ if (i == 0)
+ i = (rx_ring->count - 1);
+ else
+ 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). */
+ wmb();
+ writel(i, adapter->hw.hw_addr + rx_ring->tail);
+ }
+}
+
+/**
+ * igbvf_clean_rx_irq - Send received data up the network stack; legacy
+ * @adapter: board private structure
+ *
+ * the return value indicates whether actual cleaning was done, there
+ * is no guarantee that everything was cleaned
+ **/
+static bool igbvf_clean_rx_irq(struct igbvf_adapter *adapter,
+ int *work_done, int work_to_do)
+{
+ struct igbvf_ring *rx_ring = adapter->rx_ring;
+ struct net_device *netdev = adapter->netdev;
+ struct pci_dev *pdev = adapter->pdev;
+ union e1000_adv_rx_desc *rx_desc, *next_rxd;
+ struct igbvf_buffer *buffer_info, *next_buffer;
+ struct sk_buff *skb;
+ bool cleaned = false;
+ int cleaned_count = 0;
+ unsigned int total_bytes = 0, total_packets = 0;
+ unsigned int i;
+ u32 length, hlen, staterr;
+
+ i = rx_ring->next_to_clean;
+ rx_desc = IGBVF_RX_DESC_ADV(*rx_ring, i);
+ staterr = le32_to_cpu(rx_desc->wb.upper.status_error);
+
+ while (staterr & E1000_RXD_STAT_DD) {
+ if (*work_done >= work_to_do)
+ break;
+ (*work_done)++;
+
+ buffer_info = &rx_ring->buffer_info[i];
+
+ /* 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.
+ */
+ hlen = (le16_to_cpu(rx_desc->wb.lower.lo_dword.hs_rss.hdr_info) &
+ E1000_RXDADV_HDRBUFLEN_MASK) >> E1000_RXDADV_HDRBUFLEN_SHIFT;
+ if (hlen > adapter->rx_ps_hdr_size)
+ hlen = adapter->rx_ps_hdr_size;
+
+ length = le16_to_cpu(rx_desc->wb.upper.length);
+ cleaned = true;
+ cleaned_count++;
+
+ skb = buffer_info->skb;
+ prefetch(skb->data - NET_IP_ALIGN);
+ buffer_info->skb = NULL;
+ if (!adapter->rx_ps_hdr_size) {
+ pci_unmap_single(pdev, buffer_info->dma,
+ adapter->rx_buffer_len,
+ PCI_DMA_FROMDEVICE);
+ buffer_info->dma = 0;
+ skb_put(skb, length);
+ goto send_up;
+ }
+
+ if (!skb_shinfo(skb)->nr_frags) {
+ pci_unmap_single(pdev, buffer_info->dma,
+ adapter->rx_ps_hdr_size + NET_IP_ALIGN,
+ PCI_DMA_FROMDEVICE);
+ skb_put(skb, hlen);
+ }
+
+ if (length) {
+ pci_unmap_page(pdev, buffer_info->page_dma,
+ PAGE_SIZE / 2,
+ PCI_DMA_FROMDEVICE);
+ buffer_info->page_dma = 0;
+
+ skb_fill_page_desc(skb, skb_shinfo(skb)->nr_frags++,
+ buffer_info->page,
+ buffer_info->page_offset,
+ length);
+
+ if ((adapter->rx_buffer_len > (PAGE_SIZE / 2)) ||
+ (page_count(buffer_info->page) != 1))
+ buffer_info->page = NULL;
+ else
+ get_page(buffer_info->page);
+
+ skb->len += length;
+ skb->data_len += length;
+ skb->truesize += length;
+ }
+send_up:
+ i++;
+ if (i == rx_ring->count)
+ i = 0;
+ next_rxd = IGBVF_RX_DESC_ADV(*rx_ring, i);
+ prefetch(next_rxd);
+ next_buffer = &rx_ring->buffer_info[i];
+
+ if (!(staterr & E1000_RXD_STAT_EOP)) {
+ buffer_info->skb = next_buffer->skb;
+ buffer_info->dma = next_buffer->dma;
+ next_buffer->skb = skb;
+ next_buffer->dma = 0;
+ goto next_desc;
+ }
+
+ if (staterr & E1000_RXDEXT_ERR_FRAME_ERR_MASK) {
+ dev_kfree_skb_irq(skb);
+ goto next_desc;
+ }
+
+ total_bytes += skb->len;
+ total_packets++;
+
+ igbvf_rx_checksum_adv(adapter, staterr, skb);
+
+ skb->protocol = eth_type_trans(skb, netdev);
+
+ igbvf_receive_skb(adapter, netdev, skb, staterr,
+ rx_desc->wb.upper.vlan);
+
+ netdev->last_rx = jiffies;
+
+next_desc:
+ rx_desc->wb.upper.status_error = 0;
+
+ /* return some buffers to hardware, one at a time is too slow */
+ if (cleaned_count >= IGBVF_RX_BUFFER_WRITE) {
+ igbvf_alloc_rx_buffers(rx_ring, cleaned_count);
+ cleaned_count = 0;
+ }
+
+ /* use prefetched values */
+ rx_desc = next_rxd;
+ buffer_info = next_buffer;
+
+ staterr = le32_to_cpu(rx_desc->wb.upper.status_error);
+ }
+
+ rx_ring->next_to_clean = i;
+ cleaned_count = igbvf_desc_unused(rx_ring);
+
+ if (cleaned_count)
+ igbvf_alloc_rx_buffers(rx_ring, cleaned_count);
+
+ adapter->total_rx_packets += total_packets;
+ adapter->total_rx_bytes += total_bytes;
+ adapter->net_stats.rx_bytes += total_bytes;
+ adapter->net_stats.rx_packets += total_packets;
+ return cleaned;
+}
+
+static void igbvf_put_txbuf(struct igbvf_adapter *adapter,
+ struct igbvf_buffer *buffer_info)
+{
+ if (buffer_info->dma) {
+ pci_unmap_page(adapter->pdev, buffer_info->dma,
+ buffer_info->length, PCI_DMA_TODEVICE);
+ buffer_info->dma = 0;
+ }
+ if (buffer_info->skb) {
+ dev_kfree_skb_any(buffer_info->skb);
+ buffer_info->skb = NULL;
+ }
+ buffer_info->time_stamp = 0;
+}
+
+static void igbvf_print_tx_hang(struct igbvf_adapter *adapter)
+{
+ struct igbvf_ring *tx_ring = adapter->tx_ring;
+ unsigned int i = tx_ring->next_to_clean;
+ unsigned int eop = tx_ring->buffer_info[i].next_to_watch;
+ union e1000_adv_tx_desc *eop_desc = IGBVF_TX_DESC_ADV(*tx_ring, eop);
+
+ /* detected Tx unit hang */
+ dev_err(&adapter->pdev->dev,
+ "Detected Tx Unit Hang:\n"
+ " TDH <%x>\n"
+ " TDT <%x>\n"
+ " next_to_use <%x>\n"
+ " next_to_clean <%x>\n"
+ "buffer_info[next_to_clean]:\n"
+ " time_stamp <%lx>\n"
+ " next_to_watch <%x>\n"
+ " jiffies <%lx>\n"
+ " next_to_watch.status <%x>\n",
+ readl(adapter->hw.hw_addr + tx_ring->head),
+ readl(adapter->hw.hw_addr + tx_ring->tail),
+ tx_ring->next_to_use,
+ tx_ring->next_to_clean,
+ tx_ring->buffer_info[eop].time_stamp,
+ eop,
+ jiffies,
+ eop_desc->wb.status);
+}
+
+/**
+ * igbvf_alloc_ring_dma - allocate memory for a ring structure
+ **/
+static int igbvf_alloc_ring_dma(struct igbvf_adapter *adapter,
+ struct igbvf_ring *ring)
+{
+ struct pci_dev *pdev = adapter->pdev;
+
+ ring->desc = dma_alloc_coherent(&pdev->dev, (ring->size + sizeof(u32)), &ring->dma,
+ GFP_KERNEL);
+ if (!ring->desc)
+ return -ENOMEM;
+
+ return 0;
+}
+
+/**
+ * igbvf_setup_tx_resources - allocate Tx resources (Descriptors)
+ * @adapter: board private structure
+ *
+ * Return 0 on success, negative on failure
+ **/
+int igbvf_setup_tx_resources(struct igbvf_adapter *adapter)
+{
+ struct igbvf_ring *tx_ring = adapter->tx_ring;
+ int err = -ENOMEM, size;
+
+ size = sizeof(struct igbvf_buffer) * tx_ring->count;
+ tx_ring->buffer_info = vmalloc(size);
+ if (!tx_ring->buffer_info)
+ goto err;
+ memset(tx_ring->buffer_info, 0, size);
+
+ /* round up to nearest 4K */
+ tx_ring->size = tx_ring->count * sizeof(union e1000_adv_tx_desc);
+ tx_ring->size = ALIGN(tx_ring->size, 4096);
+
+ err = igbvf_alloc_ring_dma(adapter, tx_ring);
+ if (err)
+ goto err;
+
+ tx_ring->next_to_use = 0;
+ tx_ring->next_to_clean = 0;
+ spin_lock_init(&adapter->tx_queue_lock);
+
+ return 0;
+err:
+ vfree(tx_ring->buffer_info);
+ dev_err(&adapter->pdev->dev,
+ "Unable to allocate memory for the transmit descriptor ring\n");
+ return err;
+}
+
+/**
+ * igbvf_setup_rx_resources - allocate Rx resources (Descriptors)
+ * @adapter: board private structure
+ *
+ * Returns 0 on success, negative on failure
+ **/
+int igbvf_setup_rx_resources(struct igbvf_adapter *adapter)
+{
+ struct igbvf_ring *rx_ring = adapter->rx_ring;
+ struct pci_dev *pdev = adapter->pdev;
+ int size, desc_len;
+
+ size = sizeof(struct igbvf_buffer) * rx_ring->count;
+ rx_ring->buffer_info = vmalloc(size);
+ if (!rx_ring->buffer_info)
+ goto err;
+ memset(rx_ring->buffer_info, 0, size);
+
+ desc_len = sizeof(union e1000_adv_rx_desc);
+
+ /* Round up to nearest 4K */
+ rx_ring->size = rx_ring->count * desc_len;
+ rx_ring->size = ALIGN(rx_ring->size, 4096);
+
+ rx_ring->desc = pci_alloc_consistent(pdev, rx_ring->size,
+ &rx_ring->dma);
+
+ if (!rx_ring->desc)
+ goto err;
+
+ rx_ring->next_to_clean = 0;
+ rx_ring->next_to_use = 0;
+
+ return 0;
+
+err:
+ vfree(rx_ring->buffer_info);
+ rx_ring->buffer_info = NULL;
+ dev_err(&adapter->pdev->dev,
+ "Unable to allocate memory for the receive descriptor ring\n");
+ return -ENOMEM;
+}
+
+/**
+ * igbvf_clean_tx_ring - Free Tx Buffers
+ * @adapter: board private structure
+ **/
+static void igbvf_clean_tx_ring(struct igbvf_adapter *adapter)
+{
+ struct igbvf_ring *tx_ring = adapter->tx_ring;
+ struct igbvf_buffer *buffer_info;
+ unsigned long size;
+ unsigned int i;
+
+ for (i = 0; i < tx_ring->count; i++) {
+ buffer_info = &tx_ring->buffer_info[i];
+ igbvf_put_txbuf(adapter, buffer_info);
+ }
+
+ size = sizeof(struct igbvf_buffer) * tx_ring->count;
+ memset(tx_ring->buffer_info, 0, size);
+
+ memset(tx_ring->desc, 0, tx_ring->size);
+
+ tx_ring->next_to_use = 0;
+ tx_ring->next_to_clean = 0;
+
+ writel(0, adapter->hw.hw_addr + tx_ring->head);
+ writel(0, adapter->hw.hw_addr + tx_ring->tail);
+}
+
+/**
+ * igbvf_free_tx_resources - Free Tx Resources per Queue
+ * @adapter: board private structure
+ *
+ * Free all transmit software resources
+ **/
+void igbvf_free_tx_resources(struct igbvf_adapter *adapter)
+{
+ struct pci_dev *pdev = adapter->pdev;
+ struct igbvf_ring *tx_ring = adapter->tx_ring;
+
+ igbvf_clean_tx_ring(adapter);
+
+ vfree(tx_ring->buffer_info);
+ tx_ring->buffer_info = NULL;
+
+ dma_free_coherent(&pdev->dev, (tx_ring->size + sizeof(u32)), tx_ring->desc,
+ tx_ring->dma);
+ tx_ring->desc = NULL;
+}
+
+/**
+ * igbvf_clean_rx_ring - Free Rx Buffers per Queue
+ * @adapter: board private structure
+ **/
+static void igbvf_clean_rx_ring(struct igbvf_adapter *adapter)
+{
+ struct igbvf_ring *rx_ring = adapter->rx_ring;
+ struct igbvf_buffer *buffer_info;
+ struct pci_dev *pdev = adapter->pdev;
+ unsigned int i;
+
+ /* Free all the Rx ring sk_buffs */
+ for (i = 0; i < rx_ring->count; i++) {
+ buffer_info = &rx_ring->buffer_info[i];
+ if (buffer_info->dma) {
+ if (adapter->rx_ps_hdr_size){
+ pci_unmap_single(pdev, buffer_info->dma,
+ adapter->rx_ps_hdr_size,
+ PCI_DMA_FROMDEVICE);
+ } else {
+ pci_unmap_single(pdev, buffer_info->dma,
+ adapter->rx_buffer_len,
+ PCI_DMA_FROMDEVICE);
+ }
+ buffer_info->dma = 0;
+ }
+
+ if (buffer_info->skb) {
+ dev_kfree_skb(buffer_info->skb);
+ buffer_info->skb = NULL;
+ }
+
+ if (buffer_info->page) {
+ if (buffer_info->page_dma)
+ pci_unmap_page(pdev, buffer_info->page_dma,
+ PAGE_SIZE / 2,
+ PCI_DMA_FROMDEVICE);
+ put_page(buffer_info->page);
+ buffer_info->page = NULL;
+ buffer_info->page_dma = 0;
+ buffer_info->page_offset = 0;
+ }
+ }
+
+ /* Zero out the descriptor ring */
+ memset(rx_ring->desc, 0, rx_ring->size);
+
+ rx_ring->next_to_clean = 0;
+ rx_ring->next_to_use = 0;
+
+ writel(0, adapter->hw.hw_addr + rx_ring->head);
+ writel(0, adapter->hw.hw_addr + rx_ring->tail);
+}
+
+/**
+ * igbvf_free_rx_resources - Free Rx Resources
+ * @adapter: board private structure
+ *
+ * Free all receive software resources
+ **/
+
+void igbvf_free_rx_resources(struct igbvf_adapter *adapter)
+{
+ struct pci_dev *pdev = adapter->pdev;
+ struct igbvf_ring *rx_ring = adapter->rx_ring;
+
+ igbvf_clean_rx_ring(adapter);
+
+ vfree(rx_ring->buffer_info);
+ rx_ring->buffer_info = NULL;
+
+ dma_free_coherent(&pdev->dev, rx_ring->size, rx_ring->desc,
+ rx_ring->dma);
+ rx_ring->desc = NULL;
+}
+
+/**
+ * igbvf_update_itr - update the dynamic ITR value based on statistics
+ * @adapter: pointer to adapter
+ * @itr_setting: current adapter->itr
+ * @packets: the number of packets during this measurement interval
+ * @bytes: the number of bytes during this measurement interval
+ *
+ * Stores a new ITR value based on packets and byte
+ * counts during the last interrupt. The advantage of per interrupt
+ * computation is faster updates and more accurate ITR for the current
+ * traffic pattern. Constants in this function were computed
+ * based on theoretical maximum wire speed and thresholds were set based
+ * on testing data as well as attempting to minimize response time
+ * while increasing bulk throughput. This functionality is controlled
+ * by the InterruptThrottleRate module parameter.
+ **/
+static unsigned int igbvf_update_itr(struct igbvf_adapter *adapter,
+ u16 itr_setting, int packets,
+ int bytes)
+{
+ unsigned int retval = itr_setting;
+
+ if (packets == 0)
+ goto update_itr_done;
+
+ switch (itr_setting) {
+ case lowest_latency:
+ /* handle TSO and jumbo frames */
+ if (bytes/packets > 8000)
+ retval = bulk_latency;
+ else if ((packets < 5) && (bytes > 512))
+ retval = low_latency;
+ break;
+ case low_latency: /* 50 usec aka 20000 ints/s */
+ if (bytes > 10000) {
+ /* this if handles the TSO accounting */
+ if (bytes/packets > 8000)
+ retval = bulk_latency;
+ else if ((packets < 10) || ((bytes/packets) > 1200))
+ retval = bulk_latency;
+ else if ((packets > 35))
+ retval = lowest_latency;
+ } else if (bytes/packets > 2000) {
+ retval = bulk_latency;
+ } else if (packets <= 2 && bytes < 512) {
+ retval = lowest_latency;
+ }
+ break;
+ case bulk_latency: /* 250 usec aka 4000 ints/s */
+ if (bytes > 25000) {
+ if (packets > 35)
+ retval = low_latency;
+ } else if (bytes < 6000) {
+ retval = low_latency;
+ }
+ break;
+ }
+
+update_itr_done:
+ return retval;
+}
+
+static void igbvf_set_itr(struct igbvf_adapter *adapter)
+{
+ struct e1000_hw *hw = &adapter->hw;
+ u16 current_itr;
+ u32 new_itr = adapter->itr;
+
+ adapter->tx_itr = igbvf_update_itr(adapter, adapter->tx_itr,
+ adapter->total_tx_packets,
+ adapter->total_tx_bytes);
+ /* conservative mode (itr 3) eliminates the lowest_latency setting */
+ if (adapter->itr_setting == 3 && adapter->tx_itr == lowest_latency)
+ adapter->tx_itr = low_latency;
+
+ adapter->rx_itr = igbvf_update_itr(adapter, adapter->rx_itr,
+ adapter->total_rx_packets,
+ adapter->total_rx_bytes);
+ /* conservative mode (itr 3) eliminates the lowest_latency setting */
+ if (adapter->itr_setting == 3 && adapter->rx_itr == lowest_latency)
+ adapter->rx_itr = low_latency;
+
+ current_itr = max(adapter->rx_itr, adapter->tx_itr);
+
+ switch (current_itr) {
+ /* counts and packets in update_itr are dependent on these numbers */
+ case lowest_latency:
+ new_itr = 70000;
+ break;
+ case low_latency:
+ new_itr = 20000; /* aka hwitr = ~200 */
+ break;
+ case bulk_latency:
+ new_itr = 4000;
+ break;
+ default:
+ break;
+ }
+
+ if (new_itr != adapter->itr) {
+ /*
+ * this attempts to bias the interrupt rate towards Bulk
+ * by adding intermediate steps when interrupt rate is
+ * increasing
+ */
+ new_itr = new_itr > adapter->itr ?
+ min(adapter->itr + (new_itr >> 2), new_itr) :
+ new_itr;
+ adapter->itr = new_itr;
+ adapter->rx_ring->itr_val = 1952;
+
+ if (adapter->msix_entries)
+ adapter->rx_ring->set_itr = 1;
+ else
+ ew32(ITR, 1952);
+ }
+}
+
+/**
+ * igbvf_clean_tx_irq - Reclaim resources after transmit completes
+ * @adapter: board private structure
+ * returns true if ring is completely cleaned
+ **/
+static bool igbvf_clean_tx_irq(struct igbvf_ring *tx_ring)
+{
+ struct igbvf_adapter *adapter = tx_ring->adapter;
+ struct e1000_hw *hw = &adapter->hw;
+ struct net_device *netdev = adapter->netdev;
+ struct igbvf_buffer *buffer_info;
+ struct sk_buff *skb;
+ union e1000_adv_tx_desc *tx_desc, *eop_desc;
+ unsigned int total_bytes = 0, total_packets = 0;
+ unsigned int i, eop, count = 0;
+ bool cleaned = false;
+
+ i = tx_ring->next_to_clean;
+ eop = tx_ring->buffer_info[i].next_to_watch;
+ eop_desc = IGBVF_TX_DESC_ADV(*tx_ring, eop);
+
+ while ((eop_desc->wb.status & cpu_to_le32(E1000_TXD_STAT_DD)) &&
+ (count < tx_ring->count)) {
+ for (cleaned = false; !cleaned; count++) {
+ tx_desc = IGBVF_TX_DESC_ADV(*tx_ring, i);
+ buffer_info = &tx_ring->buffer_info[i];
+ cleaned = (i == eop);
+ skb = buffer_info->skb;
+
+ if (skb) {
+ unsigned int segs, bytecount;
+
+ /* gso_segs is currently only valid for tcp */
+ segs = skb_shinfo(skb)->gso_segs ?: 1;
+ /* multiply data chunks by size of headers */
+ bytecount = ((segs - 1) * skb_headlen(skb)) +
+ skb->len;
+ total_packets += segs;
+ total_bytes += bytecount;
+ }
+
+ igbvf_put_txbuf(adapter, buffer_info);
+ tx_desc->wb.status = 0;
+
+ i++;
+ if (i == tx_ring->count)
+ i = 0;
+ }
+ eop = tx_ring->buffer_info[i].next_to_watch;
+ eop_desc = IGBVF_TX_DESC_ADV(*tx_ring, eop);
+ }
+
+ tx_ring->next_to_clean = i;
+
+ if (unlikely(count &&
+ netif_carrier_ok(netdev) &&
+ igbvf_desc_unused(tx_ring) >= IGBVF_TX_QUEUE_WAKE)) {
+ /* Make sure that anybody stopping the queue after this
+ * sees the new next_to_clean.
+ */
+ smp_mb();
+ if (netif_queue_stopped(netdev) &&
+ !(test_bit(__IGBVF_DOWN, &adapter->state))) {
+ netif_wake_queue(netdev);
+ ++adapter->restart_queue;
+ }
+ }
+
+ if (adapter->detect_tx_hung) {
+ /* Detect a transmit hang in hardware, this serializes the
+ * check with the clearing of time_stamp and movement of i */
+ adapter->detect_tx_hung = false;
+ if (tx_ring->buffer_info[i].time_stamp &&
+ time_after(jiffies, tx_ring->buffer_info[i].time_stamp +
+ (adapter->tx_timeout_factor * HZ))
+ && !(er32(STATUS) & E1000_STATUS_TXOFF)) {
+
+ tx_desc = IGBVF_TX_DESC_ADV(*tx_ring, i);
+ /* detected Tx unit hang */
+ igbvf_print_tx_hang(adapter);
+
+ netif_stop_queue(netdev);
+ }
+ }
+ adapter->net_stats.tx_bytes += total_bytes;
+ adapter->net_stats.tx_packets += total_packets;
+ return (count < tx_ring->count);
+}
+
+static irqreturn_t igbvf_msix_other(int irq, void *data)
+{
+ struct net_device *netdev = data;
+ struct igbvf_adapter *adapter = netdev_priv(netdev);
+ struct e1000_hw *hw = &adapter->hw;
+
+ adapter->int_counter1++;
+
+ netif_carrier_off(netdev);
+ hw->mac.get_link_status = 1;
+ if (!test_bit(__IGBVF_DOWN, &adapter->state))
+ mod_timer(&adapter->watchdog_timer, jiffies + 1);
+
+ ew32(EIMS, adapter->eims_other);
+
+ return IRQ_HANDLED;
+}
+
+static irqreturn_t igbvf_intr_msix_tx(int irq, void *data)
+{
+ struct net_device *netdev = data;
+ struct igbvf_adapter *adapter = netdev_priv(netdev);
+ struct e1000_hw *hw = &adapter->hw;
+ struct igbvf_ring *tx_ring = adapter->tx_ring;
+
+
+ adapter->total_tx_bytes = 0;
+ adapter->total_tx_packets = 0;
+
+ /* auto mask will automatically reenable the interrupt when we write
+ * EICS */
+ if (!igbvf_clean_tx_irq(tx_ring))
+ /* Ring was not completely cleaned, so fire another interrupt */
+ ew32(EICS, tx_ring->eims_value);
+ else
+ ew32(EIMS, tx_ring->eims_value);
+
+ return IRQ_HANDLED;
+}
+
+static irqreturn_t igbvf_intr_msix_rx(int irq, void *data)
+{
+ struct net_device *netdev = data;
+ struct igbvf_adapter *adapter = netdev_priv(netdev);
+
+ adapter->int_counter0++;
+
+ /* Write the ITR value calculated at the end of the
+ * previous interrupt.
+ */
+ if (adapter->rx_ring->set_itr) {
+ writel(adapter->rx_ring->itr_val,
+ adapter->hw.hw_addr + adapter->rx_ring->itr_register);
+ adapter->rx_ring->set_itr = 0;
+ }
+
+ if (napi_schedule_prep(&adapter->rx_ring->napi)) {
+ adapter->total_rx_bytes = 0;
+ adapter->total_rx_packets = 0;
+ __napi_schedule(&adapter->rx_ring->napi);
+ }
+
+ return IRQ_HANDLED;
+}
+
+#define IGBVF_NO_QUEUE -1
+
+static void igbvf_assign_vector(struct igbvf_adapter *adapter, int rx_queue,
+ int tx_queue, int msix_vector)
+{
+ struct e1000_hw *hw = &adapter->hw;
+ u32 ivar, index;
+
+ /* 82576 uses a table-based method for assigning vectors.
+ Each queue has a single entry in the table to which we write
+ a vector number along with a "valid" bit. Sadly, the layout
+ of the table is somewhat counterintuitive. */
+ if (rx_queue > IGBVF_NO_QUEUE) {
+ index = (rx_queue >> 1);
+ ivar = array_er32(IVAR0, index);
+ if (rx_queue & 0x1) {
+ /* vector goes into third byte of register */
+ ivar = ivar & 0xFF00FFFF;
+ ivar |= (msix_vector | E1000_IVAR_VALID) << 16;
+ } else {
+ /* vector goes into low byte of register */
+ ivar = ivar & 0xFFFFFF00;
+ ivar |= msix_vector | E1000_IVAR_VALID;
+ }
+ adapter->rx_ring[rx_queue].eims_value = 1 << msix_vector;
+ array_ew32(IVAR0, index, ivar);
+ }
+ if (tx_queue > IGBVF_NO_QUEUE) {
+ index = (tx_queue >> 1);
+ ivar = array_er32(IVAR0, index);
+ if (tx_queue & 0x1) {
+ /* vector goes into high byte of register */
+ ivar = ivar & 0x00FFFFFF;
+ ivar |= (msix_vector | E1000_IVAR_VALID) << 24;
+ } else {
+ /* vector goes into second byte of register */
+ ivar = ivar & 0xFFFF00FF;
+ ivar |= (msix_vector | E1000_IVAR_VALID) << 8;
+ }
+ adapter->tx_ring[tx_queue].eims_value = 1 << msix_vector;
+ array_ew32(IVAR0, index, ivar);
+ }
+}
+
+/**
+ * igbvf_configure_msix - Configure MSI-X hardware
+ *
+ * igbvf_configure_msix sets up the hardware to properly
+ * generate MSI-X interrupts.
+ **/
+static void igbvf_configure_msix(struct igbvf_adapter *adapter)
+{
+ u32 tmp;
+ struct e1000_hw *hw = &adapter->hw;
+ struct igbvf_ring *tx_ring = adapter->tx_ring;
+ struct igbvf_ring *rx_ring = adapter->rx_ring;
+ int vector = 0;
+
+ adapter->eims_enable_mask = 0;
+
+ igbvf_assign_vector(adapter, IGBVF_NO_QUEUE, 0, vector++);
+ adapter->eims_enable_mask |= tx_ring->eims_value;
+ if (tx_ring->itr_val)
+ writel(tx_ring->itr_val,
+ hw->hw_addr + tx_ring->itr_register);
+ else
+ writel(1952, hw->hw_addr + tx_ring->itr_register);
+
+ igbvf_assign_vector(adapter, 0, IGBVF_NO_QUEUE, vector++);
+ adapter->eims_enable_mask |= rx_ring->eims_value;
+ if (rx_ring->itr_val)
+ writel(rx_ring->itr_val,
+ hw->hw_addr + rx_ring->itr_register);
+ else
+ writel(1952, hw->hw_addr + rx_ring->itr_register);
+
+ /* set vector for other causes, i.e. link changes */
+
+ tmp = (vector++ | E1000_IVAR_VALID);
+
+ ew32(IVAR_MISC, tmp);
+
+ adapter->eims_enable_mask = (1 << (vector)) - 1;
+ adapter->eims_other = 1 << (vector - 1);
+ e1e_flush();
+}
+
+void igbvf_reset_interrupt_capability(struct igbvf_adapter *adapter)
+{
+ if (adapter->msix_entries) {
+ pci_disable_msix(adapter->pdev);
+ kfree(adapter->msix_entries);
+ adapter->msix_entries = NULL;
+ }
+}
+
+/**
+ * igbvf_set_interrupt_capability - set MSI or MSI-X if supported
+ *
+ * Attempt to configure interrupts using the best available
+ * capabilities of the hardware and kernel.
+ **/
+void igbvf_set_interrupt_capability(struct igbvf_adapter *adapter)
+{
+ int err = -ENOMEM;
+ int i;
+
+ /* we allocate 3 vectors, 1 for tx, 1 for rx, one for pf messages */
+ adapter->msix_entries = kcalloc(3, sizeof(struct msix_entry),
+ GFP_KERNEL);
+ if (adapter->msix_entries) {
+ for (i = 0; i < 3; i++)
+ adapter->msix_entries[i].entry = i;
+
+ err = pci_enable_msix(adapter->pdev,
+ adapter->msix_entries, 3);
+ }
+
+ if (err) {
+ /* MSI-X failed */
+ dev_err(&adapter->pdev->dev,
+ "Failed to initialize MSI-X interrupts.\n");
+ igbvf_reset_interrupt_capability(adapter);
+ }
+}
+
+/**
+ * igbvf_request_msix - Initialize MSI-X interrupts
+ *
+ * igbvf_request_msix allocates MSI-X vectors and requests interrupts from the
+ * kernel.
+ **/
+static int igbvf_request_msix(struct igbvf_adapter *adapter)
+{
+ struct net_device *netdev = adapter->netdev;
+ int err = 0, vector = 0;
+
+ if (strlen(netdev->name) < (IFNAMSIZ - 5)) {
+ sprintf(adapter->tx_ring->name, "%s-tx-0", netdev->name);
+ sprintf(adapter->rx_ring->name, "%s-rx-0", netdev->name);
+ } else {
+ memcpy(adapter->tx_ring->name, netdev->name, IFNAMSIZ);
+ memcpy(adapter->rx_ring->name, netdev->name, IFNAMSIZ);
+ }
+
+ err = request_irq(adapter->msix_entries[vector].vector,
+ &igbvf_intr_msix_tx, 0, adapter->tx_ring->name,
+ netdev);
+ if (err)
+ goto out;
+
+ adapter->tx_ring->itr_register = E1000_EITR(vector);
+ adapter->tx_ring->itr_val = 1952;
+ vector++;
+
+ err = request_irq(adapter->msix_entries[vector].vector,
+ &igbvf_intr_msix_rx, 0, adapter->rx_ring->name,
+ netdev);
+ if (err)
+ goto out;
+
+ adapter->rx_ring->itr_register = E1000_EITR(vector);
+ adapter->rx_ring->itr_val = 1952;
+ vector++;
+
+ err = request_irq(adapter->msix_entries[vector].vector,
+ &igbvf_msix_other, 0, netdev->name, netdev);
+ if (err)
+ goto out;
+
+ igbvf_configure_msix(adapter);
+ return 0;
+out:
+ return err;
+}
+
+/**
+ * igbvf_alloc_queues - Allocate memory for all rings
+ * @adapter: board private structure to initialize
+ **/
+static int __devinit igbvf_alloc_queues(struct igbvf_adapter *adapter)
+{
+ struct net_device *netdev = adapter->netdev;
+
+ adapter->tx_ring = kzalloc(sizeof(struct igbvf_ring), GFP_KERNEL);
+ if (!adapter->tx_ring)
+ goto tx_err;
+ adapter->tx_ring->adapter = adapter;
+
+ adapter->rx_ring = kzalloc(sizeof(struct igbvf_ring), GFP_KERNEL);
+ if (!adapter->rx_ring)
+ goto err;
+ adapter->rx_ring->adapter = adapter;
+
+ netif_napi_add(netdev, &adapter->rx_ring->napi, igbvf_poll, 64);
+
+ return 0;
+err:
+ dev_err(&adapter->pdev->dev, "Unable to allocate memory for queues\n");
+ kfree(adapter->rx_ring);
+tx_err:
+ kfree(adapter->tx_ring);
+ return -ENOMEM;
+}
+
+/**
+ * igbvf_request_irq - initialize interrupts
+ *
+ * Attempts to configure interrupts using the best available
+ * capabilities of the hardware and kernel.
+ **/
+static int igbvf_request_irq(struct igbvf_adapter *adapter)
+{
+ int err = -1;
+
+ /* igbvf supports msi-x only */
+ if (adapter->msix_entries)
+ err = igbvf_request_msix(adapter);
+
+ if (!err)
+ return err;
+
+ dev_err(&adapter->pdev->dev,
+ "Unable to allocate interrupt, Error: %d\n", err);
+
+ return err;
+}
+
+static void igbvf_free_irq(struct igbvf_adapter *adapter)
+{
+ struct net_device *netdev = adapter->netdev;
+ int vector;
+
+ if (adapter->msix_entries) {
+ for (vector = 0; vector < 3; vector++)
+ free_irq(adapter->msix_entries[vector].vector, netdev);
+ }
+}
+
+/**
+ * igbvf_irq_disable - Mask off interrupt generation on the NIC
+ **/
+static void igbvf_irq_disable(struct igbvf_adapter *adapter)
+{
+ struct e1000_hw *hw = &adapter->hw;
+
+ ew32(EIMC, ~0);
+
+ if (adapter->msix_entries)
+ ew32(EIAC, 0);
+}
+
+/**
+ * igbvf_irq_enable - Enable default interrupt generation settings
+ **/
+static void igbvf_irq_enable(struct igbvf_adapter *adapter)
+{
+ struct e1000_hw *hw = &adapter->hw;
+
+ ew32(EIAC, adapter->eims_enable_mask);
+ ew32(EIAM, adapter->eims_enable_mask);
+ ew32(EIMS, adapter->eims_enable_mask);
+}
+
+/**
+ * igbvf_poll - NAPI Rx polling callback
+ * @napi: struct associated with this polling callback
+ * @budget: amount of packets driver is allowed to process this poll
+ **/
+static int igbvf_poll(struct napi_struct *napi, int budget)
+{
+ struct igbvf_ring *rx_ring = container_of(napi, struct igbvf_ring, napi);
+ struct igbvf_adapter *adapter = rx_ring->adapter;
+ struct e1000_hw *hw = &adapter->hw;
+ int work_done = 0;
+
+ igbvf_clean_rx_irq(adapter, &work_done, budget);
+
+ /* If not enough Rx work done, exit the polling mode */
+ if (work_done < budget) {
+ napi_complete(napi);
+
+ if (adapter->itr_setting & 3)
+ igbvf_set_itr(adapter);
+
+ if (!test_bit(__IGBVF_DOWN, &adapter->state))
+ ew32(EIMS, adapter->rx_ring->eims_value);
+ }
+
+ return work_done;
+}
+
+/**
+ * igbvf_set_rlpml - set receive large packet maximum length
+ * @adapter: board private structure
+ *
+ * Configure the maximum size of packets that will be received
+ */
+static void igbvf_set_rlpml(struct igbvf_adapter *adapter)
+{
+ int max_frame_size = adapter->max_frame_size;
+ struct e1000_hw *hw = &adapter->hw;
+
+ if (adapter->vlgrp)
+ max_frame_size += VLAN_TAG_SIZE;
+
+ e1000_rlpml_set_vf(hw, max_frame_size);
+}
+
+static void igbvf_vlan_rx_add_vid(struct net_device *netdev, u16 vid)
+{
+ struct igbvf_adapter *adapter = netdev_priv(netdev);
+ struct e1000_hw *hw = &adapter->hw;
+
+ if (hw->mac.ops.set_vfta(hw, vid, true))
+ dev_err(&adapter->pdev->dev, "Failed to add vlan id %d\n", vid);
+}
+
+static void igbvf_vlan_rx_kill_vid(struct net_device *netdev, u16 vid)
+{
+ struct igbvf_adapter *adapter = netdev_priv(netdev);
+ struct e1000_hw *hw = &adapter->hw;
+
+ igbvf_irq_disable(adapter);
+ vlan_group_set_device(adapter->vlgrp, vid, NULL);
+
+ if (!test_bit(__IGBVF_DOWN, &adapter->state))
+ igbvf_irq_enable(adapter);
+
+ if (hw->mac.ops.set_vfta(hw, vid, false))
+ dev_err(&adapter->pdev->dev,
+ "Failed to remove vlan id %d\n", vid);
+}
+
+static void igbvf_vlan_rx_register(struct net_device *netdev,
+ struct vlan_group *grp)
+{
+ struct igbvf_adapter *adapter = netdev_priv(netdev);
+
+ adapter->vlgrp = grp;
+}
+
+static void igbvf_restore_vlan(struct igbvf_adapter *adapter)
+{
+ u16 vid;
+
+ if (!adapter->vlgrp)
+ return;
+
+ for (vid = 0; vid < VLAN_GROUP_ARRAY_LEN; vid++) {
+ if (!vlan_group_get_device(adapter->vlgrp, vid))
+ continue;
+ igbvf_vlan_rx_add_vid(adapter->netdev, vid);
+ }
+
+ igbvf_set_rlpml(adapter);
+}
+
+/**
+ * igbvf_configure_tx - Configure Transmit Unit after Reset
+ * @adapter: board private structure
+ *
+ * Configure the Tx unit of the MAC after a reset.
+ **/
+static void igbvf_configure_tx(struct igbvf_adapter *adapter)
+{
+ struct e1000_hw *hw = &adapter->hw;
+ struct igbvf_ring *tx_ring = adapter->tx_ring;
+ u64 tdba;
+ u32 txdctl, dca_txctrl;
+
+ /* disable transmits */
+ txdctl = er32(TXDCTL(0));
+ ew32(TXDCTL(0), txdctl & ~E1000_TXDCTL_QUEUE_ENABLE);
+ msleep(10);
+
+ /* Setup the HW Tx Head and Tail descriptor pointers */
+ ew32(TDLEN(0), tx_ring->count * sizeof(union e1000_adv_tx_desc));
+ tdba = tx_ring->dma;
+ ew32(TDBAL(0), (tdba & DMA_32BIT_MASK));
+ ew32(TDBAH(0), (tdba >> 32));
+ ew32(TDH(0), 0);
+ ew32(TDT(0), 0);
+ tx_ring->head = E1000_TDH(0);
+ tx_ring->tail = E1000_TDT(0);
+
+ /* Turn off Relaxed Ordering on head write-backs. The writebacks
+ * MUST be delivered in order or it will completely screw up
+ * our bookeeping.
+ */
+ dca_txctrl = er32(DCA_TXCTRL(0));
+ dca_txctrl &= ~E1000_DCA_TXCTRL_TX_WB_RO_EN;
+ ew32(DCA_TXCTRL(0), dca_txctrl);
+
+ /* enable transmits */
+ txdctl |= E1000_TXDCTL_QUEUE_ENABLE;
+ ew32(TXDCTL(0), txdctl);
+
+ /* Setup Transmit Descriptor Settings for eop descriptor */
+ adapter->txd_cmd = E1000_ADVTXD_DCMD_EOP | E1000_ADVTXD_DCMD_IFCS;
+
+ /* enable Report Status bit */
+ adapter->txd_cmd |= E1000_ADVTXD_DCMD_RS;
+
+ adapter->tx_queue_len = adapter->netdev->tx_queue_len;
+}
+
+/**
+ * igbvf_setup_srrctl - configure the receive control registers
+ * @adapter: Board private structure
+ **/
+static void igbvf_setup_srrctl(struct igbvf_adapter *adapter)
+{
+ struct e1000_hw *hw = &adapter->hw;
+ u32 srrctl = 0;
+
+ srrctl &= ~(E1000_SRRCTL_DESCTYPE_MASK |
+ E1000_SRRCTL_BSIZEHDR_MASK |
+ E1000_SRRCTL_BSIZEPKT_MASK);
+
+ /* Enable queue drop to avoid head of line blocking */
+ srrctl |= E1000_SRRCTL_DROP_EN;
+
+ /* Setup buffer sizes */
+ srrctl |= ALIGN(adapter->rx_buffer_len, 1024) >>
+ E1000_SRRCTL_BSIZEPKT_SHIFT;
+
+ if (adapter->rx_buffer_len < 2048) {
+ adapter->rx_ps_hdr_size = 0;
+ srrctl |= E1000_SRRCTL_DESCTYPE_ADV_ONEBUF;
+ } else {
+ adapter->rx_ps_hdr_size = 128;
+ srrctl |= adapter->rx_ps_hdr_size <<
+ E1000_SRRCTL_BSIZEHDRSIZE_SHIFT;
+ srrctl |= E1000_SRRCTL_DESCTYPE_HDR_SPLIT_ALWAYS;
+ }
+
+ ew32(SRRCTL(0), srrctl);
+}
+
+/**
+ * igbvf_configure_rx - Configure Receive Unit after Reset
+ * @adapter: board private structure
+ *
+ * Configure the Rx unit of the MAC after a reset.
+ **/
+static void igbvf_configure_rx(struct igbvf_adapter *adapter)
+{
+ struct e1000_hw *hw = &adapter->hw;
+ struct igbvf_ring *rx_ring = adapter->rx_ring;
+ u64 rdba;
+ u32 rdlen, rxdctl;
+
+ /* disable receives */
+ rxdctl = er32(RXDCTL(0));
+ ew32(RXDCTL(0), rxdctl & ~E1000_RXDCTL_QUEUE_ENABLE);
+ msleep(10);
+
+ rdlen = rx_ring->count * sizeof(union e1000_adv_rx_desc);
+
+ /*
+ * Setup the HW Rx Head and Tail Descriptor Pointers and
+ * the Base and Length of the Rx Descriptor Ring
+ */
+ rdba = rx_ring->dma;
+ ew32(RDBAL(0), (rdba & DMA_32BIT_MASK));
+ ew32(RDBAH(0), (rdba >> 32));
+ ew32(RDLEN(0), rx_ring->count * sizeof(union e1000_adv_rx_desc));
+ rx_ring->head = E1000_RDH(0);
+ rx_ring->tail = E1000_RDT(0);
+ ew32(RDH(0), 0);
+ ew32(RDT(0), 0);
+
+ rxdctl |= E1000_RXDCTL_QUEUE_ENABLE;
+ rxdctl &= 0xFFF00000;
+ rxdctl |= IGBVF_RX_PTHRESH;
+ rxdctl |= IGBVF_RX_HTHRESH << 8;
+ rxdctl |= IGBVF_RX_WTHRESH << 16;
+
+ igbvf_set_rlpml(adapter);
+
+ /* enable receives */
+ ew32(RXDCTL(0), rxdctl);
+}
+
+/**
+ * igbvf_set_multi - Multicast and Promiscuous mode set
+ * @netdev: network interface device structure
+ *
+ * The set_multi entry point is called whenever the multicast address
+ * list or the network interface flags are updated. This routine is
+ * responsible for configuring the hardware for proper multicast,
+ * promiscuous mode, and all-multi behavior.
+ **/
+static void igbvf_set_multi(struct net_device *netdev)
+{
+ struct igbvf_adapter *adapter = netdev_priv(netdev);
+ struct e1000_hw *hw = &adapter->hw;
+ struct dev_mc_list *mc_ptr;
+ u8 *mta_list;
+ int i;
+
+ if (netdev->mc_count) {
+ mta_list = kmalloc(netdev->mc_count * 6, GFP_ATOMIC);
+ if (!mta_list)
+ return;
+
+ /* prepare a packed array of only addresses. */
+ mc_ptr = netdev->mc_list;
+
+ for (i = 0; i < netdev->mc_count; i++) {
+ if (!mc_ptr)
+ break;
+ memcpy(mta_list + (i*ETH_ALEN), mc_ptr->dmi_addr,
+ ETH_ALEN);
+ mc_ptr = mc_ptr->next;
+ }
+
+ hw->mac.ops.update_mc_addr_list(hw, mta_list, i, 0, 0);
+ kfree(mta_list);
+ } else {
+ /*
+ * if we're called from probe, we might not have
+ * anything to do here, so clear out the list
+ */
+ hw->mac.ops.update_mc_addr_list(hw, NULL, 0, 0, 0);
+ }
+}
+
+/**
+ * igbvf_configure - configure the hardware for Rx and Tx
+ * @adapter: private board structure
+ **/
+static void igbvf_configure(struct igbvf_adapter *adapter)
+{
+ igbvf_set_multi(adapter->netdev);
+
+ igbvf_restore_vlan(adapter);
+
+ igbvf_configure_tx(adapter);
+ igbvf_setup_srrctl(adapter);
+ igbvf_configure_rx(adapter);
+ igbvf_alloc_rx_buffers(adapter->rx_ring,
+ igbvf_desc_unused(adapter->rx_ring));
+}
+
+/* igbvf_reset - bring the hardware into a known good state
+ *
+ * This function boots the hardware and enables some settings that
+ * require a configuration cycle of the hardware - those cannot be
+ * set/changed during runtime. After reset the device needs to be
+ * properly configured for Rx, Tx etc.
+ */
+void igbvf_reset(struct igbvf_adapter *adapter)
+{
+ struct e1000_mac_info *mac = &adapter->hw.mac;
+ struct net_device *netdev = adapter->netdev;
+ struct e1000_hw *hw = &adapter->hw;
+
+ /* Allow time for pending master requests to run */
+ if (mac->ops.reset_hw(hw))
+ dev_err(&adapter->pdev->dev, "PF still resetting\n");
+
+ mac->ops.init_hw(hw);
+
+ if (is_valid_ether_addr(adapter->hw.mac.addr)) {
+ memcpy(netdev->dev_addr, adapter->hw.mac.addr,
+ netdev->addr_len);
+ memcpy(netdev->perm_addr, adapter->hw.mac.addr,
+ netdev->addr_len);
+ }
+}
+
+int igbvf_up(struct igbvf_adapter *adapter)
+{
+ struct e1000_hw *hw = &adapter->hw;
+
+ /* hardware has been reset, we need to reload some things */
+ igbvf_configure(adapter);
+
+ clear_bit(__IGBVF_DOWN, &adapter->state);
+
+ napi_enable(&adapter->rx_ring->napi);
+ if (adapter->msix_entries)
+ igbvf_configure_msix(adapter);
+
+ /* Clear any pending interrupts. */
+ er32(EICR);
+ igbvf_irq_enable(adapter);
+
+ /* start the watchdog */
+ hw->mac.get_link_status = 1;
+ mod_timer(&adapter->watchdog_timer, jiffies + 1);
+
+
+ return 0;
+}
+
+void igbvf_down(struct igbvf_adapter *adapter)
+{
+ struct net_device *netdev = adapter->netdev;
+
+ /*
+ * signal that we're down so the interrupt handler does not
+ * reschedule our watchdog timer
+ */
+ set_bit(__IGBVF_DOWN, &adapter->state);
+
+ netif_stop_queue(netdev);
+
+ /* FIX ME!!!! */
+ /* need to disable local queue transmit and receive */
+
+ msleep(10);
+
+ napi_disable(&adapter->rx_ring->napi);
+
+ igbvf_irq_disable(adapter);
+
+ del_timer_sync(&adapter->watchdog_timer);
+
+ netdev->tx_queue_len = adapter->tx_queue_len;
+ netif_carrier_off(netdev);
+ adapter->link_speed = 0;
+ adapter->link_duplex = 0;
+
+ igbvf_reset(adapter);
+ igbvf_clean_tx_ring(adapter);
+ igbvf_clean_rx_ring(adapter);
+}
+
+void igbvf_reinit_locked(struct igbvf_adapter *adapter)
+{
+ might_sleep();
+ while (test_and_set_bit(__IGBVF_RESETTING, &adapter->state))
+ msleep(1);
+ igbvf_down(adapter);
+ igbvf_up(adapter);
+ clear_bit(__IGBVF_RESETTING, &adapter->state);
+}
+
+/**
+ * igbvf_sw_init - Initialize general software structures (struct igbvf_adapter)
+ * @adapter: board private structure to initialize
+ *
+ * igbvf_sw_init initializes the Adapter private data structure.
+ * Fields are initialized based on PCI device information and
+ * OS network device settings (MTU size).
+ **/
+static int __devinit igbvf_sw_init(struct igbvf_adapter *adapter)
+{
+ struct net_device *netdev = adapter->netdev;
+ s32 rc;
+
+ adapter->rx_buffer_len = ETH_FRAME_LEN + VLAN_HLEN + ETH_FCS_LEN;
+ adapter->rx_ps_hdr_size = 0;
+ adapter->max_frame_size = netdev->mtu + ETH_HLEN + ETH_FCS_LEN;
+ adapter->min_frame_size = ETH_ZLEN + ETH_FCS_LEN;
+
+ adapter->tx_int_delay = 8;
+ adapter->tx_abs_int_delay = 32;
+ adapter->rx_int_delay = 0;
+ adapter->rx_abs_int_delay = 8;
+ adapter->itr_setting = 3;
+ adapter->itr = 20000;
+
+ /* Set various function pointers */
+ adapter->ei->init_ops(&adapter->hw);
+
+ rc = adapter->hw.mac.ops.init_params(&adapter->hw);
+ if (rc)
+ return rc;
+
+ rc = adapter->hw.mbx.ops.init_params(&adapter->hw);
+ if (rc)
+ return rc;
+
+ igbvf_set_interrupt_capability(adapter);
+
+ if (igbvf_alloc_queues(adapter))
+ return -ENOMEM;
+
+ spin_lock_init(&adapter->tx_queue_lock);
+
+ /* Explicitly disable IRQ since the NIC can be in any state. */
+ igbvf_irq_disable(adapter);
+
+ spin_lock_init(&adapter->stats_lock);
+
+ set_bit(__IGBVF_DOWN, &adapter->state);
+ return 0;
+}
+
+static void igbvf_initialize_last_counter_stats(struct igbvf_adapter *adapter)
+{
+ struct e1000_hw *hw = &adapter->hw;
+
+ adapter->stats.last_gprc = er32(VFGPRC);
+ adapter->stats.last_gorc = er32(VFGORC);
+ adapter->stats.last_gptc = er32(VFGPTC);
+ adapter->stats.last_gotc = er32(VFGOTC);
+ adapter->stats.last_mprc = er32(VFMPRC);
+ adapter->stats.last_gotlbc = er32(VFGOTLBC);
+ adapter->stats.last_gptlbc = er32(VFGPTLBC);
+ adapter->stats.last_gorlbc = er32(VFGORLBC);
+ adapter->stats.last_gprlbc = er32(VFGPRLBC);
+
+ adapter->stats.base_gprc = er32(VFGPRC);
+ adapter->stats.base_gorc = er32(VFGORC);
+ adapter->stats.base_gptc = er32(VFGPTC);
+ adapter->stats.base_gotc = er32(VFGOTC);
+ adapter->stats.base_mprc = er32(VFMPRC);
+ adapter->stats.base_gotlbc = er32(VFGOTLBC);
+ adapter->stats.base_gptlbc = er32(VFGPTLBC);
+ adapter->stats.base_gorlbc = er32(VFGORLBC);
+ adapter->stats.base_gprlbc = er32(VFGPRLBC);
+}
+
+/**
+ * igbvf_open - Called when a network interface is made active
+ * @netdev: network interface device structure
+ *
+ * Returns 0 on success, negative value on failure
+ *
+ * The open entry point is called when a network interface is made
+ * active by the system (IFF_UP). At this point all resources needed
+ * for transmit and receive operations are allocated, the interrupt
+ * handler is registered with the OS, the watchdog timer is started,
+ * and the stack is notified that the interface is ready.
+ **/
+static int igbvf_open(struct net_device *netdev)
+{
+ struct igbvf_adapter *adapter = netdev_priv(netdev);
+ struct e1000_hw *hw = &adapter->hw;
+ int err;
+
+ /* disallow open during test */
+ if (test_bit(__IGBVF_TESTING, &adapter->state))
+ return -EBUSY;
+
+ /* allocate transmit descriptors */
+ err = igbvf_setup_tx_resources(adapter);
+ if (err)
+ goto err_setup_tx;
+
+ /* allocate receive descriptors */
+ err = igbvf_setup_rx_resources(adapter);
+ if (err)
+ goto err_setup_rx;
+
+ /*
+ * before we allocate an interrupt, we must be ready to handle it.
+ * Setting DEBUG_SHIRQ in the kernel makes it fire an interrupt
+ * as soon as we call pci_request_irq, so we have to setup our
+ * clean_rx handler before we do so.
+ */
+ igbvf_configure(adapter);
+
+ err = igbvf_request_irq(adapter);
+ if (err)
+ goto err_req_irq;
+
+ /* From here on the code is the same as igbvf_up() */
+ clear_bit(__IGBVF_DOWN, &adapter->state);
+
+ napi_enable(&adapter->rx_ring->napi);
+
+ /* clear any pending interrupts */
+ er32(EICR);
+
+ igbvf_irq_enable(adapter);
+
+ /* start the watchdog */
+ hw->mac.get_link_status = 1;
+ mod_timer(&adapter->watchdog_timer, jiffies + 1);
+
+ return 0;
+
+err_req_irq:
+ igbvf_free_rx_resources(adapter);
+err_setup_rx:
+ igbvf_free_tx_resources(adapter);
+err_setup_tx:
+ igbvf_reset(adapter);
+
+ return err;
+}
+
+/**
+ * igbvf_close - Disables a network interface
+ * @netdev: network interface device structure
+ *
+ * Returns 0, this is not allowed to fail
+ *
+ * The close entry point is called when an interface is de-activated
+ * by the OS. The hardware is still under the drivers control, but
+ * needs to be disabled. A global MAC reset is issued to stop the
+ * hardware, and all transmit and receive resources are freed.
+ **/
+static int igbvf_close(struct net_device *netdev)
+{
+ struct igbvf_adapter *adapter = netdev_priv(netdev);
+
+ WARN_ON(test_bit(__IGBVF_RESETTING, &adapter->state));
+ igbvf_down(adapter);
+
+ igbvf_free_irq(adapter);
+
+ igbvf_free_tx_resources(adapter);
+ igbvf_free_rx_resources(adapter);
+
+ return 0;
+}
+/**
+ * igbvf_set_mac - Change the Ethernet Address of the NIC
+ * @netdev: network interface device structure
+ * @p: pointer to an address structure
+ *
+ * Returns 0 on success, negative on failure
+ **/
+static int igbvf_set_mac(struct net_device *netdev, void *p)
+{
+ struct igbvf_adapter *adapter = netdev_priv(netdev);
+ struct e1000_hw *hw = &adapter->hw;
+ struct sockaddr *addr = p;
+
+ if (!is_valid_ether_addr(addr->sa_data))
+ return -EADDRNOTAVAIL;
+
+ memcpy(hw->mac.addr, addr->sa_data, netdev->addr_len);
+
+ hw->mac.ops.rar_set(hw, hw->mac.addr, 0);
+
+ if (memcmp(addr->sa_data, hw->mac.addr, 6))
+ return -EADDRNOTAVAIL;
+
+ memcpy(netdev->dev_addr, addr->sa_data, netdev->addr_len);
+
+ return 0;
+}
+
+#define UPDATE_VF_COUNTER(reg, name) \
+ { \
+ u32 current_counter = er32(reg); \
+ if (current_counter < adapter->stats.last_##name) \
+ adapter->stats.name += 0x100000000LL; \
+ adapter->stats.last_##name = current_counter; \
+ adapter->stats.name &= 0xFFFFFFFF00000000LL; \
+ adapter->stats.name |= current_counter; \
+ }
+
+/**
+ * igbvf_update_stats - Update the board statistics counters
+ * @adapter: board private structure
+**/
+void igbvf_update_stats(struct igbvf_adapter *adapter)
+{
+ struct e1000_hw *hw = &adapter->hw;
+ struct pci_dev *pdev = adapter->pdev;
+
+ /*
+ * Prevent stats update while adapter is being reset, link is down
+ * or if the pci connection is down.
+ */
+ if (adapter->link_speed == 0)
+ return;
+
+ if (test_bit(__IGBVF_RESETTING, &adapter->state))
+ return;
+
+ if (pci_channel_offline(pdev))
+ return;
+
+ UPDATE_VF_COUNTER(VFGPRC, gprc);
+ UPDATE_VF_COUNTER(VFGORC, gorc);
+ UPDATE_VF_COUNTER(VFGPTC, gptc);
+ UPDATE_VF_COUNTER(VFGOTC, gotc);
+ UPDATE_VF_COUNTER(VFMPRC, mprc);
+ UPDATE_VF_COUNTER(VFGOTLBC, gotlbc);
+ UPDATE_VF_COUNTER(VFGPTLBC, gptlbc);
+ UPDATE_VF_COUNTER(VFGORLBC, gorlbc);
+ UPDATE_VF_COUNTER(VFGPRLBC, gprlbc);
+
+ /* Fill out the OS statistics structure */
+ adapter->net_stats.multicast = adapter->stats.mprc;
+}
+
+static void igbvf_print_link_info(struct igbvf_adapter *adapter)
+{
+ dev_info(&adapter->pdev->dev, "Link is Up %d Mbps %s\n",
+ adapter->link_speed,
+ ((adapter->link_duplex == FULL_DUPLEX) ?
+ "Full Duplex" : "Half Duplex"));
+}
+
+static bool igbvf_has_link(struct igbvf_adapter *adapter)
+{
+ struct e1000_hw *hw = &adapter->hw;
+ s32 ret_val = E1000_SUCCESS;
+ bool link_active;
+
+ ret_val = hw->mac.ops.check_for_link(hw);
+ link_active = !hw->mac.get_link_status;
+
+ /* if check for link returns error we will need to reset */
+ if (ret_val)
+ schedule_work(&adapter->reset_task);
+
+ return link_active;
+}
+
+/**
+ * igbvf_watchdog - Timer Call-back
+ * @data: pointer to adapter cast into an unsigned long
+ **/
+static void igbvf_watchdog(unsigned long data)
+{
+ struct igbvf_adapter *adapter = (struct igbvf_adapter *) data;
+
+ /* Do the rest outside of interrupt context */
+ schedule_work(&adapter->watchdog_task);
+}
+
+static void igbvf_watchdog_task(struct work_struct *work)
+{
+ struct igbvf_adapter *adapter = container_of(work,
+ struct igbvf_adapter,
+ watchdog_task);
+ struct net_device *netdev = adapter->netdev;
+ struct e1000_mac_info *mac = &adapter->hw.mac;
+ struct igbvf_ring *tx_ring = adapter->tx_ring;
+ struct e1000_hw *hw = &adapter->hw;
+ u32 link;
+ int tx_pending = 0;
+
+ link = igbvf_has_link(adapter);
+
+ if (link) {
+ if (!netif_carrier_ok(netdev)) {
+ bool txb2b = 1;
+
+ mac->ops.get_link_up_info(&adapter->hw,
+ &adapter->link_speed,
+ &adapter->link_duplex);
+ igbvf_print_link_info(adapter);
+
+ /*
+ * tweak tx_queue_len according to speed/duplex
+ * and adjust the timeout factor
+ */
+ netdev->tx_queue_len = adapter->tx_queue_len;
+ adapter->tx_timeout_factor = 1;
+ switch (adapter->link_speed) {
+ case SPEED_10:
+ txb2b = 0;
+ netdev->tx_queue_len = 10;
+ adapter->tx_timeout_factor = 16;
+ break;
+ case SPEED_100:
+ txb2b = 0;
+ netdev->tx_queue_len = 100;
+ /* maybe add some timeout factor ? */
+ break;
+ }
+
+ netif_carrier_on(netdev);
+ netif_wake_queue(netdev);
+ }
+ } else {
+ if (netif_carrier_ok(netdev)) {
+ adapter->link_speed = 0;
+ adapter->link_duplex = 0;
+ dev_info(&adapter->pdev->dev, "Link is Down\n");
+ netif_carrier_off(netdev);
+ netif_stop_queue(netdev);
+ }
+ }
+
+ if (netif_carrier_ok(netdev)) {
+ igbvf_update_stats(adapter);
+ } else {
+ tx_pending = (igbvf_desc_unused(tx_ring) + 1 <
+ tx_ring->count);
+ if (tx_pending) {
+ /*
+ * We've lost link, so the controller stops DMA,
+ * but we've got queued Tx work that's never going
+ * to get done, so reset controller to flush Tx.
+ * (Do the reset outside of interrupt context).
+ */
+ adapter->tx_timeout_count++;
+ schedule_work(&adapter->reset_task);
+ }
+ }
+
+ /* Cause software interrupt to ensure Rx ring is cleaned */
+ ew32(EICS, adapter->rx_ring->eims_value);
+
+ /* Force detection of hung controller every watchdog period */
+ adapter->detect_tx_hung = 1;
+
+ /* Reset the timer */
+ if (!test_bit(__IGBVF_DOWN, &adapter->state))
+ mod_timer(&adapter->watchdog_timer,
+ round_jiffies(jiffies + (2 * HZ)));
+}
+
+#define IGBVF_TX_FLAGS_CSUM 0x00000001
+#define IGBVF_TX_FLAGS_VLAN 0x00000002
+#define IGBVF_TX_FLAGS_TSO 0x00000004
+#define IGBVF_TX_FLAGS_IPV4 0x00000008
+#define IGBVF_TX_FLAGS_VLAN_MASK 0xffff0000
+#define IGBVF_TX_FLAGS_VLAN_SHIFT 16
+
+static int igbvf_tso(struct igbvf_adapter *adapter,
+ struct igbvf_ring *tx_ring,
+ struct sk_buff *skb, u32 tx_flags, u8 *hdr_len)
+{
+ struct e1000_adv_tx_context_desc *context_desc;
+ unsigned int i;
+ int err;
+ struct igbvf_buffer *buffer_info;
+ u32 info = 0, tu_cmd = 0;
+ u32 mss_l4len_idx, l4len;
+ *hdr_len = 0;
+
+ if (skb_header_cloned(skb)) {
+ err = pskb_expand_head(skb, 0, 0, GFP_ATOMIC);
+ if (err) {
+ dev_err(&adapter->pdev->dev,
+ "igbvf_tso returning an error\n");
+ return err;
+ }
+ }
+
+ l4len = tcp_hdrlen(skb);
+ *hdr_len += l4len;
+
+ if (skb->protocol == htons(ETH_P_IP)) {
+ struct iphdr *iph = ip_hdr(skb);
+ iph->tot_len = 0;
+ iph->check = 0;
+ tcp_hdr(skb)->check = ~csum_tcpudp_magic(iph->saddr,
+ iph->daddr, 0,
+ IPPROTO_TCP,
+ 0);
+ } else if (skb_shinfo(skb)->gso_type == SKB_GSO_TCPV6) {
+ ipv6_hdr(skb)->payload_len = 0;
+ tcp_hdr(skb)->check = ~csum_ipv6_magic(&ipv6_hdr(skb)->saddr,
+ &ipv6_hdr(skb)->daddr,
+ 0, IPPROTO_TCP, 0);
+ }
+
+ i = tx_ring->next_to_use;
+
+ buffer_info = &tx_ring->buffer_info[i];
+ context_desc = IGBVF_TX_CTXTDESC_ADV(*tx_ring, i);
+ /* VLAN MACLEN IPLEN */
+ if (tx_flags & IGBVF_TX_FLAGS_VLAN)
+ info |= (tx_flags & IGBVF_TX_FLAGS_VLAN_MASK);
+ info |= (skb_network_offset(skb) << E1000_ADVTXD_MACLEN_SHIFT);
+ *hdr_len += skb_network_offset(skb);
+ info |= (skb_transport_header(skb) - skb_network_header(skb));
+ *hdr_len += (skb_transport_header(skb) - skb_network_header(skb));
+ context_desc->vlan_macip_lens = cpu_to_le32(info);
+
+ /* ADV DTYP TUCMD MKRLOC/ISCSIHEDLEN */
+ tu_cmd |= (E1000_TXD_CMD_DEXT | E1000_ADVTXD_DTYP_CTXT);
+
+ if (skb->protocol == htons(ETH_P_IP))
+ tu_cmd |= E1000_ADVTXD_TUCMD_IPV4;
+ tu_cmd |= E1000_ADVTXD_TUCMD_L4T_TCP;
+
+ context_desc->type_tucmd_mlhl = cpu_to_le32(tu_cmd);
+
+ /* MSS L4LEN IDX */
+ mss_l4len_idx = (skb_shinfo(skb)->gso_size << E1000_ADVTXD_MSS_SHIFT);
+ mss_l4len_idx |= (l4len << E1000_ADVTXD_L4LEN_SHIFT);
+
+ context_desc->mss_l4len_idx = cpu_to_le32(mss_l4len_idx);
+ context_desc->seqnum_seed = 0;
+
+ buffer_info->time_stamp = jiffies;
+ buffer_info->dma = 0;
+ i++;
+ if (i == tx_ring->count)
+ i = 0;
+
+ tx_ring->next_to_use = i;
+
+ return true;
+}
+
+static inline bool igbvf_tx_csum(struct igbvf_adapter *adapter,
+ struct igbvf_ring *tx_ring,
+ struct sk_buff *skb, u32 tx_flags)
+{
+ struct e1000_adv_tx_context_desc *context_desc;
+ unsigned int i;
+ struct igbvf_buffer *buffer_info;
+ u32 info = 0, tu_cmd = 0;
+
+ if ((skb->ip_summed == CHECKSUM_PARTIAL) ||
+ (tx_flags & IGBVF_TX_FLAGS_VLAN)) {
+ i = tx_ring->next_to_use;
+ buffer_info = &tx_ring->buffer_info[i];
+ context_desc = IGBVF_TX_CTXTDESC_ADV(*tx_ring, i);
+
+ if (tx_flags & IGBVF_TX_FLAGS_VLAN)
+ info |= (tx_flags & IGBVF_TX_FLAGS_VLAN_MASK);
+
+ info |= (skb_network_offset(skb) << E1000_ADVTXD_MACLEN_SHIFT);
+ if (skb->ip_summed == CHECKSUM_PARTIAL)
+ info |= (skb_transport_header(skb) -
+ skb_network_header(skb));
+
+
+ context_desc->vlan_macip_lens = cpu_to_le32(info);
+
+ tu_cmd |= (E1000_TXD_CMD_DEXT | E1000_ADVTXD_DTYP_CTXT);
+
+ if (skb->ip_summed == CHECKSUM_PARTIAL) {
+ switch (skb->protocol) {
+ case __constant_htons(ETH_P_IP):
+ tu_cmd |= E1000_ADVTXD_TUCMD_IPV4;
+ if (ip_hdr(skb)->protocol == IPPROTO_TCP)
+ tu_cmd |= E1000_ADVTXD_TUCMD_L4T_TCP;
+ break;
+ case __constant_htons(ETH_P_IPV6):
+ if (ipv6_hdr(skb)->nexthdr == IPPROTO_TCP)
+ tu_cmd |= E1000_ADVTXD_TUCMD_L4T_TCP;
+ break;
+ default:
+ break;
+ }
+ }
+
+ context_desc->type_tucmd_mlhl = cpu_to_le32(tu_cmd);
+ context_desc->seqnum_seed = 0;
+ context_desc->mss_l4len_idx = 0;
+
+ buffer_info->time_stamp = jiffies;
+ buffer_info->dma = 0;
+ i++;
+ if (i == tx_ring->count)
+ i = 0;
+ tx_ring->next_to_use = i;
+
+ return true;
+ }
+
+ return false;
+}
+
+static int igbvf_maybe_stop_tx(struct net_device *netdev, int size)
+{
+ struct igbvf_adapter *adapter = netdev_priv(netdev);
+
+ /* there is enough descriptors then we don't need to worry */
+ if (igbvf_desc_unused(adapter->tx_ring) >= size)
+ return 0;
+
+ netif_stop_queue(netdev);
+
+ smp_mb();
+
+ /* We need to check again just in case room has been made available */
+ if (igbvf_desc_unused(adapter->tx_ring) < size)
+ return -EBUSY;
+
+ netif_wake_queue(netdev);
+
+ ++adapter->restart_queue;
+ return 0;
+}
+
+#define IGBVF_MAX_TXD_PWR 16
+#define IGBVF_MAX_DATA_PER_TXD (1 << IGBVF_MAX_TXD_PWR)
+
+static inline int igbvf_tx_map_adv(struct igbvf_adapter *adapter,
+ struct igbvf_ring *tx_ring,
+ struct sk_buff *skb,
+ unsigned int first)
+{
+ struct igbvf_buffer *buffer_info;
+ unsigned int len = skb_headlen(skb);
+ unsigned int count = 0, i;
+ unsigned int f;
+
+ i = tx_ring->next_to_use;
+
+ buffer_info = &tx_ring->buffer_info[i];
+ BUG_ON(len >= IGBVF_MAX_DATA_PER_TXD);
+ buffer_info->length = len;
+ /* set time_stamp *before* dma to help avoid a possible race */
+ buffer_info->time_stamp = jiffies;
+
+ buffer_info->dma = pci_map_single(adapter->pdev, skb->data, len,
+ PCI_DMA_TODEVICE);
+
+ count++;
+ i++;
+ if (i == tx_ring->count)
+ i = 0;
+
+ for (f = 0; f < skb_shinfo(skb)->nr_frags; f++) {
+ struct skb_frag_struct *frag;
+
+ frag = &skb_shinfo(skb)->frags[f];
+ len = frag->size;
+
+ buffer_info = &tx_ring->buffer_info[i];
+ BUG_ON(len >= IGBVF_MAX_DATA_PER_TXD);
+ buffer_info->length = len;
+ buffer_info->time_stamp = jiffies;
+ buffer_info->dma = pci_map_page(adapter->pdev, frag->page,
+ frag->page_offset,
+ len, PCI_DMA_TODEVICE);
+ count++;
+ i++;
+ if (i == tx_ring->count)
+ i = 0;
+ }
+
+ i = (i == 0) ? tx_ring->count - 1 : i - 1;
+ tx_ring->buffer_info[i].skb = skb;
+ tx_ring->buffer_info[first].next_to_watch = i;
+
+ return count;
+}
+
+static inline void igbvf_tx_queue_adv(struct igbvf_adapter *adapter,
+ struct igbvf_ring *tx_ring,
+ int tx_flags, int count, u32 paylen,
+ u8 hdr_len)
+{
+ union e1000_adv_tx_desc *tx_desc = NULL;
+ struct igbvf_buffer *buffer_info;
+ u32 olinfo_status = 0, cmd_type_len;
+ unsigned int i;
+
+ cmd_type_len = (E1000_ADVTXD_DTYP_DATA | E1000_ADVTXD_DCMD_IFCS |
+ E1000_ADVTXD_DCMD_DEXT);
+
+ if (tx_flags & IGBVF_TX_FLAGS_VLAN)
+ cmd_type_len |= E1000_ADVTXD_DCMD_VLE;
+
+ if (tx_flags & IGBVF_TX_FLAGS_TSO) {
+ cmd_type_len |= E1000_ADVTXD_DCMD_TSE;
+
+ /* insert tcp checksum */
+ olinfo_status |= E1000_TXD_POPTS_TXSM << 8;
+
+ /* insert ip checksum */
+ if (tx_flags & IGBVF_TX_FLAGS_IPV4)
+ olinfo_status |= E1000_TXD_POPTS_IXSM << 8;
+
+ } else if (tx_flags & IGBVF_TX_FLAGS_CSUM) {
+ olinfo_status |= E1000_TXD_POPTS_TXSM << 8;
+ }
+
+ olinfo_status |= ((paylen - hdr_len) << E1000_ADVTXD_PAYLEN_SHIFT);
+
+ i = tx_ring->next_to_use;
+ while (count--) {
+ buffer_info = &tx_ring->buffer_info[i];
+ tx_desc = IGBVF_TX_DESC_ADV(*tx_ring, i);
+ tx_desc->read.buffer_addr = cpu_to_le64(buffer_info->dma);
+ tx_desc->read.cmd_type_len =
+ cpu_to_le32(cmd_type_len | buffer_info->length);
+ tx_desc->read.olinfo_status = cpu_to_le32(olinfo_status);
+ i++;
+ if (i == tx_ring->count)
+ i = 0;
+ }
+
+ tx_desc->read.cmd_type_len |= cpu_to_le32(adapter->txd_cmd);
+ /* 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). */
+ wmb();
+
+ tx_ring->next_to_use = i;
+ writel(i, adapter->hw.hw_addr + tx_ring->tail);
+ /* we need this if more than one processor can write to our tail
+ * at a time, it syncronizes IO on IA64/Altix systems */
+ mmiowb();
+}
+
+static int igbvf_xmit_frame_ring_adv(struct sk_buff *skb,
+ struct net_device *netdev,
+ struct igbvf_ring *tx_ring)
+{
+ struct igbvf_adapter *adapter = netdev_priv(netdev);
+ unsigned int first, tx_flags = 0;
+ unsigned int len;
+ u8 hdr_len = 0;
+ int tso = 0;
+
+ len = skb_headlen(skb);
+
+ if (test_bit(__IGBVF_DOWN, &adapter->state)) {
+ dev_kfree_skb_any(skb);
+ return NETDEV_TX_OK;
+ }
+
+ if (skb->len <= 0) {
+ dev_kfree_skb_any(skb);
+ return NETDEV_TX_OK;
+ }
+
+ /*
+ * need: count + 4 desc gap to keep tail from touching
+ * + 2 desc gap to keep tail from touching head,
+ * + 1 desc for skb->data,
+ * + 1 desc for context descriptor,
+ * head, otherwise try next time
+ */
+ if (igbvf_maybe_stop_tx(netdev, skb_shinfo(skb)->nr_frags + 4)) {
+ /* this is a hard error */
+ return NETDEV_TX_BUSY;
+ }
+ skb_orphan(skb);
+
+ if (adapter->vlgrp && vlan_tx_tag_present(skb)) {
+ tx_flags |= IGBVF_TX_FLAGS_VLAN;
+ tx_flags |= (vlan_tx_tag_get(skb) << IGBVF_TX_FLAGS_VLAN_SHIFT);
+ }
+
+ if (skb->protocol == htons(ETH_P_IP))
+ tx_flags |= IGBVF_TX_FLAGS_IPV4;
+
+ first = tx_ring->next_to_use;
+
+ tso = skb_is_gso(skb) ?
+ igbvf_tso(adapter, tx_ring, skb, tx_flags, &hdr_len) : 0;
+ if (unlikely(tso < 0)) {
+ dev_kfree_skb_any(skb);
+ return NETDEV_TX_OK;
+ }
+
+ if (tso)
+ tx_flags |= IGBVF_TX_FLAGS_TSO;
+ else if (igbvf_tx_csum(adapter, tx_ring, skb, tx_flags) &&
+ (skb->ip_summed == CHECKSUM_PARTIAL))
+ tx_flags |= IGBVF_TX_FLAGS_CSUM;
+
+ igbvf_tx_queue_adv(adapter, tx_ring, tx_flags,
+ igbvf_tx_map_adv(adapter, tx_ring, skb, first),
+ skb->len, hdr_len);
+
+ netdev->trans_start = jiffies;
+
+ /* Make sure there is space in the ring for the next send. */
+ igbvf_maybe_stop_tx(netdev, MAX_SKB_FRAGS + 4);
+
+ return NETDEV_TX_OK;
+}
+
+static int igbvf_xmit_frame(struct sk_buff *skb, struct net_device *netdev)
+{
+ struct igbvf_adapter *adapter = netdev_priv(netdev);
+ struct igbvf_ring *tx_ring;
+ int retval;
+
+ if (test_bit(__IGBVF_RESETTING, &adapter->state))
+ return NETDEV_TX_BUSY;
+
+ tx_ring = &adapter->tx_ring[0];
+
+ retval = igbvf_xmit_frame_ring_adv(skb, netdev, tx_ring);
+
+ return retval;
+}
+
+/**
+ * igbvf_tx_timeout - Respond to a Tx Hang
+ * @netdev: network interface device structure
+ **/
+static void igbvf_tx_timeout(struct net_device *netdev)
+{
+ struct igbvf_adapter *adapter = netdev_priv(netdev);
+
+ /* Do the reset outside of interrupt context */
+ adapter->tx_timeout_count++;
+ schedule_work(&adapter->reset_task);
+}
+
+static void igbvf_reset_task(struct work_struct *work)
+{
+ struct igbvf_adapter *adapter;
+ adapter = container_of(work, struct igbvf_adapter, reset_task);
+
+ igbvf_reinit_locked(adapter);
+}
+
+/**
+ * igbvf_get_stats - Get System Network Statistics
+ * @netdev: network interface device structure
+ *
+ * Returns the address of the device statistics structure.
+ * The statistics are actually updated from the timer callback.
+ **/
+static struct net_device_stats *igbvf_get_stats(struct net_device *netdev)
+{
+ struct igbvf_adapter *adapter = netdev_priv(netdev);
+
+ /* only return the current stats */
+ return &adapter->net_stats;
+}
+
+/**
+ * igbvf_change_mtu - Change the Maximum Transfer Unit
+ * @netdev: network interface device structure
+ * @new_mtu: new value for maximum frame size
+ *
+ * Returns 0 on success, negative on failure
+ **/
+static int igbvf_change_mtu(struct net_device *netdev, int new_mtu)
+{
+ struct igbvf_adapter *adapter = netdev_priv(netdev);
+ int max_frame = new_mtu + ETH_HLEN + ETH_FCS_LEN;
+
+ if ((new_mtu < 68) || (max_frame > MAX_JUMBO_FRAME_SIZE)) {
+ dev_err(&adapter->pdev->dev, "Invalid MTU setting\n");
+ return -EINVAL;
+ }
+
+ /* Jumbo frame size limits */
+ if (max_frame > ETH_FRAME_LEN + ETH_FCS_LEN) {
+ if (!(adapter->flags & FLAG_HAS_JUMBO_FRAMES)) {
+ dev_err(&adapter->pdev->dev,
+ "Jumbo Frames not supported.\n");
+ return -EINVAL;
+ }
+ }
+
+#define MAX_STD_JUMBO_FRAME_SIZE 9234
+ if (max_frame > MAX_STD_JUMBO_FRAME_SIZE) {
+ dev_err(&adapter->pdev->dev, "MTU > 9216 not supported.\n");
+ return -EINVAL;
+ }
+
+ while (test_and_set_bit(__IGBVF_RESETTING, &adapter->state))
+ msleep(1);
+ /* igbvf_down has a dependency on max_frame_size */
+ adapter->max_frame_size = max_frame;
+ if (netif_running(netdev))
+ igbvf_down(adapter);
+
+ /*
+ * NOTE: netdev_alloc_skb reserves 16 bytes, and typically NET_IP_ALIGN
+ * means we reserve 2 more, this pushes us to allocate from the next
+ * larger slab size.
+ * i.e. RXBUFFER_2048 --> size-4096 slab
+ * However with the new *_jumbo_rx* routines, jumbo receives will use
+ * fragmented skbs
+ */
+
+ if (max_frame <= 1024)
+ adapter->rx_buffer_len = 1024;
+ else if (max_frame <= 2048)
+ adapter->rx_buffer_len = 2048;
+ else
+#if (PAGE_SIZE / 2) > 16384
+ adapter->rx_buffer_len = 16384;
+#else
+ adapter->rx_buffer_len = PAGE_SIZE / 2;
+#endif
+
+
+ /* adjust allocation if LPE protects us, and we aren't using SBP */
+ if ((max_frame == ETH_FRAME_LEN + ETH_FCS_LEN) ||
+ (max_frame == ETH_FRAME_LEN + VLAN_HLEN + ETH_FCS_LEN))
+ adapter->rx_buffer_len = ETH_FRAME_LEN + VLAN_HLEN +
+ ETH_FCS_LEN;
+
+ dev_info(&adapter->pdev->dev, "changing MTU from %d to %d\n",
+ netdev->mtu, new_mtu);
+ netdev->mtu = new_mtu;
+
+ if (netif_running(netdev))
+ igbvf_up(adapter);
+ else
+ igbvf_reset(adapter);
+
+ clear_bit(__IGBVF_RESETTING, &adapter->state);
+
+ return 0;
+}
+
+static int igbvf_ioctl(struct net_device *netdev, struct ifreq *ifr, int cmd)
+{
+ switch (cmd) {
+ default:
+ return -EOPNOTSUPP;
+ }
+}
+
+static int igbvf_suspend(struct pci_dev *pdev, pm_message_t state)
+{
+ struct net_device *netdev = pci_get_drvdata(pdev);
+ struct igbvf_adapter *adapter = netdev_priv(netdev);
+#ifdef CONFIG_PM
+ int retval = 0;
+#endif
+
+ netif_device_detach(netdev);
+
+ if (netif_running(netdev)) {
+ WARN_ON(test_bit(__IGBVF_RESETTING, &adapter->state));
+ igbvf_down(adapter);
+ igbvf_free_irq(adapter);
+ }
+
+#ifdef CONFIG_PM
+ retval = pci_save_state(pdev);
+ if (retval)
+ return retval;
+#endif
+
+ pci_disable_device(pdev);
+
+ return 0;
+}
+
+#ifdef CONFIG_PM
+static int igbvf_resume(struct pci_dev *pdev)
+{
+ struct net_device *netdev = pci_get_drvdata(pdev);
+ struct igbvf_adapter *adapter = netdev_priv(netdev);
+ u32 err;
+
+ pci_restore_state(pdev);
+ err = pci_enable_device_mem(pdev);
+ if (err) {
+ dev_err(&pdev->dev, "Cannot enable PCI device from suspend\n");
+ return err;
+ }
+
+ pci_set_master(pdev);
+
+ if (netif_running(netdev)) {
+ err = igbvf_request_irq(adapter);
+ if (err)
+ return err;
+ }
+
+ igbvf_reset(adapter);
+
+ if (netif_running(netdev))
+ igbvf_up(adapter);
+
+ netif_device_attach(netdev);
+
+ return 0;
+}
+#endif
+
+static void igbvf_shutdown(struct pci_dev *pdev)
+{
+ igbvf_suspend(pdev, PMSG_SUSPEND);
+}
+
+#ifdef CONFIG_NET_POLL_CONTROLLER
+/*
+ * Polling 'interrupt' - used by things like netconsole to send skbs
+ * without having to re-enable interrupts. It's not called while
+ * the interrupt routine is executing.
+ */
+static void igbvf_netpoll(struct net_device *netdev)
+{
+ struct igbvf_adapter *adapter = netdev_priv(netdev);
+
+ disable_irq(adapter->pdev->irq);
+
+ igbvf_clean_tx_irq(adapter->tx_ring);
+
+ enable_irq(adapter->pdev->irq);
+}
+#endif
+
+/**
+ * igbvf_io_error_detected - called when PCI error is detected
+ * @pdev: Pointer to PCI device
+ * @state: The current pci connection state
+ *
+ * This function is called after a PCI bus error affecting
+ * this device has been detected.
+ */
+static pci_ers_result_t igbvf_io_error_detected(struct pci_dev *pdev,
+ pci_channel_state_t state)
+{
+ struct net_device *netdev = pci_get_drvdata(pdev);
+ struct igbvf_adapter *adapter = netdev_priv(netdev);
+
+ netif_device_detach(netdev);
+
+ if (netif_running(netdev))
+ igbvf_down(adapter);
+ pci_disable_device(pdev);
+
+ /* Request a slot slot reset. */
+ return PCI_ERS_RESULT_NEED_RESET;
+}
+
+/**
+ * igbvf_io_slot_reset - called after the pci bus has been reset.
+ * @pdev: Pointer to PCI device
+ *
+ * Restart the card from scratch, as if from a cold-boot. Implementation
+ * resembles the first-half of the igbvf_resume routine.
+ */
+static pci_ers_result_t igbvf_io_slot_reset(struct pci_dev *pdev)
+{
+ struct net_device *netdev = pci_get_drvdata(pdev);
+ struct igbvf_adapter *adapter = netdev_priv(netdev);
+
+ if (pci_enable_device_mem(pdev)) {
+ dev_err(&pdev->dev,
+ "Cannot re-enable PCI device after reset.\n");
+ return PCI_ERS_RESULT_DISCONNECT;
+ }
+ pci_set_master(pdev);
+
+ igbvf_reset(adapter);
+
+ return PCI_ERS_RESULT_RECOVERED;
+}
+
+/**
+ * igbvf_io_resume - called when traffic can start flowing again.
+ * @pdev: Pointer to PCI device
+ *
+ * This callback is called when the error recovery driver tells us that
+ * its OK to resume normal operation. Implementation resembles the
+ * second-half of the igbvf_resume routine.
+ */
+static void igbvf_io_resume(struct pci_dev *pdev)
+{
+ struct net_device *netdev = pci_get_drvdata(pdev);
+ struct igbvf_adapter *adapter = netdev_priv(netdev);
+
+ if (netif_running(netdev)) {
+ if (igbvf_up(adapter)) {
+ dev_err(&pdev->dev,
+ "can't bring device back up after reset\n");
+ return;
+ }
+ }
+
+ netif_device_attach(netdev);
+}
+
+static void igbvf_print_device_info(struct igbvf_adapter *adapter)
+{
+ struct e1000_hw *hw = &adapter->hw;
+ struct net_device *netdev = adapter->netdev;
+ struct pci_dev *pdev = adapter->pdev;
+
+ dev_info(&pdev->dev, "Intel(R) 82576 Virtual Function\n");
+ dev_info(&pdev->dev, "Address: %02x:%02x:%02x:%02x:%02x:%02x\n",
+ /* MAC address */
+ netdev->dev_addr[0], netdev->dev_addr[1],
+ netdev->dev_addr[2], netdev->dev_addr[3],
+ netdev->dev_addr[4], netdev->dev_addr[5]);
+ dev_info(&pdev->dev, "MAC: %d\n", hw->mac.type);
+}
+
+static const struct net_device_ops igbvf_netdev_ops = {
+ .ndo_open = igbvf_open,
+ .ndo_stop = igbvf_close,
+ .ndo_start_xmit = igbvf_xmit_frame,
+ .ndo_get_stats = igbvf_get_stats,
+ .ndo_set_multicast_list = igbvf_set_multi,
+ .ndo_set_mac_address = igbvf_set_mac,
+ .ndo_change_mtu = igbvf_change_mtu,
+ .ndo_do_ioctl = igbvf_ioctl,
+ .ndo_tx_timeout = igbvf_tx_timeout,
+ .ndo_vlan_rx_register = igbvf_vlan_rx_register,
+ .ndo_vlan_rx_add_vid = igbvf_vlan_rx_add_vid,
+ .ndo_vlan_rx_kill_vid = igbvf_vlan_rx_kill_vid,
+#ifdef CONFIG_NET_POLL_CONTROLLER
+ .ndo_poll_controller = igbvf_netpoll,
+#endif
+};
+
+/**
+ * igbvf_probe - Device Initialization Routine
+ * @pdev: PCI device information struct
+ * @ent: entry in igbvf_pci_tbl
+ *
+ * Returns 0 on success, negative on failure
+ *
+ * igbvf_probe initializes an adapter identified by a pci_dev structure.
+ * The OS initialization, configuring of the adapter private structure,
+ * and a hardware reset occur.
+ **/
+static int __devinit igbvf_probe(struct pci_dev *pdev,
+ const struct pci_device_id *ent)
+{
+ struct net_device *netdev;
+ struct igbvf_adapter *adapter;
+ struct e1000_hw *hw;
+ const struct igbvf_info *ei = igbvf_info_tbl[ent->driver_data];
+
+ static int cards_found;
+ int err, pci_using_dac;
+
+ err = pci_enable_device_mem(pdev);
+ if (err)
+ return err;
+
+ pci_using_dac = 0;
+ err = pci_set_dma_mask(pdev, DMA_64BIT_MASK);
+ if (!err) {
+ err = pci_set_consistent_dma_mask(pdev, DMA_64BIT_MASK);
+ if (!err)
+ pci_using_dac = 1;
+ } else {
+ err = pci_set_dma_mask(pdev, DMA_32BIT_MASK);
+ if (err) {
+ err = pci_set_consistent_dma_mask(pdev, DMA_32BIT_MASK);
+ if (err) {
+ dev_err(&pdev->dev, "No usable DMA "
+ "configuration, aborting\n");
+ goto err_dma;
+ }
+ }
+ }
+
+ err = pci_request_regions(pdev, igbvf_driver_name);
+ if (err)
+ goto err_pci_reg;
+
+ pci_set_master(pdev);
+
+ err = -ENOMEM;
+ netdev = alloc_etherdev(sizeof(struct igbvf_adapter));
+ if (!netdev)
+ goto err_alloc_etherdev;
+
+ SET_NETDEV_DEV(netdev, &pdev->dev);
+
+ pci_set_drvdata(pdev, netdev);
+ adapter = netdev_priv(netdev);
+ hw = &adapter->hw;
+ adapter->netdev = netdev;
+ adapter->pdev = pdev;
+ adapter->ei = ei;
+ adapter->pba = ei->pba;
+ adapter->flags = ei->flags;
+ adapter->hw.back = adapter;
+ adapter->hw.mac.type = ei->mac;
+ adapter->msg_enable = (1 << NETIF_MSG_DRV | NETIF_MSG_PROBE) - 1;
+
+ /* PCI config space info */
+
+ hw->vendor_id = pdev->vendor;
+ hw->device_id = pdev->device;
+ hw->subsystem_vendor_id = pdev->subsystem_vendor;
+ hw->subsystem_device_id = pdev->subsystem_device;
+
+ pci_read_config_byte(pdev, PCI_REVISION_ID, &hw->revision_id);
+
+ err = -EIO;
+ adapter->hw.hw_addr = ioremap(pci_resource_start(pdev, 0),
+ pci_resource_len(pdev, 0));
+
+ if (!adapter->hw.hw_addr)
+ goto err_ioremap;
+
+ if (ei->get_variants) {
+ err = ei->get_variants(adapter);
+ if (err)
+ goto err_hw_init;
+ }
+
+ /* setup adapter struct */
+ err = igbvf_sw_init(adapter);
+ if (err)
+ goto err_sw_init;
+
+ /* construct the net_device struct */
+ netdev->netdev_ops = &igbvf_netdev_ops;
+
+ igbvf_set_ethtool_ops(netdev);
+ netdev->watchdog_timeo = 5 * HZ;
+ strncpy(netdev->name, pci_name(pdev), sizeof(netdev->name) - 1);
+
+ adapter->bd_number = cards_found++;
+
+ netdev->features = NETIF_F_SG |
+ NETIF_F_IP_CSUM |
+ NETIF_F_HW_VLAN_TX |
+ NETIF_F_HW_VLAN_RX |
+ NETIF_F_HW_VLAN_FILTER;
+
+ netdev->features |= NETIF_F_IPV6_CSUM;
+ netdev->features |= NETIF_F_TSO;
+ netdev->features |= NETIF_F_TSO6;
+
+ if (pci_using_dac)
+ netdev->features |= NETIF_F_HIGHDMA;
+
+ netdev->vlan_features |= NETIF_F_TSO;
+ netdev->vlan_features |= NETIF_F_TSO6;
+ netdev->vlan_features |= NETIF_F_IP_CSUM;
+ netdev->vlan_features |= NETIF_F_IPV6_CSUM;
+ netdev->vlan_features |= NETIF_F_SG;
+
+ /*reset the controller to put the device in a known good state */
+ err = hw->mac.ops.reset_hw(hw);
+ if (err) {
+ dev_info(&pdev->dev,
+ "PF still in reset state, assigning new address\n");
+ random_ether_addr(hw->mac.addr);
+ } else {
+ err = hw->mac.ops.read_mac_addr(hw);
+ if (err) {
+ dev_err(&pdev->dev, "Error reading MAC address\n");
+ goto err_hw_init;
+ }
+ }
+
+ memcpy(netdev->dev_addr, adapter->hw.mac.addr, netdev->addr_len);
+ memcpy(netdev->perm_addr, adapter->hw.mac.addr, netdev->addr_len);
+
+ if (!is_valid_ether_addr(netdev->perm_addr)) {
+ dev_err(&pdev->dev, "Invalid MAC Address: "
+ "%02x:%02x:%02x:%02x:%02x:%02x\n",
+ netdev->dev_addr[0], netdev->dev_addr[1],
+ netdev->dev_addr[2], netdev->dev_addr[3],
+ netdev->dev_addr[4], netdev->dev_addr[5]);
+ err = -EIO;
+ goto err_hw_init;
+ }
+
+ init_timer(&adapter->watchdog_timer);
+ adapter->watchdog_timer.function = &igbvf_watchdog;
+ adapter->watchdog_timer.data = (unsigned long) adapter;
+
+ INIT_WORK(&adapter->reset_task, igbvf_reset_task);
+ INIT_WORK(&adapter->watchdog_task, igbvf_watchdog_task);
+
+ /* ring size defaults */
+ adapter->rx_ring->count = 1024;
+ adapter->tx_ring->count = 1024;
+
+ /* reset the hardware with the new settings */
+ igbvf_reset(adapter);
+
+ /* tell the stack to leave us alone until igbvf_open() is called */
+ netif_carrier_off(netdev);
+ netif_stop_queue(netdev);
+
+ strcpy(netdev->name, "eth%d");
+ err = register_netdev(netdev);
+ if (err)
+ goto err_hw_init;
+
+ igbvf_print_device_info(adapter);
+
+ igbvf_initialize_last_counter_stats(adapter);
+
+ return 0;
+
+err_hw_init:
+ kfree(adapter->tx_ring);
+ kfree(adapter->rx_ring);
+ igbvf_reset_interrupt_capability(adapter);
+err_sw_init:
+ iounmap(adapter->hw.hw_addr);
+err_ioremap:
+ free_netdev(netdev);
+err_alloc_etherdev:
+ pci_release_regions(pdev);
+err_pci_reg:
+err_dma:
+ pci_disable_device(pdev);
+ return err;
+}
+
+/**
+ * igbvf_remove - Device Removal Routine
+ * @pdev: PCI device information struct
+ *
+ * igbvf_remove is called by the PCI subsystem to alert the driver
+ * that it should release a PCI device. The could be caused by a
+ * Hot-Plug event, or because the driver is going to be removed from
+ * memory.
+ **/
+static void __devexit igbvf_remove(struct pci_dev *pdev)
+{
+ struct net_device *netdev = pci_get_drvdata(pdev);
+ struct igbvf_adapter *adapter = netdev_priv(netdev);
+ struct e1000_hw *hw = &adapter->hw;
+
+ /*
+ * flush_scheduled work may reschedule our watchdog task, so
+ * explicitly disable watchdog tasks from being rescheduled
+ */
+ set_bit(__IGBVF_DOWN, &adapter->state);
+ del_timer_sync(&adapter->watchdog_timer);
+
+ flush_scheduled_work();
+
+ unregister_netdev(netdev);
+
+ igbvf_reset_interrupt_capability(adapter);
+ kfree(adapter->tx_ring);
+ kfree(adapter->rx_ring);
+
+ iounmap(hw->hw_addr);
+ if (hw->flash_address)
+ iounmap(hw->flash_address);
+ pci_release_regions(pdev);
+
+ free_netdev(netdev);
+
+ pci_disable_device(pdev);
+}
+
+/* PCI Error Recovery (ERS) */
+static struct pci_error_handlers igbvf_err_handler = {
+ .error_detected = igbvf_io_error_detected,
+ .slot_reset = igbvf_io_slot_reset,
+ .resume = igbvf_io_resume,
+};
+
+static struct pci_device_id igbvf_pci_tbl[] = {
+ { PCI_VDEVICE(INTEL, E1000_DEV_ID_82576_VF), board_vf },
+ { } /* terminate list */
+};
+MODULE_DEVICE_TABLE(pci, igbvf_pci_tbl);
+
+/* PCI Device API Driver */
+static struct pci_driver igbvf_driver = {
+ .name = igbvf_driver_name,
+ .id_table = igbvf_pci_tbl,
+ .probe = igbvf_probe,
+ .remove = __devexit_p(igbvf_remove),
+#ifdef CONFIG_PM
+ /* Power Management Hooks */
+ .suspend = igbvf_suspend,
+ .resume = igbvf_resume,
+#endif
+ .shutdown = igbvf_shutdown,
+ .err_handler = &igbvf_err_handler
+};
+
+/**
+ * igbvf_init_module - Driver Registration Routine
+ *
+ * igbvf_init_module is the first routine called when the driver is
+ * loaded. All it does is register with the PCI subsystem.
+ **/
+static int __init igbvf_init_module(void)
+{
+ int ret;
+ printk(KERN_INFO "%s - version %s\n",
+ igbvf_driver_string, igbvf_driver_version);
+ printk(KERN_INFO "%s\n", igbvf_copyright);
+
+ ret = pci_register_driver(&igbvf_driver);
+ pm_qos_add_requirement(PM_QOS_CPU_DMA_LATENCY, igbvf_driver_name,
+ PM_QOS_DEFAULT_VALUE);
+
+ return ret;
+}
+module_init(igbvf_init_module);
+
+/**
+ * igbvf_exit_module - Driver Exit Cleanup Routine
+ *
+ * igbvf_exit_module is called just before the driver is removed
+ * from memory.
+ **/
+static void __exit igbvf_exit_module(void)
+{
+ pci_unregister_driver(&igbvf_driver);
+ pm_qos_remove_requirement(PM_QOS_CPU_DMA_LATENCY, igbvf_driver_name);
+}
+module_exit(igbvf_exit_module);
+
+
+MODULE_AUTHOR("Intel Corporation, <[email protected]>");
+MODULE_DESCRIPTION("Intel(R) 82576 Virtual Function Network Driver");
+MODULE_LICENSE("GPL");
+MODULE_VERSION(DRV_VERSION);
+
+/* netdev.c */
diff --git a/drivers/net/igbvf/regs.h b/drivers/net/igbvf/regs.h
new file mode 100644
index 0000000..b9e24ed
--- /dev/null
+++ b/drivers/net/igbvf/regs.h
@@ -0,0 +1,108 @@
+/*******************************************************************************
+
+ Intel(R) 82576 Virtual Function Linux driver
+ Copyright(c) 2009 Intel Corporation.
+
+ This program is free software; you can redistribute it and/or modify it
+ under the terms and conditions of the GNU General Public License,
+ version 2, as published by the Free Software Foundation.
+
+ This program is distributed in the hope it will be useful, but WITHOUT
+ ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for
+ more details.
+
+ You should have received a copy of the GNU General Public License along with
+ this program; if not, write to the Free Software Foundation, Inc.,
+ 51 Franklin St - Fifth Floor, Boston, MA 02110-1301 USA.
+
+ The full GNU General Public License is included in this distribution in
+ the file called "COPYING".
+
+ Contact Information:
+ e1000-devel Mailing List <[email protected]>
+ Intel Corporation, 5200 N.E. Elam Young Parkway, Hillsboro, OR 97124-6497
+
+*******************************************************************************/
+
+#ifndef _E1000_REGS_H_
+#define _E1000_REGS_H_
+
+#define E1000_CTRL 0x00000 /* Device Control - RW */
+#define E1000_STATUS 0x00008 /* Device Status - RO */
+#define E1000_ITR 0x000C4 /* Interrupt Throttling Rate - RW */
+#define E1000_EICR 0x01580 /* Ext. Interrupt Cause Read - R/clr */
+#define E1000_EITR(_n) (0x01680 + (0x4 * (_n)))
+#define E1000_EICS 0x01520 /* Ext. Interrupt Cause Set - W0 */
+#define E1000_EIMS 0x01524 /* Ext. Interrupt Mask Set/Read - RW */
+#define E1000_EIMC 0x01528 /* Ext. Interrupt Mask Clear - WO */
+#define E1000_EIAC 0x0152C /* Ext. Interrupt Auto Clear - RW */
+#define E1000_EIAM 0x01530 /* Ext. Interrupt Ack Auto Clear Mask - RW */
+#define E1000_IVAR0 0x01700 /* Interrupt Vector Allocation (array) - RW */
+#define E1000_IVAR_MISC 0x01740 /* IVAR for "other" causes - RW */
+/*
+ * Convenience macros
+ *
+ * Note: "_n" is the queue number of the register to be written to.
+ *
+ * Example usage:
+ * E1000_RDBAL_REG(current_rx_queue)
+ */
+#define E1000_RDBAL(_n) ((_n) < 4 ? (0x02800 + ((_n) * 0x100)) : \
+ (0x0C000 + ((_n) * 0x40)))
+#define E1000_RDBAH(_n) ((_n) < 4 ? (0x02804 + ((_n) * 0x100)) : \
+ (0x0C004 + ((_n) * 0x40)))
+#define E1000_RDLEN(_n) ((_n) < 4 ? (0x02808 + ((_n) * 0x100)) : \
+ (0x0C008 + ((_n) * 0x40)))
+#define E1000_SRRCTL(_n) ((_n) < 4 ? (0x0280C + ((_n) * 0x100)) : \
+ (0x0C00C + ((_n) * 0x40)))
+#define E1000_RDH(_n) ((_n) < 4 ? (0x02810 + ((_n) * 0x100)) : \
+ (0x0C010 + ((_n) * 0x40)))
+#define E1000_RDT(_n) ((_n) < 4 ? (0x02818 + ((_n) * 0x100)) : \
+ (0x0C018 + ((_n) * 0x40)))
+#define E1000_RXDCTL(_n) ((_n) < 4 ? (0x02828 + ((_n) * 0x100)) : \
+ (0x0C028 + ((_n) * 0x40)))
+#define E1000_TDBAL(_n) ((_n) < 4 ? (0x03800 + ((_n) * 0x100)) : \
+ (0x0E000 + ((_n) * 0x40)))
+#define E1000_TDBAH(_n) ((_n) < 4 ? (0x03804 + ((_n) * 0x100)) : \
+ (0x0E004 + ((_n) * 0x40)))
+#define E1000_TDLEN(_n) ((_n) < 4 ? (0x03808 + ((_n) * 0x100)) : \
+ (0x0E008 + ((_n) * 0x40)))
+#define E1000_TDH(_n) ((_n) < 4 ? (0x03810 + ((_n) * 0x100)) : \
+ (0x0E010 + ((_n) * 0x40)))
+#define E1000_TDT(_n) ((_n) < 4 ? (0x03818 + ((_n) * 0x100)) : \
+ (0x0E018 + ((_n) * 0x40)))
+#define E1000_TXDCTL(_n) ((_n) < 4 ? (0x03828 + ((_n) * 0x100)) : \
+ (0x0E028 + ((_n) * 0x40)))
+#define E1000_DCA_TXCTRL(_n) (0x03814 + (_n << 8))
+#define E1000_DCA_RXCTRL(_n) (0x02814 + (_n << 8))
+#define E1000_RAL(_i) (((_i) <= 15) ? (0x05400 + ((_i) * 8)) : \
+ (0x054E0 + ((_i - 16) * 8)))
+#define E1000_RAH(_i) (((_i) <= 15) ? (0x05404 + ((_i) * 8)) : \
+ (0x054E4 + ((_i - 16) * 8)))
+
+/* Statistics registers */
+#define E1000_VFGPRC 0x00F10
+#define E1000_VFGORC 0x00F18
+#define E1000_VFMPRC 0x00F3C
+#define E1000_VFGPTC 0x00F14
+#define E1000_VFGOTC 0x00F34
+#define E1000_VFGOTLBC 0x00F50
+#define E1000_VFGPTLBC 0x00F44
+#define E1000_VFGORLBC 0x00F48
+#define E1000_VFGPRLBC 0x00F40
+
+/* These act per VF so an array friendly macro is used */
+#define E1000_V2PMAILBOX(_n) (0x00C40 + (4 * (_n)))
+#define E1000_VMBMEM(_n) (0x00800 + (64 * (_n)))
+
+/* Define macros for handling registers */
+#define er32(reg) readl(hw->hw_addr + E1000_##reg)
+#define ew32(reg, val) writel((val), hw->hw_addr + E1000_##reg)
+#define array_er32(reg, offset) \
+ readl(hw->hw_addr + E1000_##reg + (offset << 2))
+#define array_ew32(reg, offset, val) \
+ writel((val), hw->hw_addr + E1000_##reg + (offset << 2))
+#define e1e_flush() er32(STATUS)
+
+#endif
diff --git a/drivers/net/igbvf/vf.c b/drivers/net/igbvf/vf.c
new file mode 100644
index 0000000..aa246c9
--- /dev/null
+++ b/drivers/net/igbvf/vf.c
@@ -0,0 +1,398 @@
+/*******************************************************************************
+
+ Intel(R) 82576 Virtual Function Linux driver
+ Copyright(c) 2009 Intel Corporation.
+
+ This program is free software; you can redistribute it and/or modify it
+ under the terms and conditions of the GNU General Public License,
+ version 2, as published by the Free Software Foundation.
+
+ This program is distributed in the hope it will be useful, but WITHOUT
+ ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for
+ more details.
+
+ You should have received a copy of the GNU General Public License along with
+ this program; if not, write to the Free Software Foundation, Inc.,
+ 51 Franklin St - Fifth Floor, Boston, MA 02110-1301 USA.
+
+ The full GNU General Public License is included in this distribution in
+ the file called "COPYING".
+
+ Contact Information:
+ e1000-devel Mailing List <[email protected]>
+ Intel Corporation, 5200 N.E. Elam Young Parkway, Hillsboro, OR 97124-6497
+
+*******************************************************************************/
+
+
+#include "vf.h"
+
+static s32 e1000_check_for_link_vf(struct e1000_hw *hw);
+static s32 e1000_get_link_up_info_vf(struct e1000_hw *hw, u16 *speed,
+ u16 *duplex);
+static s32 e1000_init_hw_vf(struct e1000_hw *hw);
+static s32 e1000_reset_hw_vf(struct e1000_hw *hw);
+
+static void e1000_update_mc_addr_list_vf(struct e1000_hw *hw, u8 *,
+ u32, u32, u32);
+static void e1000_rar_set_vf(struct e1000_hw *, u8 *, u32);
+static s32 e1000_read_mac_addr_vf(struct e1000_hw *);
+static s32 e1000_set_vfta_vf(struct e1000_hw *, u16, bool);
+
+/**
+ * e1000_init_mac_params_vf - Inits MAC params
+ * @hw: pointer to the HW structure
+ **/
+s32 e1000_init_mac_params_vf(struct e1000_hw *hw)
+{
+ struct e1000_mac_info *mac = &hw->mac;
+
+ /* VF's have no MTA Registers - PF feature only */
+ mac->mta_reg_count = 128;
+ /* VF's have no access to RAR entries */
+ mac->rar_entry_count = 1;
+
+ /* Function pointers */
+ /* reset */
+ mac->ops.reset_hw = e1000_reset_hw_vf;
+ /* hw initialization */
+ mac->ops.init_hw = e1000_init_hw_vf;
+ /* check for link */
+ mac->ops.check_for_link = e1000_check_for_link_vf;
+ /* link info */
+ mac->ops.get_link_up_info = e1000_get_link_up_info_vf;
+ /* multicast address update */
+ mac->ops.update_mc_addr_list = e1000_update_mc_addr_list_vf;
+ /* set mac address */
+ mac->ops.rar_set = e1000_rar_set_vf;
+ /* read mac address */
+ mac->ops.read_mac_addr = e1000_read_mac_addr_vf;
+ /* set vlan filter table array */
+ mac->ops.set_vfta = e1000_set_vfta_vf;
+
+ return E1000_SUCCESS;
+}
+
+/**
+ * e1000_init_function_pointers_vf - Inits function pointers
+ * @hw: pointer to the HW structure
+ **/
+void e1000_init_function_pointers_vf(struct e1000_hw *hw)
+{
+ hw->mac.ops.init_params = e1000_init_mac_params_vf;
+ hw->mbx.ops.init_params = e1000_init_mbx_params_vf;
+}
+
+/**
+ * e1000_get_link_up_info_vf - Gets link info.
+ * @hw: pointer to the HW structure
+ * @speed: pointer to 16 bit value to store link speed.
+ * @duplex: pointer to 16 bit value to store duplex.
+ *
+ * Since we cannot read the PHY and get accurate link info, we must rely upon
+ * the status register's data which is often stale and inaccurate.
+ **/
+static s32 e1000_get_link_up_info_vf(struct e1000_hw *hw, u16 *speed,
+ u16 *duplex)
+{
+ s32 status;
+
+ status = er32(STATUS);
+ if (status & E1000_STATUS_SPEED_1000)
+ *speed = SPEED_1000;
+ else if (status & E1000_STATUS_SPEED_100)
+ *speed = SPEED_100;
+ else
+ *speed = SPEED_10;
+
+ if (status & E1000_STATUS_FD)
+ *duplex = FULL_DUPLEX;
+ else
+ *duplex = HALF_DUPLEX;
+
+ return E1000_SUCCESS;
+}
+
+/**
+ * e1000_reset_hw_vf - Resets the HW
+ * @hw: pointer to the HW structure
+ *
+ * VF's provide a function level reset. This is done using bit 26 of ctrl_reg.
+ * This is all the reset we can perform on a VF.
+ **/
+static s32 e1000_reset_hw_vf(struct e1000_hw *hw)
+{
+ struct e1000_mbx_info *mbx = &hw->mbx;
+ u32 timeout = E1000_VF_INIT_TIMEOUT;
+ u32 ret_val = -E1000_ERR_MAC_INIT;
+ u32 msgbuf[3];
+ u8 *addr = (u8 *)(&msgbuf[1]);
+ u32 ctrl;
+
+ /* assert vf queue/interrupt reset */
+ ctrl = er32(CTRL);
+ ew32(CTRL, ctrl | E1000_CTRL_RST);
+
+ /* we cannot initialize while the RSTI / RSTD bits are asserted */
+ while (!mbx->ops.check_for_rst(hw) && timeout) {
+ timeout--;
+ udelay(5);
+ }
+
+ if (timeout) {
+ /* mailbox timeout can now become active */
+ mbx->timeout = E1000_VF_MBX_INIT_TIMEOUT;
+
+ /* notify pf of vf reset completion */
+ msgbuf[0] = E1000_VF_RESET;
+ mbx->ops.write_posted(hw, msgbuf, 1);
+
+ msleep(10);
+
+ /* set our "perm_addr" based on info provided by PF */
+ ret_val = mbx->ops.read_posted(hw, msgbuf, 3);
+ if (!ret_val) {
+ if (msgbuf[0] == (E1000_VF_RESET | E1000_VT_MSGTYPE_ACK))
+ memcpy(hw->mac.perm_addr, addr, 6);
+ else
+ ret_val = -E1000_ERR_MAC_INIT;
+ }
+ }
+
+ return ret_val;
+}
+
+/**
+ * e1000_init_hw_vf - Inits the HW
+ * @hw: pointer to the HW structure
+ *
+ * Not much to do here except clear the PF Reset indication if there is one.
+ **/
+static s32 e1000_init_hw_vf(struct e1000_hw *hw)
+{
+ /* attempt to set and restore our mac address */
+ e1000_rar_set_vf(hw, hw->mac.addr, 0);
+
+ return E1000_SUCCESS;
+}
+
+/**
+ * e1000_hash_mc_addr_vf - Generate a multicast hash value
+ * @hw: pointer to the HW structure
+ * @mc_addr: pointer to a multicast address
+ *
+ * Generates a multicast address hash value which is used to determine
+ * the multicast filter table array address and new table value. See
+ * e1000_mta_set_generic()
+ **/
+static u32 e1000_hash_mc_addr_vf(struct e1000_hw *hw, u8 *mc_addr)
+{
+ u32 hash_value, hash_mask;
+ u8 bit_shift = 0;
+
+ /* Register count multiplied by bits per register */
+ hash_mask = (hw->mac.mta_reg_count * 32) - 1;
+
+ /*
+ * The bit_shift is the number of left-shifts
+ * where 0xFF would still fall within the hash mask.
+ */
+ while (hash_mask >> bit_shift != 0xFF)
+ bit_shift++;
+
+ hash_value = hash_mask & (((mc_addr[4] >> (8 - bit_shift)) |
+ (((u16) mc_addr[5]) << bit_shift)));
+
+ return hash_value;
+}
+
+/**
+ * e1000_update_mc_addr_list_vf - Update Multicast addresses
+ * @hw: pointer to the HW structure
+ * @mc_addr_list: array of multicast addresses to program
+ * @mc_addr_count: number of multicast addresses to program
+ * @rar_used_count: the first RAR register free to program
+ * @rar_count: total number of supported Receive Address Registers
+ *
+ * Updates the Receive Address Registers and Multicast Table Array.
+ * The caller must have a packed mc_addr_list of multicast addresses.
+ * The parameter rar_count will usually be hw->mac.rar_entry_count
+ * unless there are workarounds that change this.
+ **/
+void e1000_update_mc_addr_list_vf(struct e1000_hw *hw,
+ u8 *mc_addr_list, u32 mc_addr_count,
+ u32 rar_used_count, u32 rar_count)
+{
+ struct e1000_mbx_info *mbx = &hw->mbx;
+ u32 msgbuf[E1000_VFMAILBOX_SIZE];
+ u16 *hash_list = (u16 *)&msgbuf[1];
+ u32 hash_value;
+ u32 cnt, i;
+
+ /* Each entry in the list uses 1 16 bit word. We have 30
+ * 16 bit words available in our HW msg buffer (minus 1 for the
+ * msg type). That's 30 hash values if we pack 'em right. If
+ * there are more than 30 MC addresses to add then punt the
+ * extras for now and then add code to handle more than 30 later.
+ * It would be unusual for a server to request that many multi-cast
+ * addresses except for in large enterprise network environments.
+ */
+
+ cnt = (mc_addr_count > 30) ? 30 : mc_addr_count;
+ msgbuf[0] = E1000_VF_SET_MULTICAST;
+ msgbuf[0] |= cnt << E1000_VT_MSGINFO_SHIFT;
+
+ for (i = 0; i < cnt; i++) {
+ hash_value = e1000_hash_mc_addr_vf(hw, mc_addr_list);
+ hash_list[i] = hash_value & 0x0FFFF;
+ mc_addr_list += ETH_ADDR_LEN;
+ }
+
+ mbx->ops.write_posted(hw, msgbuf, E1000_VFMAILBOX_SIZE);
+}
+
+/**
+ * e1000_set_vfta_vf - Set/Unset vlan filter table address
+ * @hw: pointer to the HW structure
+ * @vid: determines the vfta register and bit to set/unset
+ * @set: if true then set bit, else clear bit
+ **/
+static s32 e1000_set_vfta_vf(struct e1000_hw *hw, u16 vid, bool set)
+{
+ struct e1000_mbx_info *mbx = &hw->mbx;
+ u32 msgbuf[2];
+ s32 err;
+
+ msgbuf[0] = E1000_VF_SET_VLAN;
+ msgbuf[1] = vid;
+ /* Setting the 8 bit field MSG INFO to true indicates "add" */
+ if (set)
+ msgbuf[0] |= 1 << E1000_VT_MSGINFO_SHIFT;
+
+ mbx->ops.write_posted(hw, msgbuf, 2);
+
+ err = mbx->ops.read_posted(hw, msgbuf, 2);
+
+ /* if nacked the vlan was rejected */
+ if (!err && (msgbuf[0] == (E1000_VF_SET_VLAN | E1000_VT_MSGTYPE_NACK)))
+ err = -E1000_ERR_MAC_INIT;
+
+ return err;
+}
+
+/** e1000_rlpml_set_vf - Set the maximum receive packet length
+ * @hw: pointer to the HW structure
+ * @max_size: value to assign to max frame size
+ **/
+void e1000_rlpml_set_vf(struct e1000_hw *hw, u16 max_size)
+{
+ struct e1000_mbx_info *mbx = &hw->mbx;
+ u32 msgbuf[2];
+
+ msgbuf[0] = E1000_VF_SET_LPE;
+ msgbuf[1] = max_size;
+
+ mbx->ops.write_posted(hw, msgbuf, 2);
+}
+
+/**
+ * e1000_rar_set_vf - set device MAC address
+ * @hw: pointer to the HW structure
+ * @addr: pointer to the receive address
+ * @index receive address array register
+ **/
+static void e1000_rar_set_vf(struct e1000_hw *hw, u8 * addr, u32 index)
+{
+ struct e1000_mbx_info *mbx = &hw->mbx;
+ u32 msgbuf[3];
+ u8 *msg_addr = (u8 *)(&msgbuf[1]);
+ s32 ret_val;
+
+ memset(msgbuf, 0, 12);
+ msgbuf[0] = E1000_VF_SET_MAC_ADDR;
+ memcpy(msg_addr, addr, 6);
+ ret_val = mbx->ops.write_posted(hw, msgbuf, 3);
+
+ if (!ret_val)
+ ret_val = mbx->ops.read_posted(hw, msgbuf, 3);
+
+ /* if nacked the address was rejected, use "perm_addr" */
+ if (!ret_val &&
+ (msgbuf[0] == (E1000_VF_SET_MAC_ADDR | E1000_VT_MSGTYPE_NACK)))
+ e1000_read_mac_addr_vf(hw);
+}
+
+/**
+ * e1000_read_mac_addr_vf - Read device MAC address
+ * @hw: pointer to the HW structure
+ **/
+static s32 e1000_read_mac_addr_vf(struct e1000_hw *hw)
+{
+ int i;
+
+ for (i = 0; i < ETH_ADDR_LEN; i++)
+ hw->mac.addr[i] = hw->mac.perm_addr[i];
+
+ return E1000_SUCCESS;
+}
+
+/**
+ * e1000_check_for_link_vf - Check for link for a virtual interface
+ * @hw: pointer to the HW structure
+ *
+ * Checks to see if the underlying PF is still talking to the VF and
+ * if it is then it reports the link state to the hardware, otherwise
+ * it reports link down and returns an error.
+ **/
+static s32 e1000_check_for_link_vf(struct e1000_hw *hw)
+{
+ struct e1000_mbx_info *mbx = &hw->mbx;
+ struct e1000_mac_info *mac = &hw->mac;
+ s32 ret_val = E1000_SUCCESS;
+ u32 in_msg = 0;
+
+ /*
+ * We only want to run this if there has been a rst asserted.
+ * in this case that could mean a link change, device reset,
+ * or a virtual function reset
+ */
+
+ /* If we were hit with a reset drop the link */
+ if (!mbx->ops.check_for_rst(hw))
+ mac->get_link_status = true;
+
+ if (!mac->get_link_status)
+ goto out;
+
+ /* if link status is down no point in checking to see if pf is up */
+ if (!(er32(STATUS) & E1000_STATUS_LU))
+ goto out;
+
+ /* if the read failed it could just be a mailbox collision, best wait
+ * until we are called again and don't report an error */
+ if (mbx->ops.read(hw, &in_msg, 1))
+ goto out;
+
+ /* if incoming message isn't clear to send we are waiting on response */
+ if (!(in_msg & E1000_VT_MSGTYPE_CTS)) {
+ /* message is not CTS and is NACK we must have lost CTS status */
+ if (in_msg & E1000_VT_MSGTYPE_NACK)
+ ret_val = -E1000_ERR_MAC_INIT;
+ goto out;
+ }
+
+ /* the pf is talking, if we timed out in the past we reinit */
+ if (!mbx->timeout) {
+ ret_val = -E1000_ERR_MAC_INIT;
+ goto out;
+ }
+
+ /* if we passed all the tests above then the link is up and we no
+ * longer need to check for link */
+ mac->get_link_status = false;
+
+out:
+ return ret_val;
+}
+
diff --git a/drivers/net/igbvf/vf.h b/drivers/net/igbvf/vf.h
new file mode 100644
index 0000000..ec07228
--- /dev/null
+++ b/drivers/net/igbvf/vf.h
@@ -0,0 +1,265 @@
+/*******************************************************************************
+
+ Intel(R) 82576 Virtual Function Linux driver
+ Copyright(c) 2009 Intel Corporation.
+
+ This program is free software; you can redistribute it and/or modify it
+ under the terms and conditions of the GNU General Public License,
+ version 2, as published by the Free Software Foundation.
+
+ This program is distributed in the hope it will be useful, but WITHOUT
+ ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for
+ more details.
+
+ You should have received a copy of the GNU General Public License along with
+ this program; if not, write to the Free Software Foundation, Inc.,
+ 51 Franklin St - Fifth Floor, Boston, MA 02110-1301 USA.
+
+ The full GNU General Public License is included in this distribution in
+ the file called "COPYING".
+
+ Contact Information:
+ e1000-devel Mailing List <[email protected]>
+ Intel Corporation, 5200 N.E. Elam Young Parkway, Hillsboro, OR 97124-6497
+
+*******************************************************************************/
+
+#ifndef _E1000_VF_H_
+#define _E1000_VF_H_
+
+#include <linux/pci.h>
+#include <linux/delay.h>
+#include <linux/interrupt.h>
+#include <linux/if_ether.h>
+
+#include "regs.h"
+#include "defines.h"
+
+struct e1000_hw;
+
+#define E1000_DEV_ID_82576_VF 0x10CA
+#define E1000_REVISION_0 0
+#define E1000_REVISION_1 1
+#define E1000_REVISION_2 2
+#define E1000_REVISION_3 3
+#define E1000_REVISION_4 4
+
+#define E1000_FUNC_0 0
+#define E1000_FUNC_1 1
+
+/*
+ * Receive Address Register Count
+ * Number of high/low register pairs in the RAR. The RAR (Receive Address
+ * Registers) holds the directed and multicast addresses that we monitor.
+ * These entries are also used for MAC-based filtering.
+ */
+#define E1000_RAR_ENTRIES_VF 1
+
+/* Receive Descriptor - Advanced */
+union e1000_adv_rx_desc {
+ struct {
+ u64 pkt_addr; /* Packet buffer address */
+ u64 hdr_addr; /* Header buffer address */
+ } read;
+ struct {
+ struct {
+ union {
+ u32 data;
+ struct {
+ u16 pkt_info; /* RSS/Packet type */
+ u16 hdr_info; /* Split Header,
+ * hdr buffer length */
+ } hs_rss;
+ } lo_dword;
+ union {
+ u32 rss; /* RSS Hash */
+ struct {
+ u16 ip_id; /* IP id */
+ u16 csum; /* Packet Checksum */
+ } csum_ip;
+ } hi_dword;
+ } lower;
+ struct {
+ u32 status_error; /* ext status/error */
+ u16 length; /* Packet length */
+ u16 vlan; /* VLAN tag */
+ } upper;
+ } wb; /* writeback */
+};
+
+#define E1000_RXDADV_HDRBUFLEN_MASK 0x7FE0
+#define E1000_RXDADV_HDRBUFLEN_SHIFT 5
+
+/* Transmit Descriptor - Advanced */
+union e1000_adv_tx_desc {
+ struct {
+ u64 buffer_addr; /* Address of descriptor's data buf */
+ u32 cmd_type_len;
+ u32 olinfo_status;
+ } read;
+ struct {
+ u64 rsvd; /* Reserved */
+ u32 nxtseq_seed;
+ u32 status;
+ } wb;
+};
+
+/* Adv Transmit Descriptor Config Masks */
+#define E1000_ADVTXD_DTYP_CTXT 0x00200000 /* Advanced Context Descriptor */
+#define E1000_ADVTXD_DTYP_DATA 0x00300000 /* Advanced Data Descriptor */
+#define E1000_ADVTXD_DCMD_EOP 0x01000000 /* End of Packet */
+#define E1000_ADVTXD_DCMD_IFCS 0x02000000 /* Insert FCS (Ethernet CRC) */
+#define E1000_ADVTXD_DCMD_RS 0x08000000 /* Report Status */
+#define E1000_ADVTXD_DCMD_DEXT 0x20000000 /* Descriptor extension (1=Adv) */
+#define E1000_ADVTXD_DCMD_VLE 0x40000000 /* VLAN pkt enable */
+#define E1000_ADVTXD_DCMD_TSE 0x80000000 /* TCP Seg enable */
+#define E1000_ADVTXD_PAYLEN_SHIFT 14 /* Adv desc PAYLEN shift */
+
+/* Context descriptors */
+struct e1000_adv_tx_context_desc {
+ u32 vlan_macip_lens;
+ u32 seqnum_seed;
+ u32 type_tucmd_mlhl;
+ u32 mss_l4len_idx;
+};
+
+#define E1000_ADVTXD_MACLEN_SHIFT 9 /* Adv ctxt desc mac len shift */
+#define E1000_ADVTXD_TUCMD_IPV4 0x00000400 /* IP Packet Type: 1=IPv4 */
+#define E1000_ADVTXD_TUCMD_L4T_TCP 0x00000800 /* L4 Packet TYPE of TCP */
+#define E1000_ADVTXD_L4LEN_SHIFT 8 /* Adv ctxt L4LEN shift */
+#define E1000_ADVTXD_MSS_SHIFT 16 /* Adv ctxt MSS shift */
+
+enum e1000_mac_type {
+ e1000_undefined = 0,
+ e1000_vfadapt,
+ e1000_num_macs /* List is 1-based, so subtract 1 for true count. */
+};
+
+struct e1000_vf_stats {
+ u64 base_gprc;
+ u64 base_gptc;
+ u64 base_gorc;
+ u64 base_gotc;
+ u64 base_mprc;
+ u64 base_gotlbc;
+ u64 base_gptlbc;
+ u64 base_gorlbc;
+ u64 base_gprlbc;
+
+ u32 last_gprc;
+ u32 last_gptc;
+ u32 last_gorc;
+ u32 last_gotc;
+ u32 last_mprc;
+ u32 last_gotlbc;
+ u32 last_gptlbc;
+ u32 last_gorlbc;
+ u32 last_gprlbc;
+
+ u64 gprc;
+ u64 gptc;
+ u64 gorc;
+ u64 gotc;
+ u64 mprc;
+ u64 gotlbc;
+ u64 gptlbc;
+ u64 gorlbc;
+ u64 gprlbc;
+};
+
+#include "mbx.h"
+
+struct e1000_mac_operations {
+ /* Function pointers for the MAC. */
+ s32 (*init_params)(struct e1000_hw *);
+ s32 (*check_for_link)(struct e1000_hw *);
+ void (*clear_vfta)(struct e1000_hw *);
+ s32 (*get_bus_info)(struct e1000_hw *);
+ s32 (*get_link_up_info)(struct e1000_hw *, u16 *, u16 *);
+ void (*update_mc_addr_list)(struct e1000_hw *, u8 *, u32, u32, u32);
+ s32 (*reset_hw)(struct e1000_hw *);
+ s32 (*init_hw)(struct e1000_hw *);
+ s32 (*setup_link)(struct e1000_hw *);
+ void (*write_vfta)(struct e1000_hw *, u32, u32);
+ void (*mta_set)(struct e1000_hw *, u32);
+ void (*rar_set)(struct e1000_hw *, u8*, u32);
+ s32 (*read_mac_addr)(struct e1000_hw *);
+ s32 (*set_vfta)(struct e1000_hw *, u16, bool);
+};
+
+struct e1000_mac_info {
+ struct e1000_mac_operations ops;
+ u8 addr[6];
+ u8 perm_addr[6];
+
+ enum e1000_mac_type type;
+
+ u16 mta_reg_count;
+ u16 rar_entry_count;
+
+ bool get_link_status;
+};
+
+struct e1000_mbx_operations {
+ s32 (*init_params)(struct e1000_hw *hw);
+ s32 (*read)(struct e1000_hw *, u32 *, u16);
+ s32 (*write)(struct e1000_hw *, u32 *, u16);
+ s32 (*read_posted)(struct e1000_hw *, u32 *, u16);
+ s32 (*write_posted)(struct e1000_hw *, u32 *, u16);
+ s32 (*check_for_msg)(struct e1000_hw *);
+ s32 (*check_for_ack)(struct e1000_hw *);
+ s32 (*check_for_rst)(struct e1000_hw *);
+};
+
+struct e1000_mbx_stats {
+ u32 msgs_tx;
+ u32 msgs_rx;
+
+ u32 acks;
+ u32 reqs;
+ u32 rsts;
+};
+
+struct e1000_mbx_info {
+ struct e1000_mbx_operations ops;
+ struct e1000_mbx_stats stats;
+ u32 timeout;
+ u32 usec_delay;
+ u16 size;
+};
+
+struct e1000_dev_spec_vf {
+ u32 vf_number;
+ u32 v2p_mailbox;
+};
+
+struct e1000_hw {
+ void *back;
+
+ u8 __iomem *hw_addr;
+ u8 __iomem *flash_address;
+ unsigned long io_base;
+
+ struct e1000_mac_info mac;
+ struct e1000_mbx_info mbx;
+
+ union {
+ struct e1000_dev_spec_vf vf;
+ } dev_spec;
+
+ u16 device_id;
+ u16 subsystem_vendor_id;
+ u16 subsystem_device_id;
+ u16 vendor_id;
+
+ u8 revision_id;
+};
+
+/* These functions must be implemented by drivers */
+void e1000_rlpml_set_vf(struct e1000_hw *, u16);
+void e1000_init_function_pointers_vf(struct e1000_hw *hw);
+s32 e1000_init_mac_params_vf(struct e1000_hw *hw);
+
+
+#endif /* _E1000_VF_H_ */

2009-03-11 02:11:11

by Jeff Kirsher

Subject: [net-next PATCH 2/2] igb: add PF to pool if adding a VLVF register value and the VFTA bit is

From: Alexander Duyck <[email protected]>

already set.

This patch addresses the unlikely situation that the PF adds a vlan
entry when the vlvf is full, and a vf later adds the vlan to the vlvf.

Signed-off-by: Alexander Duyck <[email protected]>
Signed-off-by: Jeff Kirsher <[email protected]>
---

drivers/net/igb/e1000_mac.c | 21 ++++++++++++++-------
drivers/net/igb/e1000_mac.h | 2 +-
drivers/net/igb/igb_main.c | 9 +++++++--
3 files changed, 22 insertions(+), 10 deletions(-)

diff --git a/drivers/net/igb/e1000_mac.c b/drivers/net/igb/e1000_mac.c
index 2804db0..f11592f 100644
--- a/drivers/net/igb/e1000_mac.c
+++ b/drivers/net/igb/e1000_mac.c
@@ -126,19 +126,26 @@ void igb_write_vfta(struct e1000_hw *hw, u32 offset, u32 value)
* Sets or clears a bit in the VLAN filter table array based on VLAN id
* and if we are adding or removing the filter
**/
-void igb_vfta_set(struct e1000_hw *hw, u32 vid, bool add)
+s32 igb_vfta_set(struct e1000_hw *hw, u32 vid, bool add)
{
u32 index = (vid >> E1000_VFTA_ENTRY_SHIFT) & E1000_VFTA_ENTRY_MASK;
u32 mask = 1 < (vid & E1000_VFTA_ENTRY_BIT_SHIFT_MASK);
- u32 vfta;
+ u32 vfta = array_rd32(E1000_VFTA, index);
+ s32 ret_val = 0;

- vfta = array_rd32(E1000_VFTA, index);
- if (add)
- vfta |= mask;
- else
- vfta &= ~mask;
+ /* bit was set/cleared before we started */
+ if ((!!(vfta & mask)) == add) {
+ ret_val = -E1000_ERR_CONFIG;
+ } else {
+ if (add)
+ vfta |= mask;
+ else
+ vfta &= ~mask;
+ }

igb_write_vfta(hw, index, vfta);
+
+ return ret_val;
}

/**
diff --git a/drivers/net/igb/e1000_mac.h b/drivers/net/igb/e1000_mac.h
index eccc353..a34de52 100644
--- a/drivers/net/igb/e1000_mac.h
+++ b/drivers/net/igb/e1000_mac.h
@@ -58,7 +58,7 @@ s32 igb_write_8bit_ctrl_reg(struct e1000_hw *hw, u32 reg,

void igb_clear_hw_cntrs_base(struct e1000_hw *hw);
void igb_clear_vfta(struct e1000_hw *hw);
-void igb_vfta_set(struct e1000_hw *hw, u32 vid, bool add);
+s32 igb_vfta_set(struct e1000_hw *hw, u32 vid, bool add);
void igb_config_collision_dist(struct e1000_hw *hw);
void igb_mta_set(struct e1000_hw *hw, u32 hash_value);
void igb_put_hw_semaphore(struct e1000_hw *hw);
diff --git a/drivers/net/igb/igb_main.c b/drivers/net/igb/igb_main.c
index 78558f8..579435d 100644
--- a/drivers/net/igb/igb_main.c
+++ b/drivers/net/igb/igb_main.c
@@ -3894,10 +3894,15 @@ static s32 igb_vlvf_set(struct igb_adapter *adapter, u32 vid, bool add, u32 vf)

/* if !enabled we need to set this up in vfta */
if (!(reg & E1000_VLVF_VLANID_ENABLE)) {
- /* add VID to filter table */
- igb_vfta_set(hw, vid, true);
+ /* add VID to filter table, if bit already set
+ * PF must have added it outside of table */
+ if (igb_vfta_set(hw, vid, true))
+ reg |= 1 << (E1000_VLVF_POOLSEL_SHIFT +
+ adapter->vfs_allocated_count);
reg |= E1000_VLVF_VLANID_ENABLE;
}
+ reg &= ~E1000_VLVF_VLANID_MASK;
+ reg |= vid;

wr32(E1000_VLVF(i), reg);
return 0;

2009-03-11 05:25:21

by Andrew Morton

Subject: Re: [net-next PATCH 1/2] igbvf: add new driver to support 82576 virtual functions

On Tue, 10 Mar 2009 19:09:28 -0700 Jeff Kirsher <[email protected]> wrote:

> From: Alexander Duyck <[email protected]>
>
> This adds an igbvf driver to handle virtual functions provided
> by the igb driver.

The drive-by reader is now wondering what a "virtual function" is.

>
> ...
>
> +#define IGBVF_STAT(m, b) sizeof(((struct igbvf_adapter *)0)->m), \
> + offsetof(struct igbvf_adapter, m), \
> + offsetof(struct igbvf_adapter, b)
>
> +static const struct igbvf_stats igbvf_gstrings_stats[] = {
> + { "rx_packets", IGBVF_STAT(stats.gprc, stats.base_gprc) },
> + { "tx_packets", IGBVF_STAT(stats.gptc, stats.base_gptc) },
> + { "rx_bytes", IGBVF_STAT(stats.gorc, stats.base_gorc) },
> + { "tx_bytes", IGBVF_STAT(stats.gotc, stats.base_gotc) },
> + { "multicast", IGBVF_STAT(stats.mprc, stats.base_mprc) },
> + { "lbrx_bytes", IGBVF_STAT(stats.gorlbc, stats.base_gorlbc) },
> + { "lbrx_packets", IGBVF_STAT(stats.gprlbc, stats.base_gprlbc) },
> + { "tx_restart_queue", IGBVF_STAT(restart_queue, zero_base) },
> + { "rx_long_byte_count", IGBVF_STAT(stats.gorc, stats.base_gorc) },
> + { "rx_csum_offload_good", IGBVF_STAT(hw_csum_good, zero_base) },
> + { "rx_csum_offload_errors", IGBVF_STAT(hw_csum_err, zero_base) },
> + { "rx_header_split", IGBVF_STAT(rx_hdr_split, zero_base) },
> + { "alloc_rx_buff_failed", IGBVF_STAT(alloc_rx_buff_failed, zero_base) },
> +};

<stares at it for a while>

It would be clearer if `m' and `b' were (much) more meaningful identifiers.

> +#define IGBVF_GLOBAL_STATS_LEN \
> + (sizeof(igbvf_gstrings_stats) / sizeof(struct igbvf_stats))

This is ARRAY_SIZE().

> +#define IGBVF_STATS_LEN (IGBVF_GLOBAL_STATS_LEN)

Why does this need to exist?

>
> ...
>
> +static int igbvf_set_ringparam(struct net_device *netdev,
> + struct ethtool_ringparam *ring)
> +{
> + struct igbvf_adapter *adapter = netdev_priv(netdev);
> + struct igbvf_ring *tx_ring, *tx_old;
> + struct igbvf_ring *rx_ring, *rx_old;
> + int err;
> +
> + if ((ring->rx_mini_pending) || (ring->rx_jumbo_pending))
> + return -EINVAL;
> +
> + while (test_and_set_bit(__IGBVF_RESETTING, &adapter->state))
> + msleep(1);

No timeout needed here? Interrupts might not be working, for example..

> + if (netif_running(adapter->netdev))
> + igbvf_down(adapter);
> +
> + tx_old = adapter->tx_ring;
> + rx_old = adapter->rx_ring;
> +
> + err = -ENOMEM;
> + tx_ring = kzalloc(sizeof(struct igbvf_ring), GFP_KERNEL);
> + if (!tx_ring)
> + goto err_alloc_tx;
> + /*
> + * use a memcpy to save any previously configured
> + * items like napi structs from having to be
> + * reinitialized
> + */
> + memcpy(tx_ring, tx_old, sizeof(struct igbvf_ring));
> +
> + rx_ring = kzalloc(sizeof(struct igbvf_ring), GFP_KERNEL);
> + if (!rx_ring)
> + goto err_alloc_rx;
> + memcpy(rx_ring, rx_old, sizeof(struct igbvf_ring));
> +
> + adapter->tx_ring = tx_ring;
> + adapter->rx_ring = rx_ring;
> +
> + rx_ring->count = max(ring->rx_pending, (u32)IGBVF_MIN_RXD);
> + rx_ring->count = min(rx_ring->count, (u32)(IGBVF_MAX_RXD));
> + rx_ring->count = ALIGN(rx_ring->count, REQ_RX_DESCRIPTOR_MULTIPLE);
> +
> + tx_ring->count = max(ring->tx_pending, (u32)IGBVF_MIN_TXD);
> + tx_ring->count = min(tx_ring->count, (u32)(IGBVF_MAX_TXD));
> + tx_ring->count = ALIGN(tx_ring->count, REQ_TX_DESCRIPTOR_MULTIPLE);
> +
> + if (netif_running(adapter->netdev)) {
> + /* Try to get new resources before deleting old */
> + err = igbvf_setup_rx_resources(adapter);
> + if (err)
> + goto err_setup_rx;
> + err = igbvf_setup_tx_resources(adapter);
> + if (err)
> + goto err_setup_tx;
> +
> + /*
> + * restore the old in order to free it,
> + * then add in the new
> + */
> + adapter->rx_ring = rx_old;
> + adapter->tx_ring = tx_old;
> + igbvf_free_rx_resources(adapter);
> + igbvf_free_tx_resources(adapter);
> + kfree(tx_old);
> + kfree(rx_old);

That's odd-looking. Why take a copy of rx_old and tx_old when we're
about to free them?

> + adapter->rx_ring = rx_ring;
> + adapter->tx_ring = tx_ring;
> + err = igbvf_up(adapter);
> + if (err)
> + goto err_setup;
> + }
> +
> + clear_bit(__IGBVF_RESETTING, &adapter->state);
> + return 0;
> +err_setup_tx:
> + igbvf_free_rx_resources(adapter);
> +err_setup_rx:
> + adapter->rx_ring = rx_old;
> + adapter->tx_ring = tx_old;
> + kfree(rx_ring);
> +err_alloc_rx:
> + kfree(tx_ring);
> +err_alloc_tx:
> + igbvf_up(adapter);
> +err_setup:
> + clear_bit(__IGBVF_RESETTING, &adapter->state);
> + return err;
> +}
> +
>
> ...
>
> +static void igbvf_diag_test(struct net_device *netdev,
> + struct ethtool_test *eth_test, u64 *data)
> +{
> + struct igbvf_adapter *adapter = netdev_priv(netdev);
> +
> + set_bit(__IGBVF_TESTING, &adapter->state);
> +
> + /*
> + * Link test performed before hardware reset so autoneg doesn't
> + * interfere with test result
> + */
> + if (igbvf_link_test(adapter, &data[0]))
> + eth_test->flags |= ETH_TEST_FL_FAILED;
> +
> + clear_bit(__IGBVF_TESTING, &adapter->state);
> + msleep_interruptible(4 * 1000);

If this process has signal_pending(), msleep_interruptible() becomes a
no-op. I expect the driver breaks?

> +}
> +
>
> ...
>
> +static void igbvf_get_ethtool_stats(struct net_device *netdev,
> + struct ethtool_stats *stats,
> + u64 *data)
> +{
> + struct igbvf_adapter *adapter = netdev_priv(netdev);
> + int i;
> +
> + igbvf_update_stats(adapter);
> + for (i = 0; i < IGBVF_GLOBAL_STATS_LEN; i++) {
> + char *p = (char *)adapter +
> + igbvf_gstrings_stats[i].stat_offset;
> + char *b = (char *)adapter +
> + igbvf_gstrings_stats[i].base_stat_offset;
> + data[i] = ((igbvf_gstrings_stats[i].sizeof_stat ==
> + sizeof(u64)) ? *(u64 *)p : *(u32 *)p) -
> + ((igbvf_gstrings_stats[i].sizeof_stat ==
> + sizeof(u64)) ? *(u64 *)b : *(u32 *)b);

yitch.

Are we using the C type system as well as possible here?

> + }
> +
> +}
> +
>
> ...
>
> +#define IGBVF_DESC_UNUSED(R) \
> + ((((R)->next_to_clean > (R)->next_to_use) ? 0 : (R)->count) + \
> + (R)->next_to_clean - (R)->next_to_use - 1)

my eyes.

This macro evaluates its argument multiple times and will misbehave (or
be slow) if passed an expreesion with side-effects.

Why not just write a plain old C function for this?

> +#define IGBVF_RX_DESC_ADV(R, i) \
> + (&(((union e1000_adv_rx_desc *)((R).desc))[i]))
> +#define IGBVF_TX_DESC_ADV(R, i) \
> + (&(((union e1000_adv_tx_desc *)((R).desc))[i]))
> +#define IGBVF_TX_CTXTDESC_ADV(R, i) \
> + (&(((struct e1000_adv_tx_context_desc *)((R).desc))[i]))

Maybe igbvf_ring.desc should have had type

union {
union e1000_adv_rx_desc;
union e1000_adv_tx_desc;
union e1000_adv_tx_context_desc;
} *

Or something.

>
> ...
>
> +/**
> + * igbvf_alloc_rx_buffers - Replace used receive buffers; packet split
> + * @rx_ring: address of ring structure to repopulate
> + * @cleaned_count: number of buffers to repopulate
> + **/
> +static void igbvf_alloc_rx_buffers(struct igbvf_ring *rx_ring,
> + int cleaned_count)
> +{
> + struct igbvf_adapter *adapter = rx_ring->adapter;
> + struct net_device *netdev = adapter->netdev;
> + struct pci_dev *pdev = adapter->pdev;
> + union e1000_adv_rx_desc *rx_desc;
> + struct igbvf_buffer *buffer_info;
> + struct sk_buff *skb;
> + unsigned int i;
> + int bufsz;
> +
> + i = rx_ring->next_to_use;
> + buffer_info = &rx_ring->buffer_info[i];
> +
> + if (adapter->rx_ps_hdr_size)
> + bufsz = adapter->rx_ps_hdr_size;
> + else
> + bufsz = adapter->rx_buffer_len;
> + bufsz += NET_IP_ALIGN;
> +
> + while (cleaned_count--) {
> + rx_desc = IGBVF_RX_DESC_ADV(*rx_ring, i);
> +
> + if (adapter->rx_ps_hdr_size && !buffer_info->page_dma) {
> + if (!buffer_info->page) {
> + buffer_info->page = alloc_page(GFP_ATOMIC);
> + if (!buffer_info->page) {
> + adapter->alloc_rx_buff_failed++;
> + goto no_buffers;
> + }
> + buffer_info->page_offset = 0;
> + } else {
> + buffer_info->page_offset ^= PAGE_SIZE / 2;
> + }
> + buffer_info->page_dma =
> + pci_map_page(pdev, buffer_info->page,
> + buffer_info->page_offset,
> + PAGE_SIZE / 2,
> + PCI_DMA_FROMDEVICE);
> + }
> +
> + if (!buffer_info->skb) {
> + skb = netdev_alloc_skb(netdev, bufsz);
> + if (!skb) {
> + adapter->alloc_rx_buff_failed++;
> + goto no_buffers;
> + }
> +
> + /* Make buffer alignment 2 beyond a 16 byte boundary
> + * this will result in a 16 byte aligned IP header after
> + * the 14 byte MAC header is removed
> + */
> + skb_reserve(skb, NET_IP_ALIGN);
> +
> + buffer_info->skb = skb;
> + buffer_info->dma = pci_map_single(pdev, skb->data,
> + bufsz,
> + PCI_DMA_FROMDEVICE);
> + }
> + /* Refresh the desc even if buffer_addrs didn't change because
> + * each write-back erases this info. */
> + if (adapter->rx_ps_hdr_size) {
> + rx_desc->read.pkt_addr =
> + cpu_to_le64(buffer_info->page_dma);
> + rx_desc->read.hdr_addr = cpu_to_le64(buffer_info->dma);
> + } else {
> + rx_desc->read.pkt_addr =
> + cpu_to_le64(buffer_info->dma);
> + rx_desc->read.hdr_addr = 0;
> + }
> +
> + i++;
> + if (i == rx_ring->count)
> + i = 0;
> + buffer_info = &rx_ring->buffer_info[i];
> + }
> +
> +no_buffers:
> + if (rx_ring->next_to_use != i) {
> + rx_ring->next_to_use = i;
> + if (i == 0)
> + i = (rx_ring->count - 1);
> + else
> + 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). */
> + wmb();
> + writel(i, adapter->hw.hw_addr + rx_ring->tail);

Does wmb() work reliably for masters other than CPUs? I lose track...

> + }
> +}
> +
>
> ...
>
> +/**
> + * igbvf_setup_tx_resources - allocate Tx resources (Descriptors)
> + * @adapter: board private structure
> + *
> + * Return 0 on success, negative on failure
> + **/
> +int igbvf_setup_tx_resources(struct igbvf_adapter *adapter)
> +{
> + struct igbvf_ring *tx_ring = adapter->tx_ring;
> + int err = -ENOMEM, size;
> +
> + size = sizeof(struct igbvf_buffer) * tx_ring->count;
> + tx_ring->buffer_info = vmalloc(size);

How large is this likely to be?

> + if (!tx_ring->buffer_info)
> + goto err;
> + memset(tx_ring->buffer_info, 0, size);
> +
> + /* round up to nearest 4K */
> + tx_ring->size = tx_ring->count * sizeof(union e1000_adv_tx_desc);
> + tx_ring->size = ALIGN(tx_ring->size, 4096);
> +
> + err = igbvf_alloc_ring_dma(adapter, tx_ring);
> + if (err)
> + goto err;
> +
> + tx_ring->next_to_use = 0;
> + tx_ring->next_to_clean = 0;
> + spin_lock_init(&adapter->tx_queue_lock);
> +
> + return 0;
> +err:
> + vfree(tx_ring->buffer_info);
> + dev_err(&adapter->pdev->dev,
> + "Unable to allocate memory for the transmit descriptor ring\n");
> + return err;
> +}
> +
>
> ...
>
> +void igbvf_set_interrupt_capability(struct igbvf_adapter *adapter)
> +{
> + int err = -ENOMEM;
> + int i;
> +
> + /* we allocate 3 vectors, 1 for tx, 1 for rx, one for pf messages */
> + adapter->msix_entries = kcalloc(3, sizeof(struct msix_entry),
> + GFP_KERNEL);
> + if (adapter->msix_entries) {
> + for (i = 0; i < 3; i++)
> + adapter->msix_entries[i].entry = i;
> +
> + err = pci_enable_msix(adapter->pdev,
> + adapter->msix_entries, 3);
> + }
> +
> + if (err) {
> + /* MSI-X failed */
> + dev_err(&adapter->pdev->dev,
> + "Failed to initialize MSI-X interrupts.\n");
> + igbvf_reset_interrupt_capability(adapter);

This can run pci_disable_msix() even if pci_enable_msix() falied. Seems odd.

> + }
> +}
> +
>
> ...
>
> +static int __devinit igbvf_alloc_queues(struct igbvf_adapter *adapter)
> +{
> + struct net_device *netdev = adapter->netdev;
> +
> + adapter->tx_ring = kzalloc(sizeof(struct igbvf_ring), GFP_KERNEL);
> + if (!adapter->tx_ring)
> + goto tx_err;
> + adapter->tx_ring->adapter = adapter;
> +
> + adapter->rx_ring = kzalloc(sizeof(struct igbvf_ring), GFP_KERNEL);
> + if (!adapter->rx_ring)
> + goto err;
> + adapter->rx_ring->adapter = adapter;
> +
> + netif_napi_add(netdev, &adapter->rx_ring->napi, igbvf_poll, 64);
> +
> + return 0;
> +err:
> + dev_err(&adapter->pdev->dev, "Unable to allocate memory for queues\n");
> + kfree(adapter->rx_ring);
> +tx_err:
> + kfree(adapter->tx_ring);
> + return -ENOMEM;
> +}

The error handling here is all mucked up.

>
> ...
>
> +/**
> + * igbvf_set_multi - Multicast and Promiscuous mode set
> + * @netdev: network interface device structure
> + *
> + * The set_multi entry point is called whenever the multicast address
> + * list or the network interface flags are updated. This routine is
> + * responsible for configuring the hardware for proper multicast,
> + * promiscuous mode, and all-multi behavior.
> + **/

It is conventional to terminate a kerneldoc comemnt with

*/

> +static void igbvf_set_multi(struct net_device *netdev)
> +{
> + struct igbvf_adapter *adapter = netdev_priv(netdev);
> + struct e1000_hw *hw = &adapter->hw;
> + struct dev_mc_list *mc_ptr;
> + u8 *mta_list;
> + int i;
> +
> + if (netdev->mc_count) {
> + mta_list = kmalloc(netdev->mc_count * 6, GFP_ATOMIC);
> + if (!mta_list)
> + return;

Shouldn't this be reported to someone?

> + /* prepare a packed array of only addresses. */
> + mc_ptr = netdev->mc_list;
> +
> + for (i = 0; i < netdev->mc_count; i++) {
> + if (!mc_ptr)
> + break;
> + memcpy(mta_list + (i*ETH_ALEN), mc_ptr->dmi_addr,
> + ETH_ALEN);
> + mc_ptr = mc_ptr->next;
> + }
> +
> + hw->mac.ops.update_mc_addr_list(hw, mta_list, i, 0, 0);
> + kfree(mta_list);
> + } else {
> + /*
> + * if we're called from probe, we might not have
> + * anything to do here, so clear out the list
> + */
> + hw->mac.ops.update_mc_addr_list(hw, NULL, 0, 0, 0);
> + }
> +}
> +
>
> ...
>
> +void igbvf_down(struct igbvf_adapter *adapter)
> +{
> + struct net_device *netdev = adapter->netdev;
> +
> + /*
> + * signal that we're down so the interrupt handler does not
> + * reschedule our watchdog timer
> + */
> + set_bit(__IGBVF_DOWN, &adapter->state);
> +
> + netif_stop_queue(netdev);
> +
> + /* FIX ME!!!! */
> + /* need to disable local queue transmit and receive */
> +
> + msleep(10);

Random mysterious msleep() deserves a code commeent.

> + napi_disable(&adapter->rx_ring->napi);
> +
> + igbvf_irq_disable(adapter);
> +
> + del_timer_sync(&adapter->watchdog_timer);
> +
> + netdev->tx_queue_len = adapter->tx_queue_len;
> + netif_carrier_off(netdev);
> + adapter->link_speed = 0;
> + adapter->link_duplex = 0;
> +
> + igbvf_reset(adapter);
> + igbvf_clean_tx_ring(adapter);
> + igbvf_clean_rx_ring(adapter);
> +}
> +
> +void igbvf_reinit_locked(struct igbvf_adapter *adapter)
> +{
> + might_sleep();
> + while (test_and_set_bit(__IGBVF_RESETTING, &adapter->state))
> + msleep(1);

timeout?

> + igbvf_down(adapter);
> + igbvf_up(adapter);
> + clear_bit(__IGBVF_RESETTING, &adapter->state);
> +}
> +
>
> ...
>
> +static int __devinit igbvf_probe(struct pci_dev *pdev,
> + const struct pci_device_id *ent)
> +{
> + struct net_device *netdev;
> + struct igbvf_adapter *adapter;
> + struct e1000_hw *hw;
> + const struct igbvf_info *ei = igbvf_info_tbl[ent->driver_data];
> +
> + static int cards_found;
> + int err, pci_using_dac;
> +
> + err = pci_enable_device_mem(pdev);
> + if (err)
> + return err;
> +
> + pci_using_dac = 0;
> + err = pci_set_dma_mask(pdev, DMA_64BIT_MASK);
> + if (!err) {
> + err = pci_set_consistent_dma_mask(pdev, DMA_64BIT_MASK);
> + if (!err)
> + pci_using_dac = 1;
> + } else {
> + err = pci_set_dma_mask(pdev, DMA_32BIT_MASK);
> + if (err) {
> + err = pci_set_consistent_dma_mask(pdev, DMA_32BIT_MASK);
> + if (err) {
> + dev_err(&pdev->dev, "No usable DMA "
> + "configuration, aborting\n");
> + goto err_dma;
> + }
> + }
> + }
> +
> + err = pci_request_regions(pdev, igbvf_driver_name);
> + if (err)
> + goto err_pci_reg;
> +
> + pci_set_master(pdev);
> +
> + err = -ENOMEM;
> + netdev = alloc_etherdev(sizeof(struct igbvf_adapter));
> + if (!netdev)
> + goto err_alloc_etherdev;
> +
> + SET_NETDEV_DEV(netdev, &pdev->dev);
> +
> + pci_set_drvdata(pdev, netdev);
> + adapter = netdev_priv(netdev);
> + hw = &adapter->hw;
> + adapter->netdev = netdev;
> + adapter->pdev = pdev;
> + adapter->ei = ei;
> + adapter->pba = ei->pba;
> + adapter->flags = ei->flags;
> + adapter->hw.back = adapter;
> + adapter->hw.mac.type = ei->mac;
> + adapter->msg_enable = (1 << NETIF_MSG_DRV | NETIF_MSG_PROBE) - 1;
> +
> + /* PCI config space info */
> +
> + hw->vendor_id = pdev->vendor;
> + hw->device_id = pdev->device;
> + hw->subsystem_vendor_id = pdev->subsystem_vendor;
> + hw->subsystem_device_id = pdev->subsystem_device;
> +
> + pci_read_config_byte(pdev, PCI_REVISION_ID, &hw->revision_id);
> +
> + err = -EIO;
> + adapter->hw.hw_addr = ioremap(pci_resource_start(pdev, 0),
> + pci_resource_len(pdev, 0));
> +
> + if (!adapter->hw.hw_addr)
> + goto err_ioremap;
> +
> + if (ei->get_variants) {
> + err = ei->get_variants(adapter);
> + if (err)
> + goto err_hw_init;
> + }
> +
> + /* setup adapter struct */
> + err = igbvf_sw_init(adapter);
> + if (err)
> + goto err_sw_init;
> +
> + /* construct the net_device struct */
> + netdev->netdev_ops = &igbvf_netdev_ops;
> +
> + igbvf_set_ethtool_ops(netdev);
> + netdev->watchdog_timeo = 5 * HZ;
> + strncpy(netdev->name, pci_name(pdev), sizeof(netdev->name) - 1);
> +
> + adapter->bd_number = cards_found++;
> +
> + netdev->features = NETIF_F_SG |
> + NETIF_F_IP_CSUM |
> + NETIF_F_HW_VLAN_TX |
> + NETIF_F_HW_VLAN_RX |
> + NETIF_F_HW_VLAN_FILTER;
> +
> + netdev->features |= NETIF_F_IPV6_CSUM;
> + netdev->features |= NETIF_F_TSO;
> + netdev->features |= NETIF_F_TSO6;
> +
> + if (pci_using_dac)
> + netdev->features |= NETIF_F_HIGHDMA;
> +
> + netdev->vlan_features |= NETIF_F_TSO;
> + netdev->vlan_features |= NETIF_F_TSO6;
> + netdev->vlan_features |= NETIF_F_IP_CSUM;
> + netdev->vlan_features |= NETIF_F_IPV6_CSUM;
> + netdev->vlan_features |= NETIF_F_SG;
> +
> + /*reset the controller to put the device in a known good state */
> + err = hw->mac.ops.reset_hw(hw);
> + if (err) {
> + dev_info(&pdev->dev,
> + "PF still in reset state, assigning new address\n");
> + random_ether_addr(hw->mac.addr);
> + } else {
> + err = hw->mac.ops.read_mac_addr(hw);
> + if (err) {
> + dev_err(&pdev->dev, "Error reading MAC address\n");
> + goto err_hw_init;
> + }
> + }
> +
> + memcpy(netdev->dev_addr, adapter->hw.mac.addr, netdev->addr_len);
> + memcpy(netdev->perm_addr, adapter->hw.mac.addr, netdev->addr_len);
> +
> + if (!is_valid_ether_addr(netdev->perm_addr)) {
> + dev_err(&pdev->dev, "Invalid MAC Address: "
> + "%02x:%02x:%02x:%02x:%02x:%02x\n",
> + netdev->dev_addr[0], netdev->dev_addr[1],
> + netdev->dev_addr[2], netdev->dev_addr[3],
> + netdev->dev_addr[4], netdev->dev_addr[5]);
> + err = -EIO;
> + goto err_hw_init;
> + }
> +
> + init_timer(&adapter->watchdog_timer);
> + adapter->watchdog_timer.function = &igbvf_watchdog;
> + adapter->watchdog_timer.data = (unsigned long) adapter;

Could use setup_timer() here.

> + INIT_WORK(&adapter->reset_task, igbvf_reset_task);
> + INIT_WORK(&adapter->watchdog_task, igbvf_watchdog_task);
> +
> + /* ring size defaults */
> + adapter->rx_ring->count = 1024;
> + adapter->tx_ring->count = 1024;
> +
> + /* reset the hardware with the new settings */
> + igbvf_reset(adapter);
> +
> + /* tell the stack to leave us alone until igbvf_open() is called */
> + netif_carrier_off(netdev);
> + netif_stop_queue(netdev);
> +
> + strcpy(netdev->name, "eth%d");
> + err = register_netdev(netdev);
> + if (err)
> + goto err_hw_init;
> +
> + igbvf_print_device_info(adapter);
> +
> + igbvf_initialize_last_counter_stats(adapter);
> +
> + return 0;
> +
> +err_hw_init:
> + kfree(adapter->tx_ring);
> + kfree(adapter->rx_ring);
> + igbvf_reset_interrupt_capability(adapter);
> +err_sw_init:
> + iounmap(adapter->hw.hw_addr);
> +err_ioremap:
> + free_netdev(netdev);
> +err_alloc_etherdev:
> + pci_release_regions(pdev);
> +err_pci_reg:
> +err_dma:
> + pci_disable_device(pdev);
> + return err;
> +}
> +
> +/**
> + * igbvf_remove - Device Removal Routine
> + * @pdev: PCI device information struct
> + *
> + * igbvf_remove is called by the PCI subsystem to alert the driver
> + * that it should release a PCI device. The could be caused by a
> + * Hot-Plug event, or because the driver is going to be removed from
> + * memory.
> + **/
> +static void __devexit igbvf_remove(struct pci_dev *pdev)
> +{
> + struct net_device *netdev = pci_get_drvdata(pdev);
> + struct igbvf_adapter *adapter = netdev_priv(netdev);
> + struct e1000_hw *hw = &adapter->hw;
> +
> + /*
> + * flush_scheduled work may reschedule our watchdog task, so
> + * explicitly disable watchdog tasks from being rescheduled
> + */
> + set_bit(__IGBVF_DOWN, &adapter->state);
> + del_timer_sync(&adapter->watchdog_timer);
> +
> + flush_scheduled_work();

flush_work() is a bit more modern - it will flush a specific work item
rather than waiting for the kernel-wide queue to drain.

> + unregister_netdev(netdev);
> +
> + igbvf_reset_interrupt_capability(adapter);
> + kfree(adapter->tx_ring);
> + kfree(adapter->rx_ring);
> +
> + iounmap(hw->hw_addr);
> + if (hw->flash_address)
> + iounmap(hw->flash_address);
> + pci_release_regions(pdev);
> +
> + free_netdev(netdev);
> +
> + pci_disable_device(pdev);
> +}
> +
>
> ...
>

What a lot of code.

2009-03-18 00:13:13

by Duyck, Alexander H

Subject: Re: [net-next PATCH 1/2] igbvf: add new driver to support 82576 virtual functions

Thanks for all the comments. I tried to incorporate most of them into
the igbvf driver and also ended up porting some over to our other
drivers, specifically igb since the igbvf driver copies much of the code.

I have added my comments inline below.

Thanks,

Alex

Andrew Morton wrote:
> On Tue, 10 Mar 2009 19:09:28 -0700 Jeff Kirsher <[email protected]> wrote:
>
>> From: Alexander Duyck <[email protected]>
>>
>> This adds an igbvf driver to handle virtual functions provided
>> by the igb driver.
>
> The drive-by reader is now wondering what a "virtual function" is.
>
>> ...
>>
>> +#define IGBVF_STAT(m, b) sizeof(((struct igbvf_adapter *)0)->m), \
>> + offsetof(struct igbvf_adapter, m), \
>> + offsetof(struct igbvf_adapter, b)
>>
>> +static const struct igbvf_stats igbvf_gstrings_stats[] = {
>> + { "rx_packets", IGBVF_STAT(stats.gprc, stats.base_gprc) },
>> + { "tx_packets", IGBVF_STAT(stats.gptc, stats.base_gptc) },
>> + { "rx_bytes", IGBVF_STAT(stats.gorc, stats.base_gorc) },
>> + { "tx_bytes", IGBVF_STAT(stats.gotc, stats.base_gotc) },
>> + { "multicast", IGBVF_STAT(stats.mprc, stats.base_mprc) },
>> + { "lbrx_bytes", IGBVF_STAT(stats.gorlbc, stats.base_gorlbc) },
>> + { "lbrx_packets", IGBVF_STAT(stats.gprlbc, stats.base_gprlbc) },
>> + { "tx_restart_queue", IGBVF_STAT(restart_queue, zero_base) },
>> + { "rx_long_byte_count", IGBVF_STAT(stats.gorc, stats.base_gorc) },
>> + { "rx_csum_offload_good", IGBVF_STAT(hw_csum_good, zero_base) },
>> + { "rx_csum_offload_errors", IGBVF_STAT(hw_csum_err, zero_base) },
>> + { "rx_header_split", IGBVF_STAT(rx_hdr_split, zero_base) },
>> + { "alloc_rx_buff_failed", IGBVF_STAT(alloc_rx_buff_failed, zero_base) },
>> +};
>
> <stares at it for a while>
>
> It would be clearer if `m' and `b' were (much) more meaningful identifiers.

I agree, the values have been changed to current and base.

>> +#define IGBVF_GLOBAL_STATS_LEN \
>> + (sizeof(igbvf_gstrings_stats) / sizeof(struct igbvf_stats))
>
> This is ARRAY_SIZE().
>
>> +#define IGBVF_STATS_LEN (IGBVF_GLOBAL_STATS_LEN)
>
> Why does this need to exist?

It doesn't, it has been dropped and references to it replaced with
IGBVF_GLOBAL_STATS_LEN.

>> ...
>>
>> +static int igbvf_set_ringparam(struct net_device *netdev,
>> + struct ethtool_ringparam *ring)
>> +{
>> + struct igbvf_adapter *adapter = netdev_priv(netdev);
>> + struct igbvf_ring *tx_ring, *tx_old;
>> + struct igbvf_ring *rx_ring, *rx_old;
>> + int err;
>> +
>> + if ((ring->rx_mini_pending) || (ring->rx_jumbo_pending))
>> + return -EINVAL;
>> +
>> + while (test_and_set_bit(__IGBVF_RESETTING, &adapter->state))
>> + msleep(1);
>
> No timeout needed here? Interrupts might not be working, for example..

This bit isn't set in interrupt context. This is always used out of
interrupt context and is just to prevent multiple setting changes at the
same time.

>> + if (netif_running(adapter->netdev))
>> + igbvf_down(adapter);
>> +
>> + tx_old = adapter->tx_ring;
>> + rx_old = adapter->rx_ring;
>> +
>> + err = -ENOMEM;
>> + tx_ring = kzalloc(sizeof(struct igbvf_ring), GFP_KERNEL);
>> + if (!tx_ring)
>> + goto err_alloc_tx;
>> + /*
>> + * use a memcpy to save any previously configured
>> + * items like napi structs from having to be
>> + * reinitialized
>> + */
>> + memcpy(tx_ring, tx_old, sizeof(struct igbvf_ring));
>> +
>> + rx_ring = kzalloc(sizeof(struct igbvf_ring), GFP_KERNEL);
>> + if (!rx_ring)
>> + goto err_alloc_rx;
>> + memcpy(rx_ring, rx_old, sizeof(struct igbvf_ring));
>> +
>> + adapter->tx_ring = tx_ring;
>> + adapter->rx_ring = rx_ring;
>> +
>> + rx_ring->count = max(ring->rx_pending, (u32)IGBVF_MIN_RXD);
>> + rx_ring->count = min(rx_ring->count, (u32)(IGBVF_MAX_RXD));
>> + rx_ring->count = ALIGN(rx_ring->count, REQ_RX_DESCRIPTOR_MULTIPLE);
>> +
>> + tx_ring->count = max(ring->tx_pending, (u32)IGBVF_MIN_TXD);
>> + tx_ring->count = min(tx_ring->count, (u32)(IGBVF_MAX_TXD));
>> + tx_ring->count = ALIGN(tx_ring->count, REQ_TX_DESCRIPTOR_MULTIPLE);
>> +
>> + if (netif_running(adapter->netdev)) {
>> + /* Try to get new resources before deleting old */
>> + err = igbvf_setup_rx_resources(adapter);
>> + if (err)
>> + goto err_setup_rx;
>> + err = igbvf_setup_tx_resources(adapter);
>> + if (err)
>> + goto err_setup_tx;
>> +
>> + /*
>> + * restore the old in order to free it,
>> + * then add in the new
>> + */
>> + adapter->rx_ring = rx_old;
>> + adapter->tx_ring = tx_old;
>> + igbvf_free_rx_resources(adapter);
>> + igbvf_free_tx_resources(adapter);
>> + kfree(tx_old);
>> + kfree(rx_old);
>
> That's odd-looking. Why take a copy of rx_old and tx_old when we're
> about to free them?

This whole section has been redone. The approach here didn't work well
and so I redid it to match how we do this in igb.

>> + adapter->rx_ring = rx_ring;
>> + adapter->tx_ring = tx_ring;
>> + err = igbvf_up(adapter);
>> + if (err)
>> + goto err_setup;
>> + }
>> +
>> + clear_bit(__IGBVF_RESETTING, &adapter->state);
>> + return 0;
>> +err_setup_tx:
>> + igbvf_free_rx_resources(adapter);
>> +err_setup_rx:
>> + adapter->rx_ring = rx_old;
>> + adapter->tx_ring = tx_old;
>> + kfree(rx_ring);
>> +err_alloc_rx:
>> + kfree(tx_ring);
>> +err_alloc_tx:
>> + igbvf_up(adapter);
>> +err_setup:
>> + clear_bit(__IGBVF_RESETTING, &adapter->state);
>> + return err;
>> +}
>> +
>>
>> ...
>>
>> +static void igbvf_diag_test(struct net_device *netdev,
>> + struct ethtool_test *eth_test, u64 *data)
>> +{
>> + struct igbvf_adapter *adapter = netdev_priv(netdev);
>> +
>> + set_bit(__IGBVF_TESTING, &adapter->state);
>> +
>> + /*
>> + * Link test performed before hardware reset so autoneg doesn't
>> + * interfere with test result
>> + */
>> + if (igbvf_link_test(adapter, &data[0]))
>> + eth_test->flags |= ETH_TEST_FL_FAILED;
>> +
>> + clear_bit(__IGBVF_TESTING, &adapter->state);
>> + msleep_interruptible(4 * 1000);
>
> If this process has signal_pending(), msleep_interruptible() becomes a
> no-op. I expect the driver breaks?

Actually it has no effect on the driver if the sleep fails. For the
most part the only reason for having this is to allow enough time to
guarantee that the link has recovered after the link test.

>> +}
>> +
>>
>> ...
>>
>> +static void igbvf_get_ethtool_stats(struct net_device *netdev,
>> + struct ethtool_stats *stats,
>> + u64 *data)
>> +{
>> + struct igbvf_adapter *adapter = netdev_priv(netdev);
>> + int i;
>> +
>> + igbvf_update_stats(adapter);
>> + for (i = 0; i < IGBVF_GLOBAL_STATS_LEN; i++) {
>> + char *p = (char *)adapter +
>> + igbvf_gstrings_stats[i].stat_offset;
>> + char *b = (char *)adapter +
>> + igbvf_gstrings_stats[i].base_stat_offset;
>> + data[i] = ((igbvf_gstrings_stats[i].sizeof_stat ==
>> + sizeof(u64)) ? *(u64 *)p : *(u32 *)p) -
>> + ((igbvf_gstrings_stats[i].sizeof_stat ==
>> + sizeof(u64)) ? *(u64 *)b : *(u32 *)b);
>
> yitch.
>
> Are we using the C type system as well as possible here?

Actually I think the ugliness is due to the fact that it is doing the
check itself twice. The code below is what it will look like after
being redone.

data[i] = ((igbvf_gstrings_stats[i].sizeof_stat ==
sizeof(u64)) ? (*(u64 *)p - *(u64 *)b) :
(*(u32 *)p - *(u32 *)b));

>> + }
>> +
>> +}
>> +
>>
>> ...
>>
>> +#define IGBVF_DESC_UNUSED(R) \
>> + ((((R)->next_to_clean > (R)->next_to_use) ? 0 : (R)->count) + \
>> + (R)->next_to_clean - (R)->next_to_use - 1)
>
> my eyes.
>
> This macro evaluates its argument multiple times and will misbehave (or
> be slow) if passed an expreesion with side-effects.
>
> Why not just write a plain old C function for this?

Actually that was already done and the macro was unused. It has been
dropped.

>> +#define IGBVF_RX_DESC_ADV(R, i) \
>> + (&(((union e1000_adv_rx_desc *)((R).desc))[i]))
>> +#define IGBVF_TX_DESC_ADV(R, i) \
>> + (&(((union e1000_adv_tx_desc *)((R).desc))[i]))
>> +#define IGBVF_TX_CTXTDESC_ADV(R, i) \
>> + (&(((struct e1000_adv_tx_context_desc *)((R).desc))[i]))
>
> Maybe igbvf_ring.desc should have had type
>
> union {
> union e1000_adv_rx_desc;
> union e1000_adv_tx_desc;
> union e1000_adv_tx_context_desc;
> } *
>
> Or something.

By switching these over to a typecast I can get away from some of the
ugliness but it is still there. Below are the changes I made to support
what you suggested.

#define IGBVF_RX_DESC_ADV(R, i) \
(&((((R).desc))[i].rx_desc))
#define IGBVF_TX_DESC_ADV(R, i) \
(&((((R).desc))[i].tx_desc))
#define IGBVF_TX_CTXTDESC_ADV(R, i) \
(&((((R).desc))[i].tx_context_desc))



union igbvf_desc {
union e1000_adv_rx_desc rx_desc;
union e1000_adv_tx_desc tx_desc;
struct e1000_adv_tx_context_desc tx_context_desc;
};

>> ...
>>
>> +/**
>> + * igbvf_alloc_rx_buffers - Replace used receive buffers; packet split
>> + * @rx_ring: address of ring structure to repopulate
>> + * @cleaned_count: number of buffers to repopulate
>> + **/
>> +static void igbvf_alloc_rx_buffers(struct igbvf_ring *rx_ring,
>> + int cleaned_count)
>> +{
>> + struct igbvf_adapter *adapter = rx_ring->adapter;
>> + struct net_device *netdev = adapter->netdev;
>> + struct pci_dev *pdev = adapter->pdev;
>> + union e1000_adv_rx_desc *rx_desc;
>> + struct igbvf_buffer *buffer_info;
>> + struct sk_buff *skb;
>> + unsigned int i;
>> + int bufsz;
>> +
>> + i = rx_ring->next_to_use;
>> + buffer_info = &rx_ring->buffer_info[i];
>> +
>> + if (adapter->rx_ps_hdr_size)
>> + bufsz = adapter->rx_ps_hdr_size;
>> + else
>> + bufsz = adapter->rx_buffer_len;
>> + bufsz += NET_IP_ALIGN;
>> +
>> + while (cleaned_count--) {
>> + rx_desc = IGBVF_RX_DESC_ADV(*rx_ring, i);
>> +
>> + if (adapter->rx_ps_hdr_size && !buffer_info->page_dma) {
>> + if (!buffer_info->page) {
>> + buffer_info->page = alloc_page(GFP_ATOMIC);
>> + if (!buffer_info->page) {
>> + adapter->alloc_rx_buff_failed++;
>> + goto no_buffers;
>> + }
>> + buffer_info->page_offset = 0;
>> + } else {
>> + buffer_info->page_offset ^= PAGE_SIZE / 2;
>> + }
>> + buffer_info->page_dma =
>> + pci_map_page(pdev, buffer_info->page,
>> + buffer_info->page_offset,
>> + PAGE_SIZE / 2,
>> + PCI_DMA_FROMDEVICE);
>> + }
>> +
>> + if (!buffer_info->skb) {
>> + skb = netdev_alloc_skb(netdev, bufsz);
>> + if (!skb) {
>> + adapter->alloc_rx_buff_failed++;
>> + goto no_buffers;
>> + }
>> +
>> + /* Make buffer alignment 2 beyond a 16 byte boundary
>> + * this will result in a 16 byte aligned IP header after
>> + * the 14 byte MAC header is removed
>> + */
>> + skb_reserve(skb, NET_IP_ALIGN);
>> +
>> + buffer_info->skb = skb;
>> + buffer_info->dma = pci_map_single(pdev, skb->data,
>> + bufsz,
>> + PCI_DMA_FROMDEVICE);
>> + }
>> + /* Refresh the desc even if buffer_addrs didn't change because
>> + * each write-back erases this info. */
>> + if (adapter->rx_ps_hdr_size) {
>> + rx_desc->read.pkt_addr =
>> + cpu_to_le64(buffer_info->page_dma);
>> + rx_desc->read.hdr_addr = cpu_to_le64(buffer_info->dma);
>> + } else {
>> + rx_desc->read.pkt_addr =
>> + cpu_to_le64(buffer_info->dma);
>> + rx_desc->read.hdr_addr = 0;
>> + }
>> +
>> + i++;
>> + if (i == rx_ring->count)
>> + i = 0;
>> + buffer_info = &rx_ring->buffer_info[i];
>> + }
>> +
>> +no_buffers:
>> + if (rx_ring->next_to_use != i) {
>> + rx_ring->next_to_use = i;
>> + if (i == 0)
>> + i = (rx_ring->count - 1);
>> + else
>> + 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). */
>> + wmb();
>> + writel(i, adapter->hw.hw_addr + rx_ring->tail);
>
> Does wmb() work reliably for masters other than CPUs? I lose track...

For SR-IOV it is always possible that someone might decide to run some
of the VFs in their DOM0 kernel so it is best to leave it in there to
support it just in case.

>> + }
>> +}
>> +
>>
>> ...
>>
>> +/**
>> + * igbvf_setup_tx_resources - allocate Tx resources (Descriptors)
>> + * @adapter: board private structure
>> + *
>> + * Return 0 on success, negative on failure
>> + **/
>> +int igbvf_setup_tx_resources(struct igbvf_adapter *adapter)
>> +{
>> + struct igbvf_ring *tx_ring = adapter->tx_ring;
>> + int err = -ENOMEM, size;
>> +
>> + size = sizeof(struct igbvf_buffer) * tx_ring->count;
>> + tx_ring->buffer_info = vmalloc(size);
>
> How large is this likely to be?

This can get to be up to 4K * size of igbvf_buffer which int total comes
to something like 32K or more.

>> + if (!tx_ring->buffer_info)
>> + goto err;
>> + memset(tx_ring->buffer_info, 0, size);
>> +
>> + /* round up to nearest 4K */
>> + tx_ring->size = tx_ring->count * sizeof(union e1000_adv_tx_desc);
>> + tx_ring->size = ALIGN(tx_ring->size, 4096);
>> +
>> + err = igbvf_alloc_ring_dma(adapter, tx_ring);
>> + if (err)
>> + goto err;
>> +
>> + tx_ring->next_to_use = 0;
>> + tx_ring->next_to_clean = 0;
>> + spin_lock_init(&adapter->tx_queue_lock);
>> +
>> + return 0;
>> +err:
>> + vfree(tx_ring->buffer_info);
>> + dev_err(&adapter->pdev->dev,
>> + "Unable to allocate memory for the transmit descriptor ring\n");
>> + return err;
>> +}
>> +
>>
>> ...
>>
>> +void igbvf_set_interrupt_capability(struct igbvf_adapter *adapter)
>> +{
>> + int err = -ENOMEM;
>> + int i;
>> +
>> + /* we allocate 3 vectors, 1 for tx, 1 for rx, one for pf messages */
>> + adapter->msix_entries = kcalloc(3, sizeof(struct msix_entry),
>> + GFP_KERNEL);
>> + if (adapter->msix_entries) {
>> + for (i = 0; i < 3; i++)
>> + adapter->msix_entries[i].entry = i;
>> +
>> + err = pci_enable_msix(adapter->pdev,
>> + adapter->msix_entries, 3);
>> + }
>> +
>> + if (err) {
>> + /* MSI-X failed */
>> + dev_err(&adapter->pdev->dev,
>> + "Failed to initialize MSI-X interrupts.\n");
>> + igbvf_reset_interrupt_capability(adapter);
>
> This can run pci_disable_msix() even if pci_enable_msix() falied. Seems odd.

Calling pci_disable_msix is safe since the first thing it checks for is
to verify if msix is enabled. If it isn't the function does nothing.

>> + }
>> +}
>> +
>>
>> ...
>>
>> +static int __devinit igbvf_alloc_queues(struct igbvf_adapter *adapter)
>> +{
>> + struct net_device *netdev = adapter->netdev;
>> +
>> + adapter->tx_ring = kzalloc(sizeof(struct igbvf_ring), GFP_KERNEL);
>> + if (!adapter->tx_ring)
>> + goto tx_err;
>> + adapter->tx_ring->adapter = adapter;
>> +
>> + adapter->rx_ring = kzalloc(sizeof(struct igbvf_ring), GFP_KERNEL);
>> + if (!adapter->rx_ring)
>> + goto err;
>> + adapter->rx_ring->adapter = adapter;
>> +
>> + netif_napi_add(netdev, &adapter->rx_ring->napi, igbvf_poll, 64);
>> +
>> + return 0;
>> +err:
>> + dev_err(&adapter->pdev->dev, "Unable to allocate memory for queues\n");
>> + kfree(adapter->rx_ring);
>> +tx_err:
>> + kfree(adapter->tx_ring);
>> + return -ENOMEM;
>> +}
>
> The error handling here is all mucked up.

I agree, it has been fixed.

>> ...
>>
>> +/**
>> + * igbvf_set_multi - Multicast and Promiscuous mode set
>> + * @netdev: network interface device structure
>> + *
>> + * The set_multi entry point is called whenever the multicast address
>> + * list or the network interface flags are updated. This routine is
>> + * responsible for configuring the hardware for proper multicast,
>> + * promiscuous mode, and all-multi behavior.
>> + **/
>
> It is conventional to terminate a kerneldoc comemnt with
>
> */

This is the coding convention we have been using this format for all of
our functions for a long time. If you check e1000 or igb you will see
they all use this format for function comments.

>> +static void igbvf_set_multi(struct net_device *netdev)
>> +{
>> + struct igbvf_adapter *adapter = netdev_priv(netdev);
>> + struct e1000_hw *hw = &adapter->hw;
>> + struct dev_mc_list *mc_ptr;
>> + u8 *mta_list;
>> + int i;
>> +
>> + if (netdev->mc_count) {
>> + mta_list = kmalloc(netdev->mc_count * 6, GFP_ATOMIC);
>> + if (!mta_list)
>> + return;
>
> Shouldn't this be reported to someone?

I've added dev_err message if this fails, but other than that it is just
a failure to set a receive filter so it isn't a showstopper. Also the
call itself is a void so I have no way of notifying the stack of the error.

>> + /* prepare a packed array of only addresses. */
>> + mc_ptr = netdev->mc_list;
>> +
>> + for (i = 0; i < netdev->mc_count; i++) {
>> + if (!mc_ptr)
>> + break;
>> + memcpy(mta_list + (i*ETH_ALEN), mc_ptr->dmi_addr,
>> + ETH_ALEN);
>> + mc_ptr = mc_ptr->next;
>> + }
>> +
>> + hw->mac.ops.update_mc_addr_list(hw, mta_list, i, 0, 0);
>> + kfree(mta_list);
>> + } else {
>> + /*
>> + * if we're called from probe, we might not have
>> + * anything to do here, so clear out the list
>> + */
>> + hw->mac.ops.update_mc_addr_list(hw, NULL, 0, 0, 0);
>> + }
>> +}
>> +
>>
>> ...
>>
>> +void igbvf_down(struct igbvf_adapter *adapter)
>> +{
>> + struct net_device *netdev = adapter->netdev;
>> +
>> + /*
>> + * signal that we're down so the interrupt handler does not
>> + * reschedule our watchdog timer
>> + */
>> + set_bit(__IGBVF_DOWN, &adapter->state);
>> +
>> + netif_stop_queue(netdev);
>> +
>> + /* FIX ME!!!! */
>> + /* need to disable local queue transmit and receive */
>> +
>> + msleep(10);
>
> Random mysterious msleep() deserves a code commeent.

I redid the entire section above. This is one spot I missed when I was
cleaning up all this code.

>> + napi_disable(&adapter->rx_ring->napi);
>> +
>> + igbvf_irq_disable(adapter);
>> +
>> + del_timer_sync(&adapter->watchdog_timer);
>> +
>> + netdev->tx_queue_len = adapter->tx_queue_len;
>> + netif_carrier_off(netdev);
>> + adapter->link_speed = 0;
>> + adapter->link_duplex = 0;
>> +
>> + igbvf_reset(adapter);
>> + igbvf_clean_tx_ring(adapter);
>> + igbvf_clean_rx_ring(adapter);
>> +}
>> +
>> +void igbvf_reinit_locked(struct igbvf_adapter *adapter)
>> +{
>> + might_sleep();
>> + while (test_and_set_bit(__IGBVF_RESETTING, &adapter->state))
>> + msleep(1);
>
> timeout?

Once again this isn't called normally in interrupt context so it is safe
here since if we set the bit we should clear it when we are done.

>> + igbvf_down(adapter);
>> + igbvf_up(adapter);
>> + clear_bit(__IGBVF_RESETTING, &adapter->state);
>> +}
>> +
>>
>> ...
>>
>> +static int __devinit igbvf_probe(struct pci_dev *pdev,
>> + const struct pci_device_id *ent)
>> +{
>> + struct net_device *netdev;
>> + struct igbvf_adapter *adapter;
>> + struct e1000_hw *hw;
>> + const struct igbvf_info *ei = igbvf_info_tbl[ent->driver_data];
>> +
>> + static int cards_found;
>> + int err, pci_using_dac;
>> +
>> + err = pci_enable_device_mem(pdev);
>> + if (err)
>> + return err;
>> +
>> + pci_using_dac = 0;
>> + err = pci_set_dma_mask(pdev, DMA_64BIT_MASK);
>> + if (!err) {
>> + err = pci_set_consistent_dma_mask(pdev, DMA_64BIT_MASK);
>> + if (!err)
>> + pci_using_dac = 1;
>> + } else {
>> + err = pci_set_dma_mask(pdev, DMA_32BIT_MASK);
>> + if (err) {
>> + err = pci_set_consistent_dma_mask(pdev, DMA_32BIT_MASK);
>> + if (err) {
>> + dev_err(&pdev->dev, "No usable DMA "
>> + "configuration, aborting\n");
>> + goto err_dma;
>> + }
>> + }
>> + }
>> +
>> + err = pci_request_regions(pdev, igbvf_driver_name);
>> + if (err)
>> + goto err_pci_reg;
>> +
>> + pci_set_master(pdev);
>> +
>> + err = -ENOMEM;
>> + netdev = alloc_etherdev(sizeof(struct igbvf_adapter));
>> + if (!netdev)
>> + goto err_alloc_etherdev;
>> +
>> + SET_NETDEV_DEV(netdev, &pdev->dev);
>> +
>> + pci_set_drvdata(pdev, netdev);
>> + adapter = netdev_priv(netdev);
>> + hw = &adapter->hw;
>> + adapter->netdev = netdev;
>> + adapter->pdev = pdev;
>> + adapter->ei = ei;
>> + adapter->pba = ei->pba;
>> + adapter->flags = ei->flags;
>> + adapter->hw.back = adapter;
>> + adapter->hw.mac.type = ei->mac;
>> + adapter->msg_enable = (1 << NETIF_MSG_DRV | NETIF_MSG_PROBE) - 1;
>> +
>> + /* PCI config space info */
>> +
>> + hw->vendor_id = pdev->vendor;
>> + hw->device_id = pdev->device;
>> + hw->subsystem_vendor_id = pdev->subsystem_vendor;
>> + hw->subsystem_device_id = pdev->subsystem_device;
>> +
>> + pci_read_config_byte(pdev, PCI_REVISION_ID, &hw->revision_id);
>> +
>> + err = -EIO;
>> + adapter->hw.hw_addr = ioremap(pci_resource_start(pdev, 0),
>> + pci_resource_len(pdev, 0));
>> +
>> + if (!adapter->hw.hw_addr)
>> + goto err_ioremap;
>> +
>> + if (ei->get_variants) {
>> + err = ei->get_variants(adapter);
>> + if (err)
>> + goto err_hw_init;
>> + }
>> +
>> + /* setup adapter struct */
>> + err = igbvf_sw_init(adapter);
>> + if (err)
>> + goto err_sw_init;
>> +
>> + /* construct the net_device struct */
>> + netdev->netdev_ops = &igbvf_netdev_ops;
>> +
>> + igbvf_set_ethtool_ops(netdev);
>> + netdev->watchdog_timeo = 5 * HZ;
>> + strncpy(netdev->name, pci_name(pdev), sizeof(netdev->name) - 1);
>> +
>> + adapter->bd_number = cards_found++;
>> +
>> + netdev->features = NETIF_F_SG |
>> + NETIF_F_IP_CSUM |
>> + NETIF_F_HW_VLAN_TX |
>> + NETIF_F_HW_VLAN_RX |
>> + NETIF_F_HW_VLAN_FILTER;
>> +
>> + netdev->features |= NETIF_F_IPV6_CSUM;
>> + netdev->features |= NETIF_F_TSO;
>> + netdev->features |= NETIF_F_TSO6;
>> +
>> + if (pci_using_dac)
>> + netdev->features |= NETIF_F_HIGHDMA;
>> +
>> + netdev->vlan_features |= NETIF_F_TSO;
>> + netdev->vlan_features |= NETIF_F_TSO6;
>> + netdev->vlan_features |= NETIF_F_IP_CSUM;
>> + netdev->vlan_features |= NETIF_F_IPV6_CSUM;
>> + netdev->vlan_features |= NETIF_F_SG;
>> +
>> + /*reset the controller to put the device in a known good state */
>> + err = hw->mac.ops.reset_hw(hw);
>> + if (err) {
>> + dev_info(&pdev->dev,
>> + "PF still in reset state, assigning new address\n");
>> + random_ether_addr(hw->mac.addr);
>> + } else {
>> + err = hw->mac.ops.read_mac_addr(hw);
>> + if (err) {
>> + dev_err(&pdev->dev, "Error reading MAC address\n");
>> + goto err_hw_init;
>> + }
>> + }
>> +
>> + memcpy(netdev->dev_addr, adapter->hw.mac.addr, netdev->addr_len);
>> + memcpy(netdev->perm_addr, adapter->hw.mac.addr, netdev->addr_len);
>> +
>> + if (!is_valid_ether_addr(netdev->perm_addr)) {
>> + dev_err(&pdev->dev, "Invalid MAC Address: "
>> + "%02x:%02x:%02x:%02x:%02x:%02x\n",
>> + netdev->dev_addr[0], netdev->dev_addr[1],
>> + netdev->dev_addr[2], netdev->dev_addr[3],
>> + netdev->dev_addr[4], netdev->dev_addr[5]);
>> + err = -EIO;
>> + goto err_hw_init;
>> + }
>> +
>> + init_timer(&adapter->watchdog_timer);
>> + adapter->watchdog_timer.function = &igbvf_watchdog;
>> + adapter->watchdog_timer.data = (unsigned long) adapter;
>
> Could use setup_timer() here.

done.

>> + INIT_WORK(&adapter->reset_task, igbvf_reset_task);
>> + INIT_WORK(&adapter->watchdog_task, igbvf_watchdog_task);
>> +
>> + /* ring size defaults */
>> + adapter->rx_ring->count = 1024;
>> + adapter->tx_ring->count = 1024;
>> +
>> + /* reset the hardware with the new settings */
>> + igbvf_reset(adapter);
>> +
>> + /* tell the stack to leave us alone until igbvf_open() is called */
>> + netif_carrier_off(netdev);
>> + netif_stop_queue(netdev);
>> +
>> + strcpy(netdev->name, "eth%d");
>> + err = register_netdev(netdev);
>> + if (err)
>> + goto err_hw_init;
>> +
>> + igbvf_print_device_info(adapter);
>> +
>> + igbvf_initialize_last_counter_stats(adapter);
>> +
>> + return 0;
>> +
>> +err_hw_init:
>> + kfree(adapter->tx_ring);
>> + kfree(adapter->rx_ring);
>> + igbvf_reset_interrupt_capability(adapter);
>> +err_sw_init:
>> + iounmap(adapter->hw.hw_addr);
>> +err_ioremap:
>> + free_netdev(netdev);
>> +err_alloc_etherdev:
>> + pci_release_regions(pdev);
>> +err_pci_reg:
>> +err_dma:
>> + pci_disable_device(pdev);
>> + return err;
>> +}
>> +
>> +/**
>> + * igbvf_remove - Device Removal Routine
>> + * @pdev: PCI device information struct
>> + *
>> + * igbvf_remove is called by the PCI subsystem to alert the driver
>> + * that it should release a PCI device. The could be caused by a
>> + * Hot-Plug event, or because the driver is going to be removed from
>> + * memory.
>> + **/
>> +static void __devexit igbvf_remove(struct pci_dev *pdev)
>> +{
>> + struct net_device *netdev = pci_get_drvdata(pdev);
>> + struct igbvf_adapter *adapter = netdev_priv(netdev);
>> + struct e1000_hw *hw = &adapter->hw;
>> +
>> + /*
>> + * flush_scheduled work may reschedule our watchdog task, so
>> + * explicitly disable watchdog tasks from being rescheduled
>> + */
>> + set_bit(__IGBVF_DOWN, &adapter->state);
>> + del_timer_sync(&adapter->watchdog_timer);
>> +
>> + flush_scheduled_work();
>
> flush_work() is a bit more modern - it will flush a specific work item
> rather than waiting for the kernel-wide queue to drain.

I think we want to use flush_scheduled_work here since there is also the
dev watchdog timer which could be in the middle of running while we are
shutting down if we don't flush all work queues.

>> + unregister_netdev(netdev);
>> +
>> + igbvf_reset_interrupt_capability(adapter);
>> + kfree(adapter->tx_ring);
>> + kfree(adapter->rx_ring);
>> +
>> + iounmap(hw->hw_addr);
>> + if (hw->flash_address)
>> + iounmap(hw->flash_address);
>> + pci_release_regions(pdev);
>> +
>> + free_netdev(netdev);
>> +
>> + pci_disable_device(pdev);
>> +}
>> +
>>
>> ...
>>
>
> What a lot of code.

I know it is a lot of code, but it is a new driver so it can't really be
helped.

2009-03-18 01:17:51

by Andrew Morton

Subject: Re: [net-next PATCH 1/2] igbvf: add new driver to support 82576 virtual functions

On Tue, 17 Mar 2009 17:12:48 -0700 Alexander Duyck <[email protected]> wrote:

> Thanks for all the comments. I tried to incorporate most of them into
> the igbvf driver and also ended up porting some over to our other
> drivers, specifically igb since the igbvf driver copies much of the code.
>
> I have added my comments inline below.
>
> Thanks,
>
> Alex
>
> Andrew Morton wrote:
> > On Tue, 10 Mar 2009 19:09:28 -0700 Jeff Kirsher <[email protected]> wrote:
> >
> >> From: Alexander Duyck <[email protected]>
> >>
> >> This adds an igbvf driver to handle virtual functions provided
> >> by the igb driver.
> >
> > The drive-by reader is now wondering what a "virtual function" is.

^^ this comment was missed.

I was indirectly asking for an overview (preferably in the changelog) of
what the whole patch actually does.

> >>
> >> +static int igbvf_set_ringparam(struct net_device *netdev,
> >> + struct ethtool_ringparam *ring)
> >> +{
> >> + struct igbvf_adapter *adapter = netdev_priv(netdev);
> >> + struct igbvf_ring *tx_ring, *tx_old;
> >> + struct igbvf_ring *rx_ring, *rx_old;
> >> + int err;
> >> +
> >> + if ((ring->rx_mini_pending) || (ring->rx_jumbo_pending))
> >> + return -EINVAL;
> >> +
> >> + while (test_and_set_bit(__IGBVF_RESETTING, &adapter->state))
> >> + msleep(1);
> >
> > No timeout needed here? Interrupts might not be working, for example..
>
> This bit isn't set in interrupt context. This is always used out of
> interrupt context and is just to prevent multiple setting changes at the
> same time.

Oh. Can't use plain old mutex_lock()?

2009-03-18 15:23:02

by Duyck, Alexander H

Subject: Re: [net-next PATCH 1/2] igbvf: add new driver to support 82576 virtual functions

Andrew Morton wrote:
> On Tue, 17 Mar 2009 17:12:48 -0700 Alexander Duyck <[email protected]> wrote:
>
>> Thanks for all the comments. I tried to incorporate most of them into
>> the igbvf driver and also ended up porting some over to our other
>> drivers, specifically igb since the igbvf driver copies much of the code.
>>
>> I have added my comments inline below.
>>
>> Thanks,
>>
>> Alex
>>
>> Andrew Morton wrote:
>>> On Tue, 10 Mar 2009 19:09:28 -0700 Jeff Kirsher <[email protected]> wrote:
>>>
>>>> From: Alexander Duyck <[email protected]>
>>>>
>>>> This adds an igbvf driver to handle virtual functions provided
>>>> by the igb driver.
>>> The drive-by reader is now wondering what a "virtual function" is.
>
> ^^ this comment was missed.
>
> I was indirectly asking for an overview (preferably in the changelog) of
> what the whole patch actually does.

Sorry, while I missed the comment in my response I had gotten to
addressing it in the next version. I updated it to more thoroughly
describe what the VF driver is doing. I also included instructions on
how to enable the VFs from the PF so that they can be tested.

>>>> +static int igbvf_set_ringparam(struct net_device *netdev,
>>>> + struct ethtool_ringparam *ring)
>>>> +{
>>>> + struct igbvf_adapter *adapter = netdev_priv(netdev);
>>>> + struct igbvf_ring *tx_ring, *tx_old;
>>>> + struct igbvf_ring *rx_ring, *rx_old;
>>>> + int err;
>>>> +
>>>> + if ((ring->rx_mini_pending) || (ring->rx_jumbo_pending))
>>>> + return -EINVAL;
>>>> +
>>>> + while (test_and_set_bit(__IGBVF_RESETTING, &adapter->state))
>>>> + msleep(1);
>>> No timeout needed here? Interrupts might not be working, for example..
>> This bit isn't set in interrupt context. This is always used out of
>> interrupt context and is just to prevent multiple setting changes at the
>> same time.
>
> Oh. Can't use plain old mutex_lock()?

We have one or two spots that actually check to see if the bit is set
and just report a warning instead of actually waiting on the bit to clear.

2009-03-18 22:03:26

by Andrew Morton

Subject: Re: [net-next PATCH 1/2] igbvf: add new driver to support 82576 virtual functions

On Wed, 18 Mar 2009 08:22:46 -0700 Alexander Duyck <[email protected]> wrote:

> >>>> +static int igbvf_set_ringparam(struct net_device *netdev,
> >>>> + struct ethtool_ringparam *ring)
> >>>> +{
> >>>> + struct igbvf_adapter *adapter = netdev_priv(netdev);
> >>>> + struct igbvf_ring *tx_ring, *tx_old;
> >>>> + struct igbvf_ring *rx_ring, *rx_old;
> >>>> + int err;
> >>>> +
> >>>> + if ((ring->rx_mini_pending) || (ring->rx_jumbo_pending))
> >>>> + return -EINVAL;
> >>>> +
> >>>> + while (test_and_set_bit(__IGBVF_RESETTING, &adapter->state))
> >>>> + msleep(1);
> >>> No timeout needed here? Interrupts might not be working, for example..
> >> This bit isn't set in interrupt context. This is always used out of
> >> interrupt context and is just to prevent multiple setting changes at the
> >> same time.
> >
> > Oh. Can't use plain old mutex_lock()?
>
> We have one or two spots that actually check to see if the bit is set
> and just report a warning instead of actually waiting on the bit to clear.

mutex_is_locked()?

2009-03-19 00:41:08

by Duyck, Alexander H

Subject: Re: [net-next PATCH 1/2] igbvf: add new driver to support 82576 virtual functions

Andrew Morton wrote:
> On Wed, 18 Mar 2009 08:22:46 -0700 Alexander Duyck <[email protected]> wrote:
>
>>>>>> +static int igbvf_set_ringparam(struct net_device *netdev,
>>>>>> + struct ethtool_ringparam *ring)
>>>>>> +{
>>>>>> + struct igbvf_adapter *adapter = netdev_priv(netdev);
>>>>>> + struct igbvf_ring *tx_ring, *tx_old;
>>>>>> + struct igbvf_ring *rx_ring, *rx_old;
>>>>>> + int err;
>>>>>> +
>>>>>> + if ((ring->rx_mini_pending) || (ring->rx_jumbo_pending))
>>>>>> + return -EINVAL;
>>>>>> +
>>>>>> + while (test_and_set_bit(__IGBVF_RESETTING, &adapter->state))
>>>>>> + msleep(1);
>>>>> No timeout needed here? Interrupts might not be working, for example..
>>>> This bit isn't set in interrupt context. This is always used out of
>>>> interrupt context and is just to prevent multiple setting changes at the
>>>> same time.
>>> Oh. Can't use plain old mutex_lock()?
>> We have one or two spots that actually check to see if the bit is set
>> and just report a warning instead of actually waiting on the bit to clear.
>
> mutex_is_locked()?

I suppose that would work, but I still would prefer to keep this bit of
code as it is. My main motivation is just to use what was already
proven, and the fact is the e1000, e1000e, igb, and several other
drivers all use this same approach and it works.

I don't think we need the extra overhead of the mutex lock since most of
the calls that end up setting the __IGBVF_RESETTING bit will already be
wrapped within rtnl_lock/unlock calls. As far as I can tell it looks
like the only two threads that would ever be competing for the lock
would be the igbvf_reinit_locked and whatever ethtool or ifconfig
requests that decide to make changes to the configuration of the netdevice.

Thanks,

Alex

2009-03-19 02:06:31

by Andrew Morton

Subject: Re: [net-next PATCH 1/2] igbvf: add new driver to support 82576 virtual functions

On Wed, 18 Mar 2009 17:40:47 -0700 Alexander Duyck <[email protected]> wrote:

> Andrew Morton wrote:
> > On Wed, 18 Mar 2009 08:22:46 -0700 Alexander Duyck <[email protected]> wrote:
> >
> >>>>>> +static int igbvf_set_ringparam(struct net_device *netdev,
> >>>>>> + struct ethtool_ringparam *ring)
> >>>>>> +{
> >>>>>> + struct igbvf_adapter *adapter = netdev_priv(netdev);
> >>>>>> + struct igbvf_ring *tx_ring, *tx_old;
> >>>>>> + struct igbvf_ring *rx_ring, *rx_old;
> >>>>>> + int err;
> >>>>>> +
> >>>>>> + if ((ring->rx_mini_pending) || (ring->rx_jumbo_pending))
> >>>>>> + return -EINVAL;
> >>>>>> +
> >>>>>> + while (test_and_set_bit(__IGBVF_RESETTING, &adapter->state))
> >>>>>> + msleep(1);
> >>>>> No timeout needed here? Interrupts might not be working, for example..
> >>>> This bit isn't set in interrupt context. This is always used out of
> >>>> interrupt context and is just to prevent multiple setting changes at the
> >>>> same time.
> >>> Oh. Can't use plain old mutex_lock()?
> >> We have one or two spots that actually check to see if the bit is set
> >> and just report a warning instead of actually waiting on the bit to clear.
> >
> > mutex_is_locked()?
>
> I suppose that would work, but I still would prefer to keep this bit of
> code as it is. My main motivation is just to use what was already
> proven, and the fact is the e1000, e1000e, igb, and several other
> drivers all use this same approach and it works.

OK, that's a reason.

> I don't think we need the extra overhead of the mutex lock since most of
> the calls that end up setting the __IGBVF_RESETTING bit will already be
> wrapped within rtnl_lock/unlock calls. As far as I can tell it looks
> like the only two threads that would ever be competing for the lock
> would be the igbvf_reinit_locked and whatever ethtool or ifconfig
> requests that decide to make changes to the configuration of the netdevice.

You may well find that mutex_lock is more efficient than setting a timer
and waking up once per millisecond. Certainly much lower latency on some
setups given clock granularities of as much as 10 milliseconds.

But it sounds like that's all a separate standalone exercise.